commit 94de5c7d77c1615bd3b4c57733bb13e9777e39bf Author: ModelHub XC Date: Thu Jul 30 06:51:18 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.05 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..64058be --- /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_reasoning.Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_sciencev00.05 +tags: +- generated_from_trainer +- trl +- open-r1 +- sft +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_sciencev00.05 + +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_reasoning.Llama3_4096toks](https://huggingface.co/datasets/Neelectric/Replay_0.01.MoT_science.wildguardmix_reasoning.Llama3_4096toks) dataset. +It has been trained using [TRL](https://github.com/huggingface/trl). + +## Quick start + +```python +from transformers import pipeline + +question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" +generator = pipeline("text-generation", model="Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.05", 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/8lv8qnxp) + + +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..82a021d --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.1067883493711675e+19, + "train_loss": 0.9522557386704232, + "train_runtime": 21515.8506, + "train_samples": 145693, + "train_samples_per_second": 13.543, + "train_steps_per_second": 0.846 +} \ 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..b231f84 --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:919b83c6ade05ba4a0ba302436353fed84d48be4401ca2ca764dc5ef22e62487 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..2458f23 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3d5dcffdb5a3bec91259c6a9c0c6722ecba8564a04bfeea3cb1a665f6fe99f1 +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..7263699 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de7af651e1db074b696b08ea6e15302f2385b0d3ef8cfd0108af05503e849f07 +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..9f56393 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8330d8d421d2d0a61c1e7fedc7926721c2c0d22393cddd5ecba081055c5d010 +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..82a021d --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.1067883493711675e+19, + "train_loss": 0.9522557386704232, + "train_runtime": 21515.8506, + "train_samples": 145693, + "train_samples_per_second": 13.543, + "train_steps_per_second": 0.846 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..ce8e0bb --- /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": 7.3010406494140625, + "learning_rate": 0.0, + "loss": 1.4841, + "mean_token_accuracy": 0.6132287979125977, + "num_tokens": 25099.0, + "step": 1 + }, + { + "epoch": 0.00021963540522732265, + "grad_norm": 6.589611053466797, + "learning_rate": 5.48847420417124e-10, + "loss": 1.4857, + "mean_token_accuracy": 0.6123285293579102, + "num_tokens": 54691.0, + "step": 2 + }, + { + "epoch": 0.000329453107840984, + "grad_norm": 6.737905502319336, + "learning_rate": 1.097694840834248e-09, + "loss": 1.4471, + "mean_token_accuracy": 0.617664098739624, + "num_tokens": 81754.0, + "step": 3 + }, + { + "epoch": 0.0004392708104546453, + "grad_norm": 7.582892894744873, + "learning_rate": 1.646542261251372e-09, + "loss": 1.5597, + "mean_token_accuracy": 0.6015436053276062, + "num_tokens": 106869.0, + "step": 4 + }, + { + "epoch": 0.0005490885130683066, + "grad_norm": 7.864223480224609, + "learning_rate": 2.195389681668496e-09, + "loss": 1.5444, + "mean_token_accuracy": 0.6106245517730713, + "num_tokens": 128527.0, + "step": 5 + }, + { + "epoch": 0.000658906215681968, + "grad_norm": 7.762698173522949, + "learning_rate": 2.7442371020856205e-09, + "loss": 1.4747, + "mean_token_accuracy": 0.6183778047561646, + "num_tokens": 151051.0, + "step": 6 + }, + { + "epoch": 0.0007687239182956292, + "grad_norm": 7.0887627601623535, + "learning_rate": 3.293084522502744e-09, + "loss": 1.4695, + "mean_token_accuracy": 0.615016758441925, + "num_tokens": 177848.0, + "step": 7 + }, + { + "epoch": 0.0008785416209092906, + "grad_norm": 7.960464954376221, + "learning_rate": 3.841931942919868e-09, + "loss": 1.5502, + "mean_token_accuracy": 0.5964752435684204, + "num_tokens": 200844.0, + "step": 8 + }, + { + "epoch": 0.000988359323522952, + "grad_norm": 6.688938140869141, + "learning_rate": 4.390779363336992e-09, + "loss": 1.3834, + "mean_token_accuracy": 0.6349782347679138, + "num_tokens": 228037.0, + "step": 9 + }, + { + "epoch": 0.0010981770261366132, + "grad_norm": 8.330236434936523, + "learning_rate": 4.939626783754116e-09, + "loss": 1.638, + "mean_token_accuracy": 0.5829293727874756, + "num_tokens": 250132.0, + "step": 10 + }, + { + "epoch": 0.0012079947287502745, + "grad_norm": 6.212152481079102, + "learning_rate": 5.488474204171241e-09, + "loss": 1.5231, + "mean_token_accuracy": 0.6064742207527161, + "num_tokens": 285875.0, + "step": 11 + }, + { + "epoch": 0.001317812431363936, + "grad_norm": 7.437734603881836, + "learning_rate": 6.0373216245883644e-09, + "loss": 1.5362, + "mean_token_accuracy": 0.6075617671012878, + "num_tokens": 309221.0, + "step": 12 + }, + { + "epoch": 0.0014276301339775973, + "grad_norm": 6.840979099273682, + "learning_rate": 6.586169045005488e-09, + "loss": 1.4987, + "mean_token_accuracy": 0.6130279302597046, + "num_tokens": 335948.0, + "step": 13 + }, + { + "epoch": 0.0015374478365912585, + "grad_norm": 7.3329176902771, + "learning_rate": 7.135016465422612e-09, + "loss": 1.5262, + "mean_token_accuracy": 0.6073921918869019, + "num_tokens": 361550.0, + "step": 14 + }, + { + "epoch": 0.0016472655392049198, + "grad_norm": 6.2873759269714355, + "learning_rate": 7.683863885839736e-09, + "loss": 1.5067, + "mean_token_accuracy": 0.606685996055603, + "num_tokens": 396626.0, + "step": 15 + }, + { + "epoch": 0.0017570832418185812, + "grad_norm": 7.572169780731201, + "learning_rate": 8.23271130625686e-09, + "loss": 1.4102, + "mean_token_accuracy": 0.6224368810653687, + "num_tokens": 420678.0, + "step": 16 + }, + { + "epoch": 0.0018669009444322424, + "grad_norm": 6.9201507568359375, + "learning_rate": 8.781558726673984e-09, + "loss": 1.4385, + "mean_token_accuracy": 0.6219155788421631, + "num_tokens": 446162.0, + "step": 17 + }, + { + "epoch": 0.001976718647045904, + "grad_norm": 8.663162231445312, + "learning_rate": 9.330406147091108e-09, + "loss": 1.5178, + "mean_token_accuracy": 0.6100188493728638, + "num_tokens": 466297.0, + "step": 18 + }, + { + "epoch": 0.002086536349659565, + "grad_norm": 7.827342510223389, + "learning_rate": 9.879253567508231e-09, + "loss": 1.4956, + "mean_token_accuracy": 0.6079697012901306, + "num_tokens": 487881.0, + "step": 19 + }, + { + "epoch": 0.0021963540522732263, + "grad_norm": 8.01622486114502, + "learning_rate": 1.0428100987925357e-08, + "loss": 1.5386, + "mean_token_accuracy": 0.60650634765625, + "num_tokens": 510688.0, + "step": 20 + }, + { + "epoch": 0.002306171754886888, + "grad_norm": 7.646853923797607, + "learning_rate": 1.0976948408342482e-08, + "loss": 1.5814, + "mean_token_accuracy": 0.5944889783859253, + "num_tokens": 534397.0, + "step": 21 + }, + { + "epoch": 0.002415989457500549, + "grad_norm": 7.02205753326416, + "learning_rate": 1.1525795828759604e-08, + "loss": 1.4032, + "mean_token_accuracy": 0.6357952952384949, + "num_tokens": 559520.0, + "step": 22 + }, + { + "epoch": 0.0025258071601142102, + "grad_norm": 6.628652572631836, + "learning_rate": 1.2074643249176729e-08, + "loss": 1.5047, + "mean_token_accuracy": 0.612226128578186, + "num_tokens": 588135.0, + "step": 23 + }, + { + "epoch": 0.002635624862727872, + "grad_norm": 7.0980544090271, + "learning_rate": 1.2623490669593852e-08, + "loss": 1.5104, + "mean_token_accuracy": 0.6049034595489502, + "num_tokens": 613878.0, + "step": 24 + }, + { + "epoch": 0.002745442565341533, + "grad_norm": 6.27535343170166, + "learning_rate": 1.3172338090010976e-08, + "loss": 1.3564, + "mean_token_accuracy": 0.6444615721702576, + "num_tokens": 642629.0, + "step": 25 + }, + { + "epoch": 0.0028552602679551946, + "grad_norm": 6.302044868469238, + "learning_rate": 1.3721185510428101e-08, + "loss": 1.5177, + "mean_token_accuracy": 0.6020737886428833, + "num_tokens": 675211.0, + "step": 26 + }, + { + "epoch": 0.0029650779705688557, + "grad_norm": 6.262789249420166, + "learning_rate": 1.4270032930845225e-08, + "loss": 1.5197, + "mean_token_accuracy": 0.6007692217826843, + "num_tokens": 706707.0, + "step": 27 + }, + { + "epoch": 0.003074895673182517, + "grad_norm": 7.039871692657471, + "learning_rate": 1.4818880351262348e-08, + "loss": 1.5015, + "mean_token_accuracy": 0.6118398904800415, + "num_tokens": 732473.0, + "step": 28 + }, + { + "epoch": 0.0031847133757961785, + "grad_norm": 7.706699848175049, + "learning_rate": 1.536772777167947e-08, + "loss": 1.5449, + "mean_token_accuracy": 0.5998638868331909, + "num_tokens": 754461.0, + "step": 29 + }, + { + "epoch": 0.0032945310784098397, + "grad_norm": 7.400781154632568, + "learning_rate": 1.5916575192096597e-08, + "loss": 1.5083, + "mean_token_accuracy": 0.6177449226379395, + "num_tokens": 777989.0, + "step": 30 + }, + { + "epoch": 0.003404348781023501, + "grad_norm": 6.966712951660156, + "learning_rate": 1.646542261251372e-08, + "loss": 1.5111, + "mean_token_accuracy": 0.6113716959953308, + "num_tokens": 806007.0, + "step": 31 + }, + { + "epoch": 0.0035141664836371624, + "grad_norm": 6.881617069244385, + "learning_rate": 1.7014270032930844e-08, + "loss": 1.4982, + "mean_token_accuracy": 0.6082711815834045, + "num_tokens": 835007.0, + "step": 32 + }, + { + "epoch": 0.0036239841862508236, + "grad_norm": 7.350600242614746, + "learning_rate": 1.756311745334797e-08, + "loss": 1.4518, + "mean_token_accuracy": 0.6213093996047974, + "num_tokens": 859964.0, + "step": 33 + }, + { + "epoch": 0.0037338018888644848, + "grad_norm": 6.988194465637207, + "learning_rate": 1.811196487376509e-08, + "loss": 1.4437, + "mean_token_accuracy": 0.6190980672836304, + "num_tokens": 885978.0, + "step": 34 + }, + { + "epoch": 0.0038436195914781464, + "grad_norm": 6.414912700653076, + "learning_rate": 1.8660812294182216e-08, + "loss": 1.4576, + "mean_token_accuracy": 0.615615725517273, + "num_tokens": 916298.0, + "step": 35 + }, + { + "epoch": 0.003953437294091808, + "grad_norm": 6.29809045791626, + "learning_rate": 1.920965971459934e-08, + "loss": 1.4311, + "mean_token_accuracy": 0.6324260234832764, + "num_tokens": 946243.0, + "step": 36 + }, + { + "epoch": 0.004063254996705469, + "grad_norm": 6.854642868041992, + "learning_rate": 1.9758507135016463e-08, + "loss": 1.508, + "mean_token_accuracy": 0.6079867482185364, + "num_tokens": 974317.0, + "step": 37 + }, + { + "epoch": 0.00417307269931913, + "grad_norm": 8.337381362915039, + "learning_rate": 2.030735455543359e-08, + "loss": 1.54, + "mean_token_accuracy": 0.5992512702941895, + "num_tokens": 995544.0, + "step": 38 + }, + { + "epoch": 0.004282890401932792, + "grad_norm": 7.9789814949035645, + "learning_rate": 2.0856201975850713e-08, + "loss": 1.5022, + "mean_token_accuracy": 0.6165128350257874, + "num_tokens": 1017587.0, + "step": 39 + }, + { + "epoch": 0.004392708104546453, + "grad_norm": 7.246042728424072, + "learning_rate": 2.1405049396267835e-08, + "loss": 1.4847, + "mean_token_accuracy": 0.6125861406326294, + "num_tokens": 1043681.0, + "step": 40 + }, + { + "epoch": 0.004502525807160114, + "grad_norm": 6.474043846130371, + "learning_rate": 2.1953896816684964e-08, + "loss": 1.5878, + "mean_token_accuracy": 0.5879368185997009, + "num_tokens": 1075881.0, + "step": 41 + }, + { + "epoch": 0.004612343509773776, + "grad_norm": 8.033190727233887, + "learning_rate": 2.2502744237102085e-08, + "loss": 1.4995, + "mean_token_accuracy": 0.6209167242050171, + "num_tokens": 1097782.0, + "step": 42 + }, + { + "epoch": 0.0047221612123874365, + "grad_norm": 7.154344081878662, + "learning_rate": 2.3051591657519207e-08, + "loss": 1.3966, + "mean_token_accuracy": 0.6405333280563354, + "num_tokens": 1122056.0, + "step": 43 + }, + { + "epoch": 0.004831978915001098, + "grad_norm": 8.031448364257812, + "learning_rate": 2.3600439077936336e-08, + "loss": 1.4994, + "mean_token_accuracy": 0.615465521812439, + "num_tokens": 1143818.0, + "step": 44 + }, + { + "epoch": 0.00494179661761476, + "grad_norm": 8.173637390136719, + "learning_rate": 2.4149286498353458e-08, + "loss": 1.5912, + "mean_token_accuracy": 0.5890518426895142, + "num_tokens": 1165248.0, + "step": 45 + }, + { + "epoch": 0.0050516143202284204, + "grad_norm": 7.009579658508301, + "learning_rate": 2.469813391877058e-08, + "loss": 1.4908, + "mean_token_accuracy": 0.6072459816932678, + "num_tokens": 1192467.0, + "step": 46 + }, + { + "epoch": 0.005161432022842082, + "grad_norm": 7.511280536651611, + "learning_rate": 2.5246981339187705e-08, + "loss": 1.3739, + "mean_token_accuracy": 0.6379960775375366, + "num_tokens": 1214641.0, + "step": 47 + }, + { + "epoch": 0.005271249725455744, + "grad_norm": 7.034018516540527, + "learning_rate": 2.579582875960483e-08, + "loss": 1.5162, + "mean_token_accuracy": 0.6041474938392639, + "num_tokens": 1242451.0, + "step": 48 + }, + { + "epoch": 0.005381067428069404, + "grad_norm": 6.287169933319092, + "learning_rate": 2.6344676180021952e-08, + "loss": 1.4035, + "mean_token_accuracy": 0.6287031173706055, + "num_tokens": 1272647.0, + "step": 49 + }, + { + "epoch": 0.005490885130683066, + "grad_norm": 7.007269859313965, + "learning_rate": 2.6893523600439077e-08, + "loss": 1.4183, + "mean_token_accuracy": 0.6315428614616394, + "num_tokens": 1299790.0, + "step": 50 + }, + { + "epoch": 0.0056007028332967276, + "grad_norm": 8.448746681213379, + "learning_rate": 2.7442371020856202e-08, + "loss": 1.4775, + "mean_token_accuracy": 0.6132835149765015, + "num_tokens": 1319751.0, + "step": 51 + }, + { + "epoch": 0.005710520535910389, + "grad_norm": 6.832853317260742, + "learning_rate": 2.7991218441273324e-08, + "loss": 1.3819, + "mean_token_accuracy": 0.6364495754241943, + "num_tokens": 1345334.0, + "step": 52 + }, + { + "epoch": 0.00582033823852405, + "grad_norm": 7.485489368438721, + "learning_rate": 2.854006586169045e-08, + "loss": 1.4697, + "mean_token_accuracy": 0.6232167482376099, + "num_tokens": 1367919.0, + "step": 53 + }, + { + "epoch": 0.0059301559411377115, + "grad_norm": 7.3876752853393555, + "learning_rate": 2.9088913282107574e-08, + "loss": 1.5109, + "mean_token_accuracy": 0.6084941029548645, + "num_tokens": 1392312.0, + "step": 54 + }, + { + "epoch": 0.006039973643751373, + "grad_norm": 6.4931535720825195, + "learning_rate": 2.9637760702524696e-08, + "loss": 1.4633, + "mean_token_accuracy": 0.6153662204742432, + "num_tokens": 1423663.0, + "step": 55 + }, + { + "epoch": 0.006149791346365034, + "grad_norm": 5.993495941162109, + "learning_rate": 3.018660812294182e-08, + "loss": 1.5734, + "mean_token_accuracy": 0.5959159135818481, + "num_tokens": 1462132.0, + "step": 56 + }, + { + "epoch": 0.006259609048978695, + "grad_norm": 7.85240364074707, + "learning_rate": 3.073545554335894e-08, + "loss": 1.5337, + "mean_token_accuracy": 0.599860429763794, + "num_tokens": 1484022.0, + "step": 57 + }, + { + "epoch": 0.006369426751592357, + "grad_norm": 8.013544082641602, + "learning_rate": 3.128430296377607e-08, + "loss": 1.4884, + "mean_token_accuracy": 0.6135703325271606, + "num_tokens": 1505743.0, + "step": 58 + }, + { + "epoch": 0.006479244454206018, + "grad_norm": 7.529724597930908, + "learning_rate": 3.1833150384193193e-08, + "loss": 1.4892, + "mean_token_accuracy": 0.615710437297821, + "num_tokens": 1530142.0, + "step": 59 + }, + { + "epoch": 0.006589062156819679, + "grad_norm": 6.557346343994141, + "learning_rate": 3.2381997804610315e-08, + "loss": 1.4592, + "mean_token_accuracy": 0.6232278347015381, + "num_tokens": 1559266.0, + "step": 60 + }, + { + "epoch": 0.006698879859433341, + "grad_norm": 6.88176155090332, + "learning_rate": 3.293084522502744e-08, + "loss": 1.4876, + "mean_token_accuracy": 0.6065075397491455, + "num_tokens": 1586877.0, + "step": 61 + }, + { + "epoch": 0.006808697562047002, + "grad_norm": 6.8972930908203125, + "learning_rate": 3.3479692645444566e-08, + "loss": 1.5109, + "mean_token_accuracy": 0.6177541017532349, + "num_tokens": 1614619.0, + "step": 62 + }, + { + "epoch": 0.006918515264660663, + "grad_norm": 8.687409400939941, + "learning_rate": 3.402854006586169e-08, + "loss": 1.4632, + "mean_token_accuracy": 0.6236600875854492, + "num_tokens": 1634147.0, + "step": 63 + }, + { + "epoch": 0.007028332967274325, + "grad_norm": 7.874764919281006, + "learning_rate": 3.4577387486278816e-08, + "loss": 1.5113, + "mean_token_accuracy": 0.6106500029563904, + "num_tokens": 1655384.0, + "step": 64 + }, + { + "epoch": 0.007138150669887986, + "grad_norm": 7.524722576141357, + "learning_rate": 3.512623490669594e-08, + "loss": 1.5886, + "mean_token_accuracy": 0.6023167967796326, + "num_tokens": 1679510.0, + "step": 65 + }, + { + "epoch": 0.007247968372501647, + "grad_norm": 7.173765182495117, + "learning_rate": 3.567508232711306e-08, + "loss": 1.4899, + "mean_token_accuracy": 0.603267252445221, + "num_tokens": 1703820.0, + "step": 66 + }, + { + "epoch": 0.007357786075115309, + "grad_norm": 6.739823341369629, + "learning_rate": 3.622392974753018e-08, + "loss": 1.4927, + "mean_token_accuracy": 0.6111637353897095, + "num_tokens": 1730049.0, + "step": 67 + }, + { + "epoch": 0.0074676037777289695, + "grad_norm": 7.718687534332275, + "learning_rate": 3.677277716794731e-08, + "loss": 1.5557, + "mean_token_accuracy": 0.6018637418746948, + "num_tokens": 1753350.0, + "step": 68 + }, + { + "epoch": 0.007577421480342631, + "grad_norm": 6.83480978012085, + "learning_rate": 3.732162458836443e-08, + "loss": 1.5119, + "mean_token_accuracy": 0.6025660037994385, + "num_tokens": 1780665.0, + "step": 69 + }, + { + "epoch": 0.007687239182956293, + "grad_norm": 6.154115200042725, + "learning_rate": 3.787047200878156e-08, + "loss": 1.3883, + "mean_token_accuracy": 0.6366411447525024, + "num_tokens": 1812205.0, + "step": 70 + }, + { + "epoch": 0.007797056885569954, + "grad_norm": 7.34127950668335, + "learning_rate": 3.841931942919868e-08, + "loss": 1.5728, + "mean_token_accuracy": 0.6030206084251404, + "num_tokens": 1835597.0, + "step": 71 + }, + { + "epoch": 0.007906874588183616, + "grad_norm": 7.240100383758545, + "learning_rate": 3.8968166849615804e-08, + "loss": 1.5972, + "mean_token_accuracy": 0.5893694162368774, + "num_tokens": 1861628.0, + "step": 72 + }, + { + "epoch": 0.008016692290797276, + "grad_norm": 7.221670150756836, + "learning_rate": 3.9517014270032926e-08, + "loss": 1.5591, + "mean_token_accuracy": 0.5948716998100281, + "num_tokens": 1885935.0, + "step": 73 + }, + { + "epoch": 0.008126509993410937, + "grad_norm": 7.25986909866333, + "learning_rate": 4.006586169045005e-08, + "loss": 1.436, + "mean_token_accuracy": 0.6173200011253357, + "num_tokens": 1909532.0, + "step": 74 + }, + { + "epoch": 0.008236327696024599, + "grad_norm": 7.47528076171875, + "learning_rate": 4.061470911086718e-08, + "loss": 1.5372, + "mean_token_accuracy": 0.6000537872314453, + "num_tokens": 1932426.0, + "step": 75 + }, + { + "epoch": 0.00834614539863826, + "grad_norm": 6.747237205505371, + "learning_rate": 4.1163556531284305e-08, + "loss": 1.4625, + "mean_token_accuracy": 0.6139432191848755, + "num_tokens": 1960194.0, + "step": 76 + }, + { + "epoch": 0.008455963101251922, + "grad_norm": 6.290520668029785, + "learning_rate": 4.1712403951701427e-08, + "loss": 1.4624, + "mean_token_accuracy": 0.6261168122291565, + "num_tokens": 1988784.0, + "step": 77 + }, + { + "epoch": 0.008565780803865584, + "grad_norm": 6.9123358726501465, + "learning_rate": 4.226125137211855e-08, + "loss": 1.4862, + "mean_token_accuracy": 0.6160770654678345, + "num_tokens": 2012797.0, + "step": 78 + }, + { + "epoch": 0.008675598506479244, + "grad_norm": 6.3855180740356445, + "learning_rate": 4.281009879253567e-08, + "loss": 1.5006, + "mean_token_accuracy": 0.6032764315605164, + "num_tokens": 2042872.0, + "step": 79 + }, + { + "epoch": 0.008785416209092905, + "grad_norm": 7.055875778198242, + "learning_rate": 4.335894621295279e-08, + "loss": 1.4295, + "mean_token_accuracy": 0.6266265511512756, + "num_tokens": 2066607.0, + "step": 80 + }, + { + "epoch": 0.008895233911706567, + "grad_norm": 6.2799506187438965, + "learning_rate": 4.390779363336993e-08, + "loss": 1.5167, + "mean_token_accuracy": 0.606131911277771, + "num_tokens": 2097763.0, + "step": 81 + }, + { + "epoch": 0.009005051614320228, + "grad_norm": 7.061357021331787, + "learning_rate": 4.445664105378705e-08, + "loss": 1.5559, + "mean_token_accuracy": 0.5898721218109131, + "num_tokens": 2123741.0, + "step": 82 + }, + { + "epoch": 0.00911486931693389, + "grad_norm": 7.899779319763184, + "learning_rate": 4.500548847420417e-08, + "loss": 1.4537, + "mean_token_accuracy": 0.6227213144302368, + "num_tokens": 2144312.0, + "step": 83 + }, + { + "epoch": 0.009224687019547552, + "grad_norm": 6.250124931335449, + "learning_rate": 4.555433589462129e-08, + "loss": 1.4422, + "mean_token_accuracy": 0.6197143197059631, + "num_tokens": 2170865.0, + "step": 84 + }, + { + "epoch": 0.009334504722161213, + "grad_norm": 7.271759986877441, + "learning_rate": 4.6103183315038415e-08, + "loss": 1.5851, + "mean_token_accuracy": 0.5867415070533752, + "num_tokens": 2196394.0, + "step": 85 + }, + { + "epoch": 0.009444322424774873, + "grad_norm": 7.2305588722229, + "learning_rate": 4.6652030735455537e-08, + "loss": 1.4671, + "mean_token_accuracy": 0.6189392805099487, + "num_tokens": 2219893.0, + "step": 86 + }, + { + "epoch": 0.009554140127388535, + "grad_norm": 6.566540718078613, + "learning_rate": 4.720087815587267e-08, + "loss": 1.4779, + "mean_token_accuracy": 0.6097551584243774, + "num_tokens": 2247625.0, + "step": 87 + }, + { + "epoch": 0.009663957830002196, + "grad_norm": 6.557201862335205, + "learning_rate": 4.7749725576289793e-08, + "loss": 1.4703, + "mean_token_accuracy": 0.6080353260040283, + "num_tokens": 2274014.0, + "step": 88 + }, + { + "epoch": 0.009773775532615858, + "grad_norm": 7.713953018188477, + "learning_rate": 4.8298572996706915e-08, + "loss": 1.4187, + "mean_token_accuracy": 0.6231116056442261, + "num_tokens": 2294470.0, + "step": 89 + }, + { + "epoch": 0.00988359323522952, + "grad_norm": 5.793415069580078, + "learning_rate": 4.884742041712404e-08, + "loss": 1.3855, + "mean_token_accuracy": 0.6340808272361755, + "num_tokens": 2322524.0, + "step": 90 + }, + { + "epoch": 0.009993410937843181, + "grad_norm": 6.364288806915283, + "learning_rate": 4.939626783754116e-08, + "loss": 1.4656, + "mean_token_accuracy": 0.6210067868232727, + "num_tokens": 2349355.0, + "step": 91 + }, + { + "epoch": 0.010103228640456841, + "grad_norm": 5.821547031402588, + "learning_rate": 4.994511525795828e-08, + "loss": 1.4078, + "mean_token_accuracy": 0.6296224594116211, + "num_tokens": 2378546.0, + "step": 92 + }, + { + "epoch": 0.010213046343070502, + "grad_norm": 5.840226173400879, + "learning_rate": 5.049396267837541e-08, + "loss": 1.4068, + "mean_token_accuracy": 0.6286621689796448, + "num_tokens": 2408884.0, + "step": 93 + }, + { + "epoch": 0.010322864045684164, + "grad_norm": 6.30141544342041, + "learning_rate": 5.104281009879254e-08, + "loss": 1.4607, + "mean_token_accuracy": 0.6101909875869751, + "num_tokens": 2435581.0, + "step": 94 + }, + { + "epoch": 0.010432681748297826, + "grad_norm": 6.46171760559082, + "learning_rate": 5.159165751920966e-08, + "loss": 1.4359, + "mean_token_accuracy": 0.6185950636863708, + "num_tokens": 2463424.0, + "step": 95 + }, + { + "epoch": 0.010542499450911487, + "grad_norm": 6.99583101272583, + "learning_rate": 5.214050493962678e-08, + "loss": 1.4975, + "mean_token_accuracy": 0.6063939332962036, + "num_tokens": 2487917.0, + "step": 96 + }, + { + "epoch": 0.010652317153525149, + "grad_norm": 6.053630352020264, + "learning_rate": 5.2689352360043903e-08, + "loss": 1.4635, + "mean_token_accuracy": 0.609929621219635, + "num_tokens": 2517351.0, + "step": 97 + }, + { + "epoch": 0.010762134856138809, + "grad_norm": 6.960501670837402, + "learning_rate": 5.3238199780461025e-08, + "loss": 1.4899, + "mean_token_accuracy": 0.6156300902366638, + "num_tokens": 2539809.0, + "step": 98 + }, + { + "epoch": 0.01087195255875247, + "grad_norm": 6.951742649078369, + "learning_rate": 5.3787047200878154e-08, + "loss": 1.5017, + "mean_token_accuracy": 0.6072380542755127, + "num_tokens": 2563630.0, + "step": 99 + }, + { + "epoch": 0.010981770261366132, + "grad_norm": 6.605172634124756, + "learning_rate": 5.433589462129528e-08, + "loss": 1.5001, + "mean_token_accuracy": 0.6037027835845947, + "num_tokens": 2589162.0, + "step": 100 + }, + { + "epoch": 0.011091587963979794, + "grad_norm": 6.642510890960693, + "learning_rate": 5.4884742041712404e-08, + "loss": 1.4807, + "mean_token_accuracy": 0.6150635480880737, + "num_tokens": 2614940.0, + "step": 101 + }, + { + "epoch": 0.011201405666593455, + "grad_norm": 6.639711380004883, + "learning_rate": 5.5433589462129526e-08, + "loss": 1.4754, + "mean_token_accuracy": 0.6146297454833984, + "num_tokens": 2640118.0, + "step": 102 + }, + { + "epoch": 0.011311223369207117, + "grad_norm": 7.084728240966797, + "learning_rate": 5.598243688254665e-08, + "loss": 1.5676, + "mean_token_accuracy": 0.5996711254119873, + "num_tokens": 2663913.0, + "step": 103 + }, + { + "epoch": 0.011421041071820778, + "grad_norm": 8.062544822692871, + "learning_rate": 5.653128430296377e-08, + "loss": 1.507, + "mean_token_accuracy": 0.6072775721549988, + "num_tokens": 2683064.0, + "step": 104 + }, + { + "epoch": 0.011530858774434438, + "grad_norm": 7.001090049743652, + "learning_rate": 5.70801317233809e-08, + "loss": 1.5254, + "mean_token_accuracy": 0.6043527126312256, + "num_tokens": 2707472.0, + "step": 105 + }, + { + "epoch": 0.0116406764770481, + "grad_norm": 6.631906986236572, + "learning_rate": 5.762897914379802e-08, + "loss": 1.3858, + "mean_token_accuracy": 0.6286760568618774, + "num_tokens": 2731830.0, + "step": 106 + }, + { + "epoch": 0.011750494179661761, + "grad_norm": 7.643786907196045, + "learning_rate": 5.817782656421515e-08, + "loss": 1.3797, + "mean_token_accuracy": 0.6353979706764221, + "num_tokens": 2751565.0, + "step": 107 + }, + { + "epoch": 0.011860311882275423, + "grad_norm": 8.140999794006348, + "learning_rate": 5.872667398463227e-08, + "loss": 1.4228, + "mean_token_accuracy": 0.6301008462905884, + "num_tokens": 2770328.0, + "step": 108 + }, + { + "epoch": 0.011970129584889085, + "grad_norm": 6.889410972595215, + "learning_rate": 5.927552140504939e-08, + "loss": 1.43, + "mean_token_accuracy": 0.6216776967048645, + "num_tokens": 2795296.0, + "step": 109 + }, + { + "epoch": 0.012079947287502746, + "grad_norm": 7.539971828460693, + "learning_rate": 5.982436882546651e-08, + "loss": 1.4834, + "mean_token_accuracy": 0.6149452924728394, + "num_tokens": 2817401.0, + "step": 110 + }, + { + "epoch": 0.012189764990116406, + "grad_norm": 6.514304161071777, + "learning_rate": 6.037321624588364e-08, + "loss": 1.5074, + "mean_token_accuracy": 0.6081002950668335, + "num_tokens": 2846286.0, + "step": 111 + }, + { + "epoch": 0.012299582692730068, + "grad_norm": 6.4149556159973145, + "learning_rate": 6.092206366630077e-08, + "loss": 1.4381, + "mean_token_accuracy": 0.6147114038467407, + "num_tokens": 2873699.0, + "step": 112 + }, + { + "epoch": 0.01240940039534373, + "grad_norm": 7.050398826599121, + "learning_rate": 6.147091108671789e-08, + "loss": 1.4129, + "mean_token_accuracy": 0.622458815574646, + "num_tokens": 2897006.0, + "step": 113 + }, + { + "epoch": 0.01251921809795739, + "grad_norm": 6.983842372894287, + "learning_rate": 6.201975850713501e-08, + "loss": 1.3675, + "mean_token_accuracy": 0.6381494402885437, + "num_tokens": 2920671.0, + "step": 114 + }, + { + "epoch": 0.012629035800571052, + "grad_norm": 6.380026817321777, + "learning_rate": 6.256860592755214e-08, + "loss": 1.4316, + "mean_token_accuracy": 0.6209526062011719, + "num_tokens": 2947946.0, + "step": 115 + }, + { + "epoch": 0.012738853503184714, + "grad_norm": 6.097968578338623, + "learning_rate": 6.311745334796927e-08, + "loss": 1.3769, + "mean_token_accuracy": 0.6295285224914551, + "num_tokens": 2976752.0, + "step": 116 + }, + { + "epoch": 0.012848671205798374, + "grad_norm": 7.460351467132568, + "learning_rate": 6.366630076838639e-08, + "loss": 1.4269, + "mean_token_accuracy": 0.6168888807296753, + "num_tokens": 2998454.0, + "step": 117 + }, + { + "epoch": 0.012958488908412035, + "grad_norm": 6.941334247589111, + "learning_rate": 6.421514818880352e-08, + "loss": 1.4587, + "mean_token_accuracy": 0.6086764335632324, + "num_tokens": 3023397.0, + "step": 118 + }, + { + "epoch": 0.013068306611025697, + "grad_norm": 6.959654331207275, + "learning_rate": 6.476399560922063e-08, + "loss": 1.3937, + "mean_token_accuracy": 0.6346122622489929, + "num_tokens": 3046641.0, + "step": 119 + }, + { + "epoch": 0.013178124313639359, + "grad_norm": 6.226715564727783, + "learning_rate": 6.531284302963776e-08, + "loss": 1.4526, + "mean_token_accuracy": 0.6123732328414917, + "num_tokens": 3075219.0, + "step": 120 + }, + { + "epoch": 0.01328794201625302, + "grad_norm": 5.504324436187744, + "learning_rate": 6.586169045005487e-08, + "loss": 1.2637, + "mean_token_accuracy": 0.6547132134437561, + "num_tokens": 3102587.0, + "step": 121 + }, + { + "epoch": 0.013397759718866682, + "grad_norm": 5.56867790222168, + "learning_rate": 6.6410537870472e-08, + "loss": 1.4866, + "mean_token_accuracy": 0.6048975586891174, + "num_tokens": 3132681.0, + "step": 122 + }, + { + "epoch": 0.013507577421480343, + "grad_norm": 5.89962911605835, + "learning_rate": 6.695938529088913e-08, + "loss": 1.3915, + "mean_token_accuracy": 0.628582239151001, + "num_tokens": 3157628.0, + "step": 123 + }, + { + "epoch": 0.013617395124094003, + "grad_norm": 6.0876874923706055, + "learning_rate": 6.750823271130625e-08, + "loss": 1.3763, + "mean_token_accuracy": 0.6292219161987305, + "num_tokens": 3181345.0, + "step": 124 + }, + { + "epoch": 0.013727212826707665, + "grad_norm": 5.5442962646484375, + "learning_rate": 6.805708013172338e-08, + "loss": 1.4072, + "mean_token_accuracy": 0.6166892051696777, + "num_tokens": 3209792.0, + "step": 125 + }, + { + "epoch": 0.013837030529321327, + "grad_norm": 5.245721340179443, + "learning_rate": 6.86059275521405e-08, + "loss": 1.3579, + "mean_token_accuracy": 0.6302279233932495, + "num_tokens": 3236860.0, + "step": 126 + }, + { + "epoch": 0.013946848231934988, + "grad_norm": 5.697200775146484, + "learning_rate": 6.915477497255763e-08, + "loss": 1.3347, + "mean_token_accuracy": 0.6397886872291565, + "num_tokens": 3260716.0, + "step": 127 + }, + { + "epoch": 0.01405666593454865, + "grad_norm": 5.57171106338501, + "learning_rate": 6.970362239297475e-08, + "loss": 1.4131, + "mean_token_accuracy": 0.6197265386581421, + "num_tokens": 3285690.0, + "step": 128 + }, + { + "epoch": 0.014166483637162311, + "grad_norm": 5.571705341339111, + "learning_rate": 7.025246981339188e-08, + "loss": 1.4691, + "mean_token_accuracy": 0.6096889972686768, + "num_tokens": 3309762.0, + "step": 129 + }, + { + "epoch": 0.014276301339775971, + "grad_norm": 5.314248085021973, + "learning_rate": 7.0801317233809e-08, + "loss": 1.3388, + "mean_token_accuracy": 0.6466895341873169, + "num_tokens": 3336166.0, + "step": 130 + }, + { + "epoch": 0.014386119042389633, + "grad_norm": 4.774181365966797, + "learning_rate": 7.135016465422612e-08, + "loss": 1.3958, + "mean_token_accuracy": 0.6289300918579102, + "num_tokens": 3367494.0, + "step": 131 + }, + { + "epoch": 0.014495936745003294, + "grad_norm": 5.271458148956299, + "learning_rate": 7.189901207464325e-08, + "loss": 1.4646, + "mean_token_accuracy": 0.6101871728897095, + "num_tokens": 3394082.0, + "step": 132 + }, + { + "epoch": 0.014605754447616956, + "grad_norm": 5.238896369934082, + "learning_rate": 7.244785949506036e-08, + "loss": 1.4381, + "mean_token_accuracy": 0.6020262241363525, + "num_tokens": 3418146.0, + "step": 133 + }, + { + "epoch": 0.014715572150230618, + "grad_norm": 4.526235580444336, + "learning_rate": 7.299670691547749e-08, + "loss": 1.4281, + "mean_token_accuracy": 0.6164277791976929, + "num_tokens": 3451874.0, + "step": 134 + }, + { + "epoch": 0.01482538985284428, + "grad_norm": 5.0412397384643555, + "learning_rate": 7.354555433589462e-08, + "loss": 1.3724, + "mean_token_accuracy": 0.6248719692230225, + "num_tokens": 3478513.0, + "step": 135 + }, + { + "epoch": 0.014935207555457939, + "grad_norm": 5.651845455169678, + "learning_rate": 7.409440175631174e-08, + "loss": 1.4254, + "mean_token_accuracy": 0.6112172603607178, + "num_tokens": 3501656.0, + "step": 136 + }, + { + "epoch": 0.0150450252580716, + "grad_norm": 5.526853084564209, + "learning_rate": 7.464324917672886e-08, + "loss": 1.4546, + "mean_token_accuracy": 0.6097128391265869, + "num_tokens": 3525786.0, + "step": 137 + }, + { + "epoch": 0.015154842960685262, + "grad_norm": 5.03640604019165, + "learning_rate": 7.519209659714599e-08, + "loss": 1.4159, + "mean_token_accuracy": 0.6097731590270996, + "num_tokens": 3553279.0, + "step": 138 + }, + { + "epoch": 0.015264660663298924, + "grad_norm": 4.616069316864014, + "learning_rate": 7.574094401756312e-08, + "loss": 1.4191, + "mean_token_accuracy": 0.6217312216758728, + "num_tokens": 3583627.0, + "step": 139 + }, + { + "epoch": 0.015374478365912585, + "grad_norm": 5.086184501647949, + "learning_rate": 7.628979143798024e-08, + "loss": 1.3571, + "mean_token_accuracy": 0.6330345273017883, + "num_tokens": 3609401.0, + "step": 140 + }, + { + "epoch": 0.015484296068526247, + "grad_norm": 5.337474822998047, + "learning_rate": 7.683863885839736e-08, + "loss": 1.3169, + "mean_token_accuracy": 0.6371541023254395, + "num_tokens": 3631857.0, + "step": 141 + }, + { + "epoch": 0.015594113771139909, + "grad_norm": 5.566058158874512, + "learning_rate": 7.738748627881449e-08, + "loss": 1.3277, + "mean_token_accuracy": 0.6318587064743042, + "num_tokens": 3654775.0, + "step": 142 + }, + { + "epoch": 0.01570393147375357, + "grad_norm": 5.376043796539307, + "learning_rate": 7.793633369923161e-08, + "loss": 1.3596, + "mean_token_accuracy": 0.6418933272361755, + "num_tokens": 3678285.0, + "step": 143 + }, + { + "epoch": 0.015813749176367232, + "grad_norm": 5.412756443023682, + "learning_rate": 7.848518111964874e-08, + "loss": 1.3012, + "mean_token_accuracy": 0.6349584460258484, + "num_tokens": 3701792.0, + "step": 144 + }, + { + "epoch": 0.01592356687898089, + "grad_norm": 5.191858768463135, + "learning_rate": 7.903402854006585e-08, + "loss": 1.4025, + "mean_token_accuracy": 0.6206223964691162, + "num_tokens": 3725458.0, + "step": 145 + }, + { + "epoch": 0.01603338458159455, + "grad_norm": 5.252474308013916, + "learning_rate": 7.958287596048298e-08, + "loss": 1.4506, + "mean_token_accuracy": 0.6051657199859619, + "num_tokens": 3747023.0, + "step": 146 + }, + { + "epoch": 0.016143202284208215, + "grad_norm": 5.023656368255615, + "learning_rate": 8.01317233809001e-08, + "loss": 1.3504, + "mean_token_accuracy": 0.6354340314865112, + "num_tokens": 3770950.0, + "step": 147 + }, + { + "epoch": 0.016253019986821875, + "grad_norm": 4.282314777374268, + "learning_rate": 8.068057080131722e-08, + "loss": 1.3588, + "mean_token_accuracy": 0.6261795163154602, + "num_tokens": 3802365.0, + "step": 148 + }, + { + "epoch": 0.016362837689435538, + "grad_norm": 5.405516624450684, + "learning_rate": 8.122941822173437e-08, + "loss": 1.3415, + "mean_token_accuracy": 0.6388635635375977, + "num_tokens": 3824576.0, + "step": 149 + }, + { + "epoch": 0.016472655392049198, + "grad_norm": 4.606717586517334, + "learning_rate": 8.177826564215148e-08, + "loss": 1.3651, + "mean_token_accuracy": 0.633628785610199, + "num_tokens": 3851964.0, + "step": 150 + }, + { + "epoch": 0.01658247309466286, + "grad_norm": 4.830141067504883, + "learning_rate": 8.232711306256861e-08, + "loss": 1.4302, + "mean_token_accuracy": 0.6067633032798767, + "num_tokens": 3880704.0, + "step": 151 + }, + { + "epoch": 0.01669229079727652, + "grad_norm": 4.719853401184082, + "learning_rate": 8.287596048298572e-08, + "loss": 1.2876, + "mean_token_accuracy": 0.648647665977478, + "num_tokens": 3905042.0, + "step": 152 + }, + { + "epoch": 0.01680210849989018, + "grad_norm": 4.814482688903809, + "learning_rate": 8.342480790340285e-08, + "loss": 1.3431, + "mean_token_accuracy": 0.6392379999160767, + "num_tokens": 3929865.0, + "step": 153 + }, + { + "epoch": 0.016911926202503844, + "grad_norm": 4.560098171234131, + "learning_rate": 8.397365532381998e-08, + "loss": 1.3542, + "mean_token_accuracy": 0.6318832039833069, + "num_tokens": 3957191.0, + "step": 154 + }, + { + "epoch": 0.017021743905117504, + "grad_norm": 4.056280136108398, + "learning_rate": 8.45225027442371e-08, + "loss": 1.4002, + "mean_token_accuracy": 0.6202133893966675, + "num_tokens": 3991325.0, + "step": 155 + }, + { + "epoch": 0.017131561607731167, + "grad_norm": 4.448657989501953, + "learning_rate": 8.507135016465423e-08, + "loss": 1.3001, + "mean_token_accuracy": 0.6413215398788452, + "num_tokens": 4016412.0, + "step": 156 + }, + { + "epoch": 0.017241379310344827, + "grad_norm": 4.5627360343933105, + "learning_rate": 8.562019758507134e-08, + "loss": 1.39, + "mean_token_accuracy": 0.6252249479293823, + "num_tokens": 4044211.0, + "step": 157 + }, + { + "epoch": 0.017351197012958487, + "grad_norm": 4.416175842285156, + "learning_rate": 8.616904500548847e-08, + "loss": 1.2905, + "mean_token_accuracy": 0.6425236463546753, + "num_tokens": 4069917.0, + "step": 158 + }, + { + "epoch": 0.01746101471557215, + "grad_norm": 4.738631248474121, + "learning_rate": 8.671789242590558e-08, + "loss": 1.3803, + "mean_token_accuracy": 0.6229529976844788, + "num_tokens": 4094897.0, + "step": 159 + }, + { + "epoch": 0.01757083241818581, + "grad_norm": 5.151134967803955, + "learning_rate": 8.726673984632271e-08, + "loss": 1.405, + "mean_token_accuracy": 0.6202230453491211, + "num_tokens": 4118104.0, + "step": 160 + }, + { + "epoch": 0.017680650120799474, + "grad_norm": 4.216309070587158, + "learning_rate": 8.781558726673985e-08, + "loss": 1.2921, + "mean_token_accuracy": 0.6489490270614624, + "num_tokens": 4146739.0, + "step": 161 + }, + { + "epoch": 0.017790467823413134, + "grad_norm": 4.387755870819092, + "learning_rate": 8.836443468715697e-08, + "loss": 1.3719, + "mean_token_accuracy": 0.6281238198280334, + "num_tokens": 4173239.0, + "step": 162 + }, + { + "epoch": 0.017900285526026797, + "grad_norm": 4.808595657348633, + "learning_rate": 8.89132821075741e-08, + "loss": 1.3746, + "mean_token_accuracy": 0.621812641620636, + "num_tokens": 4197772.0, + "step": 163 + }, + { + "epoch": 0.018010103228640457, + "grad_norm": 4.48674201965332, + "learning_rate": 8.946212952799121e-08, + "loss": 1.3165, + "mean_token_accuracy": 0.6473683714866638, + "num_tokens": 4225509.0, + "step": 164 + }, + { + "epoch": 0.018119920931254117, + "grad_norm": 4.171778202056885, + "learning_rate": 9.001097694840834e-08, + "loss": 1.3216, + "mean_token_accuracy": 0.6400130391120911, + "num_tokens": 4255099.0, + "step": 165 + }, + { + "epoch": 0.01822973863386778, + "grad_norm": 4.148672103881836, + "learning_rate": 9.055982436882546e-08, + "loss": 1.4122, + "mean_token_accuracy": 0.6165125966072083, + "num_tokens": 4284873.0, + "step": 166 + }, + { + "epoch": 0.01833955633648144, + "grad_norm": 4.571953773498535, + "learning_rate": 9.110867178924259e-08, + "loss": 1.4259, + "mean_token_accuracy": 0.6123077273368835, + "num_tokens": 4312823.0, + "step": 167 + }, + { + "epoch": 0.018449374039095103, + "grad_norm": 5.280969619750977, + "learning_rate": 9.165751920965971e-08, + "loss": 1.2808, + "mean_token_accuracy": 0.6391872763633728, + "num_tokens": 4332621.0, + "step": 168 + }, + { + "epoch": 0.018559191741708763, + "grad_norm": 4.627303600311279, + "learning_rate": 9.220636663007683e-08, + "loss": 1.3696, + "mean_token_accuracy": 0.6212846040725708, + "num_tokens": 4354905.0, + "step": 169 + }, + { + "epoch": 0.018669009444322426, + "grad_norm": 5.780825614929199, + "learning_rate": 9.275521405049396e-08, + "loss": 1.2658, + "mean_token_accuracy": 0.6508558988571167, + "num_tokens": 4373081.0, + "step": 170 + }, + { + "epoch": 0.018778827146936086, + "grad_norm": 5.447687149047852, + "learning_rate": 9.330406147091107e-08, + "loss": 1.334, + "mean_token_accuracy": 0.6350640654563904, + "num_tokens": 4391787.0, + "step": 171 + }, + { + "epoch": 0.018888644849549746, + "grad_norm": 4.42579460144043, + "learning_rate": 9.38529088913282e-08, + "loss": 1.3717, + "mean_token_accuracy": 0.6189147233963013, + "num_tokens": 4420844.0, + "step": 172 + }, + { + "epoch": 0.01899846255216341, + "grad_norm": 4.1785197257995605, + "learning_rate": 9.440175631174534e-08, + "loss": 1.3863, + "mean_token_accuracy": 0.6184136867523193, + "num_tokens": 4447614.0, + "step": 173 + }, + { + "epoch": 0.01910828025477707, + "grad_norm": 3.9345099925994873, + "learning_rate": 9.495060373216246e-08, + "loss": 1.2684, + "mean_token_accuracy": 0.6481015086174011, + "num_tokens": 4477292.0, + "step": 174 + }, + { + "epoch": 0.019218097957390733, + "grad_norm": 4.485980987548828, + "learning_rate": 9.549945115257959e-08, + "loss": 1.301, + "mean_token_accuracy": 0.6455981731414795, + "num_tokens": 4498656.0, + "step": 175 + }, + { + "epoch": 0.019327915660004392, + "grad_norm": 3.8833680152893066, + "learning_rate": 9.60482985729967e-08, + "loss": 1.3353, + "mean_token_accuracy": 0.6374614834785461, + "num_tokens": 4527867.0, + "step": 176 + }, + { + "epoch": 0.019437733362618052, + "grad_norm": 4.515674591064453, + "learning_rate": 9.659714599341383e-08, + "loss": 1.3301, + "mean_token_accuracy": 0.6361721158027649, + "num_tokens": 4551092.0, + "step": 177 + }, + { + "epoch": 0.019547551065231716, + "grad_norm": 3.86015248298645, + "learning_rate": 9.714599341383095e-08, + "loss": 1.3547, + "mean_token_accuracy": 0.628350019454956, + "num_tokens": 4579861.0, + "step": 178 + }, + { + "epoch": 0.019657368767845376, + "grad_norm": 3.839942693710327, + "learning_rate": 9.769484083424807e-08, + "loss": 1.2955, + "mean_token_accuracy": 0.6455345153808594, + "num_tokens": 4608629.0, + "step": 179 + }, + { + "epoch": 0.01976718647045904, + "grad_norm": 4.4471306800842285, + "learning_rate": 9.82436882546652e-08, + "loss": 1.3585, + "mean_token_accuracy": 0.6254092454910278, + "num_tokens": 4631075.0, + "step": 180 + }, + { + "epoch": 0.0198770041730727, + "grad_norm": 3.7461843490600586, + "learning_rate": 9.879253567508232e-08, + "loss": 1.2996, + "mean_token_accuracy": 0.6461951732635498, + "num_tokens": 4662687.0, + "step": 181 + }, + { + "epoch": 0.019986821875686362, + "grad_norm": 4.027956008911133, + "learning_rate": 9.934138309549945e-08, + "loss": 1.2857, + "mean_token_accuracy": 0.6436887979507446, + "num_tokens": 4693627.0, + "step": 182 + }, + { + "epoch": 0.020096639578300022, + "grad_norm": 4.1353535652160645, + "learning_rate": 9.989023051591656e-08, + "loss": 1.3166, + "mean_token_accuracy": 0.6336426138877869, + "num_tokens": 4723206.0, + "step": 183 + }, + { + "epoch": 0.020206457280913682, + "grad_norm": 3.856196165084839, + "learning_rate": 1.0043907793633369e-07, + "loss": 1.3244, + "mean_token_accuracy": 0.6357016563415527, + "num_tokens": 4754003.0, + "step": 184 + }, + { + "epoch": 0.020316274983527345, + "grad_norm": 4.137902736663818, + "learning_rate": 1.0098792535675082e-07, + "loss": 1.2898, + "mean_token_accuracy": 0.6448460221290588, + "num_tokens": 4778284.0, + "step": 185 + }, + { + "epoch": 0.020426092686141005, + "grad_norm": 6.167253494262695, + "learning_rate": 1.0153677277716795e-07, + "loss": 1.2946, + "mean_token_accuracy": 0.6457217335700989, + "num_tokens": 4795434.0, + "step": 186 + }, + { + "epoch": 0.02053591038875467, + "grad_norm": 4.094048500061035, + "learning_rate": 1.0208562019758508e-07, + "loss": 1.296, + "mean_token_accuracy": 0.6423852443695068, + "num_tokens": 4819690.0, + "step": 187 + }, + { + "epoch": 0.020645728091368328, + "grad_norm": 4.48492956161499, + "learning_rate": 1.0263446761800219e-07, + "loss": 1.3134, + "mean_token_accuracy": 0.6389660835266113, + "num_tokens": 4844142.0, + "step": 188 + }, + { + "epoch": 0.02075554579398199, + "grad_norm": 3.9169583320617676, + "learning_rate": 1.0318331503841932e-07, + "loss": 1.3067, + "mean_token_accuracy": 0.6410961747169495, + "num_tokens": 4870339.0, + "step": 189 + }, + { + "epoch": 0.02086536349659565, + "grad_norm": 4.9895524978637695, + "learning_rate": 1.0373216245883643e-07, + "loss": 1.4593, + "mean_token_accuracy": 0.6106641292572021, + "num_tokens": 4893907.0, + "step": 190 + }, + { + "epoch": 0.02097518119920931, + "grad_norm": 3.614288568496704, + "learning_rate": 1.0428100987925356e-07, + "loss": 1.3109, + "mean_token_accuracy": 0.6369436383247375, + "num_tokens": 4924962.0, + "step": 191 + }, + { + "epoch": 0.021084998901822975, + "grad_norm": 4.367719650268555, + "learning_rate": 1.0482985729967068e-07, + "loss": 1.3306, + "mean_token_accuracy": 0.6326977014541626, + "num_tokens": 4953272.0, + "step": 192 + }, + { + "epoch": 0.021194816604436634, + "grad_norm": 3.6910629272460938, + "learning_rate": 1.0537870472008781e-07, + "loss": 1.3454, + "mean_token_accuracy": 0.6307864189147949, + "num_tokens": 4985584.0, + "step": 193 + }, + { + "epoch": 0.021304634307050298, + "grad_norm": 5.091099739074707, + "learning_rate": 1.0592755214050494e-07, + "loss": 1.4154, + "mean_token_accuracy": 0.6154892444610596, + "num_tokens": 5008994.0, + "step": 194 + }, + { + "epoch": 0.021414452009663958, + "grad_norm": 4.805250644683838, + "learning_rate": 1.0647639956092205e-07, + "loss": 1.2818, + "mean_token_accuracy": 0.6445826292037964, + "num_tokens": 5030725.0, + "step": 195 + }, + { + "epoch": 0.021524269712277617, + "grad_norm": 4.740853786468506, + "learning_rate": 1.0702524698133918e-07, + "loss": 1.3089, + "mean_token_accuracy": 0.6440005302429199, + "num_tokens": 5053694.0, + "step": 196 + }, + { + "epoch": 0.02163408741489128, + "grad_norm": 5.141437530517578, + "learning_rate": 1.0757409440175631e-07, + "loss": 1.2861, + "mean_token_accuracy": 0.6438108682632446, + "num_tokens": 5081002.0, + "step": 197 + }, + { + "epoch": 0.02174390511750494, + "grad_norm": 5.516367435455322, + "learning_rate": 1.0812294182217344e-07, + "loss": 1.2912, + "mean_token_accuracy": 0.6439458131790161, + "num_tokens": 5105358.0, + "step": 198 + }, + { + "epoch": 0.021853722820118604, + "grad_norm": 6.233048439025879, + "learning_rate": 1.0867178924259056e-07, + "loss": 1.2201, + "mean_token_accuracy": 0.6536824703216553, + "num_tokens": 5121948.0, + "step": 199 + }, + { + "epoch": 0.021963540522732264, + "grad_norm": 4.833504676818848, + "learning_rate": 1.0922063666300768e-07, + "loss": 1.3385, + "mean_token_accuracy": 0.6269527077674866, + "num_tokens": 5149167.0, + "step": 200 + }, + { + "epoch": 0.022073358225345927, + "grad_norm": 5.204031467437744, + "learning_rate": 1.0976948408342481e-07, + "loss": 1.3285, + "mean_token_accuracy": 0.6262280344963074, + "num_tokens": 5175653.0, + "step": 201 + }, + { + "epoch": 0.022183175927959587, + "grad_norm": 4.815508842468262, + "learning_rate": 1.1031833150384192e-07, + "loss": 1.3727, + "mean_token_accuracy": 0.617740273475647, + "num_tokens": 5205056.0, + "step": 202 + }, + { + "epoch": 0.022292993630573247, + "grad_norm": 5.148724555969238, + "learning_rate": 1.1086717892425905e-07, + "loss": 1.1459, + "mean_token_accuracy": 0.6690980792045593, + "num_tokens": 5229332.0, + "step": 203 + }, + { + "epoch": 0.02240281133318691, + "grad_norm": 5.53524923324585, + "learning_rate": 1.1141602634467617e-07, + "loss": 1.2838, + "mean_token_accuracy": 0.6405318379402161, + "num_tokens": 5249726.0, + "step": 204 + }, + { + "epoch": 0.02251262903580057, + "grad_norm": 4.414306163787842, + "learning_rate": 1.119648737650933e-07, + "loss": 1.3387, + "mean_token_accuracy": 0.6339237689971924, + "num_tokens": 5277167.0, + "step": 205 + }, + { + "epoch": 0.022622446738414233, + "grad_norm": 5.115250587463379, + "learning_rate": 1.1251372118551042e-07, + "loss": 1.3101, + "mean_token_accuracy": 0.6337499022483826, + "num_tokens": 5304099.0, + "step": 206 + }, + { + "epoch": 0.022732264441027893, + "grad_norm": 4.480660915374756, + "learning_rate": 1.1306256860592754e-07, + "loss": 1.3258, + "mean_token_accuracy": 0.6365455985069275, + "num_tokens": 5330055.0, + "step": 207 + }, + { + "epoch": 0.022842082143641557, + "grad_norm": 4.682101249694824, + "learning_rate": 1.1361141602634467e-07, + "loss": 1.3207, + "mean_token_accuracy": 0.6366685032844543, + "num_tokens": 5357440.0, + "step": 208 + }, + { + "epoch": 0.022951899846255217, + "grad_norm": 4.912617206573486, + "learning_rate": 1.141602634467618e-07, + "loss": 1.3141, + "mean_token_accuracy": 0.6378239393234253, + "num_tokens": 5385688.0, + "step": 209 + }, + { + "epoch": 0.023061717548868876, + "grad_norm": 4.4219560623168945, + "learning_rate": 1.1470911086717892e-07, + "loss": 1.3737, + "mean_token_accuracy": 0.6263467073440552, + "num_tokens": 5416641.0, + "step": 210 + }, + { + "epoch": 0.02317153525148254, + "grad_norm": 4.687409400939941, + "learning_rate": 1.1525795828759604e-07, + "loss": 1.2375, + "mean_token_accuracy": 0.6529651284217834, + "num_tokens": 5442851.0, + "step": 211 + }, + { + "epoch": 0.0232813529540962, + "grad_norm": 4.368887901306152, + "learning_rate": 1.1580680570801317e-07, + "loss": 1.3605, + "mean_token_accuracy": 0.6217460632324219, + "num_tokens": 5474823.0, + "step": 212 + }, + { + "epoch": 0.023391170656709863, + "grad_norm": 4.6941022872924805, + "learning_rate": 1.163556531284303e-07, + "loss": 1.3574, + "mean_token_accuracy": 0.6383377313613892, + "num_tokens": 5502810.0, + "step": 213 + }, + { + "epoch": 0.023500988359323523, + "grad_norm": 8.8756103515625, + "learning_rate": 1.1690450054884741e-07, + "loss": 1.2434, + "mean_token_accuracy": 0.6542109251022339, + "num_tokens": 5525014.0, + "step": 214 + }, + { + "epoch": 0.023610806061937183, + "grad_norm": 4.1593170166015625, + "learning_rate": 1.1745334796926454e-07, + "loss": 1.2838, + "mean_token_accuracy": 0.6439505815505981, + "num_tokens": 5555684.0, + "step": 215 + }, + { + "epoch": 0.023720623764550846, + "grad_norm": 6.151599407196045, + "learning_rate": 1.1800219538968166e-07, + "loss": 1.2318, + "mean_token_accuracy": 0.6535921096801758, + "num_tokens": 5575527.0, + "step": 216 + }, + { + "epoch": 0.023830441467164506, + "grad_norm": 5.3587541580200195, + "learning_rate": 1.1855104281009878e-07, + "loss": 1.3499, + "mean_token_accuracy": 0.6320172548294067, + "num_tokens": 5607380.0, + "step": 217 + }, + { + "epoch": 0.02394025916977817, + "grad_norm": 4.625665664672852, + "learning_rate": 1.1909989023051591e-07, + "loss": 1.303, + "mean_token_accuracy": 0.6439661979675293, + "num_tokens": 5636453.0, + "step": 218 + }, + { + "epoch": 0.02405007687239183, + "grad_norm": 4.435354232788086, + "learning_rate": 1.1964873765093303e-07, + "loss": 1.2853, + "mean_token_accuracy": 0.6435254812240601, + "num_tokens": 5664429.0, + "step": 219 + }, + { + "epoch": 0.024159894575005492, + "grad_norm": 5.412102222442627, + "learning_rate": 1.2019758507135017e-07, + "loss": 1.2757, + "mean_token_accuracy": 0.6433618068695068, + "num_tokens": 5687467.0, + "step": 220 + }, + { + "epoch": 0.024269712277619152, + "grad_norm": 4.607097148895264, + "learning_rate": 1.2074643249176729e-07, + "loss": 1.2479, + "mean_token_accuracy": 0.6533572673797607, + "num_tokens": 5718846.0, + "step": 221 + }, + { + "epoch": 0.024379529980232812, + "grad_norm": 5.765384674072266, + "learning_rate": 1.212952799121844e-07, + "loss": 1.2831, + "mean_token_accuracy": 0.6499736309051514, + "num_tokens": 5744239.0, + "step": 222 + }, + { + "epoch": 0.024489347682846475, + "grad_norm": 6.323597431182861, + "learning_rate": 1.2184412733260154e-07, + "loss": 1.2733, + "mean_token_accuracy": 0.6459267139434814, + "num_tokens": 5765765.0, + "step": 223 + }, + { + "epoch": 0.024599165385460135, + "grad_norm": 5.514684677124023, + "learning_rate": 1.2239297475301866e-07, + "loss": 1.159, + "mean_token_accuracy": 0.666490375995636, + "num_tokens": 5787625.0, + "step": 224 + }, + { + "epoch": 0.0247089830880738, + "grad_norm": 5.673861503601074, + "learning_rate": 1.2294182217343577e-07, + "loss": 1.2361, + "mean_token_accuracy": 0.646470844745636, + "num_tokens": 5807998.0, + "step": 225 + }, + { + "epoch": 0.02481880079068746, + "grad_norm": 5.628068923950195, + "learning_rate": 1.2349066959385291e-07, + "loss": 1.2335, + "mean_token_accuracy": 0.6526831388473511, + "num_tokens": 5835823.0, + "step": 226 + }, + { + "epoch": 0.024928618493301122, + "grad_norm": 4.698570728302002, + "learning_rate": 1.2403951701427003e-07, + "loss": 1.3157, + "mean_token_accuracy": 0.6213032007217407, + "num_tokens": 5864353.0, + "step": 227 + }, + { + "epoch": 0.02503843619591478, + "grad_norm": 6.315324306488037, + "learning_rate": 1.2458836443468714e-07, + "loss": 1.2159, + "mean_token_accuracy": 0.6561124324798584, + "num_tokens": 5887612.0, + "step": 228 + }, + { + "epoch": 0.02514825389852844, + "grad_norm": 6.252568244934082, + "learning_rate": 1.2513721185510429e-07, + "loss": 1.1749, + "mean_token_accuracy": 0.6716314554214478, + "num_tokens": 5914113.0, + "step": 229 + }, + { + "epoch": 0.025258071601142105, + "grad_norm": 4.677523612976074, + "learning_rate": 1.256860592755214e-07, + "loss": 1.2174, + "mean_token_accuracy": 0.6622602939605713, + "num_tokens": 5942767.0, + "step": 230 + }, + { + "epoch": 0.025367889303755765, + "grad_norm": 4.397003650665283, + "learning_rate": 1.2623490669593854e-07, + "loss": 1.2578, + "mean_token_accuracy": 0.6505030393600464, + "num_tokens": 5971689.0, + "step": 231 + }, + { + "epoch": 0.025477707006369428, + "grad_norm": 9.53212833404541, + "learning_rate": 1.2678375411635563e-07, + "loss": 1.3384, + "mean_token_accuracy": 0.6421810388565063, + "num_tokens": 5993516.0, + "step": 232 + }, + { + "epoch": 0.025587524708983088, + "grad_norm": 7.144650936126709, + "learning_rate": 1.2733260153677277e-07, + "loss": 1.3099, + "mean_token_accuracy": 0.6296055316925049, + "num_tokens": 6015429.0, + "step": 233 + }, + { + "epoch": 0.025697342411596748, + "grad_norm": 5.976781368255615, + "learning_rate": 1.278814489571899e-07, + "loss": 1.2958, + "mean_token_accuracy": 0.6357986330986023, + "num_tokens": 6038426.0, + "step": 234 + }, + { + "epoch": 0.02580716011421041, + "grad_norm": 5.740952491760254, + "learning_rate": 1.2843029637760703e-07, + "loss": 1.204, + "mean_token_accuracy": 0.6714531183242798, + "num_tokens": 6068994.0, + "step": 235 + }, + { + "epoch": 0.02591697781682407, + "grad_norm": 6.646813869476318, + "learning_rate": 1.2897914379802412e-07, + "loss": 1.239, + "mean_token_accuracy": 0.6508788466453552, + "num_tokens": 6094921.0, + "step": 236 + }, + { + "epoch": 0.026026795519437734, + "grad_norm": 5.197638988494873, + "learning_rate": 1.2952799121844126e-07, + "loss": 1.1481, + "mean_token_accuracy": 0.6766982078552246, + "num_tokens": 6120557.0, + "step": 237 + }, + { + "epoch": 0.026136613222051394, + "grad_norm": 5.990002155303955, + "learning_rate": 1.300768386388584e-07, + "loss": 1.2003, + "mean_token_accuracy": 0.6643104553222656, + "num_tokens": 6146550.0, + "step": 238 + }, + { + "epoch": 0.026246430924665057, + "grad_norm": 9.583277702331543, + "learning_rate": 1.3062568605927552e-07, + "loss": 1.2767, + "mean_token_accuracy": 0.6444008350372314, + "num_tokens": 6173297.0, + "step": 239 + }, + { + "epoch": 0.026356248627278717, + "grad_norm": 10.299530982971191, + "learning_rate": 1.3117453347969266e-07, + "loss": 1.2418, + "mean_token_accuracy": 0.6492289304733276, + "num_tokens": 6196004.0, + "step": 240 + }, + { + "epoch": 0.026466066329892377, + "grad_norm": 5.482737064361572, + "learning_rate": 1.3172338090010975e-07, + "loss": 1.2229, + "mean_token_accuracy": 0.6585680842399597, + "num_tokens": 6217110.0, + "step": 241 + }, + { + "epoch": 0.02657588403250604, + "grad_norm": 5.453842639923096, + "learning_rate": 1.322722283205269e-07, + "loss": 1.3235, + "mean_token_accuracy": 0.6279597282409668, + "num_tokens": 6242863.0, + "step": 242 + }, + { + "epoch": 0.0266857017351197, + "grad_norm": 6.036808490753174, + "learning_rate": 1.32821075740944e-07, + "loss": 1.2387, + "mean_token_accuracy": 0.6569255590438843, + "num_tokens": 6272955.0, + "step": 243 + }, + { + "epoch": 0.026795519437733364, + "grad_norm": 7.548149108886719, + "learning_rate": 1.3336992316136115e-07, + "loss": 1.3834, + "mean_token_accuracy": 0.6140555739402771, + "num_tokens": 6299289.0, + "step": 244 + }, + { + "epoch": 0.026905337140347024, + "grad_norm": 6.487533092498779, + "learning_rate": 1.3391877058177826e-07, + "loss": 1.2183, + "mean_token_accuracy": 0.6758594512939453, + "num_tokens": 6323416.0, + "step": 245 + }, + { + "epoch": 0.027015154842960687, + "grad_norm": 5.976448059082031, + "learning_rate": 1.3446761800219538e-07, + "loss": 1.3487, + "mean_token_accuracy": 0.6296228170394897, + "num_tokens": 6348004.0, + "step": 246 + }, + { + "epoch": 0.027124972545574347, + "grad_norm": 8.272064208984375, + "learning_rate": 1.350164654226125e-07, + "loss": 1.2233, + "mean_token_accuracy": 0.6515973210334778, + "num_tokens": 6375063.0, + "step": 247 + }, + { + "epoch": 0.027234790248188007, + "grad_norm": 10.179046630859375, + "learning_rate": 1.3556531284302963e-07, + "loss": 1.1976, + "mean_token_accuracy": 0.6600548624992371, + "num_tokens": 6395849.0, + "step": 248 + }, + { + "epoch": 0.02734460795080167, + "grad_norm": 8.19560718536377, + "learning_rate": 1.3611416026344675e-07, + "loss": 1.2067, + "mean_token_accuracy": 0.6562840938568115, + "num_tokens": 6419016.0, + "step": 249 + }, + { + "epoch": 0.02745442565341533, + "grad_norm": 5.797307014465332, + "learning_rate": 1.366630076838639e-07, + "loss": 1.3017, + "mean_token_accuracy": 0.63994961977005, + "num_tokens": 6447243.0, + "step": 250 + }, + { + "epoch": 0.027564243356028993, + "grad_norm": 5.168335914611816, + "learning_rate": 1.37211855104281e-07, + "loss": 1.2016, + "mean_token_accuracy": 0.6564508676528931, + "num_tokens": 6470948.0, + "step": 251 + }, + { + "epoch": 0.027674061058642653, + "grad_norm": 5.751834869384766, + "learning_rate": 1.3776070252469812e-07, + "loss": 1.2185, + "mean_token_accuracy": 0.6503434181213379, + "num_tokens": 6494352.0, + "step": 252 + }, + { + "epoch": 0.027783878761256313, + "grad_norm": 5.216706275939941, + "learning_rate": 1.3830954994511526e-07, + "loss": 1.2737, + "mean_token_accuracy": 0.637871265411377, + "num_tokens": 6516039.0, + "step": 253 + }, + { + "epoch": 0.027893696463869976, + "grad_norm": 6.853394508361816, + "learning_rate": 1.3885839736553238e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.6835900545120239, + "num_tokens": 6537398.0, + "step": 254 + }, + { + "epoch": 0.028003514166483636, + "grad_norm": 4.73104190826416, + "learning_rate": 1.394072447859495e-07, + "loss": 1.2213, + "mean_token_accuracy": 0.6528659462928772, + "num_tokens": 6564798.0, + "step": 255 + }, + { + "epoch": 0.0281133318690973, + "grad_norm": 7.0969109535217285, + "learning_rate": 1.399560922063666e-07, + "loss": 1.1953, + "mean_token_accuracy": 0.6631855368614197, + "num_tokens": 6587453.0, + "step": 256 + }, + { + "epoch": 0.02822314957171096, + "grad_norm": 6.063892841339111, + "learning_rate": 1.4050493962678375e-07, + "loss": 1.3007, + "mean_token_accuracy": 0.6360378265380859, + "num_tokens": 6613594.0, + "step": 257 + }, + { + "epoch": 0.028332967274324623, + "grad_norm": 5.360471248626709, + "learning_rate": 1.4105378704720087e-07, + "loss": 1.2246, + "mean_token_accuracy": 0.6614041924476624, + "num_tokens": 6638843.0, + "step": 258 + }, + { + "epoch": 0.028442784976938282, + "grad_norm": 6.7408623695373535, + "learning_rate": 1.41602634467618e-07, + "loss": 1.2594, + "mean_token_accuracy": 0.6636803150177002, + "num_tokens": 6661209.0, + "step": 259 + }, + { + "epoch": 0.028552602679551942, + "grad_norm": 5.787835121154785, + "learning_rate": 1.421514818880351e-07, + "loss": 1.2597, + "mean_token_accuracy": 0.6487953066825867, + "num_tokens": 6683590.0, + "step": 260 + }, + { + "epoch": 0.028662420382165606, + "grad_norm": 4.499388694763184, + "learning_rate": 1.4270032930845224e-07, + "loss": 1.2105, + "mean_token_accuracy": 0.658724308013916, + "num_tokens": 6713938.0, + "step": 261 + }, + { + "epoch": 0.028772238084779266, + "grad_norm": 5.957883358001709, + "learning_rate": 1.4324917672886938e-07, + "loss": 1.2801, + "mean_token_accuracy": 0.6382207870483398, + "num_tokens": 6739066.0, + "step": 262 + }, + { + "epoch": 0.02888205578739293, + "grad_norm": 5.227792739868164, + "learning_rate": 1.437980241492865e-07, + "loss": 1.1893, + "mean_token_accuracy": 0.6625279188156128, + "num_tokens": 6764171.0, + "step": 263 + }, + { + "epoch": 0.02899187349000659, + "grad_norm": 5.34600305557251, + "learning_rate": 1.4434687156970364e-07, + "loss": 1.2708, + "mean_token_accuracy": 0.6413270235061646, + "num_tokens": 6789391.0, + "step": 264 + }, + { + "epoch": 0.029101691192620252, + "grad_norm": 6.226369380950928, + "learning_rate": 1.4489571899012073e-07, + "loss": 1.3123, + "mean_token_accuracy": 0.6370187401771545, + "num_tokens": 6811378.0, + "step": 265 + }, + { + "epoch": 0.029211508895233912, + "grad_norm": 5.553107261657715, + "learning_rate": 1.4544456641053787e-07, + "loss": 1.1859, + "mean_token_accuracy": 0.6673152446746826, + "num_tokens": 6837670.0, + "step": 266 + }, + { + "epoch": 0.029321326597847572, + "grad_norm": 5.0081787109375, + "learning_rate": 1.4599341383095498e-07, + "loss": 1.3266, + "mean_token_accuracy": 0.6383656859397888, + "num_tokens": 6863080.0, + "step": 267 + }, + { + "epoch": 0.029431144300461235, + "grad_norm": 5.66528844833374, + "learning_rate": 1.4654226125137212e-07, + "loss": 1.2481, + "mean_token_accuracy": 0.654227614402771, + "num_tokens": 6890200.0, + "step": 268 + }, + { + "epoch": 0.029540962003074895, + "grad_norm": 5.207780838012695, + "learning_rate": 1.4709110867178924e-07, + "loss": 1.23, + "mean_token_accuracy": 0.6546914577484131, + "num_tokens": 6917328.0, + "step": 269 + }, + { + "epoch": 0.02965077970568856, + "grad_norm": 4.421492099761963, + "learning_rate": 1.4763995609220636e-07, + "loss": 1.2676, + "mean_token_accuracy": 0.6466453671455383, + "num_tokens": 6947546.0, + "step": 270 + }, + { + "epoch": 0.029760597408302218, + "grad_norm": 6.300160884857178, + "learning_rate": 1.4818880351262347e-07, + "loss": 1.2911, + "mean_token_accuracy": 0.641746997833252, + "num_tokens": 6974425.0, + "step": 271 + }, + { + "epoch": 0.029870415110915878, + "grad_norm": 5.715331554412842, + "learning_rate": 1.487376509330406e-07, + "loss": 1.1751, + "mean_token_accuracy": 0.6725955605506897, + "num_tokens": 6998647.0, + "step": 272 + }, + { + "epoch": 0.02998023281352954, + "grad_norm": 5.9907097816467285, + "learning_rate": 1.4928649835345773e-07, + "loss": 1.3143, + "mean_token_accuracy": 0.643984317779541, + "num_tokens": 7025004.0, + "step": 273 + }, + { + "epoch": 0.0300900505161432, + "grad_norm": 5.375271320343018, + "learning_rate": 1.4983534577387484e-07, + "loss": 1.1893, + "mean_token_accuracy": 0.6666975021362305, + "num_tokens": 7051741.0, + "step": 274 + }, + { + "epoch": 0.030199868218756865, + "grad_norm": 5.355583667755127, + "learning_rate": 1.5038419319429198e-07, + "loss": 1.2173, + "mean_token_accuracy": 0.6457235813140869, + "num_tokens": 7077452.0, + "step": 275 + }, + { + "epoch": 0.030309685921370524, + "grad_norm": 6.360105991363525, + "learning_rate": 1.509330406147091e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6792569756507874, + "num_tokens": 7101494.0, + "step": 276 + }, + { + "epoch": 0.030419503623984188, + "grad_norm": 6.738111972808838, + "learning_rate": 1.5148188803512624e-07, + "loss": 1.2669, + "mean_token_accuracy": 0.6448907852172852, + "num_tokens": 7126505.0, + "step": 277 + }, + { + "epoch": 0.030529321326597848, + "grad_norm": 5.20325231552124, + "learning_rate": 1.5203073545554336e-07, + "loss": 1.2515, + "mean_token_accuracy": 0.6502450704574585, + "num_tokens": 7152737.0, + "step": 278 + }, + { + "epoch": 0.030639139029211507, + "grad_norm": 6.110866546630859, + "learning_rate": 1.5257958287596047e-07, + "loss": 1.1977, + "mean_token_accuracy": 0.662178635597229, + "num_tokens": 7176389.0, + "step": 279 + }, + { + "epoch": 0.03074895673182517, + "grad_norm": 5.594151496887207, + "learning_rate": 1.531284302963776e-07, + "loss": 1.1888, + "mean_token_accuracy": 0.6666424870491028, + "num_tokens": 7201593.0, + "step": 280 + }, + { + "epoch": 0.03085877443443883, + "grad_norm": 5.309361934661865, + "learning_rate": 1.5367727771679473e-07, + "loss": 1.3196, + "mean_token_accuracy": 0.6302840709686279, + "num_tokens": 7232508.0, + "step": 281 + }, + { + "epoch": 0.030968592137052494, + "grad_norm": 4.88066291809082, + "learning_rate": 1.5422612513721184e-07, + "loss": 1.2645, + "mean_token_accuracy": 0.6467740535736084, + "num_tokens": 7262141.0, + "step": 282 + }, + { + "epoch": 0.031078409839666154, + "grad_norm": 4.623416900634766, + "learning_rate": 1.5477497255762899e-07, + "loss": 1.2184, + "mean_token_accuracy": 0.654732882976532, + "num_tokens": 7291585.0, + "step": 283 + }, + { + "epoch": 0.031188227542279817, + "grad_norm": 4.821522235870361, + "learning_rate": 1.5532381997804607e-07, + "loss": 1.258, + "mean_token_accuracy": 0.6480942964553833, + "num_tokens": 7318085.0, + "step": 284 + }, + { + "epoch": 0.03129804524489348, + "grad_norm": 6.104188442230225, + "learning_rate": 1.5587266739846322e-07, + "loss": 1.3102, + "mean_token_accuracy": 0.6467074751853943, + "num_tokens": 7341567.0, + "step": 285 + }, + { + "epoch": 0.03140786294750714, + "grad_norm": 6.0200371742248535, + "learning_rate": 1.5642151481888036e-07, + "loss": 1.136, + "mean_token_accuracy": 0.6727185249328613, + "num_tokens": 7361400.0, + "step": 286 + }, + { + "epoch": 0.0315176806501208, + "grad_norm": 5.538603782653809, + "learning_rate": 1.5697036223929747e-07, + "loss": 1.1892, + "mean_token_accuracy": 0.6634823083877563, + "num_tokens": 7381858.0, + "step": 287 + }, + { + "epoch": 0.031627498352734464, + "grad_norm": 4.71835994720459, + "learning_rate": 1.5751920965971461e-07, + "loss": 1.2465, + "mean_token_accuracy": 0.6524161696434021, + "num_tokens": 7414913.0, + "step": 288 + }, + { + "epoch": 0.03173731605534812, + "grad_norm": 5.111032962799072, + "learning_rate": 1.580680570801317e-07, + "loss": 1.192, + "mean_token_accuracy": 0.6492413282394409, + "num_tokens": 7438328.0, + "step": 289 + }, + { + "epoch": 0.03184713375796178, + "grad_norm": 4.437552452087402, + "learning_rate": 1.5861690450054885e-07, + "loss": 1.1864, + "mean_token_accuracy": 0.6628942489624023, + "num_tokens": 7467566.0, + "step": 290 + }, + { + "epoch": 0.03195695146057544, + "grad_norm": 5.008449554443359, + "learning_rate": 1.5916575192096596e-07, + "loss": 1.2607, + "mean_token_accuracy": 0.6504977941513062, + "num_tokens": 7492126.0, + "step": 291 + }, + { + "epoch": 0.0320667691631891, + "grad_norm": 4.141957759857178, + "learning_rate": 1.597145993413831e-07, + "loss": 1.3349, + "mean_token_accuracy": 0.6266418695449829, + "num_tokens": 7524628.0, + "step": 292 + }, + { + "epoch": 0.03217658686580277, + "grad_norm": 4.426998615264893, + "learning_rate": 1.602634467618002e-07, + "loss": 1.2673, + "mean_token_accuracy": 0.6460800766944885, + "num_tokens": 7554658.0, + "step": 293 + }, + { + "epoch": 0.03228640456841643, + "grad_norm": 4.826286315917969, + "learning_rate": 1.6081229418221733e-07, + "loss": 1.2704, + "mean_token_accuracy": 0.6371636390686035, + "num_tokens": 7583236.0, + "step": 294 + }, + { + "epoch": 0.03239622227103009, + "grad_norm": 3.9418528079986572, + "learning_rate": 1.6136114160263445e-07, + "loss": 1.1693, + "mean_token_accuracy": 0.6648514270782471, + "num_tokens": 7610517.0, + "step": 295 + }, + { + "epoch": 0.03250603997364375, + "grad_norm": 4.776228904724121, + "learning_rate": 1.619099890230516e-07, + "loss": 1.1662, + "mean_token_accuracy": 0.6616625785827637, + "num_tokens": 7639060.0, + "step": 296 + }, + { + "epoch": 0.03261585767625741, + "grad_norm": 5.310563087463379, + "learning_rate": 1.6245883644346873e-07, + "loss": 1.2511, + "mean_token_accuracy": 0.641758918762207, + "num_tokens": 7664797.0, + "step": 297 + }, + { + "epoch": 0.032725675378871076, + "grad_norm": 5.481172561645508, + "learning_rate": 1.6300768386388582e-07, + "loss": 1.1098, + "mean_token_accuracy": 0.6823903918266296, + "num_tokens": 7688740.0, + "step": 298 + }, + { + "epoch": 0.032835493081484736, + "grad_norm": 5.573007106781006, + "learning_rate": 1.6355653128430296e-07, + "loss": 1.2864, + "mean_token_accuracy": 0.6449390649795532, + "num_tokens": 7711053.0, + "step": 299 + }, + { + "epoch": 0.032945310784098396, + "grad_norm": 4.5194902420043945, + "learning_rate": 1.6410537870472008e-07, + "loss": 1.1813, + "mean_token_accuracy": 0.6627769470214844, + "num_tokens": 7741221.0, + "step": 300 + }, + { + "epoch": 0.033055128486712056, + "grad_norm": 5.145860195159912, + "learning_rate": 1.6465422612513722e-07, + "loss": 1.1544, + "mean_token_accuracy": 0.6658995151519775, + "num_tokens": 7763300.0, + "step": 301 + }, + { + "epoch": 0.03316494618932572, + "grad_norm": 8.00620174407959, + "learning_rate": 1.6520307354555433e-07, + "loss": 1.1799, + "mean_token_accuracy": 0.6585037112236023, + "num_tokens": 7783216.0, + "step": 302 + }, + { + "epoch": 0.03327476389193938, + "grad_norm": 6.2629594802856445, + "learning_rate": 1.6575192096597145e-07, + "loss": 1.2734, + "mean_token_accuracy": 0.6549623012542725, + "num_tokens": 7808756.0, + "step": 303 + }, + { + "epoch": 0.03338458159455304, + "grad_norm": 4.255551815032959, + "learning_rate": 1.6630076838638856e-07, + "loss": 1.2316, + "mean_token_accuracy": 0.6519004106521606, + "num_tokens": 7841758.0, + "step": 304 + }, + { + "epoch": 0.0334943992971667, + "grad_norm": 5.454327583312988, + "learning_rate": 1.668496158068057e-07, + "loss": 1.2133, + "mean_token_accuracy": 0.6528989672660828, + "num_tokens": 7868013.0, + "step": 305 + }, + { + "epoch": 0.03360421699978036, + "grad_norm": 4.892307758331299, + "learning_rate": 1.6739846322722282e-07, + "loss": 1.1535, + "mean_token_accuracy": 0.6719813346862793, + "num_tokens": 7897750.0, + "step": 306 + }, + { + "epoch": 0.03371403470239403, + "grad_norm": 5.127041339874268, + "learning_rate": 1.6794731064763996e-07, + "loss": 1.2561, + "mean_token_accuracy": 0.6492636799812317, + "num_tokens": 7924824.0, + "step": 307 + }, + { + "epoch": 0.03382385240500769, + "grad_norm": 4.848952770233154, + "learning_rate": 1.6849615806805705e-07, + "loss": 1.2415, + "mean_token_accuracy": 0.6467645168304443, + "num_tokens": 7948465.0, + "step": 308 + }, + { + "epoch": 0.03393367010762135, + "grad_norm": 5.040799140930176, + "learning_rate": 1.690450054884742e-07, + "loss": 1.1929, + "mean_token_accuracy": 0.6619046926498413, + "num_tokens": 7972186.0, + "step": 309 + }, + { + "epoch": 0.03404348781023501, + "grad_norm": 4.79133415222168, + "learning_rate": 1.6959385290889134e-07, + "loss": 1.2418, + "mean_token_accuracy": 0.6532304286956787, + "num_tokens": 8003200.0, + "step": 310 + }, + { + "epoch": 0.03415330551284867, + "grad_norm": 4.754581451416016, + "learning_rate": 1.7014270032930845e-07, + "loss": 1.2473, + "mean_token_accuracy": 0.6465276479721069, + "num_tokens": 8038505.0, + "step": 311 + }, + { + "epoch": 0.034263123215462335, + "grad_norm": 4.169299602508545, + "learning_rate": 1.7069154774972557e-07, + "loss": 1.2551, + "mean_token_accuracy": 0.6422668099403381, + "num_tokens": 8066925.0, + "step": 312 + }, + { + "epoch": 0.034372940918075995, + "grad_norm": 5.375011444091797, + "learning_rate": 1.7124039517014268e-07, + "loss": 1.1481, + "mean_token_accuracy": 0.6790687441825867, + "num_tokens": 8091033.0, + "step": 313 + }, + { + "epoch": 0.034482758620689655, + "grad_norm": 4.372154235839844, + "learning_rate": 1.7178924259055982e-07, + "loss": 1.244, + "mean_token_accuracy": 0.6493252515792847, + "num_tokens": 8119522.0, + "step": 314 + }, + { + "epoch": 0.034592576323303315, + "grad_norm": 5.615719795227051, + "learning_rate": 1.7233809001097694e-07, + "loss": 1.1401, + "mean_token_accuracy": 0.6761949062347412, + "num_tokens": 8143617.0, + "step": 315 + }, + { + "epoch": 0.034702394025916974, + "grad_norm": 6.236863136291504, + "learning_rate": 1.7288693743139408e-07, + "loss": 1.1213, + "mean_token_accuracy": 0.6754556894302368, + "num_tokens": 8169066.0, + "step": 316 + }, + { + "epoch": 0.03481221172853064, + "grad_norm": 4.913151264190674, + "learning_rate": 1.7343578485181117e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.664522111415863, + "num_tokens": 8191383.0, + "step": 317 + }, + { + "epoch": 0.0349220294311443, + "grad_norm": 4.797080993652344, + "learning_rate": 1.739846322722283e-07, + "loss": 1.2021, + "mean_token_accuracy": 0.6607244610786438, + "num_tokens": 8221442.0, + "step": 318 + }, + { + "epoch": 0.03503184713375796, + "grad_norm": 4.657301425933838, + "learning_rate": 1.7453347969264543e-07, + "loss": 1.1139, + "mean_token_accuracy": 0.6756144165992737, + "num_tokens": 8246934.0, + "step": 319 + }, + { + "epoch": 0.03514166483637162, + "grad_norm": 5.108613014221191, + "learning_rate": 1.7508232711306257e-07, + "loss": 1.1752, + "mean_token_accuracy": 0.6688661575317383, + "num_tokens": 8267873.0, + "step": 320 + }, + { + "epoch": 0.03525148253898529, + "grad_norm": 5.478119850158691, + "learning_rate": 1.756311745334797e-07, + "loss": 1.2058, + "mean_token_accuracy": 0.6601643562316895, + "num_tokens": 8287980.0, + "step": 321 + }, + { + "epoch": 0.03536130024159895, + "grad_norm": 3.754344940185547, + "learning_rate": 1.761800219538968e-07, + "loss": 1.2054, + "mean_token_accuracy": 0.6644314527511597, + "num_tokens": 8315873.0, + "step": 322 + }, + { + "epoch": 0.03547111794421261, + "grad_norm": 5.089958667755127, + "learning_rate": 1.7672886937431394e-07, + "loss": 1.1237, + "mean_token_accuracy": 0.6684766411781311, + "num_tokens": 8333730.0, + "step": 323 + }, + { + "epoch": 0.03558093564682627, + "grad_norm": 5.286945343017578, + "learning_rate": 1.7727771679473105e-07, + "loss": 1.2792, + "mean_token_accuracy": 0.6342870593070984, + "num_tokens": 8360693.0, + "step": 324 + }, + { + "epoch": 0.03569075334943993, + "grad_norm": 5.770858287811279, + "learning_rate": 1.778265642151482e-07, + "loss": 1.2908, + "mean_token_accuracy": 0.6427783966064453, + "num_tokens": 8385306.0, + "step": 325 + }, + { + "epoch": 0.035800571052053594, + "grad_norm": 5.258806228637695, + "learning_rate": 1.7837541163556529e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6766500473022461, + "num_tokens": 8407250.0, + "step": 326 + }, + { + "epoch": 0.035910388754667254, + "grad_norm": 4.0424628257751465, + "learning_rate": 1.7892425905598243e-07, + "loss": 1.2666, + "mean_token_accuracy": 0.6436549425125122, + "num_tokens": 8434646.0, + "step": 327 + }, + { + "epoch": 0.036020206457280914, + "grad_norm": 5.683648586273193, + "learning_rate": 1.7947310647639954e-07, + "loss": 1.2124, + "mean_token_accuracy": 0.6661244630813599, + "num_tokens": 8464009.0, + "step": 328 + }, + { + "epoch": 0.03613002415989457, + "grad_norm": 3.9171226024627686, + "learning_rate": 1.8002195389681668e-07, + "loss": 1.2784, + "mean_token_accuracy": 0.6375925540924072, + "num_tokens": 8492022.0, + "step": 329 + }, + { + "epoch": 0.03623984186250823, + "grad_norm": 5.275128364562988, + "learning_rate": 1.805708013172338e-07, + "loss": 1.2988, + "mean_token_accuracy": 0.6445422768592834, + "num_tokens": 8519833.0, + "step": 330 + }, + { + "epoch": 0.0363496595651219, + "grad_norm": 5.645337104797363, + "learning_rate": 1.8111964873765091e-07, + "loss": 1.3083, + "mean_token_accuracy": 0.6318258047103882, + "num_tokens": 8544642.0, + "step": 331 + }, + { + "epoch": 0.03645947726773556, + "grad_norm": 6.019391059875488, + "learning_rate": 1.8166849615806803e-07, + "loss": 1.2282, + "mean_token_accuracy": 0.6463623046875, + "num_tokens": 8569888.0, + "step": 332 + }, + { + "epoch": 0.03656929497034922, + "grad_norm": 4.573214054107666, + "learning_rate": 1.8221734357848517e-07, + "loss": 1.1467, + "mean_token_accuracy": 0.6728915572166443, + "num_tokens": 8593324.0, + "step": 333 + }, + { + "epoch": 0.03667911267296288, + "grad_norm": 3.7564218044281006, + "learning_rate": 1.827661909989023e-07, + "loss": 1.217, + "mean_token_accuracy": 0.6579087376594543, + "num_tokens": 8624711.0, + "step": 334 + }, + { + "epoch": 0.03678893037557654, + "grad_norm": 5.095293998718262, + "learning_rate": 1.8331503841931943e-07, + "loss": 1.1291, + "mean_token_accuracy": 0.6781830787658691, + "num_tokens": 8645309.0, + "step": 335 + }, + { + "epoch": 0.036898748078190206, + "grad_norm": 4.18710470199585, + "learning_rate": 1.8386388583973654e-07, + "loss": 1.1626, + "mean_token_accuracy": 0.6633323431015015, + "num_tokens": 8665439.0, + "step": 336 + }, + { + "epoch": 0.037008565780803866, + "grad_norm": 4.013278961181641, + "learning_rate": 1.8441273326015366e-07, + "loss": 1.1587, + "mean_token_accuracy": 0.6703263521194458, + "num_tokens": 8688967.0, + "step": 337 + }, + { + "epoch": 0.037118383483417526, + "grad_norm": 4.476146697998047, + "learning_rate": 1.849615806805708e-07, + "loss": 1.1618, + "mean_token_accuracy": 0.6635377407073975, + "num_tokens": 8714799.0, + "step": 338 + }, + { + "epoch": 0.037228201186031186, + "grad_norm": 4.6193928718566895, + "learning_rate": 1.8551042810098792e-07, + "loss": 1.1489, + "mean_token_accuracy": 0.6706925630569458, + "num_tokens": 8739315.0, + "step": 339 + }, + { + "epoch": 0.03733801888864485, + "grad_norm": 4.4056172370910645, + "learning_rate": 1.8605927552140506e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6825239062309265, + "num_tokens": 8759764.0, + "step": 340 + }, + { + "epoch": 0.03744783659125851, + "grad_norm": 4.646774768829346, + "learning_rate": 1.8660812294182215e-07, + "loss": 1.2411, + "mean_token_accuracy": 0.6526185274124146, + "num_tokens": 8784971.0, + "step": 341 + }, + { + "epoch": 0.03755765429387217, + "grad_norm": 3.9222517013549805, + "learning_rate": 1.871569703622393e-07, + "loss": 1.26, + "mean_token_accuracy": 0.6398816108703613, + "num_tokens": 8815087.0, + "step": 342 + }, + { + "epoch": 0.03766747199648583, + "grad_norm": 4.533214092254639, + "learning_rate": 1.877058177826564e-07, + "loss": 1.1742, + "mean_token_accuracy": 0.6612622737884521, + "num_tokens": 8841018.0, + "step": 343 + }, + { + "epoch": 0.03777728969909949, + "grad_norm": 5.0550923347473145, + "learning_rate": 1.8825466520307354e-07, + "loss": 1.2443, + "mean_token_accuracy": 0.6478686332702637, + "num_tokens": 8865328.0, + "step": 344 + }, + { + "epoch": 0.03788710740171316, + "grad_norm": 5.412107467651367, + "learning_rate": 1.8880351262349069e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.6744738221168518, + "num_tokens": 8884254.0, + "step": 345 + }, + { + "epoch": 0.03799692510432682, + "grad_norm": 4.299205780029297, + "learning_rate": 1.8935236004390778e-07, + "loss": 1.2655, + "mean_token_accuracy": 0.6421260237693787, + "num_tokens": 8913386.0, + "step": 346 + }, + { + "epoch": 0.03810674280694048, + "grad_norm": 5.519776344299316, + "learning_rate": 1.8990120746432492e-07, + "loss": 1.2697, + "mean_token_accuracy": 0.6369417905807495, + "num_tokens": 8942684.0, + "step": 347 + }, + { + "epoch": 0.03821656050955414, + "grad_norm": 4.647257328033447, + "learning_rate": 1.9045005488474203e-07, + "loss": 1.233, + "mean_token_accuracy": 0.6543713212013245, + "num_tokens": 8974735.0, + "step": 348 + }, + { + "epoch": 0.0383263782121678, + "grad_norm": 4.860410690307617, + "learning_rate": 1.9099890230515917e-07, + "loss": 1.1722, + "mean_token_accuracy": 0.6596894264221191, + "num_tokens": 8997354.0, + "step": 349 + }, + { + "epoch": 0.038436195914781465, + "grad_norm": 5.271847248077393, + "learning_rate": 1.9154774972557626e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6775184869766235, + "num_tokens": 9016362.0, + "step": 350 + }, + { + "epoch": 0.038546013617395125, + "grad_norm": 4.628877639770508, + "learning_rate": 1.920965971459934e-07, + "loss": 1.2148, + "mean_token_accuracy": 0.6692101359367371, + "num_tokens": 9041050.0, + "step": 351 + }, + { + "epoch": 0.038655831320008785, + "grad_norm": 4.657959461212158, + "learning_rate": 1.9264544456641052e-07, + "loss": 1.1164, + "mean_token_accuracy": 0.6756889820098877, + "num_tokens": 9064296.0, + "step": 352 + }, + { + "epoch": 0.038765649022622445, + "grad_norm": 4.5909905433654785, + "learning_rate": 1.9319429198682766e-07, + "loss": 1.2979, + "mean_token_accuracy": 0.6377670168876648, + "num_tokens": 9093688.0, + "step": 353 + }, + { + "epoch": 0.038875466725236105, + "grad_norm": 5.479106903076172, + "learning_rate": 1.9374313940724478e-07, + "loss": 1.1743, + "mean_token_accuracy": 0.6635377407073975, + "num_tokens": 9116052.0, + "step": 354 + }, + { + "epoch": 0.03898528442784977, + "grad_norm": 4.317579746246338, + "learning_rate": 1.942919868276619e-07, + "loss": 1.2292, + "mean_token_accuracy": 0.649475634098053, + "num_tokens": 9138437.0, + "step": 355 + }, + { + "epoch": 0.03909510213046343, + "grad_norm": 4.659647464752197, + "learning_rate": 1.9484083424807903e-07, + "loss": 1.1764, + "mean_token_accuracy": 0.6653316020965576, + "num_tokens": 9168999.0, + "step": 356 + }, + { + "epoch": 0.03920491983307709, + "grad_norm": 4.365380764007568, + "learning_rate": 1.9538968166849615e-07, + "loss": 1.2637, + "mean_token_accuracy": 0.6520602703094482, + "num_tokens": 9195029.0, + "step": 357 + }, + { + "epoch": 0.03931473753569075, + "grad_norm": 4.741463661193848, + "learning_rate": 1.959385290889133e-07, + "loss": 1.1663, + "mean_token_accuracy": 0.6683056950569153, + "num_tokens": 9219177.0, + "step": 358 + }, + { + "epoch": 0.03942455523830442, + "grad_norm": 5.0147271156311035, + "learning_rate": 1.964873765093304e-07, + "loss": 1.1959, + "mean_token_accuracy": 0.6539356708526611, + "num_tokens": 9246239.0, + "step": 359 + }, + { + "epoch": 0.03953437294091808, + "grad_norm": 4.3269782066345215, + "learning_rate": 1.9703622392974752e-07, + "loss": 1.1828, + "mean_token_accuracy": 0.670573890209198, + "num_tokens": 9268623.0, + "step": 360 + }, + { + "epoch": 0.03964419064353174, + "grad_norm": 3.657102346420288, + "learning_rate": 1.9758507135016464e-07, + "loss": 1.1891, + "mean_token_accuracy": 0.6553224325180054, + "num_tokens": 9293450.0, + "step": 361 + }, + { + "epoch": 0.0397540083461454, + "grad_norm": 3.784867763519287, + "learning_rate": 1.9813391877058178e-07, + "loss": 1.2026, + "mean_token_accuracy": 0.6541498899459839, + "num_tokens": 9321693.0, + "step": 362 + }, + { + "epoch": 0.03986382604875906, + "grad_norm": 3.7236554622650146, + "learning_rate": 1.986827661909989e-07, + "loss": 1.211, + "mean_token_accuracy": 0.6544553637504578, + "num_tokens": 9351778.0, + "step": 363 + }, + { + "epoch": 0.039973643751372724, + "grad_norm": 3.692272663116455, + "learning_rate": 1.9923161361141603e-07, + "loss": 1.2215, + "mean_token_accuracy": 0.6654762029647827, + "num_tokens": 9376940.0, + "step": 364 + }, + { + "epoch": 0.040083461453986384, + "grad_norm": 3.312061071395874, + "learning_rate": 1.9978046103183312e-07, + "loss": 1.2615, + "mean_token_accuracy": 0.6435597538948059, + "num_tokens": 9406700.0, + "step": 365 + }, + { + "epoch": 0.040193279156600044, + "grad_norm": 5.031953811645508, + "learning_rate": 2.0032930845225027e-07, + "loss": 1.2261, + "mean_token_accuracy": 0.6442521214485168, + "num_tokens": 9427607.0, + "step": 366 + }, + { + "epoch": 0.040303096859213704, + "grad_norm": 3.668254852294922, + "learning_rate": 2.0087815587266738e-07, + "loss": 1.1688, + "mean_token_accuracy": 0.6738631725311279, + "num_tokens": 9458662.0, + "step": 367 + }, + { + "epoch": 0.040412914561827364, + "grad_norm": 3.8675010204315186, + "learning_rate": 2.0142700329308452e-07, + "loss": 1.1658, + "mean_token_accuracy": 0.6645565032958984, + "num_tokens": 9486137.0, + "step": 368 + }, + { + "epoch": 0.04052273226444103, + "grad_norm": 4.1418890953063965, + "learning_rate": 2.0197585071350164e-07, + "loss": 1.1935, + "mean_token_accuracy": 0.6814839839935303, + "num_tokens": 9510551.0, + "step": 369 + }, + { + "epoch": 0.04063254996705469, + "grad_norm": 3.540675401687622, + "learning_rate": 2.0252469813391875e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6946071982383728, + "num_tokens": 9534960.0, + "step": 370 + }, + { + "epoch": 0.04074236766966835, + "grad_norm": 4.6879377365112305, + "learning_rate": 2.030735455543359e-07, + "loss": 1.1245, + "mean_token_accuracy": 0.6922149062156677, + "num_tokens": 9558750.0, + "step": 371 + }, + { + "epoch": 0.04085218537228201, + "grad_norm": 3.1030139923095703, + "learning_rate": 2.03622392974753e-07, + "loss": 1.1522, + "mean_token_accuracy": 0.6674572229385376, + "num_tokens": 9590604.0, + "step": 372 + }, + { + "epoch": 0.04096200307489567, + "grad_norm": 4.621713638305664, + "learning_rate": 2.0417124039517015e-07, + "loss": 1.187, + "mean_token_accuracy": 0.6683238744735718, + "num_tokens": 9614048.0, + "step": 373 + }, + { + "epoch": 0.04107182077750934, + "grad_norm": 3.039022207260132, + "learning_rate": 2.0472008781558724e-07, + "loss": 1.1748, + "mean_token_accuracy": 0.6655710339546204, + "num_tokens": 9650350.0, + "step": 374 + }, + { + "epoch": 0.041181638480122996, + "grad_norm": 3.4229841232299805, + "learning_rate": 2.0526893523600438e-07, + "loss": 1.2542, + "mean_token_accuracy": 0.6431545615196228, + "num_tokens": 9682772.0, + "step": 375 + }, + { + "epoch": 0.041291456182736656, + "grad_norm": 3.501650094985962, + "learning_rate": 2.058177826564215e-07, + "loss": 1.2101, + "mean_token_accuracy": 0.6596134901046753, + "num_tokens": 9712504.0, + "step": 376 + }, + { + "epoch": 0.041401273885350316, + "grad_norm": 4.744318962097168, + "learning_rate": 2.0636663007683864e-07, + "loss": 1.1952, + "mean_token_accuracy": 0.6555055379867554, + "num_tokens": 9734221.0, + "step": 377 + }, + { + "epoch": 0.04151109158796398, + "grad_norm": 4.248786926269531, + "learning_rate": 2.0691547749725575e-07, + "loss": 1.2316, + "mean_token_accuracy": 0.6463046073913574, + "num_tokens": 9759712.0, + "step": 378 + }, + { + "epoch": 0.04162090929057764, + "grad_norm": 3.599134683609009, + "learning_rate": 2.0746432491767287e-07, + "loss": 1.2642, + "mean_token_accuracy": 0.6355037093162537, + "num_tokens": 9788920.0, + "step": 379 + }, + { + "epoch": 0.0417307269931913, + "grad_norm": 3.882537841796875, + "learning_rate": 2.0801317233809e-07, + "loss": 1.2251, + "mean_token_accuracy": 0.6540446877479553, + "num_tokens": 9817458.0, + "step": 380 + }, + { + "epoch": 0.04184054469580496, + "grad_norm": 3.625706672668457, + "learning_rate": 2.0856201975850713e-07, + "loss": 1.1926, + "mean_token_accuracy": 0.665017306804657, + "num_tokens": 9849225.0, + "step": 381 + }, + { + "epoch": 0.04195036239841862, + "grad_norm": 3.959286689758301, + "learning_rate": 2.0911086717892427e-07, + "loss": 1.1496, + "mean_token_accuracy": 0.6644473671913147, + "num_tokens": 9874546.0, + "step": 382 + }, + { + "epoch": 0.04206018010103229, + "grad_norm": 4.174918174743652, + "learning_rate": 2.0965971459934136e-07, + "loss": 1.2616, + "mean_token_accuracy": 0.6447650790214539, + "num_tokens": 9902385.0, + "step": 383 + }, + { + "epoch": 0.04216999780364595, + "grad_norm": 3.808929204940796, + "learning_rate": 2.102085620197585e-07, + "loss": 1.1569, + "mean_token_accuracy": 0.6674897074699402, + "num_tokens": 9928636.0, + "step": 384 + }, + { + "epoch": 0.04227981550625961, + "grad_norm": 4.549903392791748, + "learning_rate": 2.1075740944017561e-07, + "loss": 1.1874, + "mean_token_accuracy": 0.6600097417831421, + "num_tokens": 9955707.0, + "step": 385 + }, + { + "epoch": 0.04238963320887327, + "grad_norm": 4.274919033050537, + "learning_rate": 2.1130625686059276e-07, + "loss": 1.1928, + "mean_token_accuracy": 0.6567590236663818, + "num_tokens": 9980567.0, + "step": 386 + }, + { + "epoch": 0.04249945091148693, + "grad_norm": 3.2088677883148193, + "learning_rate": 2.1185510428100987e-07, + "loss": 1.1984, + "mean_token_accuracy": 0.6589920520782471, + "num_tokens": 10008300.0, + "step": 387 + }, + { + "epoch": 0.042609268614100596, + "grad_norm": 3.5249454975128174, + "learning_rate": 2.1240395170142699e-07, + "loss": 1.1585, + "mean_token_accuracy": 0.6681839823722839, + "num_tokens": 10035794.0, + "step": 388 + }, + { + "epoch": 0.042719086316714255, + "grad_norm": 3.9928009510040283, + "learning_rate": 2.129527991218441e-07, + "loss": 1.1374, + "mean_token_accuracy": 0.6729857921600342, + "num_tokens": 10058348.0, + "step": 389 + }, + { + "epoch": 0.042828904019327915, + "grad_norm": 4.884435653686523, + "learning_rate": 2.1350164654226124e-07, + "loss": 1.1499, + "mean_token_accuracy": 0.6644936203956604, + "num_tokens": 10077413.0, + "step": 390 + }, + { + "epoch": 0.042938721721941575, + "grad_norm": 3.867485523223877, + "learning_rate": 2.1405049396267836e-07, + "loss": 1.1483, + "mean_token_accuracy": 0.6686339378356934, + "num_tokens": 10104744.0, + "step": 391 + }, + { + "epoch": 0.043048539424555235, + "grad_norm": 4.015164375305176, + "learning_rate": 2.145993413830955e-07, + "loss": 1.1162, + "mean_token_accuracy": 0.6751184463500977, + "num_tokens": 10130760.0, + "step": 392 + }, + { + "epoch": 0.0431583571271689, + "grad_norm": 4.076845169067383, + "learning_rate": 2.1514818880351262e-07, + "loss": 1.1194, + "mean_token_accuracy": 0.6799368858337402, + "num_tokens": 10153035.0, + "step": 393 + }, + { + "epoch": 0.04326817482978256, + "grad_norm": 3.20078444480896, + "learning_rate": 2.1569703622392973e-07, + "loss": 1.2346, + "mean_token_accuracy": 0.6458620429039001, + "num_tokens": 10178048.0, + "step": 394 + }, + { + "epoch": 0.04337799253239622, + "grad_norm": 3.378406524658203, + "learning_rate": 2.1624588364434687e-07, + "loss": 1.229, + "mean_token_accuracy": 0.6583672761917114, + "num_tokens": 10207802.0, + "step": 395 + }, + { + "epoch": 0.04348781023500988, + "grad_norm": 3.6694142818450928, + "learning_rate": 2.16794731064764e-07, + "loss": 1.2797, + "mean_token_accuracy": 0.6504186391830444, + "num_tokens": 10233941.0, + "step": 396 + }, + { + "epoch": 0.04359762793762355, + "grad_norm": 4.900756359100342, + "learning_rate": 2.1734357848518113e-07, + "loss": 1.1229, + "mean_token_accuracy": 0.677718997001648, + "num_tokens": 10251742.0, + "step": 397 + }, + { + "epoch": 0.04370744564023721, + "grad_norm": 3.6742780208587646, + "learning_rate": 2.1789242590559822e-07, + "loss": 1.2521, + "mean_token_accuracy": 0.641974687576294, + "num_tokens": 10279724.0, + "step": 398 + }, + { + "epoch": 0.04381726334285087, + "grad_norm": 3.0844151973724365, + "learning_rate": 2.1844127332601536e-07, + "loss": 1.2187, + "mean_token_accuracy": 0.6504505276679993, + "num_tokens": 10312541.0, + "step": 399 + }, + { + "epoch": 0.04392708104546453, + "grad_norm": 3.153944253921509, + "learning_rate": 2.1899012074643247e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6732572317123413, + "num_tokens": 10341927.0, + "step": 400 + }, + { + "epoch": 0.04403689874807819, + "grad_norm": 4.07037878036499, + "learning_rate": 2.1953896816684962e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6882526874542236, + "num_tokens": 10364914.0, + "step": 401 + }, + { + "epoch": 0.044146716450691854, + "grad_norm": 3.330515146255493, + "learning_rate": 2.200878155872667e-07, + "loss": 1.1973, + "mean_token_accuracy": 0.6603466272354126, + "num_tokens": 10396314.0, + "step": 402 + }, + { + "epoch": 0.044256534153305514, + "grad_norm": 4.096805095672607, + "learning_rate": 2.2063666300768385e-07, + "loss": 1.2339, + "mean_token_accuracy": 0.6443521976470947, + "num_tokens": 10422147.0, + "step": 403 + }, + { + "epoch": 0.044366351855919174, + "grad_norm": 3.751723051071167, + "learning_rate": 2.21185510428101e-07, + "loss": 1.3154, + "mean_token_accuracy": 0.6384576559066772, + "num_tokens": 10451605.0, + "step": 404 + }, + { + "epoch": 0.044476169558532834, + "grad_norm": 3.450679063796997, + "learning_rate": 2.217343578485181e-07, + "loss": 1.2227, + "mean_token_accuracy": 0.6524843573570251, + "num_tokens": 10481615.0, + "step": 405 + }, + { + "epoch": 0.044585987261146494, + "grad_norm": 3.6172492504119873, + "learning_rate": 2.2228320526893525e-07, + "loss": 1.2451, + "mean_token_accuracy": 0.6475496292114258, + "num_tokens": 10508420.0, + "step": 406 + }, + { + "epoch": 0.04469580496376016, + "grad_norm": 2.8399195671081543, + "learning_rate": 2.2283205268935233e-07, + "loss": 1.1729, + "mean_token_accuracy": 0.6716769933700562, + "num_tokens": 10536423.0, + "step": 407 + }, + { + "epoch": 0.04480562266637382, + "grad_norm": 3.5181398391723633, + "learning_rate": 2.2338090010976948e-07, + "loss": 1.0971, + "mean_token_accuracy": 0.6825659871101379, + "num_tokens": 10560590.0, + "step": 408 + }, + { + "epoch": 0.04491544036898748, + "grad_norm": 3.222280263900757, + "learning_rate": 2.239297475301866e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6719815135002136, + "num_tokens": 10585761.0, + "step": 409 + }, + { + "epoch": 0.04502525807160114, + "grad_norm": 3.3230042457580566, + "learning_rate": 2.2447859495060373e-07, + "loss": 1.1912, + "mean_token_accuracy": 0.6681489944458008, + "num_tokens": 10610774.0, + "step": 410 + }, + { + "epoch": 0.0451350757742148, + "grad_norm": 3.5743682384490967, + "learning_rate": 2.2502744237102085e-07, + "loss": 1.1861, + "mean_token_accuracy": 0.6670110821723938, + "num_tokens": 10638907.0, + "step": 411 + }, + { + "epoch": 0.04524489347682847, + "grad_norm": 3.6678457260131836, + "learning_rate": 2.2557628979143796e-07, + "loss": 1.1611, + "mean_token_accuracy": 0.6771235466003418, + "num_tokens": 10661208.0, + "step": 412 + }, + { + "epoch": 0.04535471117944213, + "grad_norm": 3.3754842281341553, + "learning_rate": 2.2612513721185508e-07, + "loss": 1.1561, + "mean_token_accuracy": 0.6626687049865723, + "num_tokens": 10686180.0, + "step": 413 + }, + { + "epoch": 0.04546452888205579, + "grad_norm": 4.065917015075684, + "learning_rate": 2.2667398463227222e-07, + "loss": 1.1377, + "mean_token_accuracy": 0.6754779815673828, + "num_tokens": 10705639.0, + "step": 414 + }, + { + "epoch": 0.045574346584669446, + "grad_norm": 3.4182162284851074, + "learning_rate": 2.2722283205268934e-07, + "loss": 1.1886, + "mean_token_accuracy": 0.656450629234314, + "num_tokens": 10731797.0, + "step": 415 + }, + { + "epoch": 0.04568416428728311, + "grad_norm": 3.3261005878448486, + "learning_rate": 2.2777167947310648e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6601096391677856, + "num_tokens": 10762051.0, + "step": 416 + }, + { + "epoch": 0.04579398198989677, + "grad_norm": 4.044336795806885, + "learning_rate": 2.283205268935236e-07, + "loss": 1.1591, + "mean_token_accuracy": 0.6667532920837402, + "num_tokens": 10785075.0, + "step": 417 + }, + { + "epoch": 0.04590379969251043, + "grad_norm": 2.9511122703552246, + "learning_rate": 2.288693743139407e-07, + "loss": 1.1805, + "mean_token_accuracy": 0.6670855283737183, + "num_tokens": 10815910.0, + "step": 418 + }, + { + "epoch": 0.04601361739512409, + "grad_norm": 3.437960624694824, + "learning_rate": 2.2941822173435785e-07, + "loss": 1.1563, + "mean_token_accuracy": 0.6674416065216064, + "num_tokens": 10843686.0, + "step": 419 + }, + { + "epoch": 0.04612343509773775, + "grad_norm": 3.4877684116363525, + "learning_rate": 2.2996706915477496e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.6708109378814697, + "num_tokens": 10866115.0, + "step": 420 + }, + { + "epoch": 0.04623325280035142, + "grad_norm": 3.7728731632232666, + "learning_rate": 2.3051591657519208e-07, + "loss": 1.1023, + "mean_token_accuracy": 0.6757241487503052, + "num_tokens": 10890646.0, + "step": 421 + }, + { + "epoch": 0.04634307050296508, + "grad_norm": 4.388121604919434, + "learning_rate": 2.310647639956092e-07, + "loss": 1.2894, + "mean_token_accuracy": 0.6337904334068298, + "num_tokens": 10914719.0, + "step": 422 + }, + { + "epoch": 0.04645288820557874, + "grad_norm": 3.6479876041412354, + "learning_rate": 2.3161361141602634e-07, + "loss": 1.1999, + "mean_token_accuracy": 0.6607551574707031, + "num_tokens": 10939132.0, + "step": 423 + }, + { + "epoch": 0.0465627059081924, + "grad_norm": 3.877553701400757, + "learning_rate": 2.3216245883644345e-07, + "loss": 1.1151, + "mean_token_accuracy": 0.677558422088623, + "num_tokens": 10959273.0, + "step": 424 + }, + { + "epoch": 0.04667252361080606, + "grad_norm": 3.789616584777832, + "learning_rate": 2.327113062568606e-07, + "loss": 1.1883, + "mean_token_accuracy": 0.6569750905036926, + "num_tokens": 10981172.0, + "step": 425 + }, + { + "epoch": 0.046782341313419726, + "grad_norm": 3.551677703857422, + "learning_rate": 2.3326015367727768e-07, + "loss": 1.0899, + "mean_token_accuracy": 0.6860036849975586, + "num_tokens": 11005733.0, + "step": 426 + }, + { + "epoch": 0.046892159016033386, + "grad_norm": 3.422370672225952, + "learning_rate": 2.3380900109769482e-07, + "loss": 1.2347, + "mean_token_accuracy": 0.6574031710624695, + "num_tokens": 11031249.0, + "step": 427 + }, + { + "epoch": 0.047001976718647046, + "grad_norm": 3.178609848022461, + "learning_rate": 2.3435784851811197e-07, + "loss": 1.1271, + "mean_token_accuracy": 0.6687119603157043, + "num_tokens": 11057874.0, + "step": 428 + }, + { + "epoch": 0.047111794421260705, + "grad_norm": 3.239307165145874, + "learning_rate": 2.3490669593852908e-07, + "loss": 1.1428, + "mean_token_accuracy": 0.6680513620376587, + "num_tokens": 11086084.0, + "step": 429 + }, + { + "epoch": 0.047221612123874365, + "grad_norm": 3.122530698776245, + "learning_rate": 2.3545554335894622e-07, + "loss": 1.2595, + "mean_token_accuracy": 0.6400102376937866, + "num_tokens": 11117820.0, + "step": 430 + }, + { + "epoch": 0.04733142982648803, + "grad_norm": 3.3456263542175293, + "learning_rate": 2.360043907793633e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6746511459350586, + "num_tokens": 11142323.0, + "step": 431 + }, + { + "epoch": 0.04744124752910169, + "grad_norm": 2.8937625885009766, + "learning_rate": 2.3655323819978045e-07, + "loss": 1.1967, + "mean_token_accuracy": 0.6529737710952759, + "num_tokens": 11170146.0, + "step": 432 + }, + { + "epoch": 0.04755106523171535, + "grad_norm": 3.1056253910064697, + "learning_rate": 2.3710208562019757e-07, + "loss": 1.1375, + "mean_token_accuracy": 0.67415452003479, + "num_tokens": 11195545.0, + "step": 433 + }, + { + "epoch": 0.04766088293432901, + "grad_norm": 2.794980764389038, + "learning_rate": 2.376509330406147e-07, + "loss": 1.2223, + "mean_token_accuracy": 0.6490035057067871, + "num_tokens": 11226858.0, + "step": 434 + }, + { + "epoch": 0.04777070063694268, + "grad_norm": 3.3968803882598877, + "learning_rate": 2.3819978046103183e-07, + "loss": 1.1631, + "mean_token_accuracy": 0.6604456901550293, + "num_tokens": 11247266.0, + "step": 435 + }, + { + "epoch": 0.04788051833955634, + "grad_norm": 3.4340155124664307, + "learning_rate": 2.3874862788144894e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6771543025970459, + "num_tokens": 11267577.0, + "step": 436 + }, + { + "epoch": 0.04799033604217, + "grad_norm": 3.0786490440368652, + "learning_rate": 2.3929747530186606e-07, + "loss": 1.0941, + "mean_token_accuracy": 0.6813993453979492, + "num_tokens": 11298772.0, + "step": 437 + }, + { + "epoch": 0.04810015374478366, + "grad_norm": 3.9634082317352295, + "learning_rate": 2.3984632272228317e-07, + "loss": 1.2084, + "mean_token_accuracy": 0.6535360217094421, + "num_tokens": 11319202.0, + "step": 438 + }, + { + "epoch": 0.04820997144739732, + "grad_norm": 2.8951797485351562, + "learning_rate": 2.4039517014270034e-07, + "loss": 1.1683, + "mean_token_accuracy": 0.6569482684135437, + "num_tokens": 11347971.0, + "step": 439 + }, + { + "epoch": 0.048319789150010985, + "grad_norm": 3.6385416984558105, + "learning_rate": 2.4094401756311745e-07, + "loss": 1.2375, + "mean_token_accuracy": 0.6442484259605408, + "num_tokens": 11370660.0, + "step": 440 + }, + { + "epoch": 0.048429606852624645, + "grad_norm": 3.0462327003479004, + "learning_rate": 2.4149286498353457e-07, + "loss": 1.1819, + "mean_token_accuracy": 0.6558480262756348, + "num_tokens": 11400203.0, + "step": 441 + }, + { + "epoch": 0.048539424555238304, + "grad_norm": 3.515625476837158, + "learning_rate": 2.420417124039517e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.6736727356910706, + "num_tokens": 11421986.0, + "step": 442 + }, + { + "epoch": 0.048649242257851964, + "grad_norm": 3.05094313621521, + "learning_rate": 2.425905598243688e-07, + "loss": 1.2253, + "mean_token_accuracy": 0.6500388979911804, + "num_tokens": 11452593.0, + "step": 443 + }, + { + "epoch": 0.048759059960465624, + "grad_norm": 2.9217188358306885, + "learning_rate": 2.431394072447859e-07, + "loss": 1.1439, + "mean_token_accuracy": 0.6751363277435303, + "num_tokens": 11480419.0, + "step": 444 + }, + { + "epoch": 0.04886887766307929, + "grad_norm": 3.341031074523926, + "learning_rate": 2.436882546652031e-07, + "loss": 1.1478, + "mean_token_accuracy": 0.6719862818717957, + "num_tokens": 11503856.0, + "step": 445 + }, + { + "epoch": 0.04897869536569295, + "grad_norm": 3.2965476512908936, + "learning_rate": 2.442371020856202e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6876282691955566, + "num_tokens": 11526929.0, + "step": 446 + }, + { + "epoch": 0.04908851306830661, + "grad_norm": 3.788796901702881, + "learning_rate": 2.447859495060373e-07, + "loss": 1.1674, + "mean_token_accuracy": 0.6604946851730347, + "num_tokens": 11546677.0, + "step": 447 + }, + { + "epoch": 0.04919833077092027, + "grad_norm": 3.490683078765869, + "learning_rate": 2.4533479692645443e-07, + "loss": 1.1916, + "mean_token_accuracy": 0.6520256400108337, + "num_tokens": 11568198.0, + "step": 448 + }, + { + "epoch": 0.04930814847353393, + "grad_norm": 3.199925422668457, + "learning_rate": 2.4588364434687154e-07, + "loss": 1.1434, + "mean_token_accuracy": 0.6699213981628418, + "num_tokens": 11592941.0, + "step": 449 + }, + { + "epoch": 0.0494179661761476, + "grad_norm": 3.645038604736328, + "learning_rate": 2.4643249176728866e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6794770956039429, + "num_tokens": 11619971.0, + "step": 450 + }, + { + "epoch": 0.04952778387876126, + "grad_norm": 3.9774019718170166, + "learning_rate": 2.4698133918770583e-07, + "loss": 1.1981, + "mean_token_accuracy": 0.6563552618026733, + "num_tokens": 11642442.0, + "step": 451 + }, + { + "epoch": 0.04963760158137492, + "grad_norm": 3.4040884971618652, + "learning_rate": 2.4753018660812294e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6705517768859863, + "num_tokens": 11666064.0, + "step": 452 + }, + { + "epoch": 0.04974741928398858, + "grad_norm": 3.0721116065979004, + "learning_rate": 2.4807903402854006e-07, + "loss": 1.1887, + "mean_token_accuracy": 0.6543627977371216, + "num_tokens": 11690940.0, + "step": 453 + }, + { + "epoch": 0.049857236986602244, + "grad_norm": 3.2835757732391357, + "learning_rate": 2.486278814489572e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6598408818244934, + "num_tokens": 11713064.0, + "step": 454 + }, + { + "epoch": 0.0499670546892159, + "grad_norm": 4.040445327758789, + "learning_rate": 2.491767288693743e-07, + "loss": 1.1124, + "mean_token_accuracy": 0.6676604747772217, + "num_tokens": 11734919.0, + "step": 455 + }, + { + "epoch": 0.05007687239182956, + "grad_norm": 3.39760160446167, + "learning_rate": 2.4972557628979146e-07, + "loss": 1.2497, + "mean_token_accuracy": 0.6417697668075562, + "num_tokens": 11764924.0, + "step": 456 + }, + { + "epoch": 0.05018669009444322, + "grad_norm": 2.9585165977478027, + "learning_rate": 2.5027442371020857e-07, + "loss": 1.142, + "mean_token_accuracy": 0.6868826150894165, + "num_tokens": 11792212.0, + "step": 457 + }, + { + "epoch": 0.05029650779705688, + "grad_norm": 3.1210973262786865, + "learning_rate": 2.508232711306257e-07, + "loss": 1.1844, + "mean_token_accuracy": 0.6599335670471191, + "num_tokens": 11817359.0, + "step": 458 + }, + { + "epoch": 0.05040632549967055, + "grad_norm": 2.7316834926605225, + "learning_rate": 2.513721185510428e-07, + "loss": 1.168, + "mean_token_accuracy": 0.6690845489501953, + "num_tokens": 11844933.0, + "step": 459 + }, + { + "epoch": 0.05051614320228421, + "grad_norm": 2.6170201301574707, + "learning_rate": 2.519209659714599e-07, + "loss": 1.215, + "mean_token_accuracy": 0.6504680514335632, + "num_tokens": 11879572.0, + "step": 460 + }, + { + "epoch": 0.05062596090489787, + "grad_norm": 2.9689993858337402, + "learning_rate": 2.524698133918771e-07, + "loss": 1.1041, + "mean_token_accuracy": 0.6773097515106201, + "num_tokens": 11908862.0, + "step": 461 + }, + { + "epoch": 0.05073577860751153, + "grad_norm": 3.7472782135009766, + "learning_rate": 2.5301866081229415e-07, + "loss": 1.183, + "mean_token_accuracy": 0.6591687202453613, + "num_tokens": 11935504.0, + "step": 462 + }, + { + "epoch": 0.05084559631012519, + "grad_norm": 2.9580414295196533, + "learning_rate": 2.5356750823271126e-07, + "loss": 1.1625, + "mean_token_accuracy": 0.6671595573425293, + "num_tokens": 11966282.0, + "step": 463 + }, + { + "epoch": 0.050955414012738856, + "grad_norm": 2.992809772491455, + "learning_rate": 2.5411635565312843e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.6644184589385986, + "num_tokens": 11990732.0, + "step": 464 + }, + { + "epoch": 0.051065231715352516, + "grad_norm": 2.8793885707855225, + "learning_rate": 2.5466520307354555e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6818303465843201, + "num_tokens": 12021590.0, + "step": 465 + }, + { + "epoch": 0.051175049417966176, + "grad_norm": 3.301693916320801, + "learning_rate": 2.5521405049396266e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.6766242980957031, + "num_tokens": 12045045.0, + "step": 466 + }, + { + "epoch": 0.051284867120579836, + "grad_norm": 3.5958800315856934, + "learning_rate": 2.557628979143798e-07, + "loss": 1.1243, + "mean_token_accuracy": 0.6740423440933228, + "num_tokens": 12067800.0, + "step": 467 + }, + { + "epoch": 0.051394684823193496, + "grad_norm": 3.0918548107147217, + "learning_rate": 2.5631174533479695e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6960967779159546, + "num_tokens": 12091466.0, + "step": 468 + }, + { + "epoch": 0.05150450252580716, + "grad_norm": 2.966761827468872, + "learning_rate": 2.5686059275521406e-07, + "loss": 1.1665, + "mean_token_accuracy": 0.6663385629653931, + "num_tokens": 12116116.0, + "step": 469 + }, + { + "epoch": 0.05161432022842082, + "grad_norm": 2.7236087322235107, + "learning_rate": 2.574094401756312e-07, + "loss": 1.1692, + "mean_token_accuracy": 0.6680154204368591, + "num_tokens": 12142042.0, + "step": 470 + }, + { + "epoch": 0.05172413793103448, + "grad_norm": 3.0192484855651855, + "learning_rate": 2.5795828759604824e-07, + "loss": 1.3123, + "mean_token_accuracy": 0.6293004751205444, + "num_tokens": 12172524.0, + "step": 471 + }, + { + "epoch": 0.05183395563364814, + "grad_norm": 3.22973370552063, + "learning_rate": 2.585071350164654e-07, + "loss": 1.1457, + "mean_token_accuracy": 0.6643972396850586, + "num_tokens": 12195099.0, + "step": 472 + }, + { + "epoch": 0.05194377333626181, + "grad_norm": 2.8296382427215576, + "learning_rate": 2.590559824368825e-07, + "loss": 1.0987, + "mean_token_accuracy": 0.6790992617607117, + "num_tokens": 12220599.0, + "step": 473 + }, + { + "epoch": 0.05205359103887547, + "grad_norm": 3.022629976272583, + "learning_rate": 2.5960482985729964e-07, + "loss": 1.1148, + "mean_token_accuracy": 0.6762738227844238, + "num_tokens": 12243011.0, + "step": 474 + }, + { + "epoch": 0.05216340874148913, + "grad_norm": 2.7706663608551025, + "learning_rate": 2.601536772777168e-07, + "loss": 1.1516, + "mean_token_accuracy": 0.6734415292739868, + "num_tokens": 12271634.0, + "step": 475 + }, + { + "epoch": 0.05227322644410279, + "grad_norm": 2.895834445953369, + "learning_rate": 2.607025246981339e-07, + "loss": 1.1579, + "mean_token_accuracy": 0.6684916019439697, + "num_tokens": 12298827.0, + "step": 476 + }, + { + "epoch": 0.05238304414671645, + "grad_norm": 2.9947545528411865, + "learning_rate": 2.6125137211855104e-07, + "loss": 1.2174, + "mean_token_accuracy": 0.6583011746406555, + "num_tokens": 12325210.0, + "step": 477 + }, + { + "epoch": 0.052492861849330115, + "grad_norm": 3.0221712589263916, + "learning_rate": 2.6180021953896815e-07, + "loss": 1.1003, + "mean_token_accuracy": 0.6827609539031982, + "num_tokens": 12346787.0, + "step": 478 + }, + { + "epoch": 0.052602679551943775, + "grad_norm": 2.9393458366394043, + "learning_rate": 2.623490669593853e-07, + "loss": 1.2342, + "mean_token_accuracy": 0.647165060043335, + "num_tokens": 12371350.0, + "step": 479 + }, + { + "epoch": 0.052712497254557435, + "grad_norm": 2.7757580280303955, + "learning_rate": 2.6289791437980244e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6876219511032104, + "num_tokens": 12395825.0, + "step": 480 + }, + { + "epoch": 0.052822314957171095, + "grad_norm": 2.702286720275879, + "learning_rate": 2.634467618002195e-07, + "loss": 1.1612, + "mean_token_accuracy": 0.6623992919921875, + "num_tokens": 12425478.0, + "step": 481 + }, + { + "epoch": 0.052932132659784754, + "grad_norm": 2.7342350482940674, + "learning_rate": 2.639956092206366e-07, + "loss": 1.0385, + "mean_token_accuracy": 0.6936733722686768, + "num_tokens": 12452027.0, + "step": 482 + }, + { + "epoch": 0.05304195036239842, + "grad_norm": 2.458001136779785, + "learning_rate": 2.645444566410538e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6853656768798828, + "num_tokens": 12483706.0, + "step": 483 + }, + { + "epoch": 0.05315176806501208, + "grad_norm": 2.7410168647766113, + "learning_rate": 2.650933040614709e-07, + "loss": 1.2789, + "mean_token_accuracy": 0.6412801146507263, + "num_tokens": 12513325.0, + "step": 484 + }, + { + "epoch": 0.05326158576762574, + "grad_norm": 3.5500071048736572, + "learning_rate": 2.65642151481888e-07, + "loss": 1.1703, + "mean_token_accuracy": 0.6624087691307068, + "num_tokens": 12533625.0, + "step": 485 + }, + { + "epoch": 0.0533714034702394, + "grad_norm": 3.214707851409912, + "learning_rate": 2.6619099890230513e-07, + "loss": 1.2351, + "mean_token_accuracy": 0.6496695280075073, + "num_tokens": 12558466.0, + "step": 486 + }, + { + "epoch": 0.05348122117285306, + "grad_norm": 3.0128283500671387, + "learning_rate": 2.667398463227223e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6830190420150757, + "num_tokens": 12581048.0, + "step": 487 + }, + { + "epoch": 0.05359103887546673, + "grad_norm": 2.656106472015381, + "learning_rate": 2.672886937431394e-07, + "loss": 1.1206, + "mean_token_accuracy": 0.6781843304634094, + "num_tokens": 12610147.0, + "step": 488 + }, + { + "epoch": 0.05370085657808039, + "grad_norm": 2.9874415397644043, + "learning_rate": 2.678375411635565e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6777540445327759, + "num_tokens": 12634105.0, + "step": 489 + }, + { + "epoch": 0.05381067428069405, + "grad_norm": 2.9906208515167236, + "learning_rate": 2.6838638858397364e-07, + "loss": 1.0996, + "mean_token_accuracy": 0.6766005158424377, + "num_tokens": 12656627.0, + "step": 490 + }, + { + "epoch": 0.05392049198330771, + "grad_norm": 2.9700653553009033, + "learning_rate": 2.6893523600439076e-07, + "loss": 1.1449, + "mean_token_accuracy": 0.6706975698471069, + "num_tokens": 12681252.0, + "step": 491 + }, + { + "epoch": 0.054030309685921374, + "grad_norm": 3.518157482147217, + "learning_rate": 2.6948408342480787e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6721820831298828, + "num_tokens": 12704441.0, + "step": 492 + }, + { + "epoch": 0.054140127388535034, + "grad_norm": 2.704965114593506, + "learning_rate": 2.70032930845225e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.6815441846847534, + "num_tokens": 12731803.0, + "step": 493 + }, + { + "epoch": 0.054249945091148694, + "grad_norm": 2.926654100418091, + "learning_rate": 2.7058177826564215e-07, + "loss": 1.1859, + "mean_token_accuracy": 0.6641190052032471, + "num_tokens": 12758771.0, + "step": 494 + }, + { + "epoch": 0.05435976279376235, + "grad_norm": 3.3461036682128906, + "learning_rate": 2.7113062568605927e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.6768684983253479, + "num_tokens": 12781618.0, + "step": 495 + }, + { + "epoch": 0.05446958049637601, + "grad_norm": 2.790371894836426, + "learning_rate": 2.716794731064764e-07, + "loss": 1.1582, + "mean_token_accuracy": 0.668118953704834, + "num_tokens": 12809153.0, + "step": 496 + }, + { + "epoch": 0.05457939819898968, + "grad_norm": 2.9704442024230957, + "learning_rate": 2.722283205268935e-07, + "loss": 1.0645, + "mean_token_accuracy": 0.6906172037124634, + "num_tokens": 12830735.0, + "step": 497 + }, + { + "epoch": 0.05468921590160334, + "grad_norm": 2.6785085201263428, + "learning_rate": 2.7277716794731067e-07, + "loss": 1.2384, + "mean_token_accuracy": 0.6455684900283813, + "num_tokens": 12859477.0, + "step": 498 + }, + { + "epoch": 0.054799033604217, + "grad_norm": 2.6786441802978516, + "learning_rate": 2.733260153677278e-07, + "loss": 1.1822, + "mean_token_accuracy": 0.6591036319732666, + "num_tokens": 12888468.0, + "step": 499 + }, + { + "epoch": 0.05490885130683066, + "grad_norm": 2.896843671798706, + "learning_rate": 2.7387486278814485e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6752716898918152, + "num_tokens": 12911691.0, + "step": 500 + }, + { + "epoch": 0.05501866900944432, + "grad_norm": 2.6639323234558105, + "learning_rate": 2.74423710208562e-07, + "loss": 1.1164, + "mean_token_accuracy": 0.6747732162475586, + "num_tokens": 12936270.0, + "step": 501 + }, + { + "epoch": 0.055128486712057986, + "grad_norm": 2.4390556812286377, + "learning_rate": 2.7497255762897913e-07, + "loss": 1.2185, + "mean_token_accuracy": 0.6502364873886108, + "num_tokens": 12966906.0, + "step": 502 + }, + { + "epoch": 0.055238304414671646, + "grad_norm": 3.1848134994506836, + "learning_rate": 2.7552140504939624e-07, + "loss": 1.1684, + "mean_token_accuracy": 0.654639482498169, + "num_tokens": 12988019.0, + "step": 503 + }, + { + "epoch": 0.055348122117285306, + "grad_norm": 2.494819164276123, + "learning_rate": 2.7607025246981336e-07, + "loss": 1.1114, + "mean_token_accuracy": 0.6778311729431152, + "num_tokens": 13016080.0, + "step": 504 + }, + { + "epoch": 0.055457939819898966, + "grad_norm": 3.1881051063537598, + "learning_rate": 2.7661909989023053e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6975631713867188, + "num_tokens": 13037290.0, + "step": 505 + }, + { + "epoch": 0.055567757522512626, + "grad_norm": 2.7391064167022705, + "learning_rate": 2.7716794731064764e-07, + "loss": 1.1819, + "mean_token_accuracy": 0.6643884181976318, + "num_tokens": 13065039.0, + "step": 506 + }, + { + "epoch": 0.05567757522512629, + "grad_norm": 2.9645814895629883, + "learning_rate": 2.7771679473106476e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6829848885536194, + "num_tokens": 13088172.0, + "step": 507 + }, + { + "epoch": 0.05578739292773995, + "grad_norm": 2.9939045906066895, + "learning_rate": 2.7826564215148187e-07, + "loss": 1.1523, + "mean_token_accuracy": 0.6674472093582153, + "num_tokens": 13109554.0, + "step": 508 + }, + { + "epoch": 0.05589721063035361, + "grad_norm": 2.8582382202148438, + "learning_rate": 2.78814489571899e-07, + "loss": 1.0202, + "mean_token_accuracy": 0.6918796300888062, + "num_tokens": 13131015.0, + "step": 509 + }, + { + "epoch": 0.05600702833296727, + "grad_norm": 2.9520044326782227, + "learning_rate": 2.793633369923161e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6980718374252319, + "num_tokens": 13150815.0, + "step": 510 + }, + { + "epoch": 0.05611684603558094, + "grad_norm": 2.6719183921813965, + "learning_rate": 2.799121844127332e-07, + "loss": 1.1712, + "mean_token_accuracy": 0.6624862551689148, + "num_tokens": 13176263.0, + "step": 511 + }, + { + "epoch": 0.0562266637381946, + "grad_norm": 2.9971559047698975, + "learning_rate": 2.804610318331504e-07, + "loss": 1.1251, + "mean_token_accuracy": 0.6755516529083252, + "num_tokens": 13198459.0, + "step": 512 + }, + { + "epoch": 0.05633648144080826, + "grad_norm": 3.074840784072876, + "learning_rate": 2.810098792535675e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6924223899841309, + "num_tokens": 13219666.0, + "step": 513 + }, + { + "epoch": 0.05644629914342192, + "grad_norm": 3.6490325927734375, + "learning_rate": 2.815587266739846e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.6793038249015808, + "num_tokens": 13237420.0, + "step": 514 + }, + { + "epoch": 0.05655611684603558, + "grad_norm": 2.9455552101135254, + "learning_rate": 2.8210757409440173e-07, + "loss": 1.1052, + "mean_token_accuracy": 0.6776463389396667, + "num_tokens": 13261980.0, + "step": 515 + }, + { + "epoch": 0.056665934548649245, + "grad_norm": 2.620614767074585, + "learning_rate": 2.826564215148189e-07, + "loss": 1.1856, + "mean_token_accuracy": 0.6531254649162292, + "num_tokens": 13291204.0, + "step": 516 + }, + { + "epoch": 0.056775752251262905, + "grad_norm": 2.8178513050079346, + "learning_rate": 2.83205268935236e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.6955404281616211, + "num_tokens": 13313978.0, + "step": 517 + }, + { + "epoch": 0.056885569953876565, + "grad_norm": 2.4239654541015625, + "learning_rate": 2.8375411635565313e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.675957441329956, + "num_tokens": 13345374.0, + "step": 518 + }, + { + "epoch": 0.056995387656490225, + "grad_norm": 2.70637845993042, + "learning_rate": 2.843029637760702e-07, + "loss": 1.1911, + "mean_token_accuracy": 0.6584686040878296, + "num_tokens": 13372836.0, + "step": 519 + }, + { + "epoch": 0.057105205359103885, + "grad_norm": 2.6899681091308594, + "learning_rate": 2.8485181119648736e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6972014904022217, + "num_tokens": 13399225.0, + "step": 520 + }, + { + "epoch": 0.05721502306171755, + "grad_norm": 2.6313893795013428, + "learning_rate": 2.854006586169045e-07, + "loss": 1.1132, + "mean_token_accuracy": 0.678900957107544, + "num_tokens": 13425737.0, + "step": 521 + }, + { + "epoch": 0.05732484076433121, + "grad_norm": 2.426649570465088, + "learning_rate": 2.859495060373216e-07, + "loss": 1.208, + "mean_token_accuracy": 0.6591575145721436, + "num_tokens": 13455867.0, + "step": 522 + }, + { + "epoch": 0.05743465846694487, + "grad_norm": 2.658857822418213, + "learning_rate": 2.8649835345773876e-07, + "loss": 1.2045, + "mean_token_accuracy": 0.663352906703949, + "num_tokens": 13480346.0, + "step": 523 + }, + { + "epoch": 0.05754447616955853, + "grad_norm": 2.9578254222869873, + "learning_rate": 2.870472008781559e-07, + "loss": 1.1563, + "mean_token_accuracy": 0.660443902015686, + "num_tokens": 13502163.0, + "step": 524 + }, + { + "epoch": 0.05765429387217219, + "grad_norm": 2.7597787380218506, + "learning_rate": 2.87596048298573e-07, + "loss": 1.1808, + "mean_token_accuracy": 0.6544955968856812, + "num_tokens": 13527246.0, + "step": 525 + }, + { + "epoch": 0.05776411157478586, + "grad_norm": 2.988588809967041, + "learning_rate": 2.881448957189901e-07, + "loss": 1.1036, + "mean_token_accuracy": 0.6659939885139465, + "num_tokens": 13549764.0, + "step": 526 + }, + { + "epoch": 0.05787392927739952, + "grad_norm": 2.7976701259613037, + "learning_rate": 2.886937431394073e-07, + "loss": 1.134, + "mean_token_accuracy": 0.6658077239990234, + "num_tokens": 13572960.0, + "step": 527 + }, + { + "epoch": 0.05798374698001318, + "grad_norm": 2.578723430633545, + "learning_rate": 2.8924259055982434e-07, + "loss": 1.1715, + "mean_token_accuracy": 0.6720499396324158, + "num_tokens": 13599560.0, + "step": 528 + }, + { + "epoch": 0.05809356468262684, + "grad_norm": 3.015573740005493, + "learning_rate": 2.8979143798024145e-07, + "loss": 1.1277, + "mean_token_accuracy": 0.669548511505127, + "num_tokens": 13619816.0, + "step": 529 + }, + { + "epoch": 0.058203382385240504, + "grad_norm": 2.766928195953369, + "learning_rate": 2.9034028540065857e-07, + "loss": 1.1134, + "mean_token_accuracy": 0.6713212132453918, + "num_tokens": 13644347.0, + "step": 530 + }, + { + "epoch": 0.058313200087854164, + "grad_norm": 2.7277393341064453, + "learning_rate": 2.9088913282107574e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6786156296730042, + "num_tokens": 13671113.0, + "step": 531 + }, + { + "epoch": 0.058423017790467824, + "grad_norm": 2.623450517654419, + "learning_rate": 2.9143798024149285e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.6740940809249878, + "num_tokens": 13697555.0, + "step": 532 + }, + { + "epoch": 0.058532835493081484, + "grad_norm": 2.6168885231018066, + "learning_rate": 2.9198682766190997e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.6860982775688171, + "num_tokens": 13723707.0, + "step": 533 + }, + { + "epoch": 0.058642653195695144, + "grad_norm": 3.0028793811798096, + "learning_rate": 2.9253567508232713e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.7041282653808594, + "num_tokens": 13743334.0, + "step": 534 + }, + { + "epoch": 0.05875247089830881, + "grad_norm": 3.4743330478668213, + "learning_rate": 2.9308452250274425e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6702497005462646, + "num_tokens": 13761715.0, + "step": 535 + }, + { + "epoch": 0.05886228860092247, + "grad_norm": 2.807854413986206, + "learning_rate": 2.9363336992316136e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6805610656738281, + "num_tokens": 13786050.0, + "step": 536 + }, + { + "epoch": 0.05897210630353613, + "grad_norm": 2.724823236465454, + "learning_rate": 2.941822173435785e-07, + "loss": 1.087, + "mean_token_accuracy": 0.6788039207458496, + "num_tokens": 13810173.0, + "step": 537 + }, + { + "epoch": 0.05908192400614979, + "grad_norm": 2.358961582183838, + "learning_rate": 2.947310647639956e-07, + "loss": 1.1105, + "mean_token_accuracy": 0.6812998056411743, + "num_tokens": 13842179.0, + "step": 538 + }, + { + "epoch": 0.05919174170876345, + "grad_norm": 2.284176826477051, + "learning_rate": 2.952799121844127e-07, + "loss": 1.1766, + "mean_token_accuracy": 0.655646026134491, + "num_tokens": 13873179.0, + "step": 539 + }, + { + "epoch": 0.05930155941137712, + "grad_norm": 2.513941526412964, + "learning_rate": 2.958287596048298e-07, + "loss": 1.2053, + "mean_token_accuracy": 0.6532779335975647, + "num_tokens": 13902219.0, + "step": 540 + }, + { + "epoch": 0.059411377113990776, + "grad_norm": 2.8273403644561768, + "learning_rate": 2.9637760702524694e-07, + "loss": 1.1835, + "mean_token_accuracy": 0.6610467433929443, + "num_tokens": 13925919.0, + "step": 541 + }, + { + "epoch": 0.059521194816604436, + "grad_norm": 2.4626214504241943, + "learning_rate": 2.969264544456641e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6830787658691406, + "num_tokens": 13953041.0, + "step": 542 + }, + { + "epoch": 0.059631012519218096, + "grad_norm": 2.54952073097229, + "learning_rate": 2.974753018660812e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.6821843981742859, + "num_tokens": 13979639.0, + "step": 543 + }, + { + "epoch": 0.059740830221831756, + "grad_norm": 2.454834222793579, + "learning_rate": 2.9802414928649834e-07, + "loss": 1.1785, + "mean_token_accuracy": 0.6605499982833862, + "num_tokens": 14009368.0, + "step": 544 + }, + { + "epoch": 0.05985064792444542, + "grad_norm": 2.64323353767395, + "learning_rate": 2.9857299670691546e-07, + "loss": 1.1847, + "mean_token_accuracy": 0.6575539112091064, + "num_tokens": 14035056.0, + "step": 545 + }, + { + "epoch": 0.05996046562705908, + "grad_norm": 2.5364625453948975, + "learning_rate": 2.991218441273326e-07, + "loss": 1.1011, + "mean_token_accuracy": 0.6755093336105347, + "num_tokens": 14064325.0, + "step": 546 + }, + { + "epoch": 0.06007028332967274, + "grad_norm": 2.4501709938049316, + "learning_rate": 2.996706915477497e-07, + "loss": 1.1167, + "mean_token_accuracy": 0.6745128035545349, + "num_tokens": 14093358.0, + "step": 547 + }, + { + "epoch": 0.0601801010322864, + "grad_norm": 2.2367842197418213, + "learning_rate": 3.002195389681668e-07, + "loss": 1.2449, + "mean_token_accuracy": 0.6391915082931519, + "num_tokens": 14130425.0, + "step": 548 + }, + { + "epoch": 0.06028991873490007, + "grad_norm": 2.458012580871582, + "learning_rate": 3.0076838638858397e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6749885082244873, + "num_tokens": 14158777.0, + "step": 549 + }, + { + "epoch": 0.06039973643751373, + "grad_norm": 2.572986364364624, + "learning_rate": 3.013172338090011e-07, + "loss": 1.148, + "mean_token_accuracy": 0.6670438051223755, + "num_tokens": 14188161.0, + "step": 550 + }, + { + "epoch": 0.06050955414012739, + "grad_norm": 2.490800619125366, + "learning_rate": 3.018660812294182e-07, + "loss": 1.0783, + "mean_token_accuracy": 0.6819040179252625, + "num_tokens": 14216786.0, + "step": 551 + }, + { + "epoch": 0.06061937184274105, + "grad_norm": 2.5970709323883057, + "learning_rate": 3.024149286498353e-07, + "loss": 1.1809, + "mean_token_accuracy": 0.6579287052154541, + "num_tokens": 14242188.0, + "step": 552 + }, + { + "epoch": 0.06072918954535471, + "grad_norm": 2.4727790355682373, + "learning_rate": 3.029637760702525e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6825175285339355, + "num_tokens": 14273292.0, + "step": 553 + }, + { + "epoch": 0.060839007247968376, + "grad_norm": 2.3596832752227783, + "learning_rate": 3.035126234906696e-07, + "loss": 1.2275, + "mean_token_accuracy": 0.6506582498550415, + "num_tokens": 14306616.0, + "step": 554 + }, + { + "epoch": 0.060948824950582035, + "grad_norm": 2.7054736614227295, + "learning_rate": 3.040614709110867e-07, + "loss": 1.1601, + "mean_token_accuracy": 0.6639229655265808, + "num_tokens": 14332232.0, + "step": 555 + }, + { + "epoch": 0.061058642653195695, + "grad_norm": 2.9967598915100098, + "learning_rate": 3.046103183315038e-07, + "loss": 1.1537, + "mean_token_accuracy": 0.6705601811408997, + "num_tokens": 14352217.0, + "step": 556 + }, + { + "epoch": 0.061168460355809355, + "grad_norm": 2.726499319076538, + "learning_rate": 3.0515916575192094e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.6776540875434875, + "num_tokens": 14375174.0, + "step": 557 + }, + { + "epoch": 0.061278278058423015, + "grad_norm": 2.5130467414855957, + "learning_rate": 3.0570801317233806e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6742203235626221, + "num_tokens": 14404315.0, + "step": 558 + }, + { + "epoch": 0.06138809576103668, + "grad_norm": 2.433320999145508, + "learning_rate": 3.062568605927552e-07, + "loss": 1.0592, + "mean_token_accuracy": 0.6887427568435669, + "num_tokens": 14432028.0, + "step": 559 + }, + { + "epoch": 0.06149791346365034, + "grad_norm": 3.0641257762908936, + "learning_rate": 3.0680570801317234e-07, + "loss": 1.125, + "mean_token_accuracy": 0.671528697013855, + "num_tokens": 14456453.0, + "step": 560 + }, + { + "epoch": 0.061607731166264, + "grad_norm": 2.6266567707061768, + "learning_rate": 3.0735455543358946e-07, + "loss": 1.2008, + "mean_token_accuracy": 0.6664726734161377, + "num_tokens": 14481348.0, + "step": 561 + }, + { + "epoch": 0.06171754886887766, + "grad_norm": 2.1485607624053955, + "learning_rate": 3.0790340285400657e-07, + "loss": 1.0957, + "mean_token_accuracy": 0.6820945739746094, + "num_tokens": 14518176.0, + "step": 562 + }, + { + "epoch": 0.06182736657149132, + "grad_norm": 2.7410948276519775, + "learning_rate": 3.084522502744237e-07, + "loss": 1.0919, + "mean_token_accuracy": 0.6787691116333008, + "num_tokens": 14541407.0, + "step": 563 + }, + { + "epoch": 0.06193718427410499, + "grad_norm": 2.227301836013794, + "learning_rate": 3.0900109769484086e-07, + "loss": 1.2433, + "mean_token_accuracy": 0.642369270324707, + "num_tokens": 14576608.0, + "step": 564 + }, + { + "epoch": 0.06204700197671865, + "grad_norm": 2.6828601360321045, + "learning_rate": 3.0954994511525797e-07, + "loss": 1.083, + "mean_token_accuracy": 0.6742928624153137, + "num_tokens": 14601850.0, + "step": 565 + }, + { + "epoch": 0.06215681967933231, + "grad_norm": 4.675853729248047, + "learning_rate": 3.1009879253567503e-07, + "loss": 1.1514, + "mean_token_accuracy": 0.6675216555595398, + "num_tokens": 14624031.0, + "step": 566 + }, + { + "epoch": 0.06226663738194597, + "grad_norm": 2.1791181564331055, + "learning_rate": 3.1064763995609215e-07, + "loss": 1.1616, + "mean_token_accuracy": 0.6684722900390625, + "num_tokens": 14659898.0, + "step": 567 + }, + { + "epoch": 0.062376455084559634, + "grad_norm": 2.2837934494018555, + "learning_rate": 3.111964873765093e-07, + "loss": 1.1546, + "mean_token_accuracy": 0.6650289297103882, + "num_tokens": 14690229.0, + "step": 568 + }, + { + "epoch": 0.062486272787173294, + "grad_norm": 2.7681455612182617, + "learning_rate": 3.1174533479692643e-07, + "loss": 1.1004, + "mean_token_accuracy": 0.6809527277946472, + "num_tokens": 14713064.0, + "step": 569 + }, + { + "epoch": 0.06259609048978695, + "grad_norm": 2.21889066696167, + "learning_rate": 3.1229418221734355e-07, + "loss": 1.219, + "mean_token_accuracy": 0.652373731136322, + "num_tokens": 14747466.0, + "step": 570 + }, + { + "epoch": 0.06270590819240061, + "grad_norm": 3.109558343887329, + "learning_rate": 3.128430296377607e-07, + "loss": 1.0309, + "mean_token_accuracy": 0.6965484619140625, + "num_tokens": 14767375.0, + "step": 571 + }, + { + "epoch": 0.06281572589501427, + "grad_norm": 2.7907028198242188, + "learning_rate": 3.1339187705817783e-07, + "loss": 1.1548, + "mean_token_accuracy": 0.6531558036804199, + "num_tokens": 14793092.0, + "step": 572 + }, + { + "epoch": 0.06292554359762793, + "grad_norm": 2.754979372024536, + "learning_rate": 3.1394072447859495e-07, + "loss": 1.0798, + "mean_token_accuracy": 0.6818923354148865, + "num_tokens": 14816435.0, + "step": 573 + }, + { + "epoch": 0.0630353613002416, + "grad_norm": 2.5153255462646484, + "learning_rate": 3.1448957189901206e-07, + "loss": 1.1127, + "mean_token_accuracy": 0.6786981821060181, + "num_tokens": 14845409.0, + "step": 574 + }, + { + "epoch": 0.06314517900285525, + "grad_norm": 2.522732973098755, + "learning_rate": 3.1503841931942923e-07, + "loss": 1.1332, + "mean_token_accuracy": 0.6705083250999451, + "num_tokens": 14874379.0, + "step": 575 + }, + { + "epoch": 0.06325499670546893, + "grad_norm": 2.857145071029663, + "learning_rate": 3.155872667398463e-07, + "loss": 1.0996, + "mean_token_accuracy": 0.6786448955535889, + "num_tokens": 14896881.0, + "step": 576 + }, + { + "epoch": 0.06336481440808259, + "grad_norm": 2.1736221313476562, + "learning_rate": 3.161361141602634e-07, + "loss": 1.2995, + "mean_token_accuracy": 0.628174901008606, + "num_tokens": 14933398.0, + "step": 577 + }, + { + "epoch": 0.06347463211069625, + "grad_norm": 2.3393540382385254, + "learning_rate": 3.166849615806805e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.6757001280784607, + "num_tokens": 14963546.0, + "step": 578 + }, + { + "epoch": 0.0635844498133099, + "grad_norm": 2.5676469802856445, + "learning_rate": 3.172338090010977e-07, + "loss": 1.1827, + "mean_token_accuracy": 0.6622681617736816, + "num_tokens": 14991119.0, + "step": 579 + }, + { + "epoch": 0.06369426751592357, + "grad_norm": 2.6492879390716553, + "learning_rate": 3.177826564215148e-07, + "loss": 1.0871, + "mean_token_accuracy": 0.6849368810653687, + "num_tokens": 15017337.0, + "step": 580 + }, + { + "epoch": 0.06380408521853723, + "grad_norm": 2.788792133331299, + "learning_rate": 3.183315038419319e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6785661578178406, + "num_tokens": 15039693.0, + "step": 581 + }, + { + "epoch": 0.06391390292115089, + "grad_norm": 2.6006181240081787, + "learning_rate": 3.188803512623491e-07, + "loss": 1.2269, + "mean_token_accuracy": 0.6422544121742249, + "num_tokens": 15066340.0, + "step": 582 + }, + { + "epoch": 0.06402372062376455, + "grad_norm": 2.661752462387085, + "learning_rate": 3.194291986827662e-07, + "loss": 1.12, + "mean_token_accuracy": 0.6733411550521851, + "num_tokens": 15089896.0, + "step": 583 + }, + { + "epoch": 0.0641335383263782, + "grad_norm": 3.0014636516571045, + "learning_rate": 3.199780461031833e-07, + "loss": 1.1608, + "mean_token_accuracy": 0.6748294234275818, + "num_tokens": 15112101.0, + "step": 584 + }, + { + "epoch": 0.06424335602899188, + "grad_norm": 2.630885124206543, + "learning_rate": 3.205268935236004e-07, + "loss": 1.2141, + "mean_token_accuracy": 0.6439672708511353, + "num_tokens": 15139008.0, + "step": 585 + }, + { + "epoch": 0.06435317373160554, + "grad_norm": 3.0449635982513428, + "learning_rate": 3.2107574094401755e-07, + "loss": 1.1662, + "mean_token_accuracy": 0.6664577126502991, + "num_tokens": 15158929.0, + "step": 586 + }, + { + "epoch": 0.0644629914342192, + "grad_norm": 2.460238456726074, + "learning_rate": 3.2162458836443467e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.7025641202926636, + "num_tokens": 15185785.0, + "step": 587 + }, + { + "epoch": 0.06457280913683286, + "grad_norm": 2.5483827590942383, + "learning_rate": 3.221734357848518e-07, + "loss": 1.19, + "mean_token_accuracy": 0.6655205488204956, + "num_tokens": 15213731.0, + "step": 588 + }, + { + "epoch": 0.06468262683944652, + "grad_norm": 2.6237900257110596, + "learning_rate": 3.227222832052689e-07, + "loss": 1.1261, + "mean_token_accuracy": 0.6756885051727295, + "num_tokens": 15239928.0, + "step": 589 + }, + { + "epoch": 0.06479244454206018, + "grad_norm": 2.2704927921295166, + "learning_rate": 3.2327113062568606e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6911592483520508, + "num_tokens": 15270539.0, + "step": 590 + }, + { + "epoch": 0.06490226224467384, + "grad_norm": 2.895005226135254, + "learning_rate": 3.238199780461032e-07, + "loss": 1.0888, + "mean_token_accuracy": 0.68630450963974, + "num_tokens": 15291369.0, + "step": 591 + }, + { + "epoch": 0.0650120799472875, + "grad_norm": 2.5169432163238525, + "learning_rate": 3.243688254665203e-07, + "loss": 1.1062, + "mean_token_accuracy": 0.6769217252731323, + "num_tokens": 15318771.0, + "step": 592 + }, + { + "epoch": 0.06512189764990116, + "grad_norm": 3.042202949523926, + "learning_rate": 3.2491767288693746e-07, + "loss": 0.9208, + "mean_token_accuracy": 0.7210530638694763, + "num_tokens": 15336548.0, + "step": 593 + }, + { + "epoch": 0.06523171535251482, + "grad_norm": 2.461148977279663, + "learning_rate": 3.254665203073546e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.7036374807357788, + "num_tokens": 15363054.0, + "step": 594 + }, + { + "epoch": 0.06534153305512849, + "grad_norm": 2.6756975650787354, + "learning_rate": 3.2601536772777164e-07, + "loss": 1.1089, + "mean_token_accuracy": 0.6800655126571655, + "num_tokens": 15386868.0, + "step": 595 + }, + { + "epoch": 0.06545135075774215, + "grad_norm": 2.412820816040039, + "learning_rate": 3.2656421514818876e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.691008448600769, + "num_tokens": 15419083.0, + "step": 596 + }, + { + "epoch": 0.06556116846035581, + "grad_norm": 2.376620054244995, + "learning_rate": 3.271130625686059e-07, + "loss": 1.1229, + "mean_token_accuracy": 0.6777480244636536, + "num_tokens": 15446655.0, + "step": 597 + }, + { + "epoch": 0.06567098616296947, + "grad_norm": 2.3408257961273193, + "learning_rate": 3.2766190998902304e-07, + "loss": 1.1119, + "mean_token_accuracy": 0.6771740317344666, + "num_tokens": 15477418.0, + "step": 598 + }, + { + "epoch": 0.06578080386558313, + "grad_norm": 2.7868103981018066, + "learning_rate": 3.2821075740944015e-07, + "loss": 1.1375, + "mean_token_accuracy": 0.6748260259628296, + "num_tokens": 15499562.0, + "step": 599 + }, + { + "epoch": 0.06589062156819679, + "grad_norm": 2.783717632293701, + "learning_rate": 3.2875960482985727e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6810539364814758, + "num_tokens": 15521518.0, + "step": 600 + }, + { + "epoch": 0.06600043927081045, + "grad_norm": 2.363006353378296, + "learning_rate": 3.2930845225027444e-07, + "loss": 1.2164, + "mean_token_accuracy": 0.655285120010376, + "num_tokens": 15552337.0, + "step": 601 + }, + { + "epoch": 0.06611025697342411, + "grad_norm": 2.522510290145874, + "learning_rate": 3.2985729967069155e-07, + "loss": 1.0356, + "mean_token_accuracy": 0.6981571316719055, + "num_tokens": 15577248.0, + "step": 602 + }, + { + "epoch": 0.06622007467603777, + "grad_norm": 2.6146795749664307, + "learning_rate": 3.3040614709110867e-07, + "loss": 1.1323, + "mean_token_accuracy": 0.6723389625549316, + "num_tokens": 15601561.0, + "step": 603 + }, + { + "epoch": 0.06632989237865144, + "grad_norm": 2.188927412033081, + "learning_rate": 3.3095499451152573e-07, + "loss": 1.1159, + "mean_token_accuracy": 0.6757055521011353, + "num_tokens": 15634421.0, + "step": 604 + }, + { + "epoch": 0.0664397100812651, + "grad_norm": 2.8540990352630615, + "learning_rate": 3.315038419319429e-07, + "loss": 1.0753, + "mean_token_accuracy": 0.6799106597900391, + "num_tokens": 15654052.0, + "step": 605 + }, + { + "epoch": 0.06654952778387876, + "grad_norm": 2.4670000076293945, + "learning_rate": 3.3205268935236e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6975417137145996, + "num_tokens": 15678121.0, + "step": 606 + }, + { + "epoch": 0.06665934548649242, + "grad_norm": 2.3212289810180664, + "learning_rate": 3.3260153677277713e-07, + "loss": 1.1715, + "mean_token_accuracy": 0.6545129418373108, + "num_tokens": 15708238.0, + "step": 607 + }, + { + "epoch": 0.06676916318910608, + "grad_norm": 2.7403573989868164, + "learning_rate": 3.331503841931943e-07, + "loss": 1.1808, + "mean_token_accuracy": 0.6618822813034058, + "num_tokens": 15730572.0, + "step": 608 + }, + { + "epoch": 0.06687898089171974, + "grad_norm": 3.0717546939849854, + "learning_rate": 3.336992316136114e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6870524883270264, + "num_tokens": 15754898.0, + "step": 609 + }, + { + "epoch": 0.0669887985943334, + "grad_norm": 2.236290216445923, + "learning_rate": 3.3424807903402853e-07, + "loss": 1.1333, + "mean_token_accuracy": 0.6651558876037598, + "num_tokens": 15786096.0, + "step": 610 + }, + { + "epoch": 0.06709861629694706, + "grad_norm": 2.481175184249878, + "learning_rate": 3.3479692645444564e-07, + "loss": 1.0616, + "mean_token_accuracy": 0.695199728012085, + "num_tokens": 15812085.0, + "step": 611 + }, + { + "epoch": 0.06720843399956072, + "grad_norm": 2.6692845821380615, + "learning_rate": 3.353457738748628e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6914280652999878, + "num_tokens": 15834887.0, + "step": 612 + }, + { + "epoch": 0.06731825170217438, + "grad_norm": 2.6600418090820312, + "learning_rate": 3.3589462129527993e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6766864061355591, + "num_tokens": 15859917.0, + "step": 613 + }, + { + "epoch": 0.06742806940478806, + "grad_norm": 2.298762321472168, + "learning_rate": 3.36443468715697e-07, + "loss": 1.0854, + "mean_token_accuracy": 0.6932502388954163, + "num_tokens": 15890156.0, + "step": 614 + }, + { + "epoch": 0.06753788710740172, + "grad_norm": 2.8593807220458984, + "learning_rate": 3.369923161361141e-07, + "loss": 1.1254, + "mean_token_accuracy": 0.6802685260772705, + "num_tokens": 15910970.0, + "step": 615 + }, + { + "epoch": 0.06764770481001538, + "grad_norm": 2.490877389907837, + "learning_rate": 3.3754116355653127e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6897948980331421, + "num_tokens": 15935683.0, + "step": 616 + }, + { + "epoch": 0.06775752251262904, + "grad_norm": 2.7084224224090576, + "learning_rate": 3.380900109769484e-07, + "loss": 1.1667, + "mean_token_accuracy": 0.6619495153427124, + "num_tokens": 15958587.0, + "step": 617 + }, + { + "epoch": 0.0678673402152427, + "grad_norm": 2.572920322418213, + "learning_rate": 3.386388583973655e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.6738165020942688, + "num_tokens": 15982556.0, + "step": 618 + }, + { + "epoch": 0.06797715791785636, + "grad_norm": 2.2250096797943115, + "learning_rate": 3.3918770581778267e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6663331985473633, + "num_tokens": 16015180.0, + "step": 619 + }, + { + "epoch": 0.06808697562047002, + "grad_norm": 2.590723752975464, + "learning_rate": 3.397365532381998e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.6803073287010193, + "num_tokens": 16041947.0, + "step": 620 + }, + { + "epoch": 0.06819679332308368, + "grad_norm": 2.4779751300811768, + "learning_rate": 3.402854006586169e-07, + "loss": 1.1787, + "mean_token_accuracy": 0.6521122455596924, + "num_tokens": 16070542.0, + "step": 621 + }, + { + "epoch": 0.06830661102569734, + "grad_norm": 2.6307153701782227, + "learning_rate": 3.40834248079034e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6646641492843628, + "num_tokens": 16094643.0, + "step": 622 + }, + { + "epoch": 0.06841642872831101, + "grad_norm": 2.6761062145233154, + "learning_rate": 3.4138309549945113e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6836628913879395, + "num_tokens": 16117554.0, + "step": 623 + }, + { + "epoch": 0.06852624643092467, + "grad_norm": 2.786137819290161, + "learning_rate": 3.4193194291986825e-07, + "loss": 1.1143, + "mean_token_accuracy": 0.6828414797782898, + "num_tokens": 16139642.0, + "step": 624 + }, + { + "epoch": 0.06863606413353833, + "grad_norm": 2.8323774337768555, + "learning_rate": 3.4248079034028536e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6898580193519592, + "num_tokens": 16161654.0, + "step": 625 + }, + { + "epoch": 0.06874588183615199, + "grad_norm": 2.5455663204193115, + "learning_rate": 3.430296377607025e-07, + "loss": 1.1109, + "mean_token_accuracy": 0.6776560544967651, + "num_tokens": 16186684.0, + "step": 626 + }, + { + "epoch": 0.06885569953876565, + "grad_norm": 2.5620980262756348, + "learning_rate": 3.4357848518111965e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6877738237380981, + "num_tokens": 16211499.0, + "step": 627 + }, + { + "epoch": 0.06896551724137931, + "grad_norm": 2.8048224449157715, + "learning_rate": 3.4412733260153676e-07, + "loss": 1.1097, + "mean_token_accuracy": 0.680424690246582, + "num_tokens": 16233210.0, + "step": 628 + }, + { + "epoch": 0.06907533494399297, + "grad_norm": 2.1356306076049805, + "learning_rate": 3.446761800219539e-07, + "loss": 1.101, + "mean_token_accuracy": 0.6881267428398132, + "num_tokens": 16264417.0, + "step": 629 + }, + { + "epoch": 0.06918515264660663, + "grad_norm": 2.785360813140869, + "learning_rate": 3.4522502744237104e-07, + "loss": 1.0315, + "mean_token_accuracy": 0.7026088237762451, + "num_tokens": 16285744.0, + "step": 630 + }, + { + "epoch": 0.06929497034922029, + "grad_norm": 2.109966278076172, + "learning_rate": 3.4577387486278816e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.675920844078064, + "num_tokens": 16320284.0, + "step": 631 + }, + { + "epoch": 0.06940478805183395, + "grad_norm": 2.356943368911743, + "learning_rate": 3.463227222832053e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6768146753311157, + "num_tokens": 16348361.0, + "step": 632 + }, + { + "epoch": 0.06951460575444762, + "grad_norm": 2.4139108657836914, + "learning_rate": 3.4687156970362234e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.6924378871917725, + "num_tokens": 16373904.0, + "step": 633 + }, + { + "epoch": 0.06962442345706128, + "grad_norm": 2.945729970932007, + "learning_rate": 3.474204171240395e-07, + "loss": 1.2433, + "mean_token_accuracy": 0.6612817049026489, + "num_tokens": 16394684.0, + "step": 634 + }, + { + "epoch": 0.06973424115967494, + "grad_norm": 2.767753839492798, + "learning_rate": 3.479692645444566e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.6874748468399048, + "num_tokens": 16418808.0, + "step": 635 + }, + { + "epoch": 0.0698440588622886, + "grad_norm": 2.519524335861206, + "learning_rate": 3.4851811196487374e-07, + "loss": 1.1845, + "mean_token_accuracy": 0.6595441102981567, + "num_tokens": 16444212.0, + "step": 636 + }, + { + "epoch": 0.06995387656490226, + "grad_norm": 2.7939696311950684, + "learning_rate": 3.4906695938529085e-07, + "loss": 1.1679, + "mean_token_accuracy": 0.6606355309486389, + "num_tokens": 16468338.0, + "step": 637 + }, + { + "epoch": 0.07006369426751592, + "grad_norm": 2.2984516620635986, + "learning_rate": 3.49615806805708e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6919922828674316, + "num_tokens": 16495959.0, + "step": 638 + }, + { + "epoch": 0.07017351197012958, + "grad_norm": 2.8231241703033447, + "learning_rate": 3.5016465422612513e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6932396292686462, + "num_tokens": 16516870.0, + "step": 639 + }, + { + "epoch": 0.07028332967274324, + "grad_norm": 3.085820436477661, + "learning_rate": 3.5071350164654225e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.7016140222549438, + "num_tokens": 16536933.0, + "step": 640 + }, + { + "epoch": 0.0703931473753569, + "grad_norm": 2.247053861618042, + "learning_rate": 3.512623490669594e-07, + "loss": 1.1877, + "mean_token_accuracy": 0.6555886268615723, + "num_tokens": 16566358.0, + "step": 641 + }, + { + "epoch": 0.07050296507797058, + "grad_norm": 2.70717716217041, + "learning_rate": 3.518111964873765e-07, + "loss": 1.0526, + "mean_token_accuracy": 0.6846743226051331, + "num_tokens": 16587452.0, + "step": 642 + }, + { + "epoch": 0.07061278278058424, + "grad_norm": 3.005237102508545, + "learning_rate": 3.523600439077936e-07, + "loss": 1.0879, + "mean_token_accuracy": 0.6829128861427307, + "num_tokens": 16606828.0, + "step": 643 + }, + { + "epoch": 0.0707226004831979, + "grad_norm": 2.429506778717041, + "learning_rate": 3.529088913282107e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.669666588306427, + "num_tokens": 16632416.0, + "step": 644 + }, + { + "epoch": 0.07083241818581155, + "grad_norm": 2.5381157398223877, + "learning_rate": 3.534577387486279e-07, + "loss": 1.1195, + "mean_token_accuracy": 0.6753729581832886, + "num_tokens": 16656277.0, + "step": 645 + }, + { + "epoch": 0.07094223588842521, + "grad_norm": 2.269970655441284, + "learning_rate": 3.54006586169045e-07, + "loss": 1.179, + "mean_token_accuracy": 0.6561158895492554, + "num_tokens": 16685941.0, + "step": 646 + }, + { + "epoch": 0.07105205359103887, + "grad_norm": 2.410266399383545, + "learning_rate": 3.545554335894621e-07, + "loss": 1.2042, + "mean_token_accuracy": 0.6594665050506592, + "num_tokens": 16715612.0, + "step": 647 + }, + { + "epoch": 0.07116187129365253, + "grad_norm": 2.6361889839172363, + "learning_rate": 3.551042810098792e-07, + "loss": 1.1803, + "mean_token_accuracy": 0.669276773929596, + "num_tokens": 16739610.0, + "step": 648 + }, + { + "epoch": 0.0712716889962662, + "grad_norm": 2.568850517272949, + "learning_rate": 3.556531284302964e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.701014518737793, + "num_tokens": 16763340.0, + "step": 649 + }, + { + "epoch": 0.07138150669887985, + "grad_norm": 2.479888439178467, + "learning_rate": 3.562019758507135e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.684833288192749, + "num_tokens": 16788540.0, + "step": 650 + }, + { + "epoch": 0.07149132440149351, + "grad_norm": 2.8973209857940674, + "learning_rate": 3.5675082327113057e-07, + "loss": 1.1363, + "mean_token_accuracy": 0.6748808026313782, + "num_tokens": 16808593.0, + "step": 651 + }, + { + "epoch": 0.07160114210410719, + "grad_norm": 2.314976930618286, + "learning_rate": 3.5729967069154774e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.6846255660057068, + "num_tokens": 16837331.0, + "step": 652 + }, + { + "epoch": 0.07171095980672085, + "grad_norm": 2.395922899246216, + "learning_rate": 3.5784851811196485e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6788402199745178, + "num_tokens": 16865806.0, + "step": 653 + }, + { + "epoch": 0.07182077750933451, + "grad_norm": 2.5635969638824463, + "learning_rate": 3.5839736553238197e-07, + "loss": 1.1305, + "mean_token_accuracy": 0.6683660745620728, + "num_tokens": 16891610.0, + "step": 654 + }, + { + "epoch": 0.07193059521194817, + "grad_norm": 2.797579050064087, + "learning_rate": 3.589462129527991e-07, + "loss": 1.0273, + "mean_token_accuracy": 0.6925585269927979, + "num_tokens": 16913353.0, + "step": 655 + }, + { + "epoch": 0.07204041291456183, + "grad_norm": 2.5281319618225098, + "learning_rate": 3.5949506037321625e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6821178197860718, + "num_tokens": 16939042.0, + "step": 656 + }, + { + "epoch": 0.07215023061717549, + "grad_norm": 2.6078641414642334, + "learning_rate": 3.6004390779363337e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6545414924621582, + "num_tokens": 16962682.0, + "step": 657 + }, + { + "epoch": 0.07226004831978915, + "grad_norm": 2.2821645736694336, + "learning_rate": 3.605927552140505e-07, + "loss": 1.1685, + "mean_token_accuracy": 0.6645389795303345, + "num_tokens": 16992542.0, + "step": 658 + }, + { + "epoch": 0.0723698660224028, + "grad_norm": 2.6677443981170654, + "learning_rate": 3.611416026344676e-07, + "loss": 1.1243, + "mean_token_accuracy": 0.6702710390090942, + "num_tokens": 17014864.0, + "step": 659 + }, + { + "epoch": 0.07247968372501647, + "grad_norm": 2.5213255882263184, + "learning_rate": 3.6169045005488477e-07, + "loss": 1.1095, + "mean_token_accuracy": 0.6795868873596191, + "num_tokens": 17039718.0, + "step": 660 + }, + { + "epoch": 0.07258950142763014, + "grad_norm": 2.348700523376465, + "learning_rate": 3.6223929747530183e-07, + "loss": 1.0967, + "mean_token_accuracy": 0.6853880882263184, + "num_tokens": 17069806.0, + "step": 661 + }, + { + "epoch": 0.0726993191302438, + "grad_norm": 2.3801751136779785, + "learning_rate": 3.6278814489571894e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6728681325912476, + "num_tokens": 17096962.0, + "step": 662 + }, + { + "epoch": 0.07280913683285746, + "grad_norm": 2.3791682720184326, + "learning_rate": 3.6333699231613606e-07, + "loss": 1.1708, + "mean_token_accuracy": 0.657443106174469, + "num_tokens": 17126320.0, + "step": 663 + }, + { + "epoch": 0.07291895453547112, + "grad_norm": 2.4256057739257812, + "learning_rate": 3.6388583973655323e-07, + "loss": 1.1534, + "mean_token_accuracy": 0.6621137857437134, + "num_tokens": 17153479.0, + "step": 664 + }, + { + "epoch": 0.07302877223808478, + "grad_norm": 2.395106315612793, + "learning_rate": 3.6443468715697034e-07, + "loss": 1.0877, + "mean_token_accuracy": 0.6822353005409241, + "num_tokens": 17181187.0, + "step": 665 + }, + { + "epoch": 0.07313858994069844, + "grad_norm": 2.34136962890625, + "learning_rate": 3.6498353457738746e-07, + "loss": 1.0916, + "mean_token_accuracy": 0.6792066097259521, + "num_tokens": 17207937.0, + "step": 666 + }, + { + "epoch": 0.0732484076433121, + "grad_norm": 2.421027660369873, + "learning_rate": 3.655323819978046e-07, + "loss": 1.0654, + "mean_token_accuracy": 0.6816666126251221, + "num_tokens": 17233420.0, + "step": 667 + }, + { + "epoch": 0.07335822534592576, + "grad_norm": 2.630661964416504, + "learning_rate": 3.6608122941822174e-07, + "loss": 1.1702, + "mean_token_accuracy": 0.6716867685317993, + "num_tokens": 17257463.0, + "step": 668 + }, + { + "epoch": 0.07346804304853942, + "grad_norm": 2.720789909362793, + "learning_rate": 3.6663007683863886e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.7029726505279541, + "num_tokens": 17278214.0, + "step": 669 + }, + { + "epoch": 0.07357786075115308, + "grad_norm": 2.3844246864318848, + "learning_rate": 3.671789242590559e-07, + "loss": 1.1074, + "mean_token_accuracy": 0.6782431602478027, + "num_tokens": 17303927.0, + "step": 670 + }, + { + "epoch": 0.07368767845376675, + "grad_norm": 2.239637851715088, + "learning_rate": 3.677277716794731e-07, + "loss": 1.1066, + "mean_token_accuracy": 0.6818715333938599, + "num_tokens": 17332428.0, + "step": 671 + }, + { + "epoch": 0.07379749615638041, + "grad_norm": 2.478123426437378, + "learning_rate": 3.682766190998902e-07, + "loss": 1.1305, + "mean_token_accuracy": 0.6728582382202148, + "num_tokens": 17358551.0, + "step": 672 + }, + { + "epoch": 0.07390731385899407, + "grad_norm": 2.539393901824951, + "learning_rate": 3.688254665203073e-07, + "loss": 1.0075, + "mean_token_accuracy": 0.7042092680931091, + "num_tokens": 17382289.0, + "step": 673 + }, + { + "epoch": 0.07401713156160773, + "grad_norm": 2.339833974838257, + "learning_rate": 3.6937431394072443e-07, + "loss": 1.153, + "mean_token_accuracy": 0.6604905128479004, + "num_tokens": 17409223.0, + "step": 674 + }, + { + "epoch": 0.07412694926422139, + "grad_norm": 2.131944179534912, + "learning_rate": 3.699231613611416e-07, + "loss": 1.144, + "mean_token_accuracy": 0.6674367189407349, + "num_tokens": 17443858.0, + "step": 675 + }, + { + "epoch": 0.07423676696683505, + "grad_norm": 2.7098753452301025, + "learning_rate": 3.704720087815587e-07, + "loss": 1.0092, + "mean_token_accuracy": 0.7035195231437683, + "num_tokens": 17464562.0, + "step": 676 + }, + { + "epoch": 0.07434658466944871, + "grad_norm": 2.852755546569824, + "learning_rate": 3.7102085620197583e-07, + "loss": 1.1467, + "mean_token_accuracy": 0.6587001085281372, + "num_tokens": 17485882.0, + "step": 677 + }, + { + "epoch": 0.07445640237206237, + "grad_norm": 3.1405444145202637, + "learning_rate": 3.71569703622393e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6773390769958496, + "num_tokens": 17506036.0, + "step": 678 + }, + { + "epoch": 0.07456622007467603, + "grad_norm": 2.4215188026428223, + "learning_rate": 3.721185510428101e-07, + "loss": 1.0777, + "mean_token_accuracy": 0.6764508485794067, + "num_tokens": 17531236.0, + "step": 679 + }, + { + "epoch": 0.0746760377772897, + "grad_norm": 2.7892322540283203, + "learning_rate": 3.726673984632272e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.7037753462791443, + "num_tokens": 17554197.0, + "step": 680 + }, + { + "epoch": 0.07478585547990337, + "grad_norm": 2.4831838607788086, + "learning_rate": 3.732162458836443e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.6697279810905457, + "num_tokens": 17579317.0, + "step": 681 + }, + { + "epoch": 0.07489567318251703, + "grad_norm": 2.3880021572113037, + "learning_rate": 3.7376509330406146e-07, + "loss": 1.19, + "mean_token_accuracy": 0.6535310745239258, + "num_tokens": 17607059.0, + "step": 682 + }, + { + "epoch": 0.07500549088513069, + "grad_norm": 2.5572261810302734, + "learning_rate": 3.743139407244786e-07, + "loss": 1.0781, + "mean_token_accuracy": 0.6823660135269165, + "num_tokens": 17631869.0, + "step": 683 + }, + { + "epoch": 0.07511530858774434, + "grad_norm": 2.4556469917297363, + "learning_rate": 3.748627881448957e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6706225872039795, + "num_tokens": 17658508.0, + "step": 684 + }, + { + "epoch": 0.075225126290358, + "grad_norm": 2.6340878009796143, + "learning_rate": 3.754116355653128e-07, + "loss": 1.0631, + "mean_token_accuracy": 0.6834492087364197, + "num_tokens": 17682446.0, + "step": 685 + }, + { + "epoch": 0.07533494399297166, + "grad_norm": 2.456493616104126, + "learning_rate": 3.7596048298573e-07, + "loss": 1.233, + "mean_token_accuracy": 0.6470452547073364, + "num_tokens": 17709555.0, + "step": 686 + }, + { + "epoch": 0.07544476169558532, + "grad_norm": 2.5886614322662354, + "learning_rate": 3.765093304061471e-07, + "loss": 1.1405, + "mean_token_accuracy": 0.6733046770095825, + "num_tokens": 17733416.0, + "step": 687 + }, + { + "epoch": 0.07555457939819898, + "grad_norm": 2.3124146461486816, + "learning_rate": 3.770581778265642e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6970182657241821, + "num_tokens": 17759803.0, + "step": 688 + }, + { + "epoch": 0.07566439710081264, + "grad_norm": 2.4929912090301514, + "learning_rate": 3.7760702524698137e-07, + "loss": 1.0334, + "mean_token_accuracy": 0.6920279264450073, + "num_tokens": 17783969.0, + "step": 689 + }, + { + "epoch": 0.07577421480342632, + "grad_norm": 2.2299091815948486, + "learning_rate": 3.7815587266739844e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6832695007324219, + "num_tokens": 17812672.0, + "step": 690 + }, + { + "epoch": 0.07588403250603998, + "grad_norm": 2.502657651901245, + "learning_rate": 3.7870472008781555e-07, + "loss": 1.127, + "mean_token_accuracy": 0.6709563732147217, + "num_tokens": 17837329.0, + "step": 691 + }, + { + "epoch": 0.07599385020865364, + "grad_norm": 2.522718667984009, + "learning_rate": 3.7925356750823267e-07, + "loss": 1.0677, + "mean_token_accuracy": 0.6809372305870056, + "num_tokens": 17860395.0, + "step": 692 + }, + { + "epoch": 0.0761036679112673, + "grad_norm": 2.140423536300659, + "learning_rate": 3.7980241492864983e-07, + "loss": 1.1965, + "mean_token_accuracy": 0.6532242894172668, + "num_tokens": 17896905.0, + "step": 693 + }, + { + "epoch": 0.07621348561388096, + "grad_norm": 2.374603033065796, + "learning_rate": 3.8035126234906695e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6852961182594299, + "num_tokens": 17926874.0, + "step": 694 + }, + { + "epoch": 0.07632330331649462, + "grad_norm": 3.0878493785858154, + "learning_rate": 3.8090010976948406e-07, + "loss": 0.9839, + "mean_token_accuracy": 0.6989253163337708, + "num_tokens": 17944541.0, + "step": 695 + }, + { + "epoch": 0.07643312101910828, + "grad_norm": 2.4307641983032227, + "learning_rate": 3.814489571899012e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6760109663009644, + "num_tokens": 17971757.0, + "step": 696 + }, + { + "epoch": 0.07654293872172194, + "grad_norm": 2.591301918029785, + "learning_rate": 3.8199780461031835e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.675508439540863, + "num_tokens": 17995014.0, + "step": 697 + }, + { + "epoch": 0.0766527564243356, + "grad_norm": 2.572700023651123, + "learning_rate": 3.8254665203073546e-07, + "loss": 1.0995, + "mean_token_accuracy": 0.6755619049072266, + "num_tokens": 18018896.0, + "step": 698 + }, + { + "epoch": 0.07676257412694927, + "grad_norm": 2.2168033123016357, + "learning_rate": 3.830954994511525e-07, + "loss": 1.1635, + "mean_token_accuracy": 0.6630755066871643, + "num_tokens": 18050073.0, + "step": 699 + }, + { + "epoch": 0.07687239182956293, + "grad_norm": 2.77117657661438, + "learning_rate": 3.836443468715697e-07, + "loss": 1.1143, + "mean_token_accuracy": 0.6761640310287476, + "num_tokens": 18071162.0, + "step": 700 + }, + { + "epoch": 0.07698220953217659, + "grad_norm": 2.746065378189087, + "learning_rate": 3.841931942919868e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7062646150588989, + "num_tokens": 18090203.0, + "step": 701 + }, + { + "epoch": 0.07709202723479025, + "grad_norm": 3.002047300338745, + "learning_rate": 3.847420417124039e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.6996945142745972, + "num_tokens": 18107144.0, + "step": 702 + }, + { + "epoch": 0.07720184493740391, + "grad_norm": 2.5795986652374268, + "learning_rate": 3.8529088913282104e-07, + "loss": 0.9687, + "mean_token_accuracy": 0.7076349258422852, + "num_tokens": 18128551.0, + "step": 703 + }, + { + "epoch": 0.07731166264001757, + "grad_norm": 2.9914982318878174, + "learning_rate": 3.858397365532382e-07, + "loss": 0.9545, + "mean_token_accuracy": 0.7047800421714783, + "num_tokens": 18146076.0, + "step": 704 + }, + { + "epoch": 0.07742148034263123, + "grad_norm": 2.4004011154174805, + "learning_rate": 3.863885839736553e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.657654881477356, + "num_tokens": 18173466.0, + "step": 705 + }, + { + "epoch": 0.07753129804524489, + "grad_norm": 2.522916793823242, + "learning_rate": 3.8693743139407244e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6870692372322083, + "num_tokens": 18198025.0, + "step": 706 + }, + { + "epoch": 0.07764111574785855, + "grad_norm": 2.217928886413574, + "learning_rate": 3.8748627881448955e-07, + "loss": 1.1404, + "mean_token_accuracy": 0.666064977645874, + "num_tokens": 18230387.0, + "step": 707 + }, + { + "epoch": 0.07775093345047221, + "grad_norm": 2.474332094192505, + "learning_rate": 3.880351262349067e-07, + "loss": 1.0243, + "mean_token_accuracy": 0.6955276727676392, + "num_tokens": 18253966.0, + "step": 708 + }, + { + "epoch": 0.07786075115308588, + "grad_norm": 2.5584285259246826, + "learning_rate": 3.885839736553238e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6929574012756348, + "num_tokens": 18275726.0, + "step": 709 + }, + { + "epoch": 0.07797056885569954, + "grad_norm": 2.5311591625213623, + "learning_rate": 3.891328210757409e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.685123860836029, + "num_tokens": 18300027.0, + "step": 710 + }, + { + "epoch": 0.0780803865583132, + "grad_norm": 2.3369381427764893, + "learning_rate": 3.8968166849615807e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6745376586914062, + "num_tokens": 18329055.0, + "step": 711 + }, + { + "epoch": 0.07819020426092686, + "grad_norm": 2.163975238800049, + "learning_rate": 3.902305159165752e-07, + "loss": 1.1343, + "mean_token_accuracy": 0.6758668422698975, + "num_tokens": 18360321.0, + "step": 712 + }, + { + "epoch": 0.07830002196354052, + "grad_norm": 2.6139066219329834, + "learning_rate": 3.907793633369923e-07, + "loss": 1.0662, + "mean_token_accuracy": 0.6887204051017761, + "num_tokens": 18382767.0, + "step": 713 + }, + { + "epoch": 0.07840983966615418, + "grad_norm": 2.518386125564575, + "learning_rate": 3.913282107574094e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6827998161315918, + "num_tokens": 18403792.0, + "step": 714 + }, + { + "epoch": 0.07851965736876784, + "grad_norm": 2.4654908180236816, + "learning_rate": 3.918770581778266e-07, + "loss": 1.0977, + "mean_token_accuracy": 0.676972508430481, + "num_tokens": 18428093.0, + "step": 715 + }, + { + "epoch": 0.0786294750713815, + "grad_norm": 2.421231269836426, + "learning_rate": 3.924259055982437e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.7146577835083008, + "num_tokens": 18454771.0, + "step": 716 + }, + { + "epoch": 0.07873929277399516, + "grad_norm": 2.6103782653808594, + "learning_rate": 3.929747530186608e-07, + "loss": 1.0445, + "mean_token_accuracy": 0.6900283098220825, + "num_tokens": 18477543.0, + "step": 717 + }, + { + "epoch": 0.07884911047660884, + "grad_norm": 2.2983131408691406, + "learning_rate": 3.935236004390779e-07, + "loss": 1.081, + "mean_token_accuracy": 0.6859034299850464, + "num_tokens": 18508833.0, + "step": 718 + }, + { + "epoch": 0.0789589281792225, + "grad_norm": 2.2055726051330566, + "learning_rate": 3.9407244785949504e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.6803796291351318, + "num_tokens": 18539616.0, + "step": 719 + }, + { + "epoch": 0.07906874588183616, + "grad_norm": 2.3384344577789307, + "learning_rate": 3.9462129527991216e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.686038076877594, + "num_tokens": 18567717.0, + "step": 720 + }, + { + "epoch": 0.07917856358444982, + "grad_norm": 2.316570997238159, + "learning_rate": 3.9517014270032927e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6825223565101624, + "num_tokens": 18595201.0, + "step": 721 + }, + { + "epoch": 0.07928838128706348, + "grad_norm": 2.6781773567199707, + "learning_rate": 3.957189901207464e-07, + "loss": 1.1272, + "mean_token_accuracy": 0.6684756875038147, + "num_tokens": 18617556.0, + "step": 722 + }, + { + "epoch": 0.07939819898967714, + "grad_norm": 2.0664618015289307, + "learning_rate": 3.9626783754116356e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.677390992641449, + "num_tokens": 18648544.0, + "step": 723 + }, + { + "epoch": 0.0795080166922908, + "grad_norm": 2.3900465965270996, + "learning_rate": 3.9681668496158067e-07, + "loss": 1.0641, + "mean_token_accuracy": 0.6822684407234192, + "num_tokens": 18673668.0, + "step": 724 + }, + { + "epoch": 0.07961783439490445, + "grad_norm": 2.261446714401245, + "learning_rate": 3.973655323819978e-07, + "loss": 1.0571, + "mean_token_accuracy": 0.6999247670173645, + "num_tokens": 18702507.0, + "step": 725 + }, + { + "epoch": 0.07972765209751811, + "grad_norm": 2.4003453254699707, + "learning_rate": 3.9791437980241495e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.694583535194397, + "num_tokens": 18727212.0, + "step": 726 + }, + { + "epoch": 0.07983746980013177, + "grad_norm": 2.1780827045440674, + "learning_rate": 3.9846322722283207e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6903059482574463, + "num_tokens": 18759730.0, + "step": 727 + }, + { + "epoch": 0.07994728750274545, + "grad_norm": 2.326012134552002, + "learning_rate": 3.9901207464324913e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.699002206325531, + "num_tokens": 18787318.0, + "step": 728 + }, + { + "epoch": 0.08005710520535911, + "grad_norm": 2.606464147567749, + "learning_rate": 3.9956092206366625e-07, + "loss": 1.1194, + "mean_token_accuracy": 0.6718422174453735, + "num_tokens": 18812894.0, + "step": 729 + }, + { + "epoch": 0.08016692290797277, + "grad_norm": 3.0656065940856934, + "learning_rate": 4.001097694840834e-07, + "loss": 1.1439, + "mean_token_accuracy": 0.6869906783103943, + "num_tokens": 18831219.0, + "step": 730 + }, + { + "epoch": 0.08027674061058643, + "grad_norm": 2.427823781967163, + "learning_rate": 4.0065861690450053e-07, + "loss": 1.0817, + "mean_token_accuracy": 0.6816327571868896, + "num_tokens": 18857883.0, + "step": 731 + }, + { + "epoch": 0.08038655831320009, + "grad_norm": 2.2069215774536133, + "learning_rate": 4.0120746432491765e-07, + "loss": 1.2247, + "mean_token_accuracy": 0.6480692028999329, + "num_tokens": 18887665.0, + "step": 732 + }, + { + "epoch": 0.08049637601581375, + "grad_norm": 2.3767621517181396, + "learning_rate": 4.0175631174533476e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6818535327911377, + "num_tokens": 18916998.0, + "step": 733 + }, + { + "epoch": 0.08060619371842741, + "grad_norm": 2.2648630142211914, + "learning_rate": 4.0230515916575193e-07, + "loss": 1.1413, + "mean_token_accuracy": 0.6725637912750244, + "num_tokens": 18943423.0, + "step": 734 + }, + { + "epoch": 0.08071601142104107, + "grad_norm": 2.0856246948242188, + "learning_rate": 4.0285400658616904e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.6864229440689087, + "num_tokens": 18977108.0, + "step": 735 + }, + { + "epoch": 0.08082582912365473, + "grad_norm": 2.650080442428589, + "learning_rate": 4.0340285400658616e-07, + "loss": 1.078, + "mean_token_accuracy": 0.6821413040161133, + "num_tokens": 18999476.0, + "step": 736 + }, + { + "epoch": 0.0809356468262684, + "grad_norm": 2.619624376296997, + "learning_rate": 4.039517014270033e-07, + "loss": 0.9503, + "mean_token_accuracy": 0.7187405824661255, + "num_tokens": 19022102.0, + "step": 737 + }, + { + "epoch": 0.08104546452888206, + "grad_norm": 2.627427577972412, + "learning_rate": 4.045005488474204e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6778899431228638, + "num_tokens": 19044572.0, + "step": 738 + }, + { + "epoch": 0.08115528223149572, + "grad_norm": 2.419008493423462, + "learning_rate": 4.050493962678375e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6770764589309692, + "num_tokens": 19069768.0, + "step": 739 + }, + { + "epoch": 0.08126509993410938, + "grad_norm": 2.363037347793579, + "learning_rate": 4.055982436882546e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7130345106124878, + "num_tokens": 19093937.0, + "step": 740 + }, + { + "epoch": 0.08137491763672304, + "grad_norm": 2.2156050205230713, + "learning_rate": 4.061470911086718e-07, + "loss": 1.011, + "mean_token_accuracy": 0.6987730264663696, + "num_tokens": 19121085.0, + "step": 741 + }, + { + "epoch": 0.0814847353393367, + "grad_norm": 2.484388828277588, + "learning_rate": 4.066959385290889e-07, + "loss": 1.119, + "mean_token_accuracy": 0.6722083687782288, + "num_tokens": 19145732.0, + "step": 742 + }, + { + "epoch": 0.08159455304195036, + "grad_norm": 2.2112362384796143, + "learning_rate": 4.07244785949506e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.6747061610221863, + "num_tokens": 19174733.0, + "step": 743 + }, + { + "epoch": 0.08170437074456402, + "grad_norm": 2.2290849685668945, + "learning_rate": 4.0779363336992313e-07, + "loss": 1.0752, + "mean_token_accuracy": 0.6816835403442383, + "num_tokens": 19202240.0, + "step": 744 + }, + { + "epoch": 0.08181418844717768, + "grad_norm": 2.305635452270508, + "learning_rate": 4.083424807903403e-07, + "loss": 1.112, + "mean_token_accuracy": 0.6733685731887817, + "num_tokens": 19230765.0, + "step": 745 + }, + { + "epoch": 0.08192400614979134, + "grad_norm": 2.2872095108032227, + "learning_rate": 4.0889132821075737e-07, + "loss": 0.982, + "mean_token_accuracy": 0.7104142904281616, + "num_tokens": 19260408.0, + "step": 746 + }, + { + "epoch": 0.08203382385240501, + "grad_norm": 2.214920997619629, + "learning_rate": 4.094401756311745e-07, + "loss": 1.0123, + "mean_token_accuracy": 0.6978511810302734, + "num_tokens": 19290817.0, + "step": 747 + }, + { + "epoch": 0.08214364155501867, + "grad_norm": 2.235208034515381, + "learning_rate": 4.0998902305159165e-07, + "loss": 1.075, + "mean_token_accuracy": 0.6827152371406555, + "num_tokens": 19319256.0, + "step": 748 + }, + { + "epoch": 0.08225345925763233, + "grad_norm": 2.309936285018921, + "learning_rate": 4.1053787047200876e-07, + "loss": 1.2004, + "mean_token_accuracy": 0.6523422002792358, + "num_tokens": 19347837.0, + "step": 749 + }, + { + "epoch": 0.08236327696024599, + "grad_norm": 2.421898126602173, + "learning_rate": 4.110867178924259e-07, + "loss": 1.1387, + "mean_token_accuracy": 0.6659280061721802, + "num_tokens": 19372972.0, + "step": 750 + }, + { + "epoch": 0.08247309466285965, + "grad_norm": 2.1361160278320312, + "learning_rate": 4.11635565312843e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6900055408477783, + "num_tokens": 19404323.0, + "step": 751 + }, + { + "epoch": 0.08258291236547331, + "grad_norm": 2.0638794898986816, + "learning_rate": 4.1218441273326016e-07, + "loss": 1.1568, + "mean_token_accuracy": 0.6666438579559326, + "num_tokens": 19436333.0, + "step": 752 + }, + { + "epoch": 0.08269273006808697, + "grad_norm": 2.3339507579803467, + "learning_rate": 4.127332601536773e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.7040693163871765, + "num_tokens": 19462018.0, + "step": 753 + }, + { + "epoch": 0.08280254777070063, + "grad_norm": 2.4093616008758545, + "learning_rate": 4.132821075740944e-07, + "loss": 1.1262, + "mean_token_accuracy": 0.6789978742599487, + "num_tokens": 19489429.0, + "step": 754 + }, + { + "epoch": 0.08291236547331429, + "grad_norm": 2.5630030632019043, + "learning_rate": 4.138309549945115e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6835469007492065, + "num_tokens": 19513331.0, + "step": 755 + }, + { + "epoch": 0.08302218317592797, + "grad_norm": 2.1760900020599365, + "learning_rate": 4.143798024149286e-07, + "loss": 1.1247, + "mean_token_accuracy": 0.67546546459198, + "num_tokens": 19541778.0, + "step": 756 + }, + { + "epoch": 0.08313200087854163, + "grad_norm": 2.6457412242889404, + "learning_rate": 4.1492864983534574e-07, + "loss": 1.1081, + "mean_token_accuracy": 0.6727392077445984, + "num_tokens": 19562859.0, + "step": 757 + }, + { + "epoch": 0.08324181858115529, + "grad_norm": 2.6138062477111816, + "learning_rate": 4.1547749725576285e-07, + "loss": 1.028, + "mean_token_accuracy": 0.6947408318519592, + "num_tokens": 19583435.0, + "step": 758 + }, + { + "epoch": 0.08335163628376895, + "grad_norm": 2.26759934425354, + "learning_rate": 4.1602634467618e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.6861478090286255, + "num_tokens": 19609704.0, + "step": 759 + }, + { + "epoch": 0.0834614539863826, + "grad_norm": 2.437007427215576, + "learning_rate": 4.1657519209659714e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.7047675848007202, + "num_tokens": 19633530.0, + "step": 760 + }, + { + "epoch": 0.08357127168899627, + "grad_norm": 2.725649118423462, + "learning_rate": 4.1712403951701425e-07, + "loss": 1.0081, + "mean_token_accuracy": 0.7014360427856445, + "num_tokens": 19657148.0, + "step": 761 + }, + { + "epoch": 0.08368108939160993, + "grad_norm": 2.19480562210083, + "learning_rate": 4.1767288693743137e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6894296407699585, + "num_tokens": 19688382.0, + "step": 762 + }, + { + "epoch": 0.08379090709422359, + "grad_norm": 2.502861976623535, + "learning_rate": 4.1822173435784854e-07, + "loss": 1.0908, + "mean_token_accuracy": 0.6877384185791016, + "num_tokens": 19714262.0, + "step": 763 + }, + { + "epoch": 0.08390072479683724, + "grad_norm": 1.9183375835418701, + "learning_rate": 4.1877058177826565e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6785393953323364, + "num_tokens": 19749871.0, + "step": 764 + }, + { + "epoch": 0.0840105424994509, + "grad_norm": 2.279313802719116, + "learning_rate": 4.193194291986827e-07, + "loss": 1.1062, + "mean_token_accuracy": 0.6767182350158691, + "num_tokens": 19782694.0, + "step": 765 + }, + { + "epoch": 0.08412036020206458, + "grad_norm": 2.370185136795044, + "learning_rate": 4.1986827661909983e-07, + "loss": 1.0328, + "mean_token_accuracy": 0.6928603053092957, + "num_tokens": 19809532.0, + "step": 766 + }, + { + "epoch": 0.08423017790467824, + "grad_norm": 2.2902300357818604, + "learning_rate": 4.20417124039517e-07, + "loss": 1.132, + "mean_token_accuracy": 0.6686508655548096, + "num_tokens": 19837017.0, + "step": 767 + }, + { + "epoch": 0.0843399956072919, + "grad_norm": 2.2083239555358887, + "learning_rate": 4.209659714599341e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6681433320045471, + "num_tokens": 19866686.0, + "step": 768 + }, + { + "epoch": 0.08444981330990556, + "grad_norm": 2.2758214473724365, + "learning_rate": 4.2151481888035123e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.679768979549408, + "num_tokens": 19895021.0, + "step": 769 + }, + { + "epoch": 0.08455963101251922, + "grad_norm": 2.409475803375244, + "learning_rate": 4.2206366630076834e-07, + "loss": 1.2004, + "mean_token_accuracy": 0.6548238396644592, + "num_tokens": 19922728.0, + "step": 770 + }, + { + "epoch": 0.08466944871513288, + "grad_norm": 2.252531051635742, + "learning_rate": 4.226125137211855e-07, + "loss": 1.1637, + "mean_token_accuracy": 0.6700092554092407, + "num_tokens": 19951679.0, + "step": 771 + }, + { + "epoch": 0.08477926641774654, + "grad_norm": 2.989283561706543, + "learning_rate": 4.231613611416026e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.6944243907928467, + "num_tokens": 19969886.0, + "step": 772 + }, + { + "epoch": 0.0848890841203602, + "grad_norm": 2.1178228855133057, + "learning_rate": 4.2371020856201974e-07, + "loss": 1.0646, + "mean_token_accuracy": 0.6859607696533203, + "num_tokens": 20002739.0, + "step": 773 + }, + { + "epoch": 0.08499890182297386, + "grad_norm": 2.509751796722412, + "learning_rate": 4.242590559824369e-07, + "loss": 1.1974, + "mean_token_accuracy": 0.6560907959938049, + "num_tokens": 20029550.0, + "step": 774 + }, + { + "epoch": 0.08510871952558753, + "grad_norm": 2.246711492538452, + "learning_rate": 4.2480790340285397e-07, + "loss": 1.1939, + "mean_token_accuracy": 0.6598840355873108, + "num_tokens": 20057694.0, + "step": 775 + }, + { + "epoch": 0.08521853722820119, + "grad_norm": 2.579882860183716, + "learning_rate": 4.253567508232711e-07, + "loss": 1.1391, + "mean_token_accuracy": 0.6710803508758545, + "num_tokens": 20081038.0, + "step": 776 + }, + { + "epoch": 0.08532835493081485, + "grad_norm": 2.1639909744262695, + "learning_rate": 4.259055982436882e-07, + "loss": 1.1715, + "mean_token_accuracy": 0.663495659828186, + "num_tokens": 20111032.0, + "step": 777 + }, + { + "epoch": 0.08543817263342851, + "grad_norm": 2.3376848697662354, + "learning_rate": 4.2645444566410537e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.681607723236084, + "num_tokens": 20136264.0, + "step": 778 + }, + { + "epoch": 0.08554799033604217, + "grad_norm": 2.1180880069732666, + "learning_rate": 4.270032930845225e-07, + "loss": 1.1272, + "mean_token_accuracy": 0.6710848808288574, + "num_tokens": 20168253.0, + "step": 779 + }, + { + "epoch": 0.08565780803865583, + "grad_norm": 2.2569148540496826, + "learning_rate": 4.275521405049396e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6692781448364258, + "num_tokens": 20196641.0, + "step": 780 + }, + { + "epoch": 0.08576762574126949, + "grad_norm": 2.5442733764648438, + "learning_rate": 4.281009879253567e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.6979770660400391, + "num_tokens": 20220380.0, + "step": 781 + }, + { + "epoch": 0.08587744344388315, + "grad_norm": 2.3392229080200195, + "learning_rate": 4.286498353457739e-07, + "loss": 1.1088, + "mean_token_accuracy": 0.6739777326583862, + "num_tokens": 20246280.0, + "step": 782 + }, + { + "epoch": 0.08598726114649681, + "grad_norm": 2.369130849838257, + "learning_rate": 4.29198682766191e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.6849479675292969, + "num_tokens": 20273105.0, + "step": 783 + }, + { + "epoch": 0.08609707884911047, + "grad_norm": 2.214662790298462, + "learning_rate": 4.2974753018660806e-07, + "loss": 1.1451, + "mean_token_accuracy": 0.6752175092697144, + "num_tokens": 20301212.0, + "step": 784 + }, + { + "epoch": 0.08620689655172414, + "grad_norm": 2.634342908859253, + "learning_rate": 4.3029637760702523e-07, + "loss": 1.1834, + "mean_token_accuracy": 0.662522554397583, + "num_tokens": 20326898.0, + "step": 785 + }, + { + "epoch": 0.0863167142543378, + "grad_norm": 2.803968667984009, + "learning_rate": 4.3084522502744235e-07, + "loss": 1.0421, + "mean_token_accuracy": 0.690568745136261, + "num_tokens": 20348738.0, + "step": 786 + }, + { + "epoch": 0.08642653195695146, + "grad_norm": 2.565188407897949, + "learning_rate": 4.3139407244785946e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6886904239654541, + "num_tokens": 20371612.0, + "step": 787 + }, + { + "epoch": 0.08653634965956512, + "grad_norm": 2.6032004356384277, + "learning_rate": 4.319429198682766e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.6601462960243225, + "num_tokens": 20395811.0, + "step": 788 + }, + { + "epoch": 0.08664616736217878, + "grad_norm": 2.5222244262695312, + "learning_rate": 4.3249176728869374e-07, + "loss": 1.1314, + "mean_token_accuracy": 0.6637892127037048, + "num_tokens": 20422291.0, + "step": 789 + }, + { + "epoch": 0.08675598506479244, + "grad_norm": 2.3348066806793213, + "learning_rate": 4.3304061470911086e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6881160736083984, + "num_tokens": 20448345.0, + "step": 790 + }, + { + "epoch": 0.0868658027674061, + "grad_norm": 2.3909850120544434, + "learning_rate": 4.33589462129528e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.673900842666626, + "num_tokens": 20474414.0, + "step": 791 + }, + { + "epoch": 0.08697562047001976, + "grad_norm": 2.3794589042663574, + "learning_rate": 4.341383095499451e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6798043847084045, + "num_tokens": 20499492.0, + "step": 792 + }, + { + "epoch": 0.08708543817263342, + "grad_norm": 2.57655668258667, + "learning_rate": 4.3468715697036226e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6922095417976379, + "num_tokens": 20522454.0, + "step": 793 + }, + { + "epoch": 0.0871952558752471, + "grad_norm": 2.139578342437744, + "learning_rate": 4.352360043907793e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.6898977756500244, + "num_tokens": 20551953.0, + "step": 794 + }, + { + "epoch": 0.08730507357786076, + "grad_norm": 2.094290018081665, + "learning_rate": 4.3578485181119644e-07, + "loss": 1.1045, + "mean_token_accuracy": 0.6683666706085205, + "num_tokens": 20584482.0, + "step": 795 + }, + { + "epoch": 0.08741489128047442, + "grad_norm": 2.4363369941711426, + "learning_rate": 4.363336992316136e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6864429712295532, + "num_tokens": 20607675.0, + "step": 796 + }, + { + "epoch": 0.08752470898308808, + "grad_norm": 2.112039089202881, + "learning_rate": 4.368825466520307e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6723403334617615, + "num_tokens": 20639280.0, + "step": 797 + }, + { + "epoch": 0.08763452668570174, + "grad_norm": 2.6592776775360107, + "learning_rate": 4.3743139407244783e-07, + "loss": 1.1045, + "mean_token_accuracy": 0.672168493270874, + "num_tokens": 20662380.0, + "step": 798 + }, + { + "epoch": 0.0877443443883154, + "grad_norm": 2.4553098678588867, + "learning_rate": 4.3798024149286495e-07, + "loss": 1.141, + "mean_token_accuracy": 0.6666065454483032, + "num_tokens": 20687044.0, + "step": 799 + }, + { + "epoch": 0.08785416209092906, + "grad_norm": 2.65667986869812, + "learning_rate": 4.385290889132821e-07, + "loss": 1.0642, + "mean_token_accuracy": 0.6869840025901794, + "num_tokens": 20709471.0, + "step": 800 + }, + { + "epoch": 0.08796397979354272, + "grad_norm": 2.0557305812835693, + "learning_rate": 4.3907793633369923e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.6954424977302551, + "num_tokens": 20740205.0, + "step": 801 + }, + { + "epoch": 0.08807379749615638, + "grad_norm": 2.4421775341033936, + "learning_rate": 4.3962678375411635e-07, + "loss": 1.0257, + "mean_token_accuracy": 0.6924665570259094, + "num_tokens": 20765399.0, + "step": 802 + }, + { + "epoch": 0.08818361519877004, + "grad_norm": 2.4474716186523438, + "learning_rate": 4.401756311745334e-07, + "loss": 1.0933, + "mean_token_accuracy": 0.6697843670845032, + "num_tokens": 20790334.0, + "step": 803 + }, + { + "epoch": 0.08829343290138371, + "grad_norm": 2.402036190032959, + "learning_rate": 4.407244785949506e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6915091276168823, + "num_tokens": 20815630.0, + "step": 804 + }, + { + "epoch": 0.08840325060399737, + "grad_norm": 2.714433193206787, + "learning_rate": 4.412733260153677e-07, + "loss": 1.1358, + "mean_token_accuracy": 0.6704472899436951, + "num_tokens": 20839241.0, + "step": 805 + }, + { + "epoch": 0.08851306830661103, + "grad_norm": 2.5012943744659424, + "learning_rate": 4.418221734357848e-07, + "loss": 1.1406, + "mean_token_accuracy": 0.6646542549133301, + "num_tokens": 20864398.0, + "step": 806 + }, + { + "epoch": 0.08862288600922469, + "grad_norm": 2.42201566696167, + "learning_rate": 4.42371020856202e-07, + "loss": 1.0382, + "mean_token_accuracy": 0.6800587177276611, + "num_tokens": 20888860.0, + "step": 807 + }, + { + "epoch": 0.08873270371183835, + "grad_norm": 2.2971484661102295, + "learning_rate": 4.429198682766191e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.6740670204162598, + "num_tokens": 20916162.0, + "step": 808 + }, + { + "epoch": 0.08884252141445201, + "grad_norm": 2.4296152591705322, + "learning_rate": 4.434687156970362e-07, + "loss": 0.9522, + "mean_token_accuracy": 0.7155856490135193, + "num_tokens": 20942190.0, + "step": 809 + }, + { + "epoch": 0.08895233911706567, + "grad_norm": 2.330012321472168, + "learning_rate": 4.440175631174533e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.6689844131469727, + "num_tokens": 20968953.0, + "step": 810 + }, + { + "epoch": 0.08906215681967933, + "grad_norm": 2.457970142364502, + "learning_rate": 4.445664105378705e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6691176295280457, + "num_tokens": 20995139.0, + "step": 811 + }, + { + "epoch": 0.08917197452229299, + "grad_norm": 2.3714277744293213, + "learning_rate": 4.451152579582876e-07, + "loss": 1.0217, + "mean_token_accuracy": 0.7049256563186646, + "num_tokens": 21021761.0, + "step": 812 + }, + { + "epoch": 0.08928179222490666, + "grad_norm": 2.4437155723571777, + "learning_rate": 4.4566410537870467e-07, + "loss": 1.0057, + "mean_token_accuracy": 0.703563928604126, + "num_tokens": 21044951.0, + "step": 813 + }, + { + "epoch": 0.08939160992752032, + "grad_norm": 2.5954575538635254, + "learning_rate": 4.462129527991218e-07, + "loss": 1.145, + "mean_token_accuracy": 0.6729400157928467, + "num_tokens": 21066622.0, + "step": 814 + }, + { + "epoch": 0.08950142763013398, + "grad_norm": 2.4294674396514893, + "learning_rate": 4.4676180021953895e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6642907857894897, + "num_tokens": 21092851.0, + "step": 815 + }, + { + "epoch": 0.08961124533274764, + "grad_norm": 2.586594343185425, + "learning_rate": 4.4731064763995607e-07, + "loss": 0.9611, + "mean_token_accuracy": 0.7073994278907776, + "num_tokens": 21114268.0, + "step": 816 + }, + { + "epoch": 0.0897210630353613, + "grad_norm": 2.587435007095337, + "learning_rate": 4.478594950603732e-07, + "loss": 1.0133, + "mean_token_accuracy": 0.6971490383148193, + "num_tokens": 21135026.0, + "step": 817 + }, + { + "epoch": 0.08983088073797496, + "grad_norm": 2.275217294692993, + "learning_rate": 4.4840834248079035e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6860107183456421, + "num_tokens": 21163309.0, + "step": 818 + }, + { + "epoch": 0.08994069844058862, + "grad_norm": 2.5373380184173584, + "learning_rate": 4.4895718990120747e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6693641543388367, + "num_tokens": 21187471.0, + "step": 819 + }, + { + "epoch": 0.09005051614320228, + "grad_norm": 2.153738260269165, + "learning_rate": 4.495060373216246e-07, + "loss": 1.1084, + "mean_token_accuracy": 0.6737122535705566, + "num_tokens": 21216926.0, + "step": 820 + }, + { + "epoch": 0.09016033384581594, + "grad_norm": 2.4825046062469482, + "learning_rate": 4.500548847420417e-07, + "loss": 1.1509, + "mean_token_accuracy": 0.6667303442955017, + "num_tokens": 21242579.0, + "step": 821 + }, + { + "epoch": 0.0902701515484296, + "grad_norm": 2.480659246444702, + "learning_rate": 4.5060373216245886e-07, + "loss": 1.0093, + "mean_token_accuracy": 0.6982008218765259, + "num_tokens": 21264864.0, + "step": 822 + }, + { + "epoch": 0.09037996925104327, + "grad_norm": 2.229440689086914, + "learning_rate": 4.5115257958287593e-07, + "loss": 1.1274, + "mean_token_accuracy": 0.6728754043579102, + "num_tokens": 21293928.0, + "step": 823 + }, + { + "epoch": 0.09048978695365693, + "grad_norm": 2.46288800239563, + "learning_rate": 4.5170142700329304e-07, + "loss": 1.0804, + "mean_token_accuracy": 0.6802423596382141, + "num_tokens": 21319156.0, + "step": 824 + }, + { + "epoch": 0.0905996046562706, + "grad_norm": 2.3695600032806396, + "learning_rate": 4.5225027442371016e-07, + "loss": 1.1591, + "mean_token_accuracy": 0.6632497310638428, + "num_tokens": 21346992.0, + "step": 825 + }, + { + "epoch": 0.09070942235888425, + "grad_norm": 2.2801716327667236, + "learning_rate": 4.527991218441273e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6762323975563049, + "num_tokens": 21374631.0, + "step": 826 + }, + { + "epoch": 0.09081924006149791, + "grad_norm": 2.3054962158203125, + "learning_rate": 4.5334796926454444e-07, + "loss": 1.107, + "mean_token_accuracy": 0.6732226610183716, + "num_tokens": 21400618.0, + "step": 827 + }, + { + "epoch": 0.09092905776411157, + "grad_norm": 2.078385591506958, + "learning_rate": 4.5389681668496156e-07, + "loss": 1.192, + "mean_token_accuracy": 0.6609467267990112, + "num_tokens": 21433211.0, + "step": 828 + }, + { + "epoch": 0.09103887546672523, + "grad_norm": 2.6066718101501465, + "learning_rate": 4.5444566410537867e-07, + "loss": 1.0819, + "mean_token_accuracy": 0.6787344217300415, + "num_tokens": 21455838.0, + "step": 829 + }, + { + "epoch": 0.09114869316933889, + "grad_norm": 2.2829790115356445, + "learning_rate": 4.5499451152579584e-07, + "loss": 1.1063, + "mean_token_accuracy": 0.6716932058334351, + "num_tokens": 21481286.0, + "step": 830 + }, + { + "epoch": 0.09125851087195255, + "grad_norm": 2.3183741569519043, + "learning_rate": 4.5554335894621295e-07, + "loss": 1.0736, + "mean_token_accuracy": 0.6771422624588013, + "num_tokens": 21507354.0, + "step": 831 + }, + { + "epoch": 0.09136832857456623, + "grad_norm": 2.1377029418945312, + "learning_rate": 4.5609220636663e-07, + "loss": 1.1377, + "mean_token_accuracy": 0.6678211688995361, + "num_tokens": 21537749.0, + "step": 832 + }, + { + "epoch": 0.09147814627717989, + "grad_norm": 2.697209596633911, + "learning_rate": 4.566410537870472e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6843472719192505, + "num_tokens": 21558345.0, + "step": 833 + }, + { + "epoch": 0.09158796397979355, + "grad_norm": 2.5240070819854736, + "learning_rate": 4.571899012074643e-07, + "loss": 1.0207, + "mean_token_accuracy": 0.6935884952545166, + "num_tokens": 21579355.0, + "step": 834 + }, + { + "epoch": 0.0916977816824072, + "grad_norm": 2.323535680770874, + "learning_rate": 4.577387486278814e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.687251091003418, + "num_tokens": 21606047.0, + "step": 835 + }, + { + "epoch": 0.09180759938502087, + "grad_norm": 2.307421922683716, + "learning_rate": 4.5828759604829853e-07, + "loss": 1.0114, + "mean_token_accuracy": 0.7017228603363037, + "num_tokens": 21630214.0, + "step": 836 + }, + { + "epoch": 0.09191741708763453, + "grad_norm": 2.6467556953430176, + "learning_rate": 4.588364434687157e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6915925145149231, + "num_tokens": 21650795.0, + "step": 837 + }, + { + "epoch": 0.09202723479024819, + "grad_norm": 2.6596977710723877, + "learning_rate": 4.593852908891328e-07, + "loss": 1.1251, + "mean_token_accuracy": 0.6676136255264282, + "num_tokens": 21672060.0, + "step": 838 + }, + { + "epoch": 0.09213705249286185, + "grad_norm": 2.5070741176605225, + "learning_rate": 4.5993413830954993e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6997058987617493, + "num_tokens": 21695625.0, + "step": 839 + }, + { + "epoch": 0.0922468701954755, + "grad_norm": 2.1926817893981934, + "learning_rate": 4.6048298572996704e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6764187812805176, + "num_tokens": 21726078.0, + "step": 840 + }, + { + "epoch": 0.09235668789808917, + "grad_norm": 2.8748717308044434, + "learning_rate": 4.6103183315038416e-07, + "loss": 0.9819, + "mean_token_accuracy": 0.712313711643219, + "num_tokens": 21744944.0, + "step": 841 + }, + { + "epoch": 0.09246650560070284, + "grad_norm": 2.3266026973724365, + "learning_rate": 4.615806805708013e-07, + "loss": 1.1023, + "mean_token_accuracy": 0.6825240254402161, + "num_tokens": 21771319.0, + "step": 842 + }, + { + "epoch": 0.0925763233033165, + "grad_norm": 2.0978569984436035, + "learning_rate": 4.621295279912184e-07, + "loss": 1.1713, + "mean_token_accuracy": 0.6616482734680176, + "num_tokens": 21803090.0, + "step": 843 + }, + { + "epoch": 0.09268614100593016, + "grad_norm": 2.4831037521362305, + "learning_rate": 4.6267837541163556e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.6946195363998413, + "num_tokens": 21829113.0, + "step": 844 + }, + { + "epoch": 0.09279595870854382, + "grad_norm": 2.5974180698394775, + "learning_rate": 4.632272228320527e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.6964787840843201, + "num_tokens": 21850869.0, + "step": 845 + }, + { + "epoch": 0.09290577641115748, + "grad_norm": 2.2111759185791016, + "learning_rate": 4.637760702524698e-07, + "loss": 1.0362, + "mean_token_accuracy": 0.6985376477241516, + "num_tokens": 21879571.0, + "step": 846 + }, + { + "epoch": 0.09301559411377114, + "grad_norm": 2.564854145050049, + "learning_rate": 4.643249176728869e-07, + "loss": 1.0799, + "mean_token_accuracy": 0.6784630417823792, + "num_tokens": 21902420.0, + "step": 847 + }, + { + "epoch": 0.0931254118163848, + "grad_norm": 2.539858102798462, + "learning_rate": 4.6487376509330407e-07, + "loss": 1.1536, + "mean_token_accuracy": 0.6597084999084473, + "num_tokens": 21927389.0, + "step": 848 + }, + { + "epoch": 0.09323522951899846, + "grad_norm": 2.317765474319458, + "learning_rate": 4.654226125137212e-07, + "loss": 1.0087, + "mean_token_accuracy": 0.6991794109344482, + "num_tokens": 21951842.0, + "step": 849 + }, + { + "epoch": 0.09334504722161212, + "grad_norm": 2.829254388809204, + "learning_rate": 4.659714599341383e-07, + "loss": 1.017, + "mean_token_accuracy": 0.6886532306671143, + "num_tokens": 21969762.0, + "step": 850 + }, + { + "epoch": 0.09345486492422579, + "grad_norm": 2.2312188148498535, + "learning_rate": 4.6652030735455537e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.6810721158981323, + "num_tokens": 21997961.0, + "step": 851 + }, + { + "epoch": 0.09356468262683945, + "grad_norm": 2.2279415130615234, + "learning_rate": 4.6706915477497253e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6947228312492371, + "num_tokens": 22028961.0, + "step": 852 + }, + { + "epoch": 0.09367450032945311, + "grad_norm": 2.3287410736083984, + "learning_rate": 4.6761800219538965e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.6737366914749146, + "num_tokens": 22054905.0, + "step": 853 + }, + { + "epoch": 0.09378431803206677, + "grad_norm": 2.497230052947998, + "learning_rate": 4.6816684961580676e-07, + "loss": 1.0779, + "mean_token_accuracy": 0.6847947835922241, + "num_tokens": 22077444.0, + "step": 854 + }, + { + "epoch": 0.09389413573468043, + "grad_norm": 2.439384937286377, + "learning_rate": 4.6871569703622393e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.6934778690338135, + "num_tokens": 22100258.0, + "step": 855 + }, + { + "epoch": 0.09400395343729409, + "grad_norm": 2.3275415897369385, + "learning_rate": 4.6926454445664105e-07, + "loss": 1.0815, + "mean_token_accuracy": 0.6798056364059448, + "num_tokens": 22125359.0, + "step": 856 + }, + { + "epoch": 0.09411377113990775, + "grad_norm": 2.2404794692993164, + "learning_rate": 4.6981339187705816e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6895594000816345, + "num_tokens": 22151593.0, + "step": 857 + }, + { + "epoch": 0.09422358884252141, + "grad_norm": 2.289073944091797, + "learning_rate": 4.703622392974753e-07, + "loss": 1.0172, + "mean_token_accuracy": 0.6956802606582642, + "num_tokens": 22177647.0, + "step": 858 + }, + { + "epoch": 0.09433340654513507, + "grad_norm": 2.324702262878418, + "learning_rate": 4.7091108671789245e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.6896144151687622, + "num_tokens": 22202059.0, + "step": 859 + }, + { + "epoch": 0.09444322424774873, + "grad_norm": 2.2111589908599854, + "learning_rate": 4.714599341383095e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.676030158996582, + "num_tokens": 22229511.0, + "step": 860 + }, + { + "epoch": 0.0945530419503624, + "grad_norm": 2.3961753845214844, + "learning_rate": 4.720087815587266e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.6744182109832764, + "num_tokens": 22252867.0, + "step": 861 + }, + { + "epoch": 0.09466285965297606, + "grad_norm": 2.233245611190796, + "learning_rate": 4.7255762897914374e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6901947259902954, + "num_tokens": 22279828.0, + "step": 862 + }, + { + "epoch": 0.09477267735558972, + "grad_norm": 2.478889226913452, + "learning_rate": 4.731064763995609e-07, + "loss": 1.0934, + "mean_token_accuracy": 0.6823557019233704, + "num_tokens": 22303579.0, + "step": 863 + }, + { + "epoch": 0.09488249505820338, + "grad_norm": 2.199126958847046, + "learning_rate": 4.73655323819978e-07, + "loss": 1.0259, + "mean_token_accuracy": 0.695371150970459, + "num_tokens": 22331754.0, + "step": 864 + }, + { + "epoch": 0.09499231276081704, + "grad_norm": 1.9591832160949707, + "learning_rate": 4.7420417124039514e-07, + "loss": 1.1215, + "mean_token_accuracy": 0.6663216352462769, + "num_tokens": 22368958.0, + "step": 865 + }, + { + "epoch": 0.0951021304634307, + "grad_norm": 2.5726940631866455, + "learning_rate": 4.747530186608123e-07, + "loss": 0.99, + "mean_token_accuracy": 0.7017232179641724, + "num_tokens": 22390628.0, + "step": 866 + }, + { + "epoch": 0.09521194816604436, + "grad_norm": 2.4110100269317627, + "learning_rate": 4.753018660812294e-07, + "loss": 0.9868, + "mean_token_accuracy": 0.6977535486221313, + "num_tokens": 22412729.0, + "step": 867 + }, + { + "epoch": 0.09532176586865802, + "grad_norm": 2.4780325889587402, + "learning_rate": 4.7585071350164654e-07, + "loss": 0.9813, + "mean_token_accuracy": 0.7008484601974487, + "num_tokens": 22437639.0, + "step": 868 + }, + { + "epoch": 0.09543158357127168, + "grad_norm": 2.130146026611328, + "learning_rate": 4.7639956092206365e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.7043481469154358, + "num_tokens": 22466588.0, + "step": 869 + }, + { + "epoch": 0.09554140127388536, + "grad_norm": 2.5408518314361572, + "learning_rate": 4.769484083424808e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6753840446472168, + "num_tokens": 22489978.0, + "step": 870 + }, + { + "epoch": 0.09565121897649902, + "grad_norm": 2.8443613052368164, + "learning_rate": 4.774972557628979e-07, + "loss": 1.0686, + "mean_token_accuracy": 0.6778349876403809, + "num_tokens": 22507558.0, + "step": 871 + }, + { + "epoch": 0.09576103667911268, + "grad_norm": 2.2237460613250732, + "learning_rate": 4.78046103183315e-07, + "loss": 1.1515, + "mean_token_accuracy": 0.6677594184875488, + "num_tokens": 22535742.0, + "step": 872 + }, + { + "epoch": 0.09587085438172634, + "grad_norm": 2.683119535446167, + "learning_rate": 4.785949506037321e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.6863938570022583, + "num_tokens": 22557527.0, + "step": 873 + }, + { + "epoch": 0.09598067208434, + "grad_norm": 2.341549873352051, + "learning_rate": 4.791437980241493e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.6796650886535645, + "num_tokens": 22582479.0, + "step": 874 + }, + { + "epoch": 0.09609048978695366, + "grad_norm": 2.6724114418029785, + "learning_rate": 4.796926454445663e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6895096302032471, + "num_tokens": 22603394.0, + "step": 875 + }, + { + "epoch": 0.09620030748956732, + "grad_norm": 2.5203535556793213, + "learning_rate": 4.802414928649835e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6919255256652832, + "num_tokens": 22625624.0, + "step": 876 + }, + { + "epoch": 0.09631012519218098, + "grad_norm": 2.2454261779785156, + "learning_rate": 4.807903402854007e-07, + "loss": 1.0187, + "mean_token_accuracy": 0.691617488861084, + "num_tokens": 22652471.0, + "step": 877 + }, + { + "epoch": 0.09641994289479464, + "grad_norm": 2.4886252880096436, + "learning_rate": 4.813391877058177e-07, + "loss": 1.0123, + "mean_token_accuracy": 0.7014565467834473, + "num_tokens": 22676106.0, + "step": 878 + }, + { + "epoch": 0.0965297605974083, + "grad_norm": 2.409389019012451, + "learning_rate": 4.818880351262349e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.680571436882019, + "num_tokens": 22701106.0, + "step": 879 + }, + { + "epoch": 0.09663957830002197, + "grad_norm": 2.5238075256347656, + "learning_rate": 4.82436882546652e-07, + "loss": 1.0855, + "mean_token_accuracy": 0.6808409690856934, + "num_tokens": 22724172.0, + "step": 880 + }, + { + "epoch": 0.09674939600263563, + "grad_norm": 2.5698060989379883, + "learning_rate": 4.829857299670691e-07, + "loss": 1.0806, + "mean_token_accuracy": 0.6755663156509399, + "num_tokens": 22745983.0, + "step": 881 + }, + { + "epoch": 0.09685921370524929, + "grad_norm": 2.263284206390381, + "learning_rate": 4.835345773874863e-07, + "loss": 1.0372, + "mean_token_accuracy": 0.692600667476654, + "num_tokens": 22774421.0, + "step": 882 + }, + { + "epoch": 0.09696903140786295, + "grad_norm": 2.203406572341919, + "learning_rate": 4.840834248079034e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6906141638755798, + "num_tokens": 22803960.0, + "step": 883 + }, + { + "epoch": 0.09707884911047661, + "grad_norm": 2.5759336948394775, + "learning_rate": 4.846322722283204e-07, + "loss": 0.9997, + "mean_token_accuracy": 0.7066411972045898, + "num_tokens": 22826819.0, + "step": 884 + }, + { + "epoch": 0.09718866681309027, + "grad_norm": 2.3675663471221924, + "learning_rate": 4.851811196487376e-07, + "loss": 1.0004, + "mean_token_accuracy": 0.6990071535110474, + "num_tokens": 22852710.0, + "step": 885 + }, + { + "epoch": 0.09729848451570393, + "grad_norm": 2.0014405250549316, + "learning_rate": 4.857299670691548e-07, + "loss": 1.0332, + "mean_token_accuracy": 0.6967533826828003, + "num_tokens": 22885104.0, + "step": 886 + }, + { + "epoch": 0.09740830221831759, + "grad_norm": 2.8343355655670166, + "learning_rate": 4.862788144895718e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.6808068752288818, + "num_tokens": 22905667.0, + "step": 887 + }, + { + "epoch": 0.09751811992093125, + "grad_norm": 2.2697062492370605, + "learning_rate": 4.86827661909989e-07, + "loss": 1.0307, + "mean_token_accuracy": 0.6910250186920166, + "num_tokens": 22933692.0, + "step": 888 + }, + { + "epoch": 0.09762793762354492, + "grad_norm": 2.385450839996338, + "learning_rate": 4.873765093304062e-07, + "loss": 0.965, + "mean_token_accuracy": 0.7157467603683472, + "num_tokens": 22959847.0, + "step": 889 + }, + { + "epoch": 0.09773775532615858, + "grad_norm": 2.4662251472473145, + "learning_rate": 4.879253567508232e-07, + "loss": 0.9763, + "mean_token_accuracy": 0.7006193399429321, + "num_tokens": 22983362.0, + "step": 890 + }, + { + "epoch": 0.09784757302877224, + "grad_norm": 2.3256847858428955, + "learning_rate": 4.884742041712404e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6884794235229492, + "num_tokens": 23009525.0, + "step": 891 + }, + { + "epoch": 0.0979573907313859, + "grad_norm": 2.24880051612854, + "learning_rate": 4.890230515916576e-07, + "loss": 1.0921, + "mean_token_accuracy": 0.6784542798995972, + "num_tokens": 23036355.0, + "step": 892 + }, + { + "epoch": 0.09806720843399956, + "grad_norm": 2.229644775390625, + "learning_rate": 4.895718990120746e-07, + "loss": 1.2256, + "mean_token_accuracy": 0.6470890641212463, + "num_tokens": 23067209.0, + "step": 893 + }, + { + "epoch": 0.09817702613661322, + "grad_norm": 2.3723032474517822, + "learning_rate": 4.901207464324917e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6959242224693298, + "num_tokens": 23091740.0, + "step": 894 + }, + { + "epoch": 0.09828684383922688, + "grad_norm": 2.6809427738189697, + "learning_rate": 4.906695938529089e-07, + "loss": 1.0944, + "mean_token_accuracy": 0.6740816235542297, + "num_tokens": 23113957.0, + "step": 895 + }, + { + "epoch": 0.09839666154184054, + "grad_norm": 2.4566028118133545, + "learning_rate": 4.91218441273326e-07, + "loss": 0.9261, + "mean_token_accuracy": 0.7228519916534424, + "num_tokens": 23136971.0, + "step": 896 + }, + { + "epoch": 0.0985064792444542, + "grad_norm": 2.3327155113220215, + "learning_rate": 4.917672886937431e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.6849958300590515, + "num_tokens": 23163216.0, + "step": 897 + }, + { + "epoch": 0.09861629694706786, + "grad_norm": 2.33640193939209, + "learning_rate": 4.923161361141603e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6730111837387085, + "num_tokens": 23188018.0, + "step": 898 + }, + { + "epoch": 0.09872611464968153, + "grad_norm": 2.39003849029541, + "learning_rate": 4.928649835345773e-07, + "loss": 1.1641, + "mean_token_accuracy": 0.6844984292984009, + "num_tokens": 23213616.0, + "step": 899 + }, + { + "epoch": 0.0988359323522952, + "grad_norm": 2.3735218048095703, + "learning_rate": 4.934138309549945e-07, + "loss": 1.1343, + "mean_token_accuracy": 0.6658610701560974, + "num_tokens": 23242087.0, + "step": 900 + }, + { + "epoch": 0.09894575005490885, + "grad_norm": 2.4040098190307617, + "learning_rate": 4.939626783754117e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.672886848449707, + "num_tokens": 23267010.0, + "step": 901 + }, + { + "epoch": 0.09905556775752251, + "grad_norm": 2.104870557785034, + "learning_rate": 4.945115257958287e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6797997951507568, + "num_tokens": 23299140.0, + "step": 902 + }, + { + "epoch": 0.09916538546013617, + "grad_norm": 2.3099255561828613, + "learning_rate": 4.950603732162459e-07, + "loss": 0.955, + "mean_token_accuracy": 0.7139395475387573, + "num_tokens": 23324304.0, + "step": 903 + }, + { + "epoch": 0.09927520316274983, + "grad_norm": 2.570446252822876, + "learning_rate": 4.95609220636663e-07, + "loss": 1.0581, + "mean_token_accuracy": 0.6930422782897949, + "num_tokens": 23347361.0, + "step": 904 + }, + { + "epoch": 0.0993850208653635, + "grad_norm": 2.1447319984436035, + "learning_rate": 4.961580680570801e-07, + "loss": 1.1222, + "mean_token_accuracy": 0.6736675500869751, + "num_tokens": 23377603.0, + "step": 905 + }, + { + "epoch": 0.09949483856797715, + "grad_norm": 2.3549675941467285, + "learning_rate": 4.967069154774972e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6766946315765381, + "num_tokens": 23404986.0, + "step": 906 + }, + { + "epoch": 0.09960465627059081, + "grad_norm": 2.297924757003784, + "learning_rate": 4.972557628979143e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6764026880264282, + "num_tokens": 23430946.0, + "step": 907 + }, + { + "epoch": 0.09971447397320449, + "grad_norm": 2.2865540981292725, + "learning_rate": 4.978046103183315e-07, + "loss": 1.1049, + "mean_token_accuracy": 0.676159679889679, + "num_tokens": 23458584.0, + "step": 908 + }, + { + "epoch": 0.09982429167581815, + "grad_norm": 2.1490983963012695, + "learning_rate": 4.983534577387486e-07, + "loss": 1.0791, + "mean_token_accuracy": 0.6850111484527588, + "num_tokens": 23487578.0, + "step": 909 + }, + { + "epoch": 0.0999341093784318, + "grad_norm": 2.04343581199646, + "learning_rate": 4.989023051591657e-07, + "loss": 1.1128, + "mean_token_accuracy": 0.681627631187439, + "num_tokens": 23517487.0, + "step": 910 + }, + { + "epoch": 0.10004392708104547, + "grad_norm": 2.4713072776794434, + "learning_rate": 4.994511525795829e-07, + "loss": 1.0649, + "mean_token_accuracy": 0.6810344457626343, + "num_tokens": 23541171.0, + "step": 911 + }, + { + "epoch": 0.10015374478365913, + "grad_norm": 2.0386526584625244, + "learning_rate": 5e-07, + "loss": 1.1729, + "mean_token_accuracy": 0.6618831753730774, + "num_tokens": 23574514.0, + "step": 912 + }, + { + "epoch": 0.10026356248627279, + "grad_norm": 2.5746078491210938, + "learning_rate": 5.005488474204171e-07, + "loss": 1.14, + "mean_token_accuracy": 0.6668972373008728, + "num_tokens": 23597987.0, + "step": 913 + }, + { + "epoch": 0.10037338018888645, + "grad_norm": 2.5002658367156982, + "learning_rate": 5.010976948408342e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.6974290609359741, + "num_tokens": 23620987.0, + "step": 914 + }, + { + "epoch": 0.1004831978915001, + "grad_norm": 2.0655431747436523, + "learning_rate": 5.016465422612514e-07, + "loss": 1.112, + "mean_token_accuracy": 0.6749048829078674, + "num_tokens": 23652813.0, + "step": 915 + }, + { + "epoch": 0.10059301559411377, + "grad_norm": 2.3903536796569824, + "learning_rate": 5.021953896816685e-07, + "loss": 1.0451, + "mean_token_accuracy": 0.6874322891235352, + "num_tokens": 23676923.0, + "step": 916 + }, + { + "epoch": 0.10070283329672743, + "grad_norm": 2.491032600402832, + "learning_rate": 5.027442371020856e-07, + "loss": 1.0847, + "mean_token_accuracy": 0.6840937733650208, + "num_tokens": 23701849.0, + "step": 917 + }, + { + "epoch": 0.1008126509993411, + "grad_norm": 2.3573169708251953, + "learning_rate": 5.032930845225028e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.6998852491378784, + "num_tokens": 23728129.0, + "step": 918 + }, + { + "epoch": 0.10092246870195476, + "grad_norm": 2.185824155807495, + "learning_rate": 5.038419319429198e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6882895231246948, + "num_tokens": 23756554.0, + "step": 919 + }, + { + "epoch": 0.10103228640456842, + "grad_norm": 2.13875150680542, + "learning_rate": 5.04390779363337e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6939374208450317, + "num_tokens": 23785959.0, + "step": 920 + }, + { + "epoch": 0.10114210410718208, + "grad_norm": 2.3466384410858154, + "learning_rate": 5.049396267837542e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6954675912857056, + "num_tokens": 23812465.0, + "step": 921 + }, + { + "epoch": 0.10125192180979574, + "grad_norm": 2.3361005783081055, + "learning_rate": 5.054884742041711e-07, + "loss": 1.0762, + "mean_token_accuracy": 0.6776321530342102, + "num_tokens": 23836833.0, + "step": 922 + }, + { + "epoch": 0.1013617395124094, + "grad_norm": 2.3890905380249023, + "learning_rate": 5.060373216245883e-07, + "loss": 1.0536, + "mean_token_accuracy": 0.6862937211990356, + "num_tokens": 23861514.0, + "step": 923 + }, + { + "epoch": 0.10147155721502306, + "grad_norm": 2.04964542388916, + "learning_rate": 5.065861690450055e-07, + "loss": 1.1439, + "mean_token_accuracy": 0.664642333984375, + "num_tokens": 23892541.0, + "step": 924 + }, + { + "epoch": 0.10158137491763672, + "grad_norm": 2.3362863063812256, + "learning_rate": 5.071350164654225e-07, + "loss": 1.1461, + "mean_token_accuracy": 0.6625150442123413, + "num_tokens": 23921253.0, + "step": 925 + }, + { + "epoch": 0.10169119262025038, + "grad_norm": 2.8369698524475098, + "learning_rate": 5.076838638858397e-07, + "loss": 1.0091, + "mean_token_accuracy": 0.6982686519622803, + "num_tokens": 23940484.0, + "step": 926 + }, + { + "epoch": 0.10180101032286405, + "grad_norm": 2.4899537563323975, + "learning_rate": 5.082327113062569e-07, + "loss": 1.0222, + "mean_token_accuracy": 0.6920459270477295, + "num_tokens": 23961806.0, + "step": 927 + }, + { + "epoch": 0.10191082802547771, + "grad_norm": 2.1436519622802734, + "learning_rate": 5.087815587266739e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6890578269958496, + "num_tokens": 23991456.0, + "step": 928 + }, + { + "epoch": 0.10202064572809137, + "grad_norm": 2.4377989768981934, + "learning_rate": 5.093304061470911e-07, + "loss": 1.0885, + "mean_token_accuracy": 0.6738262176513672, + "num_tokens": 24017001.0, + "step": 929 + }, + { + "epoch": 0.10213046343070503, + "grad_norm": 2.5516793727874756, + "learning_rate": 5.098792535675082e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.690678596496582, + "num_tokens": 24040892.0, + "step": 930 + }, + { + "epoch": 0.10224028113331869, + "grad_norm": 2.304048776626587, + "learning_rate": 5.104281009879253e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6805844306945801, + "num_tokens": 24071805.0, + "step": 931 + }, + { + "epoch": 0.10235009883593235, + "grad_norm": 2.408301591873169, + "learning_rate": 5.109769484083425e-07, + "loss": 1.1323, + "mean_token_accuracy": 0.6679486036300659, + "num_tokens": 24098556.0, + "step": 932 + }, + { + "epoch": 0.10245991653854601, + "grad_norm": 2.3642876148223877, + "learning_rate": 5.115257958287596e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.6885722875595093, + "num_tokens": 24125908.0, + "step": 933 + }, + { + "epoch": 0.10256973424115967, + "grad_norm": 2.42523193359375, + "learning_rate": 5.120746432491767e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6884674429893494, + "num_tokens": 24150080.0, + "step": 934 + }, + { + "epoch": 0.10267955194377333, + "grad_norm": 2.3273723125457764, + "learning_rate": 5.126234906695939e-07, + "loss": 1.1287, + "mean_token_accuracy": 0.6741236448287964, + "num_tokens": 24176761.0, + "step": 935 + }, + { + "epoch": 0.10278936964638699, + "grad_norm": 2.4775586128234863, + "learning_rate": 5.13172338090011e-07, + "loss": 1.0819, + "mean_token_accuracy": 0.679159939289093, + "num_tokens": 24202239.0, + "step": 936 + }, + { + "epoch": 0.10289918734900066, + "grad_norm": 2.3629651069641113, + "learning_rate": 5.137211855104281e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6790528893470764, + "num_tokens": 24229532.0, + "step": 937 + }, + { + "epoch": 0.10300900505161432, + "grad_norm": 2.636235475540161, + "learning_rate": 5.142700329308453e-07, + "loss": 1.0155, + "mean_token_accuracy": 0.6955170631408691, + "num_tokens": 24250730.0, + "step": 938 + }, + { + "epoch": 0.10311882275422798, + "grad_norm": 2.772397518157959, + "learning_rate": 5.148188803512624e-07, + "loss": 0.9277, + "mean_token_accuracy": 0.7184268236160278, + "num_tokens": 24269656.0, + "step": 939 + }, + { + "epoch": 0.10322864045684164, + "grad_norm": 2.2884907722473145, + "learning_rate": 5.153677277716795e-07, + "loss": 1.035, + "mean_token_accuracy": 0.6938912272453308, + "num_tokens": 24295177.0, + "step": 940 + }, + { + "epoch": 0.1033384581594553, + "grad_norm": 2.46026873588562, + "learning_rate": 5.159165751920965e-07, + "loss": 0.9938, + "mean_token_accuracy": 0.7037017345428467, + "num_tokens": 24320724.0, + "step": 941 + }, + { + "epoch": 0.10344827586206896, + "grad_norm": 2.145505905151367, + "learning_rate": 5.164654226125136e-07, + "loss": 1.1559, + "mean_token_accuracy": 0.6582547426223755, + "num_tokens": 24351940.0, + "step": 942 + }, + { + "epoch": 0.10355809356468262, + "grad_norm": 2.2345452308654785, + "learning_rate": 5.170142700329308e-07, + "loss": 1.1163, + "mean_token_accuracy": 0.6699709892272949, + "num_tokens": 24380954.0, + "step": 943 + }, + { + "epoch": 0.10366791126729628, + "grad_norm": 2.408332586288452, + "learning_rate": 5.175631174533479e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6772923469543457, + "num_tokens": 24405316.0, + "step": 944 + }, + { + "epoch": 0.10377772896990994, + "grad_norm": 2.1422882080078125, + "learning_rate": 5.18111964873765e-07, + "loss": 1.024, + "mean_token_accuracy": 0.6951241493225098, + "num_tokens": 24435103.0, + "step": 945 + }, + { + "epoch": 0.10388754667252362, + "grad_norm": 2.208674430847168, + "learning_rate": 5.186608122941822e-07, + "loss": 0.9701, + "mean_token_accuracy": 0.7046706080436707, + "num_tokens": 24462144.0, + "step": 946 + }, + { + "epoch": 0.10399736437513728, + "grad_norm": 2.3757431507110596, + "learning_rate": 5.192096597145993e-07, + "loss": 1.0747, + "mean_token_accuracy": 0.6820845603942871, + "num_tokens": 24486997.0, + "step": 947 + }, + { + "epoch": 0.10410718207775094, + "grad_norm": 2.5838124752044678, + "learning_rate": 5.197585071350164e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.7013704776763916, + "num_tokens": 24508533.0, + "step": 948 + }, + { + "epoch": 0.1042169997803646, + "grad_norm": 2.3843400478363037, + "learning_rate": 5.203073545554336e-07, + "loss": 1.1265, + "mean_token_accuracy": 0.6588296294212341, + "num_tokens": 24532559.0, + "step": 949 + }, + { + "epoch": 0.10432681748297826, + "grad_norm": 2.1470398902893066, + "learning_rate": 5.208562019758507e-07, + "loss": 1.0974, + "mean_token_accuracy": 0.6717758774757385, + "num_tokens": 24562068.0, + "step": 950 + }, + { + "epoch": 0.10443663518559192, + "grad_norm": 2.291306495666504, + "learning_rate": 5.214050493962678e-07, + "loss": 1.1125, + "mean_token_accuracy": 0.6745741367340088, + "num_tokens": 24588448.0, + "step": 951 + }, + { + "epoch": 0.10454645288820558, + "grad_norm": 2.8084044456481934, + "learning_rate": 5.219538968166849e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6872484683990479, + "num_tokens": 24608731.0, + "step": 952 + }, + { + "epoch": 0.10465627059081924, + "grad_norm": 2.223191261291504, + "learning_rate": 5.225027442371021e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.6923182606697083, + "num_tokens": 24635366.0, + "step": 953 + }, + { + "epoch": 0.1047660882934329, + "grad_norm": 2.4219863414764404, + "learning_rate": 5.230515916575192e-07, + "loss": 0.8907, + "mean_token_accuracy": 0.7197365760803223, + "num_tokens": 24657349.0, + "step": 954 + }, + { + "epoch": 0.10487590599604656, + "grad_norm": 2.715456962585449, + "learning_rate": 5.236004390779363e-07, + "loss": 1.0009, + "mean_token_accuracy": 0.698122501373291, + "num_tokens": 24677392.0, + "step": 955 + }, + { + "epoch": 0.10498572369866023, + "grad_norm": 2.3436758518218994, + "learning_rate": 5.241492864983535e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6879056692123413, + "num_tokens": 24702485.0, + "step": 956 + }, + { + "epoch": 0.10509554140127389, + "grad_norm": 2.368443727493286, + "learning_rate": 5.246981339187706e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6758286356925964, + "num_tokens": 24729892.0, + "step": 957 + }, + { + "epoch": 0.10520535910388755, + "grad_norm": 2.191972017288208, + "learning_rate": 5.252469813391877e-07, + "loss": 1.0344, + "mean_token_accuracy": 0.6897286176681519, + "num_tokens": 24757175.0, + "step": 958 + }, + { + "epoch": 0.10531517680650121, + "grad_norm": 2.4226138591766357, + "learning_rate": 5.257958287596049e-07, + "loss": 1.041, + "mean_token_accuracy": 0.6916676759719849, + "num_tokens": 24781546.0, + "step": 959 + }, + { + "epoch": 0.10542499450911487, + "grad_norm": 2.191119432449341, + "learning_rate": 5.263446761800219e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6701521277427673, + "num_tokens": 24808263.0, + "step": 960 + }, + { + "epoch": 0.10553481221172853, + "grad_norm": 2.1311819553375244, + "learning_rate": 5.26893523600439e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.6995785236358643, + "num_tokens": 24838494.0, + "step": 961 + }, + { + "epoch": 0.10564462991434219, + "grad_norm": 2.288217067718506, + "learning_rate": 5.274423710208562e-07, + "loss": 1.0098, + "mean_token_accuracy": 0.7038345336914062, + "num_tokens": 24865320.0, + "step": 962 + }, + { + "epoch": 0.10575444761695585, + "grad_norm": 2.3826394081115723, + "learning_rate": 5.279912184412732e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.685575008392334, + "num_tokens": 24889855.0, + "step": 963 + }, + { + "epoch": 0.10586426531956951, + "grad_norm": 2.1104602813720703, + "learning_rate": 5.285400658616904e-07, + "loss": 1.1094, + "mean_token_accuracy": 0.6752517223358154, + "num_tokens": 24921939.0, + "step": 964 + }, + { + "epoch": 0.10597408302218318, + "grad_norm": 2.360926628112793, + "learning_rate": 5.290889132821076e-07, + "loss": 1.063, + "mean_token_accuracy": 0.6946674585342407, + "num_tokens": 24945957.0, + "step": 965 + }, + { + "epoch": 0.10608390072479684, + "grad_norm": 2.2566568851470947, + "learning_rate": 5.296377607025246e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6949195265769958, + "num_tokens": 24972234.0, + "step": 966 + }, + { + "epoch": 0.1061937184274105, + "grad_norm": 2.0378077030181885, + "learning_rate": 5.301866081229418e-07, + "loss": 1.0633, + "mean_token_accuracy": 0.6798534393310547, + "num_tokens": 25002716.0, + "step": 967 + }, + { + "epoch": 0.10630353613002416, + "grad_norm": 2.3433594703674316, + "learning_rate": 5.30735455543359e-07, + "loss": 1.0401, + "mean_token_accuracy": 0.6915642619132996, + "num_tokens": 25028812.0, + "step": 968 + }, + { + "epoch": 0.10641335383263782, + "grad_norm": 2.578479051589966, + "learning_rate": 5.31284302963776e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.693821907043457, + "num_tokens": 25049725.0, + "step": 969 + }, + { + "epoch": 0.10652317153525148, + "grad_norm": 2.384901285171509, + "learning_rate": 5.318331503841932e-07, + "loss": 0.9607, + "mean_token_accuracy": 0.7103067636489868, + "num_tokens": 25074074.0, + "step": 970 + }, + { + "epoch": 0.10663298923786514, + "grad_norm": 2.197627544403076, + "learning_rate": 5.323819978046103e-07, + "loss": 1.0334, + "mean_token_accuracy": 0.6920002698898315, + "num_tokens": 25103495.0, + "step": 971 + }, + { + "epoch": 0.1067428069404788, + "grad_norm": 2.599336862564087, + "learning_rate": 5.329308452250274e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6913316249847412, + "num_tokens": 25124766.0, + "step": 972 + }, + { + "epoch": 0.10685262464309246, + "grad_norm": 2.0921521186828613, + "learning_rate": 5.334796926454446e-07, + "loss": 1.0098, + "mean_token_accuracy": 0.7009889483451843, + "num_tokens": 25154265.0, + "step": 973 + }, + { + "epoch": 0.10696244234570612, + "grad_norm": 2.149738073348999, + "learning_rate": 5.340285400658617e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6830941438674927, + "num_tokens": 25182473.0, + "step": 974 + }, + { + "epoch": 0.1070722600483198, + "grad_norm": 2.5825202465057373, + "learning_rate": 5.345773874862788e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6882308721542358, + "num_tokens": 25202744.0, + "step": 975 + }, + { + "epoch": 0.10718207775093345, + "grad_norm": 2.5278127193450928, + "learning_rate": 5.35126234906696e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7056394219398499, + "num_tokens": 25224343.0, + "step": 976 + }, + { + "epoch": 0.10729189545354711, + "grad_norm": 2.1866424083709717, + "learning_rate": 5.35675082327113e-07, + "loss": 1.0812, + "mean_token_accuracy": 0.6794875860214233, + "num_tokens": 25252914.0, + "step": 977 + }, + { + "epoch": 0.10740171315616077, + "grad_norm": 2.4703941345214844, + "learning_rate": 5.362239297475302e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6718863248825073, + "num_tokens": 25275121.0, + "step": 978 + }, + { + "epoch": 0.10751153085877443, + "grad_norm": 2.3792672157287598, + "learning_rate": 5.367727771679473e-07, + "loss": 1.0295, + "mean_token_accuracy": 0.6957986354827881, + "num_tokens": 25301247.0, + "step": 979 + }, + { + "epoch": 0.1076213485613881, + "grad_norm": 2.4029555320739746, + "learning_rate": 5.373216245883643e-07, + "loss": 0.9857, + "mean_token_accuracy": 0.7019932270050049, + "num_tokens": 25325152.0, + "step": 980 + }, + { + "epoch": 0.10773116626400175, + "grad_norm": 2.1777031421661377, + "learning_rate": 5.378704720087815e-07, + "loss": 1.135, + "mean_token_accuracy": 0.6739586591720581, + "num_tokens": 25355553.0, + "step": 981 + }, + { + "epoch": 0.10784098396661541, + "grad_norm": 2.086376428604126, + "learning_rate": 5.384193194291986e-07, + "loss": 1.0601, + "mean_token_accuracy": 0.6942222118377686, + "num_tokens": 25385478.0, + "step": 982 + }, + { + "epoch": 0.10795080166922907, + "grad_norm": 2.226729154586792, + "learning_rate": 5.389681668496157e-07, + "loss": 1.1651, + "mean_token_accuracy": 0.6656048893928528, + "num_tokens": 25415349.0, + "step": 983 + }, + { + "epoch": 0.10806061937184275, + "grad_norm": 2.1168484687805176, + "learning_rate": 5.395170142700329e-07, + "loss": 1.0689, + "mean_token_accuracy": 0.6786885857582092, + "num_tokens": 25444265.0, + "step": 984 + }, + { + "epoch": 0.10817043707445641, + "grad_norm": 2.4153573513031006, + "learning_rate": 5.4006586169045e-07, + "loss": 1.0435, + "mean_token_accuracy": 0.6923635005950928, + "num_tokens": 25469508.0, + "step": 985 + }, + { + "epoch": 0.10828025477707007, + "grad_norm": 2.491246461868286, + "learning_rate": 5.406147091108671e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6823179721832275, + "num_tokens": 25492167.0, + "step": 986 + }, + { + "epoch": 0.10839007247968373, + "grad_norm": 2.4389712810516357, + "learning_rate": 5.411635565312843e-07, + "loss": 1.035, + "mean_token_accuracy": 0.6954505443572998, + "num_tokens": 25515220.0, + "step": 987 + }, + { + "epoch": 0.10849989018229739, + "grad_norm": 2.2491374015808105, + "learning_rate": 5.417124039517014e-07, + "loss": 1.1561, + "mean_token_accuracy": 0.6661851406097412, + "num_tokens": 25544608.0, + "step": 988 + }, + { + "epoch": 0.10860970788491105, + "grad_norm": 2.3387434482574463, + "learning_rate": 5.422612513721185e-07, + "loss": 1.071, + "mean_token_accuracy": 0.685352087020874, + "num_tokens": 25571261.0, + "step": 989 + }, + { + "epoch": 0.1087195255875247, + "grad_norm": 2.339442729949951, + "learning_rate": 5.428100987925357e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6917722225189209, + "num_tokens": 25597226.0, + "step": 990 + }, + { + "epoch": 0.10882934329013837, + "grad_norm": 2.453078508377075, + "learning_rate": 5.433589462129528e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.7021191120147705, + "num_tokens": 25620623.0, + "step": 991 + }, + { + "epoch": 0.10893916099275203, + "grad_norm": 2.3232357501983643, + "learning_rate": 5.439077936333699e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.6972337961196899, + "num_tokens": 25646081.0, + "step": 992 + }, + { + "epoch": 0.10904897869536569, + "grad_norm": 2.572950601577759, + "learning_rate": 5.44456641053787e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6925846338272095, + "num_tokens": 25668168.0, + "step": 993 + }, + { + "epoch": 0.10915879639797936, + "grad_norm": 2.0373432636260986, + "learning_rate": 5.450054884742042e-07, + "loss": 1.1304, + "mean_token_accuracy": 0.6613456606864929, + "num_tokens": 25700990.0, + "step": 994 + }, + { + "epoch": 0.10926861410059302, + "grad_norm": 2.506986379623413, + "learning_rate": 5.455543358946213e-07, + "loss": 0.9977, + "mean_token_accuracy": 0.7036257386207581, + "num_tokens": 25723336.0, + "step": 995 + }, + { + "epoch": 0.10937843180320668, + "grad_norm": 2.502851963043213, + "learning_rate": 5.461031833150384e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.706280529499054, + "num_tokens": 25745994.0, + "step": 996 + }, + { + "epoch": 0.10948824950582034, + "grad_norm": 2.4291980266571045, + "learning_rate": 5.466520307354556e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.687187671661377, + "num_tokens": 25770299.0, + "step": 997 + }, + { + "epoch": 0.109598067208434, + "grad_norm": 2.474764823913574, + "learning_rate": 5.472008781558726e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6942171454429626, + "num_tokens": 25792836.0, + "step": 998 + }, + { + "epoch": 0.10970788491104766, + "grad_norm": 2.4122095108032227, + "learning_rate": 5.477497255762897e-07, + "loss": 1.1718, + "mean_token_accuracy": 0.6700694561004639, + "num_tokens": 25820137.0, + "step": 999 + }, + { + "epoch": 0.10981770261366132, + "grad_norm": 2.1175360679626465, + "learning_rate": 5.482985729967069e-07, + "loss": 0.9904, + "mean_token_accuracy": 0.6990662813186646, + "num_tokens": 25847363.0, + "step": 1000 + }, + { + "epoch": 0.10992752031627498, + "grad_norm": 2.3544209003448486, + "learning_rate": 5.48847420417124e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.685491681098938, + "num_tokens": 25873573.0, + "step": 1001 + }, + { + "epoch": 0.11003733801888864, + "grad_norm": 2.1559383869171143, + "learning_rate": 5.493962678375411e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.6640099287033081, + "num_tokens": 25903805.0, + "step": 1002 + }, + { + "epoch": 0.11014715572150231, + "grad_norm": 2.3067400455474854, + "learning_rate": 5.499451152579583e-07, + "loss": 1.092, + "mean_token_accuracy": 0.6814125776290894, + "num_tokens": 25931510.0, + "step": 1003 + }, + { + "epoch": 0.11025697342411597, + "grad_norm": 2.2112436294555664, + "learning_rate": 5.504939626783753e-07, + "loss": 1.013, + "mean_token_accuracy": 0.6942495107650757, + "num_tokens": 25958538.0, + "step": 1004 + }, + { + "epoch": 0.11036679112672963, + "grad_norm": 2.3788836002349854, + "learning_rate": 5.510428100987925e-07, + "loss": 1.155, + "mean_token_accuracy": 0.655595064163208, + "num_tokens": 25983806.0, + "step": 1005 + }, + { + "epoch": 0.11047660882934329, + "grad_norm": 2.777811050415039, + "learning_rate": 5.515916575192097e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.7014333009719849, + "num_tokens": 26001943.0, + "step": 1006 + }, + { + "epoch": 0.11058642653195695, + "grad_norm": 2.377342462539673, + "learning_rate": 5.521405049396267e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6852543354034424, + "num_tokens": 26027323.0, + "step": 1007 + }, + { + "epoch": 0.11069624423457061, + "grad_norm": 2.3581089973449707, + "learning_rate": 5.526893523600439e-07, + "loss": 0.9656, + "mean_token_accuracy": 0.7067149877548218, + "num_tokens": 26050621.0, + "step": 1008 + }, + { + "epoch": 0.11080606193718427, + "grad_norm": 2.7038748264312744, + "learning_rate": 5.532381997804611e-07, + "loss": 0.9706, + "mean_token_accuracy": 0.7094982266426086, + "num_tokens": 26069705.0, + "step": 1009 + }, + { + "epoch": 0.11091587963979793, + "grad_norm": 2.343278169631958, + "learning_rate": 5.537870472008781e-07, + "loss": 1.0886, + "mean_token_accuracy": 0.6772884130477905, + "num_tokens": 26094577.0, + "step": 1010 + }, + { + "epoch": 0.11102569734241159, + "grad_norm": 2.440876007080078, + "learning_rate": 5.543358946212953e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.7046694159507751, + "num_tokens": 26117171.0, + "step": 1011 + }, + { + "epoch": 0.11113551504502525, + "grad_norm": 2.3089096546173096, + "learning_rate": 5.548847420417125e-07, + "loss": 1.0975, + "mean_token_accuracy": 0.6727151870727539, + "num_tokens": 26143835.0, + "step": 1012 + }, + { + "epoch": 0.11124533274763893, + "grad_norm": 2.0489110946655273, + "learning_rate": 5.554335894621295e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6937357783317566, + "num_tokens": 26175882.0, + "step": 1013 + }, + { + "epoch": 0.11135515045025259, + "grad_norm": 2.1561882495880127, + "learning_rate": 5.559824368825467e-07, + "loss": 1.1256, + "mean_token_accuracy": 0.6706752777099609, + "num_tokens": 26205817.0, + "step": 1014 + }, + { + "epoch": 0.11146496815286625, + "grad_norm": 2.257444143295288, + "learning_rate": 5.565312843029637e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.6870068311691284, + "num_tokens": 26232145.0, + "step": 1015 + }, + { + "epoch": 0.1115747858554799, + "grad_norm": 2.5641040802001953, + "learning_rate": 5.570801317233809e-07, + "loss": 0.9615, + "mean_token_accuracy": 0.7104686498641968, + "num_tokens": 26252234.0, + "step": 1016 + }, + { + "epoch": 0.11168460355809356, + "grad_norm": 2.3154335021972656, + "learning_rate": 5.57628979143798e-07, + "loss": 1.0589, + "mean_token_accuracy": 0.6857889890670776, + "num_tokens": 26278937.0, + "step": 1017 + }, + { + "epoch": 0.11179442126070722, + "grad_norm": 2.401758909225464, + "learning_rate": 5.58177826564215e-07, + "loss": 0.9942, + "mean_token_accuracy": 0.6943782567977905, + "num_tokens": 26302709.0, + "step": 1018 + }, + { + "epoch": 0.11190423896332088, + "grad_norm": 2.394007682800293, + "learning_rate": 5.587266739846322e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.7014590501785278, + "num_tokens": 26327001.0, + "step": 1019 + }, + { + "epoch": 0.11201405666593454, + "grad_norm": 2.2017905712127686, + "learning_rate": 5.592755214050494e-07, + "loss": 1.1276, + "mean_token_accuracy": 0.6669204235076904, + "num_tokens": 26355672.0, + "step": 1020 + }, + { + "epoch": 0.1121238743685482, + "grad_norm": 2.318930149078369, + "learning_rate": 5.598243688254664e-07, + "loss": 1.1119, + "mean_token_accuracy": 0.6720311045646667, + "num_tokens": 26384221.0, + "step": 1021 + }, + { + "epoch": 0.11223369207116188, + "grad_norm": 2.5379738807678223, + "learning_rate": 5.603732162458836e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6810629367828369, + "num_tokens": 26408024.0, + "step": 1022 + }, + { + "epoch": 0.11234350977377554, + "grad_norm": 2.1701486110687256, + "learning_rate": 5.609220636663008e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7090702056884766, + "num_tokens": 26434653.0, + "step": 1023 + }, + { + "epoch": 0.1124533274763892, + "grad_norm": 2.2974822521209717, + "learning_rate": 5.614709110867178e-07, + "loss": 1.0798, + "mean_token_accuracy": 0.6776479482650757, + "num_tokens": 26460392.0, + "step": 1024 + }, + { + "epoch": 0.11256314517900286, + "grad_norm": 2.2887959480285645, + "learning_rate": 5.62019758507135e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6910852193832397, + "num_tokens": 26487319.0, + "step": 1025 + }, + { + "epoch": 0.11267296288161652, + "grad_norm": 2.2344703674316406, + "learning_rate": 5.625686059275521e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.678625762462616, + "num_tokens": 26515077.0, + "step": 1026 + }, + { + "epoch": 0.11278278058423018, + "grad_norm": 2.4494314193725586, + "learning_rate": 5.631174533479692e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6837602853775024, + "num_tokens": 26538938.0, + "step": 1027 + }, + { + "epoch": 0.11289259828684384, + "grad_norm": 2.4960250854492188, + "learning_rate": 5.636663007683864e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6667686104774475, + "num_tokens": 26561778.0, + "step": 1028 + }, + { + "epoch": 0.1130024159894575, + "grad_norm": 2.3292548656463623, + "learning_rate": 5.642151481888035e-07, + "loss": 1.1772, + "mean_token_accuracy": 0.6551449298858643, + "num_tokens": 26590268.0, + "step": 1029 + }, + { + "epoch": 0.11311223369207116, + "grad_norm": 2.2337989807128906, + "learning_rate": 5.647639956092206e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6775429248809814, + "num_tokens": 26617560.0, + "step": 1030 + }, + { + "epoch": 0.11322205139468482, + "grad_norm": 2.5117225646972656, + "learning_rate": 5.653128430296378e-07, + "loss": 1.0317, + "mean_token_accuracy": 0.6992305517196655, + "num_tokens": 26640098.0, + "step": 1031 + }, + { + "epoch": 0.11333186909729849, + "grad_norm": 2.3357415199279785, + "learning_rate": 5.658616904500549e-07, + "loss": 1.0801, + "mean_token_accuracy": 0.6776042580604553, + "num_tokens": 26667142.0, + "step": 1032 + }, + { + "epoch": 0.11344168679991215, + "grad_norm": 2.115461826324463, + "learning_rate": 5.66410537870472e-07, + "loss": 1.1267, + "mean_token_accuracy": 0.672049880027771, + "num_tokens": 26698005.0, + "step": 1033 + }, + { + "epoch": 0.11355150450252581, + "grad_norm": 2.273535966873169, + "learning_rate": 5.669593852908892e-07, + "loss": 1.147, + "mean_token_accuracy": 0.6627565622329712, + "num_tokens": 26726899.0, + "step": 1034 + }, + { + "epoch": 0.11366132220513947, + "grad_norm": 2.389214038848877, + "learning_rate": 5.675082327113063e-07, + "loss": 1.01, + "mean_token_accuracy": 0.6962102055549622, + "num_tokens": 26750705.0, + "step": 1035 + }, + { + "epoch": 0.11377113990775313, + "grad_norm": 2.646601915359497, + "learning_rate": 5.680570801317233e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6919169425964355, + "num_tokens": 26771903.0, + "step": 1036 + }, + { + "epoch": 0.11388095761036679, + "grad_norm": 2.3187568187713623, + "learning_rate": 5.686059275521404e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.692684531211853, + "num_tokens": 26797385.0, + "step": 1037 + }, + { + "epoch": 0.11399077531298045, + "grad_norm": 2.525933265686035, + "learning_rate": 5.691547749725576e-07, + "loss": 1.0356, + "mean_token_accuracy": 0.6923536062240601, + "num_tokens": 26820459.0, + "step": 1038 + }, + { + "epoch": 0.11410059301559411, + "grad_norm": 2.2631447315216064, + "learning_rate": 5.697036223929747e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.6603925824165344, + "num_tokens": 26848972.0, + "step": 1039 + }, + { + "epoch": 0.11421041071820777, + "grad_norm": 2.271427631378174, + "learning_rate": 5.702524698133918e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.7000218629837036, + "num_tokens": 26873899.0, + "step": 1040 + }, + { + "epoch": 0.11432022842082144, + "grad_norm": 2.095316171646118, + "learning_rate": 5.70801317233809e-07, + "loss": 1.1691, + "mean_token_accuracy": 0.6509023904800415, + "num_tokens": 26907155.0, + "step": 1041 + }, + { + "epoch": 0.1144300461234351, + "grad_norm": 2.2181882858276367, + "learning_rate": 5.713501646542261e-07, + "loss": 1.0936, + "mean_token_accuracy": 0.6712911128997803, + "num_tokens": 26935075.0, + "step": 1042 + }, + { + "epoch": 0.11453986382604876, + "grad_norm": 1.9855057001113892, + "learning_rate": 5.718990120746432e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6767171025276184, + "num_tokens": 26967479.0, + "step": 1043 + }, + { + "epoch": 0.11464968152866242, + "grad_norm": 2.695836305618286, + "learning_rate": 5.724478594950604e-07, + "loss": 0.9446, + "mean_token_accuracy": 0.7113056182861328, + "num_tokens": 26985963.0, + "step": 1044 + }, + { + "epoch": 0.11475949923127608, + "grad_norm": 2.472322463989258, + "learning_rate": 5.729967069154775e-07, + "loss": 1.1157, + "mean_token_accuracy": 0.6701788902282715, + "num_tokens": 27010579.0, + "step": 1045 + }, + { + "epoch": 0.11486931693388974, + "grad_norm": 2.382490634918213, + "learning_rate": 5.735455543358946e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.7009318470954895, + "num_tokens": 27035278.0, + "step": 1046 + }, + { + "epoch": 0.1149791346365034, + "grad_norm": 2.14688777923584, + "learning_rate": 5.740944017563118e-07, + "loss": 1.0394, + "mean_token_accuracy": 0.6876481771469116, + "num_tokens": 27063950.0, + "step": 1047 + }, + { + "epoch": 0.11508895233911706, + "grad_norm": 2.1417453289031982, + "learning_rate": 5.746432491767288e-07, + "loss": 1.1326, + "mean_token_accuracy": 0.6635462641716003, + "num_tokens": 27093422.0, + "step": 1048 + }, + { + "epoch": 0.11519877004173072, + "grad_norm": 2.22611665725708, + "learning_rate": 5.75192096597146e-07, + "loss": 1.0955, + "mean_token_accuracy": 0.6721467971801758, + "num_tokens": 27122946.0, + "step": 1049 + }, + { + "epoch": 0.11530858774434438, + "grad_norm": 2.281369924545288, + "learning_rate": 5.757409440175632e-07, + "loss": 1.0172, + "mean_token_accuracy": 0.6937897205352783, + "num_tokens": 27149743.0, + "step": 1050 + }, + { + "epoch": 0.11541840544695806, + "grad_norm": 2.3429315090179443, + "learning_rate": 5.762897914379802e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6956003904342651, + "num_tokens": 27175666.0, + "step": 1051 + }, + { + "epoch": 0.11552822314957172, + "grad_norm": 2.457508087158203, + "learning_rate": 5.768386388583974e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.701339840888977, + "num_tokens": 27200894.0, + "step": 1052 + }, + { + "epoch": 0.11563804085218538, + "grad_norm": 2.430260181427002, + "learning_rate": 5.773874862788145e-07, + "loss": 1.1693, + "mean_token_accuracy": 0.6614553928375244, + "num_tokens": 27226994.0, + "step": 1053 + }, + { + "epoch": 0.11574785855479904, + "grad_norm": 2.2098450660705566, + "learning_rate": 5.779363336992316e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6639920473098755, + "num_tokens": 27256491.0, + "step": 1054 + }, + { + "epoch": 0.1158576762574127, + "grad_norm": 2.1067872047424316, + "learning_rate": 5.784851811196487e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6694446802139282, + "num_tokens": 27289247.0, + "step": 1055 + }, + { + "epoch": 0.11596749396002635, + "grad_norm": 1.954301357269287, + "learning_rate": 5.790340285400658e-07, + "loss": 1.0727, + "mean_token_accuracy": 0.688080370426178, + "num_tokens": 27323157.0, + "step": 1056 + }, + { + "epoch": 0.11607731166264001, + "grad_norm": 2.5577988624572754, + "learning_rate": 5.795828759604829e-07, + "loss": 1.1899, + "mean_token_accuracy": 0.6551192998886108, + "num_tokens": 27345497.0, + "step": 1057 + }, + { + "epoch": 0.11618712936525367, + "grad_norm": 2.3781073093414307, + "learning_rate": 5.801317233809001e-07, + "loss": 1.1043, + "mean_token_accuracy": 0.6880725622177124, + "num_tokens": 27371208.0, + "step": 1058 + }, + { + "epoch": 0.11629694706786733, + "grad_norm": 2.4016318321228027, + "learning_rate": 5.806805708013171e-07, + "loss": 1.0902, + "mean_token_accuracy": 0.6783716678619385, + "num_tokens": 27397302.0, + "step": 1059 + }, + { + "epoch": 0.11640676477048101, + "grad_norm": 2.403519630432129, + "learning_rate": 5.812294182217343e-07, + "loss": 1.051, + "mean_token_accuracy": 0.6899334192276001, + "num_tokens": 27423412.0, + "step": 1060 + }, + { + "epoch": 0.11651658247309467, + "grad_norm": 2.3044519424438477, + "learning_rate": 5.817782656421515e-07, + "loss": 1.0624, + "mean_token_accuracy": 0.6877272129058838, + "num_tokens": 27448441.0, + "step": 1061 + }, + { + "epoch": 0.11662640017570833, + "grad_norm": 2.6762521266937256, + "learning_rate": 5.823271130625685e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.6899025440216064, + "num_tokens": 27470970.0, + "step": 1062 + }, + { + "epoch": 0.11673621787832199, + "grad_norm": 2.3453991413116455, + "learning_rate": 5.828759604829857e-07, + "loss": 1.03, + "mean_token_accuracy": 0.694213330745697, + "num_tokens": 27496924.0, + "step": 1063 + }, + { + "epoch": 0.11684603558093565, + "grad_norm": 2.3866727352142334, + "learning_rate": 5.834248079034029e-07, + "loss": 1.0649, + "mean_token_accuracy": 0.6872424483299255, + "num_tokens": 27521560.0, + "step": 1064 + }, + { + "epoch": 0.11695585328354931, + "grad_norm": 2.1545844078063965, + "learning_rate": 5.839736553238199e-07, + "loss": 1.1512, + "mean_token_accuracy": 0.6751447916030884, + "num_tokens": 27552500.0, + "step": 1065 + }, + { + "epoch": 0.11706567098616297, + "grad_norm": 2.111337423324585, + "learning_rate": 5.845225027442371e-07, + "loss": 1.1524, + "mean_token_accuracy": 0.6590453386306763, + "num_tokens": 27582605.0, + "step": 1066 + }, + { + "epoch": 0.11717548868877663, + "grad_norm": 2.466740131378174, + "learning_rate": 5.850713501646543e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6980542540550232, + "num_tokens": 27607180.0, + "step": 1067 + }, + { + "epoch": 0.11728530639139029, + "grad_norm": 2.111295223236084, + "learning_rate": 5.856201975850713e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.679385244846344, + "num_tokens": 27636686.0, + "step": 1068 + }, + { + "epoch": 0.11739512409400395, + "grad_norm": 2.083693742752075, + "learning_rate": 5.861690450054885e-07, + "loss": 1.1598, + "mean_token_accuracy": 0.6575663685798645, + "num_tokens": 27669033.0, + "step": 1069 + }, + { + "epoch": 0.11750494179661762, + "grad_norm": 2.2448437213897705, + "learning_rate": 5.867178924259056e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.695425808429718, + "num_tokens": 27695232.0, + "step": 1070 + }, + { + "epoch": 0.11761475949923128, + "grad_norm": 2.579232931137085, + "learning_rate": 5.872667398463227e-07, + "loss": 1.0037, + "mean_token_accuracy": 0.6965556144714355, + "num_tokens": 27715981.0, + "step": 1071 + }, + { + "epoch": 0.11772457720184494, + "grad_norm": 2.171947717666626, + "learning_rate": 5.878155872667399e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6778603792190552, + "num_tokens": 27744682.0, + "step": 1072 + }, + { + "epoch": 0.1178343949044586, + "grad_norm": 2.66711163520813, + "learning_rate": 5.88364434687157e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6883954405784607, + "num_tokens": 27765572.0, + "step": 1073 + }, + { + "epoch": 0.11794421260707226, + "grad_norm": 2.5503108501434326, + "learning_rate": 5.88913282107574e-07, + "loss": 1.1469, + "mean_token_accuracy": 0.6795501708984375, + "num_tokens": 27788212.0, + "step": 1074 + }, + { + "epoch": 0.11805403030968592, + "grad_norm": 2.226907730102539, + "learning_rate": 5.894621295279912e-07, + "loss": 1.0378, + "mean_token_accuracy": 0.6907957196235657, + "num_tokens": 27816040.0, + "step": 1075 + }, + { + "epoch": 0.11816384801229958, + "grad_norm": 2.153531789779663, + "learning_rate": 5.900109769484083e-07, + "loss": 1.0638, + "mean_token_accuracy": 0.6807935237884521, + "num_tokens": 27843959.0, + "step": 1076 + }, + { + "epoch": 0.11827366571491324, + "grad_norm": 2.351581573486328, + "learning_rate": 5.905598243688254e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6789862513542175, + "num_tokens": 27869877.0, + "step": 1077 + }, + { + "epoch": 0.1183834834175269, + "grad_norm": 2.5989465713500977, + "learning_rate": 5.911086717892426e-07, + "loss": 1.1536, + "mean_token_accuracy": 0.666350781917572, + "num_tokens": 27893291.0, + "step": 1078 + }, + { + "epoch": 0.11849330112014057, + "grad_norm": 2.4359140396118164, + "learning_rate": 5.916575192096597e-07, + "loss": 1.1494, + "mean_token_accuracy": 0.6601415872573853, + "num_tokens": 27918164.0, + "step": 1079 + }, + { + "epoch": 0.11860311882275423, + "grad_norm": 2.053220510482788, + "learning_rate": 5.922063666300768e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.6768027544021606, + "num_tokens": 27950644.0, + "step": 1080 + }, + { + "epoch": 0.1187129365253679, + "grad_norm": 2.2682905197143555, + "learning_rate": 5.927552140504939e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.6984346508979797, + "num_tokens": 27975780.0, + "step": 1081 + }, + { + "epoch": 0.11882275422798155, + "grad_norm": 2.5469717979431152, + "learning_rate": 5.93304061470911e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6813838481903076, + "num_tokens": 27998741.0, + "step": 1082 + }, + { + "epoch": 0.11893257193059521, + "grad_norm": 2.386213779449463, + "learning_rate": 5.938529088913282e-07, + "loss": 1.0812, + "mean_token_accuracy": 0.684606671333313, + "num_tokens": 28023828.0, + "step": 1083 + }, + { + "epoch": 0.11904238963320887, + "grad_norm": 2.2877233028411865, + "learning_rate": 5.944017563117453e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.6781193017959595, + "num_tokens": 28051155.0, + "step": 1084 + }, + { + "epoch": 0.11915220733582253, + "grad_norm": 2.5296921730041504, + "learning_rate": 5.949506037321624e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6745157241821289, + "num_tokens": 28075692.0, + "step": 1085 + }, + { + "epoch": 0.11926202503843619, + "grad_norm": 2.3932952880859375, + "learning_rate": 5.954994511525796e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6818429231643677, + "num_tokens": 28100802.0, + "step": 1086 + }, + { + "epoch": 0.11937184274104985, + "grad_norm": 2.2297749519348145, + "learning_rate": 5.960482985729967e-07, + "loss": 1.1207, + "mean_token_accuracy": 0.6671159267425537, + "num_tokens": 28128849.0, + "step": 1087 + }, + { + "epoch": 0.11948166044366351, + "grad_norm": 2.304320812225342, + "learning_rate": 5.965971459934138e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6739295125007629, + "num_tokens": 28155407.0, + "step": 1088 + }, + { + "epoch": 0.11959147814627719, + "grad_norm": 2.5996196269989014, + "learning_rate": 5.971459934138309e-07, + "loss": 0.9865, + "mean_token_accuracy": 0.7008695602416992, + "num_tokens": 28177201.0, + "step": 1089 + }, + { + "epoch": 0.11970129584889085, + "grad_norm": 2.203660011291504, + "learning_rate": 5.976948408342481e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6780641078948975, + "num_tokens": 28206355.0, + "step": 1090 + }, + { + "epoch": 0.1198111135515045, + "grad_norm": 2.507641315460205, + "learning_rate": 5.982436882546652e-07, + "loss": 0.9919, + "mean_token_accuracy": 0.699578583240509, + "num_tokens": 28229230.0, + "step": 1091 + }, + { + "epoch": 0.11992093125411817, + "grad_norm": 2.263340473175049, + "learning_rate": 5.987925356750822e-07, + "loss": 1.1255, + "mean_token_accuracy": 0.6682959794998169, + "num_tokens": 28258390.0, + "step": 1092 + }, + { + "epoch": 0.12003074895673183, + "grad_norm": 2.3226795196533203, + "learning_rate": 5.993413830954994e-07, + "loss": 1.0231, + "mean_token_accuracy": 0.6979094743728638, + "num_tokens": 28284753.0, + "step": 1093 + }, + { + "epoch": 0.12014056665934549, + "grad_norm": 2.2975385189056396, + "learning_rate": 5.998902305159165e-07, + "loss": 1.0671, + "mean_token_accuracy": 0.6918848752975464, + "num_tokens": 28308276.0, + "step": 1094 + }, + { + "epoch": 0.12025038436195915, + "grad_norm": 2.210052967071533, + "learning_rate": 6.004390779363336e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.6801344156265259, + "num_tokens": 28338223.0, + "step": 1095 + }, + { + "epoch": 0.1203602020645728, + "grad_norm": 2.5475916862487793, + "learning_rate": 6.009879253567508e-07, + "loss": 1.0367, + "mean_token_accuracy": 0.6993361711502075, + "num_tokens": 28362279.0, + "step": 1096 + }, + { + "epoch": 0.12047001976718646, + "grad_norm": 2.3976967334747314, + "learning_rate": 6.015367727771679e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6942006349563599, + "num_tokens": 28387149.0, + "step": 1097 + }, + { + "epoch": 0.12057983746980014, + "grad_norm": 2.3262736797332764, + "learning_rate": 6.02085620197585e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.6935843229293823, + "num_tokens": 28413914.0, + "step": 1098 + }, + { + "epoch": 0.1206896551724138, + "grad_norm": 2.2458059787750244, + "learning_rate": 6.026344676180022e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.6783163547515869, + "num_tokens": 28442323.0, + "step": 1099 + }, + { + "epoch": 0.12079947287502746, + "grad_norm": 2.279583692550659, + "learning_rate": 6.031833150384192e-07, + "loss": 1.0668, + "mean_token_accuracy": 0.6890846490859985, + "num_tokens": 28468559.0, + "step": 1100 + }, + { + "epoch": 0.12090929057764112, + "grad_norm": 2.428530216217041, + "learning_rate": 6.037321624588364e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6687667369842529, + "num_tokens": 28494323.0, + "step": 1101 + }, + { + "epoch": 0.12101910828025478, + "grad_norm": 2.316020965576172, + "learning_rate": 6.042810098792536e-07, + "loss": 1.0779, + "mean_token_accuracy": 0.6827452778816223, + "num_tokens": 28520961.0, + "step": 1102 + }, + { + "epoch": 0.12112892598286844, + "grad_norm": 2.7019007205963135, + "learning_rate": 6.048298572996706e-07, + "loss": 0.9975, + "mean_token_accuracy": 0.7015595436096191, + "num_tokens": 28541036.0, + "step": 1103 + }, + { + "epoch": 0.1212387436854821, + "grad_norm": 2.392453908920288, + "learning_rate": 6.053787047200878e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.693213164806366, + "num_tokens": 28564651.0, + "step": 1104 + }, + { + "epoch": 0.12134856138809576, + "grad_norm": 2.5350968837738037, + "learning_rate": 6.05927552140505e-07, + "loss": 1.004, + "mean_token_accuracy": 0.7056464552879333, + "num_tokens": 28586064.0, + "step": 1105 + }, + { + "epoch": 0.12145837909070942, + "grad_norm": 2.270064115524292, + "learning_rate": 6.06476399560922e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.688461184501648, + "num_tokens": 28612141.0, + "step": 1106 + }, + { + "epoch": 0.12156819679332308, + "grad_norm": 2.5423831939697266, + "learning_rate": 6.070252469813392e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6813306212425232, + "num_tokens": 28634968.0, + "step": 1107 + }, + { + "epoch": 0.12167801449593675, + "grad_norm": 2.3699400424957275, + "learning_rate": 6.075740944017564e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6834443807601929, + "num_tokens": 28662145.0, + "step": 1108 + }, + { + "epoch": 0.12178783219855041, + "grad_norm": 2.3137283325195312, + "learning_rate": 6.081229418221734e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6855127811431885, + "num_tokens": 28687621.0, + "step": 1109 + }, + { + "epoch": 0.12189764990116407, + "grad_norm": 2.46454119682312, + "learning_rate": 6.086717892425906e-07, + "loss": 1.073, + "mean_token_accuracy": 0.6844761967658997, + "num_tokens": 28713976.0, + "step": 1110 + }, + { + "epoch": 0.12200746760377773, + "grad_norm": 2.27044415473938, + "learning_rate": 6.092206366630076e-07, + "loss": 1.0021, + "mean_token_accuracy": 0.6980367302894592, + "num_tokens": 28739178.0, + "step": 1111 + }, + { + "epoch": 0.12211728530639139, + "grad_norm": 2.3820621967315674, + "learning_rate": 6.097694840834247e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.692830502986908, + "num_tokens": 28762384.0, + "step": 1112 + }, + { + "epoch": 0.12222710300900505, + "grad_norm": 2.1904308795928955, + "learning_rate": 6.103183315038419e-07, + "loss": 1.0361, + "mean_token_accuracy": 0.7036210298538208, + "num_tokens": 28788556.0, + "step": 1113 + }, + { + "epoch": 0.12233692071161871, + "grad_norm": 2.353466272354126, + "learning_rate": 6.10867178924259e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.6768139600753784, + "num_tokens": 28814784.0, + "step": 1114 + }, + { + "epoch": 0.12244673841423237, + "grad_norm": 2.6281161308288574, + "learning_rate": 6.114160263446761e-07, + "loss": 0.9537, + "mean_token_accuracy": 0.7145497798919678, + "num_tokens": 28837092.0, + "step": 1115 + }, + { + "epoch": 0.12255655611684603, + "grad_norm": 2.245232105255127, + "learning_rate": 6.119648737650933e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6958130598068237, + "num_tokens": 28863692.0, + "step": 1116 + }, + { + "epoch": 0.1226663738194597, + "grad_norm": 2.4352807998657227, + "learning_rate": 6.125137211855103e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.6926833391189575, + "num_tokens": 28886848.0, + "step": 1117 + }, + { + "epoch": 0.12277619152207336, + "grad_norm": 2.686521291732788, + "learning_rate": 6.130625686059275e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.6684684753417969, + "num_tokens": 28908381.0, + "step": 1118 + }, + { + "epoch": 0.12288600922468702, + "grad_norm": 2.2772839069366455, + "learning_rate": 6.136114160263447e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.7027159333229065, + "num_tokens": 28932980.0, + "step": 1119 + }, + { + "epoch": 0.12299582692730068, + "grad_norm": 2.243034601211548, + "learning_rate": 6.141602634467617e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6946828365325928, + "num_tokens": 28959396.0, + "step": 1120 + }, + { + "epoch": 0.12310564462991434, + "grad_norm": 2.1572842597961426, + "learning_rate": 6.147091108671789e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6797560453414917, + "num_tokens": 28987219.0, + "step": 1121 + }, + { + "epoch": 0.123215462332528, + "grad_norm": 2.486171245574951, + "learning_rate": 6.15257958287596e-07, + "loss": 1.0332, + "mean_token_accuracy": 0.6896426677703857, + "num_tokens": 29010802.0, + "step": 1122 + }, + { + "epoch": 0.12332528003514166, + "grad_norm": 2.4014477729797363, + "learning_rate": 6.158068057080131e-07, + "loss": 1.0613, + "mean_token_accuracy": 0.6759573221206665, + "num_tokens": 29037021.0, + "step": 1123 + }, + { + "epoch": 0.12343509773775532, + "grad_norm": 2.3762619495391846, + "learning_rate": 6.163556531284303e-07, + "loss": 1.0367, + "mean_token_accuracy": 0.6839805841445923, + "num_tokens": 29061759.0, + "step": 1124 + }, + { + "epoch": 0.12354491544036898, + "grad_norm": 2.2553422451019287, + "learning_rate": 6.169045005488474e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6858190298080444, + "num_tokens": 29089050.0, + "step": 1125 + }, + { + "epoch": 0.12365473314298264, + "grad_norm": 2.376824140548706, + "learning_rate": 6.174533479692645e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.7035720348358154, + "num_tokens": 29112102.0, + "step": 1126 + }, + { + "epoch": 0.12376455084559632, + "grad_norm": 2.2959842681884766, + "learning_rate": 6.180021953896817e-07, + "loss": 1.0605, + "mean_token_accuracy": 0.683851957321167, + "num_tokens": 29138528.0, + "step": 1127 + }, + { + "epoch": 0.12387436854820998, + "grad_norm": 2.1297895908355713, + "learning_rate": 6.185510428100988e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.6599622964859009, + "num_tokens": 29168365.0, + "step": 1128 + }, + { + "epoch": 0.12398418625082364, + "grad_norm": 2.1546928882598877, + "learning_rate": 6.190998902305159e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.6913334727287292, + "num_tokens": 29196337.0, + "step": 1129 + }, + { + "epoch": 0.1240940039534373, + "grad_norm": 2.329509973526001, + "learning_rate": 6.196487376509331e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6944952011108398, + "num_tokens": 29219350.0, + "step": 1130 + }, + { + "epoch": 0.12420382165605096, + "grad_norm": 2.3308935165405273, + "learning_rate": 6.201975850713501e-07, + "loss": 1.1173, + "mean_token_accuracy": 0.6664047241210938, + "num_tokens": 29244696.0, + "step": 1131 + }, + { + "epoch": 0.12431363935866462, + "grad_norm": 2.5588016510009766, + "learning_rate": 6.207464324917672e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6842617988586426, + "num_tokens": 29269980.0, + "step": 1132 + }, + { + "epoch": 0.12442345706127828, + "grad_norm": 2.366684913635254, + "learning_rate": 6.212952799121843e-07, + "loss": 1.0528, + "mean_token_accuracy": 0.6850981712341309, + "num_tokens": 29293929.0, + "step": 1133 + }, + { + "epoch": 0.12453327476389194, + "grad_norm": 2.4365172386169434, + "learning_rate": 6.218441273326015e-07, + "loss": 1.0641, + "mean_token_accuracy": 0.6888660788536072, + "num_tokens": 29318286.0, + "step": 1134 + }, + { + "epoch": 0.1246430924665056, + "grad_norm": 2.3738760948181152, + "learning_rate": 6.223929747530186e-07, + "loss": 1.072, + "mean_token_accuracy": 0.6803975105285645, + "num_tokens": 29343741.0, + "step": 1135 + }, + { + "epoch": 0.12475291016911927, + "grad_norm": 2.4645676612854004, + "learning_rate": 6.229418221734357e-07, + "loss": 0.9878, + "mean_token_accuracy": 0.7004945278167725, + "num_tokens": 29365532.0, + "step": 1136 + }, + { + "epoch": 0.12486272787173293, + "grad_norm": 2.4728188514709473, + "learning_rate": 6.234906695938529e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6705833673477173, + "num_tokens": 29389892.0, + "step": 1137 + }, + { + "epoch": 0.12497254557434659, + "grad_norm": 3.392340660095215, + "learning_rate": 6.2403951701427e-07, + "loss": 1.0438, + "mean_token_accuracy": 0.6930882930755615, + "num_tokens": 29413035.0, + "step": 1138 + }, + { + "epoch": 0.12508236327696023, + "grad_norm": 2.3118550777435303, + "learning_rate": 6.245883644346871e-07, + "loss": 1.0021, + "mean_token_accuracy": 0.6980696320533752, + "num_tokens": 29439140.0, + "step": 1139 + }, + { + "epoch": 0.1251921809795739, + "grad_norm": 2.178699016571045, + "learning_rate": 6.251372118551043e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.695608377456665, + "num_tokens": 29465914.0, + "step": 1140 + }, + { + "epoch": 0.12530199868218758, + "grad_norm": 2.1694276332855225, + "learning_rate": 6.256860592755214e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.6681861877441406, + "num_tokens": 29494069.0, + "step": 1141 + }, + { + "epoch": 0.12541181638480123, + "grad_norm": 2.6204075813293457, + "learning_rate": 6.262349066959385e-07, + "loss": 1.1072, + "mean_token_accuracy": 0.6798648834228516, + "num_tokens": 29516960.0, + "step": 1142 + }, + { + "epoch": 0.1255216340874149, + "grad_norm": 2.329075813293457, + "learning_rate": 6.267837541163557e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.6816027760505676, + "num_tokens": 29542281.0, + "step": 1143 + }, + { + "epoch": 0.12563145179002855, + "grad_norm": 2.481334686279297, + "learning_rate": 6.273326015367727e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6923518180847168, + "num_tokens": 29564721.0, + "step": 1144 + }, + { + "epoch": 0.12574126949264222, + "grad_norm": 2.504399538040161, + "learning_rate": 6.278814489571899e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6907610297203064, + "num_tokens": 29586803.0, + "step": 1145 + }, + { + "epoch": 0.12585108719525587, + "grad_norm": 2.31760573387146, + "learning_rate": 6.284302963776071e-07, + "loss": 1.121, + "mean_token_accuracy": 0.6739479303359985, + "num_tokens": 29612222.0, + "step": 1146 + }, + { + "epoch": 0.12596090489786954, + "grad_norm": 2.4170761108398438, + "learning_rate": 6.289791437980241e-07, + "loss": 0.9152, + "mean_token_accuracy": 0.7257747054100037, + "num_tokens": 29639108.0, + "step": 1147 + }, + { + "epoch": 0.1260707226004832, + "grad_norm": 2.3322641849517822, + "learning_rate": 6.295279912184413e-07, + "loss": 1.0734, + "mean_token_accuracy": 0.6783791780471802, + "num_tokens": 29663095.0, + "step": 1148 + }, + { + "epoch": 0.12618054030309686, + "grad_norm": 2.71083664894104, + "learning_rate": 6.300768386388585e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.690373957157135, + "num_tokens": 29683089.0, + "step": 1149 + }, + { + "epoch": 0.1262903580057105, + "grad_norm": 1.890494704246521, + "learning_rate": 6.306256860592754e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.6980961561203003, + "num_tokens": 29718749.0, + "step": 1150 + }, + { + "epoch": 0.12640017570832418, + "grad_norm": 2.194661855697632, + "learning_rate": 6.311745334796926e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6710026264190674, + "num_tokens": 29748431.0, + "step": 1151 + }, + { + "epoch": 0.12650999341093785, + "grad_norm": 2.3939380645751953, + "learning_rate": 6.317233809001098e-07, + "loss": 1.0439, + "mean_token_accuracy": 0.6999430656433105, + "num_tokens": 29773742.0, + "step": 1152 + }, + { + "epoch": 0.1266198111135515, + "grad_norm": 2.268895387649536, + "learning_rate": 6.322722283205268e-07, + "loss": 1.1161, + "mean_token_accuracy": 0.6700100898742676, + "num_tokens": 29801531.0, + "step": 1153 + }, + { + "epoch": 0.12672962881616517, + "grad_norm": 2.488576650619507, + "learning_rate": 6.32821075740944e-07, + "loss": 0.9984, + "mean_token_accuracy": 0.6978114247322083, + "num_tokens": 29826165.0, + "step": 1154 + }, + { + "epoch": 0.12683944651877882, + "grad_norm": 2.1815578937530518, + "learning_rate": 6.33369923161361e-07, + "loss": 1.035, + "mean_token_accuracy": 0.6882035732269287, + "num_tokens": 29854979.0, + "step": 1155 + }, + { + "epoch": 0.1269492642213925, + "grad_norm": 2.122642993927002, + "learning_rate": 6.339187705817782e-07, + "loss": 1.0485, + "mean_token_accuracy": 0.6912128925323486, + "num_tokens": 29883755.0, + "step": 1156 + }, + { + "epoch": 0.12705908192400614, + "grad_norm": 2.243006706237793, + "learning_rate": 6.344676180021954e-07, + "loss": 0.9417, + "mean_token_accuracy": 0.7156314849853516, + "num_tokens": 29908686.0, + "step": 1157 + }, + { + "epoch": 0.1271688996266198, + "grad_norm": 2.405452251434326, + "learning_rate": 6.350164654226124e-07, + "loss": 1.1079, + "mean_token_accuracy": 0.6721410751342773, + "num_tokens": 29932878.0, + "step": 1158 + }, + { + "epoch": 0.12727871732923346, + "grad_norm": 2.355337142944336, + "learning_rate": 6.355653128430296e-07, + "loss": 1.033, + "mean_token_accuracy": 0.6909398436546326, + "num_tokens": 29956748.0, + "step": 1159 + }, + { + "epoch": 0.12738853503184713, + "grad_norm": 2.233315944671631, + "learning_rate": 6.361141602634468e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.6920101046562195, + "num_tokens": 29982810.0, + "step": 1160 + }, + { + "epoch": 0.1274983527344608, + "grad_norm": 2.1320786476135254, + "learning_rate": 6.366630076838638e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7019463181495667, + "num_tokens": 30010837.0, + "step": 1161 + }, + { + "epoch": 0.12760817043707445, + "grad_norm": 2.2579293251037598, + "learning_rate": 6.37211855104281e-07, + "loss": 1.0919, + "mean_token_accuracy": 0.6705261468887329, + "num_tokens": 30038954.0, + "step": 1162 + }, + { + "epoch": 0.12771798813968813, + "grad_norm": 2.2972748279571533, + "learning_rate": 6.377607025246982e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6763704419136047, + "num_tokens": 30065000.0, + "step": 1163 + }, + { + "epoch": 0.12782780584230177, + "grad_norm": 2.2035915851593018, + "learning_rate": 6.383095499451152e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6896023154258728, + "num_tokens": 30094684.0, + "step": 1164 + }, + { + "epoch": 0.12793762354491545, + "grad_norm": 2.255791425704956, + "learning_rate": 6.388583973655324e-07, + "loss": 1.057, + "mean_token_accuracy": 0.6886879205703735, + "num_tokens": 30124109.0, + "step": 1165 + }, + { + "epoch": 0.1280474412475291, + "grad_norm": 2.3061649799346924, + "learning_rate": 6.394072447859495e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6864258050918579, + "num_tokens": 30149104.0, + "step": 1166 + }, + { + "epoch": 0.12815725895014277, + "grad_norm": 2.201158046722412, + "learning_rate": 6.399560922063666e-07, + "loss": 0.9618, + "mean_token_accuracy": 0.7110612392425537, + "num_tokens": 30176346.0, + "step": 1167 + }, + { + "epoch": 0.1282670766527564, + "grad_norm": 2.337303638458252, + "learning_rate": 6.405049396267838e-07, + "loss": 1.004, + "mean_token_accuracy": 0.7019451856613159, + "num_tokens": 30202434.0, + "step": 1168 + }, + { + "epoch": 0.12837689435537009, + "grad_norm": 2.4144222736358643, + "learning_rate": 6.410537870472008e-07, + "loss": 1.0321, + "mean_token_accuracy": 0.6916447877883911, + "num_tokens": 30225784.0, + "step": 1169 + }, + { + "epoch": 0.12848671205798376, + "grad_norm": 2.720571756362915, + "learning_rate": 6.416026344676179e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6825323104858398, + "num_tokens": 30246017.0, + "step": 1170 + }, + { + "epoch": 0.1285965297605974, + "grad_norm": 2.1418228149414062, + "learning_rate": 6.421514818880351e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.7033500075340271, + "num_tokens": 30274483.0, + "step": 1171 + }, + { + "epoch": 0.12870634746321108, + "grad_norm": 2.4329168796539307, + "learning_rate": 6.427003293084522e-07, + "loss": 1.0035, + "mean_token_accuracy": 0.6920531392097473, + "num_tokens": 30297190.0, + "step": 1172 + }, + { + "epoch": 0.12881616516582473, + "grad_norm": 2.2048652172088623, + "learning_rate": 6.432491767288693e-07, + "loss": 1.057, + "mean_token_accuracy": 0.6857109665870667, + "num_tokens": 30323778.0, + "step": 1173 + }, + { + "epoch": 0.1289259828684384, + "grad_norm": 2.0066752433776855, + "learning_rate": 6.437980241492865e-07, + "loss": 1.0896, + "mean_token_accuracy": 0.6777154207229614, + "num_tokens": 30355444.0, + "step": 1174 + }, + { + "epoch": 0.12903580057105205, + "grad_norm": 2.480213165283203, + "learning_rate": 6.443468715697036e-07, + "loss": 1.0501, + "mean_token_accuracy": 0.6969969272613525, + "num_tokens": 30378876.0, + "step": 1175 + }, + { + "epoch": 0.12914561827366572, + "grad_norm": 2.313385009765625, + "learning_rate": 6.448957189901207e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6988660097122192, + "num_tokens": 30403770.0, + "step": 1176 + }, + { + "epoch": 0.12925543597627936, + "grad_norm": 2.4556751251220703, + "learning_rate": 6.454445664105378e-07, + "loss": 1.0794, + "mean_token_accuracy": 0.6855214834213257, + "num_tokens": 30428979.0, + "step": 1177 + }, + { + "epoch": 0.12936525367889304, + "grad_norm": 2.5516679286956787, + "learning_rate": 6.45993413830955e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6992892026901245, + "num_tokens": 30451205.0, + "step": 1178 + }, + { + "epoch": 0.1294750713815067, + "grad_norm": 2.636340856552124, + "learning_rate": 6.465422612513721e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.6880530714988708, + "num_tokens": 30471832.0, + "step": 1179 + }, + { + "epoch": 0.12958488908412036, + "grad_norm": 2.5161354541778564, + "learning_rate": 6.470911086717892e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6820878982543945, + "num_tokens": 30494698.0, + "step": 1180 + }, + { + "epoch": 0.12969470678673403, + "grad_norm": 2.1167755126953125, + "learning_rate": 6.476399560922064e-07, + "loss": 1.1529, + "mean_token_accuracy": 0.6626050472259521, + "num_tokens": 30526772.0, + "step": 1181 + }, + { + "epoch": 0.12980452448934768, + "grad_norm": 2.6938207149505615, + "learning_rate": 6.481888035126235e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.7028120756149292, + "num_tokens": 30547622.0, + "step": 1182 + }, + { + "epoch": 0.12991434219196135, + "grad_norm": 2.4757027626037598, + "learning_rate": 6.487376509330406e-07, + "loss": 0.9909, + "mean_token_accuracy": 0.70744788646698, + "num_tokens": 30570551.0, + "step": 1183 + }, + { + "epoch": 0.130024159894575, + "grad_norm": 2.1698875427246094, + "learning_rate": 6.492864983534578e-07, + "loss": 1.1626, + "mean_token_accuracy": 0.6678707003593445, + "num_tokens": 30600965.0, + "step": 1184 + }, + { + "epoch": 0.13013397759718867, + "grad_norm": 2.240346908569336, + "learning_rate": 6.498353457738749e-07, + "loss": 1.0053, + "mean_token_accuracy": 0.696221113204956, + "num_tokens": 30626176.0, + "step": 1185 + }, + { + "epoch": 0.13024379529980232, + "grad_norm": 2.1165997982025146, + "learning_rate": 6.50384193194292e-07, + "loss": 1.0476, + "mean_token_accuracy": 0.6945343017578125, + "num_tokens": 30655115.0, + "step": 1186 + }, + { + "epoch": 0.130353613002416, + "grad_norm": 2.5147366523742676, + "learning_rate": 6.509330406147092e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.6831457614898682, + "num_tokens": 30679241.0, + "step": 1187 + }, + { + "epoch": 0.13046343070502964, + "grad_norm": 3.1583592891693115, + "learning_rate": 6.514818880351261e-07, + "loss": 0.9339, + "mean_token_accuracy": 0.7172684669494629, + "num_tokens": 30694776.0, + "step": 1188 + }, + { + "epoch": 0.1305732484076433, + "grad_norm": 2.6599459648132324, + "learning_rate": 6.520307354555433e-07, + "loss": 1.1186, + "mean_token_accuracy": 0.6779919266700745, + "num_tokens": 30716241.0, + "step": 1189 + }, + { + "epoch": 0.13068306611025698, + "grad_norm": 2.549696445465088, + "learning_rate": 6.525795828759604e-07, + "loss": 1.003, + "mean_token_accuracy": 0.6959036588668823, + "num_tokens": 30738778.0, + "step": 1190 + }, + { + "epoch": 0.13079288381287063, + "grad_norm": 2.5289924144744873, + "learning_rate": 6.531284302963775e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.689943790435791, + "num_tokens": 30762993.0, + "step": 1191 + }, + { + "epoch": 0.1309027015154843, + "grad_norm": 2.4738152027130127, + "learning_rate": 6.536772777167947e-07, + "loss": 0.9893, + "mean_token_accuracy": 0.7039062976837158, + "num_tokens": 30785810.0, + "step": 1192 + }, + { + "epoch": 0.13101251921809795, + "grad_norm": 2.3419361114501953, + "learning_rate": 6.542261251372118e-07, + "loss": 1.0581, + "mean_token_accuracy": 0.6930562257766724, + "num_tokens": 30810762.0, + "step": 1193 + }, + { + "epoch": 0.13112233692071162, + "grad_norm": 2.3716182708740234, + "learning_rate": 6.547749725576289e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6979224681854248, + "num_tokens": 30835053.0, + "step": 1194 + }, + { + "epoch": 0.13123215462332527, + "grad_norm": 2.272582769393921, + "learning_rate": 6.553238199780461e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6786186695098877, + "num_tokens": 30862306.0, + "step": 1195 + }, + { + "epoch": 0.13134197232593894, + "grad_norm": 2.4447576999664307, + "learning_rate": 6.558726673984632e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.6970425248146057, + "num_tokens": 30886661.0, + "step": 1196 + }, + { + "epoch": 0.1314517900285526, + "grad_norm": 2.368319272994995, + "learning_rate": 6.564215148188803e-07, + "loss": 1.0491, + "mean_token_accuracy": 0.6901866793632507, + "num_tokens": 30910881.0, + "step": 1197 + }, + { + "epoch": 0.13156160773116626, + "grad_norm": 2.4981064796447754, + "learning_rate": 6.569703622392975e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6915544867515564, + "num_tokens": 30934291.0, + "step": 1198 + }, + { + "epoch": 0.13167142543377994, + "grad_norm": 2.2925496101379395, + "learning_rate": 6.575192096597145e-07, + "loss": 0.9596, + "mean_token_accuracy": 0.7124931812286377, + "num_tokens": 30958012.0, + "step": 1199 + }, + { + "epoch": 0.13178124313639358, + "grad_norm": 2.429196357727051, + "learning_rate": 6.580680570801317e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6948003172874451, + "num_tokens": 30981065.0, + "step": 1200 + }, + { + "epoch": 0.13189106083900726, + "grad_norm": 2.1133604049682617, + "learning_rate": 6.586169045005489e-07, + "loss": 0.9818, + "mean_token_accuracy": 0.702234148979187, + "num_tokens": 31007993.0, + "step": 1201 + }, + { + "epoch": 0.1320008785416209, + "grad_norm": 2.277515411376953, + "learning_rate": 6.591657519209659e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.7053029537200928, + "num_tokens": 31033108.0, + "step": 1202 + }, + { + "epoch": 0.13211069624423458, + "grad_norm": 2.084831714630127, + "learning_rate": 6.597145993413831e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.6714164614677429, + "num_tokens": 31064241.0, + "step": 1203 + }, + { + "epoch": 0.13222051394684822, + "grad_norm": 2.246065616607666, + "learning_rate": 6.602634467618003e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6931439638137817, + "num_tokens": 31092513.0, + "step": 1204 + }, + { + "epoch": 0.1323303316494619, + "grad_norm": 1.9303646087646484, + "learning_rate": 6.608122941822173e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.677688479423523, + "num_tokens": 31130144.0, + "step": 1205 + }, + { + "epoch": 0.13244014935207554, + "grad_norm": 2.31479811668396, + "learning_rate": 6.613611416026345e-07, + "loss": 1.0328, + "mean_token_accuracy": 0.6848405599594116, + "num_tokens": 31156122.0, + "step": 1206 + }, + { + "epoch": 0.13254996705468922, + "grad_norm": 2.241431951522827, + "learning_rate": 6.619099890230515e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6833957433700562, + "num_tokens": 31184406.0, + "step": 1207 + }, + { + "epoch": 0.1326597847573029, + "grad_norm": 2.2067558765411377, + "learning_rate": 6.624588364434686e-07, + "loss": 1.1078, + "mean_token_accuracy": 0.6779634952545166, + "num_tokens": 31212889.0, + "step": 1208 + }, + { + "epoch": 0.13276960245991654, + "grad_norm": 2.461707830429077, + "learning_rate": 6.630076838638858e-07, + "loss": 1.0619, + "mean_token_accuracy": 0.683975875377655, + "num_tokens": 31236810.0, + "step": 1209 + }, + { + "epoch": 0.1328794201625302, + "grad_norm": 2.178352117538452, + "learning_rate": 6.635565312843029e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6981410980224609, + "num_tokens": 31266648.0, + "step": 1210 + }, + { + "epoch": 0.13298923786514386, + "grad_norm": 2.049149990081787, + "learning_rate": 6.6410537870472e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.682206392288208, + "num_tokens": 31299049.0, + "step": 1211 + }, + { + "epoch": 0.13309905556775753, + "grad_norm": 2.065120220184326, + "learning_rate": 6.646542261251372e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.6785928010940552, + "num_tokens": 31331058.0, + "step": 1212 + }, + { + "epoch": 0.13320887327037118, + "grad_norm": 2.3781425952911377, + "learning_rate": 6.652030735455543e-07, + "loss": 1.0046, + "mean_token_accuracy": 0.6981070041656494, + "num_tokens": 31357243.0, + "step": 1213 + }, + { + "epoch": 0.13331869097298485, + "grad_norm": 2.254188299179077, + "learning_rate": 6.657519209659714e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6863641142845154, + "num_tokens": 31383487.0, + "step": 1214 + }, + { + "epoch": 0.1334285086755985, + "grad_norm": 2.5728507041931152, + "learning_rate": 6.663007683863886e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6974657773971558, + "num_tokens": 31405183.0, + "step": 1215 + }, + { + "epoch": 0.13353832637821217, + "grad_norm": 2.5702338218688965, + "learning_rate": 6.668496158068057e-07, + "loss": 1.07, + "mean_token_accuracy": 0.6821655035018921, + "num_tokens": 31428316.0, + "step": 1216 + }, + { + "epoch": 0.13364814408082584, + "grad_norm": 2.323514938354492, + "learning_rate": 6.673984632272228e-07, + "loss": 0.9754, + "mean_token_accuracy": 0.6992970705032349, + "num_tokens": 31452190.0, + "step": 1217 + }, + { + "epoch": 0.1337579617834395, + "grad_norm": 2.4322891235351562, + "learning_rate": 6.679473106476399e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7011460065841675, + "num_tokens": 31475238.0, + "step": 1218 + }, + { + "epoch": 0.13386777948605316, + "grad_norm": 2.1696975231170654, + "learning_rate": 6.684961580680571e-07, + "loss": 1.0656, + "mean_token_accuracy": 0.6764731407165527, + "num_tokens": 31503768.0, + "step": 1219 + }, + { + "epoch": 0.1339775971886668, + "grad_norm": 2.16890287399292, + "learning_rate": 6.690450054884742e-07, + "loss": 0.9621, + "mean_token_accuracy": 0.7106043696403503, + "num_tokens": 31532071.0, + "step": 1220 + }, + { + "epoch": 0.13408741489128048, + "grad_norm": 2.251225233078003, + "learning_rate": 6.695938529088913e-07, + "loss": 1.0305, + "mean_token_accuracy": 0.6897634267807007, + "num_tokens": 31558327.0, + "step": 1221 + }, + { + "epoch": 0.13419723259389413, + "grad_norm": 2.479668140411377, + "learning_rate": 6.701427003293085e-07, + "loss": 1.0358, + "mean_token_accuracy": 0.693688154220581, + "num_tokens": 31582070.0, + "step": 1222 + }, + { + "epoch": 0.1343070502965078, + "grad_norm": 2.133741855621338, + "learning_rate": 6.706915477497256e-07, + "loss": 0.9847, + "mean_token_accuracy": 0.7014238834381104, + "num_tokens": 31612163.0, + "step": 1223 + }, + { + "epoch": 0.13441686799912145, + "grad_norm": 2.4957053661346436, + "learning_rate": 6.712403951701427e-07, + "loss": 1.1325, + "mean_token_accuracy": 0.6687411665916443, + "num_tokens": 31636251.0, + "step": 1224 + }, + { + "epoch": 0.13452668570173512, + "grad_norm": 2.2485461235046387, + "learning_rate": 6.717892425905599e-07, + "loss": 1.1358, + "mean_token_accuracy": 0.6622223258018494, + "num_tokens": 31665393.0, + "step": 1225 + }, + { + "epoch": 0.13463650340434877, + "grad_norm": 2.7009379863739014, + "learning_rate": 6.723380900109769e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7251805067062378, + "num_tokens": 31684237.0, + "step": 1226 + }, + { + "epoch": 0.13474632110696244, + "grad_norm": 2.6877007484436035, + "learning_rate": 6.72886937431394e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6902914047241211, + "num_tokens": 31707025.0, + "step": 1227 + }, + { + "epoch": 0.13485613880957611, + "grad_norm": 2.069223165512085, + "learning_rate": 6.734357848518111e-07, + "loss": 1.0546, + "mean_token_accuracy": 0.6875495910644531, + "num_tokens": 31737026.0, + "step": 1228 + }, + { + "epoch": 0.13496595651218976, + "grad_norm": 2.231289863586426, + "learning_rate": 6.739846322722282e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.6848782896995544, + "num_tokens": 31766876.0, + "step": 1229 + }, + { + "epoch": 0.13507577421480343, + "grad_norm": 2.4432365894317627, + "learning_rate": 6.745334796926454e-07, + "loss": 1.0614, + "mean_token_accuracy": 0.6845542788505554, + "num_tokens": 31793131.0, + "step": 1230 + }, + { + "epoch": 0.13518559191741708, + "grad_norm": 2.0358774662017822, + "learning_rate": 6.750823271130625e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.6656438112258911, + "num_tokens": 31824945.0, + "step": 1231 + }, + { + "epoch": 0.13529540962003075, + "grad_norm": 2.0867862701416016, + "learning_rate": 6.756311745334796e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.6873692870140076, + "num_tokens": 31855779.0, + "step": 1232 + }, + { + "epoch": 0.1354052273226444, + "grad_norm": 2.4138195514678955, + "learning_rate": 6.761800219538968e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6785014271736145, + "num_tokens": 31880056.0, + "step": 1233 + }, + { + "epoch": 0.13551504502525807, + "grad_norm": 2.2279343605041504, + "learning_rate": 6.767288693743139e-07, + "loss": 1.0724, + "mean_token_accuracy": 0.6879022121429443, + "num_tokens": 31907657.0, + "step": 1234 + }, + { + "epoch": 0.13562486272787172, + "grad_norm": 2.3120951652526855, + "learning_rate": 6.77277716794731e-07, + "loss": 0.9676, + "mean_token_accuracy": 0.7126255631446838, + "num_tokens": 31933067.0, + "step": 1235 + }, + { + "epoch": 0.1357346804304854, + "grad_norm": 2.2625908851623535, + "learning_rate": 6.778265642151482e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6796931624412537, + "num_tokens": 31959861.0, + "step": 1236 + }, + { + "epoch": 0.13584449813309907, + "grad_norm": 2.190857410430908, + "learning_rate": 6.783754116355653e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.6834394335746765, + "num_tokens": 31988879.0, + "step": 1237 + }, + { + "epoch": 0.1359543158357127, + "grad_norm": 2.266472101211548, + "learning_rate": 6.789242590559824e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6698105931282043, + "num_tokens": 32016250.0, + "step": 1238 + }, + { + "epoch": 0.1360641335383264, + "grad_norm": 2.4210729598999023, + "learning_rate": 6.794731064763996e-07, + "loss": 1.0325, + "mean_token_accuracy": 0.6882911920547485, + "num_tokens": 32039820.0, + "step": 1239 + }, + { + "epoch": 0.13617395124094003, + "grad_norm": 2.389493465423584, + "learning_rate": 6.800219538968166e-07, + "loss": 0.997, + "mean_token_accuracy": 0.6949017643928528, + "num_tokens": 32063542.0, + "step": 1240 + }, + { + "epoch": 0.1362837689435537, + "grad_norm": 2.1811227798461914, + "learning_rate": 6.805708013172338e-07, + "loss": 0.9786, + "mean_token_accuracy": 0.7026405334472656, + "num_tokens": 32091142.0, + "step": 1241 + }, + { + "epoch": 0.13639358664616735, + "grad_norm": 2.6434926986694336, + "learning_rate": 6.81119648737651e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6931051015853882, + "num_tokens": 32112800.0, + "step": 1242 + }, + { + "epoch": 0.13650340434878103, + "grad_norm": 2.3826863765716553, + "learning_rate": 6.81668496158068e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6829262971878052, + "num_tokens": 32140067.0, + "step": 1243 + }, + { + "epoch": 0.13661322205139467, + "grad_norm": 2.2421693801879883, + "learning_rate": 6.822173435784852e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6896272897720337, + "num_tokens": 32166596.0, + "step": 1244 + }, + { + "epoch": 0.13672303975400835, + "grad_norm": 2.3535749912261963, + "learning_rate": 6.827661909989023e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6941779851913452, + "num_tokens": 32190176.0, + "step": 1245 + }, + { + "epoch": 0.13683285745662202, + "grad_norm": 2.3291304111480713, + "learning_rate": 6.833150384193193e-07, + "loss": 1.129, + "mean_token_accuracy": 0.6810532808303833, + "num_tokens": 32219286.0, + "step": 1246 + }, + { + "epoch": 0.13694267515923567, + "grad_norm": 2.2426366806030273, + "learning_rate": 6.838638858397365e-07, + "loss": 1.1025, + "mean_token_accuracy": 0.6758460998535156, + "num_tokens": 32248432.0, + "step": 1247 + }, + { + "epoch": 0.13705249286184934, + "grad_norm": 2.796079158782959, + "learning_rate": 6.844127332601537e-07, + "loss": 0.9482, + "mean_token_accuracy": 0.7117533683776855, + "num_tokens": 32267262.0, + "step": 1248 + }, + { + "epoch": 0.13716231056446299, + "grad_norm": 2.136951446533203, + "learning_rate": 6.849615806805707e-07, + "loss": 1.1506, + "mean_token_accuracy": 0.6702641248703003, + "num_tokens": 32300399.0, + "step": 1249 + }, + { + "epoch": 0.13727212826707666, + "grad_norm": 2.320467948913574, + "learning_rate": 6.855104281009879e-07, + "loss": 1.0001, + "mean_token_accuracy": 0.6978693008422852, + "num_tokens": 32328118.0, + "step": 1250 + }, + { + "epoch": 0.1373819459696903, + "grad_norm": 2.4534826278686523, + "learning_rate": 6.86059275521405e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6706653237342834, + "num_tokens": 32352316.0, + "step": 1251 + }, + { + "epoch": 0.13749176367230398, + "grad_norm": 2.3852155208587646, + "learning_rate": 6.866081229418221e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6829538345336914, + "num_tokens": 32377286.0, + "step": 1252 + }, + { + "epoch": 0.13760158137491763, + "grad_norm": 2.3225793838500977, + "learning_rate": 6.871569703622393e-07, + "loss": 0.9419, + "mean_token_accuracy": 0.7111914157867432, + "num_tokens": 32400884.0, + "step": 1253 + }, + { + "epoch": 0.1377113990775313, + "grad_norm": 2.0799779891967773, + "learning_rate": 6.877058177826564e-07, + "loss": 1.0879, + "mean_token_accuracy": 0.6836206912994385, + "num_tokens": 32432448.0, + "step": 1254 + }, + { + "epoch": 0.13782121678014497, + "grad_norm": 2.483401298522949, + "learning_rate": 6.882546652030735e-07, + "loss": 0.9466, + "mean_token_accuracy": 0.7177457213401794, + "num_tokens": 32458618.0, + "step": 1255 + }, + { + "epoch": 0.13793103448275862, + "grad_norm": 2.1822218894958496, + "learning_rate": 6.888035126234907e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6841863989830017, + "num_tokens": 32485466.0, + "step": 1256 + }, + { + "epoch": 0.1380408521853723, + "grad_norm": 1.9465945959091187, + "learning_rate": 6.893523600439078e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.6881371736526489, + "num_tokens": 32518002.0, + "step": 1257 + }, + { + "epoch": 0.13815066988798594, + "grad_norm": 2.4158332347869873, + "learning_rate": 6.899012074643249e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6854565143585205, + "num_tokens": 32544932.0, + "step": 1258 + }, + { + "epoch": 0.1382604875905996, + "grad_norm": 2.389350414276123, + "learning_rate": 6.904500548847421e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6865845322608948, + "num_tokens": 32569876.0, + "step": 1259 + }, + { + "epoch": 0.13837030529321326, + "grad_norm": 2.4267752170562744, + "learning_rate": 6.909989023051592e-07, + "loss": 1.1151, + "mean_token_accuracy": 0.6665623188018799, + "num_tokens": 32594654.0, + "step": 1260 + }, + { + "epoch": 0.13848012299582693, + "grad_norm": 2.559979200363159, + "learning_rate": 6.915477497255763e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6797416806221008, + "num_tokens": 32617185.0, + "step": 1261 + }, + { + "epoch": 0.13858994069844058, + "grad_norm": 2.587855339050293, + "learning_rate": 6.920965971459934e-07, + "loss": 1.0934, + "mean_token_accuracy": 0.6729565858840942, + "num_tokens": 32641317.0, + "step": 1262 + }, + { + "epoch": 0.13869975840105425, + "grad_norm": 2.2467033863067627, + "learning_rate": 6.926454445664105e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6887362003326416, + "num_tokens": 32669203.0, + "step": 1263 + }, + { + "epoch": 0.1388095761036679, + "grad_norm": 2.4014811515808105, + "learning_rate": 6.931942919868276e-07, + "loss": 0.9881, + "mean_token_accuracy": 0.6988037824630737, + "num_tokens": 32692581.0, + "step": 1264 + }, + { + "epoch": 0.13891939380628157, + "grad_norm": 2.307084321975708, + "learning_rate": 6.937431394072447e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6886017322540283, + "num_tokens": 32720675.0, + "step": 1265 + }, + { + "epoch": 0.13902921150889525, + "grad_norm": 2.343972682952881, + "learning_rate": 6.942919868276618e-07, + "loss": 1.0089, + "mean_token_accuracy": 0.69838947057724, + "num_tokens": 32746150.0, + "step": 1266 + }, + { + "epoch": 0.1391390292115089, + "grad_norm": 2.5071661472320557, + "learning_rate": 6.94840834248079e-07, + "loss": 1.0775, + "mean_token_accuracy": 0.6798350214958191, + "num_tokens": 32769286.0, + "step": 1267 + }, + { + "epoch": 0.13924884691412256, + "grad_norm": 2.1225130558013916, + "learning_rate": 6.953896816684961e-07, + "loss": 1.064, + "mean_token_accuracy": 0.6871559023857117, + "num_tokens": 32799613.0, + "step": 1268 + }, + { + "epoch": 0.1393586646167362, + "grad_norm": 2.3564412593841553, + "learning_rate": 6.959385290889132e-07, + "loss": 0.9274, + "mean_token_accuracy": 0.7187889814376831, + "num_tokens": 32823796.0, + "step": 1269 + }, + { + "epoch": 0.13946848231934988, + "grad_norm": 2.303948163986206, + "learning_rate": 6.964873765093304e-07, + "loss": 1.1015, + "mean_token_accuracy": 0.675574541091919, + "num_tokens": 32853135.0, + "step": 1270 + }, + { + "epoch": 0.13957830002196353, + "grad_norm": 2.30513596534729, + "learning_rate": 6.970362239297475e-07, + "loss": 0.9645, + "mean_token_accuracy": 0.7144187688827515, + "num_tokens": 32876847.0, + "step": 1271 + }, + { + "epoch": 0.1396881177245772, + "grad_norm": 2.59710955619812, + "learning_rate": 6.975850713501646e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.6822783350944519, + "num_tokens": 32904303.0, + "step": 1272 + }, + { + "epoch": 0.13979793542719085, + "grad_norm": 2.2003393173217773, + "learning_rate": 6.981339187705817e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6966398358345032, + "num_tokens": 32931913.0, + "step": 1273 + }, + { + "epoch": 0.13990775312980452, + "grad_norm": 1.9573081731796265, + "learning_rate": 6.986827661909989e-07, + "loss": 1.1057, + "mean_token_accuracy": 0.6673219203948975, + "num_tokens": 32968586.0, + "step": 1274 + }, + { + "epoch": 0.1400175708324182, + "grad_norm": 2.525826930999756, + "learning_rate": 6.99231613611416e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6969670057296753, + "num_tokens": 32990773.0, + "step": 1275 + }, + { + "epoch": 0.14012738853503184, + "grad_norm": 2.462918519973755, + "learning_rate": 6.997804610318331e-07, + "loss": 1.0599, + "mean_token_accuracy": 0.698784589767456, + "num_tokens": 33013464.0, + "step": 1276 + }, + { + "epoch": 0.14023720623764552, + "grad_norm": 2.254070520401001, + "learning_rate": 7.003293084522503e-07, + "loss": 1.1382, + "mean_token_accuracy": 0.6664584875106812, + "num_tokens": 33041466.0, + "step": 1277 + }, + { + "epoch": 0.14034702394025916, + "grad_norm": 2.671647310256958, + "learning_rate": 7.008781558726674e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6837971210479736, + "num_tokens": 33060887.0, + "step": 1278 + }, + { + "epoch": 0.14045684164287284, + "grad_norm": 2.243154287338257, + "learning_rate": 7.014270032930845e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6806493997573853, + "num_tokens": 33088003.0, + "step": 1279 + }, + { + "epoch": 0.14056665934548648, + "grad_norm": 2.4426510334014893, + "learning_rate": 7.019758507135017e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6933578252792358, + "num_tokens": 33109999.0, + "step": 1280 + }, + { + "epoch": 0.14067647704810016, + "grad_norm": 2.2426857948303223, + "learning_rate": 7.025246981339188e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.709345817565918, + "num_tokens": 33135305.0, + "step": 1281 + }, + { + "epoch": 0.1407862947507138, + "grad_norm": 2.1867942810058594, + "learning_rate": 7.030735455543358e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7084676027297974, + "num_tokens": 33160476.0, + "step": 1282 + }, + { + "epoch": 0.14089611245332748, + "grad_norm": 2.393972873687744, + "learning_rate": 7.03622392974753e-07, + "loss": 1.001, + "mean_token_accuracy": 0.70318204164505, + "num_tokens": 33184079.0, + "step": 1283 + }, + { + "epoch": 0.14100593015594115, + "grad_norm": 2.392216205596924, + "learning_rate": 7.0417124039517e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6995140314102173, + "num_tokens": 33207405.0, + "step": 1284 + }, + { + "epoch": 0.1411157478585548, + "grad_norm": 2.5029706954956055, + "learning_rate": 7.047200878155872e-07, + "loss": 1.1018, + "mean_token_accuracy": 0.6848186254501343, + "num_tokens": 33229843.0, + "step": 1285 + }, + { + "epoch": 0.14122556556116847, + "grad_norm": 2.551323413848877, + "learning_rate": 7.052689352360044e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.6997604966163635, + "num_tokens": 33251989.0, + "step": 1286 + }, + { + "epoch": 0.14133538326378212, + "grad_norm": 2.270056962966919, + "learning_rate": 7.058177826564214e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6902586221694946, + "num_tokens": 33278356.0, + "step": 1287 + }, + { + "epoch": 0.1414452009663958, + "grad_norm": 2.395988941192627, + "learning_rate": 7.063666300768386e-07, + "loss": 1.1434, + "mean_token_accuracy": 0.6619001626968384, + "num_tokens": 33307225.0, + "step": 1288 + }, + { + "epoch": 0.14155501866900944, + "grad_norm": 2.162050485610962, + "learning_rate": 7.069154774972558e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6772761344909668, + "num_tokens": 33337248.0, + "step": 1289 + }, + { + "epoch": 0.1416648363716231, + "grad_norm": 2.306509017944336, + "learning_rate": 7.074643249176728e-07, + "loss": 1.0009, + "mean_token_accuracy": 0.6980087757110596, + "num_tokens": 33361937.0, + "step": 1290 + }, + { + "epoch": 0.14177465407423676, + "grad_norm": 2.392178535461426, + "learning_rate": 7.0801317233809e-07, + "loss": 0.9468, + "mean_token_accuracy": 0.7187355756759644, + "num_tokens": 33384764.0, + "step": 1291 + }, + { + "epoch": 0.14188447177685043, + "grad_norm": 2.458216428756714, + "learning_rate": 7.085620197585072e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.6782371997833252, + "num_tokens": 33410776.0, + "step": 1292 + }, + { + "epoch": 0.1419942894794641, + "grad_norm": 1.9184764623641968, + "learning_rate": 7.091108671789242e-07, + "loss": 1.1055, + "mean_token_accuracy": 0.6738030910491943, + "num_tokens": 33444901.0, + "step": 1293 + }, + { + "epoch": 0.14210410718207775, + "grad_norm": 2.284870147705078, + "learning_rate": 7.096597145993414e-07, + "loss": 1.013, + "mean_token_accuracy": 0.6928427219390869, + "num_tokens": 33471129.0, + "step": 1294 + }, + { + "epoch": 0.14221392488469142, + "grad_norm": 2.47232723236084, + "learning_rate": 7.102085620197584e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6721687316894531, + "num_tokens": 33494088.0, + "step": 1295 + }, + { + "epoch": 0.14232374258730507, + "grad_norm": 2.2068376541137695, + "learning_rate": 7.107574094401756e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.7065798044204712, + "num_tokens": 33519106.0, + "step": 1296 + }, + { + "epoch": 0.14243356028991874, + "grad_norm": 2.341526985168457, + "learning_rate": 7.113062568605928e-07, + "loss": 0.9962, + "mean_token_accuracy": 0.6966545581817627, + "num_tokens": 33543179.0, + "step": 1297 + }, + { + "epoch": 0.1425433779925324, + "grad_norm": 2.1531214714050293, + "learning_rate": 7.118551042810098e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6847411394119263, + "num_tokens": 33570022.0, + "step": 1298 + }, + { + "epoch": 0.14265319569514606, + "grad_norm": 2.352123737335205, + "learning_rate": 7.12403951701427e-07, + "loss": 1.1402, + "mean_token_accuracy": 0.6629021763801575, + "num_tokens": 33595723.0, + "step": 1299 + }, + { + "epoch": 0.1427630133977597, + "grad_norm": 2.2029221057891846, + "learning_rate": 7.129527991218442e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.6849510073661804, + "num_tokens": 33621112.0, + "step": 1300 + }, + { + "epoch": 0.14287283110037338, + "grad_norm": 2.290186882019043, + "learning_rate": 7.135016465422611e-07, + "loss": 1.0397, + "mean_token_accuracy": 0.7006228566169739, + "num_tokens": 33646421.0, + "step": 1301 + }, + { + "epoch": 0.14298264880298703, + "grad_norm": 2.31754994392395, + "learning_rate": 7.140504939626783e-07, + "loss": 0.9103, + "mean_token_accuracy": 0.7229747772216797, + "num_tokens": 33670158.0, + "step": 1302 + }, + { + "epoch": 0.1430924665056007, + "grad_norm": 2.142179489135742, + "learning_rate": 7.145993413830955e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6980507373809814, + "num_tokens": 33698936.0, + "step": 1303 + }, + { + "epoch": 0.14320228420821438, + "grad_norm": 2.104795217514038, + "learning_rate": 7.151481888035125e-07, + "loss": 0.9966, + "mean_token_accuracy": 0.7016205787658691, + "num_tokens": 33727530.0, + "step": 1304 + }, + { + "epoch": 0.14331210191082802, + "grad_norm": 2.280703544616699, + "learning_rate": 7.156970362239297e-07, + "loss": 0.9853, + "mean_token_accuracy": 0.7054526209831238, + "num_tokens": 33752132.0, + "step": 1305 + }, + { + "epoch": 0.1434219196134417, + "grad_norm": 1.9766393899917603, + "learning_rate": 7.162458836443468e-07, + "loss": 1.0764, + "mean_token_accuracy": 0.6857814192771912, + "num_tokens": 33785610.0, + "step": 1306 + }, + { + "epoch": 0.14353173731605534, + "grad_norm": 2.2908172607421875, + "learning_rate": 7.167947310647639e-07, + "loss": 0.974, + "mean_token_accuracy": 0.7288268804550171, + "num_tokens": 33809579.0, + "step": 1307 + }, + { + "epoch": 0.14364155501866901, + "grad_norm": 1.9928685426712036, + "learning_rate": 7.173435784851811e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6790460348129272, + "num_tokens": 33840813.0, + "step": 1308 + }, + { + "epoch": 0.14375137272128266, + "grad_norm": 2.3007543087005615, + "learning_rate": 7.178924259055982e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.6840338706970215, + "num_tokens": 33866379.0, + "step": 1309 + }, + { + "epoch": 0.14386119042389633, + "grad_norm": 2.2058141231536865, + "learning_rate": 7.184412733260153e-07, + "loss": 0.9226, + "mean_token_accuracy": 0.7213126420974731, + "num_tokens": 33894579.0, + "step": 1310 + }, + { + "epoch": 0.14397100812650998, + "grad_norm": 2.3082149028778076, + "learning_rate": 7.189901207464325e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6881616711616516, + "num_tokens": 33920883.0, + "step": 1311 + }, + { + "epoch": 0.14408082582912365, + "grad_norm": 2.280940532684326, + "learning_rate": 7.195389681668496e-07, + "loss": 1.069, + "mean_token_accuracy": 0.6801354289054871, + "num_tokens": 33945682.0, + "step": 1312 + }, + { + "epoch": 0.14419064353173733, + "grad_norm": 2.4883103370666504, + "learning_rate": 7.200878155872667e-07, + "loss": 0.9367, + "mean_token_accuracy": 0.7111184597015381, + "num_tokens": 33965428.0, + "step": 1313 + }, + { + "epoch": 0.14430046123435097, + "grad_norm": 2.4055590629577637, + "learning_rate": 7.206366630076838e-07, + "loss": 1.0536, + "mean_token_accuracy": 0.685782790184021, + "num_tokens": 33989032.0, + "step": 1314 + }, + { + "epoch": 0.14441027893696465, + "grad_norm": 2.3390567302703857, + "learning_rate": 7.21185510428101e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6960591673851013, + "num_tokens": 34016149.0, + "step": 1315 + }, + { + "epoch": 0.1445200966395783, + "grad_norm": 1.9876480102539062, + "learning_rate": 7.217343578485181e-07, + "loss": 1.0189, + "mean_token_accuracy": 0.6984747052192688, + "num_tokens": 34049008.0, + "step": 1316 + }, + { + "epoch": 0.14462991434219197, + "grad_norm": 2.1662676334381104, + "learning_rate": 7.222832052689352e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.7057685852050781, + "num_tokens": 34075804.0, + "step": 1317 + }, + { + "epoch": 0.1447397320448056, + "grad_norm": 2.3589468002319336, + "learning_rate": 7.228320526893524e-07, + "loss": 1.0485, + "mean_token_accuracy": 0.6964354515075684, + "num_tokens": 34100773.0, + "step": 1318 + }, + { + "epoch": 0.1448495497474193, + "grad_norm": 2.084632635116577, + "learning_rate": 7.233809001097695e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6798384785652161, + "num_tokens": 34130591.0, + "step": 1319 + }, + { + "epoch": 0.14495936745003293, + "grad_norm": 2.2545275688171387, + "learning_rate": 7.239297475301865e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6726070046424866, + "num_tokens": 34157386.0, + "step": 1320 + }, + { + "epoch": 0.1450691851526466, + "grad_norm": 1.885910153388977, + "learning_rate": 7.244785949506037e-07, + "loss": 1.1331, + "mean_token_accuracy": 0.667427659034729, + "num_tokens": 34195099.0, + "step": 1321 + }, + { + "epoch": 0.14517900285526028, + "grad_norm": 2.5150504112243652, + "learning_rate": 7.250274423710208e-07, + "loss": 1.1066, + "mean_token_accuracy": 0.6758235692977905, + "num_tokens": 34218337.0, + "step": 1322 + }, + { + "epoch": 0.14528882055787393, + "grad_norm": 2.235403299331665, + "learning_rate": 7.255762897914379e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6859370470046997, + "num_tokens": 34245129.0, + "step": 1323 + }, + { + "epoch": 0.1453986382604876, + "grad_norm": 2.023578405380249, + "learning_rate": 7.261251372118551e-07, + "loss": 1.1168, + "mean_token_accuracy": 0.6796953678131104, + "num_tokens": 34278806.0, + "step": 1324 + }, + { + "epoch": 0.14550845596310125, + "grad_norm": 2.264937400817871, + "learning_rate": 7.266739846322721e-07, + "loss": 1.1096, + "mean_token_accuracy": 0.666857898235321, + "num_tokens": 34306548.0, + "step": 1325 + }, + { + "epoch": 0.14561827366571492, + "grad_norm": 2.3556528091430664, + "learning_rate": 7.272228320526893e-07, + "loss": 0.8958, + "mean_token_accuracy": 0.7260100245475769, + "num_tokens": 34330176.0, + "step": 1326 + }, + { + "epoch": 0.14572809136832857, + "grad_norm": 2.5399346351623535, + "learning_rate": 7.277716794731065e-07, + "loss": 0.9707, + "mean_token_accuracy": 0.705350935459137, + "num_tokens": 34351781.0, + "step": 1327 + }, + { + "epoch": 0.14583790907094224, + "grad_norm": 2.196697473526001, + "learning_rate": 7.283205268935235e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.6972188949584961, + "num_tokens": 34378573.0, + "step": 1328 + }, + { + "epoch": 0.14594772677355589, + "grad_norm": 2.4969890117645264, + "learning_rate": 7.288693743139407e-07, + "loss": 0.9975, + "mean_token_accuracy": 0.6961773633956909, + "num_tokens": 34400638.0, + "step": 1329 + }, + { + "epoch": 0.14605754447616956, + "grad_norm": 2.3901174068450928, + "learning_rate": 7.294182217343579e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.6981558203697205, + "num_tokens": 34424367.0, + "step": 1330 + }, + { + "epoch": 0.14616736217878323, + "grad_norm": 2.231799364089966, + "learning_rate": 7.299670691547749e-07, + "loss": 1.0794, + "mean_token_accuracy": 0.6851630806922913, + "num_tokens": 34450635.0, + "step": 1331 + }, + { + "epoch": 0.14627717988139688, + "grad_norm": 2.5221757888793945, + "learning_rate": 7.305159165751921e-07, + "loss": 1.0165, + "mean_token_accuracy": 0.6964387893676758, + "num_tokens": 34473315.0, + "step": 1332 + }, + { + "epoch": 0.14638699758401055, + "grad_norm": 2.3479413986206055, + "learning_rate": 7.310647639956093e-07, + "loss": 1.0957, + "mean_token_accuracy": 0.6750576496124268, + "num_tokens": 34500598.0, + "step": 1333 + }, + { + "epoch": 0.1464968152866242, + "grad_norm": 2.414536714553833, + "learning_rate": 7.316136114160263e-07, + "loss": 1.0619, + "mean_token_accuracy": 0.6851627826690674, + "num_tokens": 34525129.0, + "step": 1334 + }, + { + "epoch": 0.14660663298923787, + "grad_norm": 2.1584994792938232, + "learning_rate": 7.321624588364435e-07, + "loss": 0.9914, + "mean_token_accuracy": 0.7043607831001282, + "num_tokens": 34552939.0, + "step": 1335 + }, + { + "epoch": 0.14671645069185152, + "grad_norm": 2.2226788997650146, + "learning_rate": 7.327113062568605e-07, + "loss": 1.134, + "mean_token_accuracy": 0.6675313711166382, + "num_tokens": 34581252.0, + "step": 1336 + }, + { + "epoch": 0.1468262683944652, + "grad_norm": 2.6622860431671143, + "learning_rate": 7.332601536772777e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6799522638320923, + "num_tokens": 34603625.0, + "step": 1337 + }, + { + "epoch": 0.14693608609707884, + "grad_norm": 2.320611000061035, + "learning_rate": 7.338090010976949e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.6910641193389893, + "num_tokens": 34629634.0, + "step": 1338 + }, + { + "epoch": 0.1470459037996925, + "grad_norm": 2.3666303157806396, + "learning_rate": 7.343578485181118e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6777592897415161, + "num_tokens": 34654123.0, + "step": 1339 + }, + { + "epoch": 0.14715572150230616, + "grad_norm": 2.5041518211364746, + "learning_rate": 7.34906695938529e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6753630638122559, + "num_tokens": 34674639.0, + "step": 1340 + }, + { + "epoch": 0.14726553920491983, + "grad_norm": 2.2923576831817627, + "learning_rate": 7.354555433589462e-07, + "loss": 0.9302, + "mean_token_accuracy": 0.7135177254676819, + "num_tokens": 34696573.0, + "step": 1341 + }, + { + "epoch": 0.1473753569075335, + "grad_norm": 2.237435817718506, + "learning_rate": 7.360043907793632e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6840080618858337, + "num_tokens": 34722587.0, + "step": 1342 + }, + { + "epoch": 0.14748517461014715, + "grad_norm": 2.2295548915863037, + "learning_rate": 7.365532381997804e-07, + "loss": 1.0989, + "mean_token_accuracy": 0.6686511635780334, + "num_tokens": 34750809.0, + "step": 1343 + }, + { + "epoch": 0.14759499231276083, + "grad_norm": 2.4368298053741455, + "learning_rate": 7.371020856201976e-07, + "loss": 0.9296, + "mean_token_accuracy": 0.7112667560577393, + "num_tokens": 34772399.0, + "step": 1344 + }, + { + "epoch": 0.14770481001537447, + "grad_norm": 2.280601978302002, + "learning_rate": 7.376509330406146e-07, + "loss": 1.0164, + "mean_token_accuracy": 0.7010036110877991, + "num_tokens": 34799409.0, + "step": 1345 + }, + { + "epoch": 0.14781462771798815, + "grad_norm": 2.3643298149108887, + "learning_rate": 7.381997804610318e-07, + "loss": 1.0808, + "mean_token_accuracy": 0.6888642907142639, + "num_tokens": 34823792.0, + "step": 1346 + }, + { + "epoch": 0.1479244454206018, + "grad_norm": 2.3892993927001953, + "learning_rate": 7.387486278814489e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.6908868551254272, + "num_tokens": 34848102.0, + "step": 1347 + }, + { + "epoch": 0.14803426312321546, + "grad_norm": 2.3629565238952637, + "learning_rate": 7.39297475301866e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6742020845413208, + "num_tokens": 34873309.0, + "step": 1348 + }, + { + "epoch": 0.1481440808258291, + "grad_norm": 2.191333532333374, + "learning_rate": 7.398463227222832e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6853519082069397, + "num_tokens": 34901942.0, + "step": 1349 + }, + { + "epoch": 0.14825389852844278, + "grad_norm": 2.522627353668213, + "learning_rate": 7.403951701427003e-07, + "loss": 1.0151, + "mean_token_accuracy": 0.6927398443222046, + "num_tokens": 34924155.0, + "step": 1350 + }, + { + "epoch": 0.14836371623105646, + "grad_norm": 2.1788575649261475, + "learning_rate": 7.409440175631174e-07, + "loss": 1.0992, + "mean_token_accuracy": 0.6779065728187561, + "num_tokens": 34952940.0, + "step": 1351 + }, + { + "epoch": 0.1484735339336701, + "grad_norm": 2.501674175262451, + "learning_rate": 7.414928649835346e-07, + "loss": 0.9928, + "mean_token_accuracy": 0.6976244449615479, + "num_tokens": 34973659.0, + "step": 1352 + }, + { + "epoch": 0.14858335163628378, + "grad_norm": 2.162933349609375, + "learning_rate": 7.420417124039517e-07, + "loss": 1.0066, + "mean_token_accuracy": 0.6928379535675049, + "num_tokens": 35001334.0, + "step": 1353 + }, + { + "epoch": 0.14869316933889742, + "grad_norm": 2.2175045013427734, + "learning_rate": 7.425905598243688e-07, + "loss": 0.9791, + "mean_token_accuracy": 0.7038810849189758, + "num_tokens": 35026199.0, + "step": 1354 + }, + { + "epoch": 0.1488029870415111, + "grad_norm": 2.4070301055908203, + "learning_rate": 7.43139407244786e-07, + "loss": 1.0874, + "mean_token_accuracy": 0.6794545650482178, + "num_tokens": 35050867.0, + "step": 1355 + }, + { + "epoch": 0.14891280474412474, + "grad_norm": 2.310142993927002, + "learning_rate": 7.436882546652031e-07, + "loss": 0.9863, + "mean_token_accuracy": 0.7009339332580566, + "num_tokens": 35076362.0, + "step": 1356 + }, + { + "epoch": 0.14902262244673842, + "grad_norm": 2.02964186668396, + "learning_rate": 7.442371020856202e-07, + "loss": 1.0249, + "mean_token_accuracy": 0.6941843628883362, + "num_tokens": 35108192.0, + "step": 1357 + }, + { + "epoch": 0.14913244014935206, + "grad_norm": 2.4329349994659424, + "learning_rate": 7.447859495060372e-07, + "loss": 1.0592, + "mean_token_accuracy": 0.68939208984375, + "num_tokens": 35131006.0, + "step": 1358 + }, + { + "epoch": 0.14924225785196574, + "grad_norm": 2.1522622108459473, + "learning_rate": 7.453347969264544e-07, + "loss": 0.9967, + "mean_token_accuracy": 0.6941715478897095, + "num_tokens": 35157539.0, + "step": 1359 + }, + { + "epoch": 0.1493520755545794, + "grad_norm": 2.3516149520874023, + "learning_rate": 7.458836443468715e-07, + "loss": 1.03, + "mean_token_accuracy": 0.693190336227417, + "num_tokens": 35182383.0, + "step": 1360 + }, + { + "epoch": 0.14946189325719306, + "grad_norm": 2.206960916519165, + "learning_rate": 7.464324917672886e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6852564811706543, + "num_tokens": 35211590.0, + "step": 1361 + }, + { + "epoch": 0.14957171095980673, + "grad_norm": 2.3747541904449463, + "learning_rate": 7.469813391877058e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6783367395401001, + "num_tokens": 35237743.0, + "step": 1362 + }, + { + "epoch": 0.14968152866242038, + "grad_norm": 2.088074207305908, + "learning_rate": 7.475301866081229e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.675773561000824, + "num_tokens": 35267517.0, + "step": 1363 + }, + { + "epoch": 0.14979134636503405, + "grad_norm": 2.111032724380493, + "learning_rate": 7.4807903402854e-07, + "loss": 0.9888, + "mean_token_accuracy": 0.7022457122802734, + "num_tokens": 35297217.0, + "step": 1364 + }, + { + "epoch": 0.1499011640676477, + "grad_norm": 2.085162401199341, + "learning_rate": 7.486278814489572e-07, + "loss": 1.0511, + "mean_token_accuracy": 0.6835318207740784, + "num_tokens": 35327756.0, + "step": 1365 + }, + { + "epoch": 0.15001098177026137, + "grad_norm": 2.1751766204833984, + "learning_rate": 7.491767288693743e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.6788821220397949, + "num_tokens": 35356908.0, + "step": 1366 + }, + { + "epoch": 0.15012079947287502, + "grad_norm": 2.466546058654785, + "learning_rate": 7.497255762897914e-07, + "loss": 0.9764, + "mean_token_accuracy": 0.704409658908844, + "num_tokens": 35379932.0, + "step": 1367 + }, + { + "epoch": 0.1502306171754887, + "grad_norm": 2.2093565464019775, + "learning_rate": 7.502744237102086e-07, + "loss": 1.018, + "mean_token_accuracy": 0.6988373398780823, + "num_tokens": 35407737.0, + "step": 1368 + }, + { + "epoch": 0.15034043487810236, + "grad_norm": 2.3900201320648193, + "learning_rate": 7.508232711306256e-07, + "loss": 1.0504, + "mean_token_accuracy": 0.6855587363243103, + "num_tokens": 35431153.0, + "step": 1369 + }, + { + "epoch": 0.150450252580716, + "grad_norm": 2.34234356880188, + "learning_rate": 7.513721185510428e-07, + "loss": 1.0713, + "mean_token_accuracy": 0.679279088973999, + "num_tokens": 35458650.0, + "step": 1370 + }, + { + "epoch": 0.15056007028332968, + "grad_norm": 2.3616809844970703, + "learning_rate": 7.5192096597146e-07, + "loss": 1.0067, + "mean_token_accuracy": 0.702593207359314, + "num_tokens": 35483267.0, + "step": 1371 + }, + { + "epoch": 0.15066988798594333, + "grad_norm": 2.1625919342041016, + "learning_rate": 7.52469813391877e-07, + "loss": 1.0236, + "mean_token_accuracy": 0.6992337703704834, + "num_tokens": 35510839.0, + "step": 1372 + }, + { + "epoch": 0.150779705688557, + "grad_norm": 2.1153628826141357, + "learning_rate": 7.530186608122942e-07, + "loss": 1.1677, + "mean_token_accuracy": 0.6617355346679688, + "num_tokens": 35542698.0, + "step": 1373 + }, + { + "epoch": 0.15088952339117065, + "grad_norm": 2.6682417392730713, + "learning_rate": 7.535675082327113e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.688012957572937, + "num_tokens": 35563808.0, + "step": 1374 + }, + { + "epoch": 0.15099934109378432, + "grad_norm": 2.440830945968628, + "learning_rate": 7.541163556531284e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7015272378921509, + "num_tokens": 35586539.0, + "step": 1375 + }, + { + "epoch": 0.15110915879639797, + "grad_norm": 2.380641460418701, + "learning_rate": 7.546652030735456e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6845986843109131, + "num_tokens": 35611293.0, + "step": 1376 + }, + { + "epoch": 0.15121897649901164, + "grad_norm": 2.328315258026123, + "learning_rate": 7.552140504939627e-07, + "loss": 0.9735, + "mean_token_accuracy": 0.7027246952056885, + "num_tokens": 35634887.0, + "step": 1377 + }, + { + "epoch": 0.1513287942016253, + "grad_norm": 2.06182599067688, + "learning_rate": 7.557628979143797e-07, + "loss": 1.0949, + "mean_token_accuracy": 0.6841898560523987, + "num_tokens": 35664322.0, + "step": 1378 + }, + { + "epoch": 0.15143861190423896, + "grad_norm": 2.4042017459869385, + "learning_rate": 7.563117453347969e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.683177649974823, + "num_tokens": 35687610.0, + "step": 1379 + }, + { + "epoch": 0.15154842960685264, + "grad_norm": 2.142677068710327, + "learning_rate": 7.568605927552139e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6859540343284607, + "num_tokens": 35715234.0, + "step": 1380 + }, + { + "epoch": 0.15165824730946628, + "grad_norm": 2.4137637615203857, + "learning_rate": 7.574094401756311e-07, + "loss": 0.9729, + "mean_token_accuracy": 0.7058082222938538, + "num_tokens": 35738990.0, + "step": 1381 + }, + { + "epoch": 0.15176806501207996, + "grad_norm": 2.3567721843719482, + "learning_rate": 7.579582875960483e-07, + "loss": 1.128, + "mean_token_accuracy": 0.660635232925415, + "num_tokens": 35763339.0, + "step": 1382 + }, + { + "epoch": 0.1518778827146936, + "grad_norm": 2.2969865798950195, + "learning_rate": 7.585071350164653e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6767973899841309, + "num_tokens": 35791131.0, + "step": 1383 + }, + { + "epoch": 0.15198770041730728, + "grad_norm": 2.2089626789093018, + "learning_rate": 7.590559824368825e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7106653451919556, + "num_tokens": 35815661.0, + "step": 1384 + }, + { + "epoch": 0.15209751811992092, + "grad_norm": 2.1371824741363525, + "learning_rate": 7.596048298572997e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6681792736053467, + "num_tokens": 35845268.0, + "step": 1385 + }, + { + "epoch": 0.1522073358225346, + "grad_norm": 2.5313527584075928, + "learning_rate": 7.601536772777167e-07, + "loss": 1.0147, + "mean_token_accuracy": 0.6944137215614319, + "num_tokens": 35870076.0, + "step": 1386 + }, + { + "epoch": 0.15231715352514824, + "grad_norm": 2.01857852935791, + "learning_rate": 7.607025246981339e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.679000735282898, + "num_tokens": 35901295.0, + "step": 1387 + }, + { + "epoch": 0.15242697122776191, + "grad_norm": 2.148887872695923, + "learning_rate": 7.612513721185511e-07, + "loss": 1.0202, + "mean_token_accuracy": 0.6926007866859436, + "num_tokens": 35928627.0, + "step": 1388 + }, + { + "epoch": 0.1525367889303756, + "grad_norm": 2.353510618209839, + "learning_rate": 7.618002195389681e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6742778420448303, + "num_tokens": 35955054.0, + "step": 1389 + }, + { + "epoch": 0.15264660663298923, + "grad_norm": 2.265920639038086, + "learning_rate": 7.623490669593853e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6863700151443481, + "num_tokens": 35982902.0, + "step": 1390 + }, + { + "epoch": 0.1527564243356029, + "grad_norm": 2.4208788871765137, + "learning_rate": 7.628979143798024e-07, + "loss": 1.0207, + "mean_token_accuracy": 0.6927981376647949, + "num_tokens": 36006411.0, + "step": 1391 + }, + { + "epoch": 0.15286624203821655, + "grad_norm": 2.2819952964782715, + "learning_rate": 7.634467618002195e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.688751220703125, + "num_tokens": 36033290.0, + "step": 1392 + }, + { + "epoch": 0.15297605974083023, + "grad_norm": 2.4338185787200928, + "learning_rate": 7.639956092206367e-07, + "loss": 1.0628, + "mean_token_accuracy": 0.6833241581916809, + "num_tokens": 36056919.0, + "step": 1393 + }, + { + "epoch": 0.15308587744344387, + "grad_norm": 2.2772955894470215, + "learning_rate": 7.645444566410538e-07, + "loss": 1.1121, + "mean_token_accuracy": 0.6686204671859741, + "num_tokens": 36084272.0, + "step": 1394 + }, + { + "epoch": 0.15319569514605755, + "grad_norm": 2.211148738861084, + "learning_rate": 7.650933040614709e-07, + "loss": 0.9373, + "mean_token_accuracy": 0.7274336218833923, + "num_tokens": 36110307.0, + "step": 1395 + }, + { + "epoch": 0.1533055128486712, + "grad_norm": 2.1438634395599365, + "learning_rate": 7.656421514818881e-07, + "loss": 1.0414, + "mean_token_accuracy": 0.6914363503456116, + "num_tokens": 36139538.0, + "step": 1396 + }, + { + "epoch": 0.15341533055128487, + "grad_norm": 2.3062212467193604, + "learning_rate": 7.66190998902305e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6997917890548706, + "num_tokens": 36165960.0, + "step": 1397 + }, + { + "epoch": 0.15352514825389854, + "grad_norm": 1.9459320306777954, + "learning_rate": 7.667398463227222e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6840060949325562, + "num_tokens": 36199425.0, + "step": 1398 + }, + { + "epoch": 0.1536349659565122, + "grad_norm": 2.1839962005615234, + "learning_rate": 7.672886937431394e-07, + "loss": 1.1275, + "mean_token_accuracy": 0.6654109358787537, + "num_tokens": 36228420.0, + "step": 1399 + }, + { + "epoch": 0.15374478365912586, + "grad_norm": 2.5124998092651367, + "learning_rate": 7.678375411635564e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.6889686584472656, + "num_tokens": 36250450.0, + "step": 1400 + }, + { + "epoch": 0.1538546013617395, + "grad_norm": 2.45137619972229, + "learning_rate": 7.683863885839736e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7025374174118042, + "num_tokens": 36272755.0, + "step": 1401 + }, + { + "epoch": 0.15396441906435318, + "grad_norm": 2.1080377101898193, + "learning_rate": 7.689352360043907e-07, + "loss": 1.0463, + "mean_token_accuracy": 0.6845667958259583, + "num_tokens": 36301596.0, + "step": 1402 + }, + { + "epoch": 0.15407423676696683, + "grad_norm": 2.298405647277832, + "learning_rate": 7.694840834248078e-07, + "loss": 1.0978, + "mean_token_accuracy": 0.6861289739608765, + "num_tokens": 36327116.0, + "step": 1403 + }, + { + "epoch": 0.1541840544695805, + "grad_norm": 2.3860743045806885, + "learning_rate": 7.70032930845225e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6738840341567993, + "num_tokens": 36350676.0, + "step": 1404 + }, + { + "epoch": 0.15429387217219415, + "grad_norm": 2.182370901107788, + "learning_rate": 7.705817782656421e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7073382139205933, + "num_tokens": 36376806.0, + "step": 1405 + }, + { + "epoch": 0.15440368987480782, + "grad_norm": 2.420769453048706, + "learning_rate": 7.711306256860592e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6904512643814087, + "num_tokens": 36399293.0, + "step": 1406 + }, + { + "epoch": 0.1545135075774215, + "grad_norm": 2.7095108032226562, + "learning_rate": 7.716794731064764e-07, + "loss": 1.0333, + "mean_token_accuracy": 0.6886758804321289, + "num_tokens": 36420674.0, + "step": 1407 + }, + { + "epoch": 0.15462332528003514, + "grad_norm": 2.507371425628662, + "learning_rate": 7.722283205268935e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.7060544490814209, + "num_tokens": 36443718.0, + "step": 1408 + }, + { + "epoch": 0.1547331429826488, + "grad_norm": 2.4656779766082764, + "learning_rate": 7.727771679473106e-07, + "loss": 1.0386, + "mean_token_accuracy": 0.6880729794502258, + "num_tokens": 36468038.0, + "step": 1409 + }, + { + "epoch": 0.15484296068526246, + "grad_norm": 2.2582483291625977, + "learning_rate": 7.733260153677278e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6782144904136658, + "num_tokens": 36495140.0, + "step": 1410 + }, + { + "epoch": 0.15495277838787613, + "grad_norm": 2.3863461017608643, + "learning_rate": 7.738748627881449e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6880670785903931, + "num_tokens": 36519690.0, + "step": 1411 + }, + { + "epoch": 0.15506259609048978, + "grad_norm": 2.258880376815796, + "learning_rate": 7.74423710208562e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.7034817934036255, + "num_tokens": 36545779.0, + "step": 1412 + }, + { + "epoch": 0.15517241379310345, + "grad_norm": 2.194366455078125, + "learning_rate": 7.749725576289791e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6791149973869324, + "num_tokens": 36572099.0, + "step": 1413 + }, + { + "epoch": 0.1552822314957171, + "grad_norm": 2.425492286682129, + "learning_rate": 7.755214050493963e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6879798769950867, + "num_tokens": 36593968.0, + "step": 1414 + }, + { + "epoch": 0.15539204919833077, + "grad_norm": 2.0834245681762695, + "learning_rate": 7.760702524698134e-07, + "loss": 0.9449, + "mean_token_accuracy": 0.7122049927711487, + "num_tokens": 36620724.0, + "step": 1415 + }, + { + "epoch": 0.15550186690094442, + "grad_norm": 2.2746667861938477, + "learning_rate": 7.766190998902304e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.7042897343635559, + "num_tokens": 36646019.0, + "step": 1416 + }, + { + "epoch": 0.1556116846035581, + "grad_norm": 2.2420613765716553, + "learning_rate": 7.771679473106476e-07, + "loss": 0.9756, + "mean_token_accuracy": 0.7093145847320557, + "num_tokens": 36670676.0, + "step": 1417 + }, + { + "epoch": 0.15572150230617177, + "grad_norm": 2.2435550689697266, + "learning_rate": 7.777167947310647e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.688526451587677, + "num_tokens": 36697556.0, + "step": 1418 + }, + { + "epoch": 0.1558313200087854, + "grad_norm": 2.213421583175659, + "learning_rate": 7.782656421514818e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.6994360685348511, + "num_tokens": 36723490.0, + "step": 1419 + }, + { + "epoch": 0.15594113771139909, + "grad_norm": 2.043449640274048, + "learning_rate": 7.78814489571899e-07, + "loss": 1.1096, + "mean_token_accuracy": 0.6727668642997742, + "num_tokens": 36758197.0, + "step": 1420 + }, + { + "epoch": 0.15605095541401273, + "grad_norm": 2.142083168029785, + "learning_rate": 7.793633369923161e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6796813011169434, + "num_tokens": 36787700.0, + "step": 1421 + }, + { + "epoch": 0.1561607731166264, + "grad_norm": 2.5180513858795166, + "learning_rate": 7.799121844127332e-07, + "loss": 1.1313, + "mean_token_accuracy": 0.666046679019928, + "num_tokens": 36810721.0, + "step": 1422 + }, + { + "epoch": 0.15627059081924005, + "grad_norm": 2.193394184112549, + "learning_rate": 7.804610318331504e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6628466844558716, + "num_tokens": 36840618.0, + "step": 1423 + }, + { + "epoch": 0.15638040852185373, + "grad_norm": 2.3751752376556396, + "learning_rate": 7.810098792535674e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6942270994186401, + "num_tokens": 36864661.0, + "step": 1424 + }, + { + "epoch": 0.15649022622446737, + "grad_norm": 2.360691785812378, + "learning_rate": 7.815587266739846e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6974284052848816, + "num_tokens": 36888779.0, + "step": 1425 + }, + { + "epoch": 0.15660004392708105, + "grad_norm": 2.267709255218506, + "learning_rate": 7.821075740944018e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6897562742233276, + "num_tokens": 36916313.0, + "step": 1426 + }, + { + "epoch": 0.15670986162969472, + "grad_norm": 2.0548086166381836, + "learning_rate": 7.826564215148188e-07, + "loss": 1.0911, + "mean_token_accuracy": 0.6808615922927856, + "num_tokens": 36948943.0, + "step": 1427 + }, + { + "epoch": 0.15681967933230836, + "grad_norm": 2.442119836807251, + "learning_rate": 7.83205268935236e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.6828582286834717, + "num_tokens": 36972796.0, + "step": 1428 + }, + { + "epoch": 0.15692949703492204, + "grad_norm": 2.137317419052124, + "learning_rate": 7.837541163556532e-07, + "loss": 1.0065, + "mean_token_accuracy": 0.697424054145813, + "num_tokens": 37000283.0, + "step": 1429 + }, + { + "epoch": 0.15703931473753568, + "grad_norm": 2.157100200653076, + "learning_rate": 7.843029637760702e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7182517051696777, + "num_tokens": 37026556.0, + "step": 1430 + }, + { + "epoch": 0.15714913244014936, + "grad_norm": 2.4248509407043457, + "learning_rate": 7.848518111964874e-07, + "loss": 0.9109, + "mean_token_accuracy": 0.7174056768417358, + "num_tokens": 37047463.0, + "step": 1431 + }, + { + "epoch": 0.157258950142763, + "grad_norm": 2.603097438812256, + "learning_rate": 7.854006586169045e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6854052543640137, + "num_tokens": 37069831.0, + "step": 1432 + }, + { + "epoch": 0.15736876784537668, + "grad_norm": 2.4641714096069336, + "learning_rate": 7.859495060373216e-07, + "loss": 1.1111, + "mean_token_accuracy": 0.6746992468833923, + "num_tokens": 37094358.0, + "step": 1433 + }, + { + "epoch": 0.15747858554799032, + "grad_norm": 2.2273807525634766, + "learning_rate": 7.864983534577388e-07, + "loss": 0.9561, + "mean_token_accuracy": 0.7108427286148071, + "num_tokens": 37120750.0, + "step": 1434 + }, + { + "epoch": 0.157588403250604, + "grad_norm": 2.5061752796173096, + "learning_rate": 7.870472008781557e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.6961339712142944, + "num_tokens": 37144801.0, + "step": 1435 + }, + { + "epoch": 0.15769822095321767, + "grad_norm": 2.3162007331848145, + "learning_rate": 7.875960482985729e-07, + "loss": 0.9421, + "mean_token_accuracy": 0.7210328578948975, + "num_tokens": 37168525.0, + "step": 1436 + }, + { + "epoch": 0.15780803865583132, + "grad_norm": 2.258518695831299, + "learning_rate": 7.881448957189901e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6721923351287842, + "num_tokens": 37196945.0, + "step": 1437 + }, + { + "epoch": 0.157917856358445, + "grad_norm": 2.1722805500030518, + "learning_rate": 7.886937431394071e-07, + "loss": 1.0817, + "mean_token_accuracy": 0.6734873056411743, + "num_tokens": 37225878.0, + "step": 1438 + }, + { + "epoch": 0.15802767406105864, + "grad_norm": 2.3825843334198, + "learning_rate": 7.892425905598243e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.6971964836120605, + "num_tokens": 37249624.0, + "step": 1439 + }, + { + "epoch": 0.1581374917636723, + "grad_norm": 2.334827184677124, + "learning_rate": 7.897914379802415e-07, + "loss": 0.9453, + "mean_token_accuracy": 0.7158329486846924, + "num_tokens": 37274302.0, + "step": 1440 + }, + { + "epoch": 0.15824730946628596, + "grad_norm": 2.316854238510132, + "learning_rate": 7.903402854006585e-07, + "loss": 0.983, + "mean_token_accuracy": 0.703854501247406, + "num_tokens": 37299364.0, + "step": 1441 + }, + { + "epoch": 0.15835712716889963, + "grad_norm": 2.549060344696045, + "learning_rate": 7.908891328210757e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6887112259864807, + "num_tokens": 37323140.0, + "step": 1442 + }, + { + "epoch": 0.15846694487151328, + "grad_norm": 2.4470012187957764, + "learning_rate": 7.914379802414928e-07, + "loss": 0.9123, + "mean_token_accuracy": 0.7219023108482361, + "num_tokens": 37345343.0, + "step": 1443 + }, + { + "epoch": 0.15857676257412695, + "grad_norm": 2.0712497234344482, + "learning_rate": 7.919868276619099e-07, + "loss": 0.9394, + "mean_token_accuracy": 0.71095871925354, + "num_tokens": 37375321.0, + "step": 1444 + }, + { + "epoch": 0.15868658027674062, + "grad_norm": 2.284848690032959, + "learning_rate": 7.925356750823271e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6965764760971069, + "num_tokens": 37399684.0, + "step": 1445 + }, + { + "epoch": 0.15879639797935427, + "grad_norm": 1.9204319715499878, + "learning_rate": 7.930845225027442e-07, + "loss": 0.981, + "mean_token_accuracy": 0.7003376483917236, + "num_tokens": 37430736.0, + "step": 1446 + }, + { + "epoch": 0.15890621568196794, + "grad_norm": 2.219251871109009, + "learning_rate": 7.936333699231613e-07, + "loss": 1.0908, + "mean_token_accuracy": 0.677056610584259, + "num_tokens": 37458614.0, + "step": 1447 + }, + { + "epoch": 0.1590160333845816, + "grad_norm": 2.5725088119506836, + "learning_rate": 7.941822173435785e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.7004644870758057, + "num_tokens": 37480573.0, + "step": 1448 + }, + { + "epoch": 0.15912585108719526, + "grad_norm": 2.3243374824523926, + "learning_rate": 7.947310647639956e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6890665888786316, + "num_tokens": 37504804.0, + "step": 1449 + }, + { + "epoch": 0.1592356687898089, + "grad_norm": 2.419874668121338, + "learning_rate": 7.952799121844127e-07, + "loss": 0.9662, + "mean_token_accuracy": 0.7074979543685913, + "num_tokens": 37529385.0, + "step": 1450 + }, + { + "epoch": 0.15934548649242258, + "grad_norm": 2.3480336666107178, + "learning_rate": 7.958287596048299e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.6819173097610474, + "num_tokens": 37554273.0, + "step": 1451 + }, + { + "epoch": 0.15945530419503623, + "grad_norm": 2.0910048484802246, + "learning_rate": 7.96377607025247e-07, + "loss": 0.9645, + "mean_token_accuracy": 0.7115857601165771, + "num_tokens": 37582512.0, + "step": 1452 + }, + { + "epoch": 0.1595651218976499, + "grad_norm": 2.289750337600708, + "learning_rate": 7.969264544456641e-07, + "loss": 1.0563, + "mean_token_accuracy": 0.6890184879302979, + "num_tokens": 37606063.0, + "step": 1453 + }, + { + "epoch": 0.15967493960026355, + "grad_norm": 2.404696464538574, + "learning_rate": 7.974753018660811e-07, + "loss": 1.1457, + "mean_token_accuracy": 0.6667605638504028, + "num_tokens": 37632676.0, + "step": 1454 + }, + { + "epoch": 0.15978475730287722, + "grad_norm": 2.098749876022339, + "learning_rate": 7.980241492864983e-07, + "loss": 1.0501, + "mean_token_accuracy": 0.6933350563049316, + "num_tokens": 37660737.0, + "step": 1455 + }, + { + "epoch": 0.1598945750054909, + "grad_norm": 2.2677509784698486, + "learning_rate": 7.985729967069154e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6872488260269165, + "num_tokens": 37688663.0, + "step": 1456 + }, + { + "epoch": 0.16000439270810454, + "grad_norm": 2.423413038253784, + "learning_rate": 7.991218441273325e-07, + "loss": 0.9994, + "mean_token_accuracy": 0.6935908794403076, + "num_tokens": 37712996.0, + "step": 1457 + }, + { + "epoch": 0.16011421041071822, + "grad_norm": 2.420290470123291, + "learning_rate": 7.996706915477497e-07, + "loss": 1.0197, + "mean_token_accuracy": 0.6973977088928223, + "num_tokens": 37735608.0, + "step": 1458 + }, + { + "epoch": 0.16022402811333186, + "grad_norm": 2.041200876235962, + "learning_rate": 8.002195389681668e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6956201195716858, + "num_tokens": 37766653.0, + "step": 1459 + }, + { + "epoch": 0.16033384581594554, + "grad_norm": 2.319641351699829, + "learning_rate": 8.007683863885839e-07, + "loss": 1.0694, + "mean_token_accuracy": 0.6890512108802795, + "num_tokens": 37792746.0, + "step": 1460 + }, + { + "epoch": 0.16044366351855918, + "grad_norm": 2.168734073638916, + "learning_rate": 8.013172338090011e-07, + "loss": 0.9747, + "mean_token_accuracy": 0.7110711336135864, + "num_tokens": 37819732.0, + "step": 1461 + }, + { + "epoch": 0.16055348122117286, + "grad_norm": 2.5495645999908447, + "learning_rate": 8.018660812294182e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6834225654602051, + "num_tokens": 37841931.0, + "step": 1462 + }, + { + "epoch": 0.1606632989237865, + "grad_norm": 2.230851173400879, + "learning_rate": 8.024149286498353e-07, + "loss": 0.9901, + "mean_token_accuracy": 0.6994974613189697, + "num_tokens": 37870618.0, + "step": 1463 + }, + { + "epoch": 0.16077311662640018, + "grad_norm": 2.230595350265503, + "learning_rate": 8.029637760702525e-07, + "loss": 1.0291, + "mean_token_accuracy": 0.6920934915542603, + "num_tokens": 37898385.0, + "step": 1464 + }, + { + "epoch": 0.16088293432901385, + "grad_norm": 2.1864078044891357, + "learning_rate": 8.035126234906695e-07, + "loss": 1.0344, + "mean_token_accuracy": 0.6869654655456543, + "num_tokens": 37926304.0, + "step": 1465 + }, + { + "epoch": 0.1609927520316275, + "grad_norm": 2.066760778427124, + "learning_rate": 8.040614709110867e-07, + "loss": 1.0014, + "mean_token_accuracy": 0.7059741616249084, + "num_tokens": 37956073.0, + "step": 1466 + }, + { + "epoch": 0.16110256973424117, + "grad_norm": 2.2861015796661377, + "learning_rate": 8.046103183315039e-07, + "loss": 0.9664, + "mean_token_accuracy": 0.705944299697876, + "num_tokens": 37981975.0, + "step": 1467 + }, + { + "epoch": 0.16121238743685481, + "grad_norm": 2.359776258468628, + "learning_rate": 8.051591657519209e-07, + "loss": 0.9925, + "mean_token_accuracy": 0.7021693587303162, + "num_tokens": 38006263.0, + "step": 1468 + }, + { + "epoch": 0.1613222051394685, + "grad_norm": 2.243635416030884, + "learning_rate": 8.057080131723381e-07, + "loss": 1.1437, + "mean_token_accuracy": 0.6581497192382812, + "num_tokens": 38035725.0, + "step": 1469 + }, + { + "epoch": 0.16143202284208213, + "grad_norm": 2.4487156867980957, + "learning_rate": 8.062568605927553e-07, + "loss": 1.0045, + "mean_token_accuracy": 0.6987505555152893, + "num_tokens": 38058425.0, + "step": 1470 + }, + { + "epoch": 0.1615418405446958, + "grad_norm": 2.4559810161590576, + "learning_rate": 8.068057080131723e-07, + "loss": 0.9817, + "mean_token_accuracy": 0.7038246393203735, + "num_tokens": 38079339.0, + "step": 1471 + }, + { + "epoch": 0.16165165824730945, + "grad_norm": 2.399083137512207, + "learning_rate": 8.073545554335894e-07, + "loss": 0.9915, + "mean_token_accuracy": 0.7024737596511841, + "num_tokens": 38102338.0, + "step": 1472 + }, + { + "epoch": 0.16176147594992313, + "grad_norm": 2.2027175426483154, + "learning_rate": 8.079034028540066e-07, + "loss": 1.1212, + "mean_token_accuracy": 0.6672741174697876, + "num_tokens": 38133576.0, + "step": 1473 + }, + { + "epoch": 0.1618712936525368, + "grad_norm": 2.302802085876465, + "learning_rate": 8.084522502744236e-07, + "loss": 0.9943, + "mean_token_accuracy": 0.7008047699928284, + "num_tokens": 38161708.0, + "step": 1474 + }, + { + "epoch": 0.16198111135515045, + "grad_norm": 2.0716652870178223, + "learning_rate": 8.090010976948408e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6799691915512085, + "num_tokens": 38193054.0, + "step": 1475 + }, + { + "epoch": 0.16209092905776412, + "grad_norm": 2.384406805038452, + "learning_rate": 8.095499451152578e-07, + "loss": 1.109, + "mean_token_accuracy": 0.6744005084037781, + "num_tokens": 38217086.0, + "step": 1476 + }, + { + "epoch": 0.16220074676037777, + "grad_norm": 2.6659600734710693, + "learning_rate": 8.10098792535675e-07, + "loss": 0.9472, + "mean_token_accuracy": 0.7130717039108276, + "num_tokens": 38237100.0, + "step": 1477 + }, + { + "epoch": 0.16231056446299144, + "grad_norm": 2.502467155456543, + "learning_rate": 8.106476399560922e-07, + "loss": 1.0269, + "mean_token_accuracy": 0.6931507587432861, + "num_tokens": 38259674.0, + "step": 1478 + }, + { + "epoch": 0.1624203821656051, + "grad_norm": 2.2542877197265625, + "learning_rate": 8.111964873765092e-07, + "loss": 1.0159, + "mean_token_accuracy": 0.6961650252342224, + "num_tokens": 38284736.0, + "step": 1479 + }, + { + "epoch": 0.16253019986821876, + "grad_norm": 2.593318223953247, + "learning_rate": 8.117453347969264e-07, + "loss": 0.9428, + "mean_token_accuracy": 0.7132692337036133, + "num_tokens": 38305080.0, + "step": 1480 + }, + { + "epoch": 0.1626400175708324, + "grad_norm": 2.3921451568603516, + "learning_rate": 8.122941822173436e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.6949725151062012, + "num_tokens": 38329346.0, + "step": 1481 + }, + { + "epoch": 0.16274983527344608, + "grad_norm": 2.06760311126709, + "learning_rate": 8.128430296377606e-07, + "loss": 1.1098, + "mean_token_accuracy": 0.6682356595993042, + "num_tokens": 38361693.0, + "step": 1482 + }, + { + "epoch": 0.16285965297605975, + "grad_norm": 2.250300168991089, + "learning_rate": 8.133918770581778e-07, + "loss": 0.9608, + "mean_token_accuracy": 0.7088158130645752, + "num_tokens": 38388034.0, + "step": 1483 + }, + { + "epoch": 0.1629694706786734, + "grad_norm": 2.3077704906463623, + "learning_rate": 8.13940724478595e-07, + "loss": 0.9863, + "mean_token_accuracy": 0.7008998394012451, + "num_tokens": 38412971.0, + "step": 1484 + }, + { + "epoch": 0.16307928838128707, + "grad_norm": 2.440343141555786, + "learning_rate": 8.14489571899012e-07, + "loss": 1.0459, + "mean_token_accuracy": 0.6924830079078674, + "num_tokens": 38436424.0, + "step": 1485 + }, + { + "epoch": 0.16318910608390072, + "grad_norm": 2.3304083347320557, + "learning_rate": 8.150384193194292e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6856567859649658, + "num_tokens": 38463220.0, + "step": 1486 + }, + { + "epoch": 0.1632989237865144, + "grad_norm": 2.214980125427246, + "learning_rate": 8.155872667398463e-07, + "loss": 1.1163, + "mean_token_accuracy": 0.688467264175415, + "num_tokens": 38491239.0, + "step": 1487 + }, + { + "epoch": 0.16340874148912804, + "grad_norm": 2.9331469535827637, + "learning_rate": 8.161361141602634e-07, + "loss": 0.8662, + "mean_token_accuracy": 0.733447790145874, + "num_tokens": 38508299.0, + "step": 1488 + }, + { + "epoch": 0.1635185591917417, + "grad_norm": 2.3906610012054443, + "learning_rate": 8.166849615806806e-07, + "loss": 1.0684, + "mean_token_accuracy": 0.6806473731994629, + "num_tokens": 38533657.0, + "step": 1489 + }, + { + "epoch": 0.16362837689435536, + "grad_norm": 2.2483558654785156, + "learning_rate": 8.172338090010977e-07, + "loss": 1.0966, + "mean_token_accuracy": 0.6786621809005737, + "num_tokens": 38564589.0, + "step": 1490 + }, + { + "epoch": 0.16373819459696903, + "grad_norm": 2.29703426361084, + "learning_rate": 8.177826564215147e-07, + "loss": 1.0787, + "mean_token_accuracy": 0.6816752552986145, + "num_tokens": 38590465.0, + "step": 1491 + }, + { + "epoch": 0.16384801229958268, + "grad_norm": 2.2529094219207764, + "learning_rate": 8.183315038419319e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.7002612352371216, + "num_tokens": 38616781.0, + "step": 1492 + }, + { + "epoch": 0.16395783000219635, + "grad_norm": 2.5548276901245117, + "learning_rate": 8.18880351262349e-07, + "loss": 0.9244, + "mean_token_accuracy": 0.7141720652580261, + "num_tokens": 38636562.0, + "step": 1493 + }, + { + "epoch": 0.16406764770481003, + "grad_norm": 2.34382700920105, + "learning_rate": 8.194291986827661e-07, + "loss": 1.0418, + "mean_token_accuracy": 0.6933106184005737, + "num_tokens": 38662443.0, + "step": 1494 + }, + { + "epoch": 0.16417746540742367, + "grad_norm": 2.6613831520080566, + "learning_rate": 8.199780461031833e-07, + "loss": 0.9525, + "mean_token_accuracy": 0.7121133208274841, + "num_tokens": 38684754.0, + "step": 1495 + }, + { + "epoch": 0.16428728311003735, + "grad_norm": 2.1229069232940674, + "learning_rate": 8.205268935236004e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6958709359169006, + "num_tokens": 38713278.0, + "step": 1496 + }, + { + "epoch": 0.164397100812651, + "grad_norm": 2.5766751766204834, + "learning_rate": 8.210757409440175e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.6793776750564575, + "num_tokens": 38736256.0, + "step": 1497 + }, + { + "epoch": 0.16450691851526467, + "grad_norm": 2.5850510597229004, + "learning_rate": 8.216245883644346e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.6873959898948669, + "num_tokens": 38759521.0, + "step": 1498 + }, + { + "epoch": 0.1646167362178783, + "grad_norm": 2.330078363418579, + "learning_rate": 8.221734357848518e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.688636064529419, + "num_tokens": 38784606.0, + "step": 1499 + }, + { + "epoch": 0.16472655392049199, + "grad_norm": 2.311786413192749, + "learning_rate": 8.227222832052689e-07, + "loss": 0.958, + "mean_token_accuracy": 0.7176200747489929, + "num_tokens": 38807756.0, + "step": 1500 + }, + { + "epoch": 0.16483637162310563, + "grad_norm": 2.104017734527588, + "learning_rate": 8.23271130625686e-07, + "loss": 1.0435, + "mean_token_accuracy": 0.691443145275116, + "num_tokens": 38836631.0, + "step": 1501 + }, + { + "epoch": 0.1649461893257193, + "grad_norm": 2.2436516284942627, + "learning_rate": 8.238199780461032e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7073013782501221, + "num_tokens": 38861596.0, + "step": 1502 + }, + { + "epoch": 0.16505600702833298, + "grad_norm": 2.3616936206817627, + "learning_rate": 8.243688254665203e-07, + "loss": 0.8394, + "mean_token_accuracy": 0.7362334728240967, + "num_tokens": 38881962.0, + "step": 1503 + }, + { + "epoch": 0.16516582473094663, + "grad_norm": 2.464536428451538, + "learning_rate": 8.249176728869374e-07, + "loss": 0.9634, + "mean_token_accuracy": 0.7210573554039001, + "num_tokens": 38903574.0, + "step": 1504 + }, + { + "epoch": 0.1652756424335603, + "grad_norm": 2.4417505264282227, + "learning_rate": 8.254665203073546e-07, + "loss": 0.9383, + "mean_token_accuracy": 0.716595470905304, + "num_tokens": 38925908.0, + "step": 1505 + }, + { + "epoch": 0.16538546013617395, + "grad_norm": 2.441793203353882, + "learning_rate": 8.260153677277717e-07, + "loss": 0.9621, + "mean_token_accuracy": 0.7075765132904053, + "num_tokens": 38947603.0, + "step": 1506 + }, + { + "epoch": 0.16549527783878762, + "grad_norm": 2.5733158588409424, + "learning_rate": 8.265642151481888e-07, + "loss": 0.997, + "mean_token_accuracy": 0.697792649269104, + "num_tokens": 38968537.0, + "step": 1507 + }, + { + "epoch": 0.16560509554140126, + "grad_norm": 2.436251401901245, + "learning_rate": 8.27113062568606e-07, + "loss": 0.9514, + "mean_token_accuracy": 0.710814356803894, + "num_tokens": 38991023.0, + "step": 1508 + }, + { + "epoch": 0.16571491324401494, + "grad_norm": 2.5119001865386963, + "learning_rate": 8.27661909989023e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.6993156671524048, + "num_tokens": 39013395.0, + "step": 1509 + }, + { + "epoch": 0.16582473094662858, + "grad_norm": 2.422818660736084, + "learning_rate": 8.282107574094401e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6884182095527649, + "num_tokens": 39036355.0, + "step": 1510 + }, + { + "epoch": 0.16593454864924226, + "grad_norm": 2.2635154724121094, + "learning_rate": 8.287596048298572e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.6983024477958679, + "num_tokens": 39061923.0, + "step": 1511 + }, + { + "epoch": 0.16604436635185593, + "grad_norm": 2.2397544384002686, + "learning_rate": 8.293084522502743e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.690321147441864, + "num_tokens": 39086217.0, + "step": 1512 + }, + { + "epoch": 0.16615418405446958, + "grad_norm": 2.1307902336120605, + "learning_rate": 8.298572996706915e-07, + "loss": 1.0239, + "mean_token_accuracy": 0.6959240436553955, + "num_tokens": 39116163.0, + "step": 1513 + }, + { + "epoch": 0.16626400175708325, + "grad_norm": 2.2395918369293213, + "learning_rate": 8.304061470911086e-07, + "loss": 1.1036, + "mean_token_accuracy": 0.670574426651001, + "num_tokens": 39142392.0, + "step": 1514 + }, + { + "epoch": 0.1663738194596969, + "grad_norm": 2.464510679244995, + "learning_rate": 8.309549945115257e-07, + "loss": 1.0085, + "mean_token_accuracy": 0.7010825872421265, + "num_tokens": 39165332.0, + "step": 1515 + }, + { + "epoch": 0.16648363716231057, + "grad_norm": 2.0789010524749756, + "learning_rate": 8.315038419319429e-07, + "loss": 0.9282, + "mean_token_accuracy": 0.7196168303489685, + "num_tokens": 39192838.0, + "step": 1516 + }, + { + "epoch": 0.16659345486492422, + "grad_norm": 2.0312387943267822, + "learning_rate": 8.3205268935236e-07, + "loss": 1.0358, + "mean_token_accuracy": 0.695925235748291, + "num_tokens": 39223443.0, + "step": 1517 + }, + { + "epoch": 0.1667032725675379, + "grad_norm": 2.431405782699585, + "learning_rate": 8.326015367727771e-07, + "loss": 1.1575, + "mean_token_accuracy": 0.6610476970672607, + "num_tokens": 39247960.0, + "step": 1518 + }, + { + "epoch": 0.16681309027015154, + "grad_norm": 2.2900073528289795, + "learning_rate": 8.331503841931943e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6835548877716064, + "num_tokens": 39272785.0, + "step": 1519 + }, + { + "epoch": 0.1669229079727652, + "grad_norm": 2.232759475708008, + "learning_rate": 8.336992316136113e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6870182752609253, + "num_tokens": 39299545.0, + "step": 1520 + }, + { + "epoch": 0.16703272567537888, + "grad_norm": 2.5492560863494873, + "learning_rate": 8.342480790340285e-07, + "loss": 0.9575, + "mean_token_accuracy": 0.7137554883956909, + "num_tokens": 39320290.0, + "step": 1521 + }, + { + "epoch": 0.16714254337799253, + "grad_norm": 2.2440245151519775, + "learning_rate": 8.347969264544457e-07, + "loss": 1.0531, + "mean_token_accuracy": 0.6847922801971436, + "num_tokens": 39347080.0, + "step": 1522 + }, + { + "epoch": 0.1672523610806062, + "grad_norm": 2.291644334793091, + "learning_rate": 8.353457738748627e-07, + "loss": 1.0344, + "mean_token_accuracy": 0.6915826797485352, + "num_tokens": 39371661.0, + "step": 1523 + }, + { + "epoch": 0.16736217878321985, + "grad_norm": 2.2772438526153564, + "learning_rate": 8.358946212952799e-07, + "loss": 0.9961, + "mean_token_accuracy": 0.7112153768539429, + "num_tokens": 39398131.0, + "step": 1524 + }, + { + "epoch": 0.16747199648583352, + "grad_norm": 2.0849108695983887, + "learning_rate": 8.364434687156971e-07, + "loss": 1.052, + "mean_token_accuracy": 0.684678316116333, + "num_tokens": 39429616.0, + "step": 1525 + }, + { + "epoch": 0.16758181418844717, + "grad_norm": 2.2601544857025146, + "learning_rate": 8.369923161361141e-07, + "loss": 1.0129, + "mean_token_accuracy": 0.6978170871734619, + "num_tokens": 39454494.0, + "step": 1526 + }, + { + "epoch": 0.16769163189106084, + "grad_norm": 2.231799364089966, + "learning_rate": 8.375411635565313e-07, + "loss": 1.0102, + "mean_token_accuracy": 0.6963123083114624, + "num_tokens": 39481137.0, + "step": 1527 + }, + { + "epoch": 0.1678014495936745, + "grad_norm": 2.362236976623535, + "learning_rate": 8.380900109769485e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.7059369683265686, + "num_tokens": 39506494.0, + "step": 1528 + }, + { + "epoch": 0.16791126729628816, + "grad_norm": 2.1745712757110596, + "learning_rate": 8.386388583973654e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6985422968864441, + "num_tokens": 39534386.0, + "step": 1529 + }, + { + "epoch": 0.1680210849989018, + "grad_norm": 2.172804355621338, + "learning_rate": 8.391877058177826e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6803447008132935, + "num_tokens": 39564278.0, + "step": 1530 + }, + { + "epoch": 0.16813090270151548, + "grad_norm": 2.095689296722412, + "learning_rate": 8.397365532381997e-07, + "loss": 0.9157, + "mean_token_accuracy": 0.7305852174758911, + "num_tokens": 39591161.0, + "step": 1531 + }, + { + "epoch": 0.16824072040412916, + "grad_norm": 2.4600608348846436, + "learning_rate": 8.402854006586168e-07, + "loss": 1.026, + "mean_token_accuracy": 0.6901238560676575, + "num_tokens": 39613388.0, + "step": 1532 + }, + { + "epoch": 0.1683505381067428, + "grad_norm": 2.1770126819610596, + "learning_rate": 8.40834248079034e-07, + "loss": 1.0741, + "mean_token_accuracy": 0.6791402697563171, + "num_tokens": 39643143.0, + "step": 1533 + }, + { + "epoch": 0.16846035580935648, + "grad_norm": 2.423471212387085, + "learning_rate": 8.413830954994511e-07, + "loss": 1.0732, + "mean_token_accuracy": 0.6857718229293823, + "num_tokens": 39668405.0, + "step": 1534 + }, + { + "epoch": 0.16857017351197012, + "grad_norm": 2.609484910964966, + "learning_rate": 8.419319429198682e-07, + "loss": 1.0444, + "mean_token_accuracy": 0.6879439353942871, + "num_tokens": 39690756.0, + "step": 1535 + }, + { + "epoch": 0.1686799912145838, + "grad_norm": 2.3786187171936035, + "learning_rate": 8.424807903402854e-07, + "loss": 0.9726, + "mean_token_accuracy": 0.7077140212059021, + "num_tokens": 39716809.0, + "step": 1536 + }, + { + "epoch": 0.16878980891719744, + "grad_norm": 2.3172662258148193, + "learning_rate": 8.430296377607025e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6934388875961304, + "num_tokens": 39742370.0, + "step": 1537 + }, + { + "epoch": 0.16889962661981112, + "grad_norm": 2.5280985832214355, + "learning_rate": 8.435784851811196e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.6793931126594543, + "num_tokens": 39765340.0, + "step": 1538 + }, + { + "epoch": 0.16900944432242476, + "grad_norm": 2.3777310848236084, + "learning_rate": 8.441273326015367e-07, + "loss": 1.0536, + "mean_token_accuracy": 0.6947299242019653, + "num_tokens": 39790957.0, + "step": 1539 + }, + { + "epoch": 0.16911926202503844, + "grad_norm": 2.0682945251464844, + "learning_rate": 8.446761800219539e-07, + "loss": 1.1217, + "mean_token_accuracy": 0.6637828350067139, + "num_tokens": 39823136.0, + "step": 1540 + }, + { + "epoch": 0.1692290797276521, + "grad_norm": 2.2801761627197266, + "learning_rate": 8.45225027442371e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.675989031791687, + "num_tokens": 39848907.0, + "step": 1541 + }, + { + "epoch": 0.16933889743026576, + "grad_norm": 2.339498519897461, + "learning_rate": 8.457738748627881e-07, + "loss": 1.0368, + "mean_token_accuracy": 0.6921074390411377, + "num_tokens": 39872746.0, + "step": 1542 + }, + { + "epoch": 0.16944871513287943, + "grad_norm": 2.3205184936523438, + "learning_rate": 8.463227222832053e-07, + "loss": 0.9695, + "mean_token_accuracy": 0.7078742980957031, + "num_tokens": 39897660.0, + "step": 1543 + }, + { + "epoch": 0.16955853283549308, + "grad_norm": 2.270444631576538, + "learning_rate": 8.468715697036224e-07, + "loss": 1.048, + "mean_token_accuracy": 0.6893283724784851, + "num_tokens": 39922201.0, + "step": 1544 + }, + { + "epoch": 0.16966835053810675, + "grad_norm": 2.396655321121216, + "learning_rate": 8.474204171240395e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.6981555223464966, + "num_tokens": 39945199.0, + "step": 1545 + }, + { + "epoch": 0.1697781682407204, + "grad_norm": 2.2242767810821533, + "learning_rate": 8.479692645444567e-07, + "loss": 0.9632, + "mean_token_accuracy": 0.7067899703979492, + "num_tokens": 39969195.0, + "step": 1546 + }, + { + "epoch": 0.16988798594333407, + "grad_norm": 2.6230709552764893, + "learning_rate": 8.485181119648738e-07, + "loss": 1.067, + "mean_token_accuracy": 0.686831533908844, + "num_tokens": 39991536.0, + "step": 1547 + }, + { + "epoch": 0.16999780364594771, + "grad_norm": 2.2039597034454346, + "learning_rate": 8.490669593852908e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7013040781021118, + "num_tokens": 40017867.0, + "step": 1548 + }, + { + "epoch": 0.1701076213485614, + "grad_norm": 2.221957206726074, + "learning_rate": 8.496158068057079e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.7002183794975281, + "num_tokens": 40044381.0, + "step": 1549 + }, + { + "epoch": 0.17021743905117506, + "grad_norm": 2.344447612762451, + "learning_rate": 8.50164654226125e-07, + "loss": 0.9841, + "mean_token_accuracy": 0.7013204097747803, + "num_tokens": 40069064.0, + "step": 1550 + }, + { + "epoch": 0.1703272567537887, + "grad_norm": 2.539220094680786, + "learning_rate": 8.507135016465422e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.6922563910484314, + "num_tokens": 40090330.0, + "step": 1551 + }, + { + "epoch": 0.17043707445640238, + "grad_norm": 2.4123916625976562, + "learning_rate": 8.512623490669593e-07, + "loss": 1.0226, + "mean_token_accuracy": 0.7008834481239319, + "num_tokens": 40113542.0, + "step": 1552 + }, + { + "epoch": 0.17054689215901603, + "grad_norm": 2.2267038822174072, + "learning_rate": 8.518111964873764e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6824580430984497, + "num_tokens": 40140090.0, + "step": 1553 + }, + { + "epoch": 0.1706567098616297, + "grad_norm": 2.0680971145629883, + "learning_rate": 8.523600439077936e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.6952753663063049, + "num_tokens": 40168157.0, + "step": 1554 + }, + { + "epoch": 0.17076652756424335, + "grad_norm": 1.989874243736267, + "learning_rate": 8.529088913282107e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6734147071838379, + "num_tokens": 40201282.0, + "step": 1555 + }, + { + "epoch": 0.17087634526685702, + "grad_norm": 2.1260275840759277, + "learning_rate": 8.534577387486278e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6704449653625488, + "num_tokens": 40231174.0, + "step": 1556 + }, + { + "epoch": 0.17098616296947067, + "grad_norm": 2.1168601512908936, + "learning_rate": 8.54006586169045e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.6991404294967651, + "num_tokens": 40259217.0, + "step": 1557 + }, + { + "epoch": 0.17109598067208434, + "grad_norm": 2.17962646484375, + "learning_rate": 8.545554335894621e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.6696212887763977, + "num_tokens": 40288533.0, + "step": 1558 + }, + { + "epoch": 0.17120579837469801, + "grad_norm": 2.172524929046631, + "learning_rate": 8.551042810098792e-07, + "loss": 1.0698, + "mean_token_accuracy": 0.682542085647583, + "num_tokens": 40317522.0, + "step": 1559 + }, + { + "epoch": 0.17131561607731166, + "grad_norm": 2.2982428073883057, + "learning_rate": 8.556531284302964e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.6682136654853821, + "num_tokens": 40345757.0, + "step": 1560 + }, + { + "epoch": 0.17142543377992533, + "grad_norm": 2.3747920989990234, + "learning_rate": 8.562019758507134e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.7005839943885803, + "num_tokens": 40369248.0, + "step": 1561 + }, + { + "epoch": 0.17153525148253898, + "grad_norm": 2.1810553073883057, + "learning_rate": 8.567508232711306e-07, + "loss": 1.0631, + "mean_token_accuracy": 0.6847464442253113, + "num_tokens": 40398713.0, + "step": 1562 + }, + { + "epoch": 0.17164506918515265, + "grad_norm": 2.3673789501190186, + "learning_rate": 8.572996706915478e-07, + "loss": 0.963, + "mean_token_accuracy": 0.7112190127372742, + "num_tokens": 40422314.0, + "step": 1563 + }, + { + "epoch": 0.1717548868877663, + "grad_norm": 2.6405627727508545, + "learning_rate": 8.578485181119648e-07, + "loss": 0.933, + "mean_token_accuracy": 0.7146818041801453, + "num_tokens": 40440896.0, + "step": 1564 + }, + { + "epoch": 0.17186470459037997, + "grad_norm": 2.153846263885498, + "learning_rate": 8.58397365532382e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.697330892086029, + "num_tokens": 40470960.0, + "step": 1565 + }, + { + "epoch": 0.17197452229299362, + "grad_norm": 2.250992774963379, + "learning_rate": 8.589462129527992e-07, + "loss": 1.0767, + "mean_token_accuracy": 0.6751543283462524, + "num_tokens": 40499206.0, + "step": 1566 + }, + { + "epoch": 0.1720843399956073, + "grad_norm": 2.0649516582489014, + "learning_rate": 8.594950603732161e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6921418905258179, + "num_tokens": 40530965.0, + "step": 1567 + }, + { + "epoch": 0.17219415769822094, + "grad_norm": 2.4949140548706055, + "learning_rate": 8.600439077936333e-07, + "loss": 0.9567, + "mean_token_accuracy": 0.7086186408996582, + "num_tokens": 40553252.0, + "step": 1568 + }, + { + "epoch": 0.1723039754008346, + "grad_norm": 2.267899513244629, + "learning_rate": 8.605927552140505e-07, + "loss": 1.0004, + "mean_token_accuracy": 0.700863242149353, + "num_tokens": 40578849.0, + "step": 1569 + }, + { + "epoch": 0.1724137931034483, + "grad_norm": 2.522669553756714, + "learning_rate": 8.611416026344675e-07, + "loss": 0.9926, + "mean_token_accuracy": 0.7015402913093567, + "num_tokens": 40599435.0, + "step": 1570 + }, + { + "epoch": 0.17252361080606193, + "grad_norm": 2.2083702087402344, + "learning_rate": 8.616904500548847e-07, + "loss": 1.0559, + "mean_token_accuracy": 0.6866599321365356, + "num_tokens": 40627273.0, + "step": 1571 + }, + { + "epoch": 0.1726334285086756, + "grad_norm": 2.1985645294189453, + "learning_rate": 8.622392974753018e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7042833566665649, + "num_tokens": 40653402.0, + "step": 1572 + }, + { + "epoch": 0.17274324621128925, + "grad_norm": 2.1830172538757324, + "learning_rate": 8.627881448957189e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6731405258178711, + "num_tokens": 40684858.0, + "step": 1573 + }, + { + "epoch": 0.17285306391390293, + "grad_norm": 2.4401824474334717, + "learning_rate": 8.633369923161361e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.6953051090240479, + "num_tokens": 40706654.0, + "step": 1574 + }, + { + "epoch": 0.17296288161651657, + "grad_norm": 2.0485241413116455, + "learning_rate": 8.638858397365532e-07, + "loss": 0.9887, + "mean_token_accuracy": 0.7032544016838074, + "num_tokens": 40736248.0, + "step": 1575 + }, + { + "epoch": 0.17307269931913025, + "grad_norm": 2.4683785438537598, + "learning_rate": 8.644346871569703e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6905486583709717, + "num_tokens": 40759470.0, + "step": 1576 + }, + { + "epoch": 0.1731825170217439, + "grad_norm": 2.3072774410247803, + "learning_rate": 8.649835345773875e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.6944411993026733, + "num_tokens": 40783533.0, + "step": 1577 + }, + { + "epoch": 0.17329233472435757, + "grad_norm": 2.3341569900512695, + "learning_rate": 8.655323819978046e-07, + "loss": 1.013, + "mean_token_accuracy": 0.7032631635665894, + "num_tokens": 40811096.0, + "step": 1578 + }, + { + "epoch": 0.17340215242697124, + "grad_norm": 2.262240171432495, + "learning_rate": 8.660812294182217e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6864808201789856, + "num_tokens": 40838442.0, + "step": 1579 + }, + { + "epoch": 0.17351197012958489, + "grad_norm": 2.431162118911743, + "learning_rate": 8.666300768386389e-07, + "loss": 0.9361, + "mean_token_accuracy": 0.7164356708526611, + "num_tokens": 40861337.0, + "step": 1580 + }, + { + "epoch": 0.17362178783219856, + "grad_norm": 2.353764295578003, + "learning_rate": 8.67178924259056e-07, + "loss": 1.0591, + "mean_token_accuracy": 0.6838579177856445, + "num_tokens": 40885611.0, + "step": 1581 + }, + { + "epoch": 0.1737316055348122, + "grad_norm": 2.373192548751831, + "learning_rate": 8.677277716794731e-07, + "loss": 1.0248, + "mean_token_accuracy": 0.6913338899612427, + "num_tokens": 40909924.0, + "step": 1582 + }, + { + "epoch": 0.17384142323742588, + "grad_norm": 2.147995948791504, + "learning_rate": 8.682766190998902e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.6743961572647095, + "num_tokens": 40942186.0, + "step": 1583 + }, + { + "epoch": 0.17395124094003953, + "grad_norm": 2.0222315788269043, + "learning_rate": 8.688254665203073e-07, + "loss": 1.0534, + "mean_token_accuracy": 0.6867499351501465, + "num_tokens": 40975587.0, + "step": 1584 + }, + { + "epoch": 0.1740610586426532, + "grad_norm": 2.640129804611206, + "learning_rate": 8.693743139407245e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6652727127075195, + "num_tokens": 41003389.0, + "step": 1585 + }, + { + "epoch": 0.17417087634526685, + "grad_norm": 2.3844146728515625, + "learning_rate": 8.699231613611415e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.6974636316299438, + "num_tokens": 41026891.0, + "step": 1586 + }, + { + "epoch": 0.17428069404788052, + "grad_norm": 2.241412401199341, + "learning_rate": 8.704720087815586e-07, + "loss": 1.0664, + "mean_token_accuracy": 0.6915802955627441, + "num_tokens": 41053374.0, + "step": 1587 + }, + { + "epoch": 0.1743905117504942, + "grad_norm": 2.4400312900543213, + "learning_rate": 8.710208562019758e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.7007721662521362, + "num_tokens": 41077572.0, + "step": 1588 + }, + { + "epoch": 0.17450032945310784, + "grad_norm": 2.1929826736450195, + "learning_rate": 8.715697036223929e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6931883096694946, + "num_tokens": 41104530.0, + "step": 1589 + }, + { + "epoch": 0.1746101471557215, + "grad_norm": 2.2813334465026855, + "learning_rate": 8.7211855104281e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.7007507681846619, + "num_tokens": 41129303.0, + "step": 1590 + }, + { + "epoch": 0.17471996485833516, + "grad_norm": 2.3584415912628174, + "learning_rate": 8.726673984632272e-07, + "loss": 0.9912, + "mean_token_accuracy": 0.7027778029441833, + "num_tokens": 41154499.0, + "step": 1591 + }, + { + "epoch": 0.17482978256094883, + "grad_norm": 2.1752583980560303, + "learning_rate": 8.732162458836443e-07, + "loss": 1.1025, + "mean_token_accuracy": 0.6768195629119873, + "num_tokens": 41182403.0, + "step": 1592 + }, + { + "epoch": 0.17493960026356248, + "grad_norm": 2.24495792388916, + "learning_rate": 8.737650933040614e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.6943292021751404, + "num_tokens": 41211210.0, + "step": 1593 + }, + { + "epoch": 0.17504941796617615, + "grad_norm": 2.404047727584839, + "learning_rate": 8.743139407244785e-07, + "loss": 0.9006, + "mean_token_accuracy": 0.7212321758270264, + "num_tokens": 41232529.0, + "step": 1594 + }, + { + "epoch": 0.1751592356687898, + "grad_norm": 2.321479082107544, + "learning_rate": 8.748627881448957e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6901384592056274, + "num_tokens": 41257886.0, + "step": 1595 + }, + { + "epoch": 0.17526905337140347, + "grad_norm": 2.5609865188598633, + "learning_rate": 8.754116355653128e-07, + "loss": 1.0423, + "mean_token_accuracy": 0.6838935613632202, + "num_tokens": 41279558.0, + "step": 1596 + }, + { + "epoch": 0.17537887107401715, + "grad_norm": 2.4093997478485107, + "learning_rate": 8.759604829857299e-07, + "loss": 0.9542, + "mean_token_accuracy": 0.7095968723297119, + "num_tokens": 41303151.0, + "step": 1597 + }, + { + "epoch": 0.1754886887766308, + "grad_norm": 2.300949811935425, + "learning_rate": 8.765093304061471e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.7001299262046814, + "num_tokens": 41328539.0, + "step": 1598 + }, + { + "epoch": 0.17559850647924446, + "grad_norm": 2.0693421363830566, + "learning_rate": 8.770581778265642e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6874765157699585, + "num_tokens": 41357563.0, + "step": 1599 + }, + { + "epoch": 0.1757083241818581, + "grad_norm": 1.9734669923782349, + "learning_rate": 8.776070252469813e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.6995121240615845, + "num_tokens": 41386886.0, + "step": 1600 + }, + { + "epoch": 0.17581814188447178, + "grad_norm": 2.295571804046631, + "learning_rate": 8.781558726673985e-07, + "loss": 1.0234, + "mean_token_accuracy": 0.6959349513053894, + "num_tokens": 41411212.0, + "step": 1601 + }, + { + "epoch": 0.17592795958708543, + "grad_norm": 2.672822952270508, + "learning_rate": 8.787047200878156e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.7083730101585388, + "num_tokens": 41430136.0, + "step": 1602 + }, + { + "epoch": 0.1760377772896991, + "grad_norm": 2.152629852294922, + "learning_rate": 8.792535675082327e-07, + "loss": 1.1425, + "mean_token_accuracy": 0.6655571460723877, + "num_tokens": 41457939.0, + "step": 1603 + }, + { + "epoch": 0.17614759499231275, + "grad_norm": 2.268602132797241, + "learning_rate": 8.798024149286499e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.6944754123687744, + "num_tokens": 41483741.0, + "step": 1604 + }, + { + "epoch": 0.17625741269492642, + "grad_norm": 2.4354028701782227, + "learning_rate": 8.803512623490668e-07, + "loss": 0.9312, + "mean_token_accuracy": 0.7176679968833923, + "num_tokens": 41506679.0, + "step": 1605 + }, + { + "epoch": 0.17636723039754007, + "grad_norm": 2.076373815536499, + "learning_rate": 8.80900109769484e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6914928555488586, + "num_tokens": 41537142.0, + "step": 1606 + }, + { + "epoch": 0.17647704810015374, + "grad_norm": 2.5832793712615967, + "learning_rate": 8.814489571899012e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.7004791498184204, + "num_tokens": 41558894.0, + "step": 1607 + }, + { + "epoch": 0.17658686580276742, + "grad_norm": 2.606529951095581, + "learning_rate": 8.819978046103182e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6974630355834961, + "num_tokens": 41580474.0, + "step": 1608 + }, + { + "epoch": 0.17669668350538106, + "grad_norm": 2.2858381271362305, + "learning_rate": 8.825466520307354e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.6840444803237915, + "num_tokens": 41606527.0, + "step": 1609 + }, + { + "epoch": 0.17680650120799474, + "grad_norm": 2.6888785362243652, + "learning_rate": 8.830954994511526e-07, + "loss": 1.0058, + "mean_token_accuracy": 0.6955246329307556, + "num_tokens": 41625635.0, + "step": 1610 + }, + { + "epoch": 0.17691631891060838, + "grad_norm": 2.1631674766540527, + "learning_rate": 8.836443468715696e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6820993423461914, + "num_tokens": 41654143.0, + "step": 1611 + }, + { + "epoch": 0.17702613661322206, + "grad_norm": 2.0708160400390625, + "learning_rate": 8.841931942919868e-07, + "loss": 0.9746, + "mean_token_accuracy": 0.7034111022949219, + "num_tokens": 41682477.0, + "step": 1612 + }, + { + "epoch": 0.1771359543158357, + "grad_norm": 2.4517409801483154, + "learning_rate": 8.84742041712404e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7046013474464417, + "num_tokens": 41704881.0, + "step": 1613 + }, + { + "epoch": 0.17724577201844938, + "grad_norm": 2.3604037761688232, + "learning_rate": 8.85290889132821e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6837764382362366, + "num_tokens": 41730998.0, + "step": 1614 + }, + { + "epoch": 0.17735558972106302, + "grad_norm": 2.4384026527404785, + "learning_rate": 8.858397365532382e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.6974726319313049, + "num_tokens": 41753290.0, + "step": 1615 + }, + { + "epoch": 0.1774654074236767, + "grad_norm": 2.3707449436187744, + "learning_rate": 8.863885839736552e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6798797249794006, + "num_tokens": 41778574.0, + "step": 1616 + }, + { + "epoch": 0.17757522512629037, + "grad_norm": 2.3280980587005615, + "learning_rate": 8.869374313940724e-07, + "loss": 0.9283, + "mean_token_accuracy": 0.7186129093170166, + "num_tokens": 41800739.0, + "step": 1617 + }, + { + "epoch": 0.17768504282890402, + "grad_norm": 2.9155640602111816, + "learning_rate": 8.874862788144896e-07, + "loss": 0.8988, + "mean_token_accuracy": 0.7227168083190918, + "num_tokens": 41817764.0, + "step": 1618 + }, + { + "epoch": 0.1777948605315177, + "grad_norm": 2.394857883453369, + "learning_rate": 8.880351262349066e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6906537413597107, + "num_tokens": 41841811.0, + "step": 1619 + }, + { + "epoch": 0.17790467823413134, + "grad_norm": 2.4226789474487305, + "learning_rate": 8.885839736553238e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.7007530927658081, + "num_tokens": 41865003.0, + "step": 1620 + }, + { + "epoch": 0.178014495936745, + "grad_norm": 2.213646411895752, + "learning_rate": 8.89132821075741e-07, + "loss": 1.0575, + "mean_token_accuracy": 0.6843706369400024, + "num_tokens": 41893668.0, + "step": 1621 + }, + { + "epoch": 0.17812431363935866, + "grad_norm": 2.2760632038116455, + "learning_rate": 8.89681668496158e-07, + "loss": 1.0791, + "mean_token_accuracy": 0.6775726079940796, + "num_tokens": 41921354.0, + "step": 1622 + }, + { + "epoch": 0.17823413134197233, + "grad_norm": 2.5556271076202393, + "learning_rate": 8.902305159165752e-07, + "loss": 0.9997, + "mean_token_accuracy": 0.6955580711364746, + "num_tokens": 41941405.0, + "step": 1623 + }, + { + "epoch": 0.17834394904458598, + "grad_norm": 2.1993870735168457, + "learning_rate": 8.907793633369924e-07, + "loss": 1.0688, + "mean_token_accuracy": 0.6879730820655823, + "num_tokens": 41969650.0, + "step": 1624 + }, + { + "epoch": 0.17845376674719965, + "grad_norm": 2.1343140602111816, + "learning_rate": 8.913282107574093e-07, + "loss": 1.0216, + "mean_token_accuracy": 0.695074737071991, + "num_tokens": 41999238.0, + "step": 1625 + }, + { + "epoch": 0.17856358444981332, + "grad_norm": 2.75199556350708, + "learning_rate": 8.918770581778265e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.6961889863014221, + "num_tokens": 42018725.0, + "step": 1626 + }, + { + "epoch": 0.17867340215242697, + "grad_norm": 2.4616200923919678, + "learning_rate": 8.924259055982436e-07, + "loss": 1.0818, + "mean_token_accuracy": 0.6823850274085999, + "num_tokens": 42041856.0, + "step": 1627 + }, + { + "epoch": 0.17878321985504064, + "grad_norm": 2.6468491554260254, + "learning_rate": 8.929747530186607e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.708707869052887, + "num_tokens": 42061061.0, + "step": 1628 + }, + { + "epoch": 0.1788930375576543, + "grad_norm": 2.2109882831573486, + "learning_rate": 8.935236004390779e-07, + "loss": 1.0689, + "mean_token_accuracy": 0.6818382143974304, + "num_tokens": 42086573.0, + "step": 1629 + }, + { + "epoch": 0.17900285526026796, + "grad_norm": 2.215881109237671, + "learning_rate": 8.94072447859495e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6929303407669067, + "num_tokens": 42115567.0, + "step": 1630 + }, + { + "epoch": 0.1791126729628816, + "grad_norm": 2.295036792755127, + "learning_rate": 8.946212952799121e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.7002360820770264, + "num_tokens": 42141463.0, + "step": 1631 + }, + { + "epoch": 0.17922249066549528, + "grad_norm": 2.583893299102783, + "learning_rate": 8.951701427003293e-07, + "loss": 1.0144, + "mean_token_accuracy": 0.6923787593841553, + "num_tokens": 42161664.0, + "step": 1632 + }, + { + "epoch": 0.17933230836810893, + "grad_norm": 2.09834623336792, + "learning_rate": 8.957189901207464e-07, + "loss": 0.9989, + "mean_token_accuracy": 0.7008128762245178, + "num_tokens": 42189537.0, + "step": 1633 + }, + { + "epoch": 0.1794421260707226, + "grad_norm": 2.2944610118865967, + "learning_rate": 8.962678375411635e-07, + "loss": 1.0585, + "mean_token_accuracy": 0.6801490783691406, + "num_tokens": 42214687.0, + "step": 1634 + }, + { + "epoch": 0.17955194377333628, + "grad_norm": 2.3482742309570312, + "learning_rate": 8.968166849615807e-07, + "loss": 0.9483, + "mean_token_accuracy": 0.7144434452056885, + "num_tokens": 42237770.0, + "step": 1635 + }, + { + "epoch": 0.17966176147594992, + "grad_norm": 2.292296886444092, + "learning_rate": 8.973655323819978e-07, + "loss": 1.003, + "mean_token_accuracy": 0.7025072574615479, + "num_tokens": 42263695.0, + "step": 1636 + }, + { + "epoch": 0.1797715791785636, + "grad_norm": 2.3163857460021973, + "learning_rate": 8.979143798024149e-07, + "loss": 0.9287, + "mean_token_accuracy": 0.7180229425430298, + "num_tokens": 42287352.0, + "step": 1637 + }, + { + "epoch": 0.17988139688117724, + "grad_norm": 2.3354499340057373, + "learning_rate": 8.98463227222832e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7004472017288208, + "num_tokens": 42310732.0, + "step": 1638 + }, + { + "epoch": 0.17999121458379091, + "grad_norm": 2.267552614212036, + "learning_rate": 8.990120746432492e-07, + "loss": 1.036, + "mean_token_accuracy": 0.6965023875236511, + "num_tokens": 42336219.0, + "step": 1639 + }, + { + "epoch": 0.18010103228640456, + "grad_norm": 2.335705280303955, + "learning_rate": 8.995609220636663e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7180657386779785, + "num_tokens": 42361767.0, + "step": 1640 + }, + { + "epoch": 0.18021084998901823, + "grad_norm": 2.2318480014801025, + "learning_rate": 9.001097694840834e-07, + "loss": 1.055, + "mean_token_accuracy": 0.6861650943756104, + "num_tokens": 42391792.0, + "step": 1641 + }, + { + "epoch": 0.18032066769163188, + "grad_norm": 2.5531909465789795, + "learning_rate": 9.006586169045006e-07, + "loss": 0.8878, + "mean_token_accuracy": 0.7240060567855835, + "num_tokens": 42412205.0, + "step": 1642 + }, + { + "epoch": 0.18043048539424555, + "grad_norm": 1.9692732095718384, + "learning_rate": 9.012074643249177e-07, + "loss": 0.9689, + "mean_token_accuracy": 0.7091271281242371, + "num_tokens": 42445930.0, + "step": 1643 + }, + { + "epoch": 0.1805403030968592, + "grad_norm": 2.123889684677124, + "learning_rate": 9.017563117453347e-07, + "loss": 1.1178, + "mean_token_accuracy": 0.6694146394729614, + "num_tokens": 42476957.0, + "step": 1644 + }, + { + "epoch": 0.18065012079947287, + "grad_norm": 2.2840726375579834, + "learning_rate": 9.023051591657519e-07, + "loss": 1.0563, + "mean_token_accuracy": 0.6801490783691406, + "num_tokens": 42501717.0, + "step": 1645 + }, + { + "epoch": 0.18075993850208655, + "grad_norm": 2.3801708221435547, + "learning_rate": 9.02854006586169e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6923345327377319, + "num_tokens": 42528239.0, + "step": 1646 + }, + { + "epoch": 0.1808697562047002, + "grad_norm": 2.45285964012146, + "learning_rate": 9.034028540065861e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6712566018104553, + "num_tokens": 42552150.0, + "step": 1647 + }, + { + "epoch": 0.18097957390731387, + "grad_norm": 2.405863046646118, + "learning_rate": 9.039517014270033e-07, + "loss": 1.1363, + "mean_token_accuracy": 0.6623598337173462, + "num_tokens": 42577464.0, + "step": 1648 + }, + { + "epoch": 0.1810893916099275, + "grad_norm": 2.570178508758545, + "learning_rate": 9.045005488474203e-07, + "loss": 1.0271, + "mean_token_accuracy": 0.6913578510284424, + "num_tokens": 42597170.0, + "step": 1649 + }, + { + "epoch": 0.1811992093125412, + "grad_norm": 2.0744197368621826, + "learning_rate": 9.050493962678375e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6949768662452698, + "num_tokens": 42626780.0, + "step": 1650 + }, + { + "epoch": 0.18130902701515483, + "grad_norm": 2.4769532680511475, + "learning_rate": 9.055982436882547e-07, + "loss": 1.0015, + "mean_token_accuracy": 0.6939398646354675, + "num_tokens": 42648229.0, + "step": 1651 + }, + { + "epoch": 0.1814188447177685, + "grad_norm": 2.158506155014038, + "learning_rate": 9.061470911086717e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6796219944953918, + "num_tokens": 42675280.0, + "step": 1652 + }, + { + "epoch": 0.18152866242038215, + "grad_norm": 2.2839748859405518, + "learning_rate": 9.066959385290889e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6795511841773987, + "num_tokens": 42702215.0, + "step": 1653 + }, + { + "epoch": 0.18163848012299583, + "grad_norm": 2.1363871097564697, + "learning_rate": 9.07244785949506e-07, + "loss": 1.0696, + "mean_token_accuracy": 0.6834189295768738, + "num_tokens": 42731457.0, + "step": 1654 + }, + { + "epoch": 0.1817482978256095, + "grad_norm": 2.2500903606414795, + "learning_rate": 9.077936333699231e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.7002155780792236, + "num_tokens": 42758573.0, + "step": 1655 + }, + { + "epoch": 0.18185811552822315, + "grad_norm": 2.2282233238220215, + "learning_rate": 9.083424807903403e-07, + "loss": 0.9381, + "mean_token_accuracy": 0.7164032459259033, + "num_tokens": 42784494.0, + "step": 1656 + }, + { + "epoch": 0.18196793323083682, + "grad_norm": 2.492858409881592, + "learning_rate": 9.088913282107573e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6739277839660645, + "num_tokens": 42810428.0, + "step": 1657 + }, + { + "epoch": 0.18207775093345047, + "grad_norm": 2.3273284435272217, + "learning_rate": 9.094401756311745e-07, + "loss": 1.0309, + "mean_token_accuracy": 0.6899720430374146, + "num_tokens": 42834558.0, + "step": 1658 + }, + { + "epoch": 0.18218756863606414, + "grad_norm": 2.5403525829315186, + "learning_rate": 9.099890230515917e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6722843050956726, + "num_tokens": 42857071.0, + "step": 1659 + }, + { + "epoch": 0.18229738633867779, + "grad_norm": 2.4420485496520996, + "learning_rate": 9.105378704720087e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.7082371115684509, + "num_tokens": 42879312.0, + "step": 1660 + }, + { + "epoch": 0.18240720404129146, + "grad_norm": 2.3903260231018066, + "learning_rate": 9.110867178924259e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6828740835189819, + "num_tokens": 42902449.0, + "step": 1661 + }, + { + "epoch": 0.1825170217439051, + "grad_norm": 2.006368398666382, + "learning_rate": 9.11635565312843e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.6956490874290466, + "num_tokens": 42931635.0, + "step": 1662 + }, + { + "epoch": 0.18262683944651878, + "grad_norm": 2.6727724075317383, + "learning_rate": 9.1218441273326e-07, + "loss": 0.9832, + "mean_token_accuracy": 0.7055483460426331, + "num_tokens": 42950921.0, + "step": 1663 + }, + { + "epoch": 0.18273665714913245, + "grad_norm": 2.3185923099517822, + "learning_rate": 9.127332601536772e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7017350196838379, + "num_tokens": 42975092.0, + "step": 1664 + }, + { + "epoch": 0.1828464748517461, + "grad_norm": 2.417296886444092, + "learning_rate": 9.132821075740944e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6944403648376465, + "num_tokens": 42999736.0, + "step": 1665 + }, + { + "epoch": 0.18295629255435977, + "grad_norm": 2.4759154319763184, + "learning_rate": 9.138309549945114e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.688733696937561, + "num_tokens": 43021844.0, + "step": 1666 + }, + { + "epoch": 0.18306611025697342, + "grad_norm": 2.2852671146392822, + "learning_rate": 9.143798024149286e-07, + "loss": 0.9621, + "mean_token_accuracy": 0.7089393138885498, + "num_tokens": 43047918.0, + "step": 1667 + }, + { + "epoch": 0.1831759279595871, + "grad_norm": 2.195056676864624, + "learning_rate": 9.149286498353457e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.7009262442588806, + "num_tokens": 43076624.0, + "step": 1668 + }, + { + "epoch": 0.18328574566220074, + "grad_norm": 2.4292516708374023, + "learning_rate": 9.154774972557628e-07, + "loss": 1.0202, + "mean_token_accuracy": 0.6912022233009338, + "num_tokens": 43100758.0, + "step": 1669 + }, + { + "epoch": 0.1833955633648144, + "grad_norm": 2.362017869949341, + "learning_rate": 9.1602634467618e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6891016960144043, + "num_tokens": 43126017.0, + "step": 1670 + }, + { + "epoch": 0.18350538106742806, + "grad_norm": 2.3553571701049805, + "learning_rate": 9.165751920965971e-07, + "loss": 0.969, + "mean_token_accuracy": 0.6985521912574768, + "num_tokens": 43149091.0, + "step": 1671 + }, + { + "epoch": 0.18361519877004173, + "grad_norm": 2.3447980880737305, + "learning_rate": 9.171240395170142e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.7007318735122681, + "num_tokens": 43173267.0, + "step": 1672 + }, + { + "epoch": 0.1837250164726554, + "grad_norm": 2.1811420917510986, + "learning_rate": 9.176728869374314e-07, + "loss": 1.0793, + "mean_token_accuracy": 0.6773688793182373, + "num_tokens": 43201816.0, + "step": 1673 + }, + { + "epoch": 0.18383483417526905, + "grad_norm": 2.5655133724212646, + "learning_rate": 9.182217343578485e-07, + "loss": 1.057, + "mean_token_accuracy": 0.6942319869995117, + "num_tokens": 43223134.0, + "step": 1674 + }, + { + "epoch": 0.18394465187788273, + "grad_norm": 2.2262933254241943, + "learning_rate": 9.187705817782656e-07, + "loss": 1.0138, + "mean_token_accuracy": 0.7024996876716614, + "num_tokens": 43248449.0, + "step": 1675 + }, + { + "epoch": 0.18405446958049637, + "grad_norm": 2.332780122756958, + "learning_rate": 9.193194291986828e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6830394268035889, + "num_tokens": 43272747.0, + "step": 1676 + }, + { + "epoch": 0.18416428728311005, + "grad_norm": 2.6053643226623535, + "learning_rate": 9.198682766190999e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.7002097964286804, + "num_tokens": 43294101.0, + "step": 1677 + }, + { + "epoch": 0.1842741049857237, + "grad_norm": 2.1518280506134033, + "learning_rate": 9.20417124039517e-07, + "loss": 1.1297, + "mean_token_accuracy": 0.6673071384429932, + "num_tokens": 43323123.0, + "step": 1678 + }, + { + "epoch": 0.18438392268833736, + "grad_norm": 2.290947675704956, + "learning_rate": 9.209659714599341e-07, + "loss": 1.0776, + "mean_token_accuracy": 0.6818069219589233, + "num_tokens": 43348805.0, + "step": 1679 + }, + { + "epoch": 0.184493740390951, + "grad_norm": 2.0276708602905273, + "learning_rate": 9.215148188803513e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6756800413131714, + "num_tokens": 43380694.0, + "step": 1680 + }, + { + "epoch": 0.18460355809356468, + "grad_norm": 1.8591010570526123, + "learning_rate": 9.220636663007683e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6986803412437439, + "num_tokens": 43417481.0, + "step": 1681 + }, + { + "epoch": 0.18471337579617833, + "grad_norm": 2.467413902282715, + "learning_rate": 9.226125137211854e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6810019016265869, + "num_tokens": 43441169.0, + "step": 1682 + }, + { + "epoch": 0.184823193498792, + "grad_norm": 1.9820512533187866, + "learning_rate": 9.231613611416026e-07, + "loss": 1.075, + "mean_token_accuracy": 0.6791302561759949, + "num_tokens": 43472591.0, + "step": 1683 + }, + { + "epoch": 0.18493301120140568, + "grad_norm": 2.366220712661743, + "learning_rate": 9.237102085620197e-07, + "loss": 0.9779, + "mean_token_accuracy": 0.6993354558944702, + "num_tokens": 43495780.0, + "step": 1684 + }, + { + "epoch": 0.18504282890401932, + "grad_norm": 2.3067145347595215, + "learning_rate": 9.242590559824368e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7162474989891052, + "num_tokens": 43520905.0, + "step": 1685 + }, + { + "epoch": 0.185152646606633, + "grad_norm": 2.45027232170105, + "learning_rate": 9.24807903402854e-07, + "loss": 1.008, + "mean_token_accuracy": 0.7064981460571289, + "num_tokens": 43543358.0, + "step": 1686 + }, + { + "epoch": 0.18526246430924664, + "grad_norm": 2.11891508102417, + "learning_rate": 9.253567508232711e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6899898648262024, + "num_tokens": 43572586.0, + "step": 1687 + }, + { + "epoch": 0.18537228201186032, + "grad_norm": 2.337642192840576, + "learning_rate": 9.259055982436882e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6890581250190735, + "num_tokens": 43596703.0, + "step": 1688 + }, + { + "epoch": 0.18548209971447396, + "grad_norm": 2.348149061203003, + "learning_rate": 9.264544456641053e-07, + "loss": 0.8883, + "mean_token_accuracy": 0.7310596704483032, + "num_tokens": 43619558.0, + "step": 1689 + }, + { + "epoch": 0.18559191741708764, + "grad_norm": 2.5435383319854736, + "learning_rate": 9.270032930845224e-07, + "loss": 1.0136, + "mean_token_accuracy": 0.6956474184989929, + "num_tokens": 43640862.0, + "step": 1690 + }, + { + "epoch": 0.18570173511970128, + "grad_norm": 2.2685811519622803, + "learning_rate": 9.275521405049396e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6954634189605713, + "num_tokens": 43665839.0, + "step": 1691 + }, + { + "epoch": 0.18581155282231496, + "grad_norm": 2.534672975540161, + "learning_rate": 9.281009879253567e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.6925288438796997, + "num_tokens": 43688708.0, + "step": 1692 + }, + { + "epoch": 0.18592137052492863, + "grad_norm": 2.0161070823669434, + "learning_rate": 9.286498353457738e-07, + "loss": 0.9661, + "mean_token_accuracy": 0.7093178629875183, + "num_tokens": 43719290.0, + "step": 1693 + }, + { + "epoch": 0.18603118822754228, + "grad_norm": 2.0872106552124023, + "learning_rate": 9.29198682766191e-07, + "loss": 1.0173, + "mean_token_accuracy": 0.6938872337341309, + "num_tokens": 43748108.0, + "step": 1694 + }, + { + "epoch": 0.18614100593015595, + "grad_norm": 2.1925227642059326, + "learning_rate": 9.297475301866081e-07, + "loss": 1.046, + "mean_token_accuracy": 0.6857989430427551, + "num_tokens": 43775832.0, + "step": 1695 + }, + { + "epoch": 0.1862508236327696, + "grad_norm": 2.5378000736236572, + "learning_rate": 9.302963776070252e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6881611943244934, + "num_tokens": 43797909.0, + "step": 1696 + }, + { + "epoch": 0.18636064133538327, + "grad_norm": 2.482046365737915, + "learning_rate": 9.308452250274424e-07, + "loss": 1.0437, + "mean_token_accuracy": 0.6930134892463684, + "num_tokens": 43820057.0, + "step": 1697 + }, + { + "epoch": 0.18647045903799692, + "grad_norm": 2.38754940032959, + "learning_rate": 9.313940724478595e-07, + "loss": 1.0791, + "mean_token_accuracy": 0.6756812930107117, + "num_tokens": 43843738.0, + "step": 1698 + }, + { + "epoch": 0.1865802767406106, + "grad_norm": 2.504265546798706, + "learning_rate": 9.319429198682766e-07, + "loss": 0.9831, + "mean_token_accuracy": 0.7029518485069275, + "num_tokens": 43864446.0, + "step": 1699 + }, + { + "epoch": 0.18669009444322424, + "grad_norm": 2.118082284927368, + "learning_rate": 9.324917672886937e-07, + "loss": 1.0887, + "mean_token_accuracy": 0.6703853011131287, + "num_tokens": 43892981.0, + "step": 1700 + }, + { + "epoch": 0.1867999121458379, + "grad_norm": 2.283677577972412, + "learning_rate": 9.330406147091107e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.6827846765518188, + "num_tokens": 43919307.0, + "step": 1701 + }, + { + "epoch": 0.18690972984845158, + "grad_norm": 2.355404853820801, + "learning_rate": 9.335894621295279e-07, + "loss": 1.048, + "mean_token_accuracy": 0.6984516382217407, + "num_tokens": 43944074.0, + "step": 1702 + }, + { + "epoch": 0.18701954755106523, + "grad_norm": 2.1634397506713867, + "learning_rate": 9.341383095499451e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.6960771083831787, + "num_tokens": 43971762.0, + "step": 1703 + }, + { + "epoch": 0.1871293652536789, + "grad_norm": 2.4491539001464844, + "learning_rate": 9.346871569703621e-07, + "loss": 0.9767, + "mean_token_accuracy": 0.7021894454956055, + "num_tokens": 43993724.0, + "step": 1704 + }, + { + "epoch": 0.18723918295629255, + "grad_norm": 2.3366928100585938, + "learning_rate": 9.352360043907793e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6806435585021973, + "num_tokens": 44017263.0, + "step": 1705 + }, + { + "epoch": 0.18734900065890622, + "grad_norm": 2.4394476413726807, + "learning_rate": 9.357848518111965e-07, + "loss": 0.938, + "mean_token_accuracy": 0.7103432416915894, + "num_tokens": 44042631.0, + "step": 1706 + }, + { + "epoch": 0.18745881836151987, + "grad_norm": 2.3211448192596436, + "learning_rate": 9.363336992316135e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6812485456466675, + "num_tokens": 44074232.0, + "step": 1707 + }, + { + "epoch": 0.18756863606413354, + "grad_norm": 2.0248849391937256, + "learning_rate": 9.368825466520307e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7108293771743774, + "num_tokens": 44105160.0, + "step": 1708 + }, + { + "epoch": 0.1876784537667472, + "grad_norm": 2.174064874649048, + "learning_rate": 9.374313940724479e-07, + "loss": 1.0696, + "mean_token_accuracy": 0.6813386678695679, + "num_tokens": 44132684.0, + "step": 1709 + }, + { + "epoch": 0.18778827146936086, + "grad_norm": 2.9185993671417236, + "learning_rate": 9.379802414928649e-07, + "loss": 0.9789, + "mean_token_accuracy": 0.7028263807296753, + "num_tokens": 44151249.0, + "step": 1710 + }, + { + "epoch": 0.18789808917197454, + "grad_norm": 2.280973434448242, + "learning_rate": 9.385290889132821e-07, + "loss": 1.1098, + "mean_token_accuracy": 0.6681134700775146, + "num_tokens": 44181371.0, + "step": 1711 + }, + { + "epoch": 0.18800790687458818, + "grad_norm": 2.275254249572754, + "learning_rate": 9.390779363336992e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.6844308972358704, + "num_tokens": 44207424.0, + "step": 1712 + }, + { + "epoch": 0.18811772457720186, + "grad_norm": 2.2836809158325195, + "learning_rate": 9.396267837541163e-07, + "loss": 1.044, + "mean_token_accuracy": 0.68790602684021, + "num_tokens": 44233372.0, + "step": 1713 + }, + { + "epoch": 0.1882275422798155, + "grad_norm": 2.0311994552612305, + "learning_rate": 9.401756311745335e-07, + "loss": 1.0465, + "mean_token_accuracy": 0.694664478302002, + "num_tokens": 44267175.0, + "step": 1714 + }, + { + "epoch": 0.18833735998242918, + "grad_norm": 2.207519054412842, + "learning_rate": 9.407244785949506e-07, + "loss": 1.0192, + "mean_token_accuracy": 0.6961104869842529, + "num_tokens": 44293402.0, + "step": 1715 + }, + { + "epoch": 0.18844717768504282, + "grad_norm": 2.176581859588623, + "learning_rate": 9.412733260153677e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.6819238662719727, + "num_tokens": 44321650.0, + "step": 1716 + }, + { + "epoch": 0.1885569953876565, + "grad_norm": 2.2517409324645996, + "learning_rate": 9.418221734357849e-07, + "loss": 0.9414, + "mean_token_accuracy": 0.7178584337234497, + "num_tokens": 44346268.0, + "step": 1717 + }, + { + "epoch": 0.18866681309027014, + "grad_norm": 2.0393788814544678, + "learning_rate": 9.42371020856202e-07, + "loss": 1.0098, + "mean_token_accuracy": 0.6998518109321594, + "num_tokens": 44375795.0, + "step": 1718 + }, + { + "epoch": 0.18877663079288381, + "grad_norm": 2.5643997192382812, + "learning_rate": 9.42919868276619e-07, + "loss": 1.0165, + "mean_token_accuracy": 0.6892954111099243, + "num_tokens": 44395320.0, + "step": 1719 + }, + { + "epoch": 0.18888644849549746, + "grad_norm": 2.0478198528289795, + "learning_rate": 9.434687156970362e-07, + "loss": 0.9717, + "mean_token_accuracy": 0.6932981014251709, + "num_tokens": 44423903.0, + "step": 1720 + }, + { + "epoch": 0.18899626619811113, + "grad_norm": 2.170677661895752, + "learning_rate": 9.440175631174532e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.6984411478042603, + "num_tokens": 44451379.0, + "step": 1721 + }, + { + "epoch": 0.1891060839007248, + "grad_norm": 2.1957404613494873, + "learning_rate": 9.445664105378704e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6805759072303772, + "num_tokens": 44479035.0, + "step": 1722 + }, + { + "epoch": 0.18921590160333845, + "grad_norm": 2.079127311706543, + "learning_rate": 9.451152579582875e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6955150365829468, + "num_tokens": 44508591.0, + "step": 1723 + }, + { + "epoch": 0.18932571930595213, + "grad_norm": 2.127851724624634, + "learning_rate": 9.456641053787046e-07, + "loss": 0.9305, + "mean_token_accuracy": 0.707218587398529, + "num_tokens": 44534501.0, + "step": 1724 + }, + { + "epoch": 0.18943553700856577, + "grad_norm": 2.3727595806121826, + "learning_rate": 9.462129527991218e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.688130259513855, + "num_tokens": 44558249.0, + "step": 1725 + }, + { + "epoch": 0.18954535471117945, + "grad_norm": 2.139714002609253, + "learning_rate": 9.467618002195389e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6881107687950134, + "num_tokens": 44587953.0, + "step": 1726 + }, + { + "epoch": 0.1896551724137931, + "grad_norm": 2.3216028213500977, + "learning_rate": 9.47310647639956e-07, + "loss": 0.9399, + "mean_token_accuracy": 0.7134766578674316, + "num_tokens": 44613917.0, + "step": 1727 + }, + { + "epoch": 0.18976499011640677, + "grad_norm": 2.2077951431274414, + "learning_rate": 9.478594950603732e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6834618449211121, + "num_tokens": 44640335.0, + "step": 1728 + }, + { + "epoch": 0.1898748078190204, + "grad_norm": 2.4735724925994873, + "learning_rate": 9.484083424807903e-07, + "loss": 0.984, + "mean_token_accuracy": 0.7004914283752441, + "num_tokens": 44662548.0, + "step": 1729 + }, + { + "epoch": 0.1899846255216341, + "grad_norm": 2.0452866554260254, + "learning_rate": 9.489571899012074e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6886743307113647, + "num_tokens": 44693778.0, + "step": 1730 + }, + { + "epoch": 0.19009444322424776, + "grad_norm": 2.010411024093628, + "learning_rate": 9.495060373216246e-07, + "loss": 0.9715, + "mean_token_accuracy": 0.7037961483001709, + "num_tokens": 44724145.0, + "step": 1731 + }, + { + "epoch": 0.1902042609268614, + "grad_norm": 2.230358123779297, + "learning_rate": 9.500548847420417e-07, + "loss": 0.9935, + "mean_token_accuracy": 0.6990740299224854, + "num_tokens": 44748842.0, + "step": 1732 + }, + { + "epoch": 0.19031407862947508, + "grad_norm": 2.400355577468872, + "learning_rate": 9.506037321624588e-07, + "loss": 0.9483, + "mean_token_accuracy": 0.7084625959396362, + "num_tokens": 44771042.0, + "step": 1733 + }, + { + "epoch": 0.19042389633208873, + "grad_norm": 2.475569486618042, + "learning_rate": 9.511525795828759e-07, + "loss": 1.0368, + "mean_token_accuracy": 0.6903674006462097, + "num_tokens": 44794172.0, + "step": 1734 + }, + { + "epoch": 0.1905337140347024, + "grad_norm": 2.0218825340270996, + "learning_rate": 9.517014270032931e-07, + "loss": 1.1026, + "mean_token_accuracy": 0.6802015900611877, + "num_tokens": 44825999.0, + "step": 1735 + }, + { + "epoch": 0.19064353173731605, + "grad_norm": 2.5253031253814697, + "learning_rate": 9.522502744237102e-07, + "loss": 0.9514, + "mean_token_accuracy": 0.7125065922737122, + "num_tokens": 44847723.0, + "step": 1736 + }, + { + "epoch": 0.19075334943992972, + "grad_norm": 2.034496545791626, + "learning_rate": 9.527991218441273e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.680656909942627, + "num_tokens": 44881631.0, + "step": 1737 + }, + { + "epoch": 0.19086316714254337, + "grad_norm": 2.0246288776397705, + "learning_rate": 9.533479692645444e-07, + "loss": 1.0871, + "mean_token_accuracy": 0.6746364831924438, + "num_tokens": 44914795.0, + "step": 1738 + }, + { + "epoch": 0.19097298484515704, + "grad_norm": 2.431464910507202, + "learning_rate": 9.538968166849616e-07, + "loss": 0.9267, + "mean_token_accuracy": 0.7161364555358887, + "num_tokens": 44938257.0, + "step": 1739 + }, + { + "epoch": 0.1910828025477707, + "grad_norm": 2.5676748752593994, + "learning_rate": 9.544456641053787e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7112070322036743, + "num_tokens": 44957640.0, + "step": 1740 + }, + { + "epoch": 0.19119262025038436, + "grad_norm": 2.2435548305511475, + "learning_rate": 9.549945115257958e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6753570437431335, + "num_tokens": 44986251.0, + "step": 1741 + }, + { + "epoch": 0.19130243795299803, + "grad_norm": 2.2464237213134766, + "learning_rate": 9.55543358946213e-07, + "loss": 0.9794, + "mean_token_accuracy": 0.7006217837333679, + "num_tokens": 45012017.0, + "step": 1742 + }, + { + "epoch": 0.19141225565561168, + "grad_norm": 2.008702516555786, + "learning_rate": 9.5609220636663e-07, + "loss": 1.0368, + "mean_token_accuracy": 0.6839909553527832, + "num_tokens": 45044052.0, + "step": 1743 + }, + { + "epoch": 0.19152207335822535, + "grad_norm": 2.3187365531921387, + "learning_rate": 9.566410537870472e-07, + "loss": 1.0812, + "mean_token_accuracy": 0.674940824508667, + "num_tokens": 45069391.0, + "step": 1744 + }, + { + "epoch": 0.191631891060839, + "grad_norm": 2.1022303104400635, + "learning_rate": 9.571899012074642e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.6735266447067261, + "num_tokens": 45100478.0, + "step": 1745 + }, + { + "epoch": 0.19174170876345267, + "grad_norm": 2.297710418701172, + "learning_rate": 9.577387486278815e-07, + "loss": 1.0118, + "mean_token_accuracy": 0.6941505074501038, + "num_tokens": 45124742.0, + "step": 1746 + }, + { + "epoch": 0.19185152646606632, + "grad_norm": 2.3892836570739746, + "learning_rate": 9.582875960482986e-07, + "loss": 0.9107, + "mean_token_accuracy": 0.7181613445281982, + "num_tokens": 45147352.0, + "step": 1747 + }, + { + "epoch": 0.19196134416868, + "grad_norm": 2.1621382236480713, + "learning_rate": 9.588364434687156e-07, + "loss": 1.098, + "mean_token_accuracy": 0.6811203956604004, + "num_tokens": 45175766.0, + "step": 1748 + }, + { + "epoch": 0.19207116187129367, + "grad_norm": 2.5375049114227295, + "learning_rate": 9.593852908891327e-07, + "loss": 0.9055, + "mean_token_accuracy": 0.717212438583374, + "num_tokens": 45196045.0, + "step": 1749 + }, + { + "epoch": 0.1921809795739073, + "grad_norm": 2.2437632083892822, + "learning_rate": 9.5993413830955e-07, + "loss": 0.9872, + "mean_token_accuracy": 0.6990329027175903, + "num_tokens": 45219953.0, + "step": 1750 + }, + { + "epoch": 0.19229079727652099, + "grad_norm": 2.4227652549743652, + "learning_rate": 9.60482985729967e-07, + "loss": 1.0574, + "mean_token_accuracy": 0.6859943866729736, + "num_tokens": 45242438.0, + "step": 1751 + }, + { + "epoch": 0.19240061497913463, + "grad_norm": 2.247495412826538, + "learning_rate": 9.61031833150384e-07, + "loss": 1.0037, + "mean_token_accuracy": 0.6967213749885559, + "num_tokens": 45268879.0, + "step": 1752 + }, + { + "epoch": 0.1925104326817483, + "grad_norm": 2.5568699836730957, + "learning_rate": 9.615806805708014e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6942822337150574, + "num_tokens": 45291673.0, + "step": 1753 + }, + { + "epoch": 0.19262025038436195, + "grad_norm": 2.066736936569214, + "learning_rate": 9.621295279912184e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6773593425750732, + "num_tokens": 45320711.0, + "step": 1754 + }, + { + "epoch": 0.19273006808697563, + "grad_norm": 2.302624225616455, + "learning_rate": 9.626783754116355e-07, + "loss": 0.998, + "mean_token_accuracy": 0.7040387392044067, + "num_tokens": 45350609.0, + "step": 1755 + }, + { + "epoch": 0.19283988578958927, + "grad_norm": 2.6097300052642822, + "learning_rate": 9.632272228320525e-07, + "loss": 0.9773, + "mean_token_accuracy": 0.7027102708816528, + "num_tokens": 45369759.0, + "step": 1756 + }, + { + "epoch": 0.19294970349220295, + "grad_norm": 2.1066296100616455, + "learning_rate": 9.637760702524698e-07, + "loss": 1.062, + "mean_token_accuracy": 0.6823486089706421, + "num_tokens": 45398846.0, + "step": 1757 + }, + { + "epoch": 0.1930595211948166, + "grad_norm": 2.4008007049560547, + "learning_rate": 9.643249176728869e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6890245676040649, + "num_tokens": 45424255.0, + "step": 1758 + }, + { + "epoch": 0.19316933889743026, + "grad_norm": 2.2622532844543457, + "learning_rate": 9.64873765093304e-07, + "loss": 1.0426, + "mean_token_accuracy": 0.6863467693328857, + "num_tokens": 45451269.0, + "step": 1759 + }, + { + "epoch": 0.19327915660004394, + "grad_norm": 2.1861891746520996, + "learning_rate": 9.654226125137212e-07, + "loss": 1.04, + "mean_token_accuracy": 0.6887837648391724, + "num_tokens": 45478705.0, + "step": 1760 + }, + { + "epoch": 0.19338897430265758, + "grad_norm": 2.25431227684021, + "learning_rate": 9.659714599341383e-07, + "loss": 1.1129, + "mean_token_accuracy": 0.6796125769615173, + "num_tokens": 45507862.0, + "step": 1761 + }, + { + "epoch": 0.19349879200527126, + "grad_norm": 1.9978705644607544, + "learning_rate": 9.665203073545553e-07, + "loss": 1.1057, + "mean_token_accuracy": 0.6756068468093872, + "num_tokens": 45540085.0, + "step": 1762 + }, + { + "epoch": 0.1936086097078849, + "grad_norm": 2.4452733993530273, + "learning_rate": 9.670691547749726e-07, + "loss": 1.0589, + "mean_token_accuracy": 0.678343653678894, + "num_tokens": 45565746.0, + "step": 1763 + }, + { + "epoch": 0.19371842741049858, + "grad_norm": 2.245331287384033, + "learning_rate": 9.676180021953897e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6916415095329285, + "num_tokens": 45594424.0, + "step": 1764 + }, + { + "epoch": 0.19382824511311222, + "grad_norm": 2.304011106491089, + "learning_rate": 9.681668496158067e-07, + "loss": 0.9545, + "mean_token_accuracy": 0.7140670418739319, + "num_tokens": 45618396.0, + "step": 1765 + }, + { + "epoch": 0.1939380628157259, + "grad_norm": 2.6543028354644775, + "learning_rate": 9.68715697036224e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6919076442718506, + "num_tokens": 45638149.0, + "step": 1766 + }, + { + "epoch": 0.19404788051833954, + "grad_norm": 2.701982259750366, + "learning_rate": 9.692645444566409e-07, + "loss": 1.0687, + "mean_token_accuracy": 0.6837464570999146, + "num_tokens": 45658363.0, + "step": 1767 + }, + { + "epoch": 0.19415769822095322, + "grad_norm": 2.367452383041382, + "learning_rate": 9.698133918770581e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6931403875350952, + "num_tokens": 45681689.0, + "step": 1768 + }, + { + "epoch": 0.1942675159235669, + "grad_norm": 2.362109661102295, + "learning_rate": 9.703622392974752e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6899141073226929, + "num_tokens": 45704723.0, + "step": 1769 + }, + { + "epoch": 0.19437733362618054, + "grad_norm": 2.3084001541137695, + "learning_rate": 9.709110867178923e-07, + "loss": 0.967, + "mean_token_accuracy": 0.7067958116531372, + "num_tokens": 45727988.0, + "step": 1770 + }, + { + "epoch": 0.1944871513287942, + "grad_norm": 2.432370662689209, + "learning_rate": 9.714599341383095e-07, + "loss": 1.0814, + "mean_token_accuracy": 0.6816351413726807, + "num_tokens": 45752255.0, + "step": 1771 + }, + { + "epoch": 0.19459696903140786, + "grad_norm": 1.9745756387710571, + "learning_rate": 9.720087815587266e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.697414219379425, + "num_tokens": 45784529.0, + "step": 1772 + }, + { + "epoch": 0.19470678673402153, + "grad_norm": 2.5683529376983643, + "learning_rate": 9.725576289791437e-07, + "loss": 1.0632, + "mean_token_accuracy": 0.6796122193336487, + "num_tokens": 45807117.0, + "step": 1773 + }, + { + "epoch": 0.19481660443663518, + "grad_norm": 2.215130090713501, + "learning_rate": 9.73106476399561e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6905771493911743, + "num_tokens": 45833984.0, + "step": 1774 + }, + { + "epoch": 0.19492642213924885, + "grad_norm": 2.3747763633728027, + "learning_rate": 9.73655323819978e-07, + "loss": 1.0128, + "mean_token_accuracy": 0.6919094324111938, + "num_tokens": 45860322.0, + "step": 1775 + }, + { + "epoch": 0.1950362398418625, + "grad_norm": 2.5071794986724854, + "learning_rate": 9.74204171240395e-07, + "loss": 0.9826, + "mean_token_accuracy": 0.7029351592063904, + "num_tokens": 45882133.0, + "step": 1776 + }, + { + "epoch": 0.19514605754447617, + "grad_norm": 1.957415223121643, + "learning_rate": 9.747530186608123e-07, + "loss": 1.1145, + "mean_token_accuracy": 0.6667241454124451, + "num_tokens": 45918220.0, + "step": 1777 + }, + { + "epoch": 0.19525587524708984, + "grad_norm": 2.1139943599700928, + "learning_rate": 9.753018660812294e-07, + "loss": 1.1062, + "mean_token_accuracy": 0.6678401231765747, + "num_tokens": 45947200.0, + "step": 1778 + }, + { + "epoch": 0.1953656929497035, + "grad_norm": 2.3095221519470215, + "learning_rate": 9.758507135016465e-07, + "loss": 1.0171, + "mean_token_accuracy": 0.6907578706741333, + "num_tokens": 45971848.0, + "step": 1779 + }, + { + "epoch": 0.19547551065231716, + "grad_norm": 2.240943193435669, + "learning_rate": 9.763995609220637e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6810331344604492, + "num_tokens": 45997885.0, + "step": 1780 + }, + { + "epoch": 0.1955853283549308, + "grad_norm": 2.036709785461426, + "learning_rate": 9.769484083424808e-07, + "loss": 1.0627, + "mean_token_accuracy": 0.6851458549499512, + "num_tokens": 46030775.0, + "step": 1781 + }, + { + "epoch": 0.19569514605754448, + "grad_norm": 2.3596839904785156, + "learning_rate": 9.774972557628979e-07, + "loss": 1.1422, + "mean_token_accuracy": 0.6627873778343201, + "num_tokens": 46055476.0, + "step": 1782 + }, + { + "epoch": 0.19580496376015813, + "grad_norm": 2.194838285446167, + "learning_rate": 9.780461031833151e-07, + "loss": 1.015, + "mean_token_accuracy": 0.6941444873809814, + "num_tokens": 46082816.0, + "step": 1783 + }, + { + "epoch": 0.1959147814627718, + "grad_norm": 2.6580324172973633, + "learning_rate": 9.785949506037322e-07, + "loss": 0.9878, + "mean_token_accuracy": 0.6958792805671692, + "num_tokens": 46103126.0, + "step": 1784 + }, + { + "epoch": 0.19602459916538545, + "grad_norm": 2.078887462615967, + "learning_rate": 9.791437980241493e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.7053266167640686, + "num_tokens": 46131949.0, + "step": 1785 + }, + { + "epoch": 0.19613441686799912, + "grad_norm": 2.5111281871795654, + "learning_rate": 9.796926454445663e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6883634924888611, + "num_tokens": 46154049.0, + "step": 1786 + }, + { + "epoch": 0.1962442345706128, + "grad_norm": 2.5417356491088867, + "learning_rate": 9.802414928649834e-07, + "loss": 1.0288, + "mean_token_accuracy": 0.6879568099975586, + "num_tokens": 46176461.0, + "step": 1787 + }, + { + "epoch": 0.19635405227322644, + "grad_norm": 2.233680486679077, + "learning_rate": 9.807903402854007e-07, + "loss": 0.9497, + "mean_token_accuracy": 0.7200881242752075, + "num_tokens": 46202894.0, + "step": 1788 + }, + { + "epoch": 0.19646386997584012, + "grad_norm": 2.0838088989257812, + "learning_rate": 9.813391877058177e-07, + "loss": 1.0574, + "mean_token_accuracy": 0.6852134466171265, + "num_tokens": 46234812.0, + "step": 1789 + }, + { + "epoch": 0.19657368767845376, + "grad_norm": 2.3555867671966553, + "learning_rate": 9.818880351262348e-07, + "loss": 0.9313, + "mean_token_accuracy": 0.7138793468475342, + "num_tokens": 46256960.0, + "step": 1790 + }, + { + "epoch": 0.19668350538106744, + "grad_norm": 2.3903963565826416, + "learning_rate": 9.82436882546652e-07, + "loss": 1.0724, + "mean_token_accuracy": 0.6783753633499146, + "num_tokens": 46282216.0, + "step": 1791 + }, + { + "epoch": 0.19679332308368108, + "grad_norm": 2.2125020027160645, + "learning_rate": 9.829857299670691e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6964951753616333, + "num_tokens": 46309882.0, + "step": 1792 + }, + { + "epoch": 0.19690314078629476, + "grad_norm": 2.2402265071868896, + "learning_rate": 9.835345773874862e-07, + "loss": 0.9725, + "mean_token_accuracy": 0.7047167420387268, + "num_tokens": 46335169.0, + "step": 1793 + }, + { + "epoch": 0.1970129584889084, + "grad_norm": 2.158684730529785, + "learning_rate": 9.840834248079035e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.7104772329330444, + "num_tokens": 46365528.0, + "step": 1794 + }, + { + "epoch": 0.19712277619152208, + "grad_norm": 2.535909414291382, + "learning_rate": 9.846322722283205e-07, + "loss": 0.9311, + "mean_token_accuracy": 0.7182096838951111, + "num_tokens": 46386659.0, + "step": 1795 + }, + { + "epoch": 0.19723259389413572, + "grad_norm": 2.34110689163208, + "learning_rate": 9.851811196487376e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6837792992591858, + "num_tokens": 46410731.0, + "step": 1796 + }, + { + "epoch": 0.1973424115967494, + "grad_norm": 2.5689148902893066, + "learning_rate": 9.857299670691546e-07, + "loss": 1.0372, + "mean_token_accuracy": 0.6849459409713745, + "num_tokens": 46431363.0, + "step": 1797 + }, + { + "epoch": 0.19745222929936307, + "grad_norm": 2.176400899887085, + "learning_rate": 9.86278814489572e-07, + "loss": 0.9786, + "mean_token_accuracy": 0.7044081687927246, + "num_tokens": 46457173.0, + "step": 1798 + }, + { + "epoch": 0.19756204700197671, + "grad_norm": 2.0251057147979736, + "learning_rate": 9.86827661909989e-07, + "loss": 1.0816, + "mean_token_accuracy": 0.6789841055870056, + "num_tokens": 46488237.0, + "step": 1799 + }, + { + "epoch": 0.1976718647045904, + "grad_norm": 2.2455601692199707, + "learning_rate": 9.87376509330406e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.6958430409431458, + "num_tokens": 46515745.0, + "step": 1800 + }, + { + "epoch": 0.19778168240720403, + "grad_norm": 2.405363082885742, + "learning_rate": 9.879253567508233e-07, + "loss": 1.012, + "mean_token_accuracy": 0.6954392790794373, + "num_tokens": 46539805.0, + "step": 1801 + }, + { + "epoch": 0.1978915001098177, + "grad_norm": 2.4919497966766357, + "learning_rate": 9.884742041712404e-07, + "loss": 0.965, + "mean_token_accuracy": 0.7075890302658081, + "num_tokens": 46559834.0, + "step": 1802 + }, + { + "epoch": 0.19800131781243135, + "grad_norm": 2.274155855178833, + "learning_rate": 9.890230515916574e-07, + "loss": 1.0121, + "mean_token_accuracy": 0.695861279964447, + "num_tokens": 46585403.0, + "step": 1803 + }, + { + "epoch": 0.19811113551504503, + "grad_norm": 2.3345513343811035, + "learning_rate": 9.895718990120747e-07, + "loss": 0.9672, + "mean_token_accuracy": 0.706362247467041, + "num_tokens": 46611766.0, + "step": 1804 + }, + { + "epoch": 0.19822095321765867, + "grad_norm": 2.2830357551574707, + "learning_rate": 9.901207464324918e-07, + "loss": 0.979, + "mean_token_accuracy": 0.708478569984436, + "num_tokens": 46636512.0, + "step": 1805 + }, + { + "epoch": 0.19833077092027235, + "grad_norm": 2.300893783569336, + "learning_rate": 9.906695938529088e-07, + "loss": 0.958, + "mean_token_accuracy": 0.7160468101501465, + "num_tokens": 46661375.0, + "step": 1806 + }, + { + "epoch": 0.19844058862288602, + "grad_norm": 2.0222394466400146, + "learning_rate": 9.91218441273326e-07, + "loss": 1.0108, + "mean_token_accuracy": 0.6941244602203369, + "num_tokens": 46693626.0, + "step": 1807 + }, + { + "epoch": 0.19855040632549967, + "grad_norm": 2.3257791996002197, + "learning_rate": 9.91767288693743e-07, + "loss": 1.105, + "mean_token_accuracy": 0.692840576171875, + "num_tokens": 46720316.0, + "step": 1808 + }, + { + "epoch": 0.19866022402811334, + "grad_norm": 2.557119607925415, + "learning_rate": 9.923161361141602e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6867842078208923, + "num_tokens": 46742101.0, + "step": 1809 + }, + { + "epoch": 0.198770041730727, + "grad_norm": 2.3339836597442627, + "learning_rate": 9.928649835345773e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.6907398700714111, + "num_tokens": 46768902.0, + "step": 1810 + }, + { + "epoch": 0.19887985943334066, + "grad_norm": 2.3694357872009277, + "learning_rate": 9.934138309549944e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6815842390060425, + "num_tokens": 46794856.0, + "step": 1811 + }, + { + "epoch": 0.1989896771359543, + "grad_norm": 2.522024393081665, + "learning_rate": 9.939626783754116e-07, + "loss": 0.9643, + "mean_token_accuracy": 0.7099353075027466, + "num_tokens": 46816635.0, + "step": 1812 + }, + { + "epoch": 0.19909949483856798, + "grad_norm": 2.2982065677642822, + "learning_rate": 9.945115257958287e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6845365762710571, + "num_tokens": 46842306.0, + "step": 1813 + }, + { + "epoch": 0.19920931254118163, + "grad_norm": 2.273404121398926, + "learning_rate": 9.950603732162458e-07, + "loss": 0.9804, + "mean_token_accuracy": 0.7048953771591187, + "num_tokens": 46870306.0, + "step": 1814 + }, + { + "epoch": 0.1993191302437953, + "grad_norm": 2.2461397647857666, + "learning_rate": 9.95609220636663e-07, + "loss": 1.1532, + "mean_token_accuracy": 0.6612470746040344, + "num_tokens": 46896971.0, + "step": 1815 + }, + { + "epoch": 0.19942894794640897, + "grad_norm": 2.703758955001831, + "learning_rate": 9.9615806805708e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6915478706359863, + "num_tokens": 46921084.0, + "step": 1816 + }, + { + "epoch": 0.19953876564902262, + "grad_norm": 2.8932714462280273, + "learning_rate": 9.967069154774972e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.7052876949310303, + "num_tokens": 46937749.0, + "step": 1817 + }, + { + "epoch": 0.1996485833516363, + "grad_norm": 2.223783254623413, + "learning_rate": 9.972557628979144e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.6816633939743042, + "num_tokens": 46970453.0, + "step": 1818 + }, + { + "epoch": 0.19975840105424994, + "grad_norm": 2.6072347164154053, + "learning_rate": 9.978046103183315e-07, + "loss": 0.9263, + "mean_token_accuracy": 0.7121469974517822, + "num_tokens": 46989377.0, + "step": 1819 + }, + { + "epoch": 0.1998682187568636, + "grad_norm": 2.175729751586914, + "learning_rate": 9.983534577387486e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6867687702178955, + "num_tokens": 47018769.0, + "step": 1820 + }, + { + "epoch": 0.19997803645947726, + "grad_norm": 2.2562644481658936, + "learning_rate": 9.989023051591658e-07, + "loss": 1.0337, + "mean_token_accuracy": 0.6916120052337646, + "num_tokens": 47043861.0, + "step": 1821 + }, + { + "epoch": 0.20008785416209093, + "grad_norm": 1.9690741300582886, + "learning_rate": 9.994511525795829e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.6708359122276306, + "num_tokens": 47075955.0, + "step": 1822 + }, + { + "epoch": 0.20019767186470458, + "grad_norm": 2.3141398429870605, + "learning_rate": 1e-06, + "loss": 1.1359, + "mean_token_accuracy": 0.6650892496109009, + "num_tokens": 47101169.0, + "step": 1823 + }, + { + "epoch": 0.20030748956731825, + "grad_norm": 2.5314080715179443, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6989438533782959, + "num_tokens": 47122882.0, + "step": 1824 + }, + { + "epoch": 0.2004173072699319, + "grad_norm": 2.556128740310669, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7185449600219727, + "num_tokens": 47142761.0, + "step": 1825 + }, + { + "epoch": 0.20052712497254557, + "grad_norm": 2.3407092094421387, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.680363655090332, + "num_tokens": 47169986.0, + "step": 1826 + }, + { + "epoch": 0.20063694267515925, + "grad_norm": 2.1275510787963867, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6975085139274597, + "num_tokens": 47198575.0, + "step": 1827 + }, + { + "epoch": 0.2007467603777729, + "grad_norm": 2.3633010387420654, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6936877965927124, + "num_tokens": 47222699.0, + "step": 1828 + }, + { + "epoch": 0.20085657808038657, + "grad_norm": 2.137254238128662, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7115371823310852, + "num_tokens": 47249344.0, + "step": 1829 + }, + { + "epoch": 0.2009663957830002, + "grad_norm": 2.541733503341675, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.676663875579834, + "num_tokens": 47272433.0, + "step": 1830 + }, + { + "epoch": 0.20107621348561389, + "grad_norm": 1.9607375860214233, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7057994604110718, + "num_tokens": 47303048.0, + "step": 1831 + }, + { + "epoch": 0.20118603118822753, + "grad_norm": 2.4567785263061523, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7181744575500488, + "num_tokens": 47323885.0, + "step": 1832 + }, + { + "epoch": 0.2012958488908412, + "grad_norm": 2.227379560470581, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6923943758010864, + "num_tokens": 47350802.0, + "step": 1833 + }, + { + "epoch": 0.20140566659345485, + "grad_norm": 2.5970969200134277, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7023957371711731, + "num_tokens": 47371781.0, + "step": 1834 + }, + { + "epoch": 0.20151548429606853, + "grad_norm": 2.246225357055664, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7002476453781128, + "num_tokens": 47400746.0, + "step": 1835 + }, + { + "epoch": 0.2016253019986822, + "grad_norm": 2.6212728023529053, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7000849843025208, + "num_tokens": 47420697.0, + "step": 1836 + }, + { + "epoch": 0.20173511970129585, + "grad_norm": 2.113243579864502, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7024016976356506, + "num_tokens": 47448933.0, + "step": 1837 + }, + { + "epoch": 0.20184493740390952, + "grad_norm": 2.1446335315704346, + "learning_rate": 1e-06, + "loss": 1.1119, + "mean_token_accuracy": 0.6751524806022644, + "num_tokens": 47478889.0, + "step": 1838 + }, + { + "epoch": 0.20195475510652316, + "grad_norm": 2.4764151573181152, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7303271293640137, + "num_tokens": 47499164.0, + "step": 1839 + }, + { + "epoch": 0.20206457280913684, + "grad_norm": 2.6834805011749268, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6883862018585205, + "num_tokens": 47520384.0, + "step": 1840 + }, + { + "epoch": 0.20217439051175048, + "grad_norm": 2.4702861309051514, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6932997703552246, + "num_tokens": 47541824.0, + "step": 1841 + }, + { + "epoch": 0.20228420821436416, + "grad_norm": 2.3077456951141357, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7094550132751465, + "num_tokens": 47565373.0, + "step": 1842 + }, + { + "epoch": 0.2023940259169778, + "grad_norm": 2.326526165008545, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7089883089065552, + "num_tokens": 47589876.0, + "step": 1843 + }, + { + "epoch": 0.20250384361959148, + "grad_norm": 2.285112142562866, + "learning_rate": 1e-06, + "loss": 1.0992, + "mean_token_accuracy": 0.6746657490730286, + "num_tokens": 47615768.0, + "step": 1844 + }, + { + "epoch": 0.20261366132220515, + "grad_norm": 2.4854798316955566, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.6862969398498535, + "num_tokens": 47637439.0, + "step": 1845 + }, + { + "epoch": 0.2027234790248188, + "grad_norm": 2.0366859436035156, + "learning_rate": 1e-06, + "loss": 1.0748, + "mean_token_accuracy": 0.6735020875930786, + "num_tokens": 47667031.0, + "step": 1846 + }, + { + "epoch": 0.20283329672743247, + "grad_norm": 2.0981969833374023, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7153617143630981, + "num_tokens": 47692986.0, + "step": 1847 + }, + { + "epoch": 0.20294311443004612, + "grad_norm": 2.527834892272949, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6889652013778687, + "num_tokens": 47714630.0, + "step": 1848 + }, + { + "epoch": 0.2030529321326598, + "grad_norm": 2.43426251411438, + "learning_rate": 1e-06, + "loss": 1.1106, + "mean_token_accuracy": 0.6712252497673035, + "num_tokens": 47739066.0, + "step": 1849 + }, + { + "epoch": 0.20316274983527344, + "grad_norm": 2.4941234588623047, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6928153038024902, + "num_tokens": 47760248.0, + "step": 1850 + }, + { + "epoch": 0.2032725675378871, + "grad_norm": 2.4990344047546387, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7185670137405396, + "num_tokens": 47781223.0, + "step": 1851 + }, + { + "epoch": 0.20338238524050076, + "grad_norm": 2.110281229019165, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6798771023750305, + "num_tokens": 47807959.0, + "step": 1852 + }, + { + "epoch": 0.20349220294311443, + "grad_norm": 2.1795780658721924, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6971464157104492, + "num_tokens": 47836437.0, + "step": 1853 + }, + { + "epoch": 0.2036020206457281, + "grad_norm": 2.2493507862091064, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6797225475311279, + "num_tokens": 47863966.0, + "step": 1854 + }, + { + "epoch": 0.20371183834834175, + "grad_norm": 2.2427453994750977, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6973627805709839, + "num_tokens": 47891933.0, + "step": 1855 + }, + { + "epoch": 0.20382165605095542, + "grad_norm": 2.1075496673583984, + "learning_rate": 1e-06, + "loss": 1.0968, + "mean_token_accuracy": 0.6747036576271057, + "num_tokens": 47921106.0, + "step": 1856 + }, + { + "epoch": 0.20393147375356907, + "grad_norm": 2.060040235519409, + "learning_rate": 1e-06, + "loss": 1.0922, + "mean_token_accuracy": 0.6859272718429565, + "num_tokens": 47953234.0, + "step": 1857 + }, + { + "epoch": 0.20404129145618274, + "grad_norm": 2.232182502746582, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7000327110290527, + "num_tokens": 47979199.0, + "step": 1858 + }, + { + "epoch": 0.2041511091587964, + "grad_norm": 2.267416000366211, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6774224042892456, + "num_tokens": 48006806.0, + "step": 1859 + }, + { + "epoch": 0.20426092686141006, + "grad_norm": 2.4204506874084473, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.715157151222229, + "num_tokens": 48029631.0, + "step": 1860 + }, + { + "epoch": 0.2043707445640237, + "grad_norm": 2.4939687252044678, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.689773440361023, + "num_tokens": 48052238.0, + "step": 1861 + }, + { + "epoch": 0.20448056226663738, + "grad_norm": 2.430678606033325, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6784423589706421, + "num_tokens": 48076344.0, + "step": 1862 + }, + { + "epoch": 0.20459037996925103, + "grad_norm": 2.315659284591675, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7069861888885498, + "num_tokens": 48100222.0, + "step": 1863 + }, + { + "epoch": 0.2047001976718647, + "grad_norm": 2.147782802581787, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6975916028022766, + "num_tokens": 48130772.0, + "step": 1864 + }, + { + "epoch": 0.20481001537447838, + "grad_norm": 2.074601173400879, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7109734416007996, + "num_tokens": 48158856.0, + "step": 1865 + }, + { + "epoch": 0.20491983307709202, + "grad_norm": 2.3738479614257812, + "learning_rate": 1e-06, + "loss": 1.0877, + "mean_token_accuracy": 0.6754719614982605, + "num_tokens": 48182647.0, + "step": 1866 + }, + { + "epoch": 0.2050296507797057, + "grad_norm": 2.537712812423706, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6886967420578003, + "num_tokens": 48202781.0, + "step": 1867 + }, + { + "epoch": 0.20513946848231934, + "grad_norm": 2.293036937713623, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.697685956954956, + "num_tokens": 48229464.0, + "step": 1868 + }, + { + "epoch": 0.20524928618493302, + "grad_norm": 2.2243411540985107, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7004474401473999, + "num_tokens": 48253442.0, + "step": 1869 + }, + { + "epoch": 0.20535910388754666, + "grad_norm": 2.1598024368286133, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6873550415039062, + "num_tokens": 48283902.0, + "step": 1870 + }, + { + "epoch": 0.20546892159016034, + "grad_norm": 2.3275794982910156, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6963068246841431, + "num_tokens": 48310266.0, + "step": 1871 + }, + { + "epoch": 0.20557873929277398, + "grad_norm": 2.4129114151000977, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6862452626228333, + "num_tokens": 48334157.0, + "step": 1872 + }, + { + "epoch": 0.20568855699538766, + "grad_norm": 2.0795722007751465, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6773360967636108, + "num_tokens": 48366475.0, + "step": 1873 + }, + { + "epoch": 0.20579837469800133, + "grad_norm": 2.0281147956848145, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6980286836624146, + "num_tokens": 48399089.0, + "step": 1874 + }, + { + "epoch": 0.20590819240061498, + "grad_norm": 2.169069528579712, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7108708620071411, + "num_tokens": 48426990.0, + "step": 1875 + }, + { + "epoch": 0.20601801010322865, + "grad_norm": 2.159505605697632, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6847414970397949, + "num_tokens": 48455449.0, + "step": 1876 + }, + { + "epoch": 0.2061278278058423, + "grad_norm": 2.532038927078247, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7155501246452332, + "num_tokens": 48475625.0, + "step": 1877 + }, + { + "epoch": 0.20623764550845597, + "grad_norm": 2.0240888595581055, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6706161499023438, + "num_tokens": 48507457.0, + "step": 1878 + }, + { + "epoch": 0.20634746321106961, + "grad_norm": 2.5024213790893555, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6869663596153259, + "num_tokens": 48532190.0, + "step": 1879 + }, + { + "epoch": 0.2064572809136833, + "grad_norm": 2.2398529052734375, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7015290260314941, + "num_tokens": 48557064.0, + "step": 1880 + }, + { + "epoch": 0.20656709861629693, + "grad_norm": 2.1506497859954834, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6702972054481506, + "num_tokens": 48585898.0, + "step": 1881 + }, + { + "epoch": 0.2066769163189106, + "grad_norm": 2.2983314990997314, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6894018650054932, + "num_tokens": 48611912.0, + "step": 1882 + }, + { + "epoch": 0.20678673402152428, + "grad_norm": 2.3415071964263916, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6981804966926575, + "num_tokens": 48639480.0, + "step": 1883 + }, + { + "epoch": 0.20689655172413793, + "grad_norm": 2.3804447650909424, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.681289553642273, + "num_tokens": 48662031.0, + "step": 1884 + }, + { + "epoch": 0.2070063694267516, + "grad_norm": 2.188190460205078, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.723173975944519, + "num_tokens": 48687700.0, + "step": 1885 + }, + { + "epoch": 0.20711618712936525, + "grad_norm": 2.1833693981170654, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6880978345870972, + "num_tokens": 48717231.0, + "step": 1886 + }, + { + "epoch": 0.20722600483197892, + "grad_norm": 2.3156166076660156, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6858159303665161, + "num_tokens": 48740364.0, + "step": 1887 + }, + { + "epoch": 0.20733582253459257, + "grad_norm": 2.280902624130249, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6857059597969055, + "num_tokens": 48766553.0, + "step": 1888 + }, + { + "epoch": 0.20744564023720624, + "grad_norm": 2.647239923477173, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6970459222793579, + "num_tokens": 48788535.0, + "step": 1889 + }, + { + "epoch": 0.2075554579398199, + "grad_norm": 2.183716297149658, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6902430057525635, + "num_tokens": 48816420.0, + "step": 1890 + }, + { + "epoch": 0.20766527564243356, + "grad_norm": 2.0843820571899414, + "learning_rate": 1e-06, + "loss": 1.1354, + "mean_token_accuracy": 0.6662891507148743, + "num_tokens": 48849158.0, + "step": 1891 + }, + { + "epoch": 0.20777509334504723, + "grad_norm": 2.3976941108703613, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7098416686058044, + "num_tokens": 48871897.0, + "step": 1892 + }, + { + "epoch": 0.20788491104766088, + "grad_norm": 2.29921817779541, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6926572322845459, + "num_tokens": 48898268.0, + "step": 1893 + }, + { + "epoch": 0.20799472875027455, + "grad_norm": 2.182634115219116, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6769071221351624, + "num_tokens": 48925431.0, + "step": 1894 + }, + { + "epoch": 0.2081045464528882, + "grad_norm": 2.2955217361450195, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6780692338943481, + "num_tokens": 48950555.0, + "step": 1895 + }, + { + "epoch": 0.20821436415550187, + "grad_norm": 2.279437780380249, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.68320232629776, + "num_tokens": 48975570.0, + "step": 1896 + }, + { + "epoch": 0.20832418185811552, + "grad_norm": 2.8261847496032715, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7156237363815308, + "num_tokens": 48992995.0, + "step": 1897 + }, + { + "epoch": 0.2084339995607292, + "grad_norm": 2.773714780807495, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6868467926979065, + "num_tokens": 49012455.0, + "step": 1898 + }, + { + "epoch": 0.20854381726334284, + "grad_norm": 2.452087879180908, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.6819396018981934, + "num_tokens": 49036455.0, + "step": 1899 + }, + { + "epoch": 0.2086536349659565, + "grad_norm": 2.5009634494781494, + "learning_rate": 1e-06, + "loss": 1.1005, + "mean_token_accuracy": 0.6841752529144287, + "num_tokens": 49058836.0, + "step": 1900 + }, + { + "epoch": 0.20876345266857016, + "grad_norm": 1.9796912670135498, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6853792667388916, + "num_tokens": 49089594.0, + "step": 1901 + }, + { + "epoch": 0.20887327037118383, + "grad_norm": 2.0221707820892334, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7105711698532104, + "num_tokens": 49119711.0, + "step": 1902 + }, + { + "epoch": 0.2089830880737975, + "grad_norm": 2.493431806564331, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6934720277786255, + "num_tokens": 49139981.0, + "step": 1903 + }, + { + "epoch": 0.20909290577641115, + "grad_norm": 2.102107286453247, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.69621741771698, + "num_tokens": 49166851.0, + "step": 1904 + }, + { + "epoch": 0.20920272347902483, + "grad_norm": 2.2569901943206787, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7157481908798218, + "num_tokens": 49192191.0, + "step": 1905 + }, + { + "epoch": 0.20931254118163847, + "grad_norm": 2.352661371231079, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6994971036911011, + "num_tokens": 49214402.0, + "step": 1906 + }, + { + "epoch": 0.20942235888425215, + "grad_norm": 2.3998632431030273, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6962332129478455, + "num_tokens": 49236480.0, + "step": 1907 + }, + { + "epoch": 0.2095321765868658, + "grad_norm": 2.4707682132720947, + "learning_rate": 1e-06, + "loss": 1.0875, + "mean_token_accuracy": 0.6833639144897461, + "num_tokens": 49258229.0, + "step": 1908 + }, + { + "epoch": 0.20964199428947947, + "grad_norm": 2.2644567489624023, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6903696060180664, + "num_tokens": 49282751.0, + "step": 1909 + }, + { + "epoch": 0.2097518119920931, + "grad_norm": 2.41253924369812, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6897308230400085, + "num_tokens": 49307816.0, + "step": 1910 + }, + { + "epoch": 0.20986162969470679, + "grad_norm": 2.050058603286743, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7190377116203308, + "num_tokens": 49336332.0, + "step": 1911 + }, + { + "epoch": 0.20997144739732046, + "grad_norm": 2.5642905235290527, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7296639084815979, + "num_tokens": 49354693.0, + "step": 1912 + }, + { + "epoch": 0.2100812650999341, + "grad_norm": 2.4898319244384766, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6854008436203003, + "num_tokens": 49377459.0, + "step": 1913 + }, + { + "epoch": 0.21019108280254778, + "grad_norm": 2.1439437866210938, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6978796720504761, + "num_tokens": 49406346.0, + "step": 1914 + }, + { + "epoch": 0.21030090050516143, + "grad_norm": 2.166095018386841, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6816881895065308, + "num_tokens": 49435725.0, + "step": 1915 + }, + { + "epoch": 0.2104107182077751, + "grad_norm": 2.516584873199463, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7279311418533325, + "num_tokens": 49455546.0, + "step": 1916 + }, + { + "epoch": 0.21052053591038875, + "grad_norm": 2.2781760692596436, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.675736665725708, + "num_tokens": 49480124.0, + "step": 1917 + }, + { + "epoch": 0.21063035361300242, + "grad_norm": 2.500526189804077, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.697810173034668, + "num_tokens": 49501663.0, + "step": 1918 + }, + { + "epoch": 0.21074017131561606, + "grad_norm": 2.449380874633789, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6882633566856384, + "num_tokens": 49524109.0, + "step": 1919 + }, + { + "epoch": 0.21084998901822974, + "grad_norm": 2.1134033203125, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6894786357879639, + "num_tokens": 49555274.0, + "step": 1920 + }, + { + "epoch": 0.2109598067208434, + "grad_norm": 2.059701442718506, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6923508644104004, + "num_tokens": 49582904.0, + "step": 1921 + }, + { + "epoch": 0.21106962442345706, + "grad_norm": 2.3100411891937256, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6927655339241028, + "num_tokens": 49609415.0, + "step": 1922 + }, + { + "epoch": 0.21117944212607073, + "grad_norm": 2.1224753856658936, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7161468863487244, + "num_tokens": 49636718.0, + "step": 1923 + }, + { + "epoch": 0.21128925982868438, + "grad_norm": 2.307831287384033, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.692394495010376, + "num_tokens": 49660398.0, + "step": 1924 + }, + { + "epoch": 0.21139907753129805, + "grad_norm": 2.198296546936035, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7020289897918701, + "num_tokens": 49689059.0, + "step": 1925 + }, + { + "epoch": 0.2115088952339117, + "grad_norm": 2.3803529739379883, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7034323215484619, + "num_tokens": 49712860.0, + "step": 1926 + }, + { + "epoch": 0.21161871293652537, + "grad_norm": 2.052351474761963, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6812209486961365, + "num_tokens": 49743831.0, + "step": 1927 + }, + { + "epoch": 0.21172853063913902, + "grad_norm": 2.240525245666504, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6881847381591797, + "num_tokens": 49769479.0, + "step": 1928 + }, + { + "epoch": 0.2118383483417527, + "grad_norm": 2.3985488414764404, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6812365055084229, + "num_tokens": 49793569.0, + "step": 1929 + }, + { + "epoch": 0.21194816604436637, + "grad_norm": 2.2824783325195312, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7086283564567566, + "num_tokens": 49818135.0, + "step": 1930 + }, + { + "epoch": 0.21205798374698, + "grad_norm": 2.4176554679870605, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7345713376998901, + "num_tokens": 49838866.0, + "step": 1931 + }, + { + "epoch": 0.21216780144959368, + "grad_norm": 2.357877731323242, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6967410445213318, + "num_tokens": 49864197.0, + "step": 1932 + }, + { + "epoch": 0.21227761915220733, + "grad_norm": 2.348752975463867, + "learning_rate": 1e-06, + "loss": 1.0962, + "mean_token_accuracy": 0.676685094833374, + "num_tokens": 49888691.0, + "step": 1933 + }, + { + "epoch": 0.212387436854821, + "grad_norm": 2.024324893951416, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6913589239120483, + "num_tokens": 49920370.0, + "step": 1934 + }, + { + "epoch": 0.21249725455743465, + "grad_norm": 2.483741044998169, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7323303818702698, + "num_tokens": 49939617.0, + "step": 1935 + }, + { + "epoch": 0.21260707226004832, + "grad_norm": 2.454582452774048, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6989994049072266, + "num_tokens": 49961264.0, + "step": 1936 + }, + { + "epoch": 0.21271688996266197, + "grad_norm": 2.0539932250976562, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6959323287010193, + "num_tokens": 49992727.0, + "step": 1937 + }, + { + "epoch": 0.21282670766527564, + "grad_norm": 2.167550802230835, + "learning_rate": 1e-06, + "loss": 1.1225, + "mean_token_accuracy": 0.6679728031158447, + "num_tokens": 50022694.0, + "step": 1938 + }, + { + "epoch": 0.2129365253678893, + "grad_norm": 1.909661054611206, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.6830003261566162, + "num_tokens": 50059117.0, + "step": 1939 + }, + { + "epoch": 0.21304634307050296, + "grad_norm": 2.4635441303253174, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6874088048934937, + "num_tokens": 50081155.0, + "step": 1940 + }, + { + "epoch": 0.21315616077311664, + "grad_norm": 2.3051645755767822, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6935698390007019, + "num_tokens": 50110977.0, + "step": 1941 + }, + { + "epoch": 0.21326597847573028, + "grad_norm": 2.028697967529297, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6864513158798218, + "num_tokens": 50143960.0, + "step": 1942 + }, + { + "epoch": 0.21337579617834396, + "grad_norm": 2.356977939605713, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6982343792915344, + "num_tokens": 50168234.0, + "step": 1943 + }, + { + "epoch": 0.2134856138809576, + "grad_norm": 2.0894293785095215, + "learning_rate": 1e-06, + "loss": 1.1826, + "mean_token_accuracy": 0.6557488441467285, + "num_tokens": 50198423.0, + "step": 1944 + }, + { + "epoch": 0.21359543158357128, + "grad_norm": 2.181436061859131, + "learning_rate": 1e-06, + "loss": 1.0899, + "mean_token_accuracy": 0.6747013330459595, + "num_tokens": 50229117.0, + "step": 1945 + }, + { + "epoch": 0.21370524928618492, + "grad_norm": 2.5396370887756348, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7087562680244446, + "num_tokens": 50251050.0, + "step": 1946 + }, + { + "epoch": 0.2138150669887986, + "grad_norm": 2.1165871620178223, + "learning_rate": 1e-06, + "loss": 1.1347, + "mean_token_accuracy": 0.6612007021903992, + "num_tokens": 50283561.0, + "step": 1947 + }, + { + "epoch": 0.21392488469141224, + "grad_norm": 2.063375473022461, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7148712277412415, + "num_tokens": 50312742.0, + "step": 1948 + }, + { + "epoch": 0.21403470239402592, + "grad_norm": 2.447523593902588, + "learning_rate": 1e-06, + "loss": 1.1183, + "mean_token_accuracy": 0.666659951210022, + "num_tokens": 50335775.0, + "step": 1949 + }, + { + "epoch": 0.2141445200966396, + "grad_norm": 2.072444438934326, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6985130906105042, + "num_tokens": 50363941.0, + "step": 1950 + }, + { + "epoch": 0.21425433779925324, + "grad_norm": 2.227982997894287, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6893398761749268, + "num_tokens": 50390475.0, + "step": 1951 + }, + { + "epoch": 0.2143641555018669, + "grad_norm": 2.414372444152832, + "learning_rate": 1e-06, + "loss": 1.0806, + "mean_token_accuracy": 0.6750061511993408, + "num_tokens": 50414720.0, + "step": 1952 + }, + { + "epoch": 0.21447397320448056, + "grad_norm": 2.145209312438965, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6910707950592041, + "num_tokens": 50440691.0, + "step": 1953 + }, + { + "epoch": 0.21458379090709423, + "grad_norm": 2.600942850112915, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6998162269592285, + "num_tokens": 50459360.0, + "step": 1954 + }, + { + "epoch": 0.21469360860970788, + "grad_norm": 2.6956095695495605, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7128721475601196, + "num_tokens": 50478898.0, + "step": 1955 + }, + { + "epoch": 0.21480342631232155, + "grad_norm": 2.271672248840332, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7200330495834351, + "num_tokens": 50504233.0, + "step": 1956 + }, + { + "epoch": 0.2149132440149352, + "grad_norm": 2.1393208503723145, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6777172684669495, + "num_tokens": 50532982.0, + "step": 1957 + }, + { + "epoch": 0.21502306171754887, + "grad_norm": 2.3389041423797607, + "learning_rate": 1e-06, + "loss": 1.1203, + "mean_token_accuracy": 0.6657935380935669, + "num_tokens": 50557095.0, + "step": 1958 + }, + { + "epoch": 0.21513287942016254, + "grad_norm": 2.425910234451294, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.7033722996711731, + "num_tokens": 50580399.0, + "step": 1959 + }, + { + "epoch": 0.2152426971227762, + "grad_norm": 2.0936951637268066, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6871225833892822, + "num_tokens": 50609069.0, + "step": 1960 + }, + { + "epoch": 0.21535251482538986, + "grad_norm": 2.134704113006592, + "learning_rate": 1e-06, + "loss": 1.0702, + "mean_token_accuracy": 0.6760334372520447, + "num_tokens": 50638251.0, + "step": 1961 + }, + { + "epoch": 0.2154623325280035, + "grad_norm": 2.1463117599487305, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6794466972351074, + "num_tokens": 50667113.0, + "step": 1962 + }, + { + "epoch": 0.21557215023061718, + "grad_norm": 2.5336108207702637, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.683873176574707, + "num_tokens": 50690293.0, + "step": 1963 + }, + { + "epoch": 0.21568196793323083, + "grad_norm": 2.0181775093078613, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.707983672618866, + "num_tokens": 50717941.0, + "step": 1964 + }, + { + "epoch": 0.2157917856358445, + "grad_norm": 2.055516242980957, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6942112445831299, + "num_tokens": 50746255.0, + "step": 1965 + }, + { + "epoch": 0.21590160333845815, + "grad_norm": 2.0673155784606934, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6944658756256104, + "num_tokens": 50773974.0, + "step": 1966 + }, + { + "epoch": 0.21601142104107182, + "grad_norm": 2.3251397609710693, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7027342319488525, + "num_tokens": 50797240.0, + "step": 1967 + }, + { + "epoch": 0.2161212387436855, + "grad_norm": 2.4821786880493164, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7049990892410278, + "num_tokens": 50818483.0, + "step": 1968 + }, + { + "epoch": 0.21623105644629914, + "grad_norm": 2.249607801437378, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6977792978286743, + "num_tokens": 50844934.0, + "step": 1969 + }, + { + "epoch": 0.21634087414891282, + "grad_norm": 2.3754491806030273, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6849050521850586, + "num_tokens": 50868591.0, + "step": 1970 + }, + { + "epoch": 0.21645069185152646, + "grad_norm": 2.399648666381836, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6945484280586243, + "num_tokens": 50891145.0, + "step": 1971 + }, + { + "epoch": 0.21656050955414013, + "grad_norm": 2.1881020069122314, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6892526149749756, + "num_tokens": 50916864.0, + "step": 1972 + }, + { + "epoch": 0.21667032725675378, + "grad_norm": 2.513251304626465, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6998473405838013, + "num_tokens": 50937867.0, + "step": 1973 + }, + { + "epoch": 0.21678014495936745, + "grad_norm": 2.1927497386932373, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6764077544212341, + "num_tokens": 50964601.0, + "step": 1974 + }, + { + "epoch": 0.2168899626619811, + "grad_norm": 2.2325332164764404, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6882637739181519, + "num_tokens": 50989998.0, + "step": 1975 + }, + { + "epoch": 0.21699978036459477, + "grad_norm": 2.0867459774017334, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7144519686698914, + "num_tokens": 51015524.0, + "step": 1976 + }, + { + "epoch": 0.21710959806720842, + "grad_norm": 1.9742306470870972, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.6826363205909729, + "num_tokens": 51049972.0, + "step": 1977 + }, + { + "epoch": 0.2172194157698221, + "grad_norm": 2.080648422241211, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6871428489685059, + "num_tokens": 51077965.0, + "step": 1978 + }, + { + "epoch": 0.21732923347243577, + "grad_norm": 2.176562786102295, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6783803701400757, + "num_tokens": 51107842.0, + "step": 1979 + }, + { + "epoch": 0.2174390511750494, + "grad_norm": 2.192741632461548, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.700397253036499, + "num_tokens": 51132792.0, + "step": 1980 + }, + { + "epoch": 0.2175488688776631, + "grad_norm": 2.237051010131836, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6893057823181152, + "num_tokens": 51160228.0, + "step": 1981 + }, + { + "epoch": 0.21765868658027673, + "grad_norm": 2.2830147743225098, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.6738835573196411, + "num_tokens": 51186157.0, + "step": 1982 + }, + { + "epoch": 0.2177685042828904, + "grad_norm": 2.102134943008423, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7122358679771423, + "num_tokens": 51217730.0, + "step": 1983 + }, + { + "epoch": 0.21787832198550405, + "grad_norm": 2.4358856678009033, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6948167085647583, + "num_tokens": 51240752.0, + "step": 1984 + }, + { + "epoch": 0.21798813968811773, + "grad_norm": 1.9627323150634766, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6908152103424072, + "num_tokens": 51274024.0, + "step": 1985 + }, + { + "epoch": 0.21809795739073137, + "grad_norm": 2.68422269821167, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7238094806671143, + "num_tokens": 51292074.0, + "step": 1986 + }, + { + "epoch": 0.21820777509334505, + "grad_norm": 2.3047261238098145, + "learning_rate": 1e-06, + "loss": 1.0534, + "mean_token_accuracy": 0.6804360747337341, + "num_tokens": 51316888.0, + "step": 1987 + }, + { + "epoch": 0.21831759279595872, + "grad_norm": 2.4952659606933594, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6921126842498779, + "num_tokens": 51340259.0, + "step": 1988 + }, + { + "epoch": 0.21842741049857237, + "grad_norm": 2.327920913696289, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6930339336395264, + "num_tokens": 51366881.0, + "step": 1989 + }, + { + "epoch": 0.21853722820118604, + "grad_norm": 2.7866742610931396, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7044850587844849, + "num_tokens": 51386093.0, + "step": 1990 + }, + { + "epoch": 0.21864704590379969, + "grad_norm": 2.0110268592834473, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.693967342376709, + "num_tokens": 51417417.0, + "step": 1991 + }, + { + "epoch": 0.21875686360641336, + "grad_norm": 2.4807753562927246, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.684831976890564, + "num_tokens": 51440928.0, + "step": 1992 + }, + { + "epoch": 0.218866681309027, + "grad_norm": 2.5509064197540283, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7088305354118347, + "num_tokens": 51461057.0, + "step": 1993 + }, + { + "epoch": 0.21897649901164068, + "grad_norm": 2.2464704513549805, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.708553671836853, + "num_tokens": 51486199.0, + "step": 1994 + }, + { + "epoch": 0.21908631671425433, + "grad_norm": 2.0008950233459473, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.7000282406806946, + "num_tokens": 51517484.0, + "step": 1995 + }, + { + "epoch": 0.219196134416868, + "grad_norm": 2.2515220642089844, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6822789907455444, + "num_tokens": 51544297.0, + "step": 1996 + }, + { + "epoch": 0.21930595211948167, + "grad_norm": 2.2571728229522705, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6889485120773315, + "num_tokens": 51570516.0, + "step": 1997 + }, + { + "epoch": 0.21941576982209532, + "grad_norm": 2.2778818607330322, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7172071933746338, + "num_tokens": 51596342.0, + "step": 1998 + }, + { + "epoch": 0.219525587524709, + "grad_norm": 2.5072858333587646, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7047348022460938, + "num_tokens": 51618263.0, + "step": 1999 + }, + { + "epoch": 0.21963540522732264, + "grad_norm": 1.9306001663208008, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7277308106422424, + "num_tokens": 51649155.0, + "step": 2000 + }, + { + "epoch": 0.2197452229299363, + "grad_norm": 2.49021577835083, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6884121298789978, + "num_tokens": 51671576.0, + "step": 2001 + }, + { + "epoch": 0.21985504063254996, + "grad_norm": 2.4072768688201904, + "learning_rate": 1e-06, + "loss": 1.0968, + "mean_token_accuracy": 0.6800920367240906, + "num_tokens": 51696873.0, + "step": 2002 + }, + { + "epoch": 0.21996485833516363, + "grad_norm": 2.1893198490142822, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7129366397857666, + "num_tokens": 51723253.0, + "step": 2003 + }, + { + "epoch": 0.22007467603777728, + "grad_norm": 2.3538525104522705, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7082988023757935, + "num_tokens": 51746419.0, + "step": 2004 + }, + { + "epoch": 0.22018449374039095, + "grad_norm": 2.106628894805908, + "learning_rate": 1e-06, + "loss": 1.1315, + "mean_token_accuracy": 0.6711186170578003, + "num_tokens": 51779352.0, + "step": 2005 + }, + { + "epoch": 0.22029431144300463, + "grad_norm": 2.640599012374878, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7160677313804626, + "num_tokens": 51800386.0, + "step": 2006 + }, + { + "epoch": 0.22040412914561827, + "grad_norm": 2.606966495513916, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7049757242202759, + "num_tokens": 51820386.0, + "step": 2007 + }, + { + "epoch": 0.22051394684823195, + "grad_norm": 2.3205161094665527, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7022007703781128, + "num_tokens": 51843303.0, + "step": 2008 + }, + { + "epoch": 0.2206237645508456, + "grad_norm": 2.3178999423980713, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7150717973709106, + "num_tokens": 51867024.0, + "step": 2009 + }, + { + "epoch": 0.22073358225345927, + "grad_norm": 2.517756700515747, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.6856067180633545, + "num_tokens": 51888723.0, + "step": 2010 + }, + { + "epoch": 0.2208433999560729, + "grad_norm": 2.3956401348114014, + "learning_rate": 1e-06, + "loss": 1.075, + "mean_token_accuracy": 0.6753262877464294, + "num_tokens": 51913714.0, + "step": 2011 + }, + { + "epoch": 0.22095321765868658, + "grad_norm": 2.2436349391937256, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7114080190658569, + "num_tokens": 51940498.0, + "step": 2012 + }, + { + "epoch": 0.22106303536130023, + "grad_norm": 2.490445852279663, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6980850696563721, + "num_tokens": 51963701.0, + "step": 2013 + }, + { + "epoch": 0.2211728530639139, + "grad_norm": 1.9767701625823975, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7032231092453003, + "num_tokens": 51993543.0, + "step": 2014 + }, + { + "epoch": 0.22128267076652755, + "grad_norm": 2.0001981258392334, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6932055950164795, + "num_tokens": 52027448.0, + "step": 2015 + }, + { + "epoch": 0.22139248846914122, + "grad_norm": 2.2000892162323, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6901376247406006, + "num_tokens": 52056635.0, + "step": 2016 + }, + { + "epoch": 0.2215023061717549, + "grad_norm": 2.27919340133667, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7154002785682678, + "num_tokens": 52081956.0, + "step": 2017 + }, + { + "epoch": 0.22161212387436854, + "grad_norm": 2.524510622024536, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6885474920272827, + "num_tokens": 52104827.0, + "step": 2018 + }, + { + "epoch": 0.22172194157698222, + "grad_norm": 2.2614006996154785, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.7085971236228943, + "num_tokens": 52130853.0, + "step": 2019 + }, + { + "epoch": 0.22183175927959586, + "grad_norm": 2.1751809120178223, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6964041590690613, + "num_tokens": 52155158.0, + "step": 2020 + }, + { + "epoch": 0.22194157698220954, + "grad_norm": 2.351219892501831, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6887937784194946, + "num_tokens": 52179340.0, + "step": 2021 + }, + { + "epoch": 0.22205139468482318, + "grad_norm": 2.496023178100586, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.701045036315918, + "num_tokens": 52202003.0, + "step": 2022 + }, + { + "epoch": 0.22216121238743686, + "grad_norm": 2.3164591789245605, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7169428467750549, + "num_tokens": 52227009.0, + "step": 2023 + }, + { + "epoch": 0.2222710300900505, + "grad_norm": 2.5456695556640625, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6849697232246399, + "num_tokens": 52249066.0, + "step": 2024 + }, + { + "epoch": 0.22238084779266418, + "grad_norm": 2.4187216758728027, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6917954683303833, + "num_tokens": 52273818.0, + "step": 2025 + }, + { + "epoch": 0.22249066549527785, + "grad_norm": 2.2421202659606934, + "learning_rate": 1e-06, + "loss": 1.0907, + "mean_token_accuracy": 0.6837265491485596, + "num_tokens": 52300201.0, + "step": 2026 + }, + { + "epoch": 0.2226004831978915, + "grad_norm": 2.4403796195983887, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7025904059410095, + "num_tokens": 52323228.0, + "step": 2027 + }, + { + "epoch": 0.22271030090050517, + "grad_norm": 2.399620294570923, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6895498633384705, + "num_tokens": 52348170.0, + "step": 2028 + }, + { + "epoch": 0.22282011860311882, + "grad_norm": 2.390861749649048, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7083795070648193, + "num_tokens": 52370340.0, + "step": 2029 + }, + { + "epoch": 0.2229299363057325, + "grad_norm": 2.553152322769165, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7221633195877075, + "num_tokens": 52390486.0, + "step": 2030 + }, + { + "epoch": 0.22303975400834614, + "grad_norm": 2.1891872882843018, + "learning_rate": 1e-06, + "loss": 1.1816, + "mean_token_accuracy": 0.6586117744445801, + "num_tokens": 52421035.0, + "step": 2031 + }, + { + "epoch": 0.2231495717109598, + "grad_norm": 2.1888442039489746, + "learning_rate": 1e-06, + "loss": 1.0936, + "mean_token_accuracy": 0.6701271533966064, + "num_tokens": 52450132.0, + "step": 2032 + }, + { + "epoch": 0.22325938941357346, + "grad_norm": 2.351905584335327, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6929209232330322, + "num_tokens": 52476186.0, + "step": 2033 + }, + { + "epoch": 0.22336920711618713, + "grad_norm": 2.176661252975464, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6973096132278442, + "num_tokens": 52504668.0, + "step": 2034 + }, + { + "epoch": 0.2234790248188008, + "grad_norm": 2.0434467792510986, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.707348644733429, + "num_tokens": 52535873.0, + "step": 2035 + }, + { + "epoch": 0.22358884252141445, + "grad_norm": 2.090881824493408, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7018541097640991, + "num_tokens": 52564282.0, + "step": 2036 + }, + { + "epoch": 0.22369866022402812, + "grad_norm": 2.1425979137420654, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7052076458930969, + "num_tokens": 52591551.0, + "step": 2037 + }, + { + "epoch": 0.22380847792664177, + "grad_norm": 2.1197569370269775, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6936804056167603, + "num_tokens": 52619729.0, + "step": 2038 + }, + { + "epoch": 0.22391829562925544, + "grad_norm": 2.0738234519958496, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7153487205505371, + "num_tokens": 52647807.0, + "step": 2039 + }, + { + "epoch": 0.2240281133318691, + "grad_norm": 2.285952568054199, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7199416160583496, + "num_tokens": 52672377.0, + "step": 2040 + }, + { + "epoch": 0.22413793103448276, + "grad_norm": 2.29331111907959, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6775835156440735, + "num_tokens": 52696094.0, + "step": 2041 + }, + { + "epoch": 0.2242477487370964, + "grad_norm": 2.103109359741211, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6985684633255005, + "num_tokens": 52724157.0, + "step": 2042 + }, + { + "epoch": 0.22435756643971008, + "grad_norm": 2.497159957885742, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.724955677986145, + "num_tokens": 52746554.0, + "step": 2043 + }, + { + "epoch": 0.22446738414232376, + "grad_norm": 2.5371408462524414, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7092947959899902, + "num_tokens": 52769360.0, + "step": 2044 + }, + { + "epoch": 0.2245772018449374, + "grad_norm": 2.144953966140747, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7054743766784668, + "num_tokens": 52796307.0, + "step": 2045 + }, + { + "epoch": 0.22468701954755108, + "grad_norm": 2.274601459503174, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6917457580566406, + "num_tokens": 52821600.0, + "step": 2046 + }, + { + "epoch": 0.22479683725016472, + "grad_norm": 2.159367561340332, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6890446543693542, + "num_tokens": 52848058.0, + "step": 2047 + }, + { + "epoch": 0.2249066549527784, + "grad_norm": 2.3293986320495605, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7112159729003906, + "num_tokens": 52871863.0, + "step": 2048 + }, + { + "epoch": 0.22501647265539204, + "grad_norm": 2.7341480255126953, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7091481685638428, + "num_tokens": 52890303.0, + "step": 2049 + }, + { + "epoch": 0.22512629035800572, + "grad_norm": 2.0530011653900146, + "learning_rate": 1e-06, + "loss": 1.0832, + "mean_token_accuracy": 0.6792862415313721, + "num_tokens": 52922359.0, + "step": 2050 + }, + { + "epoch": 0.22523610806061936, + "grad_norm": 2.3090476989746094, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6912236213684082, + "num_tokens": 52945571.0, + "step": 2051 + }, + { + "epoch": 0.22534592576323303, + "grad_norm": 2.098541259765625, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6860816478729248, + "num_tokens": 52973696.0, + "step": 2052 + }, + { + "epoch": 0.22545574346584668, + "grad_norm": 2.585893154144287, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.683336079120636, + "num_tokens": 52994365.0, + "step": 2053 + }, + { + "epoch": 0.22556556116846035, + "grad_norm": 2.632777690887451, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7025629281997681, + "num_tokens": 53014978.0, + "step": 2054 + }, + { + "epoch": 0.22567537887107403, + "grad_norm": 1.9693759679794312, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7246457934379578, + "num_tokens": 53044134.0, + "step": 2055 + }, + { + "epoch": 0.22578519657368767, + "grad_norm": 2.254446268081665, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6803332567214966, + "num_tokens": 53070296.0, + "step": 2056 + }, + { + "epoch": 0.22589501427630135, + "grad_norm": 2.4145522117614746, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7001037001609802, + "num_tokens": 53091698.0, + "step": 2057 + }, + { + "epoch": 0.226004831978915, + "grad_norm": 2.544766426086426, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6966311931610107, + "num_tokens": 53113760.0, + "step": 2058 + }, + { + "epoch": 0.22611464968152867, + "grad_norm": 2.083613157272339, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6938964128494263, + "num_tokens": 53144729.0, + "step": 2059 + }, + { + "epoch": 0.2262244673841423, + "grad_norm": 2.7338273525238037, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6888638138771057, + "num_tokens": 53165887.0, + "step": 2060 + }, + { + "epoch": 0.226334285086756, + "grad_norm": 2.492577075958252, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7128058671951294, + "num_tokens": 53187239.0, + "step": 2061 + }, + { + "epoch": 0.22644410278936963, + "grad_norm": 2.1527316570281982, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6908633708953857, + "num_tokens": 53217344.0, + "step": 2062 + }, + { + "epoch": 0.2265539204919833, + "grad_norm": 2.159337043762207, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.697551965713501, + "num_tokens": 53245586.0, + "step": 2063 + }, + { + "epoch": 0.22666373819459698, + "grad_norm": 2.1898844242095947, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.688712477684021, + "num_tokens": 53273771.0, + "step": 2064 + }, + { + "epoch": 0.22677355589721063, + "grad_norm": 2.3321011066436768, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.689315676689148, + "num_tokens": 53298844.0, + "step": 2065 + }, + { + "epoch": 0.2268833735998243, + "grad_norm": 2.18461012840271, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7103944420814514, + "num_tokens": 53326136.0, + "step": 2066 + }, + { + "epoch": 0.22699319130243795, + "grad_norm": 2.535871744155884, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.681779682636261, + "num_tokens": 53351032.0, + "step": 2067 + }, + { + "epoch": 0.22710300900505162, + "grad_norm": 2.4747133255004883, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6976009011268616, + "num_tokens": 53375008.0, + "step": 2068 + }, + { + "epoch": 0.22721282670766527, + "grad_norm": 2.11387038230896, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6862179636955261, + "num_tokens": 53403423.0, + "step": 2069 + }, + { + "epoch": 0.22732264441027894, + "grad_norm": 2.293189287185669, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6913794875144958, + "num_tokens": 53429347.0, + "step": 2070 + }, + { + "epoch": 0.22743246211289259, + "grad_norm": 2.483469009399414, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6991929411888123, + "num_tokens": 53448496.0, + "step": 2071 + }, + { + "epoch": 0.22754227981550626, + "grad_norm": 2.207575798034668, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6997825503349304, + "num_tokens": 53475447.0, + "step": 2072 + }, + { + "epoch": 0.22765209751811993, + "grad_norm": 2.29792857170105, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7209329605102539, + "num_tokens": 53498154.0, + "step": 2073 + }, + { + "epoch": 0.22776191522073358, + "grad_norm": 2.1845061779022217, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6941769123077393, + "num_tokens": 53524963.0, + "step": 2074 + }, + { + "epoch": 0.22787173292334725, + "grad_norm": 2.072199583053589, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.715309739112854, + "num_tokens": 53555824.0, + "step": 2075 + }, + { + "epoch": 0.2279815506259609, + "grad_norm": 2.046696662902832, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6944242715835571, + "num_tokens": 53586236.0, + "step": 2076 + }, + { + "epoch": 0.22809136832857457, + "grad_norm": 2.1984293460845947, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6836252212524414, + "num_tokens": 53614859.0, + "step": 2077 + }, + { + "epoch": 0.22820118603118822, + "grad_norm": 2.893906831741333, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7203442454338074, + "num_tokens": 53633945.0, + "step": 2078 + }, + { + "epoch": 0.2283110037338019, + "grad_norm": 2.5426864624023438, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6867513060569763, + "num_tokens": 53655145.0, + "step": 2079 + }, + { + "epoch": 0.22842082143641554, + "grad_norm": 2.444080352783203, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6867314577102661, + "num_tokens": 53677686.0, + "step": 2080 + }, + { + "epoch": 0.2285306391390292, + "grad_norm": 2.351612091064453, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6984797716140747, + "num_tokens": 53701100.0, + "step": 2081 + }, + { + "epoch": 0.2286404568416429, + "grad_norm": 2.009812116622925, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6987684965133667, + "num_tokens": 53733348.0, + "step": 2082 + }, + { + "epoch": 0.22875027454425653, + "grad_norm": 2.3195180892944336, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.697168231010437, + "num_tokens": 53756415.0, + "step": 2083 + }, + { + "epoch": 0.2288600922468702, + "grad_norm": 2.388977289199829, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6971669793128967, + "num_tokens": 53778216.0, + "step": 2084 + }, + { + "epoch": 0.22896990994948385, + "grad_norm": 2.3790485858917236, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7174681425094604, + "num_tokens": 53801951.0, + "step": 2085 + }, + { + "epoch": 0.22907972765209753, + "grad_norm": 2.407957077026367, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7054198980331421, + "num_tokens": 53822712.0, + "step": 2086 + }, + { + "epoch": 0.22918954535471117, + "grad_norm": 2.2297916412353516, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7098484039306641, + "num_tokens": 53848372.0, + "step": 2087 + }, + { + "epoch": 0.22929936305732485, + "grad_norm": 2.374340772628784, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6855788826942444, + "num_tokens": 53872674.0, + "step": 2088 + }, + { + "epoch": 0.2294091807599385, + "grad_norm": 2.528775215148926, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7036033868789673, + "num_tokens": 53895501.0, + "step": 2089 + }, + { + "epoch": 0.22951899846255217, + "grad_norm": 1.9730417728424072, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6950953602790833, + "num_tokens": 53930371.0, + "step": 2090 + }, + { + "epoch": 0.2296288161651658, + "grad_norm": 2.4394657611846924, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7046278119087219, + "num_tokens": 53953983.0, + "step": 2091 + }, + { + "epoch": 0.22973863386777948, + "grad_norm": 2.1870815753936768, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6787817478179932, + "num_tokens": 53982606.0, + "step": 2092 + }, + { + "epoch": 0.22984845157039316, + "grad_norm": 1.9814975261688232, + "learning_rate": 1e-06, + "loss": 1.0985, + "mean_token_accuracy": 0.6746325492858887, + "num_tokens": 54013742.0, + "step": 2093 + }, + { + "epoch": 0.2299582692730068, + "grad_norm": 2.3780689239501953, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7250583171844482, + "num_tokens": 54035030.0, + "step": 2094 + }, + { + "epoch": 0.23006808697562048, + "grad_norm": 2.2532002925872803, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6987147331237793, + "num_tokens": 54061334.0, + "step": 2095 + }, + { + "epoch": 0.23017790467823412, + "grad_norm": 2.7282657623291016, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6928318738937378, + "num_tokens": 54080514.0, + "step": 2096 + }, + { + "epoch": 0.2302877223808478, + "grad_norm": 1.985225796699524, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7367237210273743, + "num_tokens": 54108595.0, + "step": 2097 + }, + { + "epoch": 0.23039754008346144, + "grad_norm": 2.7527754306793213, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7149252891540527, + "num_tokens": 54126507.0, + "step": 2098 + }, + { + "epoch": 0.23050735778607512, + "grad_norm": 2.054661512374878, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7031683921813965, + "num_tokens": 54155201.0, + "step": 2099 + }, + { + "epoch": 0.23061717548868876, + "grad_norm": 2.1511402130126953, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6929278373718262, + "num_tokens": 54185625.0, + "step": 2100 + }, + { + "epoch": 0.23072699319130244, + "grad_norm": 2.405393600463867, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6898794174194336, + "num_tokens": 54210888.0, + "step": 2101 + }, + { + "epoch": 0.2308368108939161, + "grad_norm": 2.133885622024536, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.695745587348938, + "num_tokens": 54240037.0, + "step": 2102 + }, + { + "epoch": 0.23094662859652976, + "grad_norm": 2.6640021800994873, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6947612166404724, + "num_tokens": 54261103.0, + "step": 2103 + }, + { + "epoch": 0.23105644629914343, + "grad_norm": 2.202770709991455, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6874638795852661, + "num_tokens": 54287429.0, + "step": 2104 + }, + { + "epoch": 0.23116626400175708, + "grad_norm": 2.4591963291168213, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6987625360488892, + "num_tokens": 54308282.0, + "step": 2105 + }, + { + "epoch": 0.23127608170437075, + "grad_norm": 2.0864901542663574, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6929702758789062, + "num_tokens": 54335372.0, + "step": 2106 + }, + { + "epoch": 0.2313858994069844, + "grad_norm": 2.2615842819213867, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6910749673843384, + "num_tokens": 54363029.0, + "step": 2107 + }, + { + "epoch": 0.23149571710959807, + "grad_norm": 2.1720802783966064, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6784919500350952, + "num_tokens": 54394085.0, + "step": 2108 + }, + { + "epoch": 0.23160553481221172, + "grad_norm": 2.4771454334259033, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.689974308013916, + "num_tokens": 54414397.0, + "step": 2109 + }, + { + "epoch": 0.2317153525148254, + "grad_norm": 2.2647390365600586, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7108364105224609, + "num_tokens": 54437468.0, + "step": 2110 + }, + { + "epoch": 0.23182517021743906, + "grad_norm": 2.536881923675537, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7019774913787842, + "num_tokens": 54458441.0, + "step": 2111 + }, + { + "epoch": 0.2319349879200527, + "grad_norm": 2.435492992401123, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7172410488128662, + "num_tokens": 54483055.0, + "step": 2112 + }, + { + "epoch": 0.23204480562266638, + "grad_norm": 2.402845859527588, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7032588720321655, + "num_tokens": 54505757.0, + "step": 2113 + }, + { + "epoch": 0.23215462332528003, + "grad_norm": 2.492535352706909, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6921034455299377, + "num_tokens": 54528203.0, + "step": 2114 + }, + { + "epoch": 0.2322644410278937, + "grad_norm": 2.3842742443084717, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7003737688064575, + "num_tokens": 54553515.0, + "step": 2115 + }, + { + "epoch": 0.23237425873050735, + "grad_norm": 2.2264528274536133, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6860975027084351, + "num_tokens": 54580884.0, + "step": 2116 + }, + { + "epoch": 0.23248407643312102, + "grad_norm": 2.08939790725708, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7089385986328125, + "num_tokens": 54609375.0, + "step": 2117 + }, + { + "epoch": 0.23259389413573467, + "grad_norm": 2.386341094970703, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6925106644630432, + "num_tokens": 54634531.0, + "step": 2118 + }, + { + "epoch": 0.23270371183834834, + "grad_norm": 2.3224093914031982, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6824390292167664, + "num_tokens": 54661124.0, + "step": 2119 + }, + { + "epoch": 0.23281352954096202, + "grad_norm": 2.4722235202789307, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6793211698532104, + "num_tokens": 54683693.0, + "step": 2120 + }, + { + "epoch": 0.23292334724357566, + "grad_norm": 2.2713608741760254, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7306059002876282, + "num_tokens": 54705626.0, + "step": 2121 + }, + { + "epoch": 0.23303316494618934, + "grad_norm": 2.331852912902832, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6993879079818726, + "num_tokens": 54729813.0, + "step": 2122 + }, + { + "epoch": 0.23314298264880298, + "grad_norm": 2.4873106479644775, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6953426599502563, + "num_tokens": 54753997.0, + "step": 2123 + }, + { + "epoch": 0.23325280035141666, + "grad_norm": 2.533019781112671, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6923280358314514, + "num_tokens": 54773691.0, + "step": 2124 + }, + { + "epoch": 0.2333626180540303, + "grad_norm": 2.3797402381896973, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.694779098033905, + "num_tokens": 54798426.0, + "step": 2125 + }, + { + "epoch": 0.23347243575664398, + "grad_norm": 2.181007146835327, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7006050944328308, + "num_tokens": 54825175.0, + "step": 2126 + }, + { + "epoch": 0.23358225345925762, + "grad_norm": 2.5835955142974854, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6940543055534363, + "num_tokens": 54846706.0, + "step": 2127 + }, + { + "epoch": 0.2336920711618713, + "grad_norm": 2.1287617683410645, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6900089979171753, + "num_tokens": 54876788.0, + "step": 2128 + }, + { + "epoch": 0.23380188886448494, + "grad_norm": 2.0738635063171387, + "learning_rate": 1e-06, + "loss": 1.1339, + "mean_token_accuracy": 0.6612577438354492, + "num_tokens": 54911047.0, + "step": 2129 + }, + { + "epoch": 0.23391170656709862, + "grad_norm": 2.0560288429260254, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7002678513526917, + "num_tokens": 54943308.0, + "step": 2130 + }, + { + "epoch": 0.2340215242697123, + "grad_norm": 2.3399407863616943, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7078408598899841, + "num_tokens": 54965982.0, + "step": 2131 + }, + { + "epoch": 0.23413134197232593, + "grad_norm": 2.2523226737976074, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.698694109916687, + "num_tokens": 54991467.0, + "step": 2132 + }, + { + "epoch": 0.2342411596749396, + "grad_norm": 2.234097719192505, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.704740047454834, + "num_tokens": 55017213.0, + "step": 2133 + }, + { + "epoch": 0.23435097737755325, + "grad_norm": 2.2408883571624756, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6876227259635925, + "num_tokens": 55044550.0, + "step": 2134 + }, + { + "epoch": 0.23446079508016693, + "grad_norm": 2.4626002311706543, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6996466517448425, + "num_tokens": 55067152.0, + "step": 2135 + }, + { + "epoch": 0.23457061278278057, + "grad_norm": 2.8225302696228027, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7054746150970459, + "num_tokens": 55083505.0, + "step": 2136 + }, + { + "epoch": 0.23468043048539425, + "grad_norm": 2.2130649089813232, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6874527931213379, + "num_tokens": 55109743.0, + "step": 2137 + }, + { + "epoch": 0.2347902481880079, + "grad_norm": 2.4581072330474854, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6783542037010193, + "num_tokens": 55133697.0, + "step": 2138 + }, + { + "epoch": 0.23490006589062157, + "grad_norm": 2.279792308807373, + "learning_rate": 1e-06, + "loss": 1.0789, + "mean_token_accuracy": 0.6766461730003357, + "num_tokens": 55161188.0, + "step": 2139 + }, + { + "epoch": 0.23500988359323524, + "grad_norm": 2.406154155731201, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7231480479240417, + "num_tokens": 55182028.0, + "step": 2140 + }, + { + "epoch": 0.2351197012958489, + "grad_norm": 2.085318088531494, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6996519565582275, + "num_tokens": 55212340.0, + "step": 2141 + }, + { + "epoch": 0.23522951899846256, + "grad_norm": 2.2395825386047363, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6810970306396484, + "num_tokens": 55239041.0, + "step": 2142 + }, + { + "epoch": 0.2353393367010762, + "grad_norm": 1.9287517070770264, + "learning_rate": 1e-06, + "loss": 1.0995, + "mean_token_accuracy": 0.6783239841461182, + "num_tokens": 55273480.0, + "step": 2143 + }, + { + "epoch": 0.23544915440368988, + "grad_norm": 2.053389549255371, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7195661067962646, + "num_tokens": 55303213.0, + "step": 2144 + }, + { + "epoch": 0.23555897210630353, + "grad_norm": 2.2448978424072266, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6909536719322205, + "num_tokens": 55328901.0, + "step": 2145 + }, + { + "epoch": 0.2356687898089172, + "grad_norm": 2.3429360389709473, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7029212117195129, + "num_tokens": 55354067.0, + "step": 2146 + }, + { + "epoch": 0.23577860751153085, + "grad_norm": 2.4721004962921143, + "learning_rate": 1e-06, + "loss": 1.1279, + "mean_token_accuracy": 0.6668986678123474, + "num_tokens": 55381381.0, + "step": 2147 + }, + { + "epoch": 0.23588842521414452, + "grad_norm": 1.9376893043518066, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7103787660598755, + "num_tokens": 55413293.0, + "step": 2148 + }, + { + "epoch": 0.2359982429167582, + "grad_norm": 2.4954090118408203, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7210372686386108, + "num_tokens": 55433129.0, + "step": 2149 + }, + { + "epoch": 0.23610806061937184, + "grad_norm": 2.272759437561035, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6950284242630005, + "num_tokens": 55458205.0, + "step": 2150 + }, + { + "epoch": 0.2362178783219855, + "grad_norm": 2.247143268585205, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6998485326766968, + "num_tokens": 55483978.0, + "step": 2151 + }, + { + "epoch": 0.23632769602459916, + "grad_norm": 2.218571662902832, + "learning_rate": 1e-06, + "loss": 1.0965, + "mean_token_accuracy": 0.6815308332443237, + "num_tokens": 55513758.0, + "step": 2152 + }, + { + "epoch": 0.23643751372721283, + "grad_norm": 2.2302188873291016, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7185088992118835, + "num_tokens": 55540083.0, + "step": 2153 + }, + { + "epoch": 0.23654733142982648, + "grad_norm": 2.296454429626465, + "learning_rate": 1e-06, + "loss": 1.101, + "mean_token_accuracy": 0.6747442483901978, + "num_tokens": 55565841.0, + "step": 2154 + }, + { + "epoch": 0.23665714913244015, + "grad_norm": 2.015226125717163, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7215633392333984, + "num_tokens": 55596305.0, + "step": 2155 + }, + { + "epoch": 0.2367669668350538, + "grad_norm": 2.3924317359924316, + "learning_rate": 1e-06, + "loss": 1.0791, + "mean_token_accuracy": 0.6818060278892517, + "num_tokens": 55621747.0, + "step": 2156 + }, + { + "epoch": 0.23687678453766747, + "grad_norm": 2.573497772216797, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7133760452270508, + "num_tokens": 55641471.0, + "step": 2157 + }, + { + "epoch": 0.23698660224028115, + "grad_norm": 2.2581257820129395, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6986014246940613, + "num_tokens": 55666749.0, + "step": 2158 + }, + { + "epoch": 0.2370964199428948, + "grad_norm": 2.324159860610962, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6993154287338257, + "num_tokens": 55690585.0, + "step": 2159 + }, + { + "epoch": 0.23720623764550847, + "grad_norm": 2.4581241607666016, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7041319012641907, + "num_tokens": 55714381.0, + "step": 2160 + }, + { + "epoch": 0.2373160553481221, + "grad_norm": 2.440370798110962, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7097072601318359, + "num_tokens": 55735929.0, + "step": 2161 + }, + { + "epoch": 0.2374258730507358, + "grad_norm": 2.2006049156188965, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7065392732620239, + "num_tokens": 55762786.0, + "step": 2162 + }, + { + "epoch": 0.23753569075334943, + "grad_norm": 2.455059766769409, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7110642790794373, + "num_tokens": 55785466.0, + "step": 2163 + }, + { + "epoch": 0.2376455084559631, + "grad_norm": 2.4293622970581055, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6941065788269043, + "num_tokens": 55808474.0, + "step": 2164 + }, + { + "epoch": 0.23775532615857675, + "grad_norm": 2.001474142074585, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7044382095336914, + "num_tokens": 55839259.0, + "step": 2165 + }, + { + "epoch": 0.23786514386119043, + "grad_norm": 2.228900909423828, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6947470903396606, + "num_tokens": 55865019.0, + "step": 2166 + }, + { + "epoch": 0.23797496156380407, + "grad_norm": 2.2036685943603516, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6883782148361206, + "num_tokens": 55893382.0, + "step": 2167 + }, + { + "epoch": 0.23808477926641775, + "grad_norm": 2.1396756172180176, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7055721282958984, + "num_tokens": 55921074.0, + "step": 2168 + }, + { + "epoch": 0.23819459696903142, + "grad_norm": 2.237968921661377, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6866934299468994, + "num_tokens": 55946833.0, + "step": 2169 + }, + { + "epoch": 0.23830441467164507, + "grad_norm": 2.2369189262390137, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7107365727424622, + "num_tokens": 55973218.0, + "step": 2170 + }, + { + "epoch": 0.23841423237425874, + "grad_norm": 2.3073253631591797, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6936683058738708, + "num_tokens": 55999291.0, + "step": 2171 + }, + { + "epoch": 0.23852405007687238, + "grad_norm": 2.4062089920043945, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7112094759941101, + "num_tokens": 56021326.0, + "step": 2172 + }, + { + "epoch": 0.23863386777948606, + "grad_norm": 2.121192216873169, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.69252610206604, + "num_tokens": 56049050.0, + "step": 2173 + }, + { + "epoch": 0.2387436854820997, + "grad_norm": 1.9666112661361694, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6802531480789185, + "num_tokens": 56082772.0, + "step": 2174 + }, + { + "epoch": 0.23885350318471338, + "grad_norm": 2.2199060916900635, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.7004728317260742, + "num_tokens": 56109820.0, + "step": 2175 + }, + { + "epoch": 0.23896332088732702, + "grad_norm": 2.025258779525757, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6807397603988647, + "num_tokens": 56142878.0, + "step": 2176 + }, + { + "epoch": 0.2390731385899407, + "grad_norm": 2.554342746734619, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7092458009719849, + "num_tokens": 56161722.0, + "step": 2177 + }, + { + "epoch": 0.23918295629255437, + "grad_norm": 2.4836883544921875, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.697917103767395, + "num_tokens": 56183413.0, + "step": 2178 + }, + { + "epoch": 0.23929277399516802, + "grad_norm": 2.026181221008301, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6793172359466553, + "num_tokens": 56214440.0, + "step": 2179 + }, + { + "epoch": 0.2394025916977817, + "grad_norm": 2.3717544078826904, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6907700300216675, + "num_tokens": 56239916.0, + "step": 2180 + }, + { + "epoch": 0.23951240940039534, + "grad_norm": 2.346646308898926, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6999959349632263, + "num_tokens": 56264202.0, + "step": 2181 + }, + { + "epoch": 0.239622227103009, + "grad_norm": 2.2912800312042236, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7011426687240601, + "num_tokens": 56288802.0, + "step": 2182 + }, + { + "epoch": 0.23973204480562266, + "grad_norm": 2.2599875926971436, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.6847173571586609, + "num_tokens": 56315109.0, + "step": 2183 + }, + { + "epoch": 0.23984186250823633, + "grad_norm": 2.112502098083496, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6986292004585266, + "num_tokens": 56340316.0, + "step": 2184 + }, + { + "epoch": 0.23995168021084998, + "grad_norm": 2.0898592472076416, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.700798749923706, + "num_tokens": 56368183.0, + "step": 2185 + }, + { + "epoch": 0.24006149791346365, + "grad_norm": 2.347330093383789, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7166364789009094, + "num_tokens": 56392443.0, + "step": 2186 + }, + { + "epoch": 0.24017131561607732, + "grad_norm": 2.4316060543060303, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6941742897033691, + "num_tokens": 56414488.0, + "step": 2187 + }, + { + "epoch": 0.24028113331869097, + "grad_norm": 1.9174076318740845, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6902613639831543, + "num_tokens": 56449584.0, + "step": 2188 + }, + { + "epoch": 0.24039095102130464, + "grad_norm": 2.146296977996826, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6990582346916199, + "num_tokens": 56477720.0, + "step": 2189 + }, + { + "epoch": 0.2405007687239183, + "grad_norm": 2.275601387023926, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.690841794013977, + "num_tokens": 56503137.0, + "step": 2190 + }, + { + "epoch": 0.24061058642653196, + "grad_norm": 2.1096949577331543, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7039037942886353, + "num_tokens": 56532195.0, + "step": 2191 + }, + { + "epoch": 0.2407204041291456, + "grad_norm": 2.4023261070251465, + "learning_rate": 1e-06, + "loss": 1.071, + "mean_token_accuracy": 0.6792422533035278, + "num_tokens": 56556030.0, + "step": 2192 + }, + { + "epoch": 0.24083022183175928, + "grad_norm": 2.2808518409729004, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7127516269683838, + "num_tokens": 56580286.0, + "step": 2193 + }, + { + "epoch": 0.24094003953437293, + "grad_norm": 2.3984415531158447, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.709113597869873, + "num_tokens": 56605385.0, + "step": 2194 + }, + { + "epoch": 0.2410498572369866, + "grad_norm": 2.3787567615509033, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7064076662063599, + "num_tokens": 56628373.0, + "step": 2195 + }, + { + "epoch": 0.24115967493960028, + "grad_norm": 2.1940295696258545, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6794718503952026, + "num_tokens": 56657143.0, + "step": 2196 + }, + { + "epoch": 0.24126949264221392, + "grad_norm": 2.005223274230957, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6912716627120972, + "num_tokens": 56688606.0, + "step": 2197 + }, + { + "epoch": 0.2413793103448276, + "grad_norm": 2.1816861629486084, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6908648014068604, + "num_tokens": 56716269.0, + "step": 2198 + }, + { + "epoch": 0.24148912804744124, + "grad_norm": 2.2447402477264404, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7006953358650208, + "num_tokens": 56741468.0, + "step": 2199 + }, + { + "epoch": 0.24159894575005492, + "grad_norm": 2.6220226287841797, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6910285949707031, + "num_tokens": 56763911.0, + "step": 2200 + }, + { + "epoch": 0.24170876345266856, + "grad_norm": 2.221423625946045, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7265744805335999, + "num_tokens": 56788617.0, + "step": 2201 + }, + { + "epoch": 0.24181858115528224, + "grad_norm": 2.04669189453125, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6846016645431519, + "num_tokens": 56817769.0, + "step": 2202 + }, + { + "epoch": 0.24192839885789588, + "grad_norm": 2.082106590270996, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6856358051300049, + "num_tokens": 56846220.0, + "step": 2203 + }, + { + "epoch": 0.24203821656050956, + "grad_norm": 2.1155054569244385, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6777483820915222, + "num_tokens": 56875577.0, + "step": 2204 + }, + { + "epoch": 0.2421480342631232, + "grad_norm": 2.2977046966552734, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.693943977355957, + "num_tokens": 56901161.0, + "step": 2205 + }, + { + "epoch": 0.24225785196573688, + "grad_norm": 2.5281307697296143, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6957001686096191, + "num_tokens": 56923156.0, + "step": 2206 + }, + { + "epoch": 0.24236766966835055, + "grad_norm": 2.3395681381225586, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6724457740783691, + "num_tokens": 56948396.0, + "step": 2207 + }, + { + "epoch": 0.2424774873709642, + "grad_norm": 2.0993311405181885, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7160288095474243, + "num_tokens": 56977122.0, + "step": 2208 + }, + { + "epoch": 0.24258730507357787, + "grad_norm": 2.0470898151397705, + "learning_rate": 1e-06, + "loss": 1.087, + "mean_token_accuracy": 0.6803994178771973, + "num_tokens": 57009670.0, + "step": 2209 + }, + { + "epoch": 0.24269712277619152, + "grad_norm": 2.5035552978515625, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6932633519172668, + "num_tokens": 57031672.0, + "step": 2210 + }, + { + "epoch": 0.2428069404788052, + "grad_norm": 2.212858200073242, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7058658599853516, + "num_tokens": 57057890.0, + "step": 2211 + }, + { + "epoch": 0.24291675818141883, + "grad_norm": 2.209263801574707, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7078137397766113, + "num_tokens": 57083750.0, + "step": 2212 + }, + { + "epoch": 0.2430265758840325, + "grad_norm": 2.2928662300109863, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6882067918777466, + "num_tokens": 57108567.0, + "step": 2213 + }, + { + "epoch": 0.24313639358664615, + "grad_norm": 2.098567485809326, + "learning_rate": 1e-06, + "loss": 1.1055, + "mean_token_accuracy": 0.670263946056366, + "num_tokens": 57138590.0, + "step": 2214 + }, + { + "epoch": 0.24324621128925983, + "grad_norm": 2.1253957748413086, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7029757499694824, + "num_tokens": 57167016.0, + "step": 2215 + }, + { + "epoch": 0.2433560289918735, + "grad_norm": 2.0359349250793457, + "learning_rate": 1e-06, + "loss": 1.1299, + "mean_token_accuracy": 0.6744257807731628, + "num_tokens": 57199323.0, + "step": 2216 + }, + { + "epoch": 0.24346584669448715, + "grad_norm": 2.1602022647857666, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7088528871536255, + "num_tokens": 57227445.0, + "step": 2217 + }, + { + "epoch": 0.24357566439710082, + "grad_norm": 2.378152370452881, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6920855641365051, + "num_tokens": 57251708.0, + "step": 2218 + }, + { + "epoch": 0.24368548209971447, + "grad_norm": 2.5567941665649414, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6760437488555908, + "num_tokens": 57273789.0, + "step": 2219 + }, + { + "epoch": 0.24379529980232814, + "grad_norm": 2.11203932762146, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6902851462364197, + "num_tokens": 57304287.0, + "step": 2220 + }, + { + "epoch": 0.2439051175049418, + "grad_norm": 2.3849682807922363, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7148827314376831, + "num_tokens": 57326761.0, + "step": 2221 + }, + { + "epoch": 0.24401493520755546, + "grad_norm": 2.082080602645874, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.686789870262146, + "num_tokens": 57356752.0, + "step": 2222 + }, + { + "epoch": 0.2441247529101691, + "grad_norm": 2.2965359687805176, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7099138498306274, + "num_tokens": 57379780.0, + "step": 2223 + }, + { + "epoch": 0.24423457061278278, + "grad_norm": 2.023488998413086, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6982142925262451, + "num_tokens": 57409012.0, + "step": 2224 + }, + { + "epoch": 0.24434438831539645, + "grad_norm": 2.045924186706543, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7004282474517822, + "num_tokens": 57438395.0, + "step": 2225 + }, + { + "epoch": 0.2444542060180101, + "grad_norm": 2.3314085006713867, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.70750492811203, + "num_tokens": 57463316.0, + "step": 2226 + }, + { + "epoch": 0.24456402372062377, + "grad_norm": 2.3498003482818604, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7027151584625244, + "num_tokens": 57486016.0, + "step": 2227 + }, + { + "epoch": 0.24467384142323742, + "grad_norm": 2.4089367389678955, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7171499729156494, + "num_tokens": 57506926.0, + "step": 2228 + }, + { + "epoch": 0.2447836591258511, + "grad_norm": 2.2398793697357178, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6841442584991455, + "num_tokens": 57535875.0, + "step": 2229 + }, + { + "epoch": 0.24489347682846474, + "grad_norm": 2.2792513370513916, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6980006694793701, + "num_tokens": 57559609.0, + "step": 2230 + }, + { + "epoch": 0.2450032945310784, + "grad_norm": 2.0875444412231445, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.6826435327529907, + "num_tokens": 57590705.0, + "step": 2231 + }, + { + "epoch": 0.24511311223369206, + "grad_norm": 2.3267173767089844, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6828967332839966, + "num_tokens": 57615545.0, + "step": 2232 + }, + { + "epoch": 0.24522292993630573, + "grad_norm": 2.321035146713257, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6917389631271362, + "num_tokens": 57641140.0, + "step": 2233 + }, + { + "epoch": 0.2453327476389194, + "grad_norm": 2.4855687618255615, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.6876873970031738, + "num_tokens": 57664153.0, + "step": 2234 + }, + { + "epoch": 0.24544256534153305, + "grad_norm": 2.612114667892456, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7091518640518188, + "num_tokens": 57683299.0, + "step": 2235 + }, + { + "epoch": 0.24555238304414673, + "grad_norm": 2.3515729904174805, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7120083570480347, + "num_tokens": 57706969.0, + "step": 2236 + }, + { + "epoch": 0.24566220074676037, + "grad_norm": 2.18097186088562, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.695888876914978, + "num_tokens": 57734621.0, + "step": 2237 + }, + { + "epoch": 0.24577201844937405, + "grad_norm": 2.134537696838379, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6774408221244812, + "num_tokens": 57762732.0, + "step": 2238 + }, + { + "epoch": 0.2458818361519877, + "grad_norm": 2.1861190795898438, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6976512670516968, + "num_tokens": 57790963.0, + "step": 2239 + }, + { + "epoch": 0.24599165385460137, + "grad_norm": 2.1913726329803467, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6888061761856079, + "num_tokens": 57820246.0, + "step": 2240 + }, + { + "epoch": 0.246101471557215, + "grad_norm": 2.3360257148742676, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6883586645126343, + "num_tokens": 57843504.0, + "step": 2241 + }, + { + "epoch": 0.2462112892598287, + "grad_norm": 2.4718246459960938, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7171865701675415, + "num_tokens": 57862607.0, + "step": 2242 + }, + { + "epoch": 0.24632110696244233, + "grad_norm": 2.2376973628997803, + "learning_rate": 1e-06, + "loss": 1.0897, + "mean_token_accuracy": 0.6737065315246582, + "num_tokens": 57889000.0, + "step": 2243 + }, + { + "epoch": 0.246430924665056, + "grad_norm": 2.1764020919799805, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7048946619033813, + "num_tokens": 57919218.0, + "step": 2244 + }, + { + "epoch": 0.24654074236766968, + "grad_norm": 2.285820245742798, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7144513130187988, + "num_tokens": 57944171.0, + "step": 2245 + }, + { + "epoch": 0.24665056007028333, + "grad_norm": 2.3151001930236816, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7152410745620728, + "num_tokens": 57967655.0, + "step": 2246 + }, + { + "epoch": 0.246760377772897, + "grad_norm": 1.92287278175354, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7197055816650391, + "num_tokens": 58000043.0, + "step": 2247 + }, + { + "epoch": 0.24687019547551065, + "grad_norm": 2.4972894191741943, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7038909196853638, + "num_tokens": 58022783.0, + "step": 2248 + }, + { + "epoch": 0.24698001317812432, + "grad_norm": 2.143730878829956, + "learning_rate": 1e-06, + "loss": 1.0394, + "mean_token_accuracy": 0.6852186918258667, + "num_tokens": 58049858.0, + "step": 2249 + }, + { + "epoch": 0.24708983088073797, + "grad_norm": 2.0631766319274902, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6767597198486328, + "num_tokens": 58080223.0, + "step": 2250 + }, + { + "epoch": 0.24719964858335164, + "grad_norm": 2.1843838691711426, + "learning_rate": 1e-06, + "loss": 1.0991, + "mean_token_accuracy": 0.6775201559066772, + "num_tokens": 58109944.0, + "step": 2251 + }, + { + "epoch": 0.24730946628596528, + "grad_norm": 2.1912636756896973, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7109441757202148, + "num_tokens": 58137169.0, + "step": 2252 + }, + { + "epoch": 0.24741928398857896, + "grad_norm": 2.322418689727783, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7184634208679199, + "num_tokens": 58161563.0, + "step": 2253 + }, + { + "epoch": 0.24752910169119263, + "grad_norm": 2.2865540981292725, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7100552916526794, + "num_tokens": 58185332.0, + "step": 2254 + }, + { + "epoch": 0.24763891939380628, + "grad_norm": 2.0962512493133545, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.702654242515564, + "num_tokens": 58212801.0, + "step": 2255 + }, + { + "epoch": 0.24774873709641995, + "grad_norm": 2.203570604324341, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6966158151626587, + "num_tokens": 58238997.0, + "step": 2256 + }, + { + "epoch": 0.2478585547990336, + "grad_norm": 1.9067409038543701, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6834686994552612, + "num_tokens": 58274984.0, + "step": 2257 + }, + { + "epoch": 0.24796837250164727, + "grad_norm": 2.1296679973602295, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6964091062545776, + "num_tokens": 58300934.0, + "step": 2258 + }, + { + "epoch": 0.24807819020426092, + "grad_norm": 2.2378900051116943, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6731604933738708, + "num_tokens": 58327152.0, + "step": 2259 + }, + { + "epoch": 0.2481880079068746, + "grad_norm": 2.1292991638183594, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.695470929145813, + "num_tokens": 58354560.0, + "step": 2260 + }, + { + "epoch": 0.24829782560948824, + "grad_norm": 2.150195598602295, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7061628103256226, + "num_tokens": 58380881.0, + "step": 2261 + }, + { + "epoch": 0.2484076433121019, + "grad_norm": 2.2700467109680176, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6983535289764404, + "num_tokens": 58405780.0, + "step": 2262 + }, + { + "epoch": 0.24851746101471558, + "grad_norm": 2.300940752029419, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7108293175697327, + "num_tokens": 58429036.0, + "step": 2263 + }, + { + "epoch": 0.24862727871732923, + "grad_norm": 2.085977077484131, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6824591755867004, + "num_tokens": 58460849.0, + "step": 2264 + }, + { + "epoch": 0.2487370964199429, + "grad_norm": 2.397169589996338, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6892267465591431, + "num_tokens": 58485428.0, + "step": 2265 + }, + { + "epoch": 0.24884691412255655, + "grad_norm": 2.5283756256103516, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.696410059928894, + "num_tokens": 58505691.0, + "step": 2266 + }, + { + "epoch": 0.24895673182517022, + "grad_norm": 2.48311448097229, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6944255232810974, + "num_tokens": 58527503.0, + "step": 2267 + }, + { + "epoch": 0.24906654952778387, + "grad_norm": 2.211005210876465, + "learning_rate": 1e-06, + "loss": 1.1084, + "mean_token_accuracy": 0.6760988831520081, + "num_tokens": 58555158.0, + "step": 2268 + }, + { + "epoch": 0.24917636723039754, + "grad_norm": 2.2113518714904785, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6996399164199829, + "num_tokens": 58579793.0, + "step": 2269 + }, + { + "epoch": 0.2492861849330112, + "grad_norm": 2.497539520263672, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7097227573394775, + "num_tokens": 58601113.0, + "step": 2270 + }, + { + "epoch": 0.24939600263562486, + "grad_norm": 2.0872721672058105, + "learning_rate": 1e-06, + "loss": 1.1062, + "mean_token_accuracy": 0.6725250482559204, + "num_tokens": 58632358.0, + "step": 2271 + }, + { + "epoch": 0.24950582033823854, + "grad_norm": 2.0546438694000244, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6882911920547485, + "num_tokens": 58661638.0, + "step": 2272 + }, + { + "epoch": 0.24961563804085218, + "grad_norm": 2.131930351257324, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7159479856491089, + "num_tokens": 58688276.0, + "step": 2273 + }, + { + "epoch": 0.24972545574346586, + "grad_norm": 2.272918939590454, + "learning_rate": 1e-06, + "loss": 1.1183, + "mean_token_accuracy": 0.6841590404510498, + "num_tokens": 58714149.0, + "step": 2274 + }, + { + "epoch": 0.2498352734460795, + "grad_norm": 2.4028525352478027, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6779588460922241, + "num_tokens": 58737648.0, + "step": 2275 + }, + { + "epoch": 0.24994509114869318, + "grad_norm": 2.380237102508545, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6798262596130371, + "num_tokens": 58762534.0, + "step": 2276 + }, + { + "epoch": 0.2500549088513068, + "grad_norm": 2.217552900314331, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6933132410049438, + "num_tokens": 58788512.0, + "step": 2277 + }, + { + "epoch": 0.25016472655392047, + "grad_norm": 2.1215760707855225, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7246525287628174, + "num_tokens": 58814572.0, + "step": 2278 + }, + { + "epoch": 0.25027454425653417, + "grad_norm": 2.344266891479492, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7144263982772827, + "num_tokens": 58836353.0, + "step": 2279 + }, + { + "epoch": 0.2503843619591478, + "grad_norm": 2.2394165992736816, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6929589509963989, + "num_tokens": 58862147.0, + "step": 2280 + }, + { + "epoch": 0.25049417966176146, + "grad_norm": 2.2902400493621826, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6982332468032837, + "num_tokens": 58887185.0, + "step": 2281 + }, + { + "epoch": 0.25060399736437516, + "grad_norm": 2.3965859413146973, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7094771862030029, + "num_tokens": 58910065.0, + "step": 2282 + }, + { + "epoch": 0.2507138150669888, + "grad_norm": 2.4505856037139893, + "learning_rate": 1e-06, + "loss": 1.1124, + "mean_token_accuracy": 0.667482852935791, + "num_tokens": 58933576.0, + "step": 2283 + }, + { + "epoch": 0.25082363276960246, + "grad_norm": 2.1674511432647705, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7052358984947205, + "num_tokens": 58960321.0, + "step": 2284 + }, + { + "epoch": 0.2509334504722161, + "grad_norm": 2.1836330890655518, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7191393971443176, + "num_tokens": 58986689.0, + "step": 2285 + }, + { + "epoch": 0.2510432681748298, + "grad_norm": 2.22733211517334, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7046520113945007, + "num_tokens": 59010145.0, + "step": 2286 + }, + { + "epoch": 0.25115308587744345, + "grad_norm": 2.192948341369629, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7090297937393188, + "num_tokens": 59035698.0, + "step": 2287 + }, + { + "epoch": 0.2512629035800571, + "grad_norm": 2.201167106628418, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6851138472557068, + "num_tokens": 59063768.0, + "step": 2288 + }, + { + "epoch": 0.25137272128267074, + "grad_norm": 1.9977139234542847, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6940560340881348, + "num_tokens": 59095510.0, + "step": 2289 + }, + { + "epoch": 0.25148253898528444, + "grad_norm": 2.214761257171631, + "learning_rate": 1e-06, + "loss": 1.0983, + "mean_token_accuracy": 0.6757544875144958, + "num_tokens": 59122545.0, + "step": 2290 + }, + { + "epoch": 0.2515923566878981, + "grad_norm": 2.231311321258545, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6839610934257507, + "num_tokens": 59148781.0, + "step": 2291 + }, + { + "epoch": 0.25170217439051173, + "grad_norm": 2.319638252258301, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7043545246124268, + "num_tokens": 59171520.0, + "step": 2292 + }, + { + "epoch": 0.25181199209312544, + "grad_norm": 2.430856943130493, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7143895626068115, + "num_tokens": 59192308.0, + "step": 2293 + }, + { + "epoch": 0.2519218097957391, + "grad_norm": 2.4764516353607178, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.693181574344635, + "num_tokens": 59214439.0, + "step": 2294 + }, + { + "epoch": 0.25203162749835273, + "grad_norm": 2.5136570930480957, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7031776905059814, + "num_tokens": 59236626.0, + "step": 2295 + }, + { + "epoch": 0.2521414452009664, + "grad_norm": 2.221391439437866, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6986552476882935, + "num_tokens": 59263541.0, + "step": 2296 + }, + { + "epoch": 0.2522512629035801, + "grad_norm": 2.324280261993408, + "learning_rate": 1e-06, + "loss": 1.1151, + "mean_token_accuracy": 0.6687246561050415, + "num_tokens": 59292176.0, + "step": 2297 + }, + { + "epoch": 0.2523610806061937, + "grad_norm": 2.0528135299682617, + "learning_rate": 1e-06, + "loss": 1.1259, + "mean_token_accuracy": 0.6630971431732178, + "num_tokens": 59323515.0, + "step": 2298 + }, + { + "epoch": 0.25247089830880737, + "grad_norm": 2.1431453227996826, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6872658729553223, + "num_tokens": 59351742.0, + "step": 2299 + }, + { + "epoch": 0.252580716011421, + "grad_norm": 2.149169683456421, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6895473003387451, + "num_tokens": 59380879.0, + "step": 2300 + }, + { + "epoch": 0.2526905337140347, + "grad_norm": 2.144904375076294, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6890963912010193, + "num_tokens": 59407943.0, + "step": 2301 + }, + { + "epoch": 0.25280035141664836, + "grad_norm": 2.089691400527954, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7103097438812256, + "num_tokens": 59435392.0, + "step": 2302 + }, + { + "epoch": 0.252910169119262, + "grad_norm": 2.2761569023132324, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7318888306617737, + "num_tokens": 59460009.0, + "step": 2303 + }, + { + "epoch": 0.2530199868218757, + "grad_norm": 1.9947481155395508, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7025234699249268, + "num_tokens": 59490423.0, + "step": 2304 + }, + { + "epoch": 0.25312980452448935, + "grad_norm": 2.1473560333251953, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7009248733520508, + "num_tokens": 59515653.0, + "step": 2305 + }, + { + "epoch": 0.253239622227103, + "grad_norm": 2.211273193359375, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6781065464019775, + "num_tokens": 59543488.0, + "step": 2306 + }, + { + "epoch": 0.25334943992971665, + "grad_norm": 2.3503968715667725, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6970388889312744, + "num_tokens": 59567208.0, + "step": 2307 + }, + { + "epoch": 0.25345925763233035, + "grad_norm": 2.2433364391326904, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6970237493515015, + "num_tokens": 59595157.0, + "step": 2308 + }, + { + "epoch": 0.253569075334944, + "grad_norm": 2.1443402767181396, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6985448002815247, + "num_tokens": 59623159.0, + "step": 2309 + }, + { + "epoch": 0.25367889303755764, + "grad_norm": 1.9834908246994019, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7082631587982178, + "num_tokens": 59652367.0, + "step": 2310 + }, + { + "epoch": 0.25378871074017134, + "grad_norm": 2.1715641021728516, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.7032781839370728, + "num_tokens": 59679902.0, + "step": 2311 + }, + { + "epoch": 0.253898528442785, + "grad_norm": 2.1083643436431885, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7099155187606812, + "num_tokens": 59706019.0, + "step": 2312 + }, + { + "epoch": 0.25400834614539863, + "grad_norm": 2.0020389556884766, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6961386799812317, + "num_tokens": 59736777.0, + "step": 2313 + }, + { + "epoch": 0.2541181638480123, + "grad_norm": 2.3999130725860596, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.713898777961731, + "num_tokens": 59759219.0, + "step": 2314 + }, + { + "epoch": 0.254227981550626, + "grad_norm": 2.2080090045928955, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6910272836685181, + "num_tokens": 59785176.0, + "step": 2315 + }, + { + "epoch": 0.2543377992532396, + "grad_norm": 2.336700677871704, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.701038658618927, + "num_tokens": 59809687.0, + "step": 2316 + }, + { + "epoch": 0.2544476169558533, + "grad_norm": 2.322009801864624, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6916449666023254, + "num_tokens": 59833849.0, + "step": 2317 + }, + { + "epoch": 0.2545574346584669, + "grad_norm": 2.259171485900879, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6980765461921692, + "num_tokens": 59858266.0, + "step": 2318 + }, + { + "epoch": 0.2546672523610806, + "grad_norm": 2.4547863006591797, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7211973071098328, + "num_tokens": 59879537.0, + "step": 2319 + }, + { + "epoch": 0.25477707006369427, + "grad_norm": 2.1771838665008545, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.7018054723739624, + "num_tokens": 59904961.0, + "step": 2320 + }, + { + "epoch": 0.2548868877663079, + "grad_norm": 2.0771772861480713, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6907344460487366, + "num_tokens": 59934681.0, + "step": 2321 + }, + { + "epoch": 0.2549967054689216, + "grad_norm": 2.4771392345428467, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6989715099334717, + "num_tokens": 59955528.0, + "step": 2322 + }, + { + "epoch": 0.25510652317153526, + "grad_norm": 2.285086154937744, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6776152849197388, + "num_tokens": 59979170.0, + "step": 2323 + }, + { + "epoch": 0.2552163408741489, + "grad_norm": 2.2462918758392334, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7028204202651978, + "num_tokens": 60003649.0, + "step": 2324 + }, + { + "epoch": 0.25532615857676255, + "grad_norm": 2.1103076934814453, + "learning_rate": 1e-06, + "loss": 1.0394, + "mean_token_accuracy": 0.6925048828125, + "num_tokens": 60031469.0, + "step": 2325 + }, + { + "epoch": 0.25543597627937625, + "grad_norm": 2.3965072631835938, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7225831747055054, + "num_tokens": 60052171.0, + "step": 2326 + }, + { + "epoch": 0.2555457939819899, + "grad_norm": 2.089719533920288, + "learning_rate": 1e-06, + "loss": 1.131, + "mean_token_accuracy": 0.664777934551239, + "num_tokens": 60081404.0, + "step": 2327 + }, + { + "epoch": 0.25565561168460355, + "grad_norm": 2.1475625038146973, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6959878206253052, + "num_tokens": 60108231.0, + "step": 2328 + }, + { + "epoch": 0.25576542938721725, + "grad_norm": 2.2803890705108643, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7263441681861877, + "num_tokens": 60132492.0, + "step": 2329 + }, + { + "epoch": 0.2558752470898309, + "grad_norm": 2.23850417137146, + "learning_rate": 1e-06, + "loss": 1.0908, + "mean_token_accuracy": 0.6732912063598633, + "num_tokens": 60160330.0, + "step": 2330 + }, + { + "epoch": 0.25598506479244454, + "grad_norm": 2.299774408340454, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.7084947824478149, + "num_tokens": 60184695.0, + "step": 2331 + }, + { + "epoch": 0.2560948824950582, + "grad_norm": 2.0408506393432617, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.6826627254486084, + "num_tokens": 60215168.0, + "step": 2332 + }, + { + "epoch": 0.2562047001976719, + "grad_norm": 2.4331252574920654, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7184293270111084, + "num_tokens": 60237217.0, + "step": 2333 + }, + { + "epoch": 0.25631451790028553, + "grad_norm": 2.3885533809661865, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6934524178504944, + "num_tokens": 60261410.0, + "step": 2334 + }, + { + "epoch": 0.2564243356028992, + "grad_norm": 2.312624931335449, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6809480786323547, + "num_tokens": 60287248.0, + "step": 2335 + }, + { + "epoch": 0.2565341533055128, + "grad_norm": 2.175384998321533, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7050056457519531, + "num_tokens": 60316731.0, + "step": 2336 + }, + { + "epoch": 0.2566439710081265, + "grad_norm": 2.4205172061920166, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7034503221511841, + "num_tokens": 60338959.0, + "step": 2337 + }, + { + "epoch": 0.25675378871074017, + "grad_norm": 2.150881767272949, + "learning_rate": 1e-06, + "loss": 1.0826, + "mean_token_accuracy": 0.6803547143936157, + "num_tokens": 60368860.0, + "step": 2338 + }, + { + "epoch": 0.2568636064133538, + "grad_norm": 2.223261594772339, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7081143260002136, + "num_tokens": 60394473.0, + "step": 2339 + }, + { + "epoch": 0.2569734241159675, + "grad_norm": 2.337179183959961, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6886272430419922, + "num_tokens": 60418092.0, + "step": 2340 + }, + { + "epoch": 0.25708324181858117, + "grad_norm": 2.1180458068847656, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.6717043519020081, + "num_tokens": 60446447.0, + "step": 2341 + }, + { + "epoch": 0.2571930595211948, + "grad_norm": 2.29925799369812, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6894848346710205, + "num_tokens": 60470586.0, + "step": 2342 + }, + { + "epoch": 0.25730287722380846, + "grad_norm": 2.2159156799316406, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7138614654541016, + "num_tokens": 60494810.0, + "step": 2343 + }, + { + "epoch": 0.25741269492642216, + "grad_norm": 2.325207471847534, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6834876537322998, + "num_tokens": 60519001.0, + "step": 2344 + }, + { + "epoch": 0.2575225126290358, + "grad_norm": 2.1697499752044678, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7099623084068298, + "num_tokens": 60544983.0, + "step": 2345 + }, + { + "epoch": 0.25763233033164945, + "grad_norm": 2.342846155166626, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6840147972106934, + "num_tokens": 60568020.0, + "step": 2346 + }, + { + "epoch": 0.2577421480342631, + "grad_norm": 2.6124095916748047, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7122206687927246, + "num_tokens": 60587199.0, + "step": 2347 + }, + { + "epoch": 0.2578519657368768, + "grad_norm": 2.3711369037628174, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6860393285751343, + "num_tokens": 60611967.0, + "step": 2348 + }, + { + "epoch": 0.25796178343949044, + "grad_norm": 2.0964601039886475, + "learning_rate": 1e-06, + "loss": 1.095, + "mean_token_accuracy": 0.6723437905311584, + "num_tokens": 60641762.0, + "step": 2349 + }, + { + "epoch": 0.2580716011421041, + "grad_norm": 2.358553171157837, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.7000367045402527, + "num_tokens": 60665291.0, + "step": 2350 + }, + { + "epoch": 0.2581814188447178, + "grad_norm": 2.241563558578491, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6918748021125793, + "num_tokens": 60691001.0, + "step": 2351 + }, + { + "epoch": 0.25829123654733144, + "grad_norm": 2.2165334224700928, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7030646800994873, + "num_tokens": 60718073.0, + "step": 2352 + }, + { + "epoch": 0.2584010542499451, + "grad_norm": 2.1653475761413574, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.7059882879257202, + "num_tokens": 60748071.0, + "step": 2353 + }, + { + "epoch": 0.25851087195255873, + "grad_norm": 2.5857608318328857, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7348031997680664, + "num_tokens": 60766542.0, + "step": 2354 + }, + { + "epoch": 0.25862068965517243, + "grad_norm": 2.2284514904022217, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6992921233177185, + "num_tokens": 60793103.0, + "step": 2355 + }, + { + "epoch": 0.2587305073577861, + "grad_norm": 2.3799076080322266, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6954632997512817, + "num_tokens": 60814139.0, + "step": 2356 + }, + { + "epoch": 0.2588403250603997, + "grad_norm": 2.596575975418091, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7104941606521606, + "num_tokens": 60834680.0, + "step": 2357 + }, + { + "epoch": 0.2589501427630134, + "grad_norm": 2.139164447784424, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6831586956977844, + "num_tokens": 60861523.0, + "step": 2358 + }, + { + "epoch": 0.25905996046562707, + "grad_norm": 2.202613353729248, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6968038082122803, + "num_tokens": 60888063.0, + "step": 2359 + }, + { + "epoch": 0.2591697781682407, + "grad_norm": 1.9872312545776367, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6895331144332886, + "num_tokens": 60917759.0, + "step": 2360 + }, + { + "epoch": 0.25927959587085436, + "grad_norm": 2.5894079208374023, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.714935302734375, + "num_tokens": 60936014.0, + "step": 2361 + }, + { + "epoch": 0.25938941357346806, + "grad_norm": 2.1612184047698975, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7132415175437927, + "num_tokens": 60960731.0, + "step": 2362 + }, + { + "epoch": 0.2594992312760817, + "grad_norm": 2.3150813579559326, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6982759237289429, + "num_tokens": 60985020.0, + "step": 2363 + }, + { + "epoch": 0.25960904897869536, + "grad_norm": 2.2492101192474365, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6942394971847534, + "num_tokens": 61010760.0, + "step": 2364 + }, + { + "epoch": 0.259718866681309, + "grad_norm": 2.268794536590576, + "learning_rate": 1e-06, + "loss": 1.0815, + "mean_token_accuracy": 0.6728893518447876, + "num_tokens": 61039656.0, + "step": 2365 + }, + { + "epoch": 0.2598286843839227, + "grad_norm": 2.092608690261841, + "learning_rate": 1e-06, + "loss": 1.0616, + "mean_token_accuracy": 0.6841089129447937, + "num_tokens": 61068116.0, + "step": 2366 + }, + { + "epoch": 0.25993850208653635, + "grad_norm": 2.327131509780884, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.7017263174057007, + "num_tokens": 61092024.0, + "step": 2367 + }, + { + "epoch": 0.26004831978915, + "grad_norm": 2.3306777477264404, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6820593476295471, + "num_tokens": 61117065.0, + "step": 2368 + }, + { + "epoch": 0.2601581374917637, + "grad_norm": 2.2063679695129395, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6953889727592468, + "num_tokens": 61142419.0, + "step": 2369 + }, + { + "epoch": 0.26026795519437734, + "grad_norm": 2.3943114280700684, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6918026208877563, + "num_tokens": 61166072.0, + "step": 2370 + }, + { + "epoch": 0.260377772896991, + "grad_norm": 2.5245797634124756, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.697031557559967, + "num_tokens": 61187604.0, + "step": 2371 + }, + { + "epoch": 0.26048759059960463, + "grad_norm": 1.9765626192092896, + "learning_rate": 1e-06, + "loss": 1.1087, + "mean_token_accuracy": 0.6761438846588135, + "num_tokens": 61222407.0, + "step": 2372 + }, + { + "epoch": 0.26059740830221834, + "grad_norm": 2.1553685665130615, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7229886054992676, + "num_tokens": 61246314.0, + "step": 2373 + }, + { + "epoch": 0.260707226004832, + "grad_norm": 2.3453142642974854, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6889591217041016, + "num_tokens": 61270451.0, + "step": 2374 + }, + { + "epoch": 0.26081704370744563, + "grad_norm": 2.327561378479004, + "learning_rate": 1e-06, + "loss": 1.1128, + "mean_token_accuracy": 0.6679035425186157, + "num_tokens": 61294335.0, + "step": 2375 + }, + { + "epoch": 0.2609268614100593, + "grad_norm": 2.441070079803467, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.71410071849823, + "num_tokens": 61315172.0, + "step": 2376 + }, + { + "epoch": 0.261036679112673, + "grad_norm": 2.2317004203796387, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6998625993728638, + "num_tokens": 61341229.0, + "step": 2377 + }, + { + "epoch": 0.2611464968152866, + "grad_norm": 2.4552483558654785, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7015321254730225, + "num_tokens": 61363106.0, + "step": 2378 + }, + { + "epoch": 0.26125631451790027, + "grad_norm": 2.2590808868408203, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6818060874938965, + "num_tokens": 61389569.0, + "step": 2379 + }, + { + "epoch": 0.26136613222051397, + "grad_norm": 2.2093605995178223, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6875653266906738, + "num_tokens": 61417476.0, + "step": 2380 + }, + { + "epoch": 0.2614759499231276, + "grad_norm": 2.565655469894409, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7065209746360779, + "num_tokens": 61438214.0, + "step": 2381 + }, + { + "epoch": 0.26158576762574126, + "grad_norm": 2.286374807357788, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.7033008337020874, + "num_tokens": 61465614.0, + "step": 2382 + }, + { + "epoch": 0.2616955853283549, + "grad_norm": 2.5700278282165527, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6953299641609192, + "num_tokens": 61486267.0, + "step": 2383 + }, + { + "epoch": 0.2618054030309686, + "grad_norm": 1.97044837474823, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7011287212371826, + "num_tokens": 61519433.0, + "step": 2384 + }, + { + "epoch": 0.26191522073358225, + "grad_norm": 2.110734462738037, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.685859203338623, + "num_tokens": 61547180.0, + "step": 2385 + }, + { + "epoch": 0.2620250384361959, + "grad_norm": 2.3183510303497314, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7098879218101501, + "num_tokens": 61573239.0, + "step": 2386 + }, + { + "epoch": 0.2621348561388096, + "grad_norm": 2.430783271789551, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7029111385345459, + "num_tokens": 61595423.0, + "step": 2387 + }, + { + "epoch": 0.26224467384142325, + "grad_norm": 2.25355863571167, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6885250806808472, + "num_tokens": 61620876.0, + "step": 2388 + }, + { + "epoch": 0.2623544915440369, + "grad_norm": 2.039889097213745, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6901452541351318, + "num_tokens": 61652458.0, + "step": 2389 + }, + { + "epoch": 0.26246430924665054, + "grad_norm": 2.5240588188171387, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6944581270217896, + "num_tokens": 61674859.0, + "step": 2390 + }, + { + "epoch": 0.26257412694926424, + "grad_norm": 2.2499797344207764, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6876588463783264, + "num_tokens": 61699294.0, + "step": 2391 + }, + { + "epoch": 0.2626839446518779, + "grad_norm": 2.151460886001587, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.685601532459259, + "num_tokens": 61728275.0, + "step": 2392 + }, + { + "epoch": 0.26279376235449153, + "grad_norm": 2.325702428817749, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.68622887134552, + "num_tokens": 61752341.0, + "step": 2393 + }, + { + "epoch": 0.2629035800571052, + "grad_norm": 2.1962883472442627, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.7002992033958435, + "num_tokens": 61779168.0, + "step": 2394 + }, + { + "epoch": 0.2630133977597189, + "grad_norm": 2.2916030883789062, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7121120691299438, + "num_tokens": 61802336.0, + "step": 2395 + }, + { + "epoch": 0.2631232154623325, + "grad_norm": 1.888262391090393, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6928816437721252, + "num_tokens": 61835232.0, + "step": 2396 + }, + { + "epoch": 0.2632330331649462, + "grad_norm": 2.125364303588867, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6884888410568237, + "num_tokens": 61863451.0, + "step": 2397 + }, + { + "epoch": 0.2633428508675599, + "grad_norm": 2.1419355869293213, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7056692838668823, + "num_tokens": 61890454.0, + "step": 2398 + }, + { + "epoch": 0.2634526685701735, + "grad_norm": 2.195099115371704, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6966936588287354, + "num_tokens": 61918784.0, + "step": 2399 + }, + { + "epoch": 0.26356248627278717, + "grad_norm": 2.042595624923706, + "learning_rate": 1e-06, + "loss": 1.0906, + "mean_token_accuracy": 0.6732950210571289, + "num_tokens": 61950786.0, + "step": 2400 + }, + { + "epoch": 0.2636723039754008, + "grad_norm": 2.004512310028076, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6768084764480591, + "num_tokens": 61981095.0, + "step": 2401 + }, + { + "epoch": 0.2637821216780145, + "grad_norm": 2.125494956970215, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6933797597885132, + "num_tokens": 62008529.0, + "step": 2402 + }, + { + "epoch": 0.26389193938062816, + "grad_norm": 2.101045608520508, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7162313461303711, + "num_tokens": 62036136.0, + "step": 2403 + }, + { + "epoch": 0.2640017570832418, + "grad_norm": 2.3658595085144043, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7060168981552124, + "num_tokens": 62058353.0, + "step": 2404 + }, + { + "epoch": 0.2641115747858555, + "grad_norm": 2.3049700260162354, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7088220715522766, + "num_tokens": 62083134.0, + "step": 2405 + }, + { + "epoch": 0.26422139248846915, + "grad_norm": 2.3089654445648193, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6846330165863037, + "num_tokens": 62109239.0, + "step": 2406 + }, + { + "epoch": 0.2643312101910828, + "grad_norm": 2.578106641769409, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7007532119750977, + "num_tokens": 62131840.0, + "step": 2407 + }, + { + "epoch": 0.26444102789369645, + "grad_norm": 2.253500461578369, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7226444482803345, + "num_tokens": 62156354.0, + "step": 2408 + }, + { + "epoch": 0.26455084559631015, + "grad_norm": 2.1388728618621826, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6882116794586182, + "num_tokens": 62183735.0, + "step": 2409 + }, + { + "epoch": 0.2646606632989238, + "grad_norm": 2.063368558883667, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7122186422348022, + "num_tokens": 62212553.0, + "step": 2410 + }, + { + "epoch": 0.26477048100153744, + "grad_norm": 2.3363523483276367, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6923008561134338, + "num_tokens": 62237629.0, + "step": 2411 + }, + { + "epoch": 0.2648802987041511, + "grad_norm": 2.6877427101135254, + "learning_rate": 1e-06, + "loss": 1.0732, + "mean_token_accuracy": 0.6788772344589233, + "num_tokens": 62258304.0, + "step": 2412 + }, + { + "epoch": 0.2649901164067648, + "grad_norm": 2.3245866298675537, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.6715583801269531, + "num_tokens": 62285960.0, + "step": 2413 + }, + { + "epoch": 0.26509993410937843, + "grad_norm": 2.4135773181915283, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6942824721336365, + "num_tokens": 62309222.0, + "step": 2414 + }, + { + "epoch": 0.2652097518119921, + "grad_norm": 2.6672019958496094, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7002387046813965, + "num_tokens": 62328242.0, + "step": 2415 + }, + { + "epoch": 0.2653195695146058, + "grad_norm": 2.3864376544952393, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7057990431785583, + "num_tokens": 62352418.0, + "step": 2416 + }, + { + "epoch": 0.2654293872172194, + "grad_norm": 2.1138548851013184, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.675590991973877, + "num_tokens": 62381798.0, + "step": 2417 + }, + { + "epoch": 0.26553920491983307, + "grad_norm": 1.914778470993042, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6929221153259277, + "num_tokens": 62416690.0, + "step": 2418 + }, + { + "epoch": 0.2656490226224467, + "grad_norm": 2.175252676010132, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6914165019989014, + "num_tokens": 62444218.0, + "step": 2419 + }, + { + "epoch": 0.2657588403250604, + "grad_norm": 2.1641087532043457, + "learning_rate": 1e-06, + "loss": 1.1645, + "mean_token_accuracy": 0.6661110520362854, + "num_tokens": 62473071.0, + "step": 2420 + }, + { + "epoch": 0.26586865802767407, + "grad_norm": 2.1321499347686768, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6988722085952759, + "num_tokens": 62499302.0, + "step": 2421 + }, + { + "epoch": 0.2659784757302877, + "grad_norm": 2.1342077255249023, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6938774585723877, + "num_tokens": 62529469.0, + "step": 2422 + }, + { + "epoch": 0.26608829343290136, + "grad_norm": 2.6077518463134766, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7160838842391968, + "num_tokens": 62548024.0, + "step": 2423 + }, + { + "epoch": 0.26619811113551506, + "grad_norm": 2.216510772705078, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6850734353065491, + "num_tokens": 62574421.0, + "step": 2424 + }, + { + "epoch": 0.2663079288381287, + "grad_norm": 2.514425039291382, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7121505737304688, + "num_tokens": 62596454.0, + "step": 2425 + }, + { + "epoch": 0.26641774654074235, + "grad_norm": 2.1031792163848877, + "learning_rate": 1e-06, + "loss": 1.0769, + "mean_token_accuracy": 0.6743325591087341, + "num_tokens": 62628486.0, + "step": 2426 + }, + { + "epoch": 0.26652756424335605, + "grad_norm": 2.2803471088409424, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7072607278823853, + "num_tokens": 62652447.0, + "step": 2427 + }, + { + "epoch": 0.2666373819459697, + "grad_norm": 1.9234825372695923, + "learning_rate": 1e-06, + "loss": 1.1127, + "mean_token_accuracy": 0.6741091012954712, + "num_tokens": 62687874.0, + "step": 2428 + }, + { + "epoch": 0.26674719964858334, + "grad_norm": 2.3509645462036133, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7001485824584961, + "num_tokens": 62711105.0, + "step": 2429 + }, + { + "epoch": 0.266857017351197, + "grad_norm": 2.178572416305542, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6915339231491089, + "num_tokens": 62735874.0, + "step": 2430 + }, + { + "epoch": 0.2669668350538107, + "grad_norm": 2.3349287509918213, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6948130130767822, + "num_tokens": 62760624.0, + "step": 2431 + }, + { + "epoch": 0.26707665275642434, + "grad_norm": 2.1319518089294434, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7056596279144287, + "num_tokens": 62785933.0, + "step": 2432 + }, + { + "epoch": 0.267186470459038, + "grad_norm": 2.343618631362915, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7308557033538818, + "num_tokens": 62806371.0, + "step": 2433 + }, + { + "epoch": 0.2672962881616517, + "grad_norm": 2.0218570232391357, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6853881478309631, + "num_tokens": 62839934.0, + "step": 2434 + }, + { + "epoch": 0.26740610586426533, + "grad_norm": 2.2392356395721436, + "learning_rate": 1e-06, + "loss": 1.1136, + "mean_token_accuracy": 0.6699985861778259, + "num_tokens": 62868325.0, + "step": 2435 + }, + { + "epoch": 0.267515923566879, + "grad_norm": 2.3392553329467773, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6830188632011414, + "num_tokens": 62894633.0, + "step": 2436 + }, + { + "epoch": 0.2676257412694926, + "grad_norm": 2.3163459300994873, + "learning_rate": 1e-06, + "loss": 1.079, + "mean_token_accuracy": 0.6770175695419312, + "num_tokens": 62919063.0, + "step": 2437 + }, + { + "epoch": 0.2677355589721063, + "grad_norm": 2.110191583633423, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6850354671478271, + "num_tokens": 62948061.0, + "step": 2438 + }, + { + "epoch": 0.26784537667471997, + "grad_norm": 2.3562891483306885, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7152935266494751, + "num_tokens": 62971866.0, + "step": 2439 + }, + { + "epoch": 0.2679551943773336, + "grad_norm": 2.4854979515075684, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7063655853271484, + "num_tokens": 62993392.0, + "step": 2440 + }, + { + "epoch": 0.26806501207994726, + "grad_norm": 2.266742467880249, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6874722242355347, + "num_tokens": 63018796.0, + "step": 2441 + }, + { + "epoch": 0.26817482978256096, + "grad_norm": 2.0599310398101807, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7109930515289307, + "num_tokens": 63047311.0, + "step": 2442 + }, + { + "epoch": 0.2682846474851746, + "grad_norm": 2.1297566890716553, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6998173594474792, + "num_tokens": 63073297.0, + "step": 2443 + }, + { + "epoch": 0.26839446518778826, + "grad_norm": 2.1919615268707275, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7139228582382202, + "num_tokens": 63099855.0, + "step": 2444 + }, + { + "epoch": 0.26850428289040196, + "grad_norm": 2.1270251274108887, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6874048113822937, + "num_tokens": 63128520.0, + "step": 2445 + }, + { + "epoch": 0.2686141005930156, + "grad_norm": 2.6414999961853027, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7251601219177246, + "num_tokens": 63147892.0, + "step": 2446 + }, + { + "epoch": 0.26872391829562925, + "grad_norm": 2.0255870819091797, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6960870623588562, + "num_tokens": 63178476.0, + "step": 2447 + }, + { + "epoch": 0.2688337359982429, + "grad_norm": 2.131002426147461, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6956694722175598, + "num_tokens": 63205223.0, + "step": 2448 + }, + { + "epoch": 0.2689435537008566, + "grad_norm": 2.4028854370117188, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7092064023017883, + "num_tokens": 63227640.0, + "step": 2449 + }, + { + "epoch": 0.26905337140347024, + "grad_norm": 2.296433210372925, + "learning_rate": 1e-06, + "loss": 1.0728, + "mean_token_accuracy": 0.6821365356445312, + "num_tokens": 63253812.0, + "step": 2450 + }, + { + "epoch": 0.2691631891060839, + "grad_norm": 2.3003439903259277, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.7044062614440918, + "num_tokens": 63279742.0, + "step": 2451 + }, + { + "epoch": 0.26927300680869753, + "grad_norm": 2.1476309299468994, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.7008305788040161, + "num_tokens": 63307096.0, + "step": 2452 + }, + { + "epoch": 0.26938282451131124, + "grad_norm": 2.228072166442871, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.7005294561386108, + "num_tokens": 63333937.0, + "step": 2453 + }, + { + "epoch": 0.2694926422139249, + "grad_norm": 2.1715152263641357, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.708048403263092, + "num_tokens": 63359298.0, + "step": 2454 + }, + { + "epoch": 0.26960245991653853, + "grad_norm": 2.6441757678985596, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7082767486572266, + "num_tokens": 63379801.0, + "step": 2455 + }, + { + "epoch": 0.26971227761915223, + "grad_norm": 2.03165340423584, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6978657841682434, + "num_tokens": 63410065.0, + "step": 2456 + }, + { + "epoch": 0.2698220953217659, + "grad_norm": 2.3340229988098145, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.699277400970459, + "num_tokens": 63433369.0, + "step": 2457 + }, + { + "epoch": 0.2699319130243795, + "grad_norm": 2.383875846862793, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7149590253829956, + "num_tokens": 63454984.0, + "step": 2458 + }, + { + "epoch": 0.27004173072699317, + "grad_norm": 2.5110154151916504, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7012194991111755, + "num_tokens": 63474406.0, + "step": 2459 + }, + { + "epoch": 0.27015154842960687, + "grad_norm": 1.9184472560882568, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6926854252815247, + "num_tokens": 63505566.0, + "step": 2460 + }, + { + "epoch": 0.2702613661322205, + "grad_norm": 2.168076753616333, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6873824000358582, + "num_tokens": 63532532.0, + "step": 2461 + }, + { + "epoch": 0.27037118383483416, + "grad_norm": 2.23272442817688, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7058197855949402, + "num_tokens": 63557780.0, + "step": 2462 + }, + { + "epoch": 0.27048100153744786, + "grad_norm": 2.191373825073242, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6918995380401611, + "num_tokens": 63583108.0, + "step": 2463 + }, + { + "epoch": 0.2705908192400615, + "grad_norm": 2.2297797203063965, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6929201483726501, + "num_tokens": 63609355.0, + "step": 2464 + }, + { + "epoch": 0.27070063694267515, + "grad_norm": 2.422482967376709, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6904378533363342, + "num_tokens": 63630579.0, + "step": 2465 + }, + { + "epoch": 0.2708104546452888, + "grad_norm": 2.598745822906494, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6986764669418335, + "num_tokens": 63650629.0, + "step": 2466 + }, + { + "epoch": 0.2709202723479025, + "grad_norm": 2.251478910446167, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6893923878669739, + "num_tokens": 63675924.0, + "step": 2467 + }, + { + "epoch": 0.27103009005051615, + "grad_norm": 2.2059667110443115, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6960403919219971, + "num_tokens": 63703933.0, + "step": 2468 + }, + { + "epoch": 0.2711399077531298, + "grad_norm": 2.0785586833953857, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.712672770023346, + "num_tokens": 63730841.0, + "step": 2469 + }, + { + "epoch": 0.27124972545574344, + "grad_norm": 2.1068549156188965, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6884162425994873, + "num_tokens": 63760437.0, + "step": 2470 + }, + { + "epoch": 0.27135954315835714, + "grad_norm": 2.3880584239959717, + "learning_rate": 1e-06, + "loss": 1.0936, + "mean_token_accuracy": 0.6669161319732666, + "num_tokens": 63784375.0, + "step": 2471 + }, + { + "epoch": 0.2714693608609708, + "grad_norm": 2.1266090869903564, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7061692476272583, + "num_tokens": 63811964.0, + "step": 2472 + }, + { + "epoch": 0.27157917856358443, + "grad_norm": 2.014343738555908, + "learning_rate": 1e-06, + "loss": 1.1048, + "mean_token_accuracy": 0.6650605797767639, + "num_tokens": 63845092.0, + "step": 2473 + }, + { + "epoch": 0.27168899626619814, + "grad_norm": 2.4571642875671387, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6965371966362, + "num_tokens": 63869025.0, + "step": 2474 + }, + { + "epoch": 0.2717988139688118, + "grad_norm": 2.107691526412964, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6978173851966858, + "num_tokens": 63897135.0, + "step": 2475 + }, + { + "epoch": 0.2719086316714254, + "grad_norm": 2.377901315689087, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6878422498703003, + "num_tokens": 63921256.0, + "step": 2476 + }, + { + "epoch": 0.2720184493740391, + "grad_norm": 2.1171317100524902, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7069856524467468, + "num_tokens": 63949020.0, + "step": 2477 + }, + { + "epoch": 0.2721282670766528, + "grad_norm": 2.235715627670288, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7046996355056763, + "num_tokens": 63972452.0, + "step": 2478 + }, + { + "epoch": 0.2722380847792664, + "grad_norm": 2.1717445850372314, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6958745121955872, + "num_tokens": 64000538.0, + "step": 2479 + }, + { + "epoch": 0.27234790248188007, + "grad_norm": 2.34332537651062, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6973495483398438, + "num_tokens": 64025074.0, + "step": 2480 + }, + { + "epoch": 0.27245772018449377, + "grad_norm": 2.2115862369537354, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6997119784355164, + "num_tokens": 64053186.0, + "step": 2481 + }, + { + "epoch": 0.2725675378871074, + "grad_norm": 1.9629034996032715, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6967809796333313, + "num_tokens": 64086329.0, + "step": 2482 + }, + { + "epoch": 0.27267735558972106, + "grad_norm": 2.4804556369781494, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6940837502479553, + "num_tokens": 64108256.0, + "step": 2483 + }, + { + "epoch": 0.2727871732923347, + "grad_norm": 2.0010108947753906, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6980389952659607, + "num_tokens": 64138464.0, + "step": 2484 + }, + { + "epoch": 0.2728969909949484, + "grad_norm": 2.5084171295166016, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.713110089302063, + "num_tokens": 64159097.0, + "step": 2485 + }, + { + "epoch": 0.27300680869756205, + "grad_norm": 2.219472646713257, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6847208142280579, + "num_tokens": 64184138.0, + "step": 2486 + }, + { + "epoch": 0.2731166264001757, + "grad_norm": 1.9349794387817383, + "learning_rate": 1e-06, + "loss": 1.0817, + "mean_token_accuracy": 0.6775318384170532, + "num_tokens": 64218914.0, + "step": 2487 + }, + { + "epoch": 0.27322644410278935, + "grad_norm": 2.2417213916778564, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6865782737731934, + "num_tokens": 64246462.0, + "step": 2488 + }, + { + "epoch": 0.27333626180540305, + "grad_norm": 2.1565747261047363, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.703755259513855, + "num_tokens": 64274829.0, + "step": 2489 + }, + { + "epoch": 0.2734460795080167, + "grad_norm": 2.112541913986206, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6816768646240234, + "num_tokens": 64303542.0, + "step": 2490 + }, + { + "epoch": 0.27355589721063034, + "grad_norm": 2.6087145805358887, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6712536215782166, + "num_tokens": 64323816.0, + "step": 2491 + }, + { + "epoch": 0.27366571491324404, + "grad_norm": 2.389437198638916, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7320709228515625, + "num_tokens": 64346137.0, + "step": 2492 + }, + { + "epoch": 0.2737755326158577, + "grad_norm": 2.644735336303711, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6982771754264832, + "num_tokens": 64366839.0, + "step": 2493 + }, + { + "epoch": 0.27388535031847133, + "grad_norm": 2.259606122970581, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7233859300613403, + "num_tokens": 64391342.0, + "step": 2494 + }, + { + "epoch": 0.273995168021085, + "grad_norm": 2.588787078857422, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.716132640838623, + "num_tokens": 64411224.0, + "step": 2495 + }, + { + "epoch": 0.2741049857236987, + "grad_norm": 2.178715229034424, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6865944266319275, + "num_tokens": 64440975.0, + "step": 2496 + }, + { + "epoch": 0.2742148034263123, + "grad_norm": 2.271003484725952, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7140557765960693, + "num_tokens": 64466018.0, + "step": 2497 + }, + { + "epoch": 0.27432462112892597, + "grad_norm": 2.0216736793518066, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.7017310261726379, + "num_tokens": 64497109.0, + "step": 2498 + }, + { + "epoch": 0.2744344388315396, + "grad_norm": 2.2359118461608887, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6873013973236084, + "num_tokens": 64525305.0, + "step": 2499 + }, + { + "epoch": 0.2745442565341533, + "grad_norm": 2.3072028160095215, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7047764658927917, + "num_tokens": 64549956.0, + "step": 2500 + }, + { + "epoch": 0.27465407423676697, + "grad_norm": 2.354844570159912, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7051737904548645, + "num_tokens": 64575653.0, + "step": 2501 + }, + { + "epoch": 0.2747638919393806, + "grad_norm": 2.495276927947998, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7093989849090576, + "num_tokens": 64595590.0, + "step": 2502 + }, + { + "epoch": 0.2748737096419943, + "grad_norm": 2.4885244369506836, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.683014988899231, + "num_tokens": 64619009.0, + "step": 2503 + }, + { + "epoch": 0.27498352734460796, + "grad_norm": 2.1737537384033203, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6920415163040161, + "num_tokens": 64646391.0, + "step": 2504 + }, + { + "epoch": 0.2750933450472216, + "grad_norm": 2.1784050464630127, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7029175758361816, + "num_tokens": 64672174.0, + "step": 2505 + }, + { + "epoch": 0.27520316274983525, + "grad_norm": 2.1611087322235107, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7024701237678528, + "num_tokens": 64700338.0, + "step": 2506 + }, + { + "epoch": 0.27531298045244895, + "grad_norm": 2.3393256664276123, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7023281455039978, + "num_tokens": 64724699.0, + "step": 2507 + }, + { + "epoch": 0.2754227981550626, + "grad_norm": 2.5440783500671387, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6940992474555969, + "num_tokens": 64745750.0, + "step": 2508 + }, + { + "epoch": 0.27553261585767624, + "grad_norm": 2.2927615642547607, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6972587704658508, + "num_tokens": 64770547.0, + "step": 2509 + }, + { + "epoch": 0.27564243356028995, + "grad_norm": 2.395782709121704, + "learning_rate": 1e-06, + "loss": 1.0851, + "mean_token_accuracy": 0.6717148423194885, + "num_tokens": 64792823.0, + "step": 2510 + }, + { + "epoch": 0.2757522512629036, + "grad_norm": 1.9524824619293213, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6815095543861389, + "num_tokens": 64824509.0, + "step": 2511 + }, + { + "epoch": 0.27586206896551724, + "grad_norm": 2.3183677196502686, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6980201005935669, + "num_tokens": 64848807.0, + "step": 2512 + }, + { + "epoch": 0.2759718866681309, + "grad_norm": 2.2596302032470703, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6977387070655823, + "num_tokens": 64874188.0, + "step": 2513 + }, + { + "epoch": 0.2760817043707446, + "grad_norm": 2.4059228897094727, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6970065832138062, + "num_tokens": 64899191.0, + "step": 2514 + }, + { + "epoch": 0.27619152207335823, + "grad_norm": 2.6236770153045654, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7096291184425354, + "num_tokens": 64919017.0, + "step": 2515 + }, + { + "epoch": 0.2763013397759719, + "grad_norm": 2.235435724258423, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6924608945846558, + "num_tokens": 64945959.0, + "step": 2516 + }, + { + "epoch": 0.2764111574785855, + "grad_norm": 2.49448561668396, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6912168264389038, + "num_tokens": 64968162.0, + "step": 2517 + }, + { + "epoch": 0.2765209751811992, + "grad_norm": 1.9696974754333496, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6991859078407288, + "num_tokens": 65002357.0, + "step": 2518 + }, + { + "epoch": 0.27663079288381287, + "grad_norm": 2.317485809326172, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7013013362884521, + "num_tokens": 65027743.0, + "step": 2519 + }, + { + "epoch": 0.2767406105864265, + "grad_norm": 2.194570302963257, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.69029700756073, + "num_tokens": 65055973.0, + "step": 2520 + }, + { + "epoch": 0.2768504282890402, + "grad_norm": 2.347210645675659, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6947752833366394, + "num_tokens": 65080125.0, + "step": 2521 + }, + { + "epoch": 0.27696024599165386, + "grad_norm": 2.3680434226989746, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7150743007659912, + "num_tokens": 65102925.0, + "step": 2522 + }, + { + "epoch": 0.2770700636942675, + "grad_norm": 2.1373956203460693, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6953936815261841, + "num_tokens": 65131260.0, + "step": 2523 + }, + { + "epoch": 0.27717988139688116, + "grad_norm": 2.464764356613159, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7037779092788696, + "num_tokens": 65152285.0, + "step": 2524 + }, + { + "epoch": 0.27728969909949486, + "grad_norm": 2.280667543411255, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7084994316101074, + "num_tokens": 65176178.0, + "step": 2525 + }, + { + "epoch": 0.2773995168021085, + "grad_norm": 2.4628636837005615, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6899791359901428, + "num_tokens": 65196740.0, + "step": 2526 + }, + { + "epoch": 0.27750933450472215, + "grad_norm": 2.044948101043701, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6977730393409729, + "num_tokens": 65228640.0, + "step": 2527 + }, + { + "epoch": 0.2776191522073358, + "grad_norm": 2.09145450592041, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7036658525466919, + "num_tokens": 65259440.0, + "step": 2528 + }, + { + "epoch": 0.2777289699099495, + "grad_norm": 2.1546363830566406, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7042264938354492, + "num_tokens": 65287795.0, + "step": 2529 + }, + { + "epoch": 0.27783878761256314, + "grad_norm": 2.146503448486328, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.711677074432373, + "num_tokens": 65313140.0, + "step": 2530 + }, + { + "epoch": 0.2779486053151768, + "grad_norm": 2.552994728088379, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7119708061218262, + "num_tokens": 65332172.0, + "step": 2531 + }, + { + "epoch": 0.2780584230177905, + "grad_norm": 2.183095693588257, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7041839361190796, + "num_tokens": 65358063.0, + "step": 2532 + }, + { + "epoch": 0.27816824072040414, + "grad_norm": 2.3105344772338867, + "learning_rate": 1e-06, + "loss": 1.1155, + "mean_token_accuracy": 0.6632633209228516, + "num_tokens": 65384247.0, + "step": 2533 + }, + { + "epoch": 0.2782780584230178, + "grad_norm": 2.6424477100372314, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7047666311264038, + "num_tokens": 65406748.0, + "step": 2534 + }, + { + "epoch": 0.27838787612563143, + "grad_norm": 2.090571165084839, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7251778841018677, + "num_tokens": 65433933.0, + "step": 2535 + }, + { + "epoch": 0.27849769382824513, + "grad_norm": 2.224240303039551, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7205511331558228, + "num_tokens": 65458697.0, + "step": 2536 + }, + { + "epoch": 0.2786075115308588, + "grad_norm": 2.243959665298462, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7078279256820679, + "num_tokens": 65482012.0, + "step": 2537 + }, + { + "epoch": 0.2787173292334724, + "grad_norm": 2.3121304512023926, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.7021582126617432, + "num_tokens": 65506537.0, + "step": 2538 + }, + { + "epoch": 0.2788271469360861, + "grad_norm": 2.023761749267578, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6886287927627563, + "num_tokens": 65538459.0, + "step": 2539 + }, + { + "epoch": 0.27893696463869977, + "grad_norm": 2.418968677520752, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6937419176101685, + "num_tokens": 65561564.0, + "step": 2540 + }, + { + "epoch": 0.2790467823413134, + "grad_norm": 2.031751871109009, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6848655939102173, + "num_tokens": 65591962.0, + "step": 2541 + }, + { + "epoch": 0.27915660004392706, + "grad_norm": 2.3305864334106445, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6842211484909058, + "num_tokens": 65616021.0, + "step": 2542 + }, + { + "epoch": 0.27926641774654076, + "grad_norm": 2.093228340148926, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7089740037918091, + "num_tokens": 65643416.0, + "step": 2543 + }, + { + "epoch": 0.2793762354491544, + "grad_norm": 2.4114296436309814, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6900089383125305, + "num_tokens": 65666712.0, + "step": 2544 + }, + { + "epoch": 0.27948605315176805, + "grad_norm": 2.101775646209717, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6990575790405273, + "num_tokens": 65695465.0, + "step": 2545 + }, + { + "epoch": 0.2795958708543817, + "grad_norm": 2.437445878982544, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6800665855407715, + "num_tokens": 65717539.0, + "step": 2546 + }, + { + "epoch": 0.2797056885569954, + "grad_norm": 2.3331239223480225, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6862468719482422, + "num_tokens": 65740978.0, + "step": 2547 + }, + { + "epoch": 0.27981550625960905, + "grad_norm": 2.0901613235473633, + "learning_rate": 1e-06, + "loss": 1.1039, + "mean_token_accuracy": 0.669381320476532, + "num_tokens": 65773097.0, + "step": 2548 + }, + { + "epoch": 0.2799253239622227, + "grad_norm": 2.2576792240142822, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7083859443664551, + "num_tokens": 65799034.0, + "step": 2549 + }, + { + "epoch": 0.2800351416648364, + "grad_norm": 2.289743423461914, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7001974582672119, + "num_tokens": 65824572.0, + "step": 2550 + }, + { + "epoch": 0.28014495936745004, + "grad_norm": 2.6121091842651367, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.689940333366394, + "num_tokens": 65845306.0, + "step": 2551 + }, + { + "epoch": 0.2802547770700637, + "grad_norm": 2.3156750202178955, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6865677833557129, + "num_tokens": 65869423.0, + "step": 2552 + }, + { + "epoch": 0.28036459477267733, + "grad_norm": 2.1314949989318848, + "learning_rate": 1e-06, + "loss": 1.1039, + "mean_token_accuracy": 0.6723768711090088, + "num_tokens": 65898402.0, + "step": 2553 + }, + { + "epoch": 0.28047441247529104, + "grad_norm": 2.0236964225769043, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6868985891342163, + "num_tokens": 65929264.0, + "step": 2554 + }, + { + "epoch": 0.2805842301779047, + "grad_norm": 2.3363826274871826, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7086492776870728, + "num_tokens": 65952261.0, + "step": 2555 + }, + { + "epoch": 0.2806940478805183, + "grad_norm": 2.4121177196502686, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6921762824058533, + "num_tokens": 65974770.0, + "step": 2556 + }, + { + "epoch": 0.28080386558313203, + "grad_norm": 2.147557258605957, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7008488178253174, + "num_tokens": 66003282.0, + "step": 2557 + }, + { + "epoch": 0.2809136832857457, + "grad_norm": 2.4927659034729004, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6741272211074829, + "num_tokens": 66026201.0, + "step": 2558 + }, + { + "epoch": 0.2810235009883593, + "grad_norm": 2.175222873687744, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.713988721370697, + "num_tokens": 66050667.0, + "step": 2559 + }, + { + "epoch": 0.28113331869097297, + "grad_norm": 2.3985259532928467, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7079263925552368, + "num_tokens": 66072948.0, + "step": 2560 + }, + { + "epoch": 0.28124313639358667, + "grad_norm": 2.1836960315704346, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.705634355545044, + "num_tokens": 66098205.0, + "step": 2561 + }, + { + "epoch": 0.2813529540962003, + "grad_norm": 2.148714065551758, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6821631193161011, + "num_tokens": 66125310.0, + "step": 2562 + }, + { + "epoch": 0.28146277179881396, + "grad_norm": 2.687765598297119, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7110499739646912, + "num_tokens": 66144457.0, + "step": 2563 + }, + { + "epoch": 0.2815725895014276, + "grad_norm": 2.1269304752349854, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6806932091712952, + "num_tokens": 66174887.0, + "step": 2564 + }, + { + "epoch": 0.2816824072040413, + "grad_norm": 2.2884395122528076, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7154484391212463, + "num_tokens": 66198774.0, + "step": 2565 + }, + { + "epoch": 0.28179222490665495, + "grad_norm": 2.437760353088379, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7070876955986023, + "num_tokens": 66220080.0, + "step": 2566 + }, + { + "epoch": 0.2819020426092686, + "grad_norm": 2.3717041015625, + "learning_rate": 1e-06, + "loss": 1.0978, + "mean_token_accuracy": 0.6815823912620544, + "num_tokens": 66243774.0, + "step": 2567 + }, + { + "epoch": 0.2820118603118823, + "grad_norm": 1.9535282850265503, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6896552443504333, + "num_tokens": 66273907.0, + "step": 2568 + }, + { + "epoch": 0.28212167801449595, + "grad_norm": 2.4968531131744385, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.715283989906311, + "num_tokens": 66294826.0, + "step": 2569 + }, + { + "epoch": 0.2822314957171096, + "grad_norm": 2.259622097015381, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.697750449180603, + "num_tokens": 66319302.0, + "step": 2570 + }, + { + "epoch": 0.28234131341972324, + "grad_norm": 2.2630255222320557, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7060351371765137, + "num_tokens": 66343438.0, + "step": 2571 + }, + { + "epoch": 0.28245113112233694, + "grad_norm": 2.365591526031494, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7022374868392944, + "num_tokens": 66366792.0, + "step": 2572 + }, + { + "epoch": 0.2825609488249506, + "grad_norm": 2.2338056564331055, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7295728921890259, + "num_tokens": 66389685.0, + "step": 2573 + }, + { + "epoch": 0.28267076652756423, + "grad_norm": 2.578688383102417, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6823858618736267, + "num_tokens": 66411922.0, + "step": 2574 + }, + { + "epoch": 0.2827805842301779, + "grad_norm": 2.124737024307251, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6903795003890991, + "num_tokens": 66438626.0, + "step": 2575 + }, + { + "epoch": 0.2828904019327916, + "grad_norm": 2.3411967754364014, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7038688659667969, + "num_tokens": 66461697.0, + "step": 2576 + }, + { + "epoch": 0.2830002196354052, + "grad_norm": 2.4276111125946045, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7067002058029175, + "num_tokens": 66484447.0, + "step": 2577 + }, + { + "epoch": 0.28311003733801887, + "grad_norm": 2.201432466506958, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.7016881704330444, + "num_tokens": 66511614.0, + "step": 2578 + }, + { + "epoch": 0.2832198550406326, + "grad_norm": 2.400590658187866, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7040215730667114, + "num_tokens": 66535639.0, + "step": 2579 + }, + { + "epoch": 0.2833296727432462, + "grad_norm": 2.3145411014556885, + "learning_rate": 1e-06, + "loss": 1.0709, + "mean_token_accuracy": 0.6825648546218872, + "num_tokens": 66562218.0, + "step": 2580 + }, + { + "epoch": 0.28343949044585987, + "grad_norm": 1.9729453325271606, + "learning_rate": 1e-06, + "loss": 1.0837, + "mean_token_accuracy": 0.6796630620956421, + "num_tokens": 66593955.0, + "step": 2581 + }, + { + "epoch": 0.2835493081484735, + "grad_norm": 2.1628224849700928, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.708812415599823, + "num_tokens": 66623592.0, + "step": 2582 + }, + { + "epoch": 0.2836591258510872, + "grad_norm": 2.260499954223633, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7158912420272827, + "num_tokens": 66647465.0, + "step": 2583 + }, + { + "epoch": 0.28376894355370086, + "grad_norm": 2.085193157196045, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7065041065216064, + "num_tokens": 66674798.0, + "step": 2584 + }, + { + "epoch": 0.2838787612563145, + "grad_norm": 2.5815367698669434, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.693892240524292, + "num_tokens": 66695552.0, + "step": 2585 + }, + { + "epoch": 0.2839885789589282, + "grad_norm": 2.1700692176818848, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7124919891357422, + "num_tokens": 66723633.0, + "step": 2586 + }, + { + "epoch": 0.28409839666154185, + "grad_norm": 1.8919435739517212, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7097610235214233, + "num_tokens": 66757430.0, + "step": 2587 + }, + { + "epoch": 0.2842082143641555, + "grad_norm": 2.153498411178589, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7057408094406128, + "num_tokens": 66783683.0, + "step": 2588 + }, + { + "epoch": 0.28431803206676914, + "grad_norm": 2.4273853302001953, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7006006240844727, + "num_tokens": 66805762.0, + "step": 2589 + }, + { + "epoch": 0.28442784976938285, + "grad_norm": 2.0840859413146973, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.7013078927993774, + "num_tokens": 66835964.0, + "step": 2590 + }, + { + "epoch": 0.2845376674719965, + "grad_norm": 2.200125217437744, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6963709592819214, + "num_tokens": 66861279.0, + "step": 2591 + }, + { + "epoch": 0.28464748517461014, + "grad_norm": 2.1950013637542725, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.685056746006012, + "num_tokens": 66887549.0, + "step": 2592 + }, + { + "epoch": 0.2847573028772238, + "grad_norm": 2.2132632732391357, + "learning_rate": 1e-06, + "loss": 1.0953, + "mean_token_accuracy": 0.6750653982162476, + "num_tokens": 66917139.0, + "step": 2593 + }, + { + "epoch": 0.2848671205798375, + "grad_norm": 2.3421106338500977, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6979560256004333, + "num_tokens": 66940399.0, + "step": 2594 + }, + { + "epoch": 0.28497693828245113, + "grad_norm": 2.401339530944824, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6930056810379028, + "num_tokens": 66964330.0, + "step": 2595 + }, + { + "epoch": 0.2850867559850648, + "grad_norm": 2.3397421836853027, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6940571069717407, + "num_tokens": 66988872.0, + "step": 2596 + }, + { + "epoch": 0.2851965736876785, + "grad_norm": 2.256643533706665, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7123860716819763, + "num_tokens": 67014141.0, + "step": 2597 + }, + { + "epoch": 0.2853063913902921, + "grad_norm": 2.1164281368255615, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.699055552482605, + "num_tokens": 67042288.0, + "step": 2598 + }, + { + "epoch": 0.28541620909290577, + "grad_norm": 2.373565196990967, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7143242955207825, + "num_tokens": 67065916.0, + "step": 2599 + }, + { + "epoch": 0.2855260267955194, + "grad_norm": 2.213829278945923, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6897209882736206, + "num_tokens": 67094026.0, + "step": 2600 + }, + { + "epoch": 0.2856358444981331, + "grad_norm": 2.176093816757202, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7100214958190918, + "num_tokens": 67120199.0, + "step": 2601 + }, + { + "epoch": 0.28574566220074676, + "grad_norm": 2.2092909812927246, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7025542855262756, + "num_tokens": 67144772.0, + "step": 2602 + }, + { + "epoch": 0.2858554799033604, + "grad_norm": 2.0767171382904053, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7267699241638184, + "num_tokens": 67171811.0, + "step": 2603 + }, + { + "epoch": 0.28596529760597406, + "grad_norm": 2.4887278079986572, + "learning_rate": 1e-06, + "loss": 1.0944, + "mean_token_accuracy": 0.6781119704246521, + "num_tokens": 67194424.0, + "step": 2604 + }, + { + "epoch": 0.28607511530858776, + "grad_norm": 2.1250030994415283, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.7023627758026123, + "num_tokens": 67222204.0, + "step": 2605 + }, + { + "epoch": 0.2861849330112014, + "grad_norm": 2.2528886795043945, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.705359935760498, + "num_tokens": 67247733.0, + "step": 2606 + }, + { + "epoch": 0.28629475071381505, + "grad_norm": 2.2584524154663086, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7105564475059509, + "num_tokens": 67270826.0, + "step": 2607 + }, + { + "epoch": 0.28640456841642875, + "grad_norm": 2.3164141178131104, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6923186779022217, + "num_tokens": 67295856.0, + "step": 2608 + }, + { + "epoch": 0.2865143861190424, + "grad_norm": 2.4190263748168945, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7194897532463074, + "num_tokens": 67317337.0, + "step": 2609 + }, + { + "epoch": 0.28662420382165604, + "grad_norm": 2.485265016555786, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7059313654899597, + "num_tokens": 67338239.0, + "step": 2610 + }, + { + "epoch": 0.2867340215242697, + "grad_norm": 2.1425516605377197, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6946505904197693, + "num_tokens": 67366687.0, + "step": 2611 + }, + { + "epoch": 0.2868438392268834, + "grad_norm": 1.963449478149414, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7076987028121948, + "num_tokens": 67396042.0, + "step": 2612 + }, + { + "epoch": 0.28695365692949704, + "grad_norm": 2.5563106536865234, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7092863321304321, + "num_tokens": 67417191.0, + "step": 2613 + }, + { + "epoch": 0.2870634746321107, + "grad_norm": 2.2950148582458496, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6913425922393799, + "num_tokens": 67442631.0, + "step": 2614 + }, + { + "epoch": 0.2871732923347244, + "grad_norm": 2.331770658493042, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7020630240440369, + "num_tokens": 67467379.0, + "step": 2615 + }, + { + "epoch": 0.28728311003733803, + "grad_norm": 2.241443634033203, + "learning_rate": 1e-06, + "loss": 1.1015, + "mean_token_accuracy": 0.6798645853996277, + "num_tokens": 67494142.0, + "step": 2616 + }, + { + "epoch": 0.2873929277399517, + "grad_norm": 2.553576707839966, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.689181387424469, + "num_tokens": 67514219.0, + "step": 2617 + }, + { + "epoch": 0.2875027454425653, + "grad_norm": 2.4935684204101562, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6898306012153625, + "num_tokens": 67535726.0, + "step": 2618 + }, + { + "epoch": 0.287612563145179, + "grad_norm": 2.5931625366210938, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6948595643043518, + "num_tokens": 67558408.0, + "step": 2619 + }, + { + "epoch": 0.28772238084779267, + "grad_norm": 2.2286322116851807, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6868898868560791, + "num_tokens": 67583233.0, + "step": 2620 + }, + { + "epoch": 0.2878321985504063, + "grad_norm": 2.4276323318481445, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7168546915054321, + "num_tokens": 67606442.0, + "step": 2621 + }, + { + "epoch": 0.28794201625301996, + "grad_norm": 2.4342172145843506, + "learning_rate": 1e-06, + "loss": 1.0915, + "mean_token_accuracy": 0.6789406538009644, + "num_tokens": 67632559.0, + "step": 2622 + }, + { + "epoch": 0.28805183395563366, + "grad_norm": 2.390414237976074, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.72468101978302, + "num_tokens": 67654695.0, + "step": 2623 + }, + { + "epoch": 0.2881616516582473, + "grad_norm": 2.4956817626953125, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6915521621704102, + "num_tokens": 67677728.0, + "step": 2624 + }, + { + "epoch": 0.28827146936086095, + "grad_norm": 2.0976805686950684, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.702155590057373, + "num_tokens": 67705239.0, + "step": 2625 + }, + { + "epoch": 0.28838128706347466, + "grad_norm": 2.1980302333831787, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6827535629272461, + "num_tokens": 67732491.0, + "step": 2626 + }, + { + "epoch": 0.2884911047660883, + "grad_norm": 2.4034924507141113, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7080445885658264, + "num_tokens": 67755591.0, + "step": 2627 + }, + { + "epoch": 0.28860092246870195, + "grad_norm": 2.2033307552337646, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7043716907501221, + "num_tokens": 67780989.0, + "step": 2628 + }, + { + "epoch": 0.2887107401713156, + "grad_norm": 2.1411662101745605, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6842848062515259, + "num_tokens": 67809126.0, + "step": 2629 + }, + { + "epoch": 0.2888205578739293, + "grad_norm": 2.3213489055633545, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7022768259048462, + "num_tokens": 67831633.0, + "step": 2630 + }, + { + "epoch": 0.28893037557654294, + "grad_norm": 2.204519510269165, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6818538904190063, + "num_tokens": 67857770.0, + "step": 2631 + }, + { + "epoch": 0.2890401932791566, + "grad_norm": 2.1888880729675293, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6885212659835815, + "num_tokens": 67886441.0, + "step": 2632 + }, + { + "epoch": 0.2891500109817703, + "grad_norm": 2.247835397720337, + "learning_rate": 1e-06, + "loss": 1.1076, + "mean_token_accuracy": 0.6755766868591309, + "num_tokens": 67914327.0, + "step": 2633 + }, + { + "epoch": 0.28925982868438394, + "grad_norm": 2.064044952392578, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6999560594558716, + "num_tokens": 67945120.0, + "step": 2634 + }, + { + "epoch": 0.2893696463869976, + "grad_norm": 2.223266363143921, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6976916790008545, + "num_tokens": 67970817.0, + "step": 2635 + }, + { + "epoch": 0.2894794640896112, + "grad_norm": 2.3710250854492188, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6963531374931335, + "num_tokens": 67993845.0, + "step": 2636 + }, + { + "epoch": 0.28958928179222493, + "grad_norm": 2.346996784210205, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.699607789516449, + "num_tokens": 68018204.0, + "step": 2637 + }, + { + "epoch": 0.2896990994948386, + "grad_norm": 2.3867878913879395, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7011164426803589, + "num_tokens": 68041796.0, + "step": 2638 + }, + { + "epoch": 0.2898089171974522, + "grad_norm": 2.223611354827881, + "learning_rate": 1e-06, + "loss": 1.074, + "mean_token_accuracy": 0.6803260445594788, + "num_tokens": 68069582.0, + "step": 2639 + }, + { + "epoch": 0.28991873490006587, + "grad_norm": 2.2359540462493896, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7022632360458374, + "num_tokens": 68095557.0, + "step": 2640 + }, + { + "epoch": 0.29002855260267957, + "grad_norm": 2.203047037124634, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7005206942558289, + "num_tokens": 68120918.0, + "step": 2641 + }, + { + "epoch": 0.2901383703052932, + "grad_norm": 1.931340217590332, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7208207845687866, + "num_tokens": 68153190.0, + "step": 2642 + }, + { + "epoch": 0.29024818800790686, + "grad_norm": 2.2202887535095215, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7199205756187439, + "num_tokens": 68178353.0, + "step": 2643 + }, + { + "epoch": 0.29035800571052056, + "grad_norm": 2.375102996826172, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7117083072662354, + "num_tokens": 68200514.0, + "step": 2644 + }, + { + "epoch": 0.2904678234131342, + "grad_norm": 2.1600801944732666, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6963441967964172, + "num_tokens": 68228297.0, + "step": 2645 + }, + { + "epoch": 0.29057764111574785, + "grad_norm": 2.333447217941284, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.682593047618866, + "num_tokens": 68251957.0, + "step": 2646 + }, + { + "epoch": 0.2906874588183615, + "grad_norm": 2.0088634490966797, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6887153387069702, + "num_tokens": 68283647.0, + "step": 2647 + }, + { + "epoch": 0.2907972765209752, + "grad_norm": 2.1340534687042236, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7096517086029053, + "num_tokens": 68309475.0, + "step": 2648 + }, + { + "epoch": 0.29090709422358885, + "grad_norm": 2.4744374752044678, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6928961277008057, + "num_tokens": 68333046.0, + "step": 2649 + }, + { + "epoch": 0.2910169119262025, + "grad_norm": 2.130657434463501, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6924093961715698, + "num_tokens": 68360721.0, + "step": 2650 + }, + { + "epoch": 0.29112672962881614, + "grad_norm": 2.4120216369628906, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6887108683586121, + "num_tokens": 68386221.0, + "step": 2651 + }, + { + "epoch": 0.29123654733142984, + "grad_norm": 2.3607819080352783, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6987438201904297, + "num_tokens": 68410661.0, + "step": 2652 + }, + { + "epoch": 0.2913463650340435, + "grad_norm": 2.4011919498443604, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7006452083587646, + "num_tokens": 68432343.0, + "step": 2653 + }, + { + "epoch": 0.29145618273665713, + "grad_norm": 2.5320260524749756, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7228800654411316, + "num_tokens": 68452569.0, + "step": 2654 + }, + { + "epoch": 0.29156600043927083, + "grad_norm": 2.1586997509002686, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7025578022003174, + "num_tokens": 68478721.0, + "step": 2655 + }, + { + "epoch": 0.2916758181418845, + "grad_norm": 2.710160732269287, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7103928327560425, + "num_tokens": 68499419.0, + "step": 2656 + }, + { + "epoch": 0.2917856358444981, + "grad_norm": 2.322770357131958, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6927885413169861, + "num_tokens": 68524812.0, + "step": 2657 + }, + { + "epoch": 0.29189545354711177, + "grad_norm": 2.409968137741089, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7014490365982056, + "num_tokens": 68546903.0, + "step": 2658 + }, + { + "epoch": 0.2920052712497255, + "grad_norm": 2.384960412979126, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6964099407196045, + "num_tokens": 68569572.0, + "step": 2659 + }, + { + "epoch": 0.2921150889523391, + "grad_norm": 2.3191311359405518, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7190160751342773, + "num_tokens": 68591913.0, + "step": 2660 + }, + { + "epoch": 0.29222490665495277, + "grad_norm": 2.2007641792297363, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6878015995025635, + "num_tokens": 68615429.0, + "step": 2661 + }, + { + "epoch": 0.29233472435756647, + "grad_norm": 2.291205644607544, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7113603353500366, + "num_tokens": 68640403.0, + "step": 2662 + }, + { + "epoch": 0.2924445420601801, + "grad_norm": 2.3328824043273926, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6904075145721436, + "num_tokens": 68664399.0, + "step": 2663 + }, + { + "epoch": 0.29255435976279376, + "grad_norm": 2.2064273357391357, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7246904969215393, + "num_tokens": 68689416.0, + "step": 2664 + }, + { + "epoch": 0.2926641774654074, + "grad_norm": 2.3423848152160645, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6991218328475952, + "num_tokens": 68713108.0, + "step": 2665 + }, + { + "epoch": 0.2927739951680211, + "grad_norm": 2.3813209533691406, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7124394178390503, + "num_tokens": 68734377.0, + "step": 2666 + }, + { + "epoch": 0.29288381287063475, + "grad_norm": 1.9821178913116455, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6843799352645874, + "num_tokens": 68765414.0, + "step": 2667 + }, + { + "epoch": 0.2929936305732484, + "grad_norm": 2.097163200378418, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6954277753829956, + "num_tokens": 68795093.0, + "step": 2668 + }, + { + "epoch": 0.29310344827586204, + "grad_norm": 2.0508124828338623, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6985109448432922, + "num_tokens": 68824825.0, + "step": 2669 + }, + { + "epoch": 0.29321326597847575, + "grad_norm": 2.788397789001465, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7096642255783081, + "num_tokens": 68842665.0, + "step": 2670 + }, + { + "epoch": 0.2933230836810894, + "grad_norm": 2.1630160808563232, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7051178216934204, + "num_tokens": 68868157.0, + "step": 2671 + }, + { + "epoch": 0.29343290138370304, + "grad_norm": 2.0960469245910645, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6867610216140747, + "num_tokens": 68896742.0, + "step": 2672 + }, + { + "epoch": 0.29354271908631674, + "grad_norm": 2.6828925609588623, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7038623094558716, + "num_tokens": 68915098.0, + "step": 2673 + }, + { + "epoch": 0.2936525367889304, + "grad_norm": 2.145738124847412, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7006456255912781, + "num_tokens": 68942427.0, + "step": 2674 + }, + { + "epoch": 0.29376235449154403, + "grad_norm": 2.2814128398895264, + "learning_rate": 1e-06, + "loss": 1.1113, + "mean_token_accuracy": 0.6725578904151917, + "num_tokens": 68968988.0, + "step": 2675 + }, + { + "epoch": 0.2938721721941577, + "grad_norm": 2.436642646789551, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6948084831237793, + "num_tokens": 68992486.0, + "step": 2676 + }, + { + "epoch": 0.2939819898967714, + "grad_norm": 2.118044853210449, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6840850114822388, + "num_tokens": 69024672.0, + "step": 2677 + }, + { + "epoch": 0.294091807599385, + "grad_norm": 2.5582597255706787, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7049447298049927, + "num_tokens": 69045188.0, + "step": 2678 + }, + { + "epoch": 0.29420162530199867, + "grad_norm": 2.3956658840179443, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7028151750564575, + "num_tokens": 69066225.0, + "step": 2679 + }, + { + "epoch": 0.2943114430046123, + "grad_norm": 2.4065656661987305, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6809734106063843, + "num_tokens": 69089660.0, + "step": 2680 + }, + { + "epoch": 0.294421260707226, + "grad_norm": 2.1626100540161133, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6899315714836121, + "num_tokens": 69117938.0, + "step": 2681 + }, + { + "epoch": 0.29453107840983966, + "grad_norm": 2.4587416648864746, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6930258870124817, + "num_tokens": 69137885.0, + "step": 2682 + }, + { + "epoch": 0.2946408961124533, + "grad_norm": 2.2224113941192627, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7081127762794495, + "num_tokens": 69163961.0, + "step": 2683 + }, + { + "epoch": 0.294750713815067, + "grad_norm": 2.2892563343048096, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6968937516212463, + "num_tokens": 69188259.0, + "step": 2684 + }, + { + "epoch": 0.29486053151768066, + "grad_norm": 2.402299165725708, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6766247153282166, + "num_tokens": 69210232.0, + "step": 2685 + }, + { + "epoch": 0.2949703492202943, + "grad_norm": 2.3640005588531494, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.720025897026062, + "num_tokens": 69233003.0, + "step": 2686 + }, + { + "epoch": 0.29508016692290795, + "grad_norm": 2.387371301651001, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7062334418296814, + "num_tokens": 69256133.0, + "step": 2687 + }, + { + "epoch": 0.29518998462552165, + "grad_norm": 2.2386443614959717, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.696399986743927, + "num_tokens": 69281804.0, + "step": 2688 + }, + { + "epoch": 0.2952998023281353, + "grad_norm": 2.139017343521118, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7022611498832703, + "num_tokens": 69308408.0, + "step": 2689 + }, + { + "epoch": 0.29540962003074894, + "grad_norm": 2.7666382789611816, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6943890452384949, + "num_tokens": 69325930.0, + "step": 2690 + }, + { + "epoch": 0.29551943773336264, + "grad_norm": 2.206265449523926, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6870394945144653, + "num_tokens": 69353450.0, + "step": 2691 + }, + { + "epoch": 0.2956292554359763, + "grad_norm": 2.17419171333313, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7055540084838867, + "num_tokens": 69379233.0, + "step": 2692 + }, + { + "epoch": 0.29573907313858994, + "grad_norm": 2.1470932960510254, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6830829381942749, + "num_tokens": 69408392.0, + "step": 2693 + }, + { + "epoch": 0.2958488908412036, + "grad_norm": 2.2144505977630615, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7011674642562866, + "num_tokens": 69435176.0, + "step": 2694 + }, + { + "epoch": 0.2959587085438173, + "grad_norm": 2.3862416744232178, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.702136754989624, + "num_tokens": 69458027.0, + "step": 2695 + }, + { + "epoch": 0.29606852624643093, + "grad_norm": 2.220097780227661, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7293866872787476, + "num_tokens": 69483270.0, + "step": 2696 + }, + { + "epoch": 0.2961783439490446, + "grad_norm": 2.2471225261688232, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6999580264091492, + "num_tokens": 69508542.0, + "step": 2697 + }, + { + "epoch": 0.2962881616516582, + "grad_norm": 2.0562124252319336, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6951630115509033, + "num_tokens": 69537563.0, + "step": 2698 + }, + { + "epoch": 0.2963979793542719, + "grad_norm": 2.5453927516937256, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6877301931381226, + "num_tokens": 69559403.0, + "step": 2699 + }, + { + "epoch": 0.29650779705688557, + "grad_norm": 2.2230398654937744, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6962752342224121, + "num_tokens": 69584540.0, + "step": 2700 + }, + { + "epoch": 0.2966176147594992, + "grad_norm": 2.1292665004730225, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6861860752105713, + "num_tokens": 69611803.0, + "step": 2701 + }, + { + "epoch": 0.2967274324621129, + "grad_norm": 2.1745944023132324, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7219661474227905, + "num_tokens": 69636926.0, + "step": 2702 + }, + { + "epoch": 0.29683725016472656, + "grad_norm": 2.83821702003479, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7117694020271301, + "num_tokens": 69653782.0, + "step": 2703 + }, + { + "epoch": 0.2969470678673402, + "grad_norm": 1.9634357690811157, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7230681777000427, + "num_tokens": 69682931.0, + "step": 2704 + }, + { + "epoch": 0.29705688556995385, + "grad_norm": 2.2856178283691406, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6792352199554443, + "num_tokens": 69709862.0, + "step": 2705 + }, + { + "epoch": 0.29716670327256756, + "grad_norm": 2.5722649097442627, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6871403455734253, + "num_tokens": 69730466.0, + "step": 2706 + }, + { + "epoch": 0.2972765209751812, + "grad_norm": 2.3229165077209473, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6840651035308838, + "num_tokens": 69755783.0, + "step": 2707 + }, + { + "epoch": 0.29738633867779485, + "grad_norm": 2.3054113388061523, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6997783184051514, + "num_tokens": 69781242.0, + "step": 2708 + }, + { + "epoch": 0.29749615638040855, + "grad_norm": 2.4651970863342285, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7114197015762329, + "num_tokens": 69802156.0, + "step": 2709 + }, + { + "epoch": 0.2976059740830222, + "grad_norm": 2.2183268070220947, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6876435875892639, + "num_tokens": 69829512.0, + "step": 2710 + }, + { + "epoch": 0.29771579178563584, + "grad_norm": 2.1347506046295166, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.698749840259552, + "num_tokens": 69859971.0, + "step": 2711 + }, + { + "epoch": 0.2978256094882495, + "grad_norm": 2.1063497066497803, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.69533371925354, + "num_tokens": 69887667.0, + "step": 2712 + }, + { + "epoch": 0.2979354271908632, + "grad_norm": 2.6357662677764893, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7176467180252075, + "num_tokens": 69905581.0, + "step": 2713 + }, + { + "epoch": 0.29804524489347684, + "grad_norm": 2.4674906730651855, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6819334030151367, + "num_tokens": 69927844.0, + "step": 2714 + }, + { + "epoch": 0.2981550625960905, + "grad_norm": 2.1413938999176025, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7055878043174744, + "num_tokens": 69954818.0, + "step": 2715 + }, + { + "epoch": 0.2982648802987041, + "grad_norm": 2.5320301055908203, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6805291771888733, + "num_tokens": 69978817.0, + "step": 2716 + }, + { + "epoch": 0.29837469800131783, + "grad_norm": 2.0645411014556885, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6873137950897217, + "num_tokens": 70010015.0, + "step": 2717 + }, + { + "epoch": 0.2984845157039315, + "grad_norm": 2.4104106426239014, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7120929956436157, + "num_tokens": 70031369.0, + "step": 2718 + }, + { + "epoch": 0.2985943334065451, + "grad_norm": 2.549041986465454, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7115961909294128, + "num_tokens": 70052473.0, + "step": 2719 + }, + { + "epoch": 0.2987041511091588, + "grad_norm": 2.4791243076324463, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6817399263381958, + "num_tokens": 70076647.0, + "step": 2720 + }, + { + "epoch": 0.29881396881177247, + "grad_norm": 2.2518842220306396, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6953387260437012, + "num_tokens": 70104353.0, + "step": 2721 + }, + { + "epoch": 0.2989237865143861, + "grad_norm": 2.367053985595703, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.691548764705658, + "num_tokens": 70128787.0, + "step": 2722 + }, + { + "epoch": 0.29903360421699976, + "grad_norm": 2.2362685203552246, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7174910306930542, + "num_tokens": 70151878.0, + "step": 2723 + }, + { + "epoch": 0.29914342191961346, + "grad_norm": 2.202958345413208, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7147876024246216, + "num_tokens": 70178201.0, + "step": 2724 + }, + { + "epoch": 0.2992532396222271, + "grad_norm": 2.233001470565796, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7002643942832947, + "num_tokens": 70205725.0, + "step": 2725 + }, + { + "epoch": 0.29936305732484075, + "grad_norm": 2.4496853351593018, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7004756927490234, + "num_tokens": 70227895.0, + "step": 2726 + }, + { + "epoch": 0.2994728750274544, + "grad_norm": 2.8317480087280273, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7137298583984375, + "num_tokens": 70244977.0, + "step": 2727 + }, + { + "epoch": 0.2995826927300681, + "grad_norm": 2.4258553981781006, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6911086440086365, + "num_tokens": 70269627.0, + "step": 2728 + }, + { + "epoch": 0.29969251043268175, + "grad_norm": 2.4806737899780273, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6942819356918335, + "num_tokens": 70291870.0, + "step": 2729 + }, + { + "epoch": 0.2998023281352954, + "grad_norm": 2.244109869003296, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7015226483345032, + "num_tokens": 70317252.0, + "step": 2730 + }, + { + "epoch": 0.2999121458379091, + "grad_norm": 2.288684606552124, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.7041149735450745, + "num_tokens": 70343823.0, + "step": 2731 + }, + { + "epoch": 0.30002196354052274, + "grad_norm": 2.3013617992401123, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7041323781013489, + "num_tokens": 70370000.0, + "step": 2732 + }, + { + "epoch": 0.3001317812431364, + "grad_norm": 2.4557807445526123, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6974819898605347, + "num_tokens": 70393407.0, + "step": 2733 + }, + { + "epoch": 0.30024159894575003, + "grad_norm": 2.487149715423584, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6957966089248657, + "num_tokens": 70418018.0, + "step": 2734 + }, + { + "epoch": 0.30035141664836373, + "grad_norm": 1.8632709980010986, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7033672332763672, + "num_tokens": 70455148.0, + "step": 2735 + }, + { + "epoch": 0.3004612343509774, + "grad_norm": 2.153296709060669, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6865923404693604, + "num_tokens": 70484371.0, + "step": 2736 + }, + { + "epoch": 0.300571052053591, + "grad_norm": 2.051725149154663, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7290180325508118, + "num_tokens": 70512916.0, + "step": 2737 + }, + { + "epoch": 0.3006808697562047, + "grad_norm": 2.3819894790649414, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7281641960144043, + "num_tokens": 70535045.0, + "step": 2738 + }, + { + "epoch": 0.3007906874588184, + "grad_norm": 2.2281999588012695, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6985800266265869, + "num_tokens": 70558665.0, + "step": 2739 + }, + { + "epoch": 0.300900505161432, + "grad_norm": 2.0477380752563477, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.687857985496521, + "num_tokens": 70589865.0, + "step": 2740 + }, + { + "epoch": 0.30101032286404567, + "grad_norm": 2.2256107330322266, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6805232167243958, + "num_tokens": 70618598.0, + "step": 2741 + }, + { + "epoch": 0.30112014056665937, + "grad_norm": 2.2830958366394043, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6729423999786377, + "num_tokens": 70644287.0, + "step": 2742 + }, + { + "epoch": 0.301229958269273, + "grad_norm": 2.0783236026763916, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6903148889541626, + "num_tokens": 70672724.0, + "step": 2743 + }, + { + "epoch": 0.30133977597188666, + "grad_norm": 2.3744070529937744, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7054898738861084, + "num_tokens": 70695836.0, + "step": 2744 + }, + { + "epoch": 0.3014495936745003, + "grad_norm": 1.9956474304199219, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6868711709976196, + "num_tokens": 70725296.0, + "step": 2745 + }, + { + "epoch": 0.301559411377114, + "grad_norm": 2.4973838329315186, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7124828696250916, + "num_tokens": 70747078.0, + "step": 2746 + }, + { + "epoch": 0.30166922907972765, + "grad_norm": 2.6127004623413086, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7075321674346924, + "num_tokens": 70767188.0, + "step": 2747 + }, + { + "epoch": 0.3017790467823413, + "grad_norm": 2.5398857593536377, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.679133951663971, + "num_tokens": 70791071.0, + "step": 2748 + }, + { + "epoch": 0.301888864484955, + "grad_norm": 2.419086217880249, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6870347857475281, + "num_tokens": 70814519.0, + "step": 2749 + }, + { + "epoch": 0.30199868218756865, + "grad_norm": 2.1853859424591064, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6969404816627502, + "num_tokens": 70844092.0, + "step": 2750 + }, + { + "epoch": 0.3021084998901823, + "grad_norm": 2.3377127647399902, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6890011429786682, + "num_tokens": 70867672.0, + "step": 2751 + }, + { + "epoch": 0.30221831759279594, + "grad_norm": 2.4081897735595703, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6858944892883301, + "num_tokens": 70891506.0, + "step": 2752 + }, + { + "epoch": 0.30232813529540964, + "grad_norm": 2.049337863922119, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6932470798492432, + "num_tokens": 70923128.0, + "step": 2753 + }, + { + "epoch": 0.3024379529980233, + "grad_norm": 2.1541388034820557, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7093973159790039, + "num_tokens": 70949698.0, + "step": 2754 + }, + { + "epoch": 0.30254777070063693, + "grad_norm": 2.0327229499816895, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.717363715171814, + "num_tokens": 70979202.0, + "step": 2755 + }, + { + "epoch": 0.3026575884032506, + "grad_norm": 2.2072882652282715, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6931127309799194, + "num_tokens": 71005524.0, + "step": 2756 + }, + { + "epoch": 0.3027674061058643, + "grad_norm": 2.300628423690796, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6961116790771484, + "num_tokens": 71029264.0, + "step": 2757 + }, + { + "epoch": 0.3028772238084779, + "grad_norm": 1.9655823707580566, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7033897638320923, + "num_tokens": 71059244.0, + "step": 2758 + }, + { + "epoch": 0.30298704151109157, + "grad_norm": 2.1542365550994873, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7178255319595337, + "num_tokens": 71087818.0, + "step": 2759 + }, + { + "epoch": 0.30309685921370527, + "grad_norm": 2.618834972381592, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7173728942871094, + "num_tokens": 71105927.0, + "step": 2760 + }, + { + "epoch": 0.3032066769163189, + "grad_norm": 2.2412726879119873, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.710709810256958, + "num_tokens": 71131730.0, + "step": 2761 + }, + { + "epoch": 0.30331649461893256, + "grad_norm": 2.369716167449951, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6868394017219543, + "num_tokens": 71158313.0, + "step": 2762 + }, + { + "epoch": 0.3034263123215462, + "grad_norm": 2.291417121887207, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7125017642974854, + "num_tokens": 71180428.0, + "step": 2763 + }, + { + "epoch": 0.3035361300241599, + "grad_norm": 2.383195638656616, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6838363409042358, + "num_tokens": 71203084.0, + "step": 2764 + }, + { + "epoch": 0.30364594772677356, + "grad_norm": 2.1968696117401123, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7107250690460205, + "num_tokens": 71228838.0, + "step": 2765 + }, + { + "epoch": 0.3037557654293872, + "grad_norm": 2.387272834777832, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6920168399810791, + "num_tokens": 71253617.0, + "step": 2766 + }, + { + "epoch": 0.3038655831320009, + "grad_norm": 2.19503116607666, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6960926055908203, + "num_tokens": 71282951.0, + "step": 2767 + }, + { + "epoch": 0.30397540083461455, + "grad_norm": 2.3312742710113525, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7146925926208496, + "num_tokens": 71308269.0, + "step": 2768 + }, + { + "epoch": 0.3040852185372282, + "grad_norm": 2.291191339492798, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6940736770629883, + "num_tokens": 71334265.0, + "step": 2769 + }, + { + "epoch": 0.30419503623984184, + "grad_norm": 2.4699366092681885, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6828840374946594, + "num_tokens": 71356853.0, + "step": 2770 + }, + { + "epoch": 0.30430485394245554, + "grad_norm": 2.4982407093048096, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.7014332413673401, + "num_tokens": 71379107.0, + "step": 2771 + }, + { + "epoch": 0.3044146716450692, + "grad_norm": 2.5331146717071533, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.687204897403717, + "num_tokens": 71401618.0, + "step": 2772 + }, + { + "epoch": 0.30452448934768284, + "grad_norm": 2.1086313724517822, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7041308283805847, + "num_tokens": 71430403.0, + "step": 2773 + }, + { + "epoch": 0.3046343070502965, + "grad_norm": 2.5350892543792725, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.706780195236206, + "num_tokens": 71452517.0, + "step": 2774 + }, + { + "epoch": 0.3047441247529102, + "grad_norm": 2.042313575744629, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6827452778816223, + "num_tokens": 71483294.0, + "step": 2775 + }, + { + "epoch": 0.30485394245552383, + "grad_norm": 1.9415940046310425, + "learning_rate": 1e-06, + "loss": 1.0626, + "mean_token_accuracy": 0.6848254203796387, + "num_tokens": 71516245.0, + "step": 2776 + }, + { + "epoch": 0.3049637601581375, + "grad_norm": 2.666010618209839, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7253600358963013, + "num_tokens": 71536415.0, + "step": 2777 + }, + { + "epoch": 0.3050735778607512, + "grad_norm": 2.25148344039917, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7032426595687866, + "num_tokens": 71560663.0, + "step": 2778 + }, + { + "epoch": 0.3051833955633648, + "grad_norm": 1.983282446861267, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.703040361404419, + "num_tokens": 71590181.0, + "step": 2779 + }, + { + "epoch": 0.30529321326597847, + "grad_norm": 1.9466030597686768, + "learning_rate": 1e-06, + "loss": 1.0877, + "mean_token_accuracy": 0.6815419793128967, + "num_tokens": 71624147.0, + "step": 2780 + }, + { + "epoch": 0.3054030309685921, + "grad_norm": 2.7594354152679443, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6806625127792358, + "num_tokens": 71651917.0, + "step": 2781 + }, + { + "epoch": 0.3055128486712058, + "grad_norm": 2.2035958766937256, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6913678646087646, + "num_tokens": 71679892.0, + "step": 2782 + }, + { + "epoch": 0.30562266637381946, + "grad_norm": 2.4016759395599365, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6860740780830383, + "num_tokens": 71703592.0, + "step": 2783 + }, + { + "epoch": 0.3057324840764331, + "grad_norm": 2.04240345954895, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.714634895324707, + "num_tokens": 71730045.0, + "step": 2784 + }, + { + "epoch": 0.30584230177904675, + "grad_norm": 2.0613718032836914, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7112696170806885, + "num_tokens": 71758993.0, + "step": 2785 + }, + { + "epoch": 0.30595211948166046, + "grad_norm": 1.9101316928863525, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.7039833664894104, + "num_tokens": 71794155.0, + "step": 2786 + }, + { + "epoch": 0.3060619371842741, + "grad_norm": 2.290745496749878, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6872345209121704, + "num_tokens": 71819405.0, + "step": 2787 + }, + { + "epoch": 0.30617175488688775, + "grad_norm": 2.200063467025757, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7262571454048157, + "num_tokens": 71845461.0, + "step": 2788 + }, + { + "epoch": 0.30628157258950145, + "grad_norm": 2.3075637817382812, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.696514368057251, + "num_tokens": 71870894.0, + "step": 2789 + }, + { + "epoch": 0.3063913902921151, + "grad_norm": 2.2533507347106934, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7287983894348145, + "num_tokens": 71894499.0, + "step": 2790 + }, + { + "epoch": 0.30650120799472874, + "grad_norm": 2.4730112552642822, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7277148365974426, + "num_tokens": 71915420.0, + "step": 2791 + }, + { + "epoch": 0.3066110256973424, + "grad_norm": 2.071826457977295, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7011364698410034, + "num_tokens": 71942445.0, + "step": 2792 + }, + { + "epoch": 0.3067208433999561, + "grad_norm": 2.1931231021881104, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.690630316734314, + "num_tokens": 71969314.0, + "step": 2793 + }, + { + "epoch": 0.30683066110256974, + "grad_norm": 2.437678813934326, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7119580507278442, + "num_tokens": 71990315.0, + "step": 2794 + }, + { + "epoch": 0.3069404788051834, + "grad_norm": 2.3568062782287598, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7109714150428772, + "num_tokens": 72013474.0, + "step": 2795 + }, + { + "epoch": 0.3070502965077971, + "grad_norm": 2.3373749256134033, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7020635604858398, + "num_tokens": 72038817.0, + "step": 2796 + }, + { + "epoch": 0.30716011421041073, + "grad_norm": 2.2516262531280518, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6812520623207092, + "num_tokens": 72065720.0, + "step": 2797 + }, + { + "epoch": 0.3072699319130244, + "grad_norm": 2.1771926879882812, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7100778818130493, + "num_tokens": 72093555.0, + "step": 2798 + }, + { + "epoch": 0.307379749615638, + "grad_norm": 2.343518018722534, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7035588026046753, + "num_tokens": 72117657.0, + "step": 2799 + }, + { + "epoch": 0.3074895673182517, + "grad_norm": 2.188983201980591, + "learning_rate": 1e-06, + "loss": 1.0804, + "mean_token_accuracy": 0.6901746392250061, + "num_tokens": 72146313.0, + "step": 2800 + }, + { + "epoch": 0.30759938502086537, + "grad_norm": 1.9817967414855957, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6968809366226196, + "num_tokens": 72177598.0, + "step": 2801 + }, + { + "epoch": 0.307709202723479, + "grad_norm": 2.143510341644287, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6781353950500488, + "num_tokens": 72205773.0, + "step": 2802 + }, + { + "epoch": 0.30781902042609266, + "grad_norm": 2.279268741607666, + "learning_rate": 1e-06, + "loss": 1.1385, + "mean_token_accuracy": 0.6662593483924866, + "num_tokens": 72232409.0, + "step": 2803 + }, + { + "epoch": 0.30792883812870636, + "grad_norm": 1.9949809312820435, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6846987009048462, + "num_tokens": 72264043.0, + "step": 2804 + }, + { + "epoch": 0.30803865583132, + "grad_norm": 2.5030524730682373, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7043584585189819, + "num_tokens": 72284186.0, + "step": 2805 + }, + { + "epoch": 0.30814847353393365, + "grad_norm": 2.009992837905884, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.7011421918869019, + "num_tokens": 72314080.0, + "step": 2806 + }, + { + "epoch": 0.30825829123654735, + "grad_norm": 2.345853567123413, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7242928743362427, + "num_tokens": 72336187.0, + "step": 2807 + }, + { + "epoch": 0.308368108939161, + "grad_norm": 2.0487351417541504, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7289727330207825, + "num_tokens": 72363076.0, + "step": 2808 + }, + { + "epoch": 0.30847792664177465, + "grad_norm": 2.3567516803741455, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6964386701583862, + "num_tokens": 72385579.0, + "step": 2809 + }, + { + "epoch": 0.3085877443443883, + "grad_norm": 2.204402446746826, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6919785737991333, + "num_tokens": 72410997.0, + "step": 2810 + }, + { + "epoch": 0.308697562047002, + "grad_norm": 2.1315128803253174, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6895776987075806, + "num_tokens": 72438565.0, + "step": 2811 + }, + { + "epoch": 0.30880737974961564, + "grad_norm": 2.0538008213043213, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7099101543426514, + "num_tokens": 72468051.0, + "step": 2812 + }, + { + "epoch": 0.3089171974522293, + "grad_norm": 2.084235191345215, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6768650412559509, + "num_tokens": 72499500.0, + "step": 2813 + }, + { + "epoch": 0.309027015154843, + "grad_norm": 1.9241787195205688, + "learning_rate": 1e-06, + "loss": 1.0923, + "mean_token_accuracy": 0.6734519004821777, + "num_tokens": 72534627.0, + "step": 2814 + }, + { + "epoch": 0.30913683285745663, + "grad_norm": 2.196045160293579, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6805105805397034, + "num_tokens": 72560239.0, + "step": 2815 + }, + { + "epoch": 0.3092466505600703, + "grad_norm": 2.403855562210083, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.7042680978775024, + "num_tokens": 72586042.0, + "step": 2816 + }, + { + "epoch": 0.3093564682626839, + "grad_norm": 2.188122272491455, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.7011332511901855, + "num_tokens": 72613714.0, + "step": 2817 + }, + { + "epoch": 0.3094662859652976, + "grad_norm": 2.4040629863739014, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.6994750499725342, + "num_tokens": 72635577.0, + "step": 2818 + }, + { + "epoch": 0.3095761036679113, + "grad_norm": 2.3706374168395996, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6840671896934509, + "num_tokens": 72660035.0, + "step": 2819 + }, + { + "epoch": 0.3096859213705249, + "grad_norm": 2.341973304748535, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6935346722602844, + "num_tokens": 72684013.0, + "step": 2820 + }, + { + "epoch": 0.30979573907313857, + "grad_norm": 1.959228277206421, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.687577486038208, + "num_tokens": 72717677.0, + "step": 2821 + }, + { + "epoch": 0.30990555677575227, + "grad_norm": 2.2476553916931152, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6778008937835693, + "num_tokens": 72744225.0, + "step": 2822 + }, + { + "epoch": 0.3100153744783659, + "grad_norm": 2.345348358154297, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7152806520462036, + "num_tokens": 72766822.0, + "step": 2823 + }, + { + "epoch": 0.31012519218097956, + "grad_norm": 2.166658401489258, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6970117092132568, + "num_tokens": 72792498.0, + "step": 2824 + }, + { + "epoch": 0.31023500988359326, + "grad_norm": 2.1714413166046143, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7062282562255859, + "num_tokens": 72820560.0, + "step": 2825 + }, + { + "epoch": 0.3103448275862069, + "grad_norm": 2.012526512145996, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6740922331809998, + "num_tokens": 72852398.0, + "step": 2826 + }, + { + "epoch": 0.31045464528882055, + "grad_norm": 2.649399518966675, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7223199009895325, + "num_tokens": 72869137.0, + "step": 2827 + }, + { + "epoch": 0.3105644629914342, + "grad_norm": 2.1709980964660645, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6856775283813477, + "num_tokens": 72897365.0, + "step": 2828 + }, + { + "epoch": 0.3106742806940479, + "grad_norm": 2.248919725418091, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6883786916732788, + "num_tokens": 72921887.0, + "step": 2829 + }, + { + "epoch": 0.31078409839666155, + "grad_norm": 2.0964534282684326, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6810433864593506, + "num_tokens": 72950744.0, + "step": 2830 + }, + { + "epoch": 0.3108939160992752, + "grad_norm": 2.18548583984375, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.708809494972229, + "num_tokens": 72976975.0, + "step": 2831 + }, + { + "epoch": 0.31100373380188884, + "grad_norm": 2.194673776626587, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7044822573661804, + "num_tokens": 73002969.0, + "step": 2832 + }, + { + "epoch": 0.31111355150450254, + "grad_norm": 2.295178174972534, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6973321437835693, + "num_tokens": 73029031.0, + "step": 2833 + }, + { + "epoch": 0.3112233692071162, + "grad_norm": 2.4139928817749023, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6864694952964783, + "num_tokens": 73053843.0, + "step": 2834 + }, + { + "epoch": 0.31133318690972983, + "grad_norm": 2.2853267192840576, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6785995960235596, + "num_tokens": 73078672.0, + "step": 2835 + }, + { + "epoch": 0.31144300461234353, + "grad_norm": 2.098623752593994, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7285847067832947, + "num_tokens": 73107849.0, + "step": 2836 + }, + { + "epoch": 0.3115528223149572, + "grad_norm": 2.2397079467773438, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.697033166885376, + "num_tokens": 73133804.0, + "step": 2837 + }, + { + "epoch": 0.3116626400175708, + "grad_norm": 2.292922258377075, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6920936107635498, + "num_tokens": 73158798.0, + "step": 2838 + }, + { + "epoch": 0.31177245772018447, + "grad_norm": 2.2124178409576416, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6965266466140747, + "num_tokens": 73184487.0, + "step": 2839 + }, + { + "epoch": 0.31188227542279817, + "grad_norm": 2.3234646320343018, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6962765455245972, + "num_tokens": 73206724.0, + "step": 2840 + }, + { + "epoch": 0.3119920931254118, + "grad_norm": 2.1126656532287598, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7047436237335205, + "num_tokens": 73233806.0, + "step": 2841 + }, + { + "epoch": 0.31210191082802546, + "grad_norm": 2.223536968231201, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6991193294525146, + "num_tokens": 73260450.0, + "step": 2842 + }, + { + "epoch": 0.31221172853063917, + "grad_norm": 1.8623312711715698, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6929827332496643, + "num_tokens": 73294586.0, + "step": 2843 + }, + { + "epoch": 0.3123215462332528, + "grad_norm": 2.3920810222625732, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6994614601135254, + "num_tokens": 73317507.0, + "step": 2844 + }, + { + "epoch": 0.31243136393586646, + "grad_norm": 2.1978068351745605, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6925264596939087, + "num_tokens": 73345807.0, + "step": 2845 + }, + { + "epoch": 0.3125411816384801, + "grad_norm": 2.247312307357788, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7106541395187378, + "num_tokens": 73370117.0, + "step": 2846 + }, + { + "epoch": 0.3126509993410938, + "grad_norm": 2.4322009086608887, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.7029791474342346, + "num_tokens": 73390983.0, + "step": 2847 + }, + { + "epoch": 0.31276081704370745, + "grad_norm": 2.1763858795166016, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7036164402961731, + "num_tokens": 73418253.0, + "step": 2848 + }, + { + "epoch": 0.3128706347463211, + "grad_norm": 2.234581470489502, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6856673359870911, + "num_tokens": 73444652.0, + "step": 2849 + }, + { + "epoch": 0.31298045244893474, + "grad_norm": 2.2894341945648193, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6983532905578613, + "num_tokens": 73468937.0, + "step": 2850 + }, + { + "epoch": 0.31309027015154844, + "grad_norm": 2.219724416732788, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7039169073104858, + "num_tokens": 73494408.0, + "step": 2851 + }, + { + "epoch": 0.3132000878541621, + "grad_norm": 1.9869464635849, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7021346092224121, + "num_tokens": 73522616.0, + "step": 2852 + }, + { + "epoch": 0.31330990555677574, + "grad_norm": 2.346522569656372, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6846404671669006, + "num_tokens": 73546318.0, + "step": 2853 + }, + { + "epoch": 0.31341972325938944, + "grad_norm": 2.1425180435180664, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6921296715736389, + "num_tokens": 73575150.0, + "step": 2854 + }, + { + "epoch": 0.3135295409620031, + "grad_norm": 2.0501201152801514, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7153516411781311, + "num_tokens": 73605072.0, + "step": 2855 + }, + { + "epoch": 0.31363935866461673, + "grad_norm": 2.0925352573394775, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6996179223060608, + "num_tokens": 73635707.0, + "step": 2856 + }, + { + "epoch": 0.3137491763672304, + "grad_norm": 2.2730629444122314, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6936469078063965, + "num_tokens": 73660386.0, + "step": 2857 + }, + { + "epoch": 0.3138589940698441, + "grad_norm": 1.879958987236023, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7177011966705322, + "num_tokens": 73691918.0, + "step": 2858 + }, + { + "epoch": 0.3139688117724577, + "grad_norm": 2.3152410984039307, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7259721755981445, + "num_tokens": 73717444.0, + "step": 2859 + }, + { + "epoch": 0.31407862947507137, + "grad_norm": 2.2874598503112793, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7022385001182556, + "num_tokens": 73741058.0, + "step": 2860 + }, + { + "epoch": 0.314188447177685, + "grad_norm": 2.0774574279785156, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6940597295761108, + "num_tokens": 73771549.0, + "step": 2861 + }, + { + "epoch": 0.3142982648802987, + "grad_norm": 2.312732458114624, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7057120203971863, + "num_tokens": 73794489.0, + "step": 2862 + }, + { + "epoch": 0.31440808258291236, + "grad_norm": 2.2401123046875, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6953775882720947, + "num_tokens": 73819693.0, + "step": 2863 + }, + { + "epoch": 0.314517900285526, + "grad_norm": 2.2948107719421387, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6885128617286682, + "num_tokens": 73844453.0, + "step": 2864 + }, + { + "epoch": 0.3146277179881397, + "grad_norm": 2.2449464797973633, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7139073610305786, + "num_tokens": 73869495.0, + "step": 2865 + }, + { + "epoch": 0.31473753569075336, + "grad_norm": 2.228484869003296, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6814453601837158, + "num_tokens": 73895544.0, + "step": 2866 + }, + { + "epoch": 0.314847353393367, + "grad_norm": 2.329885721206665, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7154290676116943, + "num_tokens": 73918872.0, + "step": 2867 + }, + { + "epoch": 0.31495717109598065, + "grad_norm": 2.227349281311035, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6960234642028809, + "num_tokens": 73942176.0, + "step": 2868 + }, + { + "epoch": 0.31506698879859435, + "grad_norm": 2.4801061153411865, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6978359222412109, + "num_tokens": 73964501.0, + "step": 2869 + }, + { + "epoch": 0.315176806501208, + "grad_norm": 2.201695680618286, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.691199779510498, + "num_tokens": 73988439.0, + "step": 2870 + }, + { + "epoch": 0.31528662420382164, + "grad_norm": 1.8875172138214111, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6781316995620728, + "num_tokens": 74023085.0, + "step": 2871 + }, + { + "epoch": 0.31539644190643534, + "grad_norm": 2.2331349849700928, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6895098090171814, + "num_tokens": 74049198.0, + "step": 2872 + }, + { + "epoch": 0.315506259609049, + "grad_norm": 2.162079334259033, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7075622081756592, + "num_tokens": 74075228.0, + "step": 2873 + }, + { + "epoch": 0.31561607731166264, + "grad_norm": 2.087589740753174, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7176006436347961, + "num_tokens": 74103086.0, + "step": 2874 + }, + { + "epoch": 0.3157258950142763, + "grad_norm": 2.324972152709961, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7063733339309692, + "num_tokens": 74126947.0, + "step": 2875 + }, + { + "epoch": 0.31583571271689, + "grad_norm": 2.1186327934265137, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6918543577194214, + "num_tokens": 74153925.0, + "step": 2876 + }, + { + "epoch": 0.31594553041950363, + "grad_norm": 2.3416032791137695, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6965700387954712, + "num_tokens": 74177784.0, + "step": 2877 + }, + { + "epoch": 0.3160553481221173, + "grad_norm": 2.224557399749756, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6898257732391357, + "num_tokens": 74205785.0, + "step": 2878 + }, + { + "epoch": 0.3161651658247309, + "grad_norm": 2.3847570419311523, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7232217192649841, + "num_tokens": 74229015.0, + "step": 2879 + }, + { + "epoch": 0.3162749835273446, + "grad_norm": 2.2797420024871826, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7288376092910767, + "num_tokens": 74250919.0, + "step": 2880 + }, + { + "epoch": 0.31638480122995827, + "grad_norm": 2.199533700942993, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.717870831489563, + "num_tokens": 74276051.0, + "step": 2881 + }, + { + "epoch": 0.3164946189325719, + "grad_norm": 2.1208746433258057, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6759403944015503, + "num_tokens": 74305792.0, + "step": 2882 + }, + { + "epoch": 0.3166044366351856, + "grad_norm": 2.3222768306732178, + "learning_rate": 1e-06, + "loss": 1.0842, + "mean_token_accuracy": 0.6742565035820007, + "num_tokens": 74330316.0, + "step": 2883 + }, + { + "epoch": 0.31671425433779926, + "grad_norm": 2.360769748687744, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7171240448951721, + "num_tokens": 74353932.0, + "step": 2884 + }, + { + "epoch": 0.3168240720404129, + "grad_norm": 2.341717004776001, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7206904292106628, + "num_tokens": 74375631.0, + "step": 2885 + }, + { + "epoch": 0.31693388974302655, + "grad_norm": 2.0368008613586426, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6933984756469727, + "num_tokens": 74406986.0, + "step": 2886 + }, + { + "epoch": 0.31704370744564025, + "grad_norm": 2.2509233951568604, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7019929885864258, + "num_tokens": 74432455.0, + "step": 2887 + }, + { + "epoch": 0.3171535251482539, + "grad_norm": 2.305464029312134, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7101945877075195, + "num_tokens": 74457818.0, + "step": 2888 + }, + { + "epoch": 0.31726334285086755, + "grad_norm": 2.2991669178009033, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6912297010421753, + "num_tokens": 74481505.0, + "step": 2889 + }, + { + "epoch": 0.31737316055348125, + "grad_norm": 2.018294095993042, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6943985819816589, + "num_tokens": 74512879.0, + "step": 2890 + }, + { + "epoch": 0.3174829782560949, + "grad_norm": 2.1138412952423096, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7000654339790344, + "num_tokens": 74539166.0, + "step": 2891 + }, + { + "epoch": 0.31759279595870854, + "grad_norm": 2.6424596309661865, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7060589790344238, + "num_tokens": 74559467.0, + "step": 2892 + }, + { + "epoch": 0.3177026136613222, + "grad_norm": 2.3062021732330322, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6845555901527405, + "num_tokens": 74587877.0, + "step": 2893 + }, + { + "epoch": 0.3178124313639359, + "grad_norm": 2.135079860687256, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7080562114715576, + "num_tokens": 74614073.0, + "step": 2894 + }, + { + "epoch": 0.31792224906654953, + "grad_norm": 2.402914047241211, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6867022514343262, + "num_tokens": 74640313.0, + "step": 2895 + }, + { + "epoch": 0.3180320667691632, + "grad_norm": 2.238962411880493, + "learning_rate": 1e-06, + "loss": 1.1147, + "mean_token_accuracy": 0.6779061555862427, + "num_tokens": 74670173.0, + "step": 2896 + }, + { + "epoch": 0.3181418844717768, + "grad_norm": 2.095189332962036, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6844492554664612, + "num_tokens": 74698118.0, + "step": 2897 + }, + { + "epoch": 0.3182517021743905, + "grad_norm": 2.149778127670288, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6969295740127563, + "num_tokens": 74723224.0, + "step": 2898 + }, + { + "epoch": 0.3183615198770042, + "grad_norm": 2.170947551727295, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.716420590877533, + "num_tokens": 74748921.0, + "step": 2899 + }, + { + "epoch": 0.3184713375796178, + "grad_norm": 2.410296678543091, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7045221924781799, + "num_tokens": 74771946.0, + "step": 2900 + }, + { + "epoch": 0.3185811552822315, + "grad_norm": 2.2239463329315186, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.707769513130188, + "num_tokens": 74799030.0, + "step": 2901 + }, + { + "epoch": 0.31869097298484517, + "grad_norm": 2.2037861347198486, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6873132586479187, + "num_tokens": 74826124.0, + "step": 2902 + }, + { + "epoch": 0.3188007906874588, + "grad_norm": 2.2192463874816895, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6954737901687622, + "num_tokens": 74852703.0, + "step": 2903 + }, + { + "epoch": 0.31891060839007246, + "grad_norm": 2.8534395694732666, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.694466233253479, + "num_tokens": 74871744.0, + "step": 2904 + }, + { + "epoch": 0.31902042609268616, + "grad_norm": 2.4365808963775635, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7027628421783447, + "num_tokens": 74894996.0, + "step": 2905 + }, + { + "epoch": 0.3191302437952998, + "grad_norm": 2.041466474533081, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6972193121910095, + "num_tokens": 74923271.0, + "step": 2906 + }, + { + "epoch": 0.31924006149791345, + "grad_norm": 2.2450406551361084, + "learning_rate": 1e-06, + "loss": 1.109, + "mean_token_accuracy": 0.6643778085708618, + "num_tokens": 74952035.0, + "step": 2907 + }, + { + "epoch": 0.3193498792005271, + "grad_norm": 2.328336238861084, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7222646474838257, + "num_tokens": 74976077.0, + "step": 2908 + }, + { + "epoch": 0.3194596969031408, + "grad_norm": 2.4651637077331543, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7045156359672546, + "num_tokens": 74997875.0, + "step": 2909 + }, + { + "epoch": 0.31956951460575445, + "grad_norm": 2.37601900100708, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6983495950698853, + "num_tokens": 75021357.0, + "step": 2910 + }, + { + "epoch": 0.3196793323083681, + "grad_norm": 2.2273900508880615, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7133800983428955, + "num_tokens": 75044488.0, + "step": 2911 + }, + { + "epoch": 0.3197891500109818, + "grad_norm": 2.388202428817749, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7083929777145386, + "num_tokens": 75067561.0, + "step": 2912 + }, + { + "epoch": 0.31989896771359544, + "grad_norm": 2.2584433555603027, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7079553604125977, + "num_tokens": 75092506.0, + "step": 2913 + }, + { + "epoch": 0.3200087854162091, + "grad_norm": 2.241133213043213, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6981340050697327, + "num_tokens": 75120679.0, + "step": 2914 + }, + { + "epoch": 0.32011860311882273, + "grad_norm": 2.0261831283569336, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6738959550857544, + "num_tokens": 75150476.0, + "step": 2915 + }, + { + "epoch": 0.32022842082143643, + "grad_norm": 2.51723051071167, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7163379192352295, + "num_tokens": 75171043.0, + "step": 2916 + }, + { + "epoch": 0.3203382385240501, + "grad_norm": 2.336054563522339, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7076014280319214, + "num_tokens": 75194749.0, + "step": 2917 + }, + { + "epoch": 0.3204480562266637, + "grad_norm": 2.1392242908477783, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7046011686325073, + "num_tokens": 75222305.0, + "step": 2918 + }, + { + "epoch": 0.3205578739292774, + "grad_norm": 2.0465052127838135, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7214971780776978, + "num_tokens": 75248920.0, + "step": 2919 + }, + { + "epoch": 0.32066769163189107, + "grad_norm": 2.1897096633911133, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6898910403251648, + "num_tokens": 75274663.0, + "step": 2920 + }, + { + "epoch": 0.3207775093345047, + "grad_norm": 2.477802038192749, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6996248960494995, + "num_tokens": 75295724.0, + "step": 2921 + }, + { + "epoch": 0.32088732703711836, + "grad_norm": 1.9815888404846191, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7006392478942871, + "num_tokens": 75325524.0, + "step": 2922 + }, + { + "epoch": 0.32099714473973207, + "grad_norm": 1.9579787254333496, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6943234801292419, + "num_tokens": 75357204.0, + "step": 2923 + }, + { + "epoch": 0.3211069624423457, + "grad_norm": 2.181570053100586, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6863467693328857, + "num_tokens": 75384094.0, + "step": 2924 + }, + { + "epoch": 0.32121678014495936, + "grad_norm": 2.055405616760254, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7040196657180786, + "num_tokens": 75413339.0, + "step": 2925 + }, + { + "epoch": 0.321326597847573, + "grad_norm": 2.337043046951294, + "learning_rate": 1e-06, + "loss": 1.1254, + "mean_token_accuracy": 0.6681421995162964, + "num_tokens": 75441013.0, + "step": 2926 + }, + { + "epoch": 0.3214364155501867, + "grad_norm": 2.469507932662964, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.708176851272583, + "num_tokens": 75461291.0, + "step": 2927 + }, + { + "epoch": 0.32154623325280035, + "grad_norm": 1.9946893453598022, + "learning_rate": 1e-06, + "loss": 1.114, + "mean_token_accuracy": 0.6710224747657776, + "num_tokens": 75496098.0, + "step": 2928 + }, + { + "epoch": 0.321656050955414, + "grad_norm": 2.4470720291137695, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7261612415313721, + "num_tokens": 75519445.0, + "step": 2929 + }, + { + "epoch": 0.3217658686580277, + "grad_norm": 2.3185946941375732, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7058773040771484, + "num_tokens": 75544068.0, + "step": 2930 + }, + { + "epoch": 0.32187568636064134, + "grad_norm": 2.306661605834961, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6883726119995117, + "num_tokens": 75568866.0, + "step": 2931 + }, + { + "epoch": 0.321985504063255, + "grad_norm": 2.454491376876831, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6985148191452026, + "num_tokens": 75589453.0, + "step": 2932 + }, + { + "epoch": 0.32209532176586864, + "grad_norm": 2.5832643508911133, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7064782381057739, + "num_tokens": 75608664.0, + "step": 2933 + }, + { + "epoch": 0.32220513946848234, + "grad_norm": 2.2832043170928955, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.708594799041748, + "num_tokens": 75632243.0, + "step": 2934 + }, + { + "epoch": 0.322314957171096, + "grad_norm": 2.473538398742676, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7126763463020325, + "num_tokens": 75653989.0, + "step": 2935 + }, + { + "epoch": 0.32242477487370963, + "grad_norm": 2.207929849624634, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7005876302719116, + "num_tokens": 75681860.0, + "step": 2936 + }, + { + "epoch": 0.3225345925763233, + "grad_norm": 1.9876937866210938, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.690348744392395, + "num_tokens": 75715876.0, + "step": 2937 + }, + { + "epoch": 0.322644410278937, + "grad_norm": 2.140407085418701, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6947585344314575, + "num_tokens": 75743095.0, + "step": 2938 + }, + { + "epoch": 0.3227542279815506, + "grad_norm": 2.1939375400543213, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6825278997421265, + "num_tokens": 75772816.0, + "step": 2939 + }, + { + "epoch": 0.32286404568416427, + "grad_norm": 2.207885503768921, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7192182540893555, + "num_tokens": 75797061.0, + "step": 2940 + }, + { + "epoch": 0.32297386338677797, + "grad_norm": 2.3662056922912598, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7133526802062988, + "num_tokens": 75819885.0, + "step": 2941 + }, + { + "epoch": 0.3230836810893916, + "grad_norm": 2.1151087284088135, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6861070394515991, + "num_tokens": 75848162.0, + "step": 2942 + }, + { + "epoch": 0.32319349879200526, + "grad_norm": 2.158140182495117, + "learning_rate": 1e-06, + "loss": 1.1228, + "mean_token_accuracy": 0.665877640247345, + "num_tokens": 75879295.0, + "step": 2943 + }, + { + "epoch": 0.3233033164946189, + "grad_norm": 2.501674175262451, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7085663080215454, + "num_tokens": 75900556.0, + "step": 2944 + }, + { + "epoch": 0.3234131341972326, + "grad_norm": 2.487297296524048, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6965876817703247, + "num_tokens": 75922186.0, + "step": 2945 + }, + { + "epoch": 0.32352295189984626, + "grad_norm": 2.1131880283355713, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6914610266685486, + "num_tokens": 75947861.0, + "step": 2946 + }, + { + "epoch": 0.3236327696024599, + "grad_norm": 2.109032392501831, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6834893822669983, + "num_tokens": 75978367.0, + "step": 2947 + }, + { + "epoch": 0.3237425873050736, + "grad_norm": 2.653294086456299, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7039600610733032, + "num_tokens": 75998399.0, + "step": 2948 + }, + { + "epoch": 0.32385240500768725, + "grad_norm": 2.4762697219848633, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7037975788116455, + "num_tokens": 76020578.0, + "step": 2949 + }, + { + "epoch": 0.3239622227103009, + "grad_norm": 2.3068149089813232, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6919560432434082, + "num_tokens": 76045497.0, + "step": 2950 + }, + { + "epoch": 0.32407204041291454, + "grad_norm": 2.5274946689605713, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6864248514175415, + "num_tokens": 76066312.0, + "step": 2951 + }, + { + "epoch": 0.32418185811552824, + "grad_norm": 2.5732998847961426, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.696410059928894, + "num_tokens": 76086796.0, + "step": 2952 + }, + { + "epoch": 0.3242916758181419, + "grad_norm": 2.0979838371276855, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7070170044898987, + "num_tokens": 76116544.0, + "step": 2953 + }, + { + "epoch": 0.32440149352075554, + "grad_norm": 2.207049608230591, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.681222140789032, + "num_tokens": 76143949.0, + "step": 2954 + }, + { + "epoch": 0.3245113112233692, + "grad_norm": 2.098112106323242, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7058863639831543, + "num_tokens": 76173646.0, + "step": 2955 + }, + { + "epoch": 0.3246211289259829, + "grad_norm": 2.1157314777374268, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7099952697753906, + "num_tokens": 76200853.0, + "step": 2956 + }, + { + "epoch": 0.32473094662859653, + "grad_norm": 1.9442750215530396, + "learning_rate": 1e-06, + "loss": 1.0666, + "mean_token_accuracy": 0.6884751915931702, + "num_tokens": 76230516.0, + "step": 2957 + }, + { + "epoch": 0.3248407643312102, + "grad_norm": 2.0806238651275635, + "learning_rate": 1e-06, + "loss": 1.0643, + "mean_token_accuracy": 0.6839187145233154, + "num_tokens": 76260461.0, + "step": 2958 + }, + { + "epoch": 0.3249505820338239, + "grad_norm": 2.5229828357696533, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7045673727989197, + "num_tokens": 76280303.0, + "step": 2959 + }, + { + "epoch": 0.3250603997364375, + "grad_norm": 2.285458564758301, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7071123123168945, + "num_tokens": 76304135.0, + "step": 2960 + }, + { + "epoch": 0.32517021743905117, + "grad_norm": 2.5897793769836426, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7024753093719482, + "num_tokens": 76324084.0, + "step": 2961 + }, + { + "epoch": 0.3252800351416648, + "grad_norm": 1.9913984537124634, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6964789032936096, + "num_tokens": 76353759.0, + "step": 2962 + }, + { + "epoch": 0.3253898528442785, + "grad_norm": 2.144122838973999, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6993601322174072, + "num_tokens": 76380822.0, + "step": 2963 + }, + { + "epoch": 0.32549967054689216, + "grad_norm": 2.080483913421631, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6885857582092285, + "num_tokens": 76409083.0, + "step": 2964 + }, + { + "epoch": 0.3256094882495058, + "grad_norm": 2.1210999488830566, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6806914806365967, + "num_tokens": 76437655.0, + "step": 2965 + }, + { + "epoch": 0.3257193059521195, + "grad_norm": 2.048360586166382, + "learning_rate": 1e-06, + "loss": 1.0707, + "mean_token_accuracy": 0.6864706873893738, + "num_tokens": 76468809.0, + "step": 2966 + }, + { + "epoch": 0.32582912365473315, + "grad_norm": 2.0908408164978027, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7253627181053162, + "num_tokens": 76495130.0, + "step": 2967 + }, + { + "epoch": 0.3259389413573468, + "grad_norm": 2.2114856243133545, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7406829595565796, + "num_tokens": 76520370.0, + "step": 2968 + }, + { + "epoch": 0.32604875905996045, + "grad_norm": 2.2417023181915283, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7159679532051086, + "num_tokens": 76545346.0, + "step": 2969 + }, + { + "epoch": 0.32615857676257415, + "grad_norm": 2.1715869903564453, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7069807052612305, + "num_tokens": 76572704.0, + "step": 2970 + }, + { + "epoch": 0.3262683944651878, + "grad_norm": 2.222667694091797, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7244213819503784, + "num_tokens": 76596066.0, + "step": 2971 + }, + { + "epoch": 0.32637821216780144, + "grad_norm": 2.490077495574951, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7001051902770996, + "num_tokens": 76617617.0, + "step": 2972 + }, + { + "epoch": 0.3264880298704151, + "grad_norm": 2.1797046661376953, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7102378010749817, + "num_tokens": 76643988.0, + "step": 2973 + }, + { + "epoch": 0.3265978475730288, + "grad_norm": 2.333519697189331, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6959734559059143, + "num_tokens": 76665800.0, + "step": 2974 + }, + { + "epoch": 0.32670766527564243, + "grad_norm": 2.1137664318084717, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6970123052597046, + "num_tokens": 76692818.0, + "step": 2975 + }, + { + "epoch": 0.3268174829782561, + "grad_norm": 2.735399007797241, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6912699341773987, + "num_tokens": 76713050.0, + "step": 2976 + }, + { + "epoch": 0.3269273006808698, + "grad_norm": 2.4278523921966553, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6931853294372559, + "num_tokens": 76735940.0, + "step": 2977 + }, + { + "epoch": 0.3270371183834834, + "grad_norm": 2.2649848461151123, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.698623776435852, + "num_tokens": 76762643.0, + "step": 2978 + }, + { + "epoch": 0.3271469360860971, + "grad_norm": 2.2466814517974854, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7035626769065857, + "num_tokens": 76785237.0, + "step": 2979 + }, + { + "epoch": 0.3272567537887107, + "grad_norm": 2.429884433746338, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6882016658782959, + "num_tokens": 76808033.0, + "step": 2980 + }, + { + "epoch": 0.3273665714913244, + "grad_norm": 2.1849048137664795, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7006545662879944, + "num_tokens": 76833331.0, + "step": 2981 + }, + { + "epoch": 0.32747638919393807, + "grad_norm": 2.2137084007263184, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.7021163702011108, + "num_tokens": 76859772.0, + "step": 2982 + }, + { + "epoch": 0.3275862068965517, + "grad_norm": 1.98012113571167, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6800628900527954, + "num_tokens": 76892579.0, + "step": 2983 + }, + { + "epoch": 0.32769602459916536, + "grad_norm": 2.357384443283081, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.68428635597229, + "num_tokens": 76916308.0, + "step": 2984 + }, + { + "epoch": 0.32780584230177906, + "grad_norm": 2.24430251121521, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.6851102113723755, + "num_tokens": 76941060.0, + "step": 2985 + }, + { + "epoch": 0.3279156600043927, + "grad_norm": 2.217395544052124, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7064620852470398, + "num_tokens": 76965066.0, + "step": 2986 + }, + { + "epoch": 0.32802547770700635, + "grad_norm": 1.9810936450958252, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6853587031364441, + "num_tokens": 76998143.0, + "step": 2987 + }, + { + "epoch": 0.32813529540962005, + "grad_norm": 2.091439723968506, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7073237895965576, + "num_tokens": 77024394.0, + "step": 2988 + }, + { + "epoch": 0.3282451131122337, + "grad_norm": 2.4251317977905273, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6867237091064453, + "num_tokens": 77047211.0, + "step": 2989 + }, + { + "epoch": 0.32835493081484735, + "grad_norm": 2.187753200531006, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6873447895050049, + "num_tokens": 77071790.0, + "step": 2990 + }, + { + "epoch": 0.328464748517461, + "grad_norm": 2.321141481399536, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6778768301010132, + "num_tokens": 77095754.0, + "step": 2991 + }, + { + "epoch": 0.3285745662200747, + "grad_norm": 2.2226972579956055, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7058171033859253, + "num_tokens": 77120251.0, + "step": 2992 + }, + { + "epoch": 0.32868438392268834, + "grad_norm": 2.3256752490997314, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7271384000778198, + "num_tokens": 77141003.0, + "step": 2993 + }, + { + "epoch": 0.328794201625302, + "grad_norm": 2.2400012016296387, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6803252100944519, + "num_tokens": 77168714.0, + "step": 2994 + }, + { + "epoch": 0.3289040193279157, + "grad_norm": 2.203429937362671, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7070187926292419, + "num_tokens": 77194128.0, + "step": 2995 + }, + { + "epoch": 0.32901383703052933, + "grad_norm": 2.1637468338012695, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6914628744125366, + "num_tokens": 77221463.0, + "step": 2996 + }, + { + "epoch": 0.329123654733143, + "grad_norm": 2.1744773387908936, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7025816440582275, + "num_tokens": 77248280.0, + "step": 2997 + }, + { + "epoch": 0.3292334724357566, + "grad_norm": 2.307584285736084, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6894635558128357, + "num_tokens": 77271554.0, + "step": 2998 + }, + { + "epoch": 0.3293432901383703, + "grad_norm": 2.44327712059021, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6794438362121582, + "num_tokens": 77297217.0, + "step": 2999 + }, + { + "epoch": 0.32945310784098397, + "grad_norm": 2.790201425552368, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7092496752738953, + "num_tokens": 77315963.0, + "step": 3000 + }, + { + "epoch": 0.3295629255435976, + "grad_norm": 2.281829833984375, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7058125734329224, + "num_tokens": 77340211.0, + "step": 3001 + }, + { + "epoch": 0.32967274324621126, + "grad_norm": 2.0176360607147217, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6873332262039185, + "num_tokens": 77370041.0, + "step": 3002 + }, + { + "epoch": 0.32978256094882497, + "grad_norm": 2.3038816452026367, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7126200199127197, + "num_tokens": 77396012.0, + "step": 3003 + }, + { + "epoch": 0.3298923786514386, + "grad_norm": 2.2942054271698, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.68934565782547, + "num_tokens": 77420528.0, + "step": 3004 + }, + { + "epoch": 0.33000219635405226, + "grad_norm": 2.055997133255005, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.7013257741928101, + "num_tokens": 77448705.0, + "step": 3005 + }, + { + "epoch": 0.33011201405666596, + "grad_norm": 2.1369833946228027, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6936062574386597, + "num_tokens": 77477467.0, + "step": 3006 + }, + { + "epoch": 0.3302218317592796, + "grad_norm": 2.155141830444336, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.691250205039978, + "num_tokens": 77503762.0, + "step": 3007 + }, + { + "epoch": 0.33033164946189325, + "grad_norm": 2.1827878952026367, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.7102819681167603, + "num_tokens": 77530935.0, + "step": 3008 + }, + { + "epoch": 0.3304414671645069, + "grad_norm": 2.437268018722534, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.703942060470581, + "num_tokens": 77553042.0, + "step": 3009 + }, + { + "epoch": 0.3305512848671206, + "grad_norm": 2.055429220199585, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6948639154434204, + "num_tokens": 77581663.0, + "step": 3010 + }, + { + "epoch": 0.33066110256973424, + "grad_norm": 2.1906309127807617, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6936867237091064, + "num_tokens": 77608695.0, + "step": 3011 + }, + { + "epoch": 0.3307709202723479, + "grad_norm": 2.2005422115325928, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6956524848937988, + "num_tokens": 77636207.0, + "step": 3012 + }, + { + "epoch": 0.33088073797496154, + "grad_norm": 2.4890408515930176, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6920780539512634, + "num_tokens": 77656200.0, + "step": 3013 + }, + { + "epoch": 0.33099055567757524, + "grad_norm": 2.4399702548980713, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7044920921325684, + "num_tokens": 77677849.0, + "step": 3014 + }, + { + "epoch": 0.3311003733801889, + "grad_norm": 2.1378464698791504, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7215411067008972, + "num_tokens": 77705034.0, + "step": 3015 + }, + { + "epoch": 0.33121019108280253, + "grad_norm": 2.2403156757354736, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6962595582008362, + "num_tokens": 77731323.0, + "step": 3016 + }, + { + "epoch": 0.33132000878541623, + "grad_norm": 2.4707868099212646, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7245039343833923, + "num_tokens": 77753940.0, + "step": 3017 + }, + { + "epoch": 0.3314298264880299, + "grad_norm": 2.237910509109497, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7014697790145874, + "num_tokens": 77780094.0, + "step": 3018 + }, + { + "epoch": 0.3315396441906435, + "grad_norm": 2.1603758335113525, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7117887139320374, + "num_tokens": 77807779.0, + "step": 3019 + }, + { + "epoch": 0.33164946189325717, + "grad_norm": 2.1720476150512695, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6943166851997375, + "num_tokens": 77835620.0, + "step": 3020 + }, + { + "epoch": 0.33175927959587087, + "grad_norm": 2.131121873855591, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6922627091407776, + "num_tokens": 77863304.0, + "step": 3021 + }, + { + "epoch": 0.3318690972984845, + "grad_norm": 2.2497048377990723, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7124577760696411, + "num_tokens": 77888469.0, + "step": 3022 + }, + { + "epoch": 0.33197891500109816, + "grad_norm": 2.2541122436523438, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6900464296340942, + "num_tokens": 77913898.0, + "step": 3023 + }, + { + "epoch": 0.33208873270371186, + "grad_norm": 2.3498823642730713, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7034679651260376, + "num_tokens": 77939423.0, + "step": 3024 + }, + { + "epoch": 0.3321985504063255, + "grad_norm": 2.1962995529174805, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6864434480667114, + "num_tokens": 77967766.0, + "step": 3025 + }, + { + "epoch": 0.33230836810893916, + "grad_norm": 2.0486199855804443, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7027358412742615, + "num_tokens": 77994828.0, + "step": 3026 + }, + { + "epoch": 0.3324181858115528, + "grad_norm": 2.2308459281921387, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6987676024436951, + "num_tokens": 78020315.0, + "step": 3027 + }, + { + "epoch": 0.3325280035141665, + "grad_norm": 2.1569643020629883, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7097762227058411, + "num_tokens": 78047046.0, + "step": 3028 + }, + { + "epoch": 0.33263782121678015, + "grad_norm": 1.9129767417907715, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7013837695121765, + "num_tokens": 78080980.0, + "step": 3029 + }, + { + "epoch": 0.3327476389193938, + "grad_norm": 2.279857873916626, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6866183280944824, + "num_tokens": 78107993.0, + "step": 3030 + }, + { + "epoch": 0.33285745662200744, + "grad_norm": 2.2895030975341797, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7002628445625305, + "num_tokens": 78131475.0, + "step": 3031 + }, + { + "epoch": 0.33296727432462114, + "grad_norm": 2.1544189453125, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6867884397506714, + "num_tokens": 78159754.0, + "step": 3032 + }, + { + "epoch": 0.3330770920272348, + "grad_norm": 2.074767589569092, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7007118463516235, + "num_tokens": 78190003.0, + "step": 3033 + }, + { + "epoch": 0.33318690972984844, + "grad_norm": 2.2627410888671875, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6989232301712036, + "num_tokens": 78215206.0, + "step": 3034 + }, + { + "epoch": 0.33329672743246214, + "grad_norm": 2.0859034061431885, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6941099762916565, + "num_tokens": 78243382.0, + "step": 3035 + }, + { + "epoch": 0.3334065451350758, + "grad_norm": 2.0990169048309326, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7102744579315186, + "num_tokens": 78271692.0, + "step": 3036 + }, + { + "epoch": 0.33351636283768943, + "grad_norm": 2.259965419769287, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7050171494483948, + "num_tokens": 78296399.0, + "step": 3037 + }, + { + "epoch": 0.3336261805403031, + "grad_norm": 2.4707295894622803, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6994906067848206, + "num_tokens": 78319425.0, + "step": 3038 + }, + { + "epoch": 0.3337359982429168, + "grad_norm": 2.4160282611846924, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6845672130584717, + "num_tokens": 78342245.0, + "step": 3039 + }, + { + "epoch": 0.3338458159455304, + "grad_norm": 2.1527302265167236, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7021724581718445, + "num_tokens": 78372439.0, + "step": 3040 + }, + { + "epoch": 0.33395563364814407, + "grad_norm": 1.9852839708328247, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7001529932022095, + "num_tokens": 78404306.0, + "step": 3041 + }, + { + "epoch": 0.33406545135075777, + "grad_norm": 2.101794719696045, + "learning_rate": 1e-06, + "loss": 1.0963, + "mean_token_accuracy": 0.6701035499572754, + "num_tokens": 78432587.0, + "step": 3042 + }, + { + "epoch": 0.3341752690533714, + "grad_norm": 2.279289722442627, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.685988187789917, + "num_tokens": 78460386.0, + "step": 3043 + }, + { + "epoch": 0.33428508675598506, + "grad_norm": 2.012683153152466, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6869602203369141, + "num_tokens": 78492312.0, + "step": 3044 + }, + { + "epoch": 0.3343949044585987, + "grad_norm": 2.3519532680511475, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6979051232337952, + "num_tokens": 78515653.0, + "step": 3045 + }, + { + "epoch": 0.3345047221612124, + "grad_norm": 2.666867733001709, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6988632678985596, + "num_tokens": 78534927.0, + "step": 3046 + }, + { + "epoch": 0.33461453986382605, + "grad_norm": 2.2815117835998535, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6922963261604309, + "num_tokens": 78564056.0, + "step": 3047 + }, + { + "epoch": 0.3347243575664397, + "grad_norm": 2.4038214683532715, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7250200510025024, + "num_tokens": 78586385.0, + "step": 3048 + }, + { + "epoch": 0.33483417526905335, + "grad_norm": 2.2085347175598145, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6774266958236694, + "num_tokens": 78611369.0, + "step": 3049 + }, + { + "epoch": 0.33494399297166705, + "grad_norm": 2.0318126678466797, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6900940537452698, + "num_tokens": 78642869.0, + "step": 3050 + }, + { + "epoch": 0.3350538106742807, + "grad_norm": 2.3990612030029297, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7246831059455872, + "num_tokens": 78664395.0, + "step": 3051 + }, + { + "epoch": 0.33516362837689434, + "grad_norm": 2.3009395599365234, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6895232200622559, + "num_tokens": 78688137.0, + "step": 3052 + }, + { + "epoch": 0.33527344607950804, + "grad_norm": 2.13468337059021, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6854434013366699, + "num_tokens": 78716066.0, + "step": 3053 + }, + { + "epoch": 0.3353832637821217, + "grad_norm": 2.2885093688964844, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6978453397750854, + "num_tokens": 78739237.0, + "step": 3054 + }, + { + "epoch": 0.33549308148473533, + "grad_norm": 2.0961718559265137, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7125476598739624, + "num_tokens": 78765553.0, + "step": 3055 + }, + { + "epoch": 0.335602899187349, + "grad_norm": 2.425337553024292, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7207319736480713, + "num_tokens": 78790653.0, + "step": 3056 + }, + { + "epoch": 0.3357127168899627, + "grad_norm": 2.0978400707244873, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7138715982437134, + "num_tokens": 78817755.0, + "step": 3057 + }, + { + "epoch": 0.3358225345925763, + "grad_norm": 2.0156092643737793, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6912397146224976, + "num_tokens": 78847923.0, + "step": 3058 + }, + { + "epoch": 0.33593235229519, + "grad_norm": 2.3361105918884277, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7083076238632202, + "num_tokens": 78871147.0, + "step": 3059 + }, + { + "epoch": 0.3360421699978036, + "grad_norm": 2.2416927814483643, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6919232606887817, + "num_tokens": 78895834.0, + "step": 3060 + }, + { + "epoch": 0.3361519877004173, + "grad_norm": 2.53806209564209, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7169697880744934, + "num_tokens": 78917429.0, + "step": 3061 + }, + { + "epoch": 0.33626180540303097, + "grad_norm": 2.0049493312835693, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7066969871520996, + "num_tokens": 78946836.0, + "step": 3062 + }, + { + "epoch": 0.3363716231056446, + "grad_norm": 2.32620906829834, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6855254173278809, + "num_tokens": 78971280.0, + "step": 3063 + }, + { + "epoch": 0.3364814408082583, + "grad_norm": 2.3114399909973145, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.686143159866333, + "num_tokens": 78995670.0, + "step": 3064 + }, + { + "epoch": 0.33659125851087196, + "grad_norm": 1.9512639045715332, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6849387288093567, + "num_tokens": 79026465.0, + "step": 3065 + }, + { + "epoch": 0.3367010762134856, + "grad_norm": 2.0435876846313477, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.696727991104126, + "num_tokens": 79056308.0, + "step": 3066 + }, + { + "epoch": 0.33681089391609925, + "grad_norm": 2.215190887451172, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6870208978652954, + "num_tokens": 79081220.0, + "step": 3067 + }, + { + "epoch": 0.33692071161871295, + "grad_norm": 2.4025979042053223, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7100003361701965, + "num_tokens": 79102542.0, + "step": 3068 + }, + { + "epoch": 0.3370305293213266, + "grad_norm": 2.18609356880188, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7192745208740234, + "num_tokens": 79129195.0, + "step": 3069 + }, + { + "epoch": 0.33714034702394025, + "grad_norm": 2.292410135269165, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7267578840255737, + "num_tokens": 79151873.0, + "step": 3070 + }, + { + "epoch": 0.33725016472655395, + "grad_norm": 2.260314464569092, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6988188028335571, + "num_tokens": 79177575.0, + "step": 3071 + }, + { + "epoch": 0.3373599824291676, + "grad_norm": 2.07973575592041, + "learning_rate": 1e-06, + "loss": 1.1029, + "mean_token_accuracy": 0.6691560745239258, + "num_tokens": 79207867.0, + "step": 3072 + }, + { + "epoch": 0.33746980013178124, + "grad_norm": 2.14982008934021, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.6791177988052368, + "num_tokens": 79235194.0, + "step": 3073 + }, + { + "epoch": 0.3375796178343949, + "grad_norm": 2.7268197536468506, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7194114923477173, + "num_tokens": 79255494.0, + "step": 3074 + }, + { + "epoch": 0.3376894355370086, + "grad_norm": 2.0074656009674072, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7193151116371155, + "num_tokens": 79284827.0, + "step": 3075 + }, + { + "epoch": 0.33779925323962223, + "grad_norm": 2.1091840267181396, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.694633424282074, + "num_tokens": 79312306.0, + "step": 3076 + }, + { + "epoch": 0.3379090709422359, + "grad_norm": 2.3005058765411377, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.693648099899292, + "num_tokens": 79340643.0, + "step": 3077 + }, + { + "epoch": 0.3380188886448495, + "grad_norm": 2.215301275253296, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6977013945579529, + "num_tokens": 79367367.0, + "step": 3078 + }, + { + "epoch": 0.3381287063474632, + "grad_norm": 2.2462966442108154, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7126877307891846, + "num_tokens": 79390499.0, + "step": 3079 + }, + { + "epoch": 0.33823852405007687, + "grad_norm": 2.150338888168335, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6986103653907776, + "num_tokens": 79417797.0, + "step": 3080 + }, + { + "epoch": 0.3383483417526905, + "grad_norm": 2.2253966331481934, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6889556646347046, + "num_tokens": 79443439.0, + "step": 3081 + }, + { + "epoch": 0.3384581594553042, + "grad_norm": 2.5809624195098877, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6955488920211792, + "num_tokens": 79463415.0, + "step": 3082 + }, + { + "epoch": 0.33856797715791787, + "grad_norm": 2.0830860137939453, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7138557434082031, + "num_tokens": 79493663.0, + "step": 3083 + }, + { + "epoch": 0.3386777948605315, + "grad_norm": 1.9629898071289062, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7027324438095093, + "num_tokens": 79525295.0, + "step": 3084 + }, + { + "epoch": 0.33878761256314516, + "grad_norm": 2.2813501358032227, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.704113245010376, + "num_tokens": 79548311.0, + "step": 3085 + }, + { + "epoch": 0.33889743026575886, + "grad_norm": 2.140519142150879, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7011005878448486, + "num_tokens": 79575390.0, + "step": 3086 + }, + { + "epoch": 0.3390072479683725, + "grad_norm": 2.4527087211608887, + "learning_rate": 1e-06, + "loss": 1.0786, + "mean_token_accuracy": 0.6836740970611572, + "num_tokens": 79598029.0, + "step": 3087 + }, + { + "epoch": 0.33911706567098615, + "grad_norm": 2.222041130065918, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7184730768203735, + "num_tokens": 79623682.0, + "step": 3088 + }, + { + "epoch": 0.3392268833735998, + "grad_norm": 2.237431764602661, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7046675682067871, + "num_tokens": 79649818.0, + "step": 3089 + }, + { + "epoch": 0.3393367010762135, + "grad_norm": 2.4128575325012207, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6912021636962891, + "num_tokens": 79671196.0, + "step": 3090 + }, + { + "epoch": 0.33944651877882714, + "grad_norm": 2.2427265644073486, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7151055335998535, + "num_tokens": 79697182.0, + "step": 3091 + }, + { + "epoch": 0.3395563364814408, + "grad_norm": 2.2736077308654785, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6942768096923828, + "num_tokens": 79720723.0, + "step": 3092 + }, + { + "epoch": 0.3396661541840545, + "grad_norm": 2.2166295051574707, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6938772797584534, + "num_tokens": 79746352.0, + "step": 3093 + }, + { + "epoch": 0.33977597188666814, + "grad_norm": 2.086812734603882, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7106897234916687, + "num_tokens": 79773134.0, + "step": 3094 + }, + { + "epoch": 0.3398857895892818, + "grad_norm": 2.441305637359619, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7020216584205627, + "num_tokens": 79793652.0, + "step": 3095 + }, + { + "epoch": 0.33999560729189543, + "grad_norm": 2.276266574859619, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6977211236953735, + "num_tokens": 79818279.0, + "step": 3096 + }, + { + "epoch": 0.34010542499450913, + "grad_norm": 2.466480016708374, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6965236663818359, + "num_tokens": 79841971.0, + "step": 3097 + }, + { + "epoch": 0.3402152426971228, + "grad_norm": 2.275893211364746, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6922240257263184, + "num_tokens": 79866666.0, + "step": 3098 + }, + { + "epoch": 0.3403250603997364, + "grad_norm": 1.9835683107376099, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7071321606636047, + "num_tokens": 79900068.0, + "step": 3099 + }, + { + "epoch": 0.3404348781023501, + "grad_norm": 1.990419864654541, + "learning_rate": 1e-06, + "loss": 1.149, + "mean_token_accuracy": 0.656061589717865, + "num_tokens": 79932876.0, + "step": 3100 + }, + { + "epoch": 0.34054469580496377, + "grad_norm": 2.2716057300567627, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7103056907653809, + "num_tokens": 79958452.0, + "step": 3101 + }, + { + "epoch": 0.3406545135075774, + "grad_norm": 2.3046176433563232, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6832846403121948, + "num_tokens": 79981764.0, + "step": 3102 + }, + { + "epoch": 0.34076433121019106, + "grad_norm": 2.2772600650787354, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6926384568214417, + "num_tokens": 80006775.0, + "step": 3103 + }, + { + "epoch": 0.34087414891280476, + "grad_norm": 2.234548807144165, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6978040933609009, + "num_tokens": 80030892.0, + "step": 3104 + }, + { + "epoch": 0.3409839666154184, + "grad_norm": 2.24275541305542, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6943961381912231, + "num_tokens": 80056618.0, + "step": 3105 + }, + { + "epoch": 0.34109378431803206, + "grad_norm": 2.3605380058288574, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7065497636795044, + "num_tokens": 80079403.0, + "step": 3106 + }, + { + "epoch": 0.3412036020206457, + "grad_norm": 2.3136415481567383, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6973611116409302, + "num_tokens": 80104019.0, + "step": 3107 + }, + { + "epoch": 0.3413134197232594, + "grad_norm": 2.3531479835510254, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6776583194732666, + "num_tokens": 80130735.0, + "step": 3108 + }, + { + "epoch": 0.34142323742587305, + "grad_norm": 2.387075662612915, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6904451847076416, + "num_tokens": 80156530.0, + "step": 3109 + }, + { + "epoch": 0.3415330551284867, + "grad_norm": 2.0111935138702393, + "learning_rate": 1e-06, + "loss": 1.119, + "mean_token_accuracy": 0.6653903126716614, + "num_tokens": 80188069.0, + "step": 3110 + }, + { + "epoch": 0.3416428728311004, + "grad_norm": 2.3096251487731934, + "learning_rate": 1e-06, + "loss": 1.1043, + "mean_token_accuracy": 0.6685628890991211, + "num_tokens": 80211915.0, + "step": 3111 + }, + { + "epoch": 0.34175269053371404, + "grad_norm": 2.3976054191589355, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7204967737197876, + "num_tokens": 80233495.0, + "step": 3112 + }, + { + "epoch": 0.3418625082363277, + "grad_norm": 2.1371009349823, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6876201629638672, + "num_tokens": 80262359.0, + "step": 3113 + }, + { + "epoch": 0.34197232593894134, + "grad_norm": 2.551801919937134, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7245209217071533, + "num_tokens": 80280893.0, + "step": 3114 + }, + { + "epoch": 0.34208214364155504, + "grad_norm": 2.326286554336548, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7100824117660522, + "num_tokens": 80301972.0, + "step": 3115 + }, + { + "epoch": 0.3421919613441687, + "grad_norm": 2.0956976413726807, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6889800429344177, + "num_tokens": 80329115.0, + "step": 3116 + }, + { + "epoch": 0.34230177904678233, + "grad_norm": 2.0484986305236816, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6880240440368652, + "num_tokens": 80357041.0, + "step": 3117 + }, + { + "epoch": 0.34241159674939603, + "grad_norm": 2.247593879699707, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.708584725856781, + "num_tokens": 80382380.0, + "step": 3118 + }, + { + "epoch": 0.3425214144520097, + "grad_norm": 2.1231589317321777, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7024068832397461, + "num_tokens": 80408028.0, + "step": 3119 + }, + { + "epoch": 0.3426312321546233, + "grad_norm": 2.1888387203216553, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6813201904296875, + "num_tokens": 80434981.0, + "step": 3120 + }, + { + "epoch": 0.34274104985723697, + "grad_norm": 1.9699857234954834, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7081137895584106, + "num_tokens": 80465931.0, + "step": 3121 + }, + { + "epoch": 0.34285086755985067, + "grad_norm": 2.315584182739258, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6979666948318481, + "num_tokens": 80491105.0, + "step": 3122 + }, + { + "epoch": 0.3429606852624643, + "grad_norm": 1.9742499589920044, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6771121025085449, + "num_tokens": 80522614.0, + "step": 3123 + }, + { + "epoch": 0.34307050296507796, + "grad_norm": 2.37146258354187, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7091816663742065, + "num_tokens": 80546360.0, + "step": 3124 + }, + { + "epoch": 0.3431803206676916, + "grad_norm": 1.996690273284912, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6956238150596619, + "num_tokens": 80578374.0, + "step": 3125 + }, + { + "epoch": 0.3432901383703053, + "grad_norm": 2.409364700317383, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6974350214004517, + "num_tokens": 80602844.0, + "step": 3126 + }, + { + "epoch": 0.34339995607291895, + "grad_norm": 2.566563606262207, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.712022066116333, + "num_tokens": 80623025.0, + "step": 3127 + }, + { + "epoch": 0.3435097737755326, + "grad_norm": 2.0923349857330322, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.694819986820221, + "num_tokens": 80652029.0, + "step": 3128 + }, + { + "epoch": 0.3436195914781463, + "grad_norm": 2.3712544441223145, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7013217210769653, + "num_tokens": 80675130.0, + "step": 3129 + }, + { + "epoch": 0.34372940918075995, + "grad_norm": 2.052915096282959, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6967713236808777, + "num_tokens": 80702624.0, + "step": 3130 + }, + { + "epoch": 0.3438392268833736, + "grad_norm": 2.2536935806274414, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6776730418205261, + "num_tokens": 80729912.0, + "step": 3131 + }, + { + "epoch": 0.34394904458598724, + "grad_norm": 2.2602155208587646, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6833809614181519, + "num_tokens": 80756420.0, + "step": 3132 + }, + { + "epoch": 0.34405886228860094, + "grad_norm": 2.282050848007202, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7028229236602783, + "num_tokens": 80779824.0, + "step": 3133 + }, + { + "epoch": 0.3441686799912146, + "grad_norm": 2.1224167346954346, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7004196643829346, + "num_tokens": 80806454.0, + "step": 3134 + }, + { + "epoch": 0.34427849769382823, + "grad_norm": 2.499152660369873, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7314776182174683, + "num_tokens": 80825896.0, + "step": 3135 + }, + { + "epoch": 0.3443883153964419, + "grad_norm": 2.4203836917877197, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6960965394973755, + "num_tokens": 80848763.0, + "step": 3136 + }, + { + "epoch": 0.3444981330990556, + "grad_norm": 2.564493179321289, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7033065557479858, + "num_tokens": 80869927.0, + "step": 3137 + }, + { + "epoch": 0.3446079508016692, + "grad_norm": 2.257877826690674, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7045906782150269, + "num_tokens": 80894372.0, + "step": 3138 + }, + { + "epoch": 0.3447177685042829, + "grad_norm": 2.0552616119384766, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6910562515258789, + "num_tokens": 80922693.0, + "step": 3139 + }, + { + "epoch": 0.3448275862068966, + "grad_norm": 2.0764548778533936, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7052179574966431, + "num_tokens": 80950728.0, + "step": 3140 + }, + { + "epoch": 0.3449374039095102, + "grad_norm": 2.328718900680542, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6897061467170715, + "num_tokens": 80975360.0, + "step": 3141 + }, + { + "epoch": 0.34504722161212387, + "grad_norm": 2.186711549758911, + "learning_rate": 1e-06, + "loss": 1.1072, + "mean_token_accuracy": 0.6654009222984314, + "num_tokens": 81003006.0, + "step": 3142 + }, + { + "epoch": 0.3451570393147375, + "grad_norm": 2.116159439086914, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6862322092056274, + "num_tokens": 81029831.0, + "step": 3143 + }, + { + "epoch": 0.3452668570173512, + "grad_norm": 2.3162519931793213, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7321339845657349, + "num_tokens": 81051503.0, + "step": 3144 + }, + { + "epoch": 0.34537667471996486, + "grad_norm": 2.4133660793304443, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7148917317390442, + "num_tokens": 81072652.0, + "step": 3145 + }, + { + "epoch": 0.3454864924225785, + "grad_norm": 2.218376398086548, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7249414920806885, + "num_tokens": 81097436.0, + "step": 3146 + }, + { + "epoch": 0.3455963101251922, + "grad_norm": 2.1051058769226074, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7192518711090088, + "num_tokens": 81123498.0, + "step": 3147 + }, + { + "epoch": 0.34570612782780585, + "grad_norm": 2.521519899368286, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7356486320495605, + "num_tokens": 81142822.0, + "step": 3148 + }, + { + "epoch": 0.3458159455304195, + "grad_norm": 2.123120069503784, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6965869665145874, + "num_tokens": 81170366.0, + "step": 3149 + }, + { + "epoch": 0.34592576323303315, + "grad_norm": 2.226635217666626, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7044293880462646, + "num_tokens": 81195525.0, + "step": 3150 + }, + { + "epoch": 0.34603558093564685, + "grad_norm": 2.37427020072937, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6988531351089478, + "num_tokens": 81219234.0, + "step": 3151 + }, + { + "epoch": 0.3461453986382605, + "grad_norm": 2.267101287841797, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7021185755729675, + "num_tokens": 81245266.0, + "step": 3152 + }, + { + "epoch": 0.34625521634087414, + "grad_norm": 1.9816882610321045, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6902596354484558, + "num_tokens": 81275774.0, + "step": 3153 + }, + { + "epoch": 0.3463650340434878, + "grad_norm": 2.4167699813842773, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7189483046531677, + "num_tokens": 81298591.0, + "step": 3154 + }, + { + "epoch": 0.3464748517461015, + "grad_norm": 2.251276731491089, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7219701409339905, + "num_tokens": 81323552.0, + "step": 3155 + }, + { + "epoch": 0.34658466944871513, + "grad_norm": 2.236234426498413, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6853024959564209, + "num_tokens": 81349054.0, + "step": 3156 + }, + { + "epoch": 0.3466944871513288, + "grad_norm": 1.9684138298034668, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7070696353912354, + "num_tokens": 81380180.0, + "step": 3157 + }, + { + "epoch": 0.3468043048539425, + "grad_norm": 2.497504472732544, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6982117891311646, + "num_tokens": 81401090.0, + "step": 3158 + }, + { + "epoch": 0.3469141225565561, + "grad_norm": 2.0478413105010986, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6929475665092468, + "num_tokens": 81431831.0, + "step": 3159 + }, + { + "epoch": 0.34702394025916977, + "grad_norm": 2.0325019359588623, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.682304859161377, + "num_tokens": 81462967.0, + "step": 3160 + }, + { + "epoch": 0.3471337579617834, + "grad_norm": 2.1235005855560303, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7036688327789307, + "num_tokens": 81491878.0, + "step": 3161 + }, + { + "epoch": 0.3472435756643971, + "grad_norm": 2.1368775367736816, + "learning_rate": 1e-06, + "loss": 1.0962, + "mean_token_accuracy": 0.6745163798332214, + "num_tokens": 81522737.0, + "step": 3162 + }, + { + "epoch": 0.34735339336701077, + "grad_norm": 1.9762794971466064, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6942977905273438, + "num_tokens": 81554848.0, + "step": 3163 + }, + { + "epoch": 0.3474632110696244, + "grad_norm": 2.276172399520874, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7113307118415833, + "num_tokens": 81578459.0, + "step": 3164 + }, + { + "epoch": 0.34757302877223806, + "grad_norm": 1.839676856994629, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6987542510032654, + "num_tokens": 81613750.0, + "step": 3165 + }, + { + "epoch": 0.34768284647485176, + "grad_norm": 2.119865655899048, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7074191570281982, + "num_tokens": 81639910.0, + "step": 3166 + }, + { + "epoch": 0.3477926641774654, + "grad_norm": 2.241853952407837, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6866271495819092, + "num_tokens": 81667620.0, + "step": 3167 + }, + { + "epoch": 0.34790248188007905, + "grad_norm": 2.502774715423584, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7149311304092407, + "num_tokens": 81688426.0, + "step": 3168 + }, + { + "epoch": 0.34801229958269275, + "grad_norm": 2.018758773803711, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7152484059333801, + "num_tokens": 81715157.0, + "step": 3169 + }, + { + "epoch": 0.3481221172853064, + "grad_norm": 2.611405611038208, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7111994028091431, + "num_tokens": 81734082.0, + "step": 3170 + }, + { + "epoch": 0.34823193498792004, + "grad_norm": 2.3240180015563965, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7042067646980286, + "num_tokens": 81756297.0, + "step": 3171 + }, + { + "epoch": 0.3483417526905337, + "grad_norm": 2.327507972717285, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.715599536895752, + "num_tokens": 81779881.0, + "step": 3172 + }, + { + "epoch": 0.3484515703931474, + "grad_norm": 2.122560739517212, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7238127589225769, + "num_tokens": 81806868.0, + "step": 3173 + }, + { + "epoch": 0.34856138809576104, + "grad_norm": 2.4078080654144287, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7015774846076965, + "num_tokens": 81828427.0, + "step": 3174 + }, + { + "epoch": 0.3486712057983747, + "grad_norm": 2.277326822280884, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7006470561027527, + "num_tokens": 81852917.0, + "step": 3175 + }, + { + "epoch": 0.3487810235009884, + "grad_norm": 2.1844170093536377, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7060227990150452, + "num_tokens": 81877115.0, + "step": 3176 + }, + { + "epoch": 0.34889084120360203, + "grad_norm": 2.346935987472534, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6951678395271301, + "num_tokens": 81900549.0, + "step": 3177 + }, + { + "epoch": 0.3490006589062157, + "grad_norm": 2.1498219966888428, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6991696357727051, + "num_tokens": 81926096.0, + "step": 3178 + }, + { + "epoch": 0.3491104766088293, + "grad_norm": 2.4529237747192383, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6847033500671387, + "num_tokens": 81948952.0, + "step": 3179 + }, + { + "epoch": 0.349220294311443, + "grad_norm": 2.3775746822357178, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6880491375923157, + "num_tokens": 81972235.0, + "step": 3180 + }, + { + "epoch": 0.34933011201405667, + "grad_norm": 2.4137818813323975, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7042949199676514, + "num_tokens": 81993741.0, + "step": 3181 + }, + { + "epoch": 0.3494399297166703, + "grad_norm": 2.2583413124084473, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6866868138313293, + "num_tokens": 82020457.0, + "step": 3182 + }, + { + "epoch": 0.34954974741928396, + "grad_norm": 2.2432711124420166, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6934680342674255, + "num_tokens": 82046863.0, + "step": 3183 + }, + { + "epoch": 0.34965956512189766, + "grad_norm": 1.9296795129776, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6949366331100464, + "num_tokens": 82079503.0, + "step": 3184 + }, + { + "epoch": 0.3497693828245113, + "grad_norm": 2.4102556705474854, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6988967061042786, + "num_tokens": 82103171.0, + "step": 3185 + }, + { + "epoch": 0.34987920052712496, + "grad_norm": 2.5763614177703857, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7076824903488159, + "num_tokens": 82123461.0, + "step": 3186 + }, + { + "epoch": 0.34998901822973866, + "grad_norm": 2.318870782852173, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.7023639678955078, + "num_tokens": 82148182.0, + "step": 3187 + }, + { + "epoch": 0.3500988359323523, + "grad_norm": 2.4817681312561035, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7033360004425049, + "num_tokens": 82169894.0, + "step": 3188 + }, + { + "epoch": 0.35020865363496595, + "grad_norm": 2.2828848361968994, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7165757417678833, + "num_tokens": 82192351.0, + "step": 3189 + }, + { + "epoch": 0.3503184713375796, + "grad_norm": 2.6583642959594727, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.7017303109169006, + "num_tokens": 82212475.0, + "step": 3190 + }, + { + "epoch": 0.3504282890401933, + "grad_norm": 2.114180088043213, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7042759656906128, + "num_tokens": 82240062.0, + "step": 3191 + }, + { + "epoch": 0.35053810674280694, + "grad_norm": 2.283402919769287, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6941368579864502, + "num_tokens": 82264039.0, + "step": 3192 + }, + { + "epoch": 0.3506479244454206, + "grad_norm": 2.2606194019317627, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7038987874984741, + "num_tokens": 82289065.0, + "step": 3193 + }, + { + "epoch": 0.3507577421480343, + "grad_norm": 2.3005568981170654, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6954933404922485, + "num_tokens": 82315419.0, + "step": 3194 + }, + { + "epoch": 0.35086755985064794, + "grad_norm": 2.19883394241333, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6930408477783203, + "num_tokens": 82343758.0, + "step": 3195 + }, + { + "epoch": 0.3509773775532616, + "grad_norm": 2.720388412475586, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7259335517883301, + "num_tokens": 82360737.0, + "step": 3196 + }, + { + "epoch": 0.35108719525587523, + "grad_norm": 2.2524380683898926, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7046639919281006, + "num_tokens": 82386576.0, + "step": 3197 + }, + { + "epoch": 0.35119701295848893, + "grad_norm": 2.23683500289917, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.7007524967193604, + "num_tokens": 82412169.0, + "step": 3198 + }, + { + "epoch": 0.3513068306611026, + "grad_norm": 2.1751818656921387, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6913177967071533, + "num_tokens": 82437315.0, + "step": 3199 + }, + { + "epoch": 0.3514166483637162, + "grad_norm": 2.1431868076324463, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6898543834686279, + "num_tokens": 82464846.0, + "step": 3200 + }, + { + "epoch": 0.35152646606632987, + "grad_norm": 2.545990228652954, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7162154316902161, + "num_tokens": 82485635.0, + "step": 3201 + }, + { + "epoch": 0.35163628376894357, + "grad_norm": 2.1526827812194824, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7097665071487427, + "num_tokens": 82513187.0, + "step": 3202 + }, + { + "epoch": 0.3517461014715572, + "grad_norm": 2.1228530406951904, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6847987174987793, + "num_tokens": 82541410.0, + "step": 3203 + }, + { + "epoch": 0.35185591917417086, + "grad_norm": 2.2137303352355957, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6917271614074707, + "num_tokens": 82567480.0, + "step": 3204 + }, + { + "epoch": 0.35196573687678456, + "grad_norm": 2.3296053409576416, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7095903158187866, + "num_tokens": 82591959.0, + "step": 3205 + }, + { + "epoch": 0.3520755545793982, + "grad_norm": 1.8928172588348389, + "learning_rate": 1e-06, + "loss": 1.0765, + "mean_token_accuracy": 0.6790412664413452, + "num_tokens": 82627917.0, + "step": 3206 + }, + { + "epoch": 0.35218537228201185, + "grad_norm": 2.231182813644409, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7052873969078064, + "num_tokens": 82652149.0, + "step": 3207 + }, + { + "epoch": 0.3522951899846255, + "grad_norm": 2.2009644508361816, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6862944960594177, + "num_tokens": 82679022.0, + "step": 3208 + }, + { + "epoch": 0.3524050076872392, + "grad_norm": 1.9941129684448242, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6831557154655457, + "num_tokens": 82709752.0, + "step": 3209 + }, + { + "epoch": 0.35251482538985285, + "grad_norm": 2.244149684906006, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7153542637825012, + "num_tokens": 82732210.0, + "step": 3210 + }, + { + "epoch": 0.3526246430924665, + "grad_norm": 2.2157528400421143, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6952440738677979, + "num_tokens": 82756625.0, + "step": 3211 + }, + { + "epoch": 0.35273446079508014, + "grad_norm": 2.1643917560577393, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7095510363578796, + "num_tokens": 82782102.0, + "step": 3212 + }, + { + "epoch": 0.35284427849769384, + "grad_norm": 1.9834091663360596, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7024359107017517, + "num_tokens": 82812454.0, + "step": 3213 + }, + { + "epoch": 0.3529540962003075, + "grad_norm": 2.306797981262207, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6825178265571594, + "num_tokens": 82837912.0, + "step": 3214 + }, + { + "epoch": 0.35306391390292113, + "grad_norm": 2.317671060562134, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6935404539108276, + "num_tokens": 82861389.0, + "step": 3215 + }, + { + "epoch": 0.35317373160553484, + "grad_norm": 1.8479807376861572, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6841531991958618, + "num_tokens": 82898013.0, + "step": 3216 + }, + { + "epoch": 0.3532835493081485, + "grad_norm": 2.2753946781158447, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7033569812774658, + "num_tokens": 82923868.0, + "step": 3217 + }, + { + "epoch": 0.3533933670107621, + "grad_norm": 2.246264934539795, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6980451345443726, + "num_tokens": 82948996.0, + "step": 3218 + }, + { + "epoch": 0.3535031847133758, + "grad_norm": 2.2679269313812256, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6966394186019897, + "num_tokens": 82971959.0, + "step": 3219 + }, + { + "epoch": 0.3536130024159895, + "grad_norm": 2.040602207183838, + "learning_rate": 1e-06, + "loss": 1.0394, + "mean_token_accuracy": 0.6862046718597412, + "num_tokens": 83002093.0, + "step": 3220 + }, + { + "epoch": 0.3537228201186031, + "grad_norm": 2.244387149810791, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6921752095222473, + "num_tokens": 83028000.0, + "step": 3221 + }, + { + "epoch": 0.35383263782121677, + "grad_norm": 2.118070125579834, + "learning_rate": 1e-06, + "loss": 1.1075, + "mean_token_accuracy": 0.6699917316436768, + "num_tokens": 83058008.0, + "step": 3222 + }, + { + "epoch": 0.35394245552383047, + "grad_norm": 2.19323992729187, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.691653847694397, + "num_tokens": 83082527.0, + "step": 3223 + }, + { + "epoch": 0.3540522732264441, + "grad_norm": 2.3571245670318604, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6784717440605164, + "num_tokens": 83107350.0, + "step": 3224 + }, + { + "epoch": 0.35416209092905776, + "grad_norm": 2.4299442768096924, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7244893312454224, + "num_tokens": 83129865.0, + "step": 3225 + }, + { + "epoch": 0.3542719086316714, + "grad_norm": 2.4079911708831787, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7220840454101562, + "num_tokens": 83151073.0, + "step": 3226 + }, + { + "epoch": 0.3543817263342851, + "grad_norm": 1.974870204925537, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6873799562454224, + "num_tokens": 83182618.0, + "step": 3227 + }, + { + "epoch": 0.35449154403689875, + "grad_norm": 2.3234996795654297, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7143139839172363, + "num_tokens": 83205616.0, + "step": 3228 + }, + { + "epoch": 0.3546013617395124, + "grad_norm": 2.2955477237701416, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7417010068893433, + "num_tokens": 83227008.0, + "step": 3229 + }, + { + "epoch": 0.35471117944212605, + "grad_norm": 2.2273428440093994, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7069424390792847, + "num_tokens": 83250757.0, + "step": 3230 + }, + { + "epoch": 0.35482099714473975, + "grad_norm": 2.5876941680908203, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7017358541488647, + "num_tokens": 83272243.0, + "step": 3231 + }, + { + "epoch": 0.3549308148473534, + "grad_norm": 2.1276566982269287, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6853466033935547, + "num_tokens": 83301666.0, + "step": 3232 + }, + { + "epoch": 0.35504063254996704, + "grad_norm": 2.186976194381714, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7189043760299683, + "num_tokens": 83327343.0, + "step": 3233 + }, + { + "epoch": 0.35515045025258074, + "grad_norm": 2.4459640979766846, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6978062391281128, + "num_tokens": 83351141.0, + "step": 3234 + }, + { + "epoch": 0.3552602679551944, + "grad_norm": 2.250617504119873, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7155726552009583, + "num_tokens": 83375601.0, + "step": 3235 + }, + { + "epoch": 0.35537008565780803, + "grad_norm": 2.0770671367645264, + "learning_rate": 1e-06, + "loss": 1.0841, + "mean_token_accuracy": 0.6771432161331177, + "num_tokens": 83405828.0, + "step": 3236 + }, + { + "epoch": 0.3554799033604217, + "grad_norm": 2.2005605697631836, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6987464427947998, + "num_tokens": 83433642.0, + "step": 3237 + }, + { + "epoch": 0.3555897210630354, + "grad_norm": 2.106828212738037, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.720984697341919, + "num_tokens": 83460230.0, + "step": 3238 + }, + { + "epoch": 0.355699538765649, + "grad_norm": 2.335599899291992, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7063365578651428, + "num_tokens": 83483578.0, + "step": 3239 + }, + { + "epoch": 0.35580935646826267, + "grad_norm": 1.9837939739227295, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7085657119750977, + "num_tokens": 83516602.0, + "step": 3240 + }, + { + "epoch": 0.3559191741708763, + "grad_norm": 1.8695862293243408, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7166346311569214, + "num_tokens": 83548112.0, + "step": 3241 + }, + { + "epoch": 0.35602899187349, + "grad_norm": 2.2118008136749268, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7103794813156128, + "num_tokens": 83572067.0, + "step": 3242 + }, + { + "epoch": 0.35613880957610367, + "grad_norm": 2.4098927974700928, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7046563029289246, + "num_tokens": 83595379.0, + "step": 3243 + }, + { + "epoch": 0.3562486272787173, + "grad_norm": 2.3583791255950928, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.690851628780365, + "num_tokens": 83618872.0, + "step": 3244 + }, + { + "epoch": 0.356358444981331, + "grad_norm": 2.036224365234375, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6976766586303711, + "num_tokens": 83647745.0, + "step": 3245 + }, + { + "epoch": 0.35646826268394466, + "grad_norm": 2.579676866531372, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6986071467399597, + "num_tokens": 83668542.0, + "step": 3246 + }, + { + "epoch": 0.3565780803865583, + "grad_norm": 2.039438247680664, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7128944396972656, + "num_tokens": 83698289.0, + "step": 3247 + }, + { + "epoch": 0.35668789808917195, + "grad_norm": 2.2527689933776855, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7143247723579407, + "num_tokens": 83720978.0, + "step": 3248 + }, + { + "epoch": 0.35679771579178565, + "grad_norm": 2.028792381286621, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7015483975410461, + "num_tokens": 83751192.0, + "step": 3249 + }, + { + "epoch": 0.3569075334943993, + "grad_norm": 2.2567641735076904, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6962012648582458, + "num_tokens": 83777227.0, + "step": 3250 + }, + { + "epoch": 0.35701735119701294, + "grad_norm": 2.4211463928222656, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6928752064704895, + "num_tokens": 83799067.0, + "step": 3251 + }, + { + "epoch": 0.35712716889962665, + "grad_norm": 1.9795618057250977, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7143702507019043, + "num_tokens": 83828901.0, + "step": 3252 + }, + { + "epoch": 0.3572369866022403, + "grad_norm": 2.140984058380127, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6907719373703003, + "num_tokens": 83855230.0, + "step": 3253 + }, + { + "epoch": 0.35734680430485394, + "grad_norm": 2.3287734985351562, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6902995705604553, + "num_tokens": 83877973.0, + "step": 3254 + }, + { + "epoch": 0.3574566220074676, + "grad_norm": 2.3527023792266846, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7011080384254456, + "num_tokens": 83904428.0, + "step": 3255 + }, + { + "epoch": 0.3575664397100813, + "grad_norm": 2.2073447704315186, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7045641541481018, + "num_tokens": 83930259.0, + "step": 3256 + }, + { + "epoch": 0.35767625741269493, + "grad_norm": 2.115107536315918, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6925680637359619, + "num_tokens": 83959852.0, + "step": 3257 + }, + { + "epoch": 0.3577860751153086, + "grad_norm": 2.260737419128418, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6987732648849487, + "num_tokens": 83984333.0, + "step": 3258 + }, + { + "epoch": 0.3578958928179222, + "grad_norm": 2.3333444595336914, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7047455310821533, + "num_tokens": 84006768.0, + "step": 3259 + }, + { + "epoch": 0.3580057105205359, + "grad_norm": 1.9021278619766235, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7171987295150757, + "num_tokens": 84037860.0, + "step": 3260 + }, + { + "epoch": 0.35811552822314957, + "grad_norm": 2.395416736602783, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7037833333015442, + "num_tokens": 84059567.0, + "step": 3261 + }, + { + "epoch": 0.3582253459257632, + "grad_norm": 2.258476734161377, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6995326280593872, + "num_tokens": 84084382.0, + "step": 3262 + }, + { + "epoch": 0.3583351636283769, + "grad_norm": 2.2582805156707764, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6982327699661255, + "num_tokens": 84110570.0, + "step": 3263 + }, + { + "epoch": 0.35844498133099056, + "grad_norm": 2.5179388523101807, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7181625962257385, + "num_tokens": 84131918.0, + "step": 3264 + }, + { + "epoch": 0.3585547990336042, + "grad_norm": 2.178485631942749, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6807682514190674, + "num_tokens": 84159266.0, + "step": 3265 + }, + { + "epoch": 0.35866461673621786, + "grad_norm": 2.0128753185272217, + "learning_rate": 1e-06, + "loss": 1.1084, + "mean_token_accuracy": 0.674026370048523, + "num_tokens": 84191739.0, + "step": 3266 + }, + { + "epoch": 0.35877443443883156, + "grad_norm": 2.4532768726348877, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6807734966278076, + "num_tokens": 84214072.0, + "step": 3267 + }, + { + "epoch": 0.3588842521414452, + "grad_norm": 2.183896780014038, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6774078607559204, + "num_tokens": 84242698.0, + "step": 3268 + }, + { + "epoch": 0.35899406984405885, + "grad_norm": 2.2955679893493652, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.7027203440666199, + "num_tokens": 84267032.0, + "step": 3269 + }, + { + "epoch": 0.35910388754667255, + "grad_norm": 2.050537347793579, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6999931931495667, + "num_tokens": 84296606.0, + "step": 3270 + }, + { + "epoch": 0.3592137052492862, + "grad_norm": 2.4192283153533936, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6952455639839172, + "num_tokens": 84319603.0, + "step": 3271 + }, + { + "epoch": 0.35932352295189984, + "grad_norm": 2.178037166595459, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.694341242313385, + "num_tokens": 84344464.0, + "step": 3272 + }, + { + "epoch": 0.3594333406545135, + "grad_norm": 2.4513063430786133, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7067915797233582, + "num_tokens": 84365627.0, + "step": 3273 + }, + { + "epoch": 0.3595431583571272, + "grad_norm": 2.1964213848114014, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6842053532600403, + "num_tokens": 84392647.0, + "step": 3274 + }, + { + "epoch": 0.35965297605974084, + "grad_norm": 2.4320499897003174, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6987115740776062, + "num_tokens": 84415782.0, + "step": 3275 + }, + { + "epoch": 0.3597627937623545, + "grad_norm": 2.208392381668091, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.6796939373016357, + "num_tokens": 84444286.0, + "step": 3276 + }, + { + "epoch": 0.35987261146496813, + "grad_norm": 2.45228910446167, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6934307813644409, + "num_tokens": 84465884.0, + "step": 3277 + }, + { + "epoch": 0.35998242916758183, + "grad_norm": 2.0644819736480713, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6882017850875854, + "num_tokens": 84494886.0, + "step": 3278 + }, + { + "epoch": 0.3600922468701955, + "grad_norm": 2.1915767192840576, + "learning_rate": 1e-06, + "loss": 1.0969, + "mean_token_accuracy": 0.6790242195129395, + "num_tokens": 84523152.0, + "step": 3279 + }, + { + "epoch": 0.3602020645728091, + "grad_norm": 2.5083792209625244, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7136775255203247, + "num_tokens": 84544187.0, + "step": 3280 + }, + { + "epoch": 0.3603118822754228, + "grad_norm": 2.3881664276123047, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7078636884689331, + "num_tokens": 84566417.0, + "step": 3281 + }, + { + "epoch": 0.36042169997803647, + "grad_norm": 2.44264554977417, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7095292806625366, + "num_tokens": 84588621.0, + "step": 3282 + }, + { + "epoch": 0.3605315176806501, + "grad_norm": 2.170604705810547, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7195463180541992, + "num_tokens": 84614038.0, + "step": 3283 + }, + { + "epoch": 0.36064133538326376, + "grad_norm": 2.0675175189971924, + "learning_rate": 1e-06, + "loss": 1.1109, + "mean_token_accuracy": 0.6675527095794678, + "num_tokens": 84643759.0, + "step": 3284 + }, + { + "epoch": 0.36075115308587746, + "grad_norm": 2.408066749572754, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7124336361885071, + "num_tokens": 84665687.0, + "step": 3285 + }, + { + "epoch": 0.3608609707884911, + "grad_norm": 2.5679001808166504, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7104379534721375, + "num_tokens": 84685335.0, + "step": 3286 + }, + { + "epoch": 0.36097078849110475, + "grad_norm": 2.736335277557373, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7074975967407227, + "num_tokens": 84702626.0, + "step": 3287 + }, + { + "epoch": 0.3610806061937184, + "grad_norm": 1.924810528755188, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6829306483268738, + "num_tokens": 84733509.0, + "step": 3288 + }, + { + "epoch": 0.3611904238963321, + "grad_norm": 2.1794090270996094, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7103796005249023, + "num_tokens": 84758406.0, + "step": 3289 + }, + { + "epoch": 0.36130024159894575, + "grad_norm": 2.258791923522949, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7277464866638184, + "num_tokens": 84780759.0, + "step": 3290 + }, + { + "epoch": 0.3614100593015594, + "grad_norm": 2.146876811981201, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7068725228309631, + "num_tokens": 84808210.0, + "step": 3291 + }, + { + "epoch": 0.3615198770041731, + "grad_norm": 2.2355027198791504, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6804673075675964, + "num_tokens": 84834414.0, + "step": 3292 + }, + { + "epoch": 0.36162969470678674, + "grad_norm": 2.2196171283721924, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7099621295928955, + "num_tokens": 84862592.0, + "step": 3293 + }, + { + "epoch": 0.3617395124094004, + "grad_norm": 2.2008578777313232, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7009135484695435, + "num_tokens": 84890482.0, + "step": 3294 + }, + { + "epoch": 0.36184933011201403, + "grad_norm": 2.2603039741516113, + "learning_rate": 1e-06, + "loss": 1.11, + "mean_token_accuracy": 0.6700689196586609, + "num_tokens": 84919571.0, + "step": 3295 + }, + { + "epoch": 0.36195914781462774, + "grad_norm": 2.0580434799194336, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7028911113739014, + "num_tokens": 84949846.0, + "step": 3296 + }, + { + "epoch": 0.3620689655172414, + "grad_norm": 1.9071799516677856, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.681530237197876, + "num_tokens": 84986287.0, + "step": 3297 + }, + { + "epoch": 0.362178783219855, + "grad_norm": 2.1312685012817383, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7137593030929565, + "num_tokens": 85012326.0, + "step": 3298 + }, + { + "epoch": 0.36228860092246873, + "grad_norm": 1.9989008903503418, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6903355717658997, + "num_tokens": 85045947.0, + "step": 3299 + }, + { + "epoch": 0.3623984186250824, + "grad_norm": 2.1781697273254395, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7010506987571716, + "num_tokens": 85071970.0, + "step": 3300 + }, + { + "epoch": 0.362508236327696, + "grad_norm": 2.1982905864715576, + "learning_rate": 1e-06, + "loss": 1.0938, + "mean_token_accuracy": 0.6769605875015259, + "num_tokens": 85102810.0, + "step": 3301 + }, + { + "epoch": 0.36261805403030967, + "grad_norm": 2.5399534702301025, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7208781838417053, + "num_tokens": 85123893.0, + "step": 3302 + }, + { + "epoch": 0.36272787173292337, + "grad_norm": 2.2957894802093506, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7159067392349243, + "num_tokens": 85148603.0, + "step": 3303 + }, + { + "epoch": 0.362837689435537, + "grad_norm": 2.2196998596191406, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.7000976204872131, + "num_tokens": 85175730.0, + "step": 3304 + }, + { + "epoch": 0.36294750713815066, + "grad_norm": 2.235318899154663, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7159619331359863, + "num_tokens": 85198485.0, + "step": 3305 + }, + { + "epoch": 0.3630573248407643, + "grad_norm": 2.199694871902466, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6956774592399597, + "num_tokens": 85223856.0, + "step": 3306 + }, + { + "epoch": 0.363167142543378, + "grad_norm": 2.565906047821045, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6974682211875916, + "num_tokens": 85247465.0, + "step": 3307 + }, + { + "epoch": 0.36327696024599165, + "grad_norm": 2.4470112323760986, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7012647390365601, + "num_tokens": 85271028.0, + "step": 3308 + }, + { + "epoch": 0.3633867779486053, + "grad_norm": 2.5463995933532715, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6904101967811584, + "num_tokens": 85291704.0, + "step": 3309 + }, + { + "epoch": 0.363496595651219, + "grad_norm": 2.3761398792266846, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6983246207237244, + "num_tokens": 85315335.0, + "step": 3310 + }, + { + "epoch": 0.36360641335383265, + "grad_norm": 2.216463327407837, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6974495053291321, + "num_tokens": 85340240.0, + "step": 3311 + }, + { + "epoch": 0.3637162310564463, + "grad_norm": 2.1797966957092285, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7079236507415771, + "num_tokens": 85367498.0, + "step": 3312 + }, + { + "epoch": 0.36382604875905994, + "grad_norm": 2.1658132076263428, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6992059946060181, + "num_tokens": 85395564.0, + "step": 3313 + }, + { + "epoch": 0.36393586646167364, + "grad_norm": 2.007916212081909, + "learning_rate": 1e-06, + "loss": 1.0652, + "mean_token_accuracy": 0.6894936561584473, + "num_tokens": 85427724.0, + "step": 3314 + }, + { + "epoch": 0.3640456841642873, + "grad_norm": 1.9431484937667847, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6886768341064453, + "num_tokens": 85459224.0, + "step": 3315 + }, + { + "epoch": 0.36415550186690093, + "grad_norm": 2.3824574947357178, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7048172950744629, + "num_tokens": 85481566.0, + "step": 3316 + }, + { + "epoch": 0.3642653195695146, + "grad_norm": 2.3454856872558594, + "learning_rate": 1e-06, + "loss": 1.0989, + "mean_token_accuracy": 0.6704624891281128, + "num_tokens": 85506779.0, + "step": 3317 + }, + { + "epoch": 0.3643751372721283, + "grad_norm": 2.0652525424957275, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6824414134025574, + "num_tokens": 85536050.0, + "step": 3318 + }, + { + "epoch": 0.3644849549747419, + "grad_norm": 2.0181093215942383, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6817924976348877, + "num_tokens": 85567036.0, + "step": 3319 + }, + { + "epoch": 0.36459477267735557, + "grad_norm": 2.335224151611328, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7254833579063416, + "num_tokens": 85590778.0, + "step": 3320 + }, + { + "epoch": 0.3647045903799693, + "grad_norm": 2.3575315475463867, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7028683423995972, + "num_tokens": 85615662.0, + "step": 3321 + }, + { + "epoch": 0.3648144080825829, + "grad_norm": 1.9472391605377197, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6972997188568115, + "num_tokens": 85648345.0, + "step": 3322 + }, + { + "epoch": 0.36492422578519657, + "grad_norm": 2.508164167404175, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7030770182609558, + "num_tokens": 85669125.0, + "step": 3323 + }, + { + "epoch": 0.3650340434878102, + "grad_norm": 2.2507266998291016, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7088370323181152, + "num_tokens": 85695044.0, + "step": 3324 + }, + { + "epoch": 0.3651438611904239, + "grad_norm": 2.097538709640503, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7101156711578369, + "num_tokens": 85721944.0, + "step": 3325 + }, + { + "epoch": 0.36525367889303756, + "grad_norm": 2.5179672241210938, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6958365440368652, + "num_tokens": 85743080.0, + "step": 3326 + }, + { + "epoch": 0.3653634965956512, + "grad_norm": 2.512192964553833, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7056947350502014, + "num_tokens": 85764096.0, + "step": 3327 + }, + { + "epoch": 0.3654733142982649, + "grad_norm": 2.2207460403442383, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7100839614868164, + "num_tokens": 85790882.0, + "step": 3328 + }, + { + "epoch": 0.36558313200087855, + "grad_norm": 2.371638536453247, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7088741064071655, + "num_tokens": 85814095.0, + "step": 3329 + }, + { + "epoch": 0.3656929497034922, + "grad_norm": 1.9480940103530884, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7062416076660156, + "num_tokens": 85845497.0, + "step": 3330 + }, + { + "epoch": 0.36580276740610584, + "grad_norm": 2.102612257003784, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6952446699142456, + "num_tokens": 85874112.0, + "step": 3331 + }, + { + "epoch": 0.36591258510871955, + "grad_norm": 2.528123378753662, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7107318639755249, + "num_tokens": 85893507.0, + "step": 3332 + }, + { + "epoch": 0.3660224028113332, + "grad_norm": 2.68985915184021, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6938000917434692, + "num_tokens": 85913328.0, + "step": 3333 + }, + { + "epoch": 0.36613222051394684, + "grad_norm": 2.0099387168884277, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.715183675289154, + "num_tokens": 85943051.0, + "step": 3334 + }, + { + "epoch": 0.3662420382165605, + "grad_norm": 2.0673632621765137, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7173911333084106, + "num_tokens": 85970740.0, + "step": 3335 + }, + { + "epoch": 0.3663518559191742, + "grad_norm": 2.336829423904419, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6993758678436279, + "num_tokens": 85995014.0, + "step": 3336 + }, + { + "epoch": 0.36646167362178783, + "grad_norm": 2.1076974868774414, + "learning_rate": 1e-06, + "loss": 1.0699, + "mean_token_accuracy": 0.6804921627044678, + "num_tokens": 86025015.0, + "step": 3337 + }, + { + "epoch": 0.3665714913244015, + "grad_norm": 2.0629544258117676, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7034348845481873, + "num_tokens": 86052376.0, + "step": 3338 + }, + { + "epoch": 0.3666813090270152, + "grad_norm": 2.2088887691497803, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6950749158859253, + "num_tokens": 86076889.0, + "step": 3339 + }, + { + "epoch": 0.3667911267296288, + "grad_norm": 2.3807857036590576, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7039353847503662, + "num_tokens": 86099591.0, + "step": 3340 + }, + { + "epoch": 0.36690094443224247, + "grad_norm": 2.312321662902832, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6888852119445801, + "num_tokens": 86124833.0, + "step": 3341 + }, + { + "epoch": 0.3670107621348561, + "grad_norm": 2.0483016967773438, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6888394355773926, + "num_tokens": 86154312.0, + "step": 3342 + }, + { + "epoch": 0.3671205798374698, + "grad_norm": 2.834602117538452, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7419738173484802, + "num_tokens": 86170667.0, + "step": 3343 + }, + { + "epoch": 0.36723039754008346, + "grad_norm": 2.363515853881836, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.69286048412323, + "num_tokens": 86194319.0, + "step": 3344 + }, + { + "epoch": 0.3673402152426971, + "grad_norm": 2.1946139335632324, + "learning_rate": 1e-06, + "loss": 1.0957, + "mean_token_accuracy": 0.669161319732666, + "num_tokens": 86222404.0, + "step": 3345 + }, + { + "epoch": 0.3674500329453108, + "grad_norm": 1.977504014968872, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6985313892364502, + "num_tokens": 86252513.0, + "step": 3346 + }, + { + "epoch": 0.36755985064792446, + "grad_norm": 2.2161853313446045, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7035475969314575, + "num_tokens": 86279323.0, + "step": 3347 + }, + { + "epoch": 0.3676696683505381, + "grad_norm": 2.4369237422943115, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7051978707313538, + "num_tokens": 86302361.0, + "step": 3348 + }, + { + "epoch": 0.36777948605315175, + "grad_norm": 2.373579978942871, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6916196942329407, + "num_tokens": 86325945.0, + "step": 3349 + }, + { + "epoch": 0.36788930375576545, + "grad_norm": 2.482285499572754, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7033820152282715, + "num_tokens": 86347707.0, + "step": 3350 + }, + { + "epoch": 0.3679991214583791, + "grad_norm": 2.308631420135498, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6991403102874756, + "num_tokens": 86371775.0, + "step": 3351 + }, + { + "epoch": 0.36810893916099274, + "grad_norm": 2.2310564517974854, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7075667977333069, + "num_tokens": 86397287.0, + "step": 3352 + }, + { + "epoch": 0.3682187568636064, + "grad_norm": 2.319824695587158, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7111926078796387, + "num_tokens": 86421037.0, + "step": 3353 + }, + { + "epoch": 0.3683285745662201, + "grad_norm": 2.088474750518799, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6910104751586914, + "num_tokens": 86450440.0, + "step": 3354 + }, + { + "epoch": 0.36843839226883374, + "grad_norm": 2.2768282890319824, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7027931213378906, + "num_tokens": 86473493.0, + "step": 3355 + }, + { + "epoch": 0.3685482099714474, + "grad_norm": 2.127786159515381, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7069418430328369, + "num_tokens": 86502660.0, + "step": 3356 + }, + { + "epoch": 0.3686580276740611, + "grad_norm": 2.1904702186584473, + "learning_rate": 1e-06, + "loss": 1.1136, + "mean_token_accuracy": 0.6704638004302979, + "num_tokens": 86532021.0, + "step": 3357 + }, + { + "epoch": 0.36876784537667473, + "grad_norm": 2.081956624984741, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7112675905227661, + "num_tokens": 86559004.0, + "step": 3358 + }, + { + "epoch": 0.3688776630792884, + "grad_norm": 2.428244113922119, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6870169639587402, + "num_tokens": 86579948.0, + "step": 3359 + }, + { + "epoch": 0.368987480781902, + "grad_norm": 2.3077704906463623, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6862739324569702, + "num_tokens": 86604495.0, + "step": 3360 + }, + { + "epoch": 0.3690972984845157, + "grad_norm": 2.455639362335205, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7004948854446411, + "num_tokens": 86625049.0, + "step": 3361 + }, + { + "epoch": 0.36920711618712937, + "grad_norm": 2.009989023208618, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6987300515174866, + "num_tokens": 86654312.0, + "step": 3362 + }, + { + "epoch": 0.369316933889743, + "grad_norm": 2.2391750812530518, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6864262819290161, + "num_tokens": 86681832.0, + "step": 3363 + }, + { + "epoch": 0.36942675159235666, + "grad_norm": 2.4950897693634033, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6928977370262146, + "num_tokens": 86702785.0, + "step": 3364 + }, + { + "epoch": 0.36953656929497036, + "grad_norm": 2.445114850997925, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6991971731185913, + "num_tokens": 86726223.0, + "step": 3365 + }, + { + "epoch": 0.369646386997584, + "grad_norm": 2.327428102493286, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.6780712604522705, + "num_tokens": 86751263.0, + "step": 3366 + }, + { + "epoch": 0.36975620470019765, + "grad_norm": 2.1440632343292236, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7203820943832397, + "num_tokens": 86778313.0, + "step": 3367 + }, + { + "epoch": 0.36986602240281136, + "grad_norm": 2.113633871078491, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6953961849212646, + "num_tokens": 86809337.0, + "step": 3368 + }, + { + "epoch": 0.369975840105425, + "grad_norm": 2.3359827995300293, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7006925344467163, + "num_tokens": 86832742.0, + "step": 3369 + }, + { + "epoch": 0.37008565780803865, + "grad_norm": 2.175438642501831, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7015551328659058, + "num_tokens": 86858918.0, + "step": 3370 + }, + { + "epoch": 0.3701954755106523, + "grad_norm": 2.200671672821045, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7131904363632202, + "num_tokens": 86887401.0, + "step": 3371 + }, + { + "epoch": 0.370305293213266, + "grad_norm": 2.2129523754119873, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7010505795478821, + "num_tokens": 86913846.0, + "step": 3372 + }, + { + "epoch": 0.37041511091587964, + "grad_norm": 2.1269497871398926, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6892755627632141, + "num_tokens": 86943496.0, + "step": 3373 + }, + { + "epoch": 0.3705249286184933, + "grad_norm": 2.1638553142547607, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6980169415473938, + "num_tokens": 86971140.0, + "step": 3374 + }, + { + "epoch": 0.370634746321107, + "grad_norm": 2.2226521968841553, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7010671496391296, + "num_tokens": 86996552.0, + "step": 3375 + }, + { + "epoch": 0.37074456402372064, + "grad_norm": 2.3500165939331055, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6887003183364868, + "num_tokens": 87019588.0, + "step": 3376 + }, + { + "epoch": 0.3708543817263343, + "grad_norm": 2.074510335922241, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7190811038017273, + "num_tokens": 87047452.0, + "step": 3377 + }, + { + "epoch": 0.3709641994289479, + "grad_norm": 2.373157501220703, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7256876826286316, + "num_tokens": 87068075.0, + "step": 3378 + }, + { + "epoch": 0.37107401713156163, + "grad_norm": 2.469404458999634, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7162970304489136, + "num_tokens": 87089256.0, + "step": 3379 + }, + { + "epoch": 0.3711838348341753, + "grad_norm": 1.9927459955215454, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6938840746879578, + "num_tokens": 87119727.0, + "step": 3380 + }, + { + "epoch": 0.3712936525367889, + "grad_norm": 2.254403591156006, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7013788819313049, + "num_tokens": 87145110.0, + "step": 3381 + }, + { + "epoch": 0.37140347023940257, + "grad_norm": 2.342153549194336, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7080563902854919, + "num_tokens": 87168527.0, + "step": 3382 + }, + { + "epoch": 0.37151328794201627, + "grad_norm": 2.078409194946289, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7601994276046753, + "num_tokens": 87192421.0, + "step": 3383 + }, + { + "epoch": 0.3716231056446299, + "grad_norm": 2.212850570678711, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6992839574813843, + "num_tokens": 87218087.0, + "step": 3384 + }, + { + "epoch": 0.37173292334724356, + "grad_norm": 2.465736150741577, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6953601241111755, + "num_tokens": 87242574.0, + "step": 3385 + }, + { + "epoch": 0.37184274104985726, + "grad_norm": 1.9573055505752563, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.6787867546081543, + "num_tokens": 87278713.0, + "step": 3386 + }, + { + "epoch": 0.3719525587524709, + "grad_norm": 2.474886655807495, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7080777883529663, + "num_tokens": 87302807.0, + "step": 3387 + }, + { + "epoch": 0.37206237645508455, + "grad_norm": 2.4273104667663574, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7058053016662598, + "num_tokens": 87324698.0, + "step": 3388 + }, + { + "epoch": 0.3721721941576982, + "grad_norm": 2.4841413497924805, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7040523886680603, + "num_tokens": 87345257.0, + "step": 3389 + }, + { + "epoch": 0.3722820118603119, + "grad_norm": 2.184096097946167, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7058951258659363, + "num_tokens": 87370252.0, + "step": 3390 + }, + { + "epoch": 0.37239182956292555, + "grad_norm": 2.398867607116699, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.710202693939209, + "num_tokens": 87392679.0, + "step": 3391 + }, + { + "epoch": 0.3725016472655392, + "grad_norm": 2.161041736602783, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6909757256507874, + "num_tokens": 87422129.0, + "step": 3392 + }, + { + "epoch": 0.37261146496815284, + "grad_norm": 2.318533182144165, + "learning_rate": 1e-06, + "loss": 1.0946, + "mean_token_accuracy": 0.6797225475311279, + "num_tokens": 87445631.0, + "step": 3393 + }, + { + "epoch": 0.37272128267076654, + "grad_norm": 2.2894399166107178, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7019973993301392, + "num_tokens": 87467923.0, + "step": 3394 + }, + { + "epoch": 0.3728311003733802, + "grad_norm": 2.042438268661499, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6747955083847046, + "num_tokens": 87498181.0, + "step": 3395 + }, + { + "epoch": 0.37294091807599383, + "grad_norm": 2.221186399459839, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6979097723960876, + "num_tokens": 87523749.0, + "step": 3396 + }, + { + "epoch": 0.37305073577860753, + "grad_norm": 2.0304360389709473, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6909283399581909, + "num_tokens": 87553253.0, + "step": 3397 + }, + { + "epoch": 0.3731605534812212, + "grad_norm": 2.0951712131500244, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7102013826370239, + "num_tokens": 87579169.0, + "step": 3398 + }, + { + "epoch": 0.3732703711838348, + "grad_norm": 2.2405929565429688, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7075052261352539, + "num_tokens": 87604559.0, + "step": 3399 + }, + { + "epoch": 0.37338018888644847, + "grad_norm": 2.123347282409668, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6879837512969971, + "num_tokens": 87634162.0, + "step": 3400 + }, + { + "epoch": 0.3734900065890622, + "grad_norm": 2.0109784603118896, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.7019926309585571, + "num_tokens": 87665827.0, + "step": 3401 + }, + { + "epoch": 0.3735998242916758, + "grad_norm": 2.202221155166626, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7200646996498108, + "num_tokens": 87691897.0, + "step": 3402 + }, + { + "epoch": 0.37370964199428947, + "grad_norm": 2.150979995727539, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7206356525421143, + "num_tokens": 87716262.0, + "step": 3403 + }, + { + "epoch": 0.37381945969690317, + "grad_norm": 2.037590265274048, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6967258453369141, + "num_tokens": 87746323.0, + "step": 3404 + }, + { + "epoch": 0.3739292773995168, + "grad_norm": 2.148174524307251, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7194364070892334, + "num_tokens": 87774299.0, + "step": 3405 + }, + { + "epoch": 0.37403909510213046, + "grad_norm": 2.436582565307617, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7074482440948486, + "num_tokens": 87796350.0, + "step": 3406 + }, + { + "epoch": 0.3741489128047441, + "grad_norm": 2.172363519668579, + "learning_rate": 1e-06, + "loss": 1.093, + "mean_token_accuracy": 0.6753835082054138, + "num_tokens": 87826953.0, + "step": 3407 + }, + { + "epoch": 0.3742587305073578, + "grad_norm": 2.026548385620117, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7008572816848755, + "num_tokens": 87857320.0, + "step": 3408 + }, + { + "epoch": 0.37436854820997145, + "grad_norm": 2.0954508781433105, + "learning_rate": 1e-06, + "loss": 1.0942, + "mean_token_accuracy": 0.6730766296386719, + "num_tokens": 87889552.0, + "step": 3409 + }, + { + "epoch": 0.3744783659125851, + "grad_norm": 1.9631136655807495, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6916292905807495, + "num_tokens": 87923642.0, + "step": 3410 + }, + { + "epoch": 0.37458818361519874, + "grad_norm": 2.5617988109588623, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.7016984820365906, + "num_tokens": 87944526.0, + "step": 3411 + }, + { + "epoch": 0.37469800131781245, + "grad_norm": 2.295383930206299, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6994149684906006, + "num_tokens": 87969510.0, + "step": 3412 + }, + { + "epoch": 0.3748078190204261, + "grad_norm": 2.0215260982513428, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7210351228713989, + "num_tokens": 87998804.0, + "step": 3413 + }, + { + "epoch": 0.37491763672303974, + "grad_norm": 1.9982987642288208, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.7002013921737671, + "num_tokens": 88029812.0, + "step": 3414 + }, + { + "epoch": 0.37502745442565344, + "grad_norm": 2.3212225437164307, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7114932537078857, + "num_tokens": 88053492.0, + "step": 3415 + }, + { + "epoch": 0.3751372721282671, + "grad_norm": 2.7221145629882812, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.700993001461029, + "num_tokens": 88074658.0, + "step": 3416 + }, + { + "epoch": 0.37524708983088073, + "grad_norm": 2.30668568611145, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7133240699768066, + "num_tokens": 88098660.0, + "step": 3417 + }, + { + "epoch": 0.3753569075334944, + "grad_norm": 2.436951160430908, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6956903338432312, + "num_tokens": 88123408.0, + "step": 3418 + }, + { + "epoch": 0.3754667252361081, + "grad_norm": 2.4527781009674072, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6854534149169922, + "num_tokens": 88147472.0, + "step": 3419 + }, + { + "epoch": 0.3755765429387217, + "grad_norm": 2.1603710651397705, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6985584497451782, + "num_tokens": 88173034.0, + "step": 3420 + }, + { + "epoch": 0.37568636064133537, + "grad_norm": 2.173532247543335, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6872550845146179, + "num_tokens": 88198727.0, + "step": 3421 + }, + { + "epoch": 0.37579617834394907, + "grad_norm": 2.353398084640503, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7151634693145752, + "num_tokens": 88222959.0, + "step": 3422 + }, + { + "epoch": 0.3759059960465627, + "grad_norm": 2.06463623046875, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7104458212852478, + "num_tokens": 88251576.0, + "step": 3423 + }, + { + "epoch": 0.37601581374917636, + "grad_norm": 1.8985751867294312, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7003107666969299, + "num_tokens": 88283073.0, + "step": 3424 + }, + { + "epoch": 0.37612563145179, + "grad_norm": 2.004653215408325, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6931624412536621, + "num_tokens": 88313535.0, + "step": 3425 + }, + { + "epoch": 0.3762354491544037, + "grad_norm": 2.206322193145752, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6872548460960388, + "num_tokens": 88340230.0, + "step": 3426 + }, + { + "epoch": 0.37634526685701736, + "grad_norm": 2.247570514678955, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6954503059387207, + "num_tokens": 88366285.0, + "step": 3427 + }, + { + "epoch": 0.376455084559631, + "grad_norm": 2.1506295204162598, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.7010297775268555, + "num_tokens": 88393525.0, + "step": 3428 + }, + { + "epoch": 0.37656490226224465, + "grad_norm": 2.4051249027252197, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7208294868469238, + "num_tokens": 88415088.0, + "step": 3429 + }, + { + "epoch": 0.37667471996485835, + "grad_norm": 2.082717180252075, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6847718954086304, + "num_tokens": 88445130.0, + "step": 3430 + }, + { + "epoch": 0.376784537667472, + "grad_norm": 2.0008721351623535, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7117443084716797, + "num_tokens": 88474987.0, + "step": 3431 + }, + { + "epoch": 0.37689435537008564, + "grad_norm": 2.590919017791748, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7091534733772278, + "num_tokens": 88494886.0, + "step": 3432 + }, + { + "epoch": 0.37700417307269934, + "grad_norm": 2.1021504402160645, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6903313398361206, + "num_tokens": 88523428.0, + "step": 3433 + }, + { + "epoch": 0.377113990775313, + "grad_norm": 2.318145751953125, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6930568218231201, + "num_tokens": 88547064.0, + "step": 3434 + }, + { + "epoch": 0.37722380847792664, + "grad_norm": 2.4214658737182617, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7089401483535767, + "num_tokens": 88568534.0, + "step": 3435 + }, + { + "epoch": 0.3773336261805403, + "grad_norm": 2.3127245903015137, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6933971643447876, + "num_tokens": 88591240.0, + "step": 3436 + }, + { + "epoch": 0.377443443883154, + "grad_norm": 2.2001895904541016, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7014461755752563, + "num_tokens": 88617396.0, + "step": 3437 + }, + { + "epoch": 0.37755326158576763, + "grad_norm": 2.0333974361419678, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7094371318817139, + "num_tokens": 88647858.0, + "step": 3438 + }, + { + "epoch": 0.3776630792883813, + "grad_norm": 2.3231897354125977, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7096344828605652, + "num_tokens": 88672195.0, + "step": 3439 + }, + { + "epoch": 0.3777728969909949, + "grad_norm": 1.8713048696517944, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6889094710350037, + "num_tokens": 88704658.0, + "step": 3440 + }, + { + "epoch": 0.3778827146936086, + "grad_norm": 2.066563367843628, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6942192316055298, + "num_tokens": 88732923.0, + "step": 3441 + }, + { + "epoch": 0.37799253239622227, + "grad_norm": 2.4016165733337402, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6930129528045654, + "num_tokens": 88756601.0, + "step": 3442 + }, + { + "epoch": 0.3781023500988359, + "grad_norm": 2.272390604019165, + "learning_rate": 1e-06, + "loss": 1.0646, + "mean_token_accuracy": 0.689227283000946, + "num_tokens": 88782484.0, + "step": 3443 + }, + { + "epoch": 0.3782121678014496, + "grad_norm": 2.369533061981201, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6958010196685791, + "num_tokens": 88805581.0, + "step": 3444 + }, + { + "epoch": 0.37832198550406326, + "grad_norm": 2.109570264816284, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6919628977775574, + "num_tokens": 88831736.0, + "step": 3445 + }, + { + "epoch": 0.3784318032066769, + "grad_norm": 2.198310613632202, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6811305284500122, + "num_tokens": 88859620.0, + "step": 3446 + }, + { + "epoch": 0.37854162090929055, + "grad_norm": 2.2991504669189453, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6962951421737671, + "num_tokens": 88882753.0, + "step": 3447 + }, + { + "epoch": 0.37865143861190426, + "grad_norm": 2.2297420501708984, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6957863569259644, + "num_tokens": 88909699.0, + "step": 3448 + }, + { + "epoch": 0.3787612563145179, + "grad_norm": 2.4685254096984863, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7414649128913879, + "num_tokens": 88932726.0, + "step": 3449 + }, + { + "epoch": 0.37887107401713155, + "grad_norm": 2.3317902088165283, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.7036958932876587, + "num_tokens": 88955786.0, + "step": 3450 + }, + { + "epoch": 0.37898089171974525, + "grad_norm": 2.2672271728515625, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6800941228866577, + "num_tokens": 88980538.0, + "step": 3451 + }, + { + "epoch": 0.3790907094223589, + "grad_norm": 1.9860020875930786, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7046587467193604, + "num_tokens": 89010705.0, + "step": 3452 + }, + { + "epoch": 0.37920052712497254, + "grad_norm": 1.979961633682251, + "learning_rate": 1e-06, + "loss": 1.0885, + "mean_token_accuracy": 0.678202748298645, + "num_tokens": 89042849.0, + "step": 3453 + }, + { + "epoch": 0.3793103448275862, + "grad_norm": 2.8576643466949463, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7060858011245728, + "num_tokens": 89063832.0, + "step": 3454 + }, + { + "epoch": 0.3794201625301999, + "grad_norm": 2.4057998657226562, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7127039432525635, + "num_tokens": 89085697.0, + "step": 3455 + }, + { + "epoch": 0.37952998023281354, + "grad_norm": 2.270477771759033, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7247939109802246, + "num_tokens": 89109639.0, + "step": 3456 + }, + { + "epoch": 0.3796397979354272, + "grad_norm": 2.5402557849884033, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7120809555053711, + "num_tokens": 89130660.0, + "step": 3457 + }, + { + "epoch": 0.3797496156380408, + "grad_norm": 2.3424911499023438, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.702118992805481, + "num_tokens": 89154426.0, + "step": 3458 + }, + { + "epoch": 0.37985943334065453, + "grad_norm": 2.5863378047943115, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.7066689729690552, + "num_tokens": 89176524.0, + "step": 3459 + }, + { + "epoch": 0.3799692510432682, + "grad_norm": 2.491591453552246, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7124779224395752, + "num_tokens": 89199003.0, + "step": 3460 + }, + { + "epoch": 0.3800790687458818, + "grad_norm": 2.322617292404175, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6902390122413635, + "num_tokens": 89226177.0, + "step": 3461 + }, + { + "epoch": 0.3801888864484955, + "grad_norm": 2.020094871520996, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.6811436414718628, + "num_tokens": 89258350.0, + "step": 3462 + }, + { + "epoch": 0.38029870415110917, + "grad_norm": 2.3631491661071777, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7248357534408569, + "num_tokens": 89283235.0, + "step": 3463 + }, + { + "epoch": 0.3804085218537228, + "grad_norm": 2.6576085090637207, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7055003643035889, + "num_tokens": 89303639.0, + "step": 3464 + }, + { + "epoch": 0.38051833955633646, + "grad_norm": 2.2513339519500732, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7260462641716003, + "num_tokens": 89330264.0, + "step": 3465 + }, + { + "epoch": 0.38062815725895016, + "grad_norm": 2.201817035675049, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7193719744682312, + "num_tokens": 89354842.0, + "step": 3466 + }, + { + "epoch": 0.3807379749615638, + "grad_norm": 2.0771729946136475, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6854639053344727, + "num_tokens": 89385770.0, + "step": 3467 + }, + { + "epoch": 0.38084779266417745, + "grad_norm": 2.448106050491333, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7048771381378174, + "num_tokens": 89409175.0, + "step": 3468 + }, + { + "epoch": 0.3809576103667911, + "grad_norm": 2.0836915969848633, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7131275534629822, + "num_tokens": 89436266.0, + "step": 3469 + }, + { + "epoch": 0.3810674280694048, + "grad_norm": 1.9489612579345703, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7390809059143066, + "num_tokens": 89462843.0, + "step": 3470 + }, + { + "epoch": 0.38117724577201845, + "grad_norm": 2.261678457260132, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7146016359329224, + "num_tokens": 89486440.0, + "step": 3471 + }, + { + "epoch": 0.3812870634746321, + "grad_norm": 2.3215370178222656, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6940683722496033, + "num_tokens": 89510909.0, + "step": 3472 + }, + { + "epoch": 0.3813968811772458, + "grad_norm": 2.226550817489624, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7189731001853943, + "num_tokens": 89533898.0, + "step": 3473 + }, + { + "epoch": 0.38150669887985944, + "grad_norm": 2.3648223876953125, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7010132074356079, + "num_tokens": 89558818.0, + "step": 3474 + }, + { + "epoch": 0.3816165165824731, + "grad_norm": 2.0938057899475098, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6863538026809692, + "num_tokens": 89587889.0, + "step": 3475 + }, + { + "epoch": 0.38172633428508673, + "grad_norm": 2.10738468170166, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6892518997192383, + "num_tokens": 89617069.0, + "step": 3476 + }, + { + "epoch": 0.38183615198770043, + "grad_norm": 2.38873291015625, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6952362656593323, + "num_tokens": 89639922.0, + "step": 3477 + }, + { + "epoch": 0.3819459696903141, + "grad_norm": 2.123931646347046, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7278136610984802, + "num_tokens": 89667156.0, + "step": 3478 + }, + { + "epoch": 0.3820557873929277, + "grad_norm": 2.3538053035736084, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7244670391082764, + "num_tokens": 89688905.0, + "step": 3479 + }, + { + "epoch": 0.3821656050955414, + "grad_norm": 2.1959152221679688, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.687269389629364, + "num_tokens": 89714888.0, + "step": 3480 + }, + { + "epoch": 0.3822754227981551, + "grad_norm": 2.0849287509918213, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7042725086212158, + "num_tokens": 89742883.0, + "step": 3481 + }, + { + "epoch": 0.3823852405007687, + "grad_norm": 2.3714652061462402, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7043930292129517, + "num_tokens": 89764734.0, + "step": 3482 + }, + { + "epoch": 0.38249505820338237, + "grad_norm": 2.39095139503479, + "learning_rate": 1e-06, + "loss": 1.0701, + "mean_token_accuracy": 0.6802922487258911, + "num_tokens": 89789455.0, + "step": 3483 + }, + { + "epoch": 0.38260487590599607, + "grad_norm": 2.088223695755005, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7054139375686646, + "num_tokens": 89817031.0, + "step": 3484 + }, + { + "epoch": 0.3827146936086097, + "grad_norm": 2.2754626274108887, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7237697243690491, + "num_tokens": 89839910.0, + "step": 3485 + }, + { + "epoch": 0.38282451131122336, + "grad_norm": 2.0998167991638184, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6978306174278259, + "num_tokens": 89868463.0, + "step": 3486 + }, + { + "epoch": 0.382934329013837, + "grad_norm": 2.045175313949585, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7155589461326599, + "num_tokens": 89896603.0, + "step": 3487 + }, + { + "epoch": 0.3830441467164507, + "grad_norm": 1.9959968328475952, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6958849430084229, + "num_tokens": 89928714.0, + "step": 3488 + }, + { + "epoch": 0.38315396441906435, + "grad_norm": 2.2256321907043457, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6987544298171997, + "num_tokens": 89955268.0, + "step": 3489 + }, + { + "epoch": 0.383263782121678, + "grad_norm": 2.1575286388397217, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7044425010681152, + "num_tokens": 89981018.0, + "step": 3490 + }, + { + "epoch": 0.3833735998242917, + "grad_norm": 2.3231019973754883, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7247465252876282, + "num_tokens": 90003163.0, + "step": 3491 + }, + { + "epoch": 0.38348341752690535, + "grad_norm": 2.0770535469055176, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7275904417037964, + "num_tokens": 90032401.0, + "step": 3492 + }, + { + "epoch": 0.383593235229519, + "grad_norm": 2.297208547592163, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7191731333732605, + "num_tokens": 90056873.0, + "step": 3493 + }, + { + "epoch": 0.38370305293213264, + "grad_norm": 2.0692567825317383, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.7072514295578003, + "num_tokens": 90086572.0, + "step": 3494 + }, + { + "epoch": 0.38381287063474634, + "grad_norm": 2.0350844860076904, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.6783719062805176, + "num_tokens": 90117268.0, + "step": 3495 + }, + { + "epoch": 0.38392268833736, + "grad_norm": 2.2961151599884033, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6922913789749146, + "num_tokens": 90142793.0, + "step": 3496 + }, + { + "epoch": 0.38403250603997363, + "grad_norm": 2.346078872680664, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7030766010284424, + "num_tokens": 90165590.0, + "step": 3497 + }, + { + "epoch": 0.38414232374258733, + "grad_norm": 2.354461908340454, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7290201187133789, + "num_tokens": 90188709.0, + "step": 3498 + }, + { + "epoch": 0.384252141445201, + "grad_norm": 2.532264471054077, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7117044925689697, + "num_tokens": 90208588.0, + "step": 3499 + }, + { + "epoch": 0.3843619591478146, + "grad_norm": 2.2252495288848877, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.702257513999939, + "num_tokens": 90234503.0, + "step": 3500 + }, + { + "epoch": 0.38447177685042827, + "grad_norm": 2.0668957233428955, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7126579284667969, + "num_tokens": 90262836.0, + "step": 3501 + }, + { + "epoch": 0.38458159455304197, + "grad_norm": 2.5142438411712646, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6796990633010864, + "num_tokens": 90286148.0, + "step": 3502 + }, + { + "epoch": 0.3846914122556556, + "grad_norm": 2.1143734455108643, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7184054851531982, + "num_tokens": 90313012.0, + "step": 3503 + }, + { + "epoch": 0.38480122995826926, + "grad_norm": 1.9634788036346436, + "learning_rate": 1e-06, + "loss": 1.0946, + "mean_token_accuracy": 0.6697521209716797, + "num_tokens": 90349412.0, + "step": 3504 + }, + { + "epoch": 0.3849110476608829, + "grad_norm": 2.2941713333129883, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7206814289093018, + "num_tokens": 90372594.0, + "step": 3505 + }, + { + "epoch": 0.3850208653634966, + "grad_norm": 2.277198076248169, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7091382741928101, + "num_tokens": 90397761.0, + "step": 3506 + }, + { + "epoch": 0.38513068306611026, + "grad_norm": 2.5332138538360596, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7184667587280273, + "num_tokens": 90419024.0, + "step": 3507 + }, + { + "epoch": 0.3852405007687239, + "grad_norm": 2.0719194412231445, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.714557409286499, + "num_tokens": 90447634.0, + "step": 3508 + }, + { + "epoch": 0.3853503184713376, + "grad_norm": 2.593311309814453, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7440460920333862, + "num_tokens": 90468148.0, + "step": 3509 + }, + { + "epoch": 0.38546013617395125, + "grad_norm": 2.5083420276641846, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7123807668685913, + "num_tokens": 90487570.0, + "step": 3510 + }, + { + "epoch": 0.3855699538765649, + "grad_norm": 2.3564891815185547, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7044785618782043, + "num_tokens": 90510133.0, + "step": 3511 + }, + { + "epoch": 0.38567977157917854, + "grad_norm": 2.3976078033447266, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.7003223896026611, + "num_tokens": 90532584.0, + "step": 3512 + }, + { + "epoch": 0.38578958928179224, + "grad_norm": 2.312866687774658, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7117140889167786, + "num_tokens": 90554683.0, + "step": 3513 + }, + { + "epoch": 0.3858994069844059, + "grad_norm": 2.1453824043273926, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6947885155677795, + "num_tokens": 90581094.0, + "step": 3514 + }, + { + "epoch": 0.38600922468701954, + "grad_norm": 2.2476906776428223, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7070976495742798, + "num_tokens": 90609205.0, + "step": 3515 + }, + { + "epoch": 0.3861190423896332, + "grad_norm": 2.3106300830841064, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7110608816146851, + "num_tokens": 90632726.0, + "step": 3516 + }, + { + "epoch": 0.3862288600922469, + "grad_norm": 2.653886318206787, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7157809734344482, + "num_tokens": 90651834.0, + "step": 3517 + }, + { + "epoch": 0.38633867779486053, + "grad_norm": 2.0694024562835693, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6956005096435547, + "num_tokens": 90681547.0, + "step": 3518 + }, + { + "epoch": 0.3864484954974742, + "grad_norm": 2.3727986812591553, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7155928611755371, + "num_tokens": 90701994.0, + "step": 3519 + }, + { + "epoch": 0.3865583132000879, + "grad_norm": 2.452845811843872, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7328863143920898, + "num_tokens": 90722796.0, + "step": 3520 + }, + { + "epoch": 0.3866681309027015, + "grad_norm": 1.896140217781067, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7203912138938904, + "num_tokens": 90756889.0, + "step": 3521 + }, + { + "epoch": 0.38677794860531517, + "grad_norm": 2.414264440536499, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6960780620574951, + "num_tokens": 90779843.0, + "step": 3522 + }, + { + "epoch": 0.3868877663079288, + "grad_norm": 2.2912120819091797, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6784058809280396, + "num_tokens": 90805997.0, + "step": 3523 + }, + { + "epoch": 0.3869975840105425, + "grad_norm": 2.0588085651397705, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7085423469543457, + "num_tokens": 90836671.0, + "step": 3524 + }, + { + "epoch": 0.38710740171315616, + "grad_norm": 2.20011568069458, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7021398544311523, + "num_tokens": 90860469.0, + "step": 3525 + }, + { + "epoch": 0.3872172194157698, + "grad_norm": 1.9404070377349854, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.700204610824585, + "num_tokens": 90891719.0, + "step": 3526 + }, + { + "epoch": 0.3873270371183835, + "grad_norm": 1.9936814308166504, + "learning_rate": 1e-06, + "loss": 1.1387, + "mean_token_accuracy": 0.6617885828018188, + "num_tokens": 90922730.0, + "step": 3527 + }, + { + "epoch": 0.38743685482099716, + "grad_norm": 2.556858539581299, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6900087594985962, + "num_tokens": 90943861.0, + "step": 3528 + }, + { + "epoch": 0.3875466725236108, + "grad_norm": 2.4835143089294434, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6863106489181519, + "num_tokens": 90966393.0, + "step": 3529 + }, + { + "epoch": 0.38765649022622445, + "grad_norm": 2.3322134017944336, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7098158597946167, + "num_tokens": 90991347.0, + "step": 3530 + }, + { + "epoch": 0.38776630792883815, + "grad_norm": 2.2304158210754395, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.722105085849762, + "num_tokens": 91017515.0, + "step": 3531 + }, + { + "epoch": 0.3878761256314518, + "grad_norm": 2.4768288135528564, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6892704367637634, + "num_tokens": 91040805.0, + "step": 3532 + }, + { + "epoch": 0.38798594333406544, + "grad_norm": 2.0094170570373535, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.6845696568489075, + "num_tokens": 91071871.0, + "step": 3533 + }, + { + "epoch": 0.3880957610366791, + "grad_norm": 2.368182897567749, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7248685956001282, + "num_tokens": 91093999.0, + "step": 3534 + }, + { + "epoch": 0.3882055787392928, + "grad_norm": 2.4478037357330322, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7274816036224365, + "num_tokens": 91114451.0, + "step": 3535 + }, + { + "epoch": 0.38831539644190644, + "grad_norm": 2.2341647148132324, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7030899524688721, + "num_tokens": 91141261.0, + "step": 3536 + }, + { + "epoch": 0.3884252141445201, + "grad_norm": 2.1136748790740967, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6996679902076721, + "num_tokens": 91167438.0, + "step": 3537 + }, + { + "epoch": 0.3885350318471338, + "grad_norm": 2.4350545406341553, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7003798484802246, + "num_tokens": 91189517.0, + "step": 3538 + }, + { + "epoch": 0.38864484954974743, + "grad_norm": 2.223536729812622, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6950496435165405, + "num_tokens": 91215306.0, + "step": 3539 + }, + { + "epoch": 0.3887546672523611, + "grad_norm": 1.8034720420837402, + "learning_rate": 1e-06, + "loss": 1.1039, + "mean_token_accuracy": 0.672716498374939, + "num_tokens": 91255049.0, + "step": 3540 + }, + { + "epoch": 0.3888644849549747, + "grad_norm": 2.1474215984344482, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7100332975387573, + "num_tokens": 91281904.0, + "step": 3541 + }, + { + "epoch": 0.3889743026575884, + "grad_norm": 1.8285590410232544, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6900660395622253, + "num_tokens": 91314741.0, + "step": 3542 + }, + { + "epoch": 0.38908412036020207, + "grad_norm": 2.458317995071411, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7218430042266846, + "num_tokens": 91334767.0, + "step": 3543 + }, + { + "epoch": 0.3891939380628157, + "grad_norm": 2.0623035430908203, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6911923885345459, + "num_tokens": 91363096.0, + "step": 3544 + }, + { + "epoch": 0.38930375576542936, + "grad_norm": 2.075347900390625, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6912049055099487, + "num_tokens": 91391608.0, + "step": 3545 + }, + { + "epoch": 0.38941357346804306, + "grad_norm": 2.233637571334839, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7130200862884521, + "num_tokens": 91416170.0, + "step": 3546 + }, + { + "epoch": 0.3895233911706567, + "grad_norm": 2.0799052715301514, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6790256500244141, + "num_tokens": 91445054.0, + "step": 3547 + }, + { + "epoch": 0.38963320887327035, + "grad_norm": 2.273266077041626, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7222111821174622, + "num_tokens": 91468406.0, + "step": 3548 + }, + { + "epoch": 0.38974302657588406, + "grad_norm": 2.1091361045837402, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6885482668876648, + "num_tokens": 91497661.0, + "step": 3549 + }, + { + "epoch": 0.3898528442784977, + "grad_norm": 2.2085773944854736, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.702684223651886, + "num_tokens": 91524292.0, + "step": 3550 + }, + { + "epoch": 0.38996266198111135, + "grad_norm": 1.9823596477508545, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.69508957862854, + "num_tokens": 91553935.0, + "step": 3551 + }, + { + "epoch": 0.390072479683725, + "grad_norm": 2.1283724308013916, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6994562149047852, + "num_tokens": 91581388.0, + "step": 3552 + }, + { + "epoch": 0.3901822973863387, + "grad_norm": 2.5649099349975586, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7216467261314392, + "num_tokens": 91600140.0, + "step": 3553 + }, + { + "epoch": 0.39029211508895234, + "grad_norm": 2.6145992279052734, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7393901944160461, + "num_tokens": 91618774.0, + "step": 3554 + }, + { + "epoch": 0.390401932791566, + "grad_norm": 2.214393377304077, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7171696424484253, + "num_tokens": 91643533.0, + "step": 3555 + }, + { + "epoch": 0.3905117504941797, + "grad_norm": 2.2386863231658936, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7147097587585449, + "num_tokens": 91667926.0, + "step": 3556 + }, + { + "epoch": 0.39062156819679333, + "grad_norm": 2.3149282932281494, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6924854516983032, + "num_tokens": 91691531.0, + "step": 3557 + }, + { + "epoch": 0.390731385899407, + "grad_norm": 2.3549835681915283, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6958053112030029, + "num_tokens": 91714515.0, + "step": 3558 + }, + { + "epoch": 0.3908412036020206, + "grad_norm": 2.0290186405181885, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6864502429962158, + "num_tokens": 91744628.0, + "step": 3559 + }, + { + "epoch": 0.3909510213046343, + "grad_norm": 2.112987995147705, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6779371500015259, + "num_tokens": 91773397.0, + "step": 3560 + }, + { + "epoch": 0.391060839007248, + "grad_norm": 2.6697933673858643, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7295253276824951, + "num_tokens": 91790945.0, + "step": 3561 + }, + { + "epoch": 0.3911706567098616, + "grad_norm": 2.294491767883301, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.704893946647644, + "num_tokens": 91813811.0, + "step": 3562 + }, + { + "epoch": 0.39128047441247527, + "grad_norm": 2.3131322860717773, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7114940285682678, + "num_tokens": 91837758.0, + "step": 3563 + }, + { + "epoch": 0.39139029211508897, + "grad_norm": 2.2828025817871094, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7224750518798828, + "num_tokens": 91861942.0, + "step": 3564 + }, + { + "epoch": 0.3915001098177026, + "grad_norm": 2.432971954345703, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7063971757888794, + "num_tokens": 91883797.0, + "step": 3565 + }, + { + "epoch": 0.39160992752031626, + "grad_norm": 2.1070823669433594, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.693063497543335, + "num_tokens": 91912339.0, + "step": 3566 + }, + { + "epoch": 0.39171974522292996, + "grad_norm": 2.029254674911499, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7140017747879028, + "num_tokens": 91942322.0, + "step": 3567 + }, + { + "epoch": 0.3918295629255436, + "grad_norm": 1.9684308767318726, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6930358409881592, + "num_tokens": 91975697.0, + "step": 3568 + }, + { + "epoch": 0.39193938062815725, + "grad_norm": 2.256009101867676, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.718817949295044, + "num_tokens": 91999458.0, + "step": 3569 + }, + { + "epoch": 0.3920491983307709, + "grad_norm": 2.3264458179473877, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6935973167419434, + "num_tokens": 92022604.0, + "step": 3570 + }, + { + "epoch": 0.3921590160333846, + "grad_norm": 2.500156879425049, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7037720680236816, + "num_tokens": 92043638.0, + "step": 3571 + }, + { + "epoch": 0.39226883373599825, + "grad_norm": 2.2424943447113037, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7020671963691711, + "num_tokens": 92068402.0, + "step": 3572 + }, + { + "epoch": 0.3923786514386119, + "grad_norm": 2.498493194580078, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7058323621749878, + "num_tokens": 92090556.0, + "step": 3573 + }, + { + "epoch": 0.3924884691412256, + "grad_norm": 2.5565714836120605, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.708480954170227, + "num_tokens": 92110032.0, + "step": 3574 + }, + { + "epoch": 0.39259828684383924, + "grad_norm": 2.302429676055908, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7029582262039185, + "num_tokens": 92134108.0, + "step": 3575 + }, + { + "epoch": 0.3927081045464529, + "grad_norm": 2.5313541889190674, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7204196453094482, + "num_tokens": 92154115.0, + "step": 3576 + }, + { + "epoch": 0.39281792224906653, + "grad_norm": 2.4954967498779297, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6985597610473633, + "num_tokens": 92176265.0, + "step": 3577 + }, + { + "epoch": 0.39292773995168023, + "grad_norm": 2.2273707389831543, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6903735399246216, + "num_tokens": 92204941.0, + "step": 3578 + }, + { + "epoch": 0.3930375576542939, + "grad_norm": 2.020402431488037, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6957278251647949, + "num_tokens": 92235351.0, + "step": 3579 + }, + { + "epoch": 0.3931473753569075, + "grad_norm": 2.105022430419922, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.696874737739563, + "num_tokens": 92264440.0, + "step": 3580 + }, + { + "epoch": 0.39325719305952117, + "grad_norm": 2.213787078857422, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.70888352394104, + "num_tokens": 92290196.0, + "step": 3581 + }, + { + "epoch": 0.39336701076213487, + "grad_norm": 2.3436758518218994, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6981474161148071, + "num_tokens": 92314005.0, + "step": 3582 + }, + { + "epoch": 0.3934768284647485, + "grad_norm": 2.170185089111328, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7031128406524658, + "num_tokens": 92338256.0, + "step": 3583 + }, + { + "epoch": 0.39358664616736216, + "grad_norm": 2.0713090896606445, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6912228465080261, + "num_tokens": 92366254.0, + "step": 3584 + }, + { + "epoch": 0.39369646386997587, + "grad_norm": 2.1908254623413086, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6817070841789246, + "num_tokens": 92393330.0, + "step": 3585 + }, + { + "epoch": 0.3938062815725895, + "grad_norm": 2.3085975646972656, + "learning_rate": 1e-06, + "loss": 1.0907, + "mean_token_accuracy": 0.6724405884742737, + "num_tokens": 92418604.0, + "step": 3586 + }, + { + "epoch": 0.39391609927520316, + "grad_norm": 2.062831401824951, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7033671736717224, + "num_tokens": 92446971.0, + "step": 3587 + }, + { + "epoch": 0.3940259169778168, + "grad_norm": 2.0265350341796875, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6936978101730347, + "num_tokens": 92478620.0, + "step": 3588 + }, + { + "epoch": 0.3941357346804305, + "grad_norm": 2.0706164836883545, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7087851762771606, + "num_tokens": 92506443.0, + "step": 3589 + }, + { + "epoch": 0.39424555238304415, + "grad_norm": 2.056004762649536, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6908996105194092, + "num_tokens": 92533281.0, + "step": 3590 + }, + { + "epoch": 0.3943553700856578, + "grad_norm": 2.222522020339966, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7035446763038635, + "num_tokens": 92558663.0, + "step": 3591 + }, + { + "epoch": 0.39446518778827144, + "grad_norm": 2.025498867034912, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7022464275360107, + "num_tokens": 92588856.0, + "step": 3592 + }, + { + "epoch": 0.39457500549088514, + "grad_norm": 2.262688398361206, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6919339299201965, + "num_tokens": 92612290.0, + "step": 3593 + }, + { + "epoch": 0.3946848231934988, + "grad_norm": 2.265648365020752, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6993135213851929, + "num_tokens": 92639085.0, + "step": 3594 + }, + { + "epoch": 0.39479464089611244, + "grad_norm": 2.0973639488220215, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6965605020523071, + "num_tokens": 92667613.0, + "step": 3595 + }, + { + "epoch": 0.39490445859872614, + "grad_norm": 2.033874988555908, + "learning_rate": 1e-06, + "loss": 1.1083, + "mean_token_accuracy": 0.6670224070549011, + "num_tokens": 92700563.0, + "step": 3596 + }, + { + "epoch": 0.3950142763013398, + "grad_norm": 2.008577346801758, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6946982145309448, + "num_tokens": 92730949.0, + "step": 3597 + }, + { + "epoch": 0.39512409400395343, + "grad_norm": 1.923811912536621, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.6867757439613342, + "num_tokens": 92761455.0, + "step": 3598 + }, + { + "epoch": 0.3952339117065671, + "grad_norm": 2.224740743637085, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7006494998931885, + "num_tokens": 92786098.0, + "step": 3599 + }, + { + "epoch": 0.3953437294091808, + "grad_norm": 2.186154365539551, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6808235049247742, + "num_tokens": 92814187.0, + "step": 3600 + }, + { + "epoch": 0.3954535471117944, + "grad_norm": 2.1992430686950684, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7095949649810791, + "num_tokens": 92841491.0, + "step": 3601 + }, + { + "epoch": 0.39556336481440807, + "grad_norm": 2.2547833919525146, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6883082985877991, + "num_tokens": 92868973.0, + "step": 3602 + }, + { + "epoch": 0.39567318251702177, + "grad_norm": 2.1322433948516846, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6961724162101746, + "num_tokens": 92896181.0, + "step": 3603 + }, + { + "epoch": 0.3957830002196354, + "grad_norm": 2.4250199794769287, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.688707709312439, + "num_tokens": 92919283.0, + "step": 3604 + }, + { + "epoch": 0.39589281792224906, + "grad_norm": 2.072535276412964, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6851917505264282, + "num_tokens": 92949794.0, + "step": 3605 + }, + { + "epoch": 0.3960026356248627, + "grad_norm": 2.268580436706543, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6943252682685852, + "num_tokens": 92973987.0, + "step": 3606 + }, + { + "epoch": 0.3961124533274764, + "grad_norm": 2.4498302936553955, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6796382069587708, + "num_tokens": 92995384.0, + "step": 3607 + }, + { + "epoch": 0.39622227103009006, + "grad_norm": 2.3586490154266357, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.698884129524231, + "num_tokens": 93017578.0, + "step": 3608 + }, + { + "epoch": 0.3963320887327037, + "grad_norm": 2.277651786804199, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7144584655761719, + "num_tokens": 93042519.0, + "step": 3609 + }, + { + "epoch": 0.39644190643531735, + "grad_norm": 2.203253984451294, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6833707690238953, + "num_tokens": 93068082.0, + "step": 3610 + }, + { + "epoch": 0.39655172413793105, + "grad_norm": 2.4464213848114014, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7019569277763367, + "num_tokens": 93088727.0, + "step": 3611 + }, + { + "epoch": 0.3966615418405447, + "grad_norm": 2.0856127738952637, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7086442708969116, + "num_tokens": 93114477.0, + "step": 3612 + }, + { + "epoch": 0.39677135954315834, + "grad_norm": 2.085840940475464, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6926625370979309, + "num_tokens": 93143202.0, + "step": 3613 + }, + { + "epoch": 0.39688117724577204, + "grad_norm": 2.171915292739868, + "learning_rate": 1e-06, + "loss": 1.067, + "mean_token_accuracy": 0.6848348379135132, + "num_tokens": 93171913.0, + "step": 3614 + }, + { + "epoch": 0.3969909949483857, + "grad_norm": 1.9613525867462158, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6815160512924194, + "num_tokens": 93204722.0, + "step": 3615 + }, + { + "epoch": 0.39710081265099934, + "grad_norm": 2.2528414726257324, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6853494048118591, + "num_tokens": 93230722.0, + "step": 3616 + }, + { + "epoch": 0.397210630353613, + "grad_norm": 2.0414087772369385, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7036832571029663, + "num_tokens": 93261736.0, + "step": 3617 + }, + { + "epoch": 0.3973204480562267, + "grad_norm": 2.203579902648926, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.6824461817741394, + "num_tokens": 93288761.0, + "step": 3618 + }, + { + "epoch": 0.39743026575884033, + "grad_norm": 2.1047494411468506, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7158023715019226, + "num_tokens": 93314383.0, + "step": 3619 + }, + { + "epoch": 0.397540083461454, + "grad_norm": 2.072575092315674, + "learning_rate": 1e-06, + "loss": 1.0971, + "mean_token_accuracy": 0.6786602735519409, + "num_tokens": 93344740.0, + "step": 3620 + }, + { + "epoch": 0.3976499011640676, + "grad_norm": 2.281885862350464, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7231155037879944, + "num_tokens": 93365177.0, + "step": 3621 + }, + { + "epoch": 0.3977597188666813, + "grad_norm": 2.395200252532959, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7106931209564209, + "num_tokens": 93386901.0, + "step": 3622 + }, + { + "epoch": 0.39786953656929497, + "grad_norm": 2.268885850906372, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7091563940048218, + "num_tokens": 93410775.0, + "step": 3623 + }, + { + "epoch": 0.3979793542719086, + "grad_norm": 2.3337950706481934, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7061178088188171, + "num_tokens": 93434290.0, + "step": 3624 + }, + { + "epoch": 0.3980891719745223, + "grad_norm": 2.024345636367798, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6979093551635742, + "num_tokens": 93465934.0, + "step": 3625 + }, + { + "epoch": 0.39819898967713596, + "grad_norm": 1.9619739055633545, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6893794536590576, + "num_tokens": 93498075.0, + "step": 3626 + }, + { + "epoch": 0.3983088073797496, + "grad_norm": 2.0910375118255615, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7158410549163818, + "num_tokens": 93524988.0, + "step": 3627 + }, + { + "epoch": 0.39841862508236325, + "grad_norm": 2.231602668762207, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7380455136299133, + "num_tokens": 93548860.0, + "step": 3628 + }, + { + "epoch": 0.39852844278497696, + "grad_norm": 2.132808208465576, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6984992027282715, + "num_tokens": 93575196.0, + "step": 3629 + }, + { + "epoch": 0.3986382604875906, + "grad_norm": 2.0704643726348877, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.712862491607666, + "num_tokens": 93604747.0, + "step": 3630 + }, + { + "epoch": 0.39874807819020425, + "grad_norm": 2.2288458347320557, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.699725329875946, + "num_tokens": 93631244.0, + "step": 3631 + }, + { + "epoch": 0.39885789589281795, + "grad_norm": 2.1541545391082764, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.696922779083252, + "num_tokens": 93657802.0, + "step": 3632 + }, + { + "epoch": 0.3989677135954316, + "grad_norm": 2.669959306716919, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6944038271903992, + "num_tokens": 93677958.0, + "step": 3633 + }, + { + "epoch": 0.39907753129804524, + "grad_norm": 2.1474251747131348, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7078119516372681, + "num_tokens": 93702307.0, + "step": 3634 + }, + { + "epoch": 0.3991873490006589, + "grad_norm": 1.9689092636108398, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.700049638748169, + "num_tokens": 93731906.0, + "step": 3635 + }, + { + "epoch": 0.3992971667032726, + "grad_norm": 2.544649600982666, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7226540446281433, + "num_tokens": 93751252.0, + "step": 3636 + }, + { + "epoch": 0.39940698440588623, + "grad_norm": 1.9504384994506836, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.691217839717865, + "num_tokens": 93785400.0, + "step": 3637 + }, + { + "epoch": 0.3995168021084999, + "grad_norm": 2.0649969577789307, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6903889775276184, + "num_tokens": 93815716.0, + "step": 3638 + }, + { + "epoch": 0.3996266198111135, + "grad_norm": 2.323850393295288, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7077435255050659, + "num_tokens": 93838163.0, + "step": 3639 + }, + { + "epoch": 0.3997364375137272, + "grad_norm": 2.0266506671905518, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6928442716598511, + "num_tokens": 93868816.0, + "step": 3640 + }, + { + "epoch": 0.3998462552163409, + "grad_norm": 2.090468406677246, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7131145000457764, + "num_tokens": 93896683.0, + "step": 3641 + }, + { + "epoch": 0.3999560729189545, + "grad_norm": 2.323532819747925, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7154254913330078, + "num_tokens": 93921366.0, + "step": 3642 + }, + { + "epoch": 0.4000658906215682, + "grad_norm": 1.9956719875335693, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7179807424545288, + "num_tokens": 93951399.0, + "step": 3643 + }, + { + "epoch": 0.40017570832418187, + "grad_norm": 2.089017152786255, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6999595165252686, + "num_tokens": 93980020.0, + "step": 3644 + }, + { + "epoch": 0.4002855260267955, + "grad_norm": 1.9142534732818604, + "learning_rate": 1e-06, + "loss": 1.1281, + "mean_token_accuracy": 0.6712498068809509, + "num_tokens": 94015028.0, + "step": 3645 + }, + { + "epoch": 0.40039534372940916, + "grad_norm": 2.3577191829681396, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7091201543807983, + "num_tokens": 94037005.0, + "step": 3646 + }, + { + "epoch": 0.40050516143202286, + "grad_norm": 2.2458670139312744, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.710857093334198, + "num_tokens": 94062116.0, + "step": 3647 + }, + { + "epoch": 0.4006149791346365, + "grad_norm": 2.098457098007202, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.686467170715332, + "num_tokens": 94090071.0, + "step": 3648 + }, + { + "epoch": 0.40072479683725015, + "grad_norm": 2.2162158489227295, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6828460693359375, + "num_tokens": 94115779.0, + "step": 3649 + }, + { + "epoch": 0.4008346145398638, + "grad_norm": 2.475914239883423, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6896597743034363, + "num_tokens": 94136772.0, + "step": 3650 + }, + { + "epoch": 0.4009444322424775, + "grad_norm": 2.1422600746154785, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.681756854057312, + "num_tokens": 94163870.0, + "step": 3651 + }, + { + "epoch": 0.40105424994509115, + "grad_norm": 2.085214614868164, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6999244689941406, + "num_tokens": 94191235.0, + "step": 3652 + }, + { + "epoch": 0.4011640676477048, + "grad_norm": 2.148806095123291, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7086527347564697, + "num_tokens": 94219294.0, + "step": 3653 + }, + { + "epoch": 0.4012738853503185, + "grad_norm": 2.149395227432251, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7052569389343262, + "num_tokens": 94249381.0, + "step": 3654 + }, + { + "epoch": 0.40138370305293214, + "grad_norm": 2.27966046333313, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6945967078208923, + "num_tokens": 94273491.0, + "step": 3655 + }, + { + "epoch": 0.4014935207555458, + "grad_norm": 2.2351551055908203, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7091304659843445, + "num_tokens": 94297360.0, + "step": 3656 + }, + { + "epoch": 0.40160333845815943, + "grad_norm": 2.1422886848449707, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7040896415710449, + "num_tokens": 94323498.0, + "step": 3657 + }, + { + "epoch": 0.40171315616077313, + "grad_norm": 2.1128900051116943, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6834456920623779, + "num_tokens": 94351862.0, + "step": 3658 + }, + { + "epoch": 0.4018229738633868, + "grad_norm": 2.287923574447632, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7152230739593506, + "num_tokens": 94377144.0, + "step": 3659 + }, + { + "epoch": 0.4019327915660004, + "grad_norm": 2.460562229156494, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.709325909614563, + "num_tokens": 94402697.0, + "step": 3660 + }, + { + "epoch": 0.4020426092686141, + "grad_norm": 1.9562424421310425, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6983140707015991, + "num_tokens": 94434007.0, + "step": 3661 + }, + { + "epoch": 0.40215242697122777, + "grad_norm": 2.31461238861084, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7170469760894775, + "num_tokens": 94457336.0, + "step": 3662 + }, + { + "epoch": 0.4022622446738414, + "grad_norm": 2.576307535171509, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7245877385139465, + "num_tokens": 94476392.0, + "step": 3663 + }, + { + "epoch": 0.40237206237645506, + "grad_norm": 2.439931631088257, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7002645134925842, + "num_tokens": 94498061.0, + "step": 3664 + }, + { + "epoch": 0.40248188007906877, + "grad_norm": 2.651576519012451, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7131013870239258, + "num_tokens": 94517466.0, + "step": 3665 + }, + { + "epoch": 0.4025916977816824, + "grad_norm": 1.981902003288269, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.693355143070221, + "num_tokens": 94553087.0, + "step": 3666 + }, + { + "epoch": 0.40270151548429606, + "grad_norm": 2.732771635055542, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6938872337341309, + "num_tokens": 94571900.0, + "step": 3667 + }, + { + "epoch": 0.4028113331869097, + "grad_norm": 2.1311967372894287, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6960791349411011, + "num_tokens": 94601582.0, + "step": 3668 + }, + { + "epoch": 0.4029211508895234, + "grad_norm": 2.1389942169189453, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6886035203933716, + "num_tokens": 94627581.0, + "step": 3669 + }, + { + "epoch": 0.40303096859213705, + "grad_norm": 1.8940188884735107, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6725642681121826, + "num_tokens": 94660605.0, + "step": 3670 + }, + { + "epoch": 0.4031407862947507, + "grad_norm": 2.247952938079834, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7220386862754822, + "num_tokens": 94683998.0, + "step": 3671 + }, + { + "epoch": 0.4032506039973644, + "grad_norm": 2.028801918029785, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.705753743648529, + "num_tokens": 94714139.0, + "step": 3672 + }, + { + "epoch": 0.40336042169997804, + "grad_norm": 2.1885972023010254, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6862788200378418, + "num_tokens": 94742674.0, + "step": 3673 + }, + { + "epoch": 0.4034702394025917, + "grad_norm": 2.0922491550445557, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.709656298160553, + "num_tokens": 94768981.0, + "step": 3674 + }, + { + "epoch": 0.40358005710520534, + "grad_norm": 2.512671709060669, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.696086049079895, + "num_tokens": 94790368.0, + "step": 3675 + }, + { + "epoch": 0.40368987480781904, + "grad_norm": 2.498903274536133, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7024884223937988, + "num_tokens": 94814055.0, + "step": 3676 + }, + { + "epoch": 0.4037996925104327, + "grad_norm": 2.311406135559082, + "learning_rate": 1e-06, + "loss": 1.1134, + "mean_token_accuracy": 0.6717822551727295, + "num_tokens": 94837500.0, + "step": 3677 + }, + { + "epoch": 0.40390951021304633, + "grad_norm": 2.5102691650390625, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7228979468345642, + "num_tokens": 94858619.0, + "step": 3678 + }, + { + "epoch": 0.40401932791566003, + "grad_norm": 1.846500039100647, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6831877827644348, + "num_tokens": 94895689.0, + "step": 3679 + }, + { + "epoch": 0.4041291456182737, + "grad_norm": 2.0543384552001953, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6905697584152222, + "num_tokens": 94928279.0, + "step": 3680 + }, + { + "epoch": 0.4042389633208873, + "grad_norm": 2.2748515605926514, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7053002715110779, + "num_tokens": 94951429.0, + "step": 3681 + }, + { + "epoch": 0.40434878102350097, + "grad_norm": 2.2388668060302734, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7064796090126038, + "num_tokens": 94975673.0, + "step": 3682 + }, + { + "epoch": 0.40445859872611467, + "grad_norm": 2.048414945602417, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.6848805546760559, + "num_tokens": 95004770.0, + "step": 3683 + }, + { + "epoch": 0.4045684164287283, + "grad_norm": 2.3150765895843506, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7070953845977783, + "num_tokens": 95027234.0, + "step": 3684 + }, + { + "epoch": 0.40467823413134196, + "grad_norm": 1.9468904733657837, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6958556771278381, + "num_tokens": 95059842.0, + "step": 3685 + }, + { + "epoch": 0.4047880518339556, + "grad_norm": 2.365382671356201, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6815507411956787, + "num_tokens": 95085434.0, + "step": 3686 + }, + { + "epoch": 0.4048978695365693, + "grad_norm": 2.3987534046173096, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6988818645477295, + "num_tokens": 95107366.0, + "step": 3687 + }, + { + "epoch": 0.40500768723918296, + "grad_norm": 2.1525673866271973, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.693881630897522, + "num_tokens": 95133724.0, + "step": 3688 + }, + { + "epoch": 0.4051175049417966, + "grad_norm": 2.1521244049072266, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7354028224945068, + "num_tokens": 95159848.0, + "step": 3689 + }, + { + "epoch": 0.4052273226444103, + "grad_norm": 2.3201889991760254, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7020054459571838, + "num_tokens": 95182829.0, + "step": 3690 + }, + { + "epoch": 0.40533714034702395, + "grad_norm": 2.2833614349365234, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7220933437347412, + "num_tokens": 95206987.0, + "step": 3691 + }, + { + "epoch": 0.4054469580496376, + "grad_norm": 2.793898344039917, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7243693470954895, + "num_tokens": 95223651.0, + "step": 3692 + }, + { + "epoch": 0.40555677575225124, + "grad_norm": 1.9869334697723389, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6909639835357666, + "num_tokens": 95254228.0, + "step": 3693 + }, + { + "epoch": 0.40566659345486494, + "grad_norm": 2.418822765350342, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7218772172927856, + "num_tokens": 95275667.0, + "step": 3694 + }, + { + "epoch": 0.4057764111574786, + "grad_norm": 2.1348116397857666, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6941350102424622, + "num_tokens": 95303074.0, + "step": 3695 + }, + { + "epoch": 0.40588622886009224, + "grad_norm": 2.2164533138275146, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7050479650497437, + "num_tokens": 95326800.0, + "step": 3696 + }, + { + "epoch": 0.4059960465627059, + "grad_norm": 2.2805283069610596, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6978318691253662, + "num_tokens": 95350883.0, + "step": 3697 + }, + { + "epoch": 0.4061058642653196, + "grad_norm": 2.401820182800293, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6955970525741577, + "num_tokens": 95375715.0, + "step": 3698 + }, + { + "epoch": 0.40621568196793323, + "grad_norm": 2.3058972358703613, + "learning_rate": 1e-06, + "loss": 1.0914, + "mean_token_accuracy": 0.6834943890571594, + "num_tokens": 95403702.0, + "step": 3699 + }, + { + "epoch": 0.4063254996705469, + "grad_norm": 2.186861753463745, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7067133188247681, + "num_tokens": 95430500.0, + "step": 3700 + }, + { + "epoch": 0.4064353173731606, + "grad_norm": 2.2665281295776367, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6810368299484253, + "num_tokens": 95457072.0, + "step": 3701 + }, + { + "epoch": 0.4065451350757742, + "grad_norm": 2.250279188156128, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7291799783706665, + "num_tokens": 95480216.0, + "step": 3702 + }, + { + "epoch": 0.40665495277838787, + "grad_norm": 2.3112423419952393, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6910673379898071, + "num_tokens": 95505666.0, + "step": 3703 + }, + { + "epoch": 0.4067647704810015, + "grad_norm": 2.040300130844116, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7060406804084778, + "num_tokens": 95535290.0, + "step": 3704 + }, + { + "epoch": 0.4068745881836152, + "grad_norm": 2.410761594772339, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6957656145095825, + "num_tokens": 95558809.0, + "step": 3705 + }, + { + "epoch": 0.40698440588622886, + "grad_norm": 2.7324936389923096, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7248514890670776, + "num_tokens": 95579981.0, + "step": 3706 + }, + { + "epoch": 0.4070942235888425, + "grad_norm": 2.2919676303863525, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7113251686096191, + "num_tokens": 95602796.0, + "step": 3707 + }, + { + "epoch": 0.4072040412914562, + "grad_norm": 1.976252794265747, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6878872513771057, + "num_tokens": 95634076.0, + "step": 3708 + }, + { + "epoch": 0.40731385899406986, + "grad_norm": 2.302680253982544, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6986361742019653, + "num_tokens": 95658486.0, + "step": 3709 + }, + { + "epoch": 0.4074236766966835, + "grad_norm": 2.3502233028411865, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7077481746673584, + "num_tokens": 95681127.0, + "step": 3710 + }, + { + "epoch": 0.40753349439929715, + "grad_norm": 2.269702911376953, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.704046368598938, + "num_tokens": 95706111.0, + "step": 3711 + }, + { + "epoch": 0.40764331210191085, + "grad_norm": 2.432826280593872, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7149766087532043, + "num_tokens": 95728140.0, + "step": 3712 + }, + { + "epoch": 0.4077531298045245, + "grad_norm": 2.358444929122925, + "learning_rate": 1e-06, + "loss": 1.0917, + "mean_token_accuracy": 0.6806764602661133, + "num_tokens": 95753019.0, + "step": 3713 + }, + { + "epoch": 0.40786294750713814, + "grad_norm": 2.5169448852539062, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7091896533966064, + "num_tokens": 95772623.0, + "step": 3714 + }, + { + "epoch": 0.4079727652097518, + "grad_norm": 2.165088415145874, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.710995078086853, + "num_tokens": 95798708.0, + "step": 3715 + }, + { + "epoch": 0.4080825829123655, + "grad_norm": 2.328094959259033, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.7016708850860596, + "num_tokens": 95820561.0, + "step": 3716 + }, + { + "epoch": 0.40819240061497913, + "grad_norm": 2.5268242359161377, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.723856508731842, + "num_tokens": 95840443.0, + "step": 3717 + }, + { + "epoch": 0.4083022183175928, + "grad_norm": 2.0276808738708496, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6970797777175903, + "num_tokens": 95868974.0, + "step": 3718 + }, + { + "epoch": 0.4084120360202065, + "grad_norm": 2.2268319129943848, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6918654441833496, + "num_tokens": 95894143.0, + "step": 3719 + }, + { + "epoch": 0.4085218537228201, + "grad_norm": 2.1882550716400146, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.681433916091919, + "num_tokens": 95919353.0, + "step": 3720 + }, + { + "epoch": 0.4086316714254338, + "grad_norm": 2.1708879470825195, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7182790637016296, + "num_tokens": 95944746.0, + "step": 3721 + }, + { + "epoch": 0.4087414891280474, + "grad_norm": 2.097015142440796, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6997156143188477, + "num_tokens": 95975169.0, + "step": 3722 + }, + { + "epoch": 0.4088513068306611, + "grad_norm": 2.2773280143737793, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.6993904113769531, + "num_tokens": 95998550.0, + "step": 3723 + }, + { + "epoch": 0.40896112453327477, + "grad_norm": 2.010684013366699, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6819201111793518, + "num_tokens": 96029407.0, + "step": 3724 + }, + { + "epoch": 0.4090709422358884, + "grad_norm": 2.1084225177764893, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7106136679649353, + "num_tokens": 96056979.0, + "step": 3725 + }, + { + "epoch": 0.40918075993850206, + "grad_norm": 2.2744691371917725, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.688758909702301, + "num_tokens": 96084000.0, + "step": 3726 + }, + { + "epoch": 0.40929057764111576, + "grad_norm": 2.380676507949829, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6833916306495667, + "num_tokens": 96109502.0, + "step": 3727 + }, + { + "epoch": 0.4094003953437294, + "grad_norm": 2.3273324966430664, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.7029403448104858, + "num_tokens": 96132475.0, + "step": 3728 + }, + { + "epoch": 0.40951021304634305, + "grad_norm": 2.050635576248169, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6909719109535217, + "num_tokens": 96160307.0, + "step": 3729 + }, + { + "epoch": 0.40962003074895675, + "grad_norm": 2.203195333480835, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6939194202423096, + "num_tokens": 96186187.0, + "step": 3730 + }, + { + "epoch": 0.4097298484515704, + "grad_norm": 2.2912843227386475, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6944655776023865, + "num_tokens": 96210572.0, + "step": 3731 + }, + { + "epoch": 0.40983966615418405, + "grad_norm": 2.260943651199341, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7033542394638062, + "num_tokens": 96236024.0, + "step": 3732 + }, + { + "epoch": 0.4099494838567977, + "grad_norm": 2.305260181427002, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7331427931785583, + "num_tokens": 96257483.0, + "step": 3733 + }, + { + "epoch": 0.4100593015594114, + "grad_norm": 2.559107542037964, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7325120568275452, + "num_tokens": 96277344.0, + "step": 3734 + }, + { + "epoch": 0.41016911926202504, + "grad_norm": 2.479282855987549, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7213001251220703, + "num_tokens": 96298625.0, + "step": 3735 + }, + { + "epoch": 0.4102789369646387, + "grad_norm": 2.305968999862671, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6971741914749146, + "num_tokens": 96321881.0, + "step": 3736 + }, + { + "epoch": 0.4103887546672524, + "grad_norm": 2.3537235260009766, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7180188894271851, + "num_tokens": 96342862.0, + "step": 3737 + }, + { + "epoch": 0.41049857236986603, + "grad_norm": 2.1510114669799805, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7304633259773254, + "num_tokens": 96368341.0, + "step": 3738 + }, + { + "epoch": 0.4106083900724797, + "grad_norm": 2.08274245262146, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6847860813140869, + "num_tokens": 96396788.0, + "step": 3739 + }, + { + "epoch": 0.4107182077750933, + "grad_norm": 2.189014434814453, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7062823176383972, + "num_tokens": 96421984.0, + "step": 3740 + }, + { + "epoch": 0.410828025477707, + "grad_norm": 2.4298672676086426, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6927186250686646, + "num_tokens": 96443177.0, + "step": 3741 + }, + { + "epoch": 0.41093784318032067, + "grad_norm": 2.1677234172821045, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7036478519439697, + "num_tokens": 96468900.0, + "step": 3742 + }, + { + "epoch": 0.4110476608829343, + "grad_norm": 2.096188545227051, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6873142719268799, + "num_tokens": 96499680.0, + "step": 3743 + }, + { + "epoch": 0.41115747858554796, + "grad_norm": 2.189467430114746, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6876137256622314, + "num_tokens": 96527875.0, + "step": 3744 + }, + { + "epoch": 0.41126729628816167, + "grad_norm": 2.3856329917907715, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7005611658096313, + "num_tokens": 96549900.0, + "step": 3745 + }, + { + "epoch": 0.4113771139907753, + "grad_norm": 2.085115432739258, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7094656229019165, + "num_tokens": 96575724.0, + "step": 3746 + }, + { + "epoch": 0.41148693169338896, + "grad_norm": 2.0671262741088867, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6982487440109253, + "num_tokens": 96604379.0, + "step": 3747 + }, + { + "epoch": 0.41159674939600266, + "grad_norm": 2.0636913776397705, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6918987035751343, + "num_tokens": 96632373.0, + "step": 3748 + }, + { + "epoch": 0.4117065670986163, + "grad_norm": 1.9918780326843262, + "learning_rate": 1e-06, + "loss": 1.0825, + "mean_token_accuracy": 0.6733529567718506, + "num_tokens": 96663582.0, + "step": 3749 + }, + { + "epoch": 0.41181638480122995, + "grad_norm": 2.3444008827209473, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6999684572219849, + "num_tokens": 96687273.0, + "step": 3750 + }, + { + "epoch": 0.4119262025038436, + "grad_norm": 2.086422920227051, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6848565340042114, + "num_tokens": 96716391.0, + "step": 3751 + }, + { + "epoch": 0.4120360202064573, + "grad_norm": 1.835047960281372, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.710579514503479, + "num_tokens": 96750957.0, + "step": 3752 + }, + { + "epoch": 0.41214583790907094, + "grad_norm": 2.177560567855835, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7035000324249268, + "num_tokens": 96776920.0, + "step": 3753 + }, + { + "epoch": 0.4122556556116846, + "grad_norm": 2.2129125595092773, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.7039579153060913, + "num_tokens": 96802204.0, + "step": 3754 + }, + { + "epoch": 0.4123654733142983, + "grad_norm": 2.216710329055786, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7091341018676758, + "num_tokens": 96828116.0, + "step": 3755 + }, + { + "epoch": 0.41247529101691194, + "grad_norm": 2.0284597873687744, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7277485132217407, + "num_tokens": 96854187.0, + "step": 3756 + }, + { + "epoch": 0.4125851087195256, + "grad_norm": 2.4270529747009277, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7066558599472046, + "num_tokens": 96877609.0, + "step": 3757 + }, + { + "epoch": 0.41269492642213923, + "grad_norm": 2.2581560611724854, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6861717700958252, + "num_tokens": 96906243.0, + "step": 3758 + }, + { + "epoch": 0.41280474412475293, + "grad_norm": 2.007234811782837, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6866648197174072, + "num_tokens": 96937800.0, + "step": 3759 + }, + { + "epoch": 0.4129145618273666, + "grad_norm": 2.249591588973999, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7046306133270264, + "num_tokens": 96963616.0, + "step": 3760 + }, + { + "epoch": 0.4130243795299802, + "grad_norm": 2.323166847229004, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7069370746612549, + "num_tokens": 96988221.0, + "step": 3761 + }, + { + "epoch": 0.41313419723259387, + "grad_norm": 2.1956467628479004, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6970781683921814, + "num_tokens": 97015712.0, + "step": 3762 + }, + { + "epoch": 0.41324401493520757, + "grad_norm": 2.089508056640625, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.6812767386436462, + "num_tokens": 97045974.0, + "step": 3763 + }, + { + "epoch": 0.4133538326378212, + "grad_norm": 2.1674444675445557, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.700330376625061, + "num_tokens": 97074150.0, + "step": 3764 + }, + { + "epoch": 0.41346365034043486, + "grad_norm": 2.204441785812378, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.710477352142334, + "num_tokens": 97099405.0, + "step": 3765 + }, + { + "epoch": 0.41357346804304856, + "grad_norm": 2.286754846572876, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6861709356307983, + "num_tokens": 97123704.0, + "step": 3766 + }, + { + "epoch": 0.4136832857456622, + "grad_norm": 2.1306586265563965, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6923477649688721, + "num_tokens": 97151453.0, + "step": 3767 + }, + { + "epoch": 0.41379310344827586, + "grad_norm": 2.3651888370513916, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7073168158531189, + "num_tokens": 97177275.0, + "step": 3768 + }, + { + "epoch": 0.4139029211508895, + "grad_norm": 2.2577528953552246, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6991813778877258, + "num_tokens": 97201539.0, + "step": 3769 + }, + { + "epoch": 0.4140127388535032, + "grad_norm": 2.3199589252471924, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6913929581642151, + "num_tokens": 97226967.0, + "step": 3770 + }, + { + "epoch": 0.41412255655611685, + "grad_norm": 2.086977005004883, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6973364353179932, + "num_tokens": 97254157.0, + "step": 3771 + }, + { + "epoch": 0.4142323742587305, + "grad_norm": 2.105048418045044, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7209550142288208, + "num_tokens": 97282796.0, + "step": 3772 + }, + { + "epoch": 0.41434219196134414, + "grad_norm": 2.4167020320892334, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6946096420288086, + "num_tokens": 97304315.0, + "step": 3773 + }, + { + "epoch": 0.41445200966395784, + "grad_norm": 2.112548589706421, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7055752277374268, + "num_tokens": 97333699.0, + "step": 3774 + }, + { + "epoch": 0.4145618273665715, + "grad_norm": 2.3904333114624023, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7252909541130066, + "num_tokens": 97357171.0, + "step": 3775 + }, + { + "epoch": 0.41467164506918514, + "grad_norm": 2.2296197414398193, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6926453113555908, + "num_tokens": 97385450.0, + "step": 3776 + }, + { + "epoch": 0.41478146277179884, + "grad_norm": 2.227976083755493, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7126826047897339, + "num_tokens": 97409370.0, + "step": 3777 + }, + { + "epoch": 0.4148912804744125, + "grad_norm": 2.0510170459747314, + "learning_rate": 1e-06, + "loss": 1.1019, + "mean_token_accuracy": 0.6698103547096252, + "num_tokens": 97439978.0, + "step": 3778 + }, + { + "epoch": 0.41500109817702613, + "grad_norm": 2.154755115509033, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6991179585456848, + "num_tokens": 97466968.0, + "step": 3779 + }, + { + "epoch": 0.4151109158796398, + "grad_norm": 2.639526128768921, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6995089054107666, + "num_tokens": 97487150.0, + "step": 3780 + }, + { + "epoch": 0.4152207335822535, + "grad_norm": 2.6531317234039307, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.70981764793396, + "num_tokens": 97506056.0, + "step": 3781 + }, + { + "epoch": 0.4153305512848671, + "grad_norm": 2.089266777038574, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6909292936325073, + "num_tokens": 97538059.0, + "step": 3782 + }, + { + "epoch": 0.41544036898748077, + "grad_norm": 2.3083178997039795, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7135065197944641, + "num_tokens": 97560457.0, + "step": 3783 + }, + { + "epoch": 0.41555018669009447, + "grad_norm": 2.2001805305480957, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6881805658340454, + "num_tokens": 97585237.0, + "step": 3784 + }, + { + "epoch": 0.4156600043927081, + "grad_norm": 2.489190101623535, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6964497566223145, + "num_tokens": 97607155.0, + "step": 3785 + }, + { + "epoch": 0.41576982209532176, + "grad_norm": 2.1229312419891357, + "learning_rate": 1e-06, + "loss": 1.0718, + "mean_token_accuracy": 0.6868051886558533, + "num_tokens": 97632634.0, + "step": 3786 + }, + { + "epoch": 0.4158796397979354, + "grad_norm": 1.9817558526992798, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7246693968772888, + "num_tokens": 97660021.0, + "step": 3787 + }, + { + "epoch": 0.4159894575005491, + "grad_norm": 2.0910046100616455, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7006400227546692, + "num_tokens": 97687998.0, + "step": 3788 + }, + { + "epoch": 0.41609927520316276, + "grad_norm": 2.4055545330047607, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6871086955070496, + "num_tokens": 97712028.0, + "step": 3789 + }, + { + "epoch": 0.4162090929057764, + "grad_norm": 2.3890645503997803, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7244142293930054, + "num_tokens": 97732785.0, + "step": 3790 + }, + { + "epoch": 0.41631891060839005, + "grad_norm": 2.291133403778076, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7159087061882019, + "num_tokens": 97753692.0, + "step": 3791 + }, + { + "epoch": 0.41642872831100375, + "grad_norm": 2.025095224380493, + "learning_rate": 1e-06, + "loss": 1.0824, + "mean_token_accuracy": 0.6737291812896729, + "num_tokens": 97784355.0, + "step": 3792 + }, + { + "epoch": 0.4165385460136174, + "grad_norm": 1.930920958518982, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7095961570739746, + "num_tokens": 97815111.0, + "step": 3793 + }, + { + "epoch": 0.41664836371623104, + "grad_norm": 1.9383007287979126, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7101887464523315, + "num_tokens": 97847890.0, + "step": 3794 + }, + { + "epoch": 0.41675818141884474, + "grad_norm": 2.126680612564087, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6872240900993347, + "num_tokens": 97875099.0, + "step": 3795 + }, + { + "epoch": 0.4168679991214584, + "grad_norm": 2.0392489433288574, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7015495896339417, + "num_tokens": 97903616.0, + "step": 3796 + }, + { + "epoch": 0.41697781682407203, + "grad_norm": 2.199708938598633, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6938201189041138, + "num_tokens": 97930112.0, + "step": 3797 + }, + { + "epoch": 0.4170876345266857, + "grad_norm": 2.4005286693573, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7085343599319458, + "num_tokens": 97951334.0, + "step": 3798 + }, + { + "epoch": 0.4171974522292994, + "grad_norm": 2.6777536869049072, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7199786901473999, + "num_tokens": 97969072.0, + "step": 3799 + }, + { + "epoch": 0.417307269931913, + "grad_norm": 2.1523654460906982, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7143179178237915, + "num_tokens": 97998273.0, + "step": 3800 + }, + { + "epoch": 0.4174170876345267, + "grad_norm": 2.4113311767578125, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.6968643665313721, + "num_tokens": 98020408.0, + "step": 3801 + }, + { + "epoch": 0.4175269053371403, + "grad_norm": 2.088040828704834, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7051538228988647, + "num_tokens": 98048085.0, + "step": 3802 + }, + { + "epoch": 0.417636723039754, + "grad_norm": 2.3434808254241943, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7139257192611694, + "num_tokens": 98071839.0, + "step": 3803 + }, + { + "epoch": 0.41774654074236767, + "grad_norm": 2.2494256496429443, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7476288676261902, + "num_tokens": 98096379.0, + "step": 3804 + }, + { + "epoch": 0.4178563584449813, + "grad_norm": 2.1408448219299316, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.7048678398132324, + "num_tokens": 98123245.0, + "step": 3805 + }, + { + "epoch": 0.417966176147595, + "grad_norm": 2.3941612243652344, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7060965299606323, + "num_tokens": 98145861.0, + "step": 3806 + }, + { + "epoch": 0.41807599385020866, + "grad_norm": 1.9911452531814575, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6986023783683777, + "num_tokens": 98176395.0, + "step": 3807 + }, + { + "epoch": 0.4181858115528223, + "grad_norm": 2.19619083404541, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7321831583976746, + "num_tokens": 98200400.0, + "step": 3808 + }, + { + "epoch": 0.41829562925543595, + "grad_norm": 2.4200844764709473, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6942472457885742, + "num_tokens": 98222184.0, + "step": 3809 + }, + { + "epoch": 0.41840544695804965, + "grad_norm": 2.2967753410339355, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6892181038856506, + "num_tokens": 98246918.0, + "step": 3810 + }, + { + "epoch": 0.4185152646606633, + "grad_norm": 1.8707553148269653, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7269638180732727, + "num_tokens": 98282039.0, + "step": 3811 + }, + { + "epoch": 0.41862508236327695, + "grad_norm": 2.2422995567321777, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7016685009002686, + "num_tokens": 98306789.0, + "step": 3812 + }, + { + "epoch": 0.41873490006589065, + "grad_norm": 2.1712608337402344, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.68343186378479, + "num_tokens": 98333984.0, + "step": 3813 + }, + { + "epoch": 0.4188447177685043, + "grad_norm": 2.4045636653900146, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6951820254325867, + "num_tokens": 98356729.0, + "step": 3814 + }, + { + "epoch": 0.41895453547111794, + "grad_norm": 1.9004408121109009, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6890066266059875, + "num_tokens": 98389776.0, + "step": 3815 + }, + { + "epoch": 0.4190643531737316, + "grad_norm": 2.23901104927063, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6975427865982056, + "num_tokens": 98413833.0, + "step": 3816 + }, + { + "epoch": 0.4191741708763453, + "grad_norm": 2.19351863861084, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7144006490707397, + "num_tokens": 98439400.0, + "step": 3817 + }, + { + "epoch": 0.41928398857895893, + "grad_norm": 2.1310067176818848, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7231035232543945, + "num_tokens": 98466611.0, + "step": 3818 + }, + { + "epoch": 0.4193938062815726, + "grad_norm": 2.5216879844665527, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6921557188034058, + "num_tokens": 98488397.0, + "step": 3819 + }, + { + "epoch": 0.4195036239841862, + "grad_norm": 2.3487675189971924, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.704734206199646, + "num_tokens": 98513790.0, + "step": 3820 + }, + { + "epoch": 0.4196134416867999, + "grad_norm": 2.2248315811157227, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7232701182365417, + "num_tokens": 98538191.0, + "step": 3821 + }, + { + "epoch": 0.41972325938941357, + "grad_norm": 2.428109884262085, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7270423173904419, + "num_tokens": 98558551.0, + "step": 3822 + }, + { + "epoch": 0.4198330770920272, + "grad_norm": 2.3038220405578613, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.707068681716919, + "num_tokens": 98582157.0, + "step": 3823 + }, + { + "epoch": 0.4199428947946409, + "grad_norm": 2.24648118019104, + "learning_rate": 1e-06, + "loss": 1.0935, + "mean_token_accuracy": 0.6736630201339722, + "num_tokens": 98607916.0, + "step": 3824 + }, + { + "epoch": 0.42005271249725457, + "grad_norm": 2.2493033409118652, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7287964820861816, + "num_tokens": 98632435.0, + "step": 3825 + }, + { + "epoch": 0.4201625301998682, + "grad_norm": 2.3735766410827637, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.730766773223877, + "num_tokens": 98654168.0, + "step": 3826 + }, + { + "epoch": 0.42027234790248186, + "grad_norm": 2.286802053451538, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6936193704605103, + "num_tokens": 98680486.0, + "step": 3827 + }, + { + "epoch": 0.42038216560509556, + "grad_norm": 2.549217700958252, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.705274224281311, + "num_tokens": 98700896.0, + "step": 3828 + }, + { + "epoch": 0.4204919833077092, + "grad_norm": 2.2020747661590576, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6967278122901917, + "num_tokens": 98728052.0, + "step": 3829 + }, + { + "epoch": 0.42060180101032285, + "grad_norm": 2.292991876602173, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6823498606681824, + "num_tokens": 98751858.0, + "step": 3830 + }, + { + "epoch": 0.42071161871293655, + "grad_norm": 2.4315996170043945, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.712775707244873, + "num_tokens": 98774127.0, + "step": 3831 + }, + { + "epoch": 0.4208214364155502, + "grad_norm": 2.123298406600952, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6981872916221619, + "num_tokens": 98799279.0, + "step": 3832 + }, + { + "epoch": 0.42093125411816384, + "grad_norm": 2.2000081539154053, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7082743644714355, + "num_tokens": 98824948.0, + "step": 3833 + }, + { + "epoch": 0.4210410718207775, + "grad_norm": 2.3741953372955322, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6902211904525757, + "num_tokens": 98848812.0, + "step": 3834 + }, + { + "epoch": 0.4211508895233912, + "grad_norm": 2.113943338394165, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7019091844558716, + "num_tokens": 98877205.0, + "step": 3835 + }, + { + "epoch": 0.42126070722600484, + "grad_norm": 2.161886692047119, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7146503925323486, + "num_tokens": 98903539.0, + "step": 3836 + }, + { + "epoch": 0.4213705249286185, + "grad_norm": 2.2067348957061768, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6933751106262207, + "num_tokens": 98930935.0, + "step": 3837 + }, + { + "epoch": 0.42148034263123213, + "grad_norm": 2.1228106021881104, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7158695459365845, + "num_tokens": 98958786.0, + "step": 3838 + }, + { + "epoch": 0.42159016033384583, + "grad_norm": 1.936156988143921, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7097827196121216, + "num_tokens": 98988130.0, + "step": 3839 + }, + { + "epoch": 0.4216999780364595, + "grad_norm": 2.3460540771484375, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7020829916000366, + "num_tokens": 99012448.0, + "step": 3840 + }, + { + "epoch": 0.4218097957390731, + "grad_norm": 2.2897799015045166, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7077991962432861, + "num_tokens": 99034898.0, + "step": 3841 + }, + { + "epoch": 0.4219196134416868, + "grad_norm": 1.921103596687317, + "learning_rate": 1e-06, + "loss": 1.0745, + "mean_token_accuracy": 0.6826128959655762, + "num_tokens": 99069960.0, + "step": 3842 + }, + { + "epoch": 0.42202943114430047, + "grad_norm": 2.0123825073242188, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.720139741897583, + "num_tokens": 99099128.0, + "step": 3843 + }, + { + "epoch": 0.4221392488469141, + "grad_norm": 2.1943228244781494, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6998914480209351, + "num_tokens": 99123862.0, + "step": 3844 + }, + { + "epoch": 0.42224906654952776, + "grad_norm": 2.0415401458740234, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6824027299880981, + "num_tokens": 99153367.0, + "step": 3845 + }, + { + "epoch": 0.42235888425214146, + "grad_norm": 2.178081512451172, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7005663514137268, + "num_tokens": 99180589.0, + "step": 3846 + }, + { + "epoch": 0.4224687019547551, + "grad_norm": 2.150456666946411, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7050089836120605, + "num_tokens": 99205457.0, + "step": 3847 + }, + { + "epoch": 0.42257851965736876, + "grad_norm": 2.1531786918640137, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7083832025527954, + "num_tokens": 99232568.0, + "step": 3848 + }, + { + "epoch": 0.4226883373599824, + "grad_norm": 2.189265251159668, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6974977254867554, + "num_tokens": 99259609.0, + "step": 3849 + }, + { + "epoch": 0.4227981550625961, + "grad_norm": 2.090578556060791, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6993422508239746, + "num_tokens": 99287325.0, + "step": 3850 + }, + { + "epoch": 0.42290797276520975, + "grad_norm": 2.3301875591278076, + "learning_rate": 1e-06, + "loss": 1.0839, + "mean_token_accuracy": 0.6858333349227905, + "num_tokens": 99312208.0, + "step": 3851 + }, + { + "epoch": 0.4230177904678234, + "grad_norm": 2.2813425064086914, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7242469787597656, + "num_tokens": 99334419.0, + "step": 3852 + }, + { + "epoch": 0.4231276081704371, + "grad_norm": 2.246154546737671, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6894041299819946, + "num_tokens": 99359494.0, + "step": 3853 + }, + { + "epoch": 0.42323742587305074, + "grad_norm": 2.2310714721679688, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.705877423286438, + "num_tokens": 99384979.0, + "step": 3854 + }, + { + "epoch": 0.4233472435756644, + "grad_norm": 2.1434292793273926, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7239435911178589, + "num_tokens": 99411064.0, + "step": 3855 + }, + { + "epoch": 0.42345706127827804, + "grad_norm": 2.3685615062713623, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7057443857192993, + "num_tokens": 99434251.0, + "step": 3856 + }, + { + "epoch": 0.42356687898089174, + "grad_norm": 2.2478439807891846, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.707994818687439, + "num_tokens": 99457956.0, + "step": 3857 + }, + { + "epoch": 0.4236766966835054, + "grad_norm": 2.236900806427002, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6935232877731323, + "num_tokens": 99484020.0, + "step": 3858 + }, + { + "epoch": 0.42378651438611903, + "grad_norm": 2.2515909671783447, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7189033031463623, + "num_tokens": 99508778.0, + "step": 3859 + }, + { + "epoch": 0.42389633208873273, + "grad_norm": 1.94222092628479, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6945381760597229, + "num_tokens": 99539835.0, + "step": 3860 + }, + { + "epoch": 0.4240061497913464, + "grad_norm": 2.093703031539917, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7042181491851807, + "num_tokens": 99565805.0, + "step": 3861 + }, + { + "epoch": 0.42411596749396, + "grad_norm": 2.0708060264587402, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.705461323261261, + "num_tokens": 99595428.0, + "step": 3862 + }, + { + "epoch": 0.42422578519657367, + "grad_norm": 2.6695563793182373, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7130168676376343, + "num_tokens": 99614400.0, + "step": 3863 + }, + { + "epoch": 0.42433560289918737, + "grad_norm": 2.016296625137329, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6783995628356934, + "num_tokens": 99645184.0, + "step": 3864 + }, + { + "epoch": 0.424445420601801, + "grad_norm": 2.3317854404449463, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7042024731636047, + "num_tokens": 99668161.0, + "step": 3865 + }, + { + "epoch": 0.42455523830441466, + "grad_norm": 2.101422071456909, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6922869682312012, + "num_tokens": 99698079.0, + "step": 3866 + }, + { + "epoch": 0.4246650560070283, + "grad_norm": 2.055980920791626, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6935935020446777, + "num_tokens": 99727555.0, + "step": 3867 + }, + { + "epoch": 0.424774873709642, + "grad_norm": 2.0535709857940674, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7261120080947876, + "num_tokens": 99755401.0, + "step": 3868 + }, + { + "epoch": 0.42488469141225566, + "grad_norm": 2.369138479232788, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7276978492736816, + "num_tokens": 99776256.0, + "step": 3869 + }, + { + "epoch": 0.4249945091148693, + "grad_norm": 2.099210262298584, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6962441205978394, + "num_tokens": 99803870.0, + "step": 3870 + }, + { + "epoch": 0.425104326817483, + "grad_norm": 2.4234509468078613, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7138044834136963, + "num_tokens": 99827403.0, + "step": 3871 + }, + { + "epoch": 0.42521414452009665, + "grad_norm": 1.973258376121521, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6929692029953003, + "num_tokens": 99857457.0, + "step": 3872 + }, + { + "epoch": 0.4253239622227103, + "grad_norm": 2.186399459838867, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7056302428245544, + "num_tokens": 99883362.0, + "step": 3873 + }, + { + "epoch": 0.42543377992532394, + "grad_norm": 2.462373733520508, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7040956020355225, + "num_tokens": 99904590.0, + "step": 3874 + }, + { + "epoch": 0.42554359762793764, + "grad_norm": 2.1015613079071045, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6921278834342957, + "num_tokens": 99934440.0, + "step": 3875 + }, + { + "epoch": 0.4256534153305513, + "grad_norm": 1.9414002895355225, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.694092333316803, + "num_tokens": 99969459.0, + "step": 3876 + }, + { + "epoch": 0.42576323303316493, + "grad_norm": 2.154709815979004, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6992195844650269, + "num_tokens": 99994582.0, + "step": 3877 + }, + { + "epoch": 0.4258730507357786, + "grad_norm": 2.4561188220977783, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7200766205787659, + "num_tokens": 100014999.0, + "step": 3878 + }, + { + "epoch": 0.4259828684383923, + "grad_norm": 2.309431314468384, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.689243733882904, + "num_tokens": 100039785.0, + "step": 3879 + }, + { + "epoch": 0.4260926861410059, + "grad_norm": 2.2855942249298096, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6821173429489136, + "num_tokens": 100065680.0, + "step": 3880 + }, + { + "epoch": 0.4262025038436196, + "grad_norm": 2.0710248947143555, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7134287357330322, + "num_tokens": 100092924.0, + "step": 3881 + }, + { + "epoch": 0.4263123215462333, + "grad_norm": 2.0954949855804443, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6961627006530762, + "num_tokens": 100119304.0, + "step": 3882 + }, + { + "epoch": 0.4264221392488469, + "grad_norm": 2.1744797229766846, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6918225288391113, + "num_tokens": 100147293.0, + "step": 3883 + }, + { + "epoch": 0.42653195695146057, + "grad_norm": 2.401395320892334, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7252547740936279, + "num_tokens": 100169251.0, + "step": 3884 + }, + { + "epoch": 0.4266417746540742, + "grad_norm": 2.2675862312316895, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6917674541473389, + "num_tokens": 100196836.0, + "step": 3885 + }, + { + "epoch": 0.4267515923566879, + "grad_norm": 2.035898208618164, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7158626317977905, + "num_tokens": 100223312.0, + "step": 3886 + }, + { + "epoch": 0.42686141005930156, + "grad_norm": 2.1434497833251953, + "learning_rate": 1e-06, + "loss": 1.12, + "mean_token_accuracy": 0.6688846349716187, + "num_tokens": 100251709.0, + "step": 3887 + }, + { + "epoch": 0.4269712277619152, + "grad_norm": 2.2430834770202637, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6862309575080872, + "num_tokens": 100277977.0, + "step": 3888 + }, + { + "epoch": 0.4270810454645289, + "grad_norm": 2.3168954849243164, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6997394561767578, + "num_tokens": 100302111.0, + "step": 3889 + }, + { + "epoch": 0.42719086316714255, + "grad_norm": 2.2363646030426025, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6913464069366455, + "num_tokens": 100327982.0, + "step": 3890 + }, + { + "epoch": 0.4273006808697562, + "grad_norm": 2.3924455642700195, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.706877589225769, + "num_tokens": 100349849.0, + "step": 3891 + }, + { + "epoch": 0.42741049857236985, + "grad_norm": 2.446200370788574, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7298334836959839, + "num_tokens": 100370780.0, + "step": 3892 + }, + { + "epoch": 0.42752031627498355, + "grad_norm": 2.4213383197784424, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7121611833572388, + "num_tokens": 100394033.0, + "step": 3893 + }, + { + "epoch": 0.4276301339775972, + "grad_norm": 1.9508814811706543, + "learning_rate": 1e-06, + "loss": 1.0967, + "mean_token_accuracy": 0.6773126721382141, + "num_tokens": 100427849.0, + "step": 3894 + }, + { + "epoch": 0.42773995168021084, + "grad_norm": 2.009345769882202, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6976860761642456, + "num_tokens": 100460057.0, + "step": 3895 + }, + { + "epoch": 0.4278497693828245, + "grad_norm": 2.4759552478790283, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7133632898330688, + "num_tokens": 100480221.0, + "step": 3896 + }, + { + "epoch": 0.4279595870854382, + "grad_norm": 2.5011346340179443, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7027918100357056, + "num_tokens": 100500905.0, + "step": 3897 + }, + { + "epoch": 0.42806940478805183, + "grad_norm": 1.8960298299789429, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6837822198867798, + "num_tokens": 100535759.0, + "step": 3898 + }, + { + "epoch": 0.4281792224906655, + "grad_norm": 2.359464406967163, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7070997953414917, + "num_tokens": 100558756.0, + "step": 3899 + }, + { + "epoch": 0.4282890401932792, + "grad_norm": 2.049722671508789, + "learning_rate": 1e-06, + "loss": 1.1147, + "mean_token_accuracy": 0.672266960144043, + "num_tokens": 100591428.0, + "step": 3900 + }, + { + "epoch": 0.4283988578958928, + "grad_norm": 2.5537056922912598, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7145633697509766, + "num_tokens": 100611135.0, + "step": 3901 + }, + { + "epoch": 0.42850867559850647, + "grad_norm": 1.8726063966751099, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.713585376739502, + "num_tokens": 100644859.0, + "step": 3902 + }, + { + "epoch": 0.4286184933011201, + "grad_norm": 2.287261486053467, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.697931706905365, + "num_tokens": 100669046.0, + "step": 3903 + }, + { + "epoch": 0.4287283110037338, + "grad_norm": 2.124331474304199, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7215827107429504, + "num_tokens": 100695450.0, + "step": 3904 + }, + { + "epoch": 0.42883812870634747, + "grad_norm": 2.210496187210083, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6824694871902466, + "num_tokens": 100722704.0, + "step": 3905 + }, + { + "epoch": 0.4289479464089611, + "grad_norm": 1.9982444047927856, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6874740719795227, + "num_tokens": 100753240.0, + "step": 3906 + }, + { + "epoch": 0.4290577641115748, + "grad_norm": 1.9862719774246216, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6996020078659058, + "num_tokens": 100785087.0, + "step": 3907 + }, + { + "epoch": 0.42916758181418846, + "grad_norm": 2.426476240158081, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7088544368743896, + "num_tokens": 100807584.0, + "step": 3908 + }, + { + "epoch": 0.4292773995168021, + "grad_norm": 2.140801191329956, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6821885704994202, + "num_tokens": 100835174.0, + "step": 3909 + }, + { + "epoch": 0.42938721721941575, + "grad_norm": 2.2401716709136963, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7313674688339233, + "num_tokens": 100858189.0, + "step": 3910 + }, + { + "epoch": 0.42949703492202945, + "grad_norm": 2.452341318130493, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7223081588745117, + "num_tokens": 100880251.0, + "step": 3911 + }, + { + "epoch": 0.4296068526246431, + "grad_norm": 2.37312650680542, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.713475227355957, + "num_tokens": 100903346.0, + "step": 3912 + }, + { + "epoch": 0.42971667032725674, + "grad_norm": 2.0328640937805176, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.693501353263855, + "num_tokens": 100930321.0, + "step": 3913 + }, + { + "epoch": 0.4298264880298704, + "grad_norm": 2.505378007888794, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7114500999450684, + "num_tokens": 100951502.0, + "step": 3914 + }, + { + "epoch": 0.4299363057324841, + "grad_norm": 1.9625416994094849, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6931915283203125, + "num_tokens": 100984492.0, + "step": 3915 + }, + { + "epoch": 0.43004612343509774, + "grad_norm": 2.2972254753112793, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7203962802886963, + "num_tokens": 101008470.0, + "step": 3916 + }, + { + "epoch": 0.4301559411377114, + "grad_norm": 1.9897651672363281, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.712074875831604, + "num_tokens": 101038368.0, + "step": 3917 + }, + { + "epoch": 0.4302657588403251, + "grad_norm": 2.0003974437713623, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6968361139297485, + "num_tokens": 101069930.0, + "step": 3918 + }, + { + "epoch": 0.43037557654293873, + "grad_norm": 2.6243672370910645, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7011290192604065, + "num_tokens": 101089456.0, + "step": 3919 + }, + { + "epoch": 0.4304853942455524, + "grad_norm": 2.31689715385437, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6997342109680176, + "num_tokens": 101114559.0, + "step": 3920 + }, + { + "epoch": 0.430595211948166, + "grad_norm": 2.415555238723755, + "learning_rate": 1e-06, + "loss": 1.0613, + "mean_token_accuracy": 0.6769847869873047, + "num_tokens": 101139308.0, + "step": 3921 + }, + { + "epoch": 0.4307050296507797, + "grad_norm": 2.1901392936706543, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7006949186325073, + "num_tokens": 101166776.0, + "step": 3922 + }, + { + "epoch": 0.43081484735339337, + "grad_norm": 2.0158941745758057, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.694524347782135, + "num_tokens": 101196521.0, + "step": 3923 + }, + { + "epoch": 0.430924665056007, + "grad_norm": 2.0992794036865234, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6979236602783203, + "num_tokens": 101223832.0, + "step": 3924 + }, + { + "epoch": 0.43103448275862066, + "grad_norm": 2.244405508041382, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7153371572494507, + "num_tokens": 101248087.0, + "step": 3925 + }, + { + "epoch": 0.43114430046123436, + "grad_norm": 2.0127005577087402, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7068058252334595, + "num_tokens": 101279175.0, + "step": 3926 + }, + { + "epoch": 0.431254118163848, + "grad_norm": 2.267249345779419, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6970616579055786, + "num_tokens": 101304772.0, + "step": 3927 + }, + { + "epoch": 0.43136393586646166, + "grad_norm": 2.517832040786743, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6926898956298828, + "num_tokens": 101327487.0, + "step": 3928 + }, + { + "epoch": 0.43147375356907536, + "grad_norm": 2.3095200061798096, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6875874996185303, + "num_tokens": 101351295.0, + "step": 3929 + }, + { + "epoch": 0.431583571271689, + "grad_norm": 2.055248260498047, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6993274092674255, + "num_tokens": 101382037.0, + "step": 3930 + }, + { + "epoch": 0.43169338897430265, + "grad_norm": 2.626981496810913, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7179175615310669, + "num_tokens": 101401013.0, + "step": 3931 + }, + { + "epoch": 0.4318032066769163, + "grad_norm": 2.031968593597412, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6772790551185608, + "num_tokens": 101432531.0, + "step": 3932 + }, + { + "epoch": 0.43191302437953, + "grad_norm": 2.5397326946258545, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7031253576278687, + "num_tokens": 101452195.0, + "step": 3933 + }, + { + "epoch": 0.43202284208214364, + "grad_norm": 2.309326171875, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.7015434503555298, + "num_tokens": 101477943.0, + "step": 3934 + }, + { + "epoch": 0.4321326597847573, + "grad_norm": 2.078875780105591, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7236558198928833, + "num_tokens": 101504610.0, + "step": 3935 + }, + { + "epoch": 0.432242477487371, + "grad_norm": 2.4482977390289307, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7049988508224487, + "num_tokens": 101524412.0, + "step": 3936 + }, + { + "epoch": 0.43235229518998464, + "grad_norm": 2.454918384552002, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6959810256958008, + "num_tokens": 101546064.0, + "step": 3937 + }, + { + "epoch": 0.4324621128925983, + "grad_norm": 2.2415525913238525, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7106702327728271, + "num_tokens": 101570077.0, + "step": 3938 + }, + { + "epoch": 0.43257193059521193, + "grad_norm": 2.397528886795044, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6973540186882019, + "num_tokens": 101592230.0, + "step": 3939 + }, + { + "epoch": 0.43268174829782563, + "grad_norm": 2.0148987770080566, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7271245121955872, + "num_tokens": 101620098.0, + "step": 3940 + }, + { + "epoch": 0.4327915660004393, + "grad_norm": 2.4623000621795654, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6958388686180115, + "num_tokens": 101641933.0, + "step": 3941 + }, + { + "epoch": 0.4329013837030529, + "grad_norm": 2.509535074234009, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.708031415939331, + "num_tokens": 101661020.0, + "step": 3942 + }, + { + "epoch": 0.43301120140566657, + "grad_norm": 2.351806402206421, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.69652259349823, + "num_tokens": 101684501.0, + "step": 3943 + }, + { + "epoch": 0.43312101910828027, + "grad_norm": 2.3002395629882812, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7091661095619202, + "num_tokens": 101707516.0, + "step": 3944 + }, + { + "epoch": 0.4332308368108939, + "grad_norm": 2.4238109588623047, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.694936990737915, + "num_tokens": 101731267.0, + "step": 3945 + }, + { + "epoch": 0.43334065451350756, + "grad_norm": 2.5667052268981934, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7171695232391357, + "num_tokens": 101750152.0, + "step": 3946 + }, + { + "epoch": 0.43345047221612126, + "grad_norm": 2.1823205947875977, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6994726657867432, + "num_tokens": 101775807.0, + "step": 3947 + }, + { + "epoch": 0.4335602899187349, + "grad_norm": 2.2297399044036865, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7143545746803284, + "num_tokens": 101801654.0, + "step": 3948 + }, + { + "epoch": 0.43367010762134856, + "grad_norm": 2.4236879348754883, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7110111117362976, + "num_tokens": 101824066.0, + "step": 3949 + }, + { + "epoch": 0.4337799253239622, + "grad_norm": 2.305166721343994, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6906107664108276, + "num_tokens": 101847958.0, + "step": 3950 + }, + { + "epoch": 0.4338897430265759, + "grad_norm": 2.3554134368896484, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7209051847457886, + "num_tokens": 101869432.0, + "step": 3951 + }, + { + "epoch": 0.43399956072918955, + "grad_norm": 2.0916552543640137, + "learning_rate": 1e-06, + "loss": 1.1319, + "mean_token_accuracy": 0.6611286401748657, + "num_tokens": 101901182.0, + "step": 3952 + }, + { + "epoch": 0.4341093784318032, + "grad_norm": 2.729830026626587, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7164559364318848, + "num_tokens": 101919128.0, + "step": 3953 + }, + { + "epoch": 0.43421919613441684, + "grad_norm": 2.1913440227508545, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6934000253677368, + "num_tokens": 101945398.0, + "step": 3954 + }, + { + "epoch": 0.43432901383703054, + "grad_norm": 2.1441562175750732, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7232751250267029, + "num_tokens": 101969771.0, + "step": 3955 + }, + { + "epoch": 0.4344388315396442, + "grad_norm": 2.5518102645874023, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7132818698883057, + "num_tokens": 101988835.0, + "step": 3956 + }, + { + "epoch": 0.43454864924225783, + "grad_norm": 2.2138595581054688, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.683440625667572, + "num_tokens": 102017061.0, + "step": 3957 + }, + { + "epoch": 0.43465846694487154, + "grad_norm": 2.3522863388061523, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.686548113822937, + "num_tokens": 102040434.0, + "step": 3958 + }, + { + "epoch": 0.4347682846474852, + "grad_norm": 2.058521032333374, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7082728147506714, + "num_tokens": 102068857.0, + "step": 3959 + }, + { + "epoch": 0.4348781023500988, + "grad_norm": 2.194244146347046, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7023507952690125, + "num_tokens": 102096157.0, + "step": 3960 + }, + { + "epoch": 0.4349879200527125, + "grad_norm": 2.262279987335205, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6853809356689453, + "num_tokens": 102120502.0, + "step": 3961 + }, + { + "epoch": 0.4350977377553262, + "grad_norm": 2.5854146480560303, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.703387439250946, + "num_tokens": 102140402.0, + "step": 3962 + }, + { + "epoch": 0.4352075554579398, + "grad_norm": 2.502462387084961, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7308918237686157, + "num_tokens": 102160242.0, + "step": 3963 + }, + { + "epoch": 0.43531737316055347, + "grad_norm": 2.210268020629883, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6949862837791443, + "num_tokens": 102185639.0, + "step": 3964 + }, + { + "epoch": 0.43542719086316717, + "grad_norm": 2.203132152557373, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7036483287811279, + "num_tokens": 102211456.0, + "step": 3965 + }, + { + "epoch": 0.4355370085657808, + "grad_norm": 2.1629016399383545, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6896231174468994, + "num_tokens": 102239365.0, + "step": 3966 + }, + { + "epoch": 0.43564682626839446, + "grad_norm": 2.1001198291778564, + "learning_rate": 1e-06, + "loss": 1.076, + "mean_token_accuracy": 0.6782528758049011, + "num_tokens": 102268941.0, + "step": 3967 + }, + { + "epoch": 0.4357566439710081, + "grad_norm": 2.155463695526123, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7156844139099121, + "num_tokens": 102295733.0, + "step": 3968 + }, + { + "epoch": 0.4358664616736218, + "grad_norm": 2.1407809257507324, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6907006502151489, + "num_tokens": 102323561.0, + "step": 3969 + }, + { + "epoch": 0.43597627937623545, + "grad_norm": 2.2842769622802734, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6879207491874695, + "num_tokens": 102349209.0, + "step": 3970 + }, + { + "epoch": 0.4360860970788491, + "grad_norm": 2.634209156036377, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7221450805664062, + "num_tokens": 102367932.0, + "step": 3971 + }, + { + "epoch": 0.43619591478146275, + "grad_norm": 2.00407075881958, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7054832577705383, + "num_tokens": 102396181.0, + "step": 3972 + }, + { + "epoch": 0.43630573248407645, + "grad_norm": 2.034763813018799, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.7013403177261353, + "num_tokens": 102427524.0, + "step": 3973 + }, + { + "epoch": 0.4364155501866901, + "grad_norm": 2.0961203575134277, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7049033641815186, + "num_tokens": 102454203.0, + "step": 3974 + }, + { + "epoch": 0.43652536788930374, + "grad_norm": 2.187136650085449, + "learning_rate": 1e-06, + "loss": 1.1334, + "mean_token_accuracy": 0.6695563793182373, + "num_tokens": 102479389.0, + "step": 3975 + }, + { + "epoch": 0.43663518559191744, + "grad_norm": 2.3821048736572266, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7133364081382751, + "num_tokens": 102501427.0, + "step": 3976 + }, + { + "epoch": 0.4367450032945311, + "grad_norm": 2.473822832107544, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7217829823493958, + "num_tokens": 102523424.0, + "step": 3977 + }, + { + "epoch": 0.43685482099714473, + "grad_norm": 2.077465057373047, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6881488561630249, + "num_tokens": 102553104.0, + "step": 3978 + }, + { + "epoch": 0.4369646386997584, + "grad_norm": 2.386678695678711, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7070518732070923, + "num_tokens": 102575304.0, + "step": 3979 + }, + { + "epoch": 0.4370744564023721, + "grad_norm": 2.3107750415802, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7276159524917603, + "num_tokens": 102596716.0, + "step": 3980 + }, + { + "epoch": 0.4371842741049857, + "grad_norm": 2.1751725673675537, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7075878381729126, + "num_tokens": 102621696.0, + "step": 3981 + }, + { + "epoch": 0.43729409180759937, + "grad_norm": 2.1486291885375977, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7112572193145752, + "num_tokens": 102649454.0, + "step": 3982 + }, + { + "epoch": 0.4374039095102131, + "grad_norm": 2.276219129562378, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7014399170875549, + "num_tokens": 102673200.0, + "step": 3983 + }, + { + "epoch": 0.4375137272128267, + "grad_norm": 2.1561615467071533, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7074922323226929, + "num_tokens": 102698792.0, + "step": 3984 + }, + { + "epoch": 0.43762354491544037, + "grad_norm": 2.5068624019622803, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7045494318008423, + "num_tokens": 102721060.0, + "step": 3985 + }, + { + "epoch": 0.437733362618054, + "grad_norm": 2.1649563312530518, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6887323260307312, + "num_tokens": 102747676.0, + "step": 3986 + }, + { + "epoch": 0.4378431803206677, + "grad_norm": 2.2503793239593506, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6892375946044922, + "num_tokens": 102772408.0, + "step": 3987 + }, + { + "epoch": 0.43795299802328136, + "grad_norm": 2.465752363204956, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6988627910614014, + "num_tokens": 102794596.0, + "step": 3988 + }, + { + "epoch": 0.438062815725895, + "grad_norm": 2.295968770980835, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7289494872093201, + "num_tokens": 102820048.0, + "step": 3989 + }, + { + "epoch": 0.43817263342850865, + "grad_norm": 2.1987743377685547, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6902608275413513, + "num_tokens": 102846102.0, + "step": 3990 + }, + { + "epoch": 0.43828245113112235, + "grad_norm": 2.2096827030181885, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7183813452720642, + "num_tokens": 102871582.0, + "step": 3991 + }, + { + "epoch": 0.438392268833736, + "grad_norm": 2.144704580307007, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7071065902709961, + "num_tokens": 102898491.0, + "step": 3992 + }, + { + "epoch": 0.43850208653634964, + "grad_norm": 2.2352333068847656, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6948204636573792, + "num_tokens": 102924356.0, + "step": 3993 + }, + { + "epoch": 0.43861190423896335, + "grad_norm": 1.8121229410171509, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7062707543373108, + "num_tokens": 102959823.0, + "step": 3994 + }, + { + "epoch": 0.438721721941577, + "grad_norm": 2.590367317199707, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.704123318195343, + "num_tokens": 102980638.0, + "step": 3995 + }, + { + "epoch": 0.43883153964419064, + "grad_norm": 2.3183605670928955, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7153067588806152, + "num_tokens": 103001969.0, + "step": 3996 + }, + { + "epoch": 0.4389413573468043, + "grad_norm": 2.4787437915802, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7089990973472595, + "num_tokens": 103021811.0, + "step": 3997 + }, + { + "epoch": 0.439051175049418, + "grad_norm": 2.3777716159820557, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6814332604408264, + "num_tokens": 103045801.0, + "step": 3998 + }, + { + "epoch": 0.43916099275203163, + "grad_norm": 2.45758318901062, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7059420943260193, + "num_tokens": 103067124.0, + "step": 3999 + }, + { + "epoch": 0.4392708104546453, + "grad_norm": 2.226613998413086, + "learning_rate": 1e-06, + "loss": 1.1341, + "mean_token_accuracy": 0.6723923683166504, + "num_tokens": 103095906.0, + "step": 4000 + }, + { + "epoch": 0.4393806281572589, + "grad_norm": 2.236572027206421, + "learning_rate": 1e-06, + "loss": 1.0862, + "mean_token_accuracy": 0.6741431951522827, + "num_tokens": 103122734.0, + "step": 4001 + }, + { + "epoch": 0.4394904458598726, + "grad_norm": 2.1579511165618896, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.681565523147583, + "num_tokens": 103148732.0, + "step": 4002 + }, + { + "epoch": 0.43960026356248627, + "grad_norm": 2.1798601150512695, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6856507658958435, + "num_tokens": 103177685.0, + "step": 4003 + }, + { + "epoch": 0.4397100812650999, + "grad_norm": 2.435617208480835, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7035492658615112, + "num_tokens": 103199045.0, + "step": 4004 + }, + { + "epoch": 0.4398198989677136, + "grad_norm": 2.1654956340789795, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6984583139419556, + "num_tokens": 103225235.0, + "step": 4005 + }, + { + "epoch": 0.43992971667032726, + "grad_norm": 2.266253709793091, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7048602104187012, + "num_tokens": 103250170.0, + "step": 4006 + }, + { + "epoch": 0.4400395343729409, + "grad_norm": 2.01116681098938, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7112764120101929, + "num_tokens": 103279224.0, + "step": 4007 + }, + { + "epoch": 0.44014935207555456, + "grad_norm": 2.4444961547851562, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6821430921554565, + "num_tokens": 103303096.0, + "step": 4008 + }, + { + "epoch": 0.44025916977816826, + "grad_norm": 2.1966733932495117, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7013514041900635, + "num_tokens": 103329306.0, + "step": 4009 + }, + { + "epoch": 0.4403689874807819, + "grad_norm": 2.0104689598083496, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6960698366165161, + "num_tokens": 103360379.0, + "step": 4010 + }, + { + "epoch": 0.44047880518339555, + "grad_norm": 2.240110158920288, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6946606636047363, + "num_tokens": 103388929.0, + "step": 4011 + }, + { + "epoch": 0.44058862288600925, + "grad_norm": 2.125898838043213, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6998140811920166, + "num_tokens": 103413922.0, + "step": 4012 + }, + { + "epoch": 0.4406984405886229, + "grad_norm": 2.5019097328186035, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7031397223472595, + "num_tokens": 103433681.0, + "step": 4013 + }, + { + "epoch": 0.44080825829123654, + "grad_norm": 2.1884169578552246, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6966654658317566, + "num_tokens": 103458452.0, + "step": 4014 + }, + { + "epoch": 0.4409180759938502, + "grad_norm": 2.040328025817871, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6956809759140015, + "num_tokens": 103488543.0, + "step": 4015 + }, + { + "epoch": 0.4410278936964639, + "grad_norm": 2.2855446338653564, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7215262651443481, + "num_tokens": 103511916.0, + "step": 4016 + }, + { + "epoch": 0.44113771139907754, + "grad_norm": 2.3554041385650635, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7474905848503113, + "num_tokens": 103533386.0, + "step": 4017 + }, + { + "epoch": 0.4412475291016912, + "grad_norm": 1.9584654569625854, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6953094601631165, + "num_tokens": 103565168.0, + "step": 4018 + }, + { + "epoch": 0.44135734680430483, + "grad_norm": 2.248427629470825, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6967657804489136, + "num_tokens": 103591130.0, + "step": 4019 + }, + { + "epoch": 0.44146716450691853, + "grad_norm": 2.185474395751953, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7468052506446838, + "num_tokens": 103617147.0, + "step": 4020 + }, + { + "epoch": 0.4415769822095322, + "grad_norm": 2.6694114208221436, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7208819389343262, + "num_tokens": 103633890.0, + "step": 4021 + }, + { + "epoch": 0.4416867999121458, + "grad_norm": 2.1522719860076904, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.7008845806121826, + "num_tokens": 103659442.0, + "step": 4022 + }, + { + "epoch": 0.4417966176147595, + "grad_norm": 2.100344657897949, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6787564158439636, + "num_tokens": 103688558.0, + "step": 4023 + }, + { + "epoch": 0.44190643531737317, + "grad_norm": 2.491478443145752, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.6974220871925354, + "num_tokens": 103709587.0, + "step": 4024 + }, + { + "epoch": 0.4420162530199868, + "grad_norm": 2.2398979663848877, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6951889395713806, + "num_tokens": 103736068.0, + "step": 4025 + }, + { + "epoch": 0.44212607072260046, + "grad_norm": 2.337944984436035, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6806049942970276, + "num_tokens": 103761126.0, + "step": 4026 + }, + { + "epoch": 0.44223588842521416, + "grad_norm": 2.5854461193084717, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7079062461853027, + "num_tokens": 103780873.0, + "step": 4027 + }, + { + "epoch": 0.4423457061278278, + "grad_norm": 2.3293628692626953, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7127640843391418, + "num_tokens": 103802979.0, + "step": 4028 + }, + { + "epoch": 0.44245552383044146, + "grad_norm": 2.29925274848938, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7103176712989807, + "num_tokens": 103826469.0, + "step": 4029 + }, + { + "epoch": 0.4425653415330551, + "grad_norm": 1.950658917427063, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7162846326828003, + "num_tokens": 103858088.0, + "step": 4030 + }, + { + "epoch": 0.4426751592356688, + "grad_norm": 2.089967966079712, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6910527944564819, + "num_tokens": 103888096.0, + "step": 4031 + }, + { + "epoch": 0.44278497693828245, + "grad_norm": 2.0461480617523193, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6976901292800903, + "num_tokens": 103917431.0, + "step": 4032 + }, + { + "epoch": 0.4428947946408961, + "grad_norm": 2.6176843643188477, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.693088948726654, + "num_tokens": 103935888.0, + "step": 4033 + }, + { + "epoch": 0.4430046123435098, + "grad_norm": 2.305410146713257, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.705216109752655, + "num_tokens": 103959705.0, + "step": 4034 + }, + { + "epoch": 0.44311443004612344, + "grad_norm": 2.2093615531921387, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7002490758895874, + "num_tokens": 103985613.0, + "step": 4035 + }, + { + "epoch": 0.4432242477487371, + "grad_norm": 2.241274356842041, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7090717554092407, + "num_tokens": 104010224.0, + "step": 4036 + }, + { + "epoch": 0.44333406545135073, + "grad_norm": 2.163595199584961, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6952266693115234, + "num_tokens": 104037578.0, + "step": 4037 + }, + { + "epoch": 0.44344388315396444, + "grad_norm": 2.336324453353882, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6944606304168701, + "num_tokens": 104062462.0, + "step": 4038 + }, + { + "epoch": 0.4435537008565781, + "grad_norm": 2.368384599685669, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6891087293624878, + "num_tokens": 104085096.0, + "step": 4039 + }, + { + "epoch": 0.4436635185591917, + "grad_norm": 1.7715847492218018, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7033698558807373, + "num_tokens": 104120717.0, + "step": 4040 + }, + { + "epoch": 0.44377333626180543, + "grad_norm": 2.3064470291137695, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.71274733543396, + "num_tokens": 104146000.0, + "step": 4041 + }, + { + "epoch": 0.4438831539644191, + "grad_norm": 2.3358328342437744, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6995629072189331, + "num_tokens": 104169632.0, + "step": 4042 + }, + { + "epoch": 0.4439929716670327, + "grad_norm": 2.163653612136841, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6913748979568481, + "num_tokens": 104195879.0, + "step": 4043 + }, + { + "epoch": 0.44410278936964637, + "grad_norm": 2.3277132511138916, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7294621467590332, + "num_tokens": 104216413.0, + "step": 4044 + }, + { + "epoch": 0.44421260707226007, + "grad_norm": 2.3586254119873047, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7103733420372009, + "num_tokens": 104239928.0, + "step": 4045 + }, + { + "epoch": 0.4443224247748737, + "grad_norm": 2.400064706802368, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6964399814605713, + "num_tokens": 104263046.0, + "step": 4046 + }, + { + "epoch": 0.44443224247748736, + "grad_norm": 2.2916676998138428, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6920983791351318, + "num_tokens": 104290272.0, + "step": 4047 + }, + { + "epoch": 0.444542060180101, + "grad_norm": 2.2982683181762695, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.704878032207489, + "num_tokens": 104312096.0, + "step": 4048 + }, + { + "epoch": 0.4446518778827147, + "grad_norm": 2.049548625946045, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.7053244113922119, + "num_tokens": 104341847.0, + "step": 4049 + }, + { + "epoch": 0.44476169558532835, + "grad_norm": 2.1606733798980713, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6954174041748047, + "num_tokens": 104368684.0, + "step": 4050 + }, + { + "epoch": 0.444871513287942, + "grad_norm": 2.288365364074707, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6906735301017761, + "num_tokens": 104392961.0, + "step": 4051 + }, + { + "epoch": 0.4449813309905557, + "grad_norm": 2.2557125091552734, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7037372589111328, + "num_tokens": 104417287.0, + "step": 4052 + }, + { + "epoch": 0.44509114869316935, + "grad_norm": 2.077279806137085, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6935606598854065, + "num_tokens": 104446072.0, + "step": 4053 + }, + { + "epoch": 0.445200966395783, + "grad_norm": 2.154433250427246, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7156445384025574, + "num_tokens": 104475273.0, + "step": 4054 + }, + { + "epoch": 0.44531078409839664, + "grad_norm": 2.2415356636047363, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7089473009109497, + "num_tokens": 104500923.0, + "step": 4055 + }, + { + "epoch": 0.44542060180101034, + "grad_norm": 2.120523452758789, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7150527834892273, + "num_tokens": 104528476.0, + "step": 4056 + }, + { + "epoch": 0.445530419503624, + "grad_norm": 2.2639501094818115, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6875886917114258, + "num_tokens": 104552884.0, + "step": 4057 + }, + { + "epoch": 0.44564023720623763, + "grad_norm": 2.363133430480957, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7069764137268066, + "num_tokens": 104576592.0, + "step": 4058 + }, + { + "epoch": 0.44575005490885133, + "grad_norm": 2.531795024871826, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6832262277603149, + "num_tokens": 104599134.0, + "step": 4059 + }, + { + "epoch": 0.445859872611465, + "grad_norm": 2.176614999771118, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6952536106109619, + "num_tokens": 104627292.0, + "step": 4060 + }, + { + "epoch": 0.4459696903140786, + "grad_norm": 2.276093006134033, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7024700045585632, + "num_tokens": 104651700.0, + "step": 4061 + }, + { + "epoch": 0.44607950801669227, + "grad_norm": 2.2074105739593506, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7089095115661621, + "num_tokens": 104677686.0, + "step": 4062 + }, + { + "epoch": 0.446189325719306, + "grad_norm": 2.356339454650879, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7077916264533997, + "num_tokens": 104701611.0, + "step": 4063 + }, + { + "epoch": 0.4462991434219196, + "grad_norm": 2.3880293369293213, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6935986876487732, + "num_tokens": 104725866.0, + "step": 4064 + }, + { + "epoch": 0.44640896112453327, + "grad_norm": 2.0216715335845947, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7049856185913086, + "num_tokens": 104755239.0, + "step": 4065 + }, + { + "epoch": 0.4465187788271469, + "grad_norm": 2.065981149673462, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6814898252487183, + "num_tokens": 104783434.0, + "step": 4066 + }, + { + "epoch": 0.4466285965297606, + "grad_norm": 2.0575785636901855, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6836967468261719, + "num_tokens": 104812157.0, + "step": 4067 + }, + { + "epoch": 0.44673841423237426, + "grad_norm": 2.3675851821899414, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7045560479164124, + "num_tokens": 104834733.0, + "step": 4068 + }, + { + "epoch": 0.4468482319349879, + "grad_norm": 2.6319658756256104, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7366341352462769, + "num_tokens": 104854222.0, + "step": 4069 + }, + { + "epoch": 0.4469580496376016, + "grad_norm": 2.1606268882751465, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7017316818237305, + "num_tokens": 104883117.0, + "step": 4070 + }, + { + "epoch": 0.44706786734021525, + "grad_norm": 2.3592545986175537, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6797601580619812, + "num_tokens": 104907980.0, + "step": 4071 + }, + { + "epoch": 0.4471776850428289, + "grad_norm": 2.1735658645629883, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7033393383026123, + "num_tokens": 104933174.0, + "step": 4072 + }, + { + "epoch": 0.44728750274544254, + "grad_norm": 2.33205509185791, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7039355039596558, + "num_tokens": 104956816.0, + "step": 4073 + }, + { + "epoch": 0.44739732044805625, + "grad_norm": 2.4704477787017822, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6985977292060852, + "num_tokens": 104980214.0, + "step": 4074 + }, + { + "epoch": 0.4475071381506699, + "grad_norm": 2.3383235931396484, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6785318851470947, + "num_tokens": 105006013.0, + "step": 4075 + }, + { + "epoch": 0.44761695585328354, + "grad_norm": 2.359323263168335, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7171502113342285, + "num_tokens": 105029313.0, + "step": 4076 + }, + { + "epoch": 0.4477267735558972, + "grad_norm": 2.5909290313720703, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7173675298690796, + "num_tokens": 105049892.0, + "step": 4077 + }, + { + "epoch": 0.4478365912585109, + "grad_norm": 2.284942865371704, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7062168121337891, + "num_tokens": 105074852.0, + "step": 4078 + }, + { + "epoch": 0.44794640896112453, + "grad_norm": 2.116973638534546, + "learning_rate": 1e-06, + "loss": 1.0827, + "mean_token_accuracy": 0.685116171836853, + "num_tokens": 105105425.0, + "step": 4079 + }, + { + "epoch": 0.4480562266637382, + "grad_norm": 2.3460652828216553, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7024600505828857, + "num_tokens": 105129982.0, + "step": 4080 + }, + { + "epoch": 0.4481660443663519, + "grad_norm": 2.476222276687622, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7120670676231384, + "num_tokens": 105150612.0, + "step": 4081 + }, + { + "epoch": 0.4482758620689655, + "grad_norm": 2.234431028366089, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6892697811126709, + "num_tokens": 105175307.0, + "step": 4082 + }, + { + "epoch": 0.44838567977157917, + "grad_norm": 2.233494520187378, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7060312032699585, + "num_tokens": 105201136.0, + "step": 4083 + }, + { + "epoch": 0.4484954974741928, + "grad_norm": 2.199099540710449, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7094019651412964, + "num_tokens": 105226406.0, + "step": 4084 + }, + { + "epoch": 0.4486053151768065, + "grad_norm": 2.1371920108795166, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6965113878250122, + "num_tokens": 105254003.0, + "step": 4085 + }, + { + "epoch": 0.44871513287942016, + "grad_norm": 2.194525718688965, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.696175217628479, + "num_tokens": 105279881.0, + "step": 4086 + }, + { + "epoch": 0.4488249505820338, + "grad_norm": 2.2883543968200684, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7159578204154968, + "num_tokens": 105304784.0, + "step": 4087 + }, + { + "epoch": 0.4489347682846475, + "grad_norm": 2.231387138366699, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7042215466499329, + "num_tokens": 105327660.0, + "step": 4088 + }, + { + "epoch": 0.44904458598726116, + "grad_norm": 2.504107713699341, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7214727401733398, + "num_tokens": 105347722.0, + "step": 4089 + }, + { + "epoch": 0.4491544036898748, + "grad_norm": 2.5043601989746094, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7022510170936584, + "num_tokens": 105369400.0, + "step": 4090 + }, + { + "epoch": 0.44926422139248845, + "grad_norm": 2.142242431640625, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.684053897857666, + "num_tokens": 105398011.0, + "step": 4091 + }, + { + "epoch": 0.44937403909510215, + "grad_norm": 1.8305505514144897, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6986954808235168, + "num_tokens": 105435526.0, + "step": 4092 + }, + { + "epoch": 0.4494838567977158, + "grad_norm": 2.1599080562591553, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7170549631118774, + "num_tokens": 105463225.0, + "step": 4093 + }, + { + "epoch": 0.44959367450032944, + "grad_norm": 2.2182421684265137, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.708103597164154, + "num_tokens": 105488403.0, + "step": 4094 + }, + { + "epoch": 0.4497034922029431, + "grad_norm": 2.032581329345703, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7091784477233887, + "num_tokens": 105518093.0, + "step": 4095 + }, + { + "epoch": 0.4498133099055568, + "grad_norm": 2.1633031368255615, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7336046695709229, + "num_tokens": 105542585.0, + "step": 4096 + }, + { + "epoch": 0.44992312760817044, + "grad_norm": 2.28145432472229, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7073395252227783, + "num_tokens": 105568633.0, + "step": 4097 + }, + { + "epoch": 0.4500329453107841, + "grad_norm": 2.167710304260254, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.721548318862915, + "num_tokens": 105592348.0, + "step": 4098 + }, + { + "epoch": 0.4501427630133978, + "grad_norm": 1.9794760942459106, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.698079526424408, + "num_tokens": 105622413.0, + "step": 4099 + }, + { + "epoch": 0.45025258071601143, + "grad_norm": 2.2879059314727783, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7059265971183777, + "num_tokens": 105649271.0, + "step": 4100 + }, + { + "epoch": 0.4503623984186251, + "grad_norm": 2.125384569168091, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7116650342941284, + "num_tokens": 105677615.0, + "step": 4101 + }, + { + "epoch": 0.4504722161212387, + "grad_norm": 1.9511921405792236, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7242233753204346, + "num_tokens": 105704887.0, + "step": 4102 + }, + { + "epoch": 0.4505820338238524, + "grad_norm": 2.0996816158294678, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6929250955581665, + "num_tokens": 105733202.0, + "step": 4103 + }, + { + "epoch": 0.45069185152646607, + "grad_norm": 2.10919189453125, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6897116899490356, + "num_tokens": 105760790.0, + "step": 4104 + }, + { + "epoch": 0.4508016692290797, + "grad_norm": 2.4006810188293457, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6898504495620728, + "num_tokens": 105783607.0, + "step": 4105 + }, + { + "epoch": 0.45091148693169336, + "grad_norm": 2.3064746856689453, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7130051851272583, + "num_tokens": 105805748.0, + "step": 4106 + }, + { + "epoch": 0.45102130463430706, + "grad_norm": 2.0631096363067627, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6993361115455627, + "num_tokens": 105834055.0, + "step": 4107 + }, + { + "epoch": 0.4511311223369207, + "grad_norm": 2.4147045612335205, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7224863171577454, + "num_tokens": 105857189.0, + "step": 4108 + }, + { + "epoch": 0.45124094003953436, + "grad_norm": 2.225369691848755, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7325127124786377, + "num_tokens": 105881831.0, + "step": 4109 + }, + { + "epoch": 0.45135075774214806, + "grad_norm": 2.2951974868774414, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7063440084457397, + "num_tokens": 105904485.0, + "step": 4110 + }, + { + "epoch": 0.4514605754447617, + "grad_norm": 2.229407787322998, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6959867477416992, + "num_tokens": 105931306.0, + "step": 4111 + }, + { + "epoch": 0.45157039314737535, + "grad_norm": 2.068662166595459, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.7009317874908447, + "num_tokens": 105960032.0, + "step": 4112 + }, + { + "epoch": 0.451680210849989, + "grad_norm": 2.4787158966064453, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6992842555046082, + "num_tokens": 105982251.0, + "step": 4113 + }, + { + "epoch": 0.4517900285526027, + "grad_norm": 1.9903404712677002, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6870442628860474, + "num_tokens": 106015119.0, + "step": 4114 + }, + { + "epoch": 0.45189984625521634, + "grad_norm": 1.9326410293579102, + "learning_rate": 1e-06, + "loss": 1.084, + "mean_token_accuracy": 0.6758811473846436, + "num_tokens": 106051086.0, + "step": 4115 + }, + { + "epoch": 0.45200966395783, + "grad_norm": 2.341020345687866, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6981279850006104, + "num_tokens": 106074355.0, + "step": 4116 + }, + { + "epoch": 0.4521194816604437, + "grad_norm": 2.046088695526123, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6929703950881958, + "num_tokens": 106104091.0, + "step": 4117 + }, + { + "epoch": 0.45222929936305734, + "grad_norm": 2.226449489593506, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7062338590621948, + "num_tokens": 106127974.0, + "step": 4118 + }, + { + "epoch": 0.452339117065671, + "grad_norm": 2.0432159900665283, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.681510329246521, + "num_tokens": 106158851.0, + "step": 4119 + }, + { + "epoch": 0.4524489347682846, + "grad_norm": 2.064844846725464, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6906546950340271, + "num_tokens": 106187368.0, + "step": 4120 + }, + { + "epoch": 0.45255875247089833, + "grad_norm": 2.0575389862060547, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7237322926521301, + "num_tokens": 106213437.0, + "step": 4121 + }, + { + "epoch": 0.452668570173512, + "grad_norm": 2.066899061203003, + "learning_rate": 1e-06, + "loss": 1.159, + "mean_token_accuracy": 0.649700403213501, + "num_tokens": 106246498.0, + "step": 4122 + }, + { + "epoch": 0.4527783878761256, + "grad_norm": 2.096750497817993, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7081307172775269, + "num_tokens": 106274260.0, + "step": 4123 + }, + { + "epoch": 0.45288820557873927, + "grad_norm": 2.472749948501587, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7167807817459106, + "num_tokens": 106293940.0, + "step": 4124 + }, + { + "epoch": 0.45299802328135297, + "grad_norm": 2.0752320289611816, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7048044204711914, + "num_tokens": 106321862.0, + "step": 4125 + }, + { + "epoch": 0.4531078409839666, + "grad_norm": 2.3486979007720947, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7040383815765381, + "num_tokens": 106346130.0, + "step": 4126 + }, + { + "epoch": 0.45321765868658026, + "grad_norm": 2.4160189628601074, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.706723690032959, + "num_tokens": 106368426.0, + "step": 4127 + }, + { + "epoch": 0.45332747638919396, + "grad_norm": 2.239037275314331, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7052095532417297, + "num_tokens": 106393287.0, + "step": 4128 + }, + { + "epoch": 0.4534372940918076, + "grad_norm": 2.2089455127716064, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6878496408462524, + "num_tokens": 106418751.0, + "step": 4129 + }, + { + "epoch": 0.45354711179442125, + "grad_norm": 2.369497060775757, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.71787029504776, + "num_tokens": 106441185.0, + "step": 4130 + }, + { + "epoch": 0.4536569294970349, + "grad_norm": 2.2571890354156494, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6967563629150391, + "num_tokens": 106464716.0, + "step": 4131 + }, + { + "epoch": 0.4537667471996486, + "grad_norm": 2.209879159927368, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7040472626686096, + "num_tokens": 106491181.0, + "step": 4132 + }, + { + "epoch": 0.45387656490226225, + "grad_norm": 2.555839776992798, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7008116245269775, + "num_tokens": 106511268.0, + "step": 4133 + }, + { + "epoch": 0.4539863826048759, + "grad_norm": 2.095764398574829, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6874667406082153, + "num_tokens": 106539304.0, + "step": 4134 + }, + { + "epoch": 0.4540962003074896, + "grad_norm": 2.256908655166626, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7057616710662842, + "num_tokens": 106565116.0, + "step": 4135 + }, + { + "epoch": 0.45420601801010324, + "grad_norm": 2.248871088027954, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7044767737388611, + "num_tokens": 106589238.0, + "step": 4136 + }, + { + "epoch": 0.4543158357127169, + "grad_norm": 2.2921969890594482, + "learning_rate": 1e-06, + "loss": 1.0926, + "mean_token_accuracy": 0.6806166768074036, + "num_tokens": 106614543.0, + "step": 4137 + }, + { + "epoch": 0.45442565341533053, + "grad_norm": 2.4509689807891846, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7242482900619507, + "num_tokens": 106633708.0, + "step": 4138 + }, + { + "epoch": 0.45453547111794423, + "grad_norm": 2.504709243774414, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7116037607192993, + "num_tokens": 106654967.0, + "step": 4139 + }, + { + "epoch": 0.4546452888205579, + "grad_norm": 2.261777639389038, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6915780305862427, + "num_tokens": 106681675.0, + "step": 4140 + }, + { + "epoch": 0.4547551065231715, + "grad_norm": 2.032258987426758, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6910485029220581, + "num_tokens": 106710569.0, + "step": 4141 + }, + { + "epoch": 0.45486492422578517, + "grad_norm": 2.235633611679077, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6960257291793823, + "num_tokens": 106733996.0, + "step": 4142 + }, + { + "epoch": 0.4549747419283989, + "grad_norm": 2.2758188247680664, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7257190942764282, + "num_tokens": 106756790.0, + "step": 4143 + }, + { + "epoch": 0.4550845596310125, + "grad_norm": 2.4644596576690674, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6890754103660583, + "num_tokens": 106779588.0, + "step": 4144 + }, + { + "epoch": 0.45519437733362617, + "grad_norm": 2.130537748336792, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7100791931152344, + "num_tokens": 106805617.0, + "step": 4145 + }, + { + "epoch": 0.45530419503623987, + "grad_norm": 2.435009717941284, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7119869589805603, + "num_tokens": 106826018.0, + "step": 4146 + }, + { + "epoch": 0.4554140127388535, + "grad_norm": 2.0367190837860107, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6930916905403137, + "num_tokens": 106856975.0, + "step": 4147 + }, + { + "epoch": 0.45552383044146716, + "grad_norm": 2.0408823490142822, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7016001343727112, + "num_tokens": 106886312.0, + "step": 4148 + }, + { + "epoch": 0.4556336481440808, + "grad_norm": 2.5206146240234375, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7138535976409912, + "num_tokens": 106905454.0, + "step": 4149 + }, + { + "epoch": 0.4557434658466945, + "grad_norm": 2.1004087924957275, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7014105319976807, + "num_tokens": 106933141.0, + "step": 4150 + }, + { + "epoch": 0.45585328354930815, + "grad_norm": 2.419504404067993, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7148276567459106, + "num_tokens": 106954407.0, + "step": 4151 + }, + { + "epoch": 0.4559631012519218, + "grad_norm": 2.2511303424835205, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7028881311416626, + "num_tokens": 106979424.0, + "step": 4152 + }, + { + "epoch": 0.45607291895453544, + "grad_norm": 2.075490713119507, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.7169148921966553, + "num_tokens": 107007592.0, + "step": 4153 + }, + { + "epoch": 0.45618273665714915, + "grad_norm": 2.2304463386535645, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.70982825756073, + "num_tokens": 107032322.0, + "step": 4154 + }, + { + "epoch": 0.4562925543597628, + "grad_norm": 2.029071092605591, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7139513492584229, + "num_tokens": 107060475.0, + "step": 4155 + }, + { + "epoch": 0.45640237206237644, + "grad_norm": 2.0672104358673096, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7189521193504333, + "num_tokens": 107089978.0, + "step": 4156 + }, + { + "epoch": 0.45651218976499014, + "grad_norm": 2.2531962394714355, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7350946664810181, + "num_tokens": 107112718.0, + "step": 4157 + }, + { + "epoch": 0.4566220074676038, + "grad_norm": 1.9897871017456055, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7050048112869263, + "num_tokens": 107141776.0, + "step": 4158 + }, + { + "epoch": 0.45673182517021743, + "grad_norm": 2.45945143699646, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7112966775894165, + "num_tokens": 107160985.0, + "step": 4159 + }, + { + "epoch": 0.4568416428728311, + "grad_norm": 2.178784132003784, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7082182168960571, + "num_tokens": 107187169.0, + "step": 4160 + }, + { + "epoch": 0.4569514605754448, + "grad_norm": 1.7583799362182617, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7206888794898987, + "num_tokens": 107221879.0, + "step": 4161 + }, + { + "epoch": 0.4570612782780584, + "grad_norm": 2.10298490524292, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6982759833335876, + "num_tokens": 107248064.0, + "step": 4162 + }, + { + "epoch": 0.45717109598067207, + "grad_norm": 2.056306838989258, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7148799896240234, + "num_tokens": 107275098.0, + "step": 4163 + }, + { + "epoch": 0.4572809136832858, + "grad_norm": 2.369786024093628, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7271881103515625, + "num_tokens": 107297243.0, + "step": 4164 + }, + { + "epoch": 0.4573907313858994, + "grad_norm": 2.4887821674346924, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.706577718257904, + "num_tokens": 107318027.0, + "step": 4165 + }, + { + "epoch": 0.45750054908851306, + "grad_norm": 2.019524097442627, + "learning_rate": 1e-06, + "loss": 1.0827, + "mean_token_accuracy": 0.6794009804725647, + "num_tokens": 107348552.0, + "step": 4166 + }, + { + "epoch": 0.4576103667911267, + "grad_norm": 2.1637439727783203, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7068860530853271, + "num_tokens": 107374144.0, + "step": 4167 + }, + { + "epoch": 0.4577201844937404, + "grad_norm": 2.03432297706604, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.70247882604599, + "num_tokens": 107404035.0, + "step": 4168 + }, + { + "epoch": 0.45783000219635406, + "grad_norm": 2.214848756790161, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7020025849342346, + "num_tokens": 107429070.0, + "step": 4169 + }, + { + "epoch": 0.4579398198989677, + "grad_norm": 2.392346143722534, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7339882850646973, + "num_tokens": 107449799.0, + "step": 4170 + }, + { + "epoch": 0.45804963760158135, + "grad_norm": 2.372307777404785, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7188957333564758, + "num_tokens": 107470308.0, + "step": 4171 + }, + { + "epoch": 0.45815945530419505, + "grad_norm": 2.3662564754486084, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7190061807632446, + "num_tokens": 107491327.0, + "step": 4172 + }, + { + "epoch": 0.4582692730068087, + "grad_norm": 2.1442766189575195, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6910442113876343, + "num_tokens": 107518917.0, + "step": 4173 + }, + { + "epoch": 0.45837909070942234, + "grad_norm": 2.003915309906006, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7048778533935547, + "num_tokens": 107547858.0, + "step": 4174 + }, + { + "epoch": 0.45848890841203604, + "grad_norm": 2.151806354522705, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7016209959983826, + "num_tokens": 107574899.0, + "step": 4175 + }, + { + "epoch": 0.4585987261146497, + "grad_norm": 2.2538199424743652, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6987353563308716, + "num_tokens": 107601591.0, + "step": 4176 + }, + { + "epoch": 0.45870854381726334, + "grad_norm": 2.3415396213531494, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6938462257385254, + "num_tokens": 107624300.0, + "step": 4177 + }, + { + "epoch": 0.458818361519877, + "grad_norm": 2.329636573791504, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7102587223052979, + "num_tokens": 107647950.0, + "step": 4178 + }, + { + "epoch": 0.4589281792224907, + "grad_norm": 2.1757798194885254, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6945698857307434, + "num_tokens": 107674448.0, + "step": 4179 + }, + { + "epoch": 0.45903799692510433, + "grad_norm": 2.284602642059326, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7252078652381897, + "num_tokens": 107696559.0, + "step": 4180 + }, + { + "epoch": 0.459147814627718, + "grad_norm": 2.4111526012420654, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7090877294540405, + "num_tokens": 107718820.0, + "step": 4181 + }, + { + "epoch": 0.4592576323303316, + "grad_norm": 2.1027636528015137, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.693769097328186, + "num_tokens": 107747096.0, + "step": 4182 + }, + { + "epoch": 0.4593674500329453, + "grad_norm": 2.3045504093170166, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7011897563934326, + "num_tokens": 107771801.0, + "step": 4183 + }, + { + "epoch": 0.45947726773555897, + "grad_norm": 2.3010575771331787, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7019520401954651, + "num_tokens": 107796689.0, + "step": 4184 + }, + { + "epoch": 0.4595870854381726, + "grad_norm": 2.0905685424804688, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6989336013793945, + "num_tokens": 107826372.0, + "step": 4185 + }, + { + "epoch": 0.4596969031407863, + "grad_norm": 2.3636319637298584, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7013034820556641, + "num_tokens": 107849953.0, + "step": 4186 + }, + { + "epoch": 0.45980672084339996, + "grad_norm": 2.171532392501831, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6873023509979248, + "num_tokens": 107875957.0, + "step": 4187 + }, + { + "epoch": 0.4599165385460136, + "grad_norm": 2.119562864303589, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7065308690071106, + "num_tokens": 107906459.0, + "step": 4188 + }, + { + "epoch": 0.46002635624862726, + "grad_norm": 2.4813854694366455, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7062702178955078, + "num_tokens": 107928439.0, + "step": 4189 + }, + { + "epoch": 0.46013617395124096, + "grad_norm": 2.3021883964538574, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7041894197463989, + "num_tokens": 107952823.0, + "step": 4190 + }, + { + "epoch": 0.4602459916538546, + "grad_norm": 2.096449136734009, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6982560157775879, + "num_tokens": 107981294.0, + "step": 4191 + }, + { + "epoch": 0.46035580935646825, + "grad_norm": 2.2003731727600098, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6923109292984009, + "num_tokens": 108009062.0, + "step": 4192 + }, + { + "epoch": 0.46046562705908195, + "grad_norm": 2.3359367847442627, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7124511003494263, + "num_tokens": 108032534.0, + "step": 4193 + }, + { + "epoch": 0.4605754447616956, + "grad_norm": 2.263054370880127, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.733500599861145, + "num_tokens": 108055493.0, + "step": 4194 + }, + { + "epoch": 0.46068526246430924, + "grad_norm": 1.9468861818313599, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6780929565429688, + "num_tokens": 108086791.0, + "step": 4195 + }, + { + "epoch": 0.4607950801669229, + "grad_norm": 2.2020182609558105, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6949601173400879, + "num_tokens": 108112193.0, + "step": 4196 + }, + { + "epoch": 0.4609048978695366, + "grad_norm": 2.4361753463745117, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7051585912704468, + "num_tokens": 108135252.0, + "step": 4197 + }, + { + "epoch": 0.46101471557215024, + "grad_norm": 2.221134901046753, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7081238031387329, + "num_tokens": 108160668.0, + "step": 4198 + }, + { + "epoch": 0.4611245332747639, + "grad_norm": 2.315734386444092, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7012280225753784, + "num_tokens": 108184239.0, + "step": 4199 + }, + { + "epoch": 0.4612343509773775, + "grad_norm": 2.2759177684783936, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7105268836021423, + "num_tokens": 108208010.0, + "step": 4200 + }, + { + "epoch": 0.46134416867999123, + "grad_norm": 1.933605432510376, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.6856229901313782, + "num_tokens": 108241746.0, + "step": 4201 + }, + { + "epoch": 0.4614539863826049, + "grad_norm": 1.8182895183563232, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6866245865821838, + "num_tokens": 108277773.0, + "step": 4202 + }, + { + "epoch": 0.4615638040852185, + "grad_norm": 2.197930097579956, + "learning_rate": 1e-06, + "loss": 1.1466, + "mean_token_accuracy": 0.6580252647399902, + "num_tokens": 108304032.0, + "step": 4203 + }, + { + "epoch": 0.4616736217878322, + "grad_norm": 1.8612425327301025, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7158251404762268, + "num_tokens": 108336096.0, + "step": 4204 + }, + { + "epoch": 0.46178343949044587, + "grad_norm": 2.059993267059326, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6841193437576294, + "num_tokens": 108367259.0, + "step": 4205 + }, + { + "epoch": 0.4618932571930595, + "grad_norm": 2.4486083984375, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7001795768737793, + "num_tokens": 108390465.0, + "step": 4206 + }, + { + "epoch": 0.46200307489567316, + "grad_norm": 2.2403571605682373, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6939677000045776, + "num_tokens": 108415120.0, + "step": 4207 + }, + { + "epoch": 0.46211289259828686, + "grad_norm": 2.0106420516967773, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7125231027603149, + "num_tokens": 108443780.0, + "step": 4208 + }, + { + "epoch": 0.4622227103009005, + "grad_norm": 2.057803153991699, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.701795220375061, + "num_tokens": 108472446.0, + "step": 4209 + }, + { + "epoch": 0.46233252800351415, + "grad_norm": 2.1100223064422607, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7014002799987793, + "num_tokens": 108497610.0, + "step": 4210 + }, + { + "epoch": 0.46244234570612786, + "grad_norm": 2.378680944442749, + "learning_rate": 1e-06, + "loss": 1.1013, + "mean_token_accuracy": 0.6762545704841614, + "num_tokens": 108520504.0, + "step": 4211 + }, + { + "epoch": 0.4625521634087415, + "grad_norm": 2.116213321685791, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6950287818908691, + "num_tokens": 108546863.0, + "step": 4212 + }, + { + "epoch": 0.46266198111135515, + "grad_norm": 2.168161630630493, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7015013098716736, + "num_tokens": 108573576.0, + "step": 4213 + }, + { + "epoch": 0.4627717988139688, + "grad_norm": 2.1503658294677734, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7254040241241455, + "num_tokens": 108598069.0, + "step": 4214 + }, + { + "epoch": 0.4628816165165825, + "grad_norm": 2.289677858352661, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7185872793197632, + "num_tokens": 108623402.0, + "step": 4215 + }, + { + "epoch": 0.46299143421919614, + "grad_norm": 1.9902113676071167, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7258591651916504, + "num_tokens": 108651448.0, + "step": 4216 + }, + { + "epoch": 0.4631012519218098, + "grad_norm": 2.2176809310913086, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7132126092910767, + "num_tokens": 108675808.0, + "step": 4217 + }, + { + "epoch": 0.46321106962442343, + "grad_norm": 2.3982229232788086, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7227266430854797, + "num_tokens": 108699218.0, + "step": 4218 + }, + { + "epoch": 0.46332088732703713, + "grad_norm": 2.0647408962249756, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7024540901184082, + "num_tokens": 108728653.0, + "step": 4219 + }, + { + "epoch": 0.4634307050296508, + "grad_norm": 1.870071291923523, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.696130096912384, + "num_tokens": 108765409.0, + "step": 4220 + }, + { + "epoch": 0.4635405227322644, + "grad_norm": 2.3236727714538574, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7060816884040833, + "num_tokens": 108788635.0, + "step": 4221 + }, + { + "epoch": 0.4636503404348781, + "grad_norm": 2.2362916469573975, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7157633304595947, + "num_tokens": 108813511.0, + "step": 4222 + }, + { + "epoch": 0.4637601581374918, + "grad_norm": 2.0442404747009277, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6952181458473206, + "num_tokens": 108843406.0, + "step": 4223 + }, + { + "epoch": 0.4638699758401054, + "grad_norm": 2.3232569694519043, + "learning_rate": 1e-06, + "loss": 1.124, + "mean_token_accuracy": 0.6692008972167969, + "num_tokens": 108873254.0, + "step": 4224 + }, + { + "epoch": 0.46397979354271907, + "grad_norm": 2.152921438217163, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7104313969612122, + "num_tokens": 108899764.0, + "step": 4225 + }, + { + "epoch": 0.46408961124533277, + "grad_norm": 2.140078067779541, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6990241408348083, + "num_tokens": 108929554.0, + "step": 4226 + }, + { + "epoch": 0.4641994289479464, + "grad_norm": 2.157062292098999, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7313234806060791, + "num_tokens": 108954783.0, + "step": 4227 + }, + { + "epoch": 0.46430924665056006, + "grad_norm": 2.3986103534698486, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6882438659667969, + "num_tokens": 108976319.0, + "step": 4228 + }, + { + "epoch": 0.4644190643531737, + "grad_norm": 2.415829658508301, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6983996033668518, + "num_tokens": 109001145.0, + "step": 4229 + }, + { + "epoch": 0.4645288820557874, + "grad_norm": 1.9951786994934082, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.698704183101654, + "num_tokens": 109030478.0, + "step": 4230 + }, + { + "epoch": 0.46463869975840105, + "grad_norm": 2.3420395851135254, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.723493754863739, + "num_tokens": 109051522.0, + "step": 4231 + }, + { + "epoch": 0.4647485174610147, + "grad_norm": 2.334994077682495, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7066189050674438, + "num_tokens": 109075143.0, + "step": 4232 + }, + { + "epoch": 0.4648583351636284, + "grad_norm": 2.13000750541687, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7228352427482605, + "num_tokens": 109102434.0, + "step": 4233 + }, + { + "epoch": 0.46496815286624205, + "grad_norm": 2.0613579750061035, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7013213634490967, + "num_tokens": 109130097.0, + "step": 4234 + }, + { + "epoch": 0.4650779705688557, + "grad_norm": 2.0718653202056885, + "learning_rate": 1e-06, + "loss": 1.1262, + "mean_token_accuracy": 0.6721882820129395, + "num_tokens": 109160526.0, + "step": 4235 + }, + { + "epoch": 0.46518778827146934, + "grad_norm": 2.011425256729126, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7221848368644714, + "num_tokens": 109190010.0, + "step": 4236 + }, + { + "epoch": 0.46529760597408304, + "grad_norm": 2.254258394241333, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7011152505874634, + "num_tokens": 109214813.0, + "step": 4237 + }, + { + "epoch": 0.4654074236766967, + "grad_norm": 2.035205602645874, + "learning_rate": 1e-06, + "loss": 1.0657, + "mean_token_accuracy": 0.6796387434005737, + "num_tokens": 109244424.0, + "step": 4238 + }, + { + "epoch": 0.46551724137931033, + "grad_norm": 2.1332876682281494, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7067570090293884, + "num_tokens": 109272102.0, + "step": 4239 + }, + { + "epoch": 0.46562705908192403, + "grad_norm": 2.3142786026000977, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.7030372619628906, + "num_tokens": 109297357.0, + "step": 4240 + }, + { + "epoch": 0.4657368767845377, + "grad_norm": 2.1841213703155518, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6965857148170471, + "num_tokens": 109322081.0, + "step": 4241 + }, + { + "epoch": 0.4658466944871513, + "grad_norm": 2.293426990509033, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7172495126724243, + "num_tokens": 109343521.0, + "step": 4242 + }, + { + "epoch": 0.46595651218976497, + "grad_norm": 2.336932897567749, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.7018766403198242, + "num_tokens": 109366904.0, + "step": 4243 + }, + { + "epoch": 0.4660663298923787, + "grad_norm": 2.2084059715270996, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6977598667144775, + "num_tokens": 109393576.0, + "step": 4244 + }, + { + "epoch": 0.4661761475949923, + "grad_norm": 1.9884662628173828, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7086536288261414, + "num_tokens": 109422946.0, + "step": 4245 + }, + { + "epoch": 0.46628596529760596, + "grad_norm": 2.605753183364868, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7055575847625732, + "num_tokens": 109442966.0, + "step": 4246 + }, + { + "epoch": 0.4663957830002196, + "grad_norm": 1.9441554546356201, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7026180624961853, + "num_tokens": 109473042.0, + "step": 4247 + }, + { + "epoch": 0.4665056007028333, + "grad_norm": 2.563471555709839, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7047956585884094, + "num_tokens": 109494776.0, + "step": 4248 + }, + { + "epoch": 0.46661541840544696, + "grad_norm": 2.1584537029266357, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6907356977462769, + "num_tokens": 109523261.0, + "step": 4249 + }, + { + "epoch": 0.4667252361080606, + "grad_norm": 2.364185094833374, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7023321390151978, + "num_tokens": 109545361.0, + "step": 4250 + }, + { + "epoch": 0.4668350538106743, + "grad_norm": 2.4456593990325928, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6940556168556213, + "num_tokens": 109568174.0, + "step": 4251 + }, + { + "epoch": 0.46694487151328795, + "grad_norm": 2.468104124069214, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7227562665939331, + "num_tokens": 109588035.0, + "step": 4252 + }, + { + "epoch": 0.4670546892159016, + "grad_norm": 2.1011340618133545, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7090436816215515, + "num_tokens": 109614180.0, + "step": 4253 + }, + { + "epoch": 0.46716450691851524, + "grad_norm": 2.1447372436523438, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.706760048866272, + "num_tokens": 109638594.0, + "step": 4254 + }, + { + "epoch": 0.46727432462112894, + "grad_norm": 2.374762535095215, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7125357985496521, + "num_tokens": 109660171.0, + "step": 4255 + }, + { + "epoch": 0.4673841423237426, + "grad_norm": 2.3116767406463623, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7211328744888306, + "num_tokens": 109684171.0, + "step": 4256 + }, + { + "epoch": 0.46749396002635624, + "grad_norm": 2.100900650024414, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6851378679275513, + "num_tokens": 109712700.0, + "step": 4257 + }, + { + "epoch": 0.4676037777289699, + "grad_norm": 2.3940775394439697, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7130063772201538, + "num_tokens": 109734990.0, + "step": 4258 + }, + { + "epoch": 0.4677135954315836, + "grad_norm": 2.2657370567321777, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6982678174972534, + "num_tokens": 109759808.0, + "step": 4259 + }, + { + "epoch": 0.46782341313419723, + "grad_norm": 2.3064377307891846, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7075403928756714, + "num_tokens": 109782888.0, + "step": 4260 + }, + { + "epoch": 0.4679332308368109, + "grad_norm": 2.1467719078063965, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6921693682670593, + "num_tokens": 109810133.0, + "step": 4261 + }, + { + "epoch": 0.4680430485394246, + "grad_norm": 2.436255931854248, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.703502357006073, + "num_tokens": 109830723.0, + "step": 4262 + }, + { + "epoch": 0.4681528662420382, + "grad_norm": 2.5378777980804443, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7044932246208191, + "num_tokens": 109850512.0, + "step": 4263 + }, + { + "epoch": 0.46826268394465187, + "grad_norm": 2.2825608253479004, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7016653418540955, + "num_tokens": 109874884.0, + "step": 4264 + }, + { + "epoch": 0.4683725016472655, + "grad_norm": 2.2910563945770264, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7040282487869263, + "num_tokens": 109900711.0, + "step": 4265 + }, + { + "epoch": 0.4684823193498792, + "grad_norm": 2.143834352493286, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.715896487236023, + "num_tokens": 109928230.0, + "step": 4266 + }, + { + "epoch": 0.46859213705249286, + "grad_norm": 2.5548348426818848, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7270544767379761, + "num_tokens": 109947895.0, + "step": 4267 + }, + { + "epoch": 0.4687019547551065, + "grad_norm": 2.4786243438720703, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7161449193954468, + "num_tokens": 109968151.0, + "step": 4268 + }, + { + "epoch": 0.4688117724577202, + "grad_norm": 2.204367160797119, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7087932825088501, + "num_tokens": 109994500.0, + "step": 4269 + }, + { + "epoch": 0.46892159016033386, + "grad_norm": 2.26769757270813, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7041891813278198, + "num_tokens": 110018241.0, + "step": 4270 + }, + { + "epoch": 0.4690314078629475, + "grad_norm": 2.248250961303711, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.7048072814941406, + "num_tokens": 110043074.0, + "step": 4271 + }, + { + "epoch": 0.46914122556556115, + "grad_norm": 2.2247090339660645, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6904983520507812, + "num_tokens": 110068180.0, + "step": 4272 + }, + { + "epoch": 0.46925104326817485, + "grad_norm": 2.0715301036834717, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.714040994644165, + "num_tokens": 110095584.0, + "step": 4273 + }, + { + "epoch": 0.4693608609707885, + "grad_norm": 2.265427827835083, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7059374451637268, + "num_tokens": 110119325.0, + "step": 4274 + }, + { + "epoch": 0.46947067867340214, + "grad_norm": 2.019953966140747, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7101887464523315, + "num_tokens": 110146157.0, + "step": 4275 + }, + { + "epoch": 0.4695804963760158, + "grad_norm": 2.2302768230438232, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6995055675506592, + "num_tokens": 110171593.0, + "step": 4276 + }, + { + "epoch": 0.4696903140786295, + "grad_norm": 2.497138261795044, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7015858888626099, + "num_tokens": 110192329.0, + "step": 4277 + }, + { + "epoch": 0.46980013178124314, + "grad_norm": 2.019388198852539, + "learning_rate": 1e-06, + "loss": 1.0759, + "mean_token_accuracy": 0.6762692928314209, + "num_tokens": 110221783.0, + "step": 4278 + }, + { + "epoch": 0.4699099494838568, + "grad_norm": 2.263045310974121, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7082022428512573, + "num_tokens": 110246276.0, + "step": 4279 + }, + { + "epoch": 0.4700197671864705, + "grad_norm": 1.9955224990844727, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6833937764167786, + "num_tokens": 110275105.0, + "step": 4280 + }, + { + "epoch": 0.47012958488908413, + "grad_norm": 2.4361541271209717, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6996349096298218, + "num_tokens": 110296757.0, + "step": 4281 + }, + { + "epoch": 0.4702394025916978, + "grad_norm": 2.0122199058532715, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6923406720161438, + "num_tokens": 110326095.0, + "step": 4282 + }, + { + "epoch": 0.4703492202943114, + "grad_norm": 2.4293689727783203, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7246614694595337, + "num_tokens": 110349434.0, + "step": 4283 + }, + { + "epoch": 0.4704590379969251, + "grad_norm": 2.2078988552093506, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7097456455230713, + "num_tokens": 110375283.0, + "step": 4284 + }, + { + "epoch": 0.47056885569953877, + "grad_norm": 2.285775899887085, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6892622709274292, + "num_tokens": 110399053.0, + "step": 4285 + }, + { + "epoch": 0.4706786734021524, + "grad_norm": 2.2433159351348877, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6904902458190918, + "num_tokens": 110425585.0, + "step": 4286 + }, + { + "epoch": 0.4707884911047661, + "grad_norm": 2.3353443145751953, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6911131143569946, + "num_tokens": 110450676.0, + "step": 4287 + }, + { + "epoch": 0.47089830880737976, + "grad_norm": 2.17281436920166, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6827205419540405, + "num_tokens": 110479432.0, + "step": 4288 + }, + { + "epoch": 0.4710081265099934, + "grad_norm": 1.841239333152771, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6985688805580139, + "num_tokens": 110513068.0, + "step": 4289 + }, + { + "epoch": 0.47111794421260705, + "grad_norm": 2.3318886756896973, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6969634890556335, + "num_tokens": 110539497.0, + "step": 4290 + }, + { + "epoch": 0.47122776191522076, + "grad_norm": 2.3966400623321533, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7120972871780396, + "num_tokens": 110560454.0, + "step": 4291 + }, + { + "epoch": 0.4713375796178344, + "grad_norm": 1.962507963180542, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6959336996078491, + "num_tokens": 110592651.0, + "step": 4292 + }, + { + "epoch": 0.47144739732044805, + "grad_norm": 2.016411304473877, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7013298273086548, + "num_tokens": 110625178.0, + "step": 4293 + }, + { + "epoch": 0.4715572150230617, + "grad_norm": 2.127373695373535, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6985375285148621, + "num_tokens": 110651159.0, + "step": 4294 + }, + { + "epoch": 0.4716670327256754, + "grad_norm": 2.141164779663086, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.721023440361023, + "num_tokens": 110675010.0, + "step": 4295 + }, + { + "epoch": 0.47177685042828904, + "grad_norm": 2.114741802215576, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7004140019416809, + "num_tokens": 110705022.0, + "step": 4296 + }, + { + "epoch": 0.4718866681309027, + "grad_norm": 2.1669952869415283, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7289954423904419, + "num_tokens": 110731822.0, + "step": 4297 + }, + { + "epoch": 0.4719964858335164, + "grad_norm": 2.1979072093963623, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6912648677825928, + "num_tokens": 110757304.0, + "step": 4298 + }, + { + "epoch": 0.47210630353613003, + "grad_norm": 2.280139446258545, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6898320913314819, + "num_tokens": 110783987.0, + "step": 4299 + }, + { + "epoch": 0.4722161212387437, + "grad_norm": 2.010988712310791, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6892312169075012, + "num_tokens": 110814709.0, + "step": 4300 + }, + { + "epoch": 0.4723259389413573, + "grad_norm": 2.05157470703125, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7092908024787903, + "num_tokens": 110842647.0, + "step": 4301 + }, + { + "epoch": 0.472435756643971, + "grad_norm": 2.4204390048980713, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7158182859420776, + "num_tokens": 110865288.0, + "step": 4302 + }, + { + "epoch": 0.4725455743465847, + "grad_norm": 2.121873617172241, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6977242827415466, + "num_tokens": 110893033.0, + "step": 4303 + }, + { + "epoch": 0.4726553920491983, + "grad_norm": 2.085798978805542, + "learning_rate": 1e-06, + "loss": 1.1091, + "mean_token_accuracy": 0.671903133392334, + "num_tokens": 110924799.0, + "step": 4304 + }, + { + "epoch": 0.47276520975181197, + "grad_norm": 2.171114206314087, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6800701022148132, + "num_tokens": 110954489.0, + "step": 4305 + }, + { + "epoch": 0.47287502745442567, + "grad_norm": 2.2352302074432373, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7126326560974121, + "num_tokens": 110981489.0, + "step": 4306 + }, + { + "epoch": 0.4729848451570393, + "grad_norm": 2.2786049842834473, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6965852379798889, + "num_tokens": 111007886.0, + "step": 4307 + }, + { + "epoch": 0.47309466285965296, + "grad_norm": 2.4200541973114014, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7254970073699951, + "num_tokens": 111029952.0, + "step": 4308 + }, + { + "epoch": 0.47320448056226666, + "grad_norm": 2.4876549243927, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6946777105331421, + "num_tokens": 111051846.0, + "step": 4309 + }, + { + "epoch": 0.4733142982648803, + "grad_norm": 2.278191089630127, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7098767757415771, + "num_tokens": 111080998.0, + "step": 4310 + }, + { + "epoch": 0.47342411596749395, + "grad_norm": 2.1084632873535156, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.6984310150146484, + "num_tokens": 111107095.0, + "step": 4311 + }, + { + "epoch": 0.4735339336701076, + "grad_norm": 2.1133761405944824, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7069962620735168, + "num_tokens": 111133003.0, + "step": 4312 + }, + { + "epoch": 0.4736437513727213, + "grad_norm": 2.195467472076416, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7240171432495117, + "num_tokens": 111158759.0, + "step": 4313 + }, + { + "epoch": 0.47375356907533495, + "grad_norm": 1.877242088317871, + "learning_rate": 1e-06, + "loss": 1.0792, + "mean_token_accuracy": 0.6861321330070496, + "num_tokens": 111191137.0, + "step": 4314 + }, + { + "epoch": 0.4738633867779486, + "grad_norm": 2.316622495651245, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7102864980697632, + "num_tokens": 111214592.0, + "step": 4315 + }, + { + "epoch": 0.4739732044805623, + "grad_norm": 2.4300198554992676, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7163515090942383, + "num_tokens": 111236924.0, + "step": 4316 + }, + { + "epoch": 0.47408302218317594, + "grad_norm": 2.10792875289917, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6867971420288086, + "num_tokens": 111266464.0, + "step": 4317 + }, + { + "epoch": 0.4741928398857896, + "grad_norm": 2.0403225421905518, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6975981593132019, + "num_tokens": 111293513.0, + "step": 4318 + }, + { + "epoch": 0.47430265758840323, + "grad_norm": 2.025221109390259, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6826932430267334, + "num_tokens": 111324314.0, + "step": 4319 + }, + { + "epoch": 0.47441247529101693, + "grad_norm": 2.019935131072998, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7132858037948608, + "num_tokens": 111352006.0, + "step": 4320 + }, + { + "epoch": 0.4745222929936306, + "grad_norm": 2.0565805435180664, + "learning_rate": 1e-06, + "loss": 1.0878, + "mean_token_accuracy": 0.6774547100067139, + "num_tokens": 111382996.0, + "step": 4321 + }, + { + "epoch": 0.4746321106962442, + "grad_norm": 2.116079807281494, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7187293767929077, + "num_tokens": 111408844.0, + "step": 4322 + }, + { + "epoch": 0.47474192839885787, + "grad_norm": 2.207458734512329, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7038390636444092, + "num_tokens": 111434351.0, + "step": 4323 + }, + { + "epoch": 0.4748517461014716, + "grad_norm": 2.2339677810668945, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6933110356330872, + "num_tokens": 111459396.0, + "step": 4324 + }, + { + "epoch": 0.4749615638040852, + "grad_norm": 2.2594645023345947, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7041975855827332, + "num_tokens": 111484632.0, + "step": 4325 + }, + { + "epoch": 0.47507138150669886, + "grad_norm": 2.575984477996826, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7129088044166565, + "num_tokens": 111503606.0, + "step": 4326 + }, + { + "epoch": 0.47518119920931257, + "grad_norm": 2.0588643550872803, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.68342125415802, + "num_tokens": 111532476.0, + "step": 4327 + }, + { + "epoch": 0.4752910169119262, + "grad_norm": 1.9905807971954346, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6948604583740234, + "num_tokens": 111565850.0, + "step": 4328 + }, + { + "epoch": 0.47540083461453986, + "grad_norm": 2.418872356414795, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6940625309944153, + "num_tokens": 111589186.0, + "step": 4329 + }, + { + "epoch": 0.4755106523171535, + "grad_norm": 2.2452802658081055, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6872419118881226, + "num_tokens": 111614130.0, + "step": 4330 + }, + { + "epoch": 0.4756204700197672, + "grad_norm": 2.299823522567749, + "learning_rate": 1e-06, + "loss": 1.0811, + "mean_token_accuracy": 0.678274393081665, + "num_tokens": 111640266.0, + "step": 4331 + }, + { + "epoch": 0.47573028772238085, + "grad_norm": 2.0714187622070312, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7149847745895386, + "num_tokens": 111667215.0, + "step": 4332 + }, + { + "epoch": 0.4758401054249945, + "grad_norm": 2.421555280685425, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7056546211242676, + "num_tokens": 111690425.0, + "step": 4333 + }, + { + "epoch": 0.47594992312760814, + "grad_norm": 2.1712868213653564, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7191866040229797, + "num_tokens": 111715001.0, + "step": 4334 + }, + { + "epoch": 0.47605974083022184, + "grad_norm": 2.285749673843384, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7067304253578186, + "num_tokens": 111737171.0, + "step": 4335 + }, + { + "epoch": 0.4761695585328355, + "grad_norm": 2.3044373989105225, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6871747970581055, + "num_tokens": 111762637.0, + "step": 4336 + }, + { + "epoch": 0.47627937623544914, + "grad_norm": 2.2121050357818604, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7197582721710205, + "num_tokens": 111786350.0, + "step": 4337 + }, + { + "epoch": 0.47638919393806284, + "grad_norm": 2.0731215476989746, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6930288076400757, + "num_tokens": 111814055.0, + "step": 4338 + }, + { + "epoch": 0.4764990116406765, + "grad_norm": 2.671503782272339, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7164448499679565, + "num_tokens": 111834110.0, + "step": 4339 + }, + { + "epoch": 0.47660882934329013, + "grad_norm": 2.507997989654541, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7061711549758911, + "num_tokens": 111855673.0, + "step": 4340 + }, + { + "epoch": 0.4767186470459038, + "grad_norm": 2.691044330596924, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.712685227394104, + "num_tokens": 111873513.0, + "step": 4341 + }, + { + "epoch": 0.4768284647485175, + "grad_norm": 1.977911114692688, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7191541194915771, + "num_tokens": 111903503.0, + "step": 4342 + }, + { + "epoch": 0.4769382824511311, + "grad_norm": 2.4936115741729736, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7253303527832031, + "num_tokens": 111924726.0, + "step": 4343 + }, + { + "epoch": 0.47704810015374477, + "grad_norm": 2.433579683303833, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.689848780632019, + "num_tokens": 111946561.0, + "step": 4344 + }, + { + "epoch": 0.47715791785635847, + "grad_norm": 1.9907516241073608, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6857346892356873, + "num_tokens": 111978664.0, + "step": 4345 + }, + { + "epoch": 0.4772677355589721, + "grad_norm": 2.130263090133667, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7209345102310181, + "num_tokens": 112004162.0, + "step": 4346 + }, + { + "epoch": 0.47737755326158576, + "grad_norm": 2.289606809616089, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7078536152839661, + "num_tokens": 112029492.0, + "step": 4347 + }, + { + "epoch": 0.4774873709641994, + "grad_norm": 2.079946994781494, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7051684856414795, + "num_tokens": 112056263.0, + "step": 4348 + }, + { + "epoch": 0.4775971886668131, + "grad_norm": 2.4577884674072266, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7147128582000732, + "num_tokens": 112075601.0, + "step": 4349 + }, + { + "epoch": 0.47770700636942676, + "grad_norm": 1.8913016319274902, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.695946991443634, + "num_tokens": 112108539.0, + "step": 4350 + }, + { + "epoch": 0.4778168240720404, + "grad_norm": 2.339460849761963, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7027206420898438, + "num_tokens": 112133419.0, + "step": 4351 + }, + { + "epoch": 0.47792664177465405, + "grad_norm": 2.1383118629455566, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6901795268058777, + "num_tokens": 112159388.0, + "step": 4352 + }, + { + "epoch": 0.47803645947726775, + "grad_norm": 2.001129388809204, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6965729594230652, + "num_tokens": 112192643.0, + "step": 4353 + }, + { + "epoch": 0.4781462771798814, + "grad_norm": 2.4458184242248535, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7069859504699707, + "num_tokens": 112214186.0, + "step": 4354 + }, + { + "epoch": 0.47825609488249504, + "grad_norm": 2.4517998695373535, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7154996991157532, + "num_tokens": 112233967.0, + "step": 4355 + }, + { + "epoch": 0.47836591258510874, + "grad_norm": 2.1993765830993652, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7215672135353088, + "num_tokens": 112257768.0, + "step": 4356 + }, + { + "epoch": 0.4784757302877224, + "grad_norm": 2.12320876121521, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7007126808166504, + "num_tokens": 112284749.0, + "step": 4357 + }, + { + "epoch": 0.47858554799033604, + "grad_norm": 2.0466151237487793, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6973583698272705, + "num_tokens": 112315550.0, + "step": 4358 + }, + { + "epoch": 0.4786953656929497, + "grad_norm": 2.31725811958313, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.703488290309906, + "num_tokens": 112339910.0, + "step": 4359 + }, + { + "epoch": 0.4788051833955634, + "grad_norm": 2.143549680709839, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6998918056488037, + "num_tokens": 112368293.0, + "step": 4360 + }, + { + "epoch": 0.47891500109817703, + "grad_norm": 2.605360507965088, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7183799743652344, + "num_tokens": 112386768.0, + "step": 4361 + }, + { + "epoch": 0.4790248188007907, + "grad_norm": 2.4429779052734375, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7120852470397949, + "num_tokens": 112406877.0, + "step": 4362 + }, + { + "epoch": 0.4791346365034044, + "grad_norm": 2.381978988647461, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6944941282272339, + "num_tokens": 112430669.0, + "step": 4363 + }, + { + "epoch": 0.479244454206018, + "grad_norm": 2.268388271331787, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7235107421875, + "num_tokens": 112454611.0, + "step": 4364 + }, + { + "epoch": 0.47935427190863167, + "grad_norm": 2.1028690338134766, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7086197733879089, + "num_tokens": 112482286.0, + "step": 4365 + }, + { + "epoch": 0.4794640896112453, + "grad_norm": 2.1304707527160645, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6891322135925293, + "num_tokens": 112510243.0, + "step": 4366 + }, + { + "epoch": 0.479573907313859, + "grad_norm": 2.0604515075683594, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6850992441177368, + "num_tokens": 112539310.0, + "step": 4367 + }, + { + "epoch": 0.47968372501647266, + "grad_norm": 2.2343389987945557, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6840460300445557, + "num_tokens": 112563546.0, + "step": 4368 + }, + { + "epoch": 0.4797935427190863, + "grad_norm": 2.0826175212860107, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6826056838035583, + "num_tokens": 112593100.0, + "step": 4369 + }, + { + "epoch": 0.47990336042169995, + "grad_norm": 2.1496055126190186, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.68923020362854, + "num_tokens": 112620633.0, + "step": 4370 + }, + { + "epoch": 0.48001317812431366, + "grad_norm": 2.191755771636963, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7269275188446045, + "num_tokens": 112644537.0, + "step": 4371 + }, + { + "epoch": 0.4801229958269273, + "grad_norm": 2.220472812652588, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7024855017662048, + "num_tokens": 112670740.0, + "step": 4372 + }, + { + "epoch": 0.48023281352954095, + "grad_norm": 2.46720552444458, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6941333413124084, + "num_tokens": 112693247.0, + "step": 4373 + }, + { + "epoch": 0.48034263123215465, + "grad_norm": 2.1076788902282715, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6805815696716309, + "num_tokens": 112722431.0, + "step": 4374 + }, + { + "epoch": 0.4804524489347683, + "grad_norm": 2.0734221935272217, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7051382660865784, + "num_tokens": 112749897.0, + "step": 4375 + }, + { + "epoch": 0.48056226663738194, + "grad_norm": 2.058939218521118, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6972548961639404, + "num_tokens": 112777373.0, + "step": 4376 + }, + { + "epoch": 0.4806720843399956, + "grad_norm": 2.371955633163452, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7142660617828369, + "num_tokens": 112799062.0, + "step": 4377 + }, + { + "epoch": 0.4807819020426093, + "grad_norm": 2.1476426124572754, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7235515713691711, + "num_tokens": 112824561.0, + "step": 4378 + }, + { + "epoch": 0.48089171974522293, + "grad_norm": 2.451115369796753, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7102954387664795, + "num_tokens": 112847084.0, + "step": 4379 + }, + { + "epoch": 0.4810015374478366, + "grad_norm": 2.4978229999542236, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6913924813270569, + "num_tokens": 112867952.0, + "step": 4380 + }, + { + "epoch": 0.4811113551504502, + "grad_norm": 2.205024480819702, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.705734372138977, + "num_tokens": 112893494.0, + "step": 4381 + }, + { + "epoch": 0.4812211728530639, + "grad_norm": 2.0644235610961914, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7018532752990723, + "num_tokens": 112919961.0, + "step": 4382 + }, + { + "epoch": 0.4813309905556776, + "grad_norm": 2.2953946590423584, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7148066163063049, + "num_tokens": 112943402.0, + "step": 4383 + }, + { + "epoch": 0.4814408082582912, + "grad_norm": 2.2802722454071045, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6967178583145142, + "num_tokens": 112969079.0, + "step": 4384 + }, + { + "epoch": 0.4815506259609049, + "grad_norm": 2.379589557647705, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6994776725769043, + "num_tokens": 112990743.0, + "step": 4385 + }, + { + "epoch": 0.48166044366351857, + "grad_norm": 2.1477489471435547, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7344484329223633, + "num_tokens": 113015952.0, + "step": 4386 + }, + { + "epoch": 0.4817702613661322, + "grad_norm": 2.6183109283447266, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7093641757965088, + "num_tokens": 113035604.0, + "step": 4387 + }, + { + "epoch": 0.48188007906874586, + "grad_norm": 2.2176225185394287, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7163822650909424, + "num_tokens": 113058214.0, + "step": 4388 + }, + { + "epoch": 0.48198989677135956, + "grad_norm": 2.2347352504730225, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.712065577507019, + "num_tokens": 113081697.0, + "step": 4389 + }, + { + "epoch": 0.4820997144739732, + "grad_norm": 2.5273053646087646, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7005945444107056, + "num_tokens": 113103929.0, + "step": 4390 + }, + { + "epoch": 0.48220953217658685, + "grad_norm": 2.1720635890960693, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7129209041595459, + "num_tokens": 113130083.0, + "step": 4391 + }, + { + "epoch": 0.48231934987920055, + "grad_norm": 2.0750529766082764, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.69419926404953, + "num_tokens": 113158389.0, + "step": 4392 + }, + { + "epoch": 0.4824291675818142, + "grad_norm": 2.4658467769622803, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7103493213653564, + "num_tokens": 113178545.0, + "step": 4393 + }, + { + "epoch": 0.48253898528442785, + "grad_norm": 2.781705379486084, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7042632699012756, + "num_tokens": 113196008.0, + "step": 4394 + }, + { + "epoch": 0.4826488029870415, + "grad_norm": 2.0122361183166504, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7033027410507202, + "num_tokens": 113224619.0, + "step": 4395 + }, + { + "epoch": 0.4827586206896552, + "grad_norm": 2.0242385864257812, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7226650714874268, + "num_tokens": 113250605.0, + "step": 4396 + }, + { + "epoch": 0.48286843839226884, + "grad_norm": 2.4758214950561523, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7095286846160889, + "num_tokens": 113272426.0, + "step": 4397 + }, + { + "epoch": 0.4829782560948825, + "grad_norm": 2.45625901222229, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6935277581214905, + "num_tokens": 113293668.0, + "step": 4398 + }, + { + "epoch": 0.48308807379749613, + "grad_norm": 2.514096736907959, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.705689013004303, + "num_tokens": 113314345.0, + "step": 4399 + }, + { + "epoch": 0.48319789150010983, + "grad_norm": 2.265324354171753, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7300690412521362, + "num_tokens": 113336734.0, + "step": 4400 + }, + { + "epoch": 0.4833077092027235, + "grad_norm": 2.1081833839416504, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7088289260864258, + "num_tokens": 113363902.0, + "step": 4401 + }, + { + "epoch": 0.4834175269053371, + "grad_norm": 2.1603875160217285, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6740983724594116, + "num_tokens": 113389755.0, + "step": 4402 + }, + { + "epoch": 0.4835273446079508, + "grad_norm": 2.1485867500305176, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6777675151824951, + "num_tokens": 113415551.0, + "step": 4403 + }, + { + "epoch": 0.4836371623105645, + "grad_norm": 2.055034637451172, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7040402293205261, + "num_tokens": 113443284.0, + "step": 4404 + }, + { + "epoch": 0.4837469800131781, + "grad_norm": 2.0534205436706543, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7169055342674255, + "num_tokens": 113471355.0, + "step": 4405 + }, + { + "epoch": 0.48385679771579176, + "grad_norm": 2.0889105796813965, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7112709879875183, + "num_tokens": 113500453.0, + "step": 4406 + }, + { + "epoch": 0.48396661541840547, + "grad_norm": 2.239372730255127, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6883038282394409, + "num_tokens": 113528298.0, + "step": 4407 + }, + { + "epoch": 0.4840764331210191, + "grad_norm": 2.2202935218811035, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7002121210098267, + "num_tokens": 113553140.0, + "step": 4408 + }, + { + "epoch": 0.48418625082363276, + "grad_norm": 2.3595993518829346, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6945275068283081, + "num_tokens": 113575680.0, + "step": 4409 + }, + { + "epoch": 0.4842960685262464, + "grad_norm": 2.1594760417938232, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7003233432769775, + "num_tokens": 113604720.0, + "step": 4410 + }, + { + "epoch": 0.4844058862288601, + "grad_norm": 2.4086055755615234, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7202693223953247, + "num_tokens": 113627519.0, + "step": 4411 + }, + { + "epoch": 0.48451570393147375, + "grad_norm": 2.3059017658233643, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7157495021820068, + "num_tokens": 113651427.0, + "step": 4412 + }, + { + "epoch": 0.4846255216340874, + "grad_norm": 2.369431734085083, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7086343765258789, + "num_tokens": 113672832.0, + "step": 4413 + }, + { + "epoch": 0.4847353393367011, + "grad_norm": 2.3560423851013184, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7112036943435669, + "num_tokens": 113696213.0, + "step": 4414 + }, + { + "epoch": 0.48484515703931474, + "grad_norm": 2.290518283843994, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6971755027770996, + "num_tokens": 113722011.0, + "step": 4415 + }, + { + "epoch": 0.4849549747419284, + "grad_norm": 2.6879427433013916, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7198834419250488, + "num_tokens": 113739800.0, + "step": 4416 + }, + { + "epoch": 0.48506479244454204, + "grad_norm": 2.36800217628479, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7115689516067505, + "num_tokens": 113761602.0, + "step": 4417 + }, + { + "epoch": 0.48517461014715574, + "grad_norm": 2.2182228565216064, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6991810202598572, + "num_tokens": 113786737.0, + "step": 4418 + }, + { + "epoch": 0.4852844278497694, + "grad_norm": 2.1873316764831543, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7253819704055786, + "num_tokens": 113811332.0, + "step": 4419 + }, + { + "epoch": 0.48539424555238303, + "grad_norm": 2.10897159576416, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6931244730949402, + "num_tokens": 113837566.0, + "step": 4420 + }, + { + "epoch": 0.48550406325499673, + "grad_norm": 2.500488042831421, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6951615810394287, + "num_tokens": 113859675.0, + "step": 4421 + }, + { + "epoch": 0.4856138809576104, + "grad_norm": 2.493300199508667, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7006492614746094, + "num_tokens": 113882462.0, + "step": 4422 + }, + { + "epoch": 0.485723698660224, + "grad_norm": 1.996612548828125, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6879227161407471, + "num_tokens": 113911733.0, + "step": 4423 + }, + { + "epoch": 0.48583351636283767, + "grad_norm": 2.366046667098999, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7390011548995972, + "num_tokens": 113934043.0, + "step": 4424 + }, + { + "epoch": 0.48594333406545137, + "grad_norm": 2.2468061447143555, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7037893533706665, + "num_tokens": 113957028.0, + "step": 4425 + }, + { + "epoch": 0.486053151768065, + "grad_norm": 2.566790819168091, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7119922637939453, + "num_tokens": 113976424.0, + "step": 4426 + }, + { + "epoch": 0.48616296947067866, + "grad_norm": 1.8979613780975342, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6867009997367859, + "num_tokens": 114007782.0, + "step": 4427 + }, + { + "epoch": 0.4862727871732923, + "grad_norm": 2.0924763679504395, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7015508413314819, + "num_tokens": 114036297.0, + "step": 4428 + }, + { + "epoch": 0.486382604875906, + "grad_norm": 2.106048583984375, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7019882202148438, + "num_tokens": 114062603.0, + "step": 4429 + }, + { + "epoch": 0.48649242257851966, + "grad_norm": 2.232235908508301, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6972577571868896, + "num_tokens": 114088742.0, + "step": 4430 + }, + { + "epoch": 0.4866022402811333, + "grad_norm": 1.9866100549697876, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7049680948257446, + "num_tokens": 114118383.0, + "step": 4431 + }, + { + "epoch": 0.486712057983747, + "grad_norm": 1.9691890478134155, + "learning_rate": 1e-06, + "loss": 1.1032, + "mean_token_accuracy": 0.674850583076477, + "num_tokens": 114152253.0, + "step": 4432 + }, + { + "epoch": 0.48682187568636065, + "grad_norm": 2.229206085205078, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6825733780860901, + "num_tokens": 114176746.0, + "step": 4433 + }, + { + "epoch": 0.4869316933889743, + "grad_norm": 2.057313919067383, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7067320346832275, + "num_tokens": 114205325.0, + "step": 4434 + }, + { + "epoch": 0.48704151109158794, + "grad_norm": 2.3721604347229004, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7069891095161438, + "num_tokens": 114230268.0, + "step": 4435 + }, + { + "epoch": 0.48715132879420164, + "grad_norm": 1.8804147243499756, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.7006769180297852, + "num_tokens": 114262380.0, + "step": 4436 + }, + { + "epoch": 0.4872611464968153, + "grad_norm": 2.2767884731292725, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6910735368728638, + "num_tokens": 114285293.0, + "step": 4437 + }, + { + "epoch": 0.48737096419942894, + "grad_norm": 1.9834829568862915, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7022058367729187, + "num_tokens": 114317579.0, + "step": 4438 + }, + { + "epoch": 0.48748078190204264, + "grad_norm": 2.0924558639526367, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6920916438102722, + "num_tokens": 114346083.0, + "step": 4439 + }, + { + "epoch": 0.4875905996046563, + "grad_norm": 2.2157700061798096, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.700226366519928, + "num_tokens": 114371533.0, + "step": 4440 + }, + { + "epoch": 0.48770041730726993, + "grad_norm": 2.114023208618164, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7328169941902161, + "num_tokens": 114396518.0, + "step": 4441 + }, + { + "epoch": 0.4878102350098836, + "grad_norm": 2.256784439086914, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7144352197647095, + "num_tokens": 114420856.0, + "step": 4442 + }, + { + "epoch": 0.4879200527124973, + "grad_norm": 2.116658926010132, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6981574296951294, + "num_tokens": 114448177.0, + "step": 4443 + }, + { + "epoch": 0.4880298704151109, + "grad_norm": 2.4497885704040527, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7301323413848877, + "num_tokens": 114466869.0, + "step": 4444 + }, + { + "epoch": 0.48813968811772457, + "grad_norm": 2.123347282409668, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6958963871002197, + "num_tokens": 114492990.0, + "step": 4445 + }, + { + "epoch": 0.4882495058203382, + "grad_norm": 2.355518341064453, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7024781703948975, + "num_tokens": 114515272.0, + "step": 4446 + }, + { + "epoch": 0.4883593235229519, + "grad_norm": 2.3993072509765625, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7076193690299988, + "num_tokens": 114538221.0, + "step": 4447 + }, + { + "epoch": 0.48846914122556556, + "grad_norm": 2.428292989730835, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.721097469329834, + "num_tokens": 114558025.0, + "step": 4448 + }, + { + "epoch": 0.4885789589281792, + "grad_norm": 2.126537322998047, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.699461817741394, + "num_tokens": 114584756.0, + "step": 4449 + }, + { + "epoch": 0.4886887766307929, + "grad_norm": 2.368326187133789, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7215386033058167, + "num_tokens": 114606390.0, + "step": 4450 + }, + { + "epoch": 0.48879859433340656, + "grad_norm": 2.100093364715576, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6959351897239685, + "num_tokens": 114635182.0, + "step": 4451 + }, + { + "epoch": 0.4889084120360202, + "grad_norm": 2.0756208896636963, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6809674501419067, + "num_tokens": 114663296.0, + "step": 4452 + }, + { + "epoch": 0.48901822973863385, + "grad_norm": 1.9656531810760498, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7120401263237, + "num_tokens": 114692895.0, + "step": 4453 + }, + { + "epoch": 0.48912804744124755, + "grad_norm": 2.1705145835876465, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7215783596038818, + "num_tokens": 114716667.0, + "step": 4454 + }, + { + "epoch": 0.4892378651438612, + "grad_norm": 2.142866849899292, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6848535537719727, + "num_tokens": 114745004.0, + "step": 4455 + }, + { + "epoch": 0.48934768284647484, + "grad_norm": 2.399930715560913, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6871370077133179, + "num_tokens": 114768918.0, + "step": 4456 + }, + { + "epoch": 0.4894575005490885, + "grad_norm": 2.1658010482788086, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6831566691398621, + "num_tokens": 114794632.0, + "step": 4457 + }, + { + "epoch": 0.4895673182517022, + "grad_norm": 2.1684041023254395, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7140242457389832, + "num_tokens": 114822960.0, + "step": 4458 + }, + { + "epoch": 0.48967713595431583, + "grad_norm": 2.4493861198425293, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6856791377067566, + "num_tokens": 114844290.0, + "step": 4459 + }, + { + "epoch": 0.4897869536569295, + "grad_norm": 2.5020225048065186, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7179861664772034, + "num_tokens": 114864642.0, + "step": 4460 + }, + { + "epoch": 0.4898967713595432, + "grad_norm": 2.2917823791503906, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6864328384399414, + "num_tokens": 114889747.0, + "step": 4461 + }, + { + "epoch": 0.4900065890621568, + "grad_norm": 2.031113386154175, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7014422416687012, + "num_tokens": 114918934.0, + "step": 4462 + }, + { + "epoch": 0.4901164067647705, + "grad_norm": 2.097844362258911, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7086925506591797, + "num_tokens": 114946500.0, + "step": 4463 + }, + { + "epoch": 0.4902262244673841, + "grad_norm": 2.3967132568359375, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7107648849487305, + "num_tokens": 114968851.0, + "step": 4464 + }, + { + "epoch": 0.4903360421699978, + "grad_norm": 2.8760743141174316, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7206429243087769, + "num_tokens": 114985110.0, + "step": 4465 + }, + { + "epoch": 0.49044585987261147, + "grad_norm": 2.1151692867279053, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7193145751953125, + "num_tokens": 115009590.0, + "step": 4466 + }, + { + "epoch": 0.4905556775752251, + "grad_norm": 2.0976781845092773, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.7009478211402893, + "num_tokens": 115037674.0, + "step": 4467 + }, + { + "epoch": 0.4906654952778388, + "grad_norm": 1.7867447137832642, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6945525407791138, + "num_tokens": 115074234.0, + "step": 4468 + }, + { + "epoch": 0.49077531298045246, + "grad_norm": 2.134337902069092, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6972765922546387, + "num_tokens": 115101352.0, + "step": 4469 + }, + { + "epoch": 0.4908851306830661, + "grad_norm": 1.791002869606018, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6758401989936829, + "num_tokens": 115141273.0, + "step": 4470 + }, + { + "epoch": 0.49099494838567975, + "grad_norm": 2.2750744819641113, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6953675150871277, + "num_tokens": 115164760.0, + "step": 4471 + }, + { + "epoch": 0.49110476608829345, + "grad_norm": 2.0264177322387695, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6875702142715454, + "num_tokens": 115194114.0, + "step": 4472 + }, + { + "epoch": 0.4912145837909071, + "grad_norm": 2.001056671142578, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6946592330932617, + "num_tokens": 115222579.0, + "step": 4473 + }, + { + "epoch": 0.49132440149352075, + "grad_norm": 2.7809524536132812, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7194510698318481, + "num_tokens": 115239976.0, + "step": 4474 + }, + { + "epoch": 0.4914342191961344, + "grad_norm": 1.9846807718276978, + "learning_rate": 1e-06, + "loss": 1.0819, + "mean_token_accuracy": 0.669640064239502, + "num_tokens": 115271278.0, + "step": 4475 + }, + { + "epoch": 0.4915440368987481, + "grad_norm": 2.3999686241149902, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6941664218902588, + "num_tokens": 115293100.0, + "step": 4476 + }, + { + "epoch": 0.49165385460136174, + "grad_norm": 2.1041510105133057, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7120095491409302, + "num_tokens": 115319935.0, + "step": 4477 + }, + { + "epoch": 0.4917636723039754, + "grad_norm": 2.041759729385376, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6913576722145081, + "num_tokens": 115351829.0, + "step": 4478 + }, + { + "epoch": 0.4918734900065891, + "grad_norm": 1.9980610609054565, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.698833703994751, + "num_tokens": 115381540.0, + "step": 4479 + }, + { + "epoch": 0.49198330770920273, + "grad_norm": 2.2925238609313965, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.709510326385498, + "num_tokens": 115405949.0, + "step": 4480 + }, + { + "epoch": 0.4920931254118164, + "grad_norm": 2.0916588306427, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.69710373878479, + "num_tokens": 115433520.0, + "step": 4481 + }, + { + "epoch": 0.49220294311443, + "grad_norm": 2.0635204315185547, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7210105657577515, + "num_tokens": 115462481.0, + "step": 4482 + }, + { + "epoch": 0.4923127608170437, + "grad_norm": 2.0362918376922607, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6966730356216431, + "num_tokens": 115492240.0, + "step": 4483 + }, + { + "epoch": 0.4924225785196574, + "grad_norm": 2.203632354736328, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6894966959953308, + "num_tokens": 115517053.0, + "step": 4484 + }, + { + "epoch": 0.492532396222271, + "grad_norm": 2.5570380687713623, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7180330753326416, + "num_tokens": 115537860.0, + "step": 4485 + }, + { + "epoch": 0.49264221392488466, + "grad_norm": 2.255284547805786, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6971048712730408, + "num_tokens": 115564696.0, + "step": 4486 + }, + { + "epoch": 0.49275203162749837, + "grad_norm": 2.058396100997925, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7063738107681274, + "num_tokens": 115594142.0, + "step": 4487 + }, + { + "epoch": 0.492861849330112, + "grad_norm": 2.335667371749878, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7064521908760071, + "num_tokens": 115618419.0, + "step": 4488 + }, + { + "epoch": 0.49297166703272566, + "grad_norm": 2.2081549167633057, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7108510136604309, + "num_tokens": 115643047.0, + "step": 4489 + }, + { + "epoch": 0.49308148473533936, + "grad_norm": 2.238434076309204, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7020387649536133, + "num_tokens": 115667540.0, + "step": 4490 + }, + { + "epoch": 0.493191302437953, + "grad_norm": 2.2135190963745117, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7005463242530823, + "num_tokens": 115692962.0, + "step": 4491 + }, + { + "epoch": 0.49330112014056665, + "grad_norm": 2.0876030921936035, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7012588977813721, + "num_tokens": 115720878.0, + "step": 4492 + }, + { + "epoch": 0.4934109378431803, + "grad_norm": 2.2323102951049805, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6975734829902649, + "num_tokens": 115746834.0, + "step": 4493 + }, + { + "epoch": 0.493520755545794, + "grad_norm": 2.392042636871338, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7096198797225952, + "num_tokens": 115770298.0, + "step": 4494 + }, + { + "epoch": 0.49363057324840764, + "grad_norm": 2.248746395111084, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.680479884147644, + "num_tokens": 115794733.0, + "step": 4495 + }, + { + "epoch": 0.4937403909510213, + "grad_norm": 1.9196356534957886, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6811990141868591, + "num_tokens": 115829895.0, + "step": 4496 + }, + { + "epoch": 0.493850208653635, + "grad_norm": 2.4059157371520996, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.7005395889282227, + "num_tokens": 115853035.0, + "step": 4497 + }, + { + "epoch": 0.49396002635624864, + "grad_norm": 2.4088311195373535, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7233462333679199, + "num_tokens": 115874497.0, + "step": 4498 + }, + { + "epoch": 0.4940698440588623, + "grad_norm": 2.2567427158355713, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7104760408401489, + "num_tokens": 115898395.0, + "step": 4499 + }, + { + "epoch": 0.49417966176147593, + "grad_norm": 2.194127321243286, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6886048913002014, + "num_tokens": 115925492.0, + "step": 4500 + }, + { + "epoch": 0.49428947946408963, + "grad_norm": 2.2392659187316895, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6995916366577148, + "num_tokens": 115949195.0, + "step": 4501 + }, + { + "epoch": 0.4943992971667033, + "grad_norm": 2.0485689640045166, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6983246803283691, + "num_tokens": 115978400.0, + "step": 4502 + }, + { + "epoch": 0.4945091148693169, + "grad_norm": 2.2134408950805664, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7083372473716736, + "num_tokens": 116003792.0, + "step": 4503 + }, + { + "epoch": 0.49461893257193057, + "grad_norm": 2.153625249862671, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7139826416969299, + "num_tokens": 116028741.0, + "step": 4504 + }, + { + "epoch": 0.49472875027454427, + "grad_norm": 2.284329891204834, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7357291579246521, + "num_tokens": 116051344.0, + "step": 4505 + }, + { + "epoch": 0.4948385679771579, + "grad_norm": 2.2329039573669434, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7118805050849915, + "num_tokens": 116073656.0, + "step": 4506 + }, + { + "epoch": 0.49494838567977156, + "grad_norm": 2.239725351333618, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6903365850448608, + "num_tokens": 116098834.0, + "step": 4507 + }, + { + "epoch": 0.49505820338238526, + "grad_norm": 2.232968330383301, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7090005874633789, + "num_tokens": 116122318.0, + "step": 4508 + }, + { + "epoch": 0.4951680210849989, + "grad_norm": 2.006152391433716, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7207432985305786, + "num_tokens": 116150455.0, + "step": 4509 + }, + { + "epoch": 0.49527783878761256, + "grad_norm": 2.2403929233551025, + "learning_rate": 1e-06, + "loss": 1.086, + "mean_token_accuracy": 0.6810002326965332, + "num_tokens": 116178142.0, + "step": 4510 + }, + { + "epoch": 0.4953876564902262, + "grad_norm": 2.0973875522613525, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7082724571228027, + "num_tokens": 116206827.0, + "step": 4511 + }, + { + "epoch": 0.4954974741928399, + "grad_norm": 2.431072235107422, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6948789358139038, + "num_tokens": 116230374.0, + "step": 4512 + }, + { + "epoch": 0.49560729189545355, + "grad_norm": 2.0968520641326904, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6792519092559814, + "num_tokens": 116258795.0, + "step": 4513 + }, + { + "epoch": 0.4957171095980672, + "grad_norm": 2.218911647796631, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7470075488090515, + "num_tokens": 116280521.0, + "step": 4514 + }, + { + "epoch": 0.4958269273006809, + "grad_norm": 2.431098461151123, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7134246230125427, + "num_tokens": 116300623.0, + "step": 4515 + }, + { + "epoch": 0.49593674500329454, + "grad_norm": 2.081798553466797, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.705769419670105, + "num_tokens": 116329573.0, + "step": 4516 + }, + { + "epoch": 0.4960465627059082, + "grad_norm": 2.3312411308288574, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.700141429901123, + "num_tokens": 116351126.0, + "step": 4517 + }, + { + "epoch": 0.49615638040852184, + "grad_norm": 2.28471302986145, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7050473093986511, + "num_tokens": 116375562.0, + "step": 4518 + }, + { + "epoch": 0.49626619811113554, + "grad_norm": 2.3249988555908203, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7032632231712341, + "num_tokens": 116397604.0, + "step": 4519 + }, + { + "epoch": 0.4963760158137492, + "grad_norm": 2.4603564739227295, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7239032983779907, + "num_tokens": 116419341.0, + "step": 4520 + }, + { + "epoch": 0.49648583351636283, + "grad_norm": 2.0447487831115723, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7054641246795654, + "num_tokens": 116448385.0, + "step": 4521 + }, + { + "epoch": 0.4965956512189765, + "grad_norm": 2.1302390098571777, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6928577423095703, + "num_tokens": 116478638.0, + "step": 4522 + }, + { + "epoch": 0.4967054689215902, + "grad_norm": 2.3252968788146973, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6994925737380981, + "num_tokens": 116502581.0, + "step": 4523 + }, + { + "epoch": 0.4968152866242038, + "grad_norm": 2.054563283920288, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7164760828018188, + "num_tokens": 116531451.0, + "step": 4524 + }, + { + "epoch": 0.49692510432681747, + "grad_norm": 2.413248062133789, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7193863391876221, + "num_tokens": 116552374.0, + "step": 4525 + }, + { + "epoch": 0.49703492202943117, + "grad_norm": 2.460930824279785, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7181657552719116, + "num_tokens": 116574782.0, + "step": 4526 + }, + { + "epoch": 0.4971447397320448, + "grad_norm": 2.1912765502929688, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7033791542053223, + "num_tokens": 116599681.0, + "step": 4527 + }, + { + "epoch": 0.49725455743465846, + "grad_norm": 2.474684476852417, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7015488147735596, + "num_tokens": 116619884.0, + "step": 4528 + }, + { + "epoch": 0.4973643751372721, + "grad_norm": 2.217219591140747, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7092211842536926, + "num_tokens": 116643485.0, + "step": 4529 + }, + { + "epoch": 0.4974741928398858, + "grad_norm": 2.374859094619751, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.702896237373352, + "num_tokens": 116667034.0, + "step": 4530 + }, + { + "epoch": 0.49758401054249946, + "grad_norm": 2.219184160232544, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7079321146011353, + "num_tokens": 116691041.0, + "step": 4531 + }, + { + "epoch": 0.4976938282451131, + "grad_norm": 2.1991233825683594, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7088648080825806, + "num_tokens": 116717067.0, + "step": 4532 + }, + { + "epoch": 0.49780364594772675, + "grad_norm": 2.1232666969299316, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7029887437820435, + "num_tokens": 116742824.0, + "step": 4533 + }, + { + "epoch": 0.49791346365034045, + "grad_norm": 2.2942452430725098, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7185102701187134, + "num_tokens": 116765762.0, + "step": 4534 + }, + { + "epoch": 0.4980232813529541, + "grad_norm": 2.309037685394287, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6968318223953247, + "num_tokens": 116787837.0, + "step": 4535 + }, + { + "epoch": 0.49813309905556774, + "grad_norm": 2.132868766784668, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7230790257453918, + "num_tokens": 116814980.0, + "step": 4536 + }, + { + "epoch": 0.49824291675818144, + "grad_norm": 2.373706102371216, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7082486748695374, + "num_tokens": 116838976.0, + "step": 4537 + }, + { + "epoch": 0.4983527344607951, + "grad_norm": 2.1949691772460938, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7011654376983643, + "num_tokens": 116864023.0, + "step": 4538 + }, + { + "epoch": 0.49846255216340873, + "grad_norm": 2.0899553298950195, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7160957455635071, + "num_tokens": 116892079.0, + "step": 4539 + }, + { + "epoch": 0.4985723698660224, + "grad_norm": 2.250265598297119, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6886167526245117, + "num_tokens": 116917681.0, + "step": 4540 + }, + { + "epoch": 0.4986821875686361, + "grad_norm": 2.167431354522705, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7090216875076294, + "num_tokens": 116942728.0, + "step": 4541 + }, + { + "epoch": 0.4987920052712497, + "grad_norm": 2.552978515625, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6985470652580261, + "num_tokens": 116962449.0, + "step": 4542 + }, + { + "epoch": 0.4989018229738634, + "grad_norm": 2.2976841926574707, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7061198949813843, + "num_tokens": 116985165.0, + "step": 4543 + }, + { + "epoch": 0.4990116406764771, + "grad_norm": 2.0429294109344482, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7127199172973633, + "num_tokens": 117011751.0, + "step": 4544 + }, + { + "epoch": 0.4991214583790907, + "grad_norm": 2.200300455093384, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6866557598114014, + "num_tokens": 117037859.0, + "step": 4545 + }, + { + "epoch": 0.49923127608170437, + "grad_norm": 2.124769449234009, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.69745934009552, + "num_tokens": 117065077.0, + "step": 4546 + }, + { + "epoch": 0.499341093784318, + "grad_norm": 2.238388776779175, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6864936351776123, + "num_tokens": 117092078.0, + "step": 4547 + }, + { + "epoch": 0.4994509114869317, + "grad_norm": 2.081200361251831, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7030055522918701, + "num_tokens": 117119882.0, + "step": 4548 + }, + { + "epoch": 0.49956072918954536, + "grad_norm": 2.198456287384033, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.692923903465271, + "num_tokens": 117144274.0, + "step": 4549 + }, + { + "epoch": 0.499670546892159, + "grad_norm": 1.9929686784744263, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7059717178344727, + "num_tokens": 117172257.0, + "step": 4550 + }, + { + "epoch": 0.49978036459477265, + "grad_norm": 1.9985084533691406, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6818846464157104, + "num_tokens": 117201919.0, + "step": 4551 + }, + { + "epoch": 0.49989018229738635, + "grad_norm": 2.299156665802002, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7055193185806274, + "num_tokens": 117224579.0, + "step": 4552 + }, + { + "epoch": 0.5, + "grad_norm": 2.0628044605255127, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.707129955291748, + "num_tokens": 117251720.0, + "step": 4553 + }, + { + "epoch": 0.5001098177026136, + "grad_norm": 2.1624486446380615, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7107616662979126, + "num_tokens": 117276484.0, + "step": 4554 + }, + { + "epoch": 0.5002196354052273, + "grad_norm": 2.099186420440674, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7120582461357117, + "num_tokens": 117302748.0, + "step": 4555 + }, + { + "epoch": 0.5003294531078409, + "grad_norm": 2.128863573074341, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6759061813354492, + "num_tokens": 117330125.0, + "step": 4556 + }, + { + "epoch": 0.5004392708104547, + "grad_norm": 2.32373309135437, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6887342929840088, + "num_tokens": 117354577.0, + "step": 4557 + }, + { + "epoch": 0.5005490885130683, + "grad_norm": 1.9458380937576294, + "learning_rate": 1e-06, + "loss": 1.1061, + "mean_token_accuracy": 0.6697925329208374, + "num_tokens": 117385813.0, + "step": 4558 + }, + { + "epoch": 0.500658906215682, + "grad_norm": 2.1883199214935303, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.704116702079773, + "num_tokens": 117411822.0, + "step": 4559 + }, + { + "epoch": 0.5007687239182956, + "grad_norm": 2.689944267272949, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7330520153045654, + "num_tokens": 117428094.0, + "step": 4560 + }, + { + "epoch": 0.5008785416209093, + "grad_norm": 2.2701053619384766, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6938050985336304, + "num_tokens": 117451216.0, + "step": 4561 + }, + { + "epoch": 0.5009883593235229, + "grad_norm": 2.4767637252807617, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7001389265060425, + "num_tokens": 117470897.0, + "step": 4562 + }, + { + "epoch": 0.5010981770261366, + "grad_norm": 2.132601737976074, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7146382331848145, + "num_tokens": 117495731.0, + "step": 4563 + }, + { + "epoch": 0.5012079947287503, + "grad_norm": 2.4230761528015137, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.7084462642669678, + "num_tokens": 117516847.0, + "step": 4564 + }, + { + "epoch": 0.501317812431364, + "grad_norm": 2.2020375728607178, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6975885629653931, + "num_tokens": 117541969.0, + "step": 4565 + }, + { + "epoch": 0.5014276301339776, + "grad_norm": 2.560748338699341, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6915414929389954, + "num_tokens": 117561910.0, + "step": 4566 + }, + { + "epoch": 0.5015374478365913, + "grad_norm": 2.474472761154175, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6917765140533447, + "num_tokens": 117584841.0, + "step": 4567 + }, + { + "epoch": 0.5016472655392049, + "grad_norm": 2.393503189086914, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7056819200515747, + "num_tokens": 117609607.0, + "step": 4568 + }, + { + "epoch": 0.5017570832418186, + "grad_norm": 2.2806177139282227, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7190632820129395, + "num_tokens": 117633399.0, + "step": 4569 + }, + { + "epoch": 0.5018669009444322, + "grad_norm": 2.3275768756866455, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6878248453140259, + "num_tokens": 117657685.0, + "step": 4570 + }, + { + "epoch": 0.5019767186470458, + "grad_norm": 2.2228925228118896, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7199559211730957, + "num_tokens": 117682688.0, + "step": 4571 + }, + { + "epoch": 0.5020865363496596, + "grad_norm": 2.172297716140747, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6872175931930542, + "num_tokens": 117710089.0, + "step": 4572 + }, + { + "epoch": 0.5021963540522733, + "grad_norm": 2.194833278656006, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6922793388366699, + "num_tokens": 117735497.0, + "step": 4573 + }, + { + "epoch": 0.5023061717548869, + "grad_norm": 2.4104082584381104, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6925488710403442, + "num_tokens": 117758054.0, + "step": 4574 + }, + { + "epoch": 0.5024159894575005, + "grad_norm": 2.168863534927368, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6978112459182739, + "num_tokens": 117785239.0, + "step": 4575 + }, + { + "epoch": 0.5025258071601142, + "grad_norm": 2.4308996200561523, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7058091759681702, + "num_tokens": 117808295.0, + "step": 4576 + }, + { + "epoch": 0.5026356248627278, + "grad_norm": 2.242331027984619, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7028864622116089, + "num_tokens": 117831555.0, + "step": 4577 + }, + { + "epoch": 0.5027454425653415, + "grad_norm": 2.5209755897521973, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7282699346542358, + "num_tokens": 117851710.0, + "step": 4578 + }, + { + "epoch": 0.5028552602679552, + "grad_norm": 2.5154967308044434, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7149975299835205, + "num_tokens": 117873583.0, + "step": 4579 + }, + { + "epoch": 0.5029650779705689, + "grad_norm": 2.273090124130249, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6984653472900391, + "num_tokens": 117896672.0, + "step": 4580 + }, + { + "epoch": 0.5030748956731825, + "grad_norm": 2.03543758392334, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7120720744132996, + "num_tokens": 117923405.0, + "step": 4581 + }, + { + "epoch": 0.5031847133757962, + "grad_norm": 2.2799041271209717, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7098941802978516, + "num_tokens": 117947153.0, + "step": 4582 + }, + { + "epoch": 0.5032945310784098, + "grad_norm": 2.2468295097351074, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6916464567184448, + "num_tokens": 117973793.0, + "step": 4583 + }, + { + "epoch": 0.5034043487810235, + "grad_norm": 2.3849947452545166, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7093081474304199, + "num_tokens": 117996260.0, + "step": 4584 + }, + { + "epoch": 0.5035141664836371, + "grad_norm": 1.8767657279968262, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7137205004692078, + "num_tokens": 118027954.0, + "step": 4585 + }, + { + "epoch": 0.5036239841862509, + "grad_norm": 1.9812541007995605, + "learning_rate": 1e-06, + "loss": 1.0798, + "mean_token_accuracy": 0.6803930401802063, + "num_tokens": 118059917.0, + "step": 4586 + }, + { + "epoch": 0.5037338018888645, + "grad_norm": 1.8881278038024902, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6952131986618042, + "num_tokens": 118091115.0, + "step": 4587 + }, + { + "epoch": 0.5038436195914782, + "grad_norm": 2.216893196105957, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.70538729429245, + "num_tokens": 118116240.0, + "step": 4588 + }, + { + "epoch": 0.5039534372940918, + "grad_norm": 2.1678733825683594, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6954472064971924, + "num_tokens": 118142617.0, + "step": 4589 + }, + { + "epoch": 0.5040632549967055, + "grad_norm": 1.831138253211975, + "learning_rate": 1e-06, + "loss": 1.1408, + "mean_token_accuracy": 0.6724689602851868, + "num_tokens": 118180436.0, + "step": 4590 + }, + { + "epoch": 0.5041730726993191, + "grad_norm": 2.011707067489624, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7022466063499451, + "num_tokens": 118210976.0, + "step": 4591 + }, + { + "epoch": 0.5042828904019327, + "grad_norm": 2.2223618030548096, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7143963575363159, + "num_tokens": 118236374.0, + "step": 4592 + }, + { + "epoch": 0.5043927081045465, + "grad_norm": 2.1310245990753174, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.7023850083351135, + "num_tokens": 118264276.0, + "step": 4593 + }, + { + "epoch": 0.5045025258071602, + "grad_norm": 2.474477767944336, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7122067809104919, + "num_tokens": 118286981.0, + "step": 4594 + }, + { + "epoch": 0.5046123435097738, + "grad_norm": 2.2204091548919678, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6931303143501282, + "num_tokens": 118313980.0, + "step": 4595 + }, + { + "epoch": 0.5047221612123874, + "grad_norm": 2.146498918533325, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7283174991607666, + "num_tokens": 118340306.0, + "step": 4596 + }, + { + "epoch": 0.5048319789150011, + "grad_norm": 2.2010862827301025, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7041659355163574, + "num_tokens": 118365905.0, + "step": 4597 + }, + { + "epoch": 0.5049417966176147, + "grad_norm": 2.2806921005249023, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7024303674697876, + "num_tokens": 118388755.0, + "step": 4598 + }, + { + "epoch": 0.5050516143202284, + "grad_norm": 2.1964993476867676, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6986638903617859, + "num_tokens": 118414720.0, + "step": 4599 + }, + { + "epoch": 0.505161432022842, + "grad_norm": 2.1309995651245117, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6989319920539856, + "num_tokens": 118442354.0, + "step": 4600 + }, + { + "epoch": 0.5052712497254558, + "grad_norm": 2.3432388305664062, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6921815872192383, + "num_tokens": 118465452.0, + "step": 4601 + }, + { + "epoch": 0.5053810674280694, + "grad_norm": 2.116185426712036, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6918795108795166, + "num_tokens": 118492099.0, + "step": 4602 + }, + { + "epoch": 0.5054908851306831, + "grad_norm": 2.1721138954162598, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6912795901298523, + "num_tokens": 118518505.0, + "step": 4603 + }, + { + "epoch": 0.5056007028332967, + "grad_norm": 3.209824800491333, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7156211733818054, + "num_tokens": 118541675.0, + "step": 4604 + }, + { + "epoch": 0.5057105205359104, + "grad_norm": 2.24399471282959, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7191280126571655, + "num_tokens": 118566941.0, + "step": 4605 + }, + { + "epoch": 0.505820338238524, + "grad_norm": 2.1901893615722656, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.688825249671936, + "num_tokens": 118594242.0, + "step": 4606 + }, + { + "epoch": 0.5059301559411377, + "grad_norm": 2.0985898971557617, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7082567811012268, + "num_tokens": 118621155.0, + "step": 4607 + }, + { + "epoch": 0.5060399736437514, + "grad_norm": 2.105569839477539, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6936749815940857, + "num_tokens": 118647630.0, + "step": 4608 + }, + { + "epoch": 0.5061497913463651, + "grad_norm": 2.2803895473480225, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6982371211051941, + "num_tokens": 118673026.0, + "step": 4609 + }, + { + "epoch": 0.5062596090489787, + "grad_norm": 2.013151168823242, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6990393996238708, + "num_tokens": 118704586.0, + "step": 4610 + }, + { + "epoch": 0.5063694267515924, + "grad_norm": 2.19677472114563, + "learning_rate": 1e-06, + "loss": 1.1115, + "mean_token_accuracy": 0.6686004400253296, + "num_tokens": 118733087.0, + "step": 4611 + }, + { + "epoch": 0.506479244454206, + "grad_norm": 1.9930193424224854, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.693252682685852, + "num_tokens": 118765599.0, + "step": 4612 + }, + { + "epoch": 0.5065890621568196, + "grad_norm": 2.0371458530426025, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7248536348342896, + "num_tokens": 118793165.0, + "step": 4613 + }, + { + "epoch": 0.5066988798594333, + "grad_norm": 2.2203619480133057, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7059586048126221, + "num_tokens": 118818637.0, + "step": 4614 + }, + { + "epoch": 0.506808697562047, + "grad_norm": 2.184030532836914, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6801550984382629, + "num_tokens": 118845749.0, + "step": 4615 + }, + { + "epoch": 0.5069185152646607, + "grad_norm": 2.2406632900238037, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6850347518920898, + "num_tokens": 118871963.0, + "step": 4616 + }, + { + "epoch": 0.5070283329672743, + "grad_norm": 1.976312279701233, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6935935020446777, + "num_tokens": 118901453.0, + "step": 4617 + }, + { + "epoch": 0.507138150669888, + "grad_norm": 2.122912883758545, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6892603635787964, + "num_tokens": 118928955.0, + "step": 4618 + }, + { + "epoch": 0.5072479683725016, + "grad_norm": 2.3342669010162354, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7131909728050232, + "num_tokens": 118954622.0, + "step": 4619 + }, + { + "epoch": 0.5073577860751153, + "grad_norm": 2.3255398273468018, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7260432243347168, + "num_tokens": 118975568.0, + "step": 4620 + }, + { + "epoch": 0.5074676037777289, + "grad_norm": 2.2510581016540527, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7036171555519104, + "num_tokens": 118999411.0, + "step": 4621 + }, + { + "epoch": 0.5075774214803427, + "grad_norm": 2.4001526832580566, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6927083730697632, + "num_tokens": 119022937.0, + "step": 4622 + }, + { + "epoch": 0.5076872391829563, + "grad_norm": 2.0434060096740723, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7088341116905212, + "num_tokens": 119050323.0, + "step": 4623 + }, + { + "epoch": 0.50779705688557, + "grad_norm": 2.192519187927246, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7067587375640869, + "num_tokens": 119075220.0, + "step": 4624 + }, + { + "epoch": 0.5079068745881836, + "grad_norm": 2.0466690063476562, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7035969495773315, + "num_tokens": 119104355.0, + "step": 4625 + }, + { + "epoch": 0.5080166922907973, + "grad_norm": 2.3390750885009766, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.691920280456543, + "num_tokens": 119128255.0, + "step": 4626 + }, + { + "epoch": 0.5081265099934109, + "grad_norm": 2.159579277038574, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6866211891174316, + "num_tokens": 119155630.0, + "step": 4627 + }, + { + "epoch": 0.5082363276960246, + "grad_norm": 2.029402256011963, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6788012981414795, + "num_tokens": 119185030.0, + "step": 4628 + }, + { + "epoch": 0.5083461453986382, + "grad_norm": 1.9886631965637207, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.709076464176178, + "num_tokens": 119215332.0, + "step": 4629 + }, + { + "epoch": 0.508455963101252, + "grad_norm": 2.1445958614349365, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7054051160812378, + "num_tokens": 119242010.0, + "step": 4630 + }, + { + "epoch": 0.5085657808038656, + "grad_norm": 2.1554617881774902, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6882895231246948, + "num_tokens": 119270247.0, + "step": 4631 + }, + { + "epoch": 0.5086755985064793, + "grad_norm": 2.172102689743042, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6943738460540771, + "num_tokens": 119297940.0, + "step": 4632 + }, + { + "epoch": 0.5087854162090929, + "grad_norm": 2.2491629123687744, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6959657669067383, + "num_tokens": 119322654.0, + "step": 4633 + }, + { + "epoch": 0.5088952339117065, + "grad_norm": 2.354001760482788, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7043835520744324, + "num_tokens": 119346596.0, + "step": 4634 + }, + { + "epoch": 0.5090050516143202, + "grad_norm": 2.2416493892669678, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.732030987739563, + "num_tokens": 119369710.0, + "step": 4635 + }, + { + "epoch": 0.5091148693169338, + "grad_norm": 2.224522352218628, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7092700600624084, + "num_tokens": 119395025.0, + "step": 4636 + }, + { + "epoch": 0.5092246870195476, + "grad_norm": 2.250476360321045, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6862140893936157, + "num_tokens": 119421152.0, + "step": 4637 + }, + { + "epoch": 0.5093345047221612, + "grad_norm": 2.033047914505005, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7153515815734863, + "num_tokens": 119448223.0, + "step": 4638 + }, + { + "epoch": 0.5094443224247749, + "grad_norm": 1.9477990865707397, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6966425180435181, + "num_tokens": 119481213.0, + "step": 4639 + }, + { + "epoch": 0.5095541401273885, + "grad_norm": 2.2978150844573975, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6902168989181519, + "num_tokens": 119506311.0, + "step": 4640 + }, + { + "epoch": 0.5096639578300022, + "grad_norm": 2.26003098487854, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.704002320766449, + "num_tokens": 119530977.0, + "step": 4641 + }, + { + "epoch": 0.5097737755326158, + "grad_norm": 2.149331569671631, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7017161846160889, + "num_tokens": 119557722.0, + "step": 4642 + }, + { + "epoch": 0.5098835932352295, + "grad_norm": 1.9702342748641968, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.716478705406189, + "num_tokens": 119588831.0, + "step": 4643 + }, + { + "epoch": 0.5099934109378432, + "grad_norm": 2.453824758529663, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7127417325973511, + "num_tokens": 119609927.0, + "step": 4644 + }, + { + "epoch": 0.5101032286404569, + "grad_norm": 1.9549373388290405, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.704422116279602, + "num_tokens": 119641358.0, + "step": 4645 + }, + { + "epoch": 0.5102130463430705, + "grad_norm": 2.525489330291748, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6856812238693237, + "num_tokens": 119663094.0, + "step": 4646 + }, + { + "epoch": 0.5103228640456842, + "grad_norm": 2.444905996322632, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7136223912239075, + "num_tokens": 119684967.0, + "step": 4647 + }, + { + "epoch": 0.5104326817482978, + "grad_norm": 2.00508975982666, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7001117467880249, + "num_tokens": 119713475.0, + "step": 4648 + }, + { + "epoch": 0.5105424994509115, + "grad_norm": 2.230933666229248, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6899127960205078, + "num_tokens": 119739021.0, + "step": 4649 + }, + { + "epoch": 0.5106523171535251, + "grad_norm": 2.0646584033966064, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7043185234069824, + "num_tokens": 119765968.0, + "step": 4650 + }, + { + "epoch": 0.5107621348561389, + "grad_norm": 2.2251522541046143, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7312514781951904, + "num_tokens": 119788688.0, + "step": 4651 + }, + { + "epoch": 0.5108719525587525, + "grad_norm": 2.4553442001342773, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6934406757354736, + "num_tokens": 119809762.0, + "step": 4652 + }, + { + "epoch": 0.5109817702613662, + "grad_norm": 2.204082727432251, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7062957286834717, + "num_tokens": 119834676.0, + "step": 4653 + }, + { + "epoch": 0.5110915879639798, + "grad_norm": 2.103902816772461, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.714982807636261, + "num_tokens": 119860527.0, + "step": 4654 + }, + { + "epoch": 0.5112014056665934, + "grad_norm": 2.081444263458252, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7226534485816956, + "num_tokens": 119886008.0, + "step": 4655 + }, + { + "epoch": 0.5113112233692071, + "grad_norm": 1.9010132551193237, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7282237410545349, + "num_tokens": 119915669.0, + "step": 4656 + }, + { + "epoch": 0.5114210410718207, + "grad_norm": 1.8940805196762085, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6989582180976868, + "num_tokens": 119948271.0, + "step": 4657 + }, + { + "epoch": 0.5115308587744345, + "grad_norm": 2.1708807945251465, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6841154098510742, + "num_tokens": 119975623.0, + "step": 4658 + }, + { + "epoch": 0.5116406764770481, + "grad_norm": 2.5901594161987305, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7041593790054321, + "num_tokens": 119995227.0, + "step": 4659 + }, + { + "epoch": 0.5117504941796618, + "grad_norm": 2.064840078353882, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6924952268600464, + "num_tokens": 120025666.0, + "step": 4660 + }, + { + "epoch": 0.5118603118822754, + "grad_norm": 1.9725998640060425, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7104315161705017, + "num_tokens": 120054805.0, + "step": 4661 + }, + { + "epoch": 0.5119701295848891, + "grad_norm": 2.167271375656128, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6933286786079407, + "num_tokens": 120083013.0, + "step": 4662 + }, + { + "epoch": 0.5120799472875027, + "grad_norm": 1.997175693511963, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.7002173066139221, + "num_tokens": 120113919.0, + "step": 4663 + }, + { + "epoch": 0.5121897649901164, + "grad_norm": 2.2774832248687744, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7160859704017639, + "num_tokens": 120136552.0, + "step": 4664 + }, + { + "epoch": 0.51229958269273, + "grad_norm": 2.3893418312072754, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7079509496688843, + "num_tokens": 120158680.0, + "step": 4665 + }, + { + "epoch": 0.5124094003953438, + "grad_norm": 1.996982216835022, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7013764381408691, + "num_tokens": 120189717.0, + "step": 4666 + }, + { + "epoch": 0.5125192180979574, + "grad_norm": 2.381256103515625, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.7091949582099915, + "num_tokens": 120213725.0, + "step": 4667 + }, + { + "epoch": 0.5126290358005711, + "grad_norm": 1.9992457628250122, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7171043157577515, + "num_tokens": 120241683.0, + "step": 4668 + }, + { + "epoch": 0.5127388535031847, + "grad_norm": 2.5350496768951416, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7344641089439392, + "num_tokens": 120259528.0, + "step": 4669 + }, + { + "epoch": 0.5128486712057984, + "grad_norm": 2.1622841358184814, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7100067138671875, + "num_tokens": 120285980.0, + "step": 4670 + }, + { + "epoch": 0.512958488908412, + "grad_norm": 2.3021304607391357, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.702451765537262, + "num_tokens": 120309002.0, + "step": 4671 + }, + { + "epoch": 0.5130683066110256, + "grad_norm": 2.532705783843994, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6958036422729492, + "num_tokens": 120330674.0, + "step": 4672 + }, + { + "epoch": 0.5131781243136394, + "grad_norm": 2.339613914489746, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7069417238235474, + "num_tokens": 120353906.0, + "step": 4673 + }, + { + "epoch": 0.513287942016253, + "grad_norm": 2.140136241912842, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6943731307983398, + "num_tokens": 120381571.0, + "step": 4674 + }, + { + "epoch": 0.5133977597188667, + "grad_norm": 2.4820735454559326, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7158752679824829, + "num_tokens": 120402263.0, + "step": 4675 + }, + { + "epoch": 0.5135075774214803, + "grad_norm": 2.4163739681243896, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7098244428634644, + "num_tokens": 120423605.0, + "step": 4676 + }, + { + "epoch": 0.513617395124094, + "grad_norm": 1.9329712390899658, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7038552761077881, + "num_tokens": 120455124.0, + "step": 4677 + }, + { + "epoch": 0.5137272128267076, + "grad_norm": 2.288862466812134, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7295747995376587, + "num_tokens": 120478556.0, + "step": 4678 + }, + { + "epoch": 0.5138370305293213, + "grad_norm": 2.7618372440338135, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7022637128829956, + "num_tokens": 120497571.0, + "step": 4679 + }, + { + "epoch": 0.513946848231935, + "grad_norm": 2.675306797027588, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.713585376739502, + "num_tokens": 120516894.0, + "step": 4680 + }, + { + "epoch": 0.5140566659345487, + "grad_norm": 2.512955904006958, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7218381762504578, + "num_tokens": 120536057.0, + "step": 4681 + }, + { + "epoch": 0.5141664836371623, + "grad_norm": 2.446671962738037, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7105898261070251, + "num_tokens": 120557022.0, + "step": 4682 + }, + { + "epoch": 0.514276301339776, + "grad_norm": 2.105447769165039, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.7070869207382202, + "num_tokens": 120584336.0, + "step": 4683 + }, + { + "epoch": 0.5143861190423896, + "grad_norm": 2.1621532440185547, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7166469097137451, + "num_tokens": 120610505.0, + "step": 4684 + }, + { + "epoch": 0.5144959367450033, + "grad_norm": 2.152198553085327, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7296010255813599, + "num_tokens": 120636614.0, + "step": 4685 + }, + { + "epoch": 0.5146057544476169, + "grad_norm": 2.0331573486328125, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7085576057434082, + "num_tokens": 120665238.0, + "step": 4686 + }, + { + "epoch": 0.5147155721502307, + "grad_norm": 2.2861671447753906, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6964526176452637, + "num_tokens": 120688825.0, + "step": 4687 + }, + { + "epoch": 0.5148253898528443, + "grad_norm": 2.571239948272705, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6928005814552307, + "num_tokens": 120710383.0, + "step": 4688 + }, + { + "epoch": 0.514935207555458, + "grad_norm": 2.177769184112549, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.692929744720459, + "num_tokens": 120735666.0, + "step": 4689 + }, + { + "epoch": 0.5150450252580716, + "grad_norm": 2.115403175354004, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7055649161338806, + "num_tokens": 120762519.0, + "step": 4690 + }, + { + "epoch": 0.5151548429606853, + "grad_norm": 2.0980682373046875, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7113120555877686, + "num_tokens": 120789899.0, + "step": 4691 + }, + { + "epoch": 0.5152646606632989, + "grad_norm": 1.9968889951705933, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6759376525878906, + "num_tokens": 120819991.0, + "step": 4692 + }, + { + "epoch": 0.5153744783659125, + "grad_norm": 2.3299906253814697, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6833145618438721, + "num_tokens": 120845239.0, + "step": 4693 + }, + { + "epoch": 0.5154842960685262, + "grad_norm": 2.514091730117798, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.721822202205658, + "num_tokens": 120864593.0, + "step": 4694 + }, + { + "epoch": 0.51559411377114, + "grad_norm": 2.417369842529297, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6961215734481812, + "num_tokens": 120888306.0, + "step": 4695 + }, + { + "epoch": 0.5157039314737536, + "grad_norm": 2.2692112922668457, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7262253761291504, + "num_tokens": 120911235.0, + "step": 4696 + }, + { + "epoch": 0.5158137491763672, + "grad_norm": 2.207395076751709, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7217382788658142, + "num_tokens": 120936460.0, + "step": 4697 + }, + { + "epoch": 0.5159235668789809, + "grad_norm": 2.257784366607666, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7340735197067261, + "num_tokens": 120960452.0, + "step": 4698 + }, + { + "epoch": 0.5160333845815945, + "grad_norm": 1.9878778457641602, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6910632252693176, + "num_tokens": 120993193.0, + "step": 4699 + }, + { + "epoch": 0.5161432022842082, + "grad_norm": 2.119993209838867, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7205818891525269, + "num_tokens": 121019505.0, + "step": 4700 + }, + { + "epoch": 0.5162530199868218, + "grad_norm": 2.197683572769165, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7009379863739014, + "num_tokens": 121045236.0, + "step": 4701 + }, + { + "epoch": 0.5163628376894356, + "grad_norm": 2.1214261054992676, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.703550398349762, + "num_tokens": 121072613.0, + "step": 4702 + }, + { + "epoch": 0.5164726553920492, + "grad_norm": 2.0766892433166504, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7123014330863953, + "num_tokens": 121100654.0, + "step": 4703 + }, + { + "epoch": 0.5165824730946629, + "grad_norm": 1.8438726663589478, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6967902183532715, + "num_tokens": 121134319.0, + "step": 4704 + }, + { + "epoch": 0.5166922907972765, + "grad_norm": 2.1840627193450928, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7137500047683716, + "num_tokens": 121158808.0, + "step": 4705 + }, + { + "epoch": 0.5168021084998902, + "grad_norm": 2.192091464996338, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7092403173446655, + "num_tokens": 121184981.0, + "step": 4706 + }, + { + "epoch": 0.5169119262025038, + "grad_norm": 2.3429458141326904, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7081811428070068, + "num_tokens": 121207248.0, + "step": 4707 + }, + { + "epoch": 0.5170217439051175, + "grad_norm": 2.6557741165161133, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7148574590682983, + "num_tokens": 121228805.0, + "step": 4708 + }, + { + "epoch": 0.5171315616077312, + "grad_norm": 2.3322887420654297, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7204310297966003, + "num_tokens": 121250496.0, + "step": 4709 + }, + { + "epoch": 0.5172413793103449, + "grad_norm": 2.4722912311553955, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7268868684768677, + "num_tokens": 121269977.0, + "step": 4710 + }, + { + "epoch": 0.5173511970129585, + "grad_norm": 2.1162455081939697, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7232846021652222, + "num_tokens": 121297010.0, + "step": 4711 + }, + { + "epoch": 0.5174610147155722, + "grad_norm": 2.4175784587860107, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.701748788356781, + "num_tokens": 121318815.0, + "step": 4712 + }, + { + "epoch": 0.5175708324181858, + "grad_norm": 2.3972601890563965, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7183487415313721, + "num_tokens": 121339611.0, + "step": 4713 + }, + { + "epoch": 0.5176806501207994, + "grad_norm": 2.1444690227508545, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6931968927383423, + "num_tokens": 121364719.0, + "step": 4714 + }, + { + "epoch": 0.5177904678234131, + "grad_norm": 2.176218032836914, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7017039060592651, + "num_tokens": 121390079.0, + "step": 4715 + }, + { + "epoch": 0.5179002855260268, + "grad_norm": 2.260993242263794, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6988748908042908, + "num_tokens": 121415243.0, + "step": 4716 + }, + { + "epoch": 0.5180101032286405, + "grad_norm": 2.1073570251464844, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6906256675720215, + "num_tokens": 121443860.0, + "step": 4717 + }, + { + "epoch": 0.5181199209312541, + "grad_norm": 2.5416929721832275, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7092311978340149, + "num_tokens": 121465165.0, + "step": 4718 + }, + { + "epoch": 0.5182297386338678, + "grad_norm": 2.041963815689087, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7028013467788696, + "num_tokens": 121495825.0, + "step": 4719 + }, + { + "epoch": 0.5183395563364814, + "grad_norm": 2.079291582107544, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7051233649253845, + "num_tokens": 121522566.0, + "step": 4720 + }, + { + "epoch": 0.5184493740390951, + "grad_norm": 1.8339797258377075, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7041808366775513, + "num_tokens": 121554353.0, + "step": 4721 + }, + { + "epoch": 0.5185591917417087, + "grad_norm": 2.003450393676758, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7188318967819214, + "num_tokens": 121584196.0, + "step": 4722 + }, + { + "epoch": 0.5186690094443224, + "grad_norm": 2.162177324295044, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7008424997329712, + "num_tokens": 121610935.0, + "step": 4723 + }, + { + "epoch": 0.5187788271469361, + "grad_norm": 2.1415224075317383, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6953566074371338, + "num_tokens": 121636990.0, + "step": 4724 + }, + { + "epoch": 0.5188886448495498, + "grad_norm": 2.048039674758911, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6915228962898254, + "num_tokens": 121666640.0, + "step": 4725 + }, + { + "epoch": 0.5189984625521634, + "grad_norm": 2.3512425422668457, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7026817798614502, + "num_tokens": 121689242.0, + "step": 4726 + }, + { + "epoch": 0.5191082802547771, + "grad_norm": 2.457707405090332, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7195405960083008, + "num_tokens": 121709837.0, + "step": 4727 + }, + { + "epoch": 0.5192180979573907, + "grad_norm": 2.1011059284210205, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6977953910827637, + "num_tokens": 121738722.0, + "step": 4728 + }, + { + "epoch": 0.5193279156600044, + "grad_norm": 2.1104469299316406, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6929928064346313, + "num_tokens": 121766725.0, + "step": 4729 + }, + { + "epoch": 0.519437733362618, + "grad_norm": 2.434603691101074, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7179755568504333, + "num_tokens": 121787791.0, + "step": 4730 + }, + { + "epoch": 0.5195475510652318, + "grad_norm": 2.189922332763672, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7297686338424683, + "num_tokens": 121812194.0, + "step": 4731 + }, + { + "epoch": 0.5196573687678454, + "grad_norm": 2.1567413806915283, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7147911787033081, + "num_tokens": 121839073.0, + "step": 4732 + }, + { + "epoch": 0.519767186470459, + "grad_norm": 2.0575361251831055, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6948263049125671, + "num_tokens": 121865308.0, + "step": 4733 + }, + { + "epoch": 0.5198770041730727, + "grad_norm": 2.1955833435058594, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7268013954162598, + "num_tokens": 121891750.0, + "step": 4734 + }, + { + "epoch": 0.5199868218756863, + "grad_norm": 2.3053500652313232, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6688778400421143, + "num_tokens": 121914739.0, + "step": 4735 + }, + { + "epoch": 0.5200966395783, + "grad_norm": 2.1281447410583496, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6889081597328186, + "num_tokens": 121943479.0, + "step": 4736 + }, + { + "epoch": 0.5202064572809136, + "grad_norm": 2.21781325340271, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7166916131973267, + "num_tokens": 121967141.0, + "step": 4737 + }, + { + "epoch": 0.5203162749835274, + "grad_norm": 2.147716760635376, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.714835524559021, + "num_tokens": 121993210.0, + "step": 4738 + }, + { + "epoch": 0.520426092686141, + "grad_norm": 2.1667256355285645, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7001272439956665, + "num_tokens": 122019996.0, + "step": 4739 + }, + { + "epoch": 0.5205359103887547, + "grad_norm": 2.2944962978363037, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7039180397987366, + "num_tokens": 122044643.0, + "step": 4740 + }, + { + "epoch": 0.5206457280913683, + "grad_norm": 2.1066458225250244, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7188462615013123, + "num_tokens": 122070081.0, + "step": 4741 + }, + { + "epoch": 0.520755545793982, + "grad_norm": 2.2750890254974365, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7122130393981934, + "num_tokens": 122092671.0, + "step": 4742 + }, + { + "epoch": 0.5208653634965956, + "grad_norm": 2.0176398754119873, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7182748913764954, + "num_tokens": 122123225.0, + "step": 4743 + }, + { + "epoch": 0.5209751811992093, + "grad_norm": 2.352210760116577, + "learning_rate": 1e-06, + "loss": 1.0888, + "mean_token_accuracy": 0.6764141321182251, + "num_tokens": 122147871.0, + "step": 4744 + }, + { + "epoch": 0.521084998901823, + "grad_norm": 2.17512583732605, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7128371000289917, + "num_tokens": 122173305.0, + "step": 4745 + }, + { + "epoch": 0.5211948166044367, + "grad_norm": 2.271421194076538, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6875386238098145, + "num_tokens": 122197816.0, + "step": 4746 + }, + { + "epoch": 0.5213046343070503, + "grad_norm": 2.288191080093384, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7165178060531616, + "num_tokens": 122220916.0, + "step": 4747 + }, + { + "epoch": 0.521414452009664, + "grad_norm": 2.3365869522094727, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6995052099227905, + "num_tokens": 122243604.0, + "step": 4748 + }, + { + "epoch": 0.5215242697122776, + "grad_norm": 1.8671295642852783, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.691491961479187, + "num_tokens": 122278297.0, + "step": 4749 + }, + { + "epoch": 0.5216340874148913, + "grad_norm": 2.450091600418091, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7054259181022644, + "num_tokens": 122298234.0, + "step": 4750 + }, + { + "epoch": 0.5217439051175049, + "grad_norm": 2.2899413108825684, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.713905930519104, + "num_tokens": 122321702.0, + "step": 4751 + }, + { + "epoch": 0.5218537228201185, + "grad_norm": 2.3024797439575195, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7184453010559082, + "num_tokens": 122345677.0, + "step": 4752 + }, + { + "epoch": 0.5219635405227323, + "grad_norm": 2.1029953956604004, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6892531514167786, + "num_tokens": 122373897.0, + "step": 4753 + }, + { + "epoch": 0.522073358225346, + "grad_norm": 2.04289174079895, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7040168642997742, + "num_tokens": 122404115.0, + "step": 4754 + }, + { + "epoch": 0.5221831759279596, + "grad_norm": 1.9650733470916748, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6894879341125488, + "num_tokens": 122436729.0, + "step": 4755 + }, + { + "epoch": 0.5222929936305732, + "grad_norm": 2.2470908164978027, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7012484669685364, + "num_tokens": 122461520.0, + "step": 4756 + }, + { + "epoch": 0.5224028113331869, + "grad_norm": 2.08735990524292, + "learning_rate": 1e-06, + "loss": 1.083, + "mean_token_accuracy": 0.6735889911651611, + "num_tokens": 122491491.0, + "step": 4757 + }, + { + "epoch": 0.5225126290358005, + "grad_norm": 2.054244041442871, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6935051679611206, + "num_tokens": 122518766.0, + "step": 4758 + }, + { + "epoch": 0.5226224467384142, + "grad_norm": 1.9607027769088745, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6814999580383301, + "num_tokens": 122553201.0, + "step": 4759 + }, + { + "epoch": 0.5227322644410279, + "grad_norm": 2.0900940895080566, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7282199859619141, + "num_tokens": 122579004.0, + "step": 4760 + }, + { + "epoch": 0.5228420821436416, + "grad_norm": 1.9735267162322998, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7115744948387146, + "num_tokens": 122608271.0, + "step": 4761 + }, + { + "epoch": 0.5229518998462552, + "grad_norm": 2.253079414367676, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.720514714717865, + "num_tokens": 122632595.0, + "step": 4762 + }, + { + "epoch": 0.5230617175488689, + "grad_norm": 2.182319164276123, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7321333885192871, + "num_tokens": 122655691.0, + "step": 4763 + }, + { + "epoch": 0.5231715352514825, + "grad_norm": 2.390932321548462, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7165411710739136, + "num_tokens": 122678144.0, + "step": 4764 + }, + { + "epoch": 0.5232813529540962, + "grad_norm": 2.3752024173736572, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7099236249923706, + "num_tokens": 122698731.0, + "step": 4765 + }, + { + "epoch": 0.5233911706567098, + "grad_norm": 2.203115940093994, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6981661319732666, + "num_tokens": 122724913.0, + "step": 4766 + }, + { + "epoch": 0.5235009883593236, + "grad_norm": 2.0798425674438477, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.700835108757019, + "num_tokens": 122753057.0, + "step": 4767 + }, + { + "epoch": 0.5236108060619372, + "grad_norm": 2.5238149166107178, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7014288306236267, + "num_tokens": 122773081.0, + "step": 4768 + }, + { + "epoch": 0.5237206237645509, + "grad_norm": 2.2296390533447266, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7061303853988647, + "num_tokens": 122799994.0, + "step": 4769 + }, + { + "epoch": 0.5238304414671645, + "grad_norm": 1.982675313949585, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7103282809257507, + "num_tokens": 122829669.0, + "step": 4770 + }, + { + "epoch": 0.5239402591697782, + "grad_norm": 2.2241315841674805, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7357409000396729, + "num_tokens": 122852938.0, + "step": 4771 + }, + { + "epoch": 0.5240500768723918, + "grad_norm": 1.8904303312301636, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7259576320648193, + "num_tokens": 122883820.0, + "step": 4772 + }, + { + "epoch": 0.5241598945750054, + "grad_norm": 2.228193759918213, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7206874489784241, + "num_tokens": 122905814.0, + "step": 4773 + }, + { + "epoch": 0.5242697122776192, + "grad_norm": 2.165605068206787, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7086709141731262, + "num_tokens": 122930957.0, + "step": 4774 + }, + { + "epoch": 0.5243795299802329, + "grad_norm": 2.2283871173858643, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7007334232330322, + "num_tokens": 122958880.0, + "step": 4775 + }, + { + "epoch": 0.5244893476828465, + "grad_norm": 2.020725727081299, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6943129897117615, + "num_tokens": 122987089.0, + "step": 4776 + }, + { + "epoch": 0.5245991653854601, + "grad_norm": 2.326241970062256, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7347224354743958, + "num_tokens": 123008614.0, + "step": 4777 + }, + { + "epoch": 0.5247089830880738, + "grad_norm": 2.2281553745269775, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7404126524925232, + "num_tokens": 123032657.0, + "step": 4778 + }, + { + "epoch": 0.5248188007906874, + "grad_norm": 2.7088632583618164, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6958478689193726, + "num_tokens": 123052315.0, + "step": 4779 + }, + { + "epoch": 0.5249286184933011, + "grad_norm": 2.114910364151001, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7075121402740479, + "num_tokens": 123079896.0, + "step": 4780 + }, + { + "epoch": 0.5250384361959147, + "grad_norm": 2.158747434616089, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6948540210723877, + "num_tokens": 123107709.0, + "step": 4781 + }, + { + "epoch": 0.5251482538985285, + "grad_norm": 2.2955007553100586, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6991687417030334, + "num_tokens": 123132228.0, + "step": 4782 + }, + { + "epoch": 0.5252580716011421, + "grad_norm": 2.185704231262207, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6965847611427307, + "num_tokens": 123159951.0, + "step": 4783 + }, + { + "epoch": 0.5253678893037558, + "grad_norm": 2.0689308643341064, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6817616820335388, + "num_tokens": 123191960.0, + "step": 4784 + }, + { + "epoch": 0.5254777070063694, + "grad_norm": 2.419705390930176, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7210625410079956, + "num_tokens": 123211687.0, + "step": 4785 + }, + { + "epoch": 0.5255875247089831, + "grad_norm": 2.3262522220611572, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.695849597454071, + "num_tokens": 123235353.0, + "step": 4786 + }, + { + "epoch": 0.5256973424115967, + "grad_norm": 2.2780089378356934, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6951714754104614, + "num_tokens": 123257848.0, + "step": 4787 + }, + { + "epoch": 0.5258071601142104, + "grad_norm": 2.2425546646118164, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7108871936798096, + "num_tokens": 123282110.0, + "step": 4788 + }, + { + "epoch": 0.5259169778168241, + "grad_norm": 2.2331247329711914, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7243551015853882, + "num_tokens": 123304820.0, + "step": 4789 + }, + { + "epoch": 0.5260267955194378, + "grad_norm": 2.3941256999969482, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7033624053001404, + "num_tokens": 123326579.0, + "step": 4790 + }, + { + "epoch": 0.5261366132220514, + "grad_norm": 2.3360824584960938, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7055765986442566, + "num_tokens": 123351878.0, + "step": 4791 + }, + { + "epoch": 0.526246430924665, + "grad_norm": 2.3416237831115723, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6973972916603088, + "num_tokens": 123376614.0, + "step": 4792 + }, + { + "epoch": 0.5263562486272787, + "grad_norm": 2.356840133666992, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7084507942199707, + "num_tokens": 123399007.0, + "step": 4793 + }, + { + "epoch": 0.5264660663298923, + "grad_norm": 2.15627384185791, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6975095868110657, + "num_tokens": 123425230.0, + "step": 4794 + }, + { + "epoch": 0.526575884032506, + "grad_norm": 2.0158298015594482, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7001217007637024, + "num_tokens": 123452895.0, + "step": 4795 + }, + { + "epoch": 0.5266857017351197, + "grad_norm": 2.1498425006866455, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6812454462051392, + "num_tokens": 123479008.0, + "step": 4796 + }, + { + "epoch": 0.5267955194377334, + "grad_norm": 2.181114673614502, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7170491814613342, + "num_tokens": 123502618.0, + "step": 4797 + }, + { + "epoch": 0.526905337140347, + "grad_norm": 2.263517379760742, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7184287905693054, + "num_tokens": 123525182.0, + "step": 4798 + }, + { + "epoch": 0.5270151548429607, + "grad_norm": 2.2412071228027344, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.699826180934906, + "num_tokens": 123552187.0, + "step": 4799 + }, + { + "epoch": 0.5271249725455743, + "grad_norm": 2.4180662631988525, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7006481885910034, + "num_tokens": 123574637.0, + "step": 4800 + }, + { + "epoch": 0.527234790248188, + "grad_norm": 2.096783399581909, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6928340196609497, + "num_tokens": 123599750.0, + "step": 4801 + }, + { + "epoch": 0.5273446079508016, + "grad_norm": 2.3858330249786377, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.7036648988723755, + "num_tokens": 123622922.0, + "step": 4802 + }, + { + "epoch": 0.5274544256534154, + "grad_norm": 2.3048477172851562, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6962543725967407, + "num_tokens": 123646564.0, + "step": 4803 + }, + { + "epoch": 0.527564243356029, + "grad_norm": 2.1348466873168945, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6779134273529053, + "num_tokens": 123673541.0, + "step": 4804 + }, + { + "epoch": 0.5276740610586427, + "grad_norm": 2.1641087532043457, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7170206308364868, + "num_tokens": 123699348.0, + "step": 4805 + }, + { + "epoch": 0.5277838787612563, + "grad_norm": 2.242908477783203, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7182968258857727, + "num_tokens": 123723424.0, + "step": 4806 + }, + { + "epoch": 0.52789369646387, + "grad_norm": 2.1565747261047363, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7110270261764526, + "num_tokens": 123754081.0, + "step": 4807 + }, + { + "epoch": 0.5280035141664836, + "grad_norm": 2.392151355743408, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.701561450958252, + "num_tokens": 123776049.0, + "step": 4808 + }, + { + "epoch": 0.5281133318690973, + "grad_norm": 1.9810105562210083, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6984021663665771, + "num_tokens": 123805223.0, + "step": 4809 + }, + { + "epoch": 0.528223149571711, + "grad_norm": 2.0969905853271484, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7085853815078735, + "num_tokens": 123833584.0, + "step": 4810 + }, + { + "epoch": 0.5283329672743247, + "grad_norm": 2.1306135654449463, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6943986415863037, + "num_tokens": 123860944.0, + "step": 4811 + }, + { + "epoch": 0.5284427849769383, + "grad_norm": 2.263467311859131, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7130670547485352, + "num_tokens": 123886598.0, + "step": 4812 + }, + { + "epoch": 0.528552602679552, + "grad_norm": 2.1890439987182617, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7026095986366272, + "num_tokens": 123913792.0, + "step": 4813 + }, + { + "epoch": 0.5286624203821656, + "grad_norm": 2.4597980976104736, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7085175514221191, + "num_tokens": 123934901.0, + "step": 4814 + }, + { + "epoch": 0.5287722380847792, + "grad_norm": 2.0008578300476074, + "learning_rate": 1e-06, + "loss": 1.0806, + "mean_token_accuracy": 0.6819159388542175, + "num_tokens": 123964115.0, + "step": 4815 + }, + { + "epoch": 0.5288820557873929, + "grad_norm": 2.1823863983154297, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6946216821670532, + "num_tokens": 123990575.0, + "step": 4816 + }, + { + "epoch": 0.5289918734900065, + "grad_norm": 2.2869458198547363, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7066109776496887, + "num_tokens": 124012887.0, + "step": 4817 + }, + { + "epoch": 0.5291016911926203, + "grad_norm": 1.9098774194717407, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6769036054611206, + "num_tokens": 124045380.0, + "step": 4818 + }, + { + "epoch": 0.5292115088952339, + "grad_norm": 2.2548398971557617, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6950996518135071, + "num_tokens": 124072934.0, + "step": 4819 + }, + { + "epoch": 0.5293213265978476, + "grad_norm": 2.285494804382324, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.702332615852356, + "num_tokens": 124098626.0, + "step": 4820 + }, + { + "epoch": 0.5294311443004612, + "grad_norm": 2.150088310241699, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6929053068161011, + "num_tokens": 124126215.0, + "step": 4821 + }, + { + "epoch": 0.5295409620030749, + "grad_norm": 1.9609591960906982, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6978332996368408, + "num_tokens": 124156333.0, + "step": 4822 + }, + { + "epoch": 0.5296507797056885, + "grad_norm": 2.1031713485717773, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7018284201622009, + "num_tokens": 124184787.0, + "step": 4823 + }, + { + "epoch": 0.5297605974083022, + "grad_norm": 2.149684190750122, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7145718336105347, + "num_tokens": 124211275.0, + "step": 4824 + }, + { + "epoch": 0.5298704151109159, + "grad_norm": 2.1872289180755615, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7043654918670654, + "num_tokens": 124237666.0, + "step": 4825 + }, + { + "epoch": 0.5299802328135296, + "grad_norm": 2.2500362396240234, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7126886248588562, + "num_tokens": 124261637.0, + "step": 4826 + }, + { + "epoch": 0.5300900505161432, + "grad_norm": 2.439164400100708, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.716747522354126, + "num_tokens": 124285112.0, + "step": 4827 + }, + { + "epoch": 0.5301998682187569, + "grad_norm": 2.1626486778259277, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7018017768859863, + "num_tokens": 124310149.0, + "step": 4828 + }, + { + "epoch": 0.5303096859213705, + "grad_norm": 2.2013049125671387, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7098098993301392, + "num_tokens": 124334406.0, + "step": 4829 + }, + { + "epoch": 0.5304195036239842, + "grad_norm": 2.071958065032959, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7002304792404175, + "num_tokens": 124365026.0, + "step": 4830 + }, + { + "epoch": 0.5305293213265978, + "grad_norm": 2.0837769508361816, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7246644496917725, + "num_tokens": 124391977.0, + "step": 4831 + }, + { + "epoch": 0.5306391390292116, + "grad_norm": 2.2197418212890625, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6964079141616821, + "num_tokens": 124416860.0, + "step": 4832 + }, + { + "epoch": 0.5307489567318252, + "grad_norm": 2.635209083557129, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.697769820690155, + "num_tokens": 124437484.0, + "step": 4833 + }, + { + "epoch": 0.5308587744344389, + "grad_norm": 2.296701431274414, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7376742362976074, + "num_tokens": 124460205.0, + "step": 4834 + }, + { + "epoch": 0.5309685921370525, + "grad_norm": 2.3355836868286133, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6980297565460205, + "num_tokens": 124485968.0, + "step": 4835 + }, + { + "epoch": 0.5310784098396661, + "grad_norm": 2.3043389320373535, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6948368549346924, + "num_tokens": 124509521.0, + "step": 4836 + }, + { + "epoch": 0.5311882275422798, + "grad_norm": 2.2178900241851807, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7189850807189941, + "num_tokens": 124534382.0, + "step": 4837 + }, + { + "epoch": 0.5312980452448934, + "grad_norm": 2.154815196990967, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7132028937339783, + "num_tokens": 124560008.0, + "step": 4838 + }, + { + "epoch": 0.5314078629475072, + "grad_norm": 2.197622537612915, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6921634078025818, + "num_tokens": 124586155.0, + "step": 4839 + }, + { + "epoch": 0.5315176806501208, + "grad_norm": 2.3879029750823975, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7178835868835449, + "num_tokens": 124606279.0, + "step": 4840 + }, + { + "epoch": 0.5316274983527345, + "grad_norm": 1.9867808818817139, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6819816827774048, + "num_tokens": 124636569.0, + "step": 4841 + }, + { + "epoch": 0.5317373160553481, + "grad_norm": 2.306863784790039, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7016605138778687, + "num_tokens": 124659577.0, + "step": 4842 + }, + { + "epoch": 0.5318471337579618, + "grad_norm": 2.1217193603515625, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7220170497894287, + "num_tokens": 124683502.0, + "step": 4843 + }, + { + "epoch": 0.5319569514605754, + "grad_norm": 2.440958023071289, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.70976722240448, + "num_tokens": 124704139.0, + "step": 4844 + }, + { + "epoch": 0.5320667691631891, + "grad_norm": 2.2570085525512695, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6796772480010986, + "num_tokens": 124729778.0, + "step": 4845 + }, + { + "epoch": 0.5321765868658027, + "grad_norm": 2.2785122394561768, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6933121085166931, + "num_tokens": 124753299.0, + "step": 4846 + }, + { + "epoch": 0.5322864045684165, + "grad_norm": 2.195805549621582, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.707202136516571, + "num_tokens": 124776437.0, + "step": 4847 + }, + { + "epoch": 0.5323962222710301, + "grad_norm": 1.91879403591156, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6992335319519043, + "num_tokens": 124806468.0, + "step": 4848 + }, + { + "epoch": 0.5325060399736438, + "grad_norm": 2.245918035507202, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7223671674728394, + "num_tokens": 124831359.0, + "step": 4849 + }, + { + "epoch": 0.5326158576762574, + "grad_norm": 2.1157312393188477, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6851789951324463, + "num_tokens": 124859276.0, + "step": 4850 + }, + { + "epoch": 0.532725675378871, + "grad_norm": 2.168124198913574, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.709793746471405, + "num_tokens": 124886721.0, + "step": 4851 + }, + { + "epoch": 0.5328354930814847, + "grad_norm": 2.293686628341675, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7028213739395142, + "num_tokens": 124908562.0, + "step": 4852 + }, + { + "epoch": 0.5329453107840983, + "grad_norm": 2.1431844234466553, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6825443506240845, + "num_tokens": 124937540.0, + "step": 4853 + }, + { + "epoch": 0.5330551284867121, + "grad_norm": 2.687992572784424, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7030353546142578, + "num_tokens": 124956848.0, + "step": 4854 + }, + { + "epoch": 0.5331649461893258, + "grad_norm": 2.118661642074585, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7002267241477966, + "num_tokens": 124985503.0, + "step": 4855 + }, + { + "epoch": 0.5332747638919394, + "grad_norm": 2.0584206581115723, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7053351402282715, + "num_tokens": 125012474.0, + "step": 4856 + }, + { + "epoch": 0.533384581594553, + "grad_norm": 2.106642723083496, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7066439986228943, + "num_tokens": 125038720.0, + "step": 4857 + }, + { + "epoch": 0.5334943992971667, + "grad_norm": 2.281796932220459, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7199634313583374, + "num_tokens": 125063538.0, + "step": 4858 + }, + { + "epoch": 0.5336042169997803, + "grad_norm": 2.0134594440460205, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7017488479614258, + "num_tokens": 125091291.0, + "step": 4859 + }, + { + "epoch": 0.533714034702394, + "grad_norm": 2.0092649459838867, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.699411153793335, + "num_tokens": 125122020.0, + "step": 4860 + }, + { + "epoch": 0.5338238524050077, + "grad_norm": 2.0680997371673584, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.696872889995575, + "num_tokens": 125151032.0, + "step": 4861 + }, + { + "epoch": 0.5339336701076214, + "grad_norm": 2.215299606323242, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7050895690917969, + "num_tokens": 125176611.0, + "step": 4862 + }, + { + "epoch": 0.534043487810235, + "grad_norm": 1.8031713962554932, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6953802108764648, + "num_tokens": 125213639.0, + "step": 4863 + }, + { + "epoch": 0.5341533055128487, + "grad_norm": 2.0030555725097656, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.7017480731010437, + "num_tokens": 125241203.0, + "step": 4864 + }, + { + "epoch": 0.5342631232154623, + "grad_norm": 2.1213088035583496, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6969255208969116, + "num_tokens": 125268666.0, + "step": 4865 + }, + { + "epoch": 0.534372940918076, + "grad_norm": 2.5364813804626465, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7071068286895752, + "num_tokens": 125289559.0, + "step": 4866 + }, + { + "epoch": 0.5344827586206896, + "grad_norm": 2.4941232204437256, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7046910524368286, + "num_tokens": 125313150.0, + "step": 4867 + }, + { + "epoch": 0.5345925763233034, + "grad_norm": 2.5381431579589844, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7085334062576294, + "num_tokens": 125334513.0, + "step": 4868 + }, + { + "epoch": 0.534702394025917, + "grad_norm": 2.1037755012512207, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.721164882183075, + "num_tokens": 125359812.0, + "step": 4869 + }, + { + "epoch": 0.5348122117285307, + "grad_norm": 2.219587802886963, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7054442167282104, + "num_tokens": 125385152.0, + "step": 4870 + }, + { + "epoch": 0.5349220294311443, + "grad_norm": 2.226886510848999, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6925617456436157, + "num_tokens": 125412707.0, + "step": 4871 + }, + { + "epoch": 0.535031847133758, + "grad_norm": 2.2133309841156006, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7219170331954956, + "num_tokens": 125434940.0, + "step": 4872 + }, + { + "epoch": 0.5351416648363716, + "grad_norm": 2.1077444553375244, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6770684719085693, + "num_tokens": 125465714.0, + "step": 4873 + }, + { + "epoch": 0.5352514825389852, + "grad_norm": 2.158493757247925, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7064847350120544, + "num_tokens": 125490974.0, + "step": 4874 + }, + { + "epoch": 0.5353613002415989, + "grad_norm": 2.699504852294922, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.703730583190918, + "num_tokens": 125510417.0, + "step": 4875 + }, + { + "epoch": 0.5354711179442126, + "grad_norm": 2.758629083633423, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7159115076065063, + "num_tokens": 125529198.0, + "step": 4876 + }, + { + "epoch": 0.5355809356468263, + "grad_norm": 2.279203176498413, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6860348582267761, + "num_tokens": 125553738.0, + "step": 4877 + }, + { + "epoch": 0.5356907533494399, + "grad_norm": 2.081148147583008, + "learning_rate": 1e-06, + "loss": 1.0894, + "mean_token_accuracy": 0.674263060092926, + "num_tokens": 125585911.0, + "step": 4878 + }, + { + "epoch": 0.5358005710520536, + "grad_norm": 2.0455892086029053, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6945351958274841, + "num_tokens": 125613759.0, + "step": 4879 + }, + { + "epoch": 0.5359103887546672, + "grad_norm": 2.201059579849243, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6832401156425476, + "num_tokens": 125639774.0, + "step": 4880 + }, + { + "epoch": 0.5360202064572809, + "grad_norm": 2.078646421432495, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7083150148391724, + "num_tokens": 125665957.0, + "step": 4881 + }, + { + "epoch": 0.5361300241598945, + "grad_norm": 2.392791271209717, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7046812772750854, + "num_tokens": 125687042.0, + "step": 4882 + }, + { + "epoch": 0.5362398418625083, + "grad_norm": 2.3337626457214355, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7293387055397034, + "num_tokens": 125710217.0, + "step": 4883 + }, + { + "epoch": 0.5363496595651219, + "grad_norm": 2.384050130844116, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7200980186462402, + "num_tokens": 125731386.0, + "step": 4884 + }, + { + "epoch": 0.5364594772677356, + "grad_norm": 2.40334153175354, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.706987738609314, + "num_tokens": 125754304.0, + "step": 4885 + }, + { + "epoch": 0.5365692949703492, + "grad_norm": 2.0614025592803955, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7095913290977478, + "num_tokens": 125780438.0, + "step": 4886 + }, + { + "epoch": 0.5366791126729629, + "grad_norm": 2.3098745346069336, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7034531831741333, + "num_tokens": 125804307.0, + "step": 4887 + }, + { + "epoch": 0.5367889303755765, + "grad_norm": 2.050352096557617, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7098209261894226, + "num_tokens": 125835306.0, + "step": 4888 + }, + { + "epoch": 0.5368987480781902, + "grad_norm": 2.38657546043396, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6828985214233398, + "num_tokens": 125860781.0, + "step": 4889 + }, + { + "epoch": 0.5370085657808039, + "grad_norm": 1.9818084239959717, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6891258955001831, + "num_tokens": 125893228.0, + "step": 4890 + }, + { + "epoch": 0.5371183834834176, + "grad_norm": 2.519390344619751, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7029882073402405, + "num_tokens": 125916477.0, + "step": 4891 + }, + { + "epoch": 0.5372282011860312, + "grad_norm": 2.384234666824341, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7189472913742065, + "num_tokens": 125940166.0, + "step": 4892 + }, + { + "epoch": 0.5373380188886449, + "grad_norm": 1.9120681285858154, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6853733062744141, + "num_tokens": 125972942.0, + "step": 4893 + }, + { + "epoch": 0.5374478365912585, + "grad_norm": 2.1328253746032715, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7101956605911255, + "num_tokens": 125998727.0, + "step": 4894 + }, + { + "epoch": 0.5375576542938721, + "grad_norm": 2.126037120819092, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6909854412078857, + "num_tokens": 126025730.0, + "step": 4895 + }, + { + "epoch": 0.5376674719964858, + "grad_norm": 2.2953412532806396, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7138291597366333, + "num_tokens": 126047731.0, + "step": 4896 + }, + { + "epoch": 0.5377772896990995, + "grad_norm": 2.197309732437134, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.698610782623291, + "num_tokens": 126076887.0, + "step": 4897 + }, + { + "epoch": 0.5378871074017132, + "grad_norm": 2.1516127586364746, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6885337829589844, + "num_tokens": 126104452.0, + "step": 4898 + }, + { + "epoch": 0.5379969251043268, + "grad_norm": 2.48055362701416, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7062873840332031, + "num_tokens": 126126422.0, + "step": 4899 + }, + { + "epoch": 0.5381067428069405, + "grad_norm": 2.281953811645508, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7072035074234009, + "num_tokens": 126151244.0, + "step": 4900 + }, + { + "epoch": 0.5382165605095541, + "grad_norm": 2.2867465019226074, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6933860182762146, + "num_tokens": 126177798.0, + "step": 4901 + }, + { + "epoch": 0.5383263782121678, + "grad_norm": 2.1539835929870605, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6928945183753967, + "num_tokens": 126203076.0, + "step": 4902 + }, + { + "epoch": 0.5384361959147814, + "grad_norm": 2.01353120803833, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6939430236816406, + "num_tokens": 126233893.0, + "step": 4903 + }, + { + "epoch": 0.5385460136173951, + "grad_norm": 2.1297216415405273, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7035351991653442, + "num_tokens": 126262998.0, + "step": 4904 + }, + { + "epoch": 0.5386558313200088, + "grad_norm": 2.254711627960205, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6808546781539917, + "num_tokens": 126288580.0, + "step": 4905 + }, + { + "epoch": 0.5387656490226225, + "grad_norm": 2.4526102542877197, + "learning_rate": 1e-06, + "loss": 1.0917, + "mean_token_accuracy": 0.6786562204360962, + "num_tokens": 126311339.0, + "step": 4906 + }, + { + "epoch": 0.5388754667252361, + "grad_norm": 2.4038429260253906, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7223416566848755, + "num_tokens": 126332795.0, + "step": 4907 + }, + { + "epoch": 0.5389852844278498, + "grad_norm": 2.2016165256500244, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7263826727867126, + "num_tokens": 126357093.0, + "step": 4908 + }, + { + "epoch": 0.5390951021304634, + "grad_norm": 1.9646998643875122, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7119132280349731, + "num_tokens": 126386852.0, + "step": 4909 + }, + { + "epoch": 0.5392049198330771, + "grad_norm": 2.212501049041748, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7199397087097168, + "num_tokens": 126413834.0, + "step": 4910 + }, + { + "epoch": 0.5393147375356907, + "grad_norm": 2.2099008560180664, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7069323658943176, + "num_tokens": 126437896.0, + "step": 4911 + }, + { + "epoch": 0.5394245552383045, + "grad_norm": 2.430393934249878, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6953091025352478, + "num_tokens": 126459705.0, + "step": 4912 + }, + { + "epoch": 0.5395343729409181, + "grad_norm": 2.1996798515319824, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6987360715866089, + "num_tokens": 126485410.0, + "step": 4913 + }, + { + "epoch": 0.5396441906435318, + "grad_norm": 2.175480842590332, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.7014076709747314, + "num_tokens": 126511957.0, + "step": 4914 + }, + { + "epoch": 0.5397540083461454, + "grad_norm": 2.088625192642212, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7042161226272583, + "num_tokens": 126537976.0, + "step": 4915 + }, + { + "epoch": 0.539863826048759, + "grad_norm": 2.069888114929199, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7147292494773865, + "num_tokens": 126565582.0, + "step": 4916 + }, + { + "epoch": 0.5399736437513727, + "grad_norm": 2.1443326473236084, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6918688416481018, + "num_tokens": 126591714.0, + "step": 4917 + }, + { + "epoch": 0.5400834614539863, + "grad_norm": 2.299163818359375, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7114359140396118, + "num_tokens": 126614265.0, + "step": 4918 + }, + { + "epoch": 0.5401932791566001, + "grad_norm": 2.4059958457946777, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6720545887947083, + "num_tokens": 126637409.0, + "step": 4919 + }, + { + "epoch": 0.5403030968592137, + "grad_norm": 2.6362576484680176, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7028290033340454, + "num_tokens": 126656624.0, + "step": 4920 + }, + { + "epoch": 0.5404129145618274, + "grad_norm": 2.000070333480835, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7042890191078186, + "num_tokens": 126686957.0, + "step": 4921 + }, + { + "epoch": 0.540522732264441, + "grad_norm": 1.9763423204421997, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6872225403785706, + "num_tokens": 126716757.0, + "step": 4922 + }, + { + "epoch": 0.5406325499670547, + "grad_norm": 2.310100555419922, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6964004039764404, + "num_tokens": 126741772.0, + "step": 4923 + }, + { + "epoch": 0.5407423676696683, + "grad_norm": 2.6215662956237793, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6993740797042847, + "num_tokens": 126761783.0, + "step": 4924 + }, + { + "epoch": 0.540852185372282, + "grad_norm": 2.081374406814575, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6856178045272827, + "num_tokens": 126791983.0, + "step": 4925 + }, + { + "epoch": 0.5409620030748957, + "grad_norm": 2.2663016319274902, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7228302955627441, + "num_tokens": 126814885.0, + "step": 4926 + }, + { + "epoch": 0.5410718207775094, + "grad_norm": 2.155468702316284, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6824068427085876, + "num_tokens": 126841028.0, + "step": 4927 + }, + { + "epoch": 0.541181638480123, + "grad_norm": 2.148545503616333, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7225404977798462, + "num_tokens": 126866738.0, + "step": 4928 + }, + { + "epoch": 0.5412914561827367, + "grad_norm": 2.1482629776000977, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6866454482078552, + "num_tokens": 126894608.0, + "step": 4929 + }, + { + "epoch": 0.5414012738853503, + "grad_norm": 2.5982913970947266, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7108073234558105, + "num_tokens": 126913837.0, + "step": 4930 + }, + { + "epoch": 0.541511091587964, + "grad_norm": 2.3283398151397705, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7152641415596008, + "num_tokens": 126935259.0, + "step": 4931 + }, + { + "epoch": 0.5416209092905776, + "grad_norm": 2.2643067836761475, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6986886262893677, + "num_tokens": 126959184.0, + "step": 4932 + }, + { + "epoch": 0.5417307269931912, + "grad_norm": 2.066657543182373, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7053999900817871, + "num_tokens": 126985989.0, + "step": 4933 + }, + { + "epoch": 0.541840544695805, + "grad_norm": 2.129380702972412, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.689263105392456, + "num_tokens": 127014165.0, + "step": 4934 + }, + { + "epoch": 0.5419503623984187, + "grad_norm": 2.498947858810425, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7137658596038818, + "num_tokens": 127033991.0, + "step": 4935 + }, + { + "epoch": 0.5420601801010323, + "grad_norm": 1.9876551628112793, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7268701195716858, + "num_tokens": 127062437.0, + "step": 4936 + }, + { + "epoch": 0.5421699978036459, + "grad_norm": 2.265232801437378, + "learning_rate": 1e-06, + "loss": 1.0946, + "mean_token_accuracy": 0.6750102043151855, + "num_tokens": 127086553.0, + "step": 4937 + }, + { + "epoch": 0.5422798155062596, + "grad_norm": 2.1095426082611084, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6972547173500061, + "num_tokens": 127115673.0, + "step": 4938 + }, + { + "epoch": 0.5423896332088732, + "grad_norm": 2.1821835041046143, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.7012466192245483, + "num_tokens": 127142908.0, + "step": 4939 + }, + { + "epoch": 0.5424994509114869, + "grad_norm": 2.200371026992798, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7036327719688416, + "num_tokens": 127168258.0, + "step": 4940 + }, + { + "epoch": 0.5426092686141006, + "grad_norm": 2.3922042846679688, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6817324757575989, + "num_tokens": 127192736.0, + "step": 4941 + }, + { + "epoch": 0.5427190863167143, + "grad_norm": 2.2325263023376465, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7259742617607117, + "num_tokens": 127216223.0, + "step": 4942 + }, + { + "epoch": 0.5428289040193279, + "grad_norm": 2.012066125869751, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6947736740112305, + "num_tokens": 127246079.0, + "step": 4943 + }, + { + "epoch": 0.5429387217219416, + "grad_norm": 2.2389748096466064, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.721678614616394, + "num_tokens": 127270037.0, + "step": 4944 + }, + { + "epoch": 0.5430485394245552, + "grad_norm": 2.2102444171905518, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7167056202888489, + "num_tokens": 127294514.0, + "step": 4945 + }, + { + "epoch": 0.5431583571271689, + "grad_norm": 1.9344156980514526, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.702876091003418, + "num_tokens": 127326654.0, + "step": 4946 + }, + { + "epoch": 0.5432681748297825, + "grad_norm": 2.3621296882629395, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7044681310653687, + "num_tokens": 127349463.0, + "step": 4947 + }, + { + "epoch": 0.5433779925323963, + "grad_norm": 2.2868611812591553, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7071804404258728, + "num_tokens": 127372451.0, + "step": 4948 + }, + { + "epoch": 0.5434878102350099, + "grad_norm": 2.1773173809051514, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6922142505645752, + "num_tokens": 127397888.0, + "step": 4949 + }, + { + "epoch": 0.5435976279376236, + "grad_norm": 2.0263917446136475, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7051852345466614, + "num_tokens": 127427050.0, + "step": 4950 + }, + { + "epoch": 0.5437074456402372, + "grad_norm": 2.2974050045013428, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7006670832633972, + "num_tokens": 127450917.0, + "step": 4951 + }, + { + "epoch": 0.5438172633428509, + "grad_norm": 2.5675528049468994, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7042309641838074, + "num_tokens": 127471266.0, + "step": 4952 + }, + { + "epoch": 0.5439270810454645, + "grad_norm": 2.0999159812927246, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7120383977890015, + "num_tokens": 127498643.0, + "step": 4953 + }, + { + "epoch": 0.5440368987480781, + "grad_norm": 2.0044753551483154, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6764565110206604, + "num_tokens": 127530168.0, + "step": 4954 + }, + { + "epoch": 0.5441467164506919, + "grad_norm": 1.941611886024475, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.702039361000061, + "num_tokens": 127560308.0, + "step": 4955 + }, + { + "epoch": 0.5442565341533055, + "grad_norm": 2.2246246337890625, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7052868008613586, + "num_tokens": 127586459.0, + "step": 4956 + }, + { + "epoch": 0.5443663518559192, + "grad_norm": 2.381174325942993, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7160186767578125, + "num_tokens": 127608287.0, + "step": 4957 + }, + { + "epoch": 0.5444761695585328, + "grad_norm": 2.1778035163879395, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6876963973045349, + "num_tokens": 127634569.0, + "step": 4958 + }, + { + "epoch": 0.5445859872611465, + "grad_norm": 2.486605644226074, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7201476097106934, + "num_tokens": 127655294.0, + "step": 4959 + }, + { + "epoch": 0.5446958049637601, + "grad_norm": 2.232983112335205, + "learning_rate": 1e-06, + "loss": 1.0953, + "mean_token_accuracy": 0.6724286079406738, + "num_tokens": 127682043.0, + "step": 4960 + }, + { + "epoch": 0.5448056226663738, + "grad_norm": 2.201190233230591, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6817386150360107, + "num_tokens": 127707227.0, + "step": 4961 + }, + { + "epoch": 0.5449154403689875, + "grad_norm": 2.5106568336486816, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7169172167778015, + "num_tokens": 127727704.0, + "step": 4962 + }, + { + "epoch": 0.5450252580716012, + "grad_norm": 2.355881929397583, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.713388204574585, + "num_tokens": 127750669.0, + "step": 4963 + }, + { + "epoch": 0.5451350757742148, + "grad_norm": 2.283137559890747, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7005746364593506, + "num_tokens": 127775051.0, + "step": 4964 + }, + { + "epoch": 0.5452448934768285, + "grad_norm": 2.548996925354004, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7150341868400574, + "num_tokens": 127796119.0, + "step": 4965 + }, + { + "epoch": 0.5453547111794421, + "grad_norm": 2.2298481464385986, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7026853561401367, + "num_tokens": 127820843.0, + "step": 4966 + }, + { + "epoch": 0.5454645288820558, + "grad_norm": 2.041753053665161, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7061454653739929, + "num_tokens": 127851943.0, + "step": 4967 + }, + { + "epoch": 0.5455743465846694, + "grad_norm": 2.459277629852295, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6778148412704468, + "num_tokens": 127873826.0, + "step": 4968 + }, + { + "epoch": 0.5456841642872831, + "grad_norm": 2.104696273803711, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7087658643722534, + "num_tokens": 127902720.0, + "step": 4969 + }, + { + "epoch": 0.5457939819898968, + "grad_norm": 2.194108486175537, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7135331630706787, + "num_tokens": 127928215.0, + "step": 4970 + }, + { + "epoch": 0.5459037996925105, + "grad_norm": 2.633751630783081, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7215149402618408, + "num_tokens": 127947966.0, + "step": 4971 + }, + { + "epoch": 0.5460136173951241, + "grad_norm": 2.046971321105957, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7050879001617432, + "num_tokens": 127979113.0, + "step": 4972 + }, + { + "epoch": 0.5461234350977378, + "grad_norm": 2.388353109359741, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.704201340675354, + "num_tokens": 128001965.0, + "step": 4973 + }, + { + "epoch": 0.5462332528003514, + "grad_norm": 2.1466786861419678, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7018502950668335, + "num_tokens": 128028629.0, + "step": 4974 + }, + { + "epoch": 0.546343070502965, + "grad_norm": 2.1488044261932373, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.707298994064331, + "num_tokens": 128054078.0, + "step": 4975 + }, + { + "epoch": 0.5464528882055787, + "grad_norm": 2.105454206466675, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7010971307754517, + "num_tokens": 128081736.0, + "step": 4976 + }, + { + "epoch": 0.5465627059081924, + "grad_norm": 2.23006534576416, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.68995201587677, + "num_tokens": 128107716.0, + "step": 4977 + }, + { + "epoch": 0.5466725236108061, + "grad_norm": 2.076343536376953, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6914370059967041, + "num_tokens": 128136784.0, + "step": 4978 + }, + { + "epoch": 0.5467823413134197, + "grad_norm": 2.0709006786346436, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6884088516235352, + "num_tokens": 128164712.0, + "step": 4979 + }, + { + "epoch": 0.5468921590160334, + "grad_norm": 2.0752739906311035, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7172114849090576, + "num_tokens": 128189137.0, + "step": 4980 + }, + { + "epoch": 0.547001976718647, + "grad_norm": 2.4633233547210693, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.694015383720398, + "num_tokens": 128210434.0, + "step": 4981 + }, + { + "epoch": 0.5471117944212607, + "grad_norm": 2.354558229446411, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7185806035995483, + "num_tokens": 128231836.0, + "step": 4982 + }, + { + "epoch": 0.5472216121238743, + "grad_norm": 2.3870277404785156, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6875238418579102, + "num_tokens": 128253618.0, + "step": 4983 + }, + { + "epoch": 0.5473314298264881, + "grad_norm": 2.0676000118255615, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7191964387893677, + "num_tokens": 128281406.0, + "step": 4984 + }, + { + "epoch": 0.5474412475291017, + "grad_norm": 2.585085391998291, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7115776538848877, + "num_tokens": 128301957.0, + "step": 4985 + }, + { + "epoch": 0.5475510652317154, + "grad_norm": 2.129092216491699, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7252844572067261, + "num_tokens": 128328205.0, + "step": 4986 + }, + { + "epoch": 0.547660882934329, + "grad_norm": 2.386625051498413, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7116762399673462, + "num_tokens": 128351234.0, + "step": 4987 + }, + { + "epoch": 0.5477707006369427, + "grad_norm": 2.1253304481506348, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6939049363136292, + "num_tokens": 128378641.0, + "step": 4988 + }, + { + "epoch": 0.5478805183395563, + "grad_norm": 2.1300954818725586, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7129716277122498, + "num_tokens": 128405587.0, + "step": 4989 + }, + { + "epoch": 0.54799033604217, + "grad_norm": 2.255159616470337, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7209488153457642, + "num_tokens": 128431801.0, + "step": 4990 + }, + { + "epoch": 0.5481001537447837, + "grad_norm": 2.566748857498169, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6985055208206177, + "num_tokens": 128452154.0, + "step": 4991 + }, + { + "epoch": 0.5482099714473974, + "grad_norm": 2.3249473571777344, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7109543681144714, + "num_tokens": 128475776.0, + "step": 4992 + }, + { + "epoch": 0.548319789150011, + "grad_norm": 2.079608678817749, + "learning_rate": 1e-06, + "loss": 1.1001, + "mean_token_accuracy": 0.6723541617393494, + "num_tokens": 128504810.0, + "step": 4993 + }, + { + "epoch": 0.5484296068526247, + "grad_norm": 2.214975595474243, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7249097228050232, + "num_tokens": 128528567.0, + "step": 4994 + }, + { + "epoch": 0.5485394245552383, + "grad_norm": 1.869337558746338, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7060317993164062, + "num_tokens": 128560275.0, + "step": 4995 + }, + { + "epoch": 0.5486492422578519, + "grad_norm": 2.103472948074341, + "learning_rate": 1e-06, + "loss": 1.1004, + "mean_token_accuracy": 0.6678544282913208, + "num_tokens": 128589932.0, + "step": 4996 + }, + { + "epoch": 0.5487590599604656, + "grad_norm": 2.203227996826172, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7066385746002197, + "num_tokens": 128615428.0, + "step": 4997 + }, + { + "epoch": 0.5488688776630792, + "grad_norm": 2.35556697845459, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7132493257522583, + "num_tokens": 128639561.0, + "step": 4998 + }, + { + "epoch": 0.548978695365693, + "grad_norm": 2.0638904571533203, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6932891607284546, + "num_tokens": 128666813.0, + "step": 4999 + }, + { + "epoch": 0.5490885130683066, + "grad_norm": 2.62912917137146, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6934987902641296, + "num_tokens": 128685770.0, + "step": 5000 + }, + { + "epoch": 0.5491983307709203, + "grad_norm": 2.0657060146331787, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7128210067749023, + "num_tokens": 128713940.0, + "step": 5001 + }, + { + "epoch": 0.5493081484735339, + "grad_norm": 2.0948402881622314, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7026340961456299, + "num_tokens": 128741223.0, + "step": 5002 + }, + { + "epoch": 0.5494179661761476, + "grad_norm": 2.290370225906372, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7234431505203247, + "num_tokens": 128762770.0, + "step": 5003 + }, + { + "epoch": 0.5495277838787612, + "grad_norm": 2.0149025917053223, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7232263088226318, + "num_tokens": 128792894.0, + "step": 5004 + }, + { + "epoch": 0.5496376015813749, + "grad_norm": 2.2017407417297363, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7300891876220703, + "num_tokens": 128817232.0, + "step": 5005 + }, + { + "epoch": 0.5497474192839886, + "grad_norm": 2.134892225265503, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.719336211681366, + "num_tokens": 128841671.0, + "step": 5006 + }, + { + "epoch": 0.5498572369866023, + "grad_norm": 2.2080771923065186, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7136905789375305, + "num_tokens": 128866230.0, + "step": 5007 + }, + { + "epoch": 0.5499670546892159, + "grad_norm": 2.39453125, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.696776807308197, + "num_tokens": 128889279.0, + "step": 5008 + }, + { + "epoch": 0.5500768723918296, + "grad_norm": 2.2810003757476807, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7085456848144531, + "num_tokens": 128912095.0, + "step": 5009 + }, + { + "epoch": 0.5501866900944432, + "grad_norm": 2.4530556201934814, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7086246609687805, + "num_tokens": 128932195.0, + "step": 5010 + }, + { + "epoch": 0.5502965077970569, + "grad_norm": 2.4122068881988525, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.724067211151123, + "num_tokens": 128952749.0, + "step": 5011 + }, + { + "epoch": 0.5504063254996705, + "grad_norm": 2.140415906906128, + "learning_rate": 1e-06, + "loss": 1.0633, + "mean_token_accuracy": 0.6901156306266785, + "num_tokens": 128981500.0, + "step": 5012 + }, + { + "epoch": 0.5505161432022843, + "grad_norm": 2.173826217651367, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6914713382720947, + "num_tokens": 129006140.0, + "step": 5013 + }, + { + "epoch": 0.5506259609048979, + "grad_norm": 2.124356508255005, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7345837354660034, + "num_tokens": 129031240.0, + "step": 5014 + }, + { + "epoch": 0.5507357786075116, + "grad_norm": 2.208226203918457, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7193980813026428, + "num_tokens": 129056431.0, + "step": 5015 + }, + { + "epoch": 0.5508455963101252, + "grad_norm": 2.136693239212036, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.6849609613418579, + "num_tokens": 129082913.0, + "step": 5016 + }, + { + "epoch": 0.5509554140127388, + "grad_norm": 2.4104807376861572, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.713912844657898, + "num_tokens": 129104689.0, + "step": 5017 + }, + { + "epoch": 0.5510652317153525, + "grad_norm": 2.4046366214752197, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7129781246185303, + "num_tokens": 129127696.0, + "step": 5018 + }, + { + "epoch": 0.5511750494179661, + "grad_norm": 2.097791910171509, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7032663822174072, + "num_tokens": 129154700.0, + "step": 5019 + }, + { + "epoch": 0.5512848671205799, + "grad_norm": 2.286522626876831, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7189385890960693, + "num_tokens": 129179738.0, + "step": 5020 + }, + { + "epoch": 0.5513946848231935, + "grad_norm": 2.1000163555145264, + "learning_rate": 1e-06, + "loss": 1.1033, + "mean_token_accuracy": 0.677667498588562, + "num_tokens": 129209103.0, + "step": 5021 + }, + { + "epoch": 0.5515045025258072, + "grad_norm": 2.403088331222534, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7137231826782227, + "num_tokens": 129230493.0, + "step": 5022 + }, + { + "epoch": 0.5516143202284208, + "grad_norm": 2.421980142593384, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7316173315048218, + "num_tokens": 129251698.0, + "step": 5023 + }, + { + "epoch": 0.5517241379310345, + "grad_norm": 2.0077736377716064, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6936559081077576, + "num_tokens": 129281564.0, + "step": 5024 + }, + { + "epoch": 0.5518339556336481, + "grad_norm": 2.142768621444702, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6965968608856201, + "num_tokens": 129309106.0, + "step": 5025 + }, + { + "epoch": 0.5519437733362618, + "grad_norm": 2.3947031497955322, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7133594155311584, + "num_tokens": 129331439.0, + "step": 5026 + }, + { + "epoch": 0.5520535910388754, + "grad_norm": 2.117765426635742, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7136111259460449, + "num_tokens": 129358548.0, + "step": 5027 + }, + { + "epoch": 0.5521634087414892, + "grad_norm": 2.361194133758545, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7122563123703003, + "num_tokens": 129383259.0, + "step": 5028 + }, + { + "epoch": 0.5522732264441028, + "grad_norm": 2.608252763748169, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6768897771835327, + "num_tokens": 129403441.0, + "step": 5029 + }, + { + "epoch": 0.5523830441467165, + "grad_norm": 2.38836669921875, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.699989914894104, + "num_tokens": 129425106.0, + "step": 5030 + }, + { + "epoch": 0.5524928618493301, + "grad_norm": 2.410031795501709, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6954199075698853, + "num_tokens": 129448439.0, + "step": 5031 + }, + { + "epoch": 0.5526026795519438, + "grad_norm": 2.0006017684936523, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7106719017028809, + "num_tokens": 129477655.0, + "step": 5032 + }, + { + "epoch": 0.5527124972545574, + "grad_norm": 2.147143840789795, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6961000561714172, + "num_tokens": 129505783.0, + "step": 5033 + }, + { + "epoch": 0.552822314957171, + "grad_norm": 2.055647850036621, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7055844068527222, + "num_tokens": 129532568.0, + "step": 5034 + }, + { + "epoch": 0.5529321326597848, + "grad_norm": 2.3549928665161133, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7116140723228455, + "num_tokens": 129553520.0, + "step": 5035 + }, + { + "epoch": 0.5530419503623984, + "grad_norm": 2.121004104614258, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7085455060005188, + "num_tokens": 129582139.0, + "step": 5036 + }, + { + "epoch": 0.5531517680650121, + "grad_norm": 2.5468432903289795, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7300401926040649, + "num_tokens": 129600355.0, + "step": 5037 + }, + { + "epoch": 0.5532615857676257, + "grad_norm": 2.0868618488311768, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6865350008010864, + "num_tokens": 129629388.0, + "step": 5038 + }, + { + "epoch": 0.5533714034702394, + "grad_norm": 2.0691897869110107, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6990069150924683, + "num_tokens": 129658867.0, + "step": 5039 + }, + { + "epoch": 0.553481221172853, + "grad_norm": 2.0995850563049316, + "learning_rate": 1e-06, + "loss": 1.1288, + "mean_token_accuracy": 0.6647897362709045, + "num_tokens": 129687643.0, + "step": 5040 + }, + { + "epoch": 0.5535910388754667, + "grad_norm": 2.053062677383423, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6918584704399109, + "num_tokens": 129717963.0, + "step": 5041 + }, + { + "epoch": 0.5537008565780804, + "grad_norm": 2.4264824390411377, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7178335189819336, + "num_tokens": 129739744.0, + "step": 5042 + }, + { + "epoch": 0.5538106742806941, + "grad_norm": 2.1764745712280273, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.679955780506134, + "num_tokens": 129765797.0, + "step": 5043 + }, + { + "epoch": 0.5539204919833077, + "grad_norm": 2.164076328277588, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7059369087219238, + "num_tokens": 129791612.0, + "step": 5044 + }, + { + "epoch": 0.5540303096859214, + "grad_norm": 1.9272929430007935, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7048097848892212, + "num_tokens": 129822726.0, + "step": 5045 + }, + { + "epoch": 0.554140127388535, + "grad_norm": 2.3369760513305664, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7095255851745605, + "num_tokens": 129846263.0, + "step": 5046 + }, + { + "epoch": 0.5542499450911487, + "grad_norm": 2.2079215049743652, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7095028162002563, + "num_tokens": 129874022.0, + "step": 5047 + }, + { + "epoch": 0.5543597627937623, + "grad_norm": 2.1360466480255127, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7155886292457581, + "num_tokens": 129900351.0, + "step": 5048 + }, + { + "epoch": 0.5544695804963761, + "grad_norm": 2.172656774520874, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.719401478767395, + "num_tokens": 129924175.0, + "step": 5049 + }, + { + "epoch": 0.5545793981989897, + "grad_norm": 2.4359183311462402, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6989967823028564, + "num_tokens": 129946527.0, + "step": 5050 + }, + { + "epoch": 0.5546892159016034, + "grad_norm": 2.581597089767456, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7042036056518555, + "num_tokens": 129965318.0, + "step": 5051 + }, + { + "epoch": 0.554799033604217, + "grad_norm": 1.9676146507263184, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7186987996101379, + "num_tokens": 129995136.0, + "step": 5052 + }, + { + "epoch": 0.5549088513068307, + "grad_norm": 2.1562719345092773, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6993952989578247, + "num_tokens": 130021825.0, + "step": 5053 + }, + { + "epoch": 0.5550186690094443, + "grad_norm": 2.1372201442718506, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7104402184486389, + "num_tokens": 130047832.0, + "step": 5054 + }, + { + "epoch": 0.555128486712058, + "grad_norm": 2.3599557876586914, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6949285864830017, + "num_tokens": 130072942.0, + "step": 5055 + }, + { + "epoch": 0.5552383044146716, + "grad_norm": 2.1780426502227783, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6979165077209473, + "num_tokens": 130099461.0, + "step": 5056 + }, + { + "epoch": 0.5553481221172853, + "grad_norm": 1.9933217763900757, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6973710060119629, + "num_tokens": 130131189.0, + "step": 5057 + }, + { + "epoch": 0.555457939819899, + "grad_norm": 2.286862850189209, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6983184814453125, + "num_tokens": 130154958.0, + "step": 5058 + }, + { + "epoch": 0.5555677575225126, + "grad_norm": 2.3913211822509766, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6901235580444336, + "num_tokens": 130177459.0, + "step": 5059 + }, + { + "epoch": 0.5556775752251263, + "grad_norm": 2.094884157180786, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6867392063140869, + "num_tokens": 130205392.0, + "step": 5060 + }, + { + "epoch": 0.5557873929277399, + "grad_norm": 2.3716719150543213, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6917905807495117, + "num_tokens": 130227884.0, + "step": 5061 + }, + { + "epoch": 0.5558972106303536, + "grad_norm": 2.078989028930664, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6875477433204651, + "num_tokens": 130258485.0, + "step": 5062 + }, + { + "epoch": 0.5560070283329672, + "grad_norm": 2.2339048385620117, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.6803551912307739, + "num_tokens": 130285548.0, + "step": 5063 + }, + { + "epoch": 0.556116846035581, + "grad_norm": 2.3202555179595947, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7043635249137878, + "num_tokens": 130309989.0, + "step": 5064 + }, + { + "epoch": 0.5562266637381946, + "grad_norm": 1.967708945274353, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6878811120986938, + "num_tokens": 130342383.0, + "step": 5065 + }, + { + "epoch": 0.5563364814408083, + "grad_norm": 2.384629011154175, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.694907546043396, + "num_tokens": 130365736.0, + "step": 5066 + }, + { + "epoch": 0.5564462991434219, + "grad_norm": 2.3299317359924316, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7184809446334839, + "num_tokens": 130390045.0, + "step": 5067 + }, + { + "epoch": 0.5565561168460356, + "grad_norm": 2.1310765743255615, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.6802016496658325, + "num_tokens": 130421307.0, + "step": 5068 + }, + { + "epoch": 0.5566659345486492, + "grad_norm": 2.235966682434082, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6947605609893799, + "num_tokens": 130444431.0, + "step": 5069 + }, + { + "epoch": 0.5567757522512629, + "grad_norm": 2.174516201019287, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.6976917386054993, + "num_tokens": 130470502.0, + "step": 5070 + }, + { + "epoch": 0.5568855699538766, + "grad_norm": 2.5378947257995605, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7356580495834351, + "num_tokens": 130489040.0, + "step": 5071 + }, + { + "epoch": 0.5569953876564903, + "grad_norm": 2.1883885860443115, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6845114231109619, + "num_tokens": 130513563.0, + "step": 5072 + }, + { + "epoch": 0.5571052053591039, + "grad_norm": 2.097045660018921, + "learning_rate": 1e-06, + "loss": 1.0849, + "mean_token_accuracy": 0.6769955158233643, + "num_tokens": 130540342.0, + "step": 5073 + }, + { + "epoch": 0.5572150230617176, + "grad_norm": 2.0552334785461426, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6957577466964722, + "num_tokens": 130568605.0, + "step": 5074 + }, + { + "epoch": 0.5573248407643312, + "grad_norm": 2.2642040252685547, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6949557662010193, + "num_tokens": 130592204.0, + "step": 5075 + }, + { + "epoch": 0.5574346584669448, + "grad_norm": 2.3864433765411377, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7173199653625488, + "num_tokens": 130614611.0, + "step": 5076 + }, + { + "epoch": 0.5575444761695585, + "grad_norm": 2.2137515544891357, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6860907077789307, + "num_tokens": 130639848.0, + "step": 5077 + }, + { + "epoch": 0.5576542938721722, + "grad_norm": 2.164454936981201, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7174674272537231, + "num_tokens": 130666284.0, + "step": 5078 + }, + { + "epoch": 0.5577641115747859, + "grad_norm": 2.206295967102051, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6995105743408203, + "num_tokens": 130691229.0, + "step": 5079 + }, + { + "epoch": 0.5578739292773995, + "grad_norm": 2.229111671447754, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7025953531265259, + "num_tokens": 130716000.0, + "step": 5080 + }, + { + "epoch": 0.5579837469800132, + "grad_norm": 2.227806806564331, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7131021022796631, + "num_tokens": 130741113.0, + "step": 5081 + }, + { + "epoch": 0.5580935646826268, + "grad_norm": 2.1839821338653564, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7201225757598877, + "num_tokens": 130764762.0, + "step": 5082 + }, + { + "epoch": 0.5582033823852405, + "grad_norm": 2.1233768463134766, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.701248288154602, + "num_tokens": 130790668.0, + "step": 5083 + }, + { + "epoch": 0.5583132000878541, + "grad_norm": 2.178602457046509, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6994513273239136, + "num_tokens": 130816106.0, + "step": 5084 + }, + { + "epoch": 0.5584230177904678, + "grad_norm": 2.1413795948028564, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7031522989273071, + "num_tokens": 130842617.0, + "step": 5085 + }, + { + "epoch": 0.5585328354930815, + "grad_norm": 2.3254239559173584, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6889053583145142, + "num_tokens": 130865752.0, + "step": 5086 + }, + { + "epoch": 0.5586426531956952, + "grad_norm": 2.029053211212158, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7063988447189331, + "num_tokens": 130894537.0, + "step": 5087 + }, + { + "epoch": 0.5587524708983088, + "grad_norm": 2.3851864337921143, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6888142824172974, + "num_tokens": 130917309.0, + "step": 5088 + }, + { + "epoch": 0.5588622886009225, + "grad_norm": 2.4673078060150146, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6945732831954956, + "num_tokens": 130938111.0, + "step": 5089 + }, + { + "epoch": 0.5589721063035361, + "grad_norm": 2.1655545234680176, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6888837814331055, + "num_tokens": 130963494.0, + "step": 5090 + }, + { + "epoch": 0.5590819240061498, + "grad_norm": 2.012995719909668, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7027616500854492, + "num_tokens": 130990207.0, + "step": 5091 + }, + { + "epoch": 0.5591917417087634, + "grad_norm": 2.266638994216919, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6902331709861755, + "num_tokens": 131018376.0, + "step": 5092 + }, + { + "epoch": 0.5593015594113772, + "grad_norm": 1.994059681892395, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7097906470298767, + "num_tokens": 131046078.0, + "step": 5093 + }, + { + "epoch": 0.5594113771139908, + "grad_norm": 2.0652215480804443, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6923633217811584, + "num_tokens": 131075806.0, + "step": 5094 + }, + { + "epoch": 0.5595211948166045, + "grad_norm": 2.0947277545928955, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6911027431488037, + "num_tokens": 131104239.0, + "step": 5095 + }, + { + "epoch": 0.5596310125192181, + "grad_norm": 2.0575153827667236, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7005151510238647, + "num_tokens": 131134003.0, + "step": 5096 + }, + { + "epoch": 0.5597408302218317, + "grad_norm": 1.9595553874969482, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6833746433258057, + "num_tokens": 131165257.0, + "step": 5097 + }, + { + "epoch": 0.5598506479244454, + "grad_norm": 2.2984344959259033, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6914723515510559, + "num_tokens": 131189969.0, + "step": 5098 + }, + { + "epoch": 0.559960465627059, + "grad_norm": 2.1399004459381104, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7040414214134216, + "num_tokens": 131220020.0, + "step": 5099 + }, + { + "epoch": 0.5600702833296728, + "grad_norm": 2.2049808502197266, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.6980931162834167, + "num_tokens": 131247516.0, + "step": 5100 + }, + { + "epoch": 0.5601801010322864, + "grad_norm": 2.39715576171875, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7068066596984863, + "num_tokens": 131269898.0, + "step": 5101 + }, + { + "epoch": 0.5602899187349001, + "grad_norm": 2.278182029724121, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6924192905426025, + "num_tokens": 131293647.0, + "step": 5102 + }, + { + "epoch": 0.5603997364375137, + "grad_norm": 2.1842124462127686, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6975046992301941, + "num_tokens": 131320881.0, + "step": 5103 + }, + { + "epoch": 0.5605095541401274, + "grad_norm": 2.303391218185425, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6939572691917419, + "num_tokens": 131343639.0, + "step": 5104 + }, + { + "epoch": 0.560619371842741, + "grad_norm": 2.0482358932495117, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.686469554901123, + "num_tokens": 131374635.0, + "step": 5105 + }, + { + "epoch": 0.5607291895453547, + "grad_norm": 2.1800084114074707, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6960783004760742, + "num_tokens": 131399767.0, + "step": 5106 + }, + { + "epoch": 0.5608390072479684, + "grad_norm": 2.226074695587158, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.732559323310852, + "num_tokens": 131426652.0, + "step": 5107 + }, + { + "epoch": 0.5609488249505821, + "grad_norm": 2.4138476848602295, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7032831907272339, + "num_tokens": 131448670.0, + "step": 5108 + }, + { + "epoch": 0.5610586426531957, + "grad_norm": 2.2176685333251953, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6829183101654053, + "num_tokens": 131474906.0, + "step": 5109 + }, + { + "epoch": 0.5611684603558094, + "grad_norm": 2.4710025787353516, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7051099538803101, + "num_tokens": 131495489.0, + "step": 5110 + }, + { + "epoch": 0.561278278058423, + "grad_norm": 1.9847638607025146, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6948491930961609, + "num_tokens": 131524055.0, + "step": 5111 + }, + { + "epoch": 0.5613880957610367, + "grad_norm": 2.163292407989502, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6948540210723877, + "num_tokens": 131550009.0, + "step": 5112 + }, + { + "epoch": 0.5614979134636503, + "grad_norm": 2.4051430225372314, + "learning_rate": 1e-06, + "loss": 1.0859, + "mean_token_accuracy": 0.6792408227920532, + "num_tokens": 131573640.0, + "step": 5113 + }, + { + "epoch": 0.5616077311662641, + "grad_norm": 2.353947639465332, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7061178088188171, + "num_tokens": 131595238.0, + "step": 5114 + }, + { + "epoch": 0.5617175488688777, + "grad_norm": 2.143188714981079, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6955272555351257, + "num_tokens": 131622333.0, + "step": 5115 + }, + { + "epoch": 0.5618273665714913, + "grad_norm": 2.0742833614349365, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7291369438171387, + "num_tokens": 131648966.0, + "step": 5116 + }, + { + "epoch": 0.561937184274105, + "grad_norm": 2.2492735385894775, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7260542511940002, + "num_tokens": 131672891.0, + "step": 5117 + }, + { + "epoch": 0.5620470019767186, + "grad_norm": 1.937347173690796, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6738876700401306, + "num_tokens": 131707171.0, + "step": 5118 + }, + { + "epoch": 0.5621568196793323, + "grad_norm": 2.27530837059021, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7023243308067322, + "num_tokens": 131731739.0, + "step": 5119 + }, + { + "epoch": 0.5622666373819459, + "grad_norm": 2.103311777114868, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7033034563064575, + "num_tokens": 131761004.0, + "step": 5120 + }, + { + "epoch": 0.5623764550845596, + "grad_norm": 2.0012009143829346, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6923728585243225, + "num_tokens": 131791173.0, + "step": 5121 + }, + { + "epoch": 0.5624862727871733, + "grad_norm": 2.113248825073242, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7201583981513977, + "num_tokens": 131815880.0, + "step": 5122 + }, + { + "epoch": 0.562596090489787, + "grad_norm": 2.027492046356201, + "learning_rate": 1e-06, + "loss": 1.0833, + "mean_token_accuracy": 0.6749738454818726, + "num_tokens": 131849604.0, + "step": 5123 + }, + { + "epoch": 0.5627059081924006, + "grad_norm": 2.220791816711426, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7212831974029541, + "num_tokens": 131874698.0, + "step": 5124 + }, + { + "epoch": 0.5628157258950143, + "grad_norm": 2.390545129776001, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6928979754447937, + "num_tokens": 131899383.0, + "step": 5125 + }, + { + "epoch": 0.5629255435976279, + "grad_norm": 2.0998213291168213, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7012538909912109, + "num_tokens": 131927161.0, + "step": 5126 + }, + { + "epoch": 0.5630353613002416, + "grad_norm": 2.3322372436523438, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7130005955696106, + "num_tokens": 131949628.0, + "step": 5127 + }, + { + "epoch": 0.5631451790028552, + "grad_norm": 2.464693784713745, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7170567512512207, + "num_tokens": 131969916.0, + "step": 5128 + }, + { + "epoch": 0.563254996705469, + "grad_norm": 2.096059560775757, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7096015810966492, + "num_tokens": 131998706.0, + "step": 5129 + }, + { + "epoch": 0.5633648144080826, + "grad_norm": 2.0817370414733887, + "learning_rate": 1e-06, + "loss": 1.1117, + "mean_token_accuracy": 0.6727818846702576, + "num_tokens": 132029952.0, + "step": 5130 + }, + { + "epoch": 0.5634746321106963, + "grad_norm": 1.859932780265808, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7219629287719727, + "num_tokens": 132063256.0, + "step": 5131 + }, + { + "epoch": 0.5635844498133099, + "grad_norm": 2.2432000637054443, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6914857625961304, + "num_tokens": 132090381.0, + "step": 5132 + }, + { + "epoch": 0.5636942675159236, + "grad_norm": 2.3203492164611816, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7051265239715576, + "num_tokens": 132112956.0, + "step": 5133 + }, + { + "epoch": 0.5638040852185372, + "grad_norm": 1.995676040649414, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7209525108337402, + "num_tokens": 132143330.0, + "step": 5134 + }, + { + "epoch": 0.5639139029211508, + "grad_norm": 2.0674431324005127, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7056348323822021, + "num_tokens": 132171926.0, + "step": 5135 + }, + { + "epoch": 0.5640237206237646, + "grad_norm": 2.253495454788208, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7003939747810364, + "num_tokens": 132195634.0, + "step": 5136 + }, + { + "epoch": 0.5641335383263782, + "grad_norm": 2.0807721614837646, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7072097063064575, + "num_tokens": 132223325.0, + "step": 5137 + }, + { + "epoch": 0.5642433560289919, + "grad_norm": 2.3959310054779053, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7041252851486206, + "num_tokens": 132247187.0, + "step": 5138 + }, + { + "epoch": 0.5643531737316055, + "grad_norm": 2.0771172046661377, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7156500816345215, + "num_tokens": 132273485.0, + "step": 5139 + }, + { + "epoch": 0.5644629914342192, + "grad_norm": 2.148751974105835, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6918485760688782, + "num_tokens": 132301830.0, + "step": 5140 + }, + { + "epoch": 0.5645728091368328, + "grad_norm": 2.092471122741699, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7297555208206177, + "num_tokens": 132331318.0, + "step": 5141 + }, + { + "epoch": 0.5646826268394465, + "grad_norm": 2.2769875526428223, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6847153306007385, + "num_tokens": 132356614.0, + "step": 5142 + }, + { + "epoch": 0.5647924445420602, + "grad_norm": 2.052830219268799, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7078632712364197, + "num_tokens": 132387376.0, + "step": 5143 + }, + { + "epoch": 0.5649022622446739, + "grad_norm": 2.4028451442718506, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.72041916847229, + "num_tokens": 132408207.0, + "step": 5144 + }, + { + "epoch": 0.5650120799472875, + "grad_norm": 2.2379603385925293, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7074484825134277, + "num_tokens": 132432623.0, + "step": 5145 + }, + { + "epoch": 0.5651218976499012, + "grad_norm": 2.1605019569396973, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7040143609046936, + "num_tokens": 132460390.0, + "step": 5146 + }, + { + "epoch": 0.5652317153525148, + "grad_norm": 2.274672746658325, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6828916072845459, + "num_tokens": 132485236.0, + "step": 5147 + }, + { + "epoch": 0.5653415330551285, + "grad_norm": 2.12762188911438, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7063044905662537, + "num_tokens": 132511585.0, + "step": 5148 + }, + { + "epoch": 0.5654513507577421, + "grad_norm": 2.0409975051879883, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7228492498397827, + "num_tokens": 132539954.0, + "step": 5149 + }, + { + "epoch": 0.5655611684603558, + "grad_norm": 2.2660720348358154, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7070143222808838, + "num_tokens": 132563221.0, + "step": 5150 + }, + { + "epoch": 0.5656709861629695, + "grad_norm": 2.365414619445801, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7077445983886719, + "num_tokens": 132587069.0, + "step": 5151 + }, + { + "epoch": 0.5657808038655832, + "grad_norm": 2.171886920928955, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7135586738586426, + "num_tokens": 132611250.0, + "step": 5152 + }, + { + "epoch": 0.5658906215681968, + "grad_norm": 2.195462465286255, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6974235773086548, + "num_tokens": 132638381.0, + "step": 5153 + }, + { + "epoch": 0.5660004392708105, + "grad_norm": 2.007053852081299, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7136538028717041, + "num_tokens": 132667387.0, + "step": 5154 + }, + { + "epoch": 0.5661102569734241, + "grad_norm": 2.138927936553955, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7092127799987793, + "num_tokens": 132695711.0, + "step": 5155 + }, + { + "epoch": 0.5662200746760377, + "grad_norm": 2.2210121154785156, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6860418319702148, + "num_tokens": 132722591.0, + "step": 5156 + }, + { + "epoch": 0.5663298923786514, + "grad_norm": 2.2364094257354736, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7188200354576111, + "num_tokens": 132748930.0, + "step": 5157 + }, + { + "epoch": 0.5664397100812651, + "grad_norm": 2.134705066680908, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7096556425094604, + "num_tokens": 132775780.0, + "step": 5158 + }, + { + "epoch": 0.5665495277838788, + "grad_norm": 2.6825015544891357, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7003652453422546, + "num_tokens": 132793773.0, + "step": 5159 + }, + { + "epoch": 0.5666593454864924, + "grad_norm": 2.270780086517334, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7284995913505554, + "num_tokens": 132818054.0, + "step": 5160 + }, + { + "epoch": 0.5667691631891061, + "grad_norm": 2.641169309616089, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7403568029403687, + "num_tokens": 132834522.0, + "step": 5161 + }, + { + "epoch": 0.5668789808917197, + "grad_norm": 2.259726047515869, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7186300158500671, + "num_tokens": 132857162.0, + "step": 5162 + }, + { + "epoch": 0.5669887985943334, + "grad_norm": 1.9419586658477783, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6991379261016846, + "num_tokens": 132888246.0, + "step": 5163 + }, + { + "epoch": 0.567098616296947, + "grad_norm": 1.9860038757324219, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7147124409675598, + "num_tokens": 132918590.0, + "step": 5164 + }, + { + "epoch": 0.5672084339995608, + "grad_norm": 2.3657848834991455, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7002868056297302, + "num_tokens": 132940477.0, + "step": 5165 + }, + { + "epoch": 0.5673182517021744, + "grad_norm": 2.059126853942871, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6998445987701416, + "num_tokens": 132971546.0, + "step": 5166 + }, + { + "epoch": 0.5674280694047881, + "grad_norm": 2.1007919311523438, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6980234980583191, + "num_tokens": 133000498.0, + "step": 5167 + }, + { + "epoch": 0.5675378871074017, + "grad_norm": 2.490260362625122, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7194632291793823, + "num_tokens": 133020089.0, + "step": 5168 + }, + { + "epoch": 0.5676477048100154, + "grad_norm": 1.926693081855774, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6912713646888733, + "num_tokens": 133051265.0, + "step": 5169 + }, + { + "epoch": 0.567757522512629, + "grad_norm": 2.517702102661133, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7102007865905762, + "num_tokens": 133071614.0, + "step": 5170 + }, + { + "epoch": 0.5678673402152427, + "grad_norm": 2.170046091079712, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7149888277053833, + "num_tokens": 133097551.0, + "step": 5171 + }, + { + "epoch": 0.5679771579178564, + "grad_norm": 2.46205472946167, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6969988346099854, + "num_tokens": 133119352.0, + "step": 5172 + }, + { + "epoch": 0.5680869756204701, + "grad_norm": 2.0852503776550293, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7041918039321899, + "num_tokens": 133147121.0, + "step": 5173 + }, + { + "epoch": 0.5681967933230837, + "grad_norm": 2.0985453128814697, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6999524831771851, + "num_tokens": 133176491.0, + "step": 5174 + }, + { + "epoch": 0.5683066110256974, + "grad_norm": 2.460310697555542, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.722496747970581, + "num_tokens": 133197494.0, + "step": 5175 + }, + { + "epoch": 0.568416428728311, + "grad_norm": 2.1052653789520264, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6962001323699951, + "num_tokens": 133226417.0, + "step": 5176 + }, + { + "epoch": 0.5685262464309246, + "grad_norm": 2.4590911865234375, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7093302607536316, + "num_tokens": 133249745.0, + "step": 5177 + }, + { + "epoch": 0.5686360641335383, + "grad_norm": 1.8790221214294434, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6872545480728149, + "num_tokens": 133284473.0, + "step": 5178 + }, + { + "epoch": 0.5687458818361519, + "grad_norm": 2.1668107509613037, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.701598584651947, + "num_tokens": 133311350.0, + "step": 5179 + }, + { + "epoch": 0.5688556995387657, + "grad_norm": 2.0382132530212402, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6946014165878296, + "num_tokens": 133340454.0, + "step": 5180 + }, + { + "epoch": 0.5689655172413793, + "grad_norm": 2.127133369445801, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6926276683807373, + "num_tokens": 133367997.0, + "step": 5181 + }, + { + "epoch": 0.569075334943993, + "grad_norm": 2.120737314224243, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6863261461257935, + "num_tokens": 133396756.0, + "step": 5182 + }, + { + "epoch": 0.5691851526466066, + "grad_norm": 1.8893342018127441, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7031896114349365, + "num_tokens": 133429374.0, + "step": 5183 + }, + { + "epoch": 0.5692949703492203, + "grad_norm": 2.365959644317627, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6999339461326599, + "num_tokens": 133453946.0, + "step": 5184 + }, + { + "epoch": 0.5694047880518339, + "grad_norm": 2.0060155391693115, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7005882263183594, + "num_tokens": 133484045.0, + "step": 5185 + }, + { + "epoch": 0.5695146057544476, + "grad_norm": 2.104914903640747, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7303681373596191, + "num_tokens": 133510853.0, + "step": 5186 + }, + { + "epoch": 0.5696244234570613, + "grad_norm": 2.344900131225586, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6944146752357483, + "num_tokens": 133535263.0, + "step": 5187 + }, + { + "epoch": 0.569734241159675, + "grad_norm": 2.1656293869018555, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7298275232315063, + "num_tokens": 133561404.0, + "step": 5188 + }, + { + "epoch": 0.5698440588622886, + "grad_norm": 2.522991418838501, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7220042943954468, + "num_tokens": 133581644.0, + "step": 5189 + }, + { + "epoch": 0.5699538765649023, + "grad_norm": 2.2164149284362793, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7095191478729248, + "num_tokens": 133606271.0, + "step": 5190 + }, + { + "epoch": 0.5700636942675159, + "grad_norm": 2.0847904682159424, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6895833015441895, + "num_tokens": 133634658.0, + "step": 5191 + }, + { + "epoch": 0.5701735119701296, + "grad_norm": 2.0084526538848877, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6960753798484802, + "num_tokens": 133662625.0, + "step": 5192 + }, + { + "epoch": 0.5702833296727432, + "grad_norm": 2.285339117050171, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6997584700584412, + "num_tokens": 133686127.0, + "step": 5193 + }, + { + "epoch": 0.570393147375357, + "grad_norm": 2.5450940132141113, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7273464202880859, + "num_tokens": 133705544.0, + "step": 5194 + }, + { + "epoch": 0.5705029650779706, + "grad_norm": 1.8013249635696411, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6878777742385864, + "num_tokens": 133743840.0, + "step": 5195 + }, + { + "epoch": 0.5706127827805842, + "grad_norm": 2.056701898574829, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7076466083526611, + "num_tokens": 133770515.0, + "step": 5196 + }, + { + "epoch": 0.5707226004831979, + "grad_norm": 2.1966562271118164, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6956150531768799, + "num_tokens": 133795967.0, + "step": 5197 + }, + { + "epoch": 0.5708324181858115, + "grad_norm": 2.333129405975342, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7014239430427551, + "num_tokens": 133817709.0, + "step": 5198 + }, + { + "epoch": 0.5709422358884252, + "grad_norm": 2.2772257328033447, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6993237137794495, + "num_tokens": 133842727.0, + "step": 5199 + }, + { + "epoch": 0.5710520535910388, + "grad_norm": 2.198530673980713, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7201673984527588, + "num_tokens": 133867520.0, + "step": 5200 + }, + { + "epoch": 0.5711618712936526, + "grad_norm": 2.194920301437378, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6918342113494873, + "num_tokens": 133895611.0, + "step": 5201 + }, + { + "epoch": 0.5712716889962662, + "grad_norm": 1.9745806455612183, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6988936066627502, + "num_tokens": 133927970.0, + "step": 5202 + }, + { + "epoch": 0.5713815066988799, + "grad_norm": 2.1138124465942383, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7095036506652832, + "num_tokens": 133953020.0, + "step": 5203 + }, + { + "epoch": 0.5714913244014935, + "grad_norm": 2.1566314697265625, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7024645209312439, + "num_tokens": 133979548.0, + "step": 5204 + }, + { + "epoch": 0.5716011421041072, + "grad_norm": 2.5999271869659424, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7124550342559814, + "num_tokens": 133999082.0, + "step": 5205 + }, + { + "epoch": 0.5717109598067208, + "grad_norm": 2.2301902770996094, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6914458274841309, + "num_tokens": 134025610.0, + "step": 5206 + }, + { + "epoch": 0.5718207775093345, + "grad_norm": 2.0953369140625, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7320191860198975, + "num_tokens": 134052346.0, + "step": 5207 + }, + { + "epoch": 0.5719305952119481, + "grad_norm": 2.323272705078125, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7006083726882935, + "num_tokens": 134075559.0, + "step": 5208 + }, + { + "epoch": 0.5720404129145619, + "grad_norm": 2.2042088508605957, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6862931847572327, + "num_tokens": 134101774.0, + "step": 5209 + }, + { + "epoch": 0.5721502306171755, + "grad_norm": 2.1571645736694336, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.696057915687561, + "num_tokens": 134128045.0, + "step": 5210 + }, + { + "epoch": 0.5722600483197892, + "grad_norm": 2.4743549823760986, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.699285626411438, + "num_tokens": 134149856.0, + "step": 5211 + }, + { + "epoch": 0.5723698660224028, + "grad_norm": 2.4160585403442383, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7094370722770691, + "num_tokens": 134171155.0, + "step": 5212 + }, + { + "epoch": 0.5724796837250165, + "grad_norm": 2.0664727687835693, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.6997207403182983, + "num_tokens": 134201983.0, + "step": 5213 + }, + { + "epoch": 0.5725895014276301, + "grad_norm": 2.085265636444092, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7152107954025269, + "num_tokens": 134228599.0, + "step": 5214 + }, + { + "epoch": 0.5726993191302437, + "grad_norm": 2.2912437915802, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6932690143585205, + "num_tokens": 134253206.0, + "step": 5215 + }, + { + "epoch": 0.5728091368328575, + "grad_norm": 2.0867362022399902, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6977609395980835, + "num_tokens": 134280502.0, + "step": 5216 + }, + { + "epoch": 0.5729189545354711, + "grad_norm": 2.3301644325256348, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6865898966789246, + "num_tokens": 134306955.0, + "step": 5217 + }, + { + "epoch": 0.5730287722380848, + "grad_norm": 2.0727009773254395, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6905863285064697, + "num_tokens": 134337434.0, + "step": 5218 + }, + { + "epoch": 0.5731385899406984, + "grad_norm": 2.3250370025634766, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.702113926410675, + "num_tokens": 134361031.0, + "step": 5219 + }, + { + "epoch": 0.5732484076433121, + "grad_norm": 2.4403562545776367, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.721450686454773, + "num_tokens": 134381500.0, + "step": 5220 + }, + { + "epoch": 0.5733582253459257, + "grad_norm": 2.198143482208252, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.699515700340271, + "num_tokens": 134407180.0, + "step": 5221 + }, + { + "epoch": 0.5734680430485394, + "grad_norm": 2.1463875770568848, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6967955827713013, + "num_tokens": 134433087.0, + "step": 5222 + }, + { + "epoch": 0.5735778607511531, + "grad_norm": 2.462311029434204, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7147829532623291, + "num_tokens": 134454539.0, + "step": 5223 + }, + { + "epoch": 0.5736876784537668, + "grad_norm": 2.14212965965271, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.690049946308136, + "num_tokens": 134483820.0, + "step": 5224 + }, + { + "epoch": 0.5737974961563804, + "grad_norm": 2.0072615146636963, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6764807105064392, + "num_tokens": 134514455.0, + "step": 5225 + }, + { + "epoch": 0.5739073138589941, + "grad_norm": 2.150000810623169, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7058377265930176, + "num_tokens": 134542742.0, + "step": 5226 + }, + { + "epoch": 0.5740171315616077, + "grad_norm": 2.3765780925750732, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7150167226791382, + "num_tokens": 134564761.0, + "step": 5227 + }, + { + "epoch": 0.5741269492642214, + "grad_norm": 2.106070041656494, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6913547515869141, + "num_tokens": 134593630.0, + "step": 5228 + }, + { + "epoch": 0.574236766966835, + "grad_norm": 2.481736898422241, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7052359580993652, + "num_tokens": 134614287.0, + "step": 5229 + }, + { + "epoch": 0.5743465846694488, + "grad_norm": 2.015221118927002, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.6879304647445679, + "num_tokens": 134646823.0, + "step": 5230 + }, + { + "epoch": 0.5744564023720624, + "grad_norm": 2.2563090324401855, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7072679996490479, + "num_tokens": 134672159.0, + "step": 5231 + }, + { + "epoch": 0.5745662200746761, + "grad_norm": 2.4025611877441406, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7155013680458069, + "num_tokens": 134693255.0, + "step": 5232 + }, + { + "epoch": 0.5746760377772897, + "grad_norm": 2.305466651916504, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6828609704971313, + "num_tokens": 134717270.0, + "step": 5233 + }, + { + "epoch": 0.5747858554799034, + "grad_norm": 1.9758063554763794, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7099555730819702, + "num_tokens": 134748438.0, + "step": 5234 + }, + { + "epoch": 0.574895673182517, + "grad_norm": 2.262357234954834, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7117747664451599, + "num_tokens": 134772403.0, + "step": 5235 + }, + { + "epoch": 0.5750054908851306, + "grad_norm": 2.007534980773926, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7137957811355591, + "num_tokens": 134800945.0, + "step": 5236 + }, + { + "epoch": 0.5751153085877443, + "grad_norm": 2.520540714263916, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7134023308753967, + "num_tokens": 134820379.0, + "step": 5237 + }, + { + "epoch": 0.575225126290358, + "grad_norm": 1.9955233335494995, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7070976495742798, + "num_tokens": 134850220.0, + "step": 5238 + }, + { + "epoch": 0.5753349439929717, + "grad_norm": 2.2084996700286865, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.694059431552887, + "num_tokens": 134875883.0, + "step": 5239 + }, + { + "epoch": 0.5754447616955853, + "grad_norm": 2.016329050064087, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6979448795318604, + "num_tokens": 134907209.0, + "step": 5240 + }, + { + "epoch": 0.575554579398199, + "grad_norm": 2.4320499897003174, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7076902389526367, + "num_tokens": 134928630.0, + "step": 5241 + }, + { + "epoch": 0.5756643971008126, + "grad_norm": 1.8806403875350952, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6873224973678589, + "num_tokens": 134962808.0, + "step": 5242 + }, + { + "epoch": 0.5757742148034263, + "grad_norm": 2.4633102416992188, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.7002935409545898, + "num_tokens": 134987499.0, + "step": 5243 + }, + { + "epoch": 0.5758840325060399, + "grad_norm": 2.4892382621765137, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7235152721405029, + "num_tokens": 135007088.0, + "step": 5244 + }, + { + "epoch": 0.5759938502086537, + "grad_norm": 2.3556110858917236, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.713882565498352, + "num_tokens": 135029871.0, + "step": 5245 + }, + { + "epoch": 0.5761036679112673, + "grad_norm": 2.246401309967041, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7253682613372803, + "num_tokens": 135053315.0, + "step": 5246 + }, + { + "epoch": 0.576213485613881, + "grad_norm": 2.2457799911499023, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7044233083724976, + "num_tokens": 135077517.0, + "step": 5247 + }, + { + "epoch": 0.5763233033164946, + "grad_norm": 1.9785929918289185, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7004810571670532, + "num_tokens": 135106891.0, + "step": 5248 + }, + { + "epoch": 0.5764331210191083, + "grad_norm": 1.985498309135437, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.684539794921875, + "num_tokens": 135140641.0, + "step": 5249 + }, + { + "epoch": 0.5765429387217219, + "grad_norm": 2.398108959197998, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7044369578361511, + "num_tokens": 135164513.0, + "step": 5250 + }, + { + "epoch": 0.5766527564243356, + "grad_norm": 2.1981711387634277, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7013869285583496, + "num_tokens": 135189311.0, + "step": 5251 + }, + { + "epoch": 0.5767625741269493, + "grad_norm": 2.218975067138672, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6992676854133606, + "num_tokens": 135212875.0, + "step": 5252 + }, + { + "epoch": 0.576872391829563, + "grad_norm": 2.1122164726257324, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6962787508964539, + "num_tokens": 135240723.0, + "step": 5253 + }, + { + "epoch": 0.5769822095321766, + "grad_norm": 2.4199490547180176, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.704462468624115, + "num_tokens": 135265075.0, + "step": 5254 + }, + { + "epoch": 0.5770920272347903, + "grad_norm": 2.3256618976593018, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6907778978347778, + "num_tokens": 135288751.0, + "step": 5255 + }, + { + "epoch": 0.5772018449374039, + "grad_norm": 2.1661159992218018, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7063629031181335, + "num_tokens": 135314913.0, + "step": 5256 + }, + { + "epoch": 0.5773116626400175, + "grad_norm": 2.5350372791290283, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7192361950874329, + "num_tokens": 135336345.0, + "step": 5257 + }, + { + "epoch": 0.5774214803426312, + "grad_norm": 2.2729787826538086, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6988650560379028, + "num_tokens": 135362359.0, + "step": 5258 + }, + { + "epoch": 0.577531298045245, + "grad_norm": 2.167062759399414, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7224308252334595, + "num_tokens": 135388435.0, + "step": 5259 + }, + { + "epoch": 0.5776411157478586, + "grad_norm": 1.975696325302124, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7168865203857422, + "num_tokens": 135418404.0, + "step": 5260 + }, + { + "epoch": 0.5777509334504722, + "grad_norm": 1.9815491437911987, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6960601210594177, + "num_tokens": 135448720.0, + "step": 5261 + }, + { + "epoch": 0.5778607511530859, + "grad_norm": 2.538878917694092, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7165683507919312, + "num_tokens": 135467924.0, + "step": 5262 + }, + { + "epoch": 0.5779705688556995, + "grad_norm": 2.1476891040802, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7156299352645874, + "num_tokens": 135494399.0, + "step": 5263 + }, + { + "epoch": 0.5780803865583132, + "grad_norm": 2.3497838973999023, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7037805318832397, + "num_tokens": 135516978.0, + "step": 5264 + }, + { + "epoch": 0.5781902042609268, + "grad_norm": 2.615626335144043, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7096661329269409, + "num_tokens": 135536566.0, + "step": 5265 + }, + { + "epoch": 0.5783000219635406, + "grad_norm": 2.144845724105835, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7033333778381348, + "num_tokens": 135565265.0, + "step": 5266 + }, + { + "epoch": 0.5784098396661542, + "grad_norm": 2.301482915878296, + "learning_rate": 1e-06, + "loss": 1.1137, + "mean_token_accuracy": 0.6773727536201477, + "num_tokens": 135591031.0, + "step": 5267 + }, + { + "epoch": 0.5785196573687679, + "grad_norm": 1.9509730339050293, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7039376497268677, + "num_tokens": 135623107.0, + "step": 5268 + }, + { + "epoch": 0.5786294750713815, + "grad_norm": 2.6273298263549805, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6863434314727783, + "num_tokens": 135644145.0, + "step": 5269 + }, + { + "epoch": 0.5787392927739952, + "grad_norm": 2.312647581100464, + "learning_rate": 1e-06, + "loss": 1.0698, + "mean_token_accuracy": 0.6840113997459412, + "num_tokens": 135667847.0, + "step": 5270 + }, + { + "epoch": 0.5788491104766088, + "grad_norm": 2.143718957901001, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7017185688018799, + "num_tokens": 135694762.0, + "step": 5271 + }, + { + "epoch": 0.5789589281792225, + "grad_norm": 2.2119269371032715, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7083180546760559, + "num_tokens": 135719753.0, + "step": 5272 + }, + { + "epoch": 0.5790687458818361, + "grad_norm": 2.0099403858184814, + "learning_rate": 1e-06, + "loss": 1.0617, + "mean_token_accuracy": 0.6831653714179993, + "num_tokens": 135749786.0, + "step": 5273 + }, + { + "epoch": 0.5791785635844499, + "grad_norm": 2.3908042907714844, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7115368843078613, + "num_tokens": 135772176.0, + "step": 5274 + }, + { + "epoch": 0.5792883812870635, + "grad_norm": 2.372298240661621, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7305502891540527, + "num_tokens": 135793478.0, + "step": 5275 + }, + { + "epoch": 0.5793981989896771, + "grad_norm": 2.3927321434020996, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7170090675354004, + "num_tokens": 135814219.0, + "step": 5276 + }, + { + "epoch": 0.5795080166922908, + "grad_norm": 2.325742721557617, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.694810152053833, + "num_tokens": 135837450.0, + "step": 5277 + }, + { + "epoch": 0.5796178343949044, + "grad_norm": 2.3572332859039307, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7167083024978638, + "num_tokens": 135859392.0, + "step": 5278 + }, + { + "epoch": 0.5797276520975181, + "grad_norm": 2.0663397312164307, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6922125816345215, + "num_tokens": 135887678.0, + "step": 5279 + }, + { + "epoch": 0.5798374698001317, + "grad_norm": 2.4373910427093506, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6988775730133057, + "num_tokens": 135910902.0, + "step": 5280 + }, + { + "epoch": 0.5799472875027455, + "grad_norm": 2.132366895675659, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7081531882286072, + "num_tokens": 135936969.0, + "step": 5281 + }, + { + "epoch": 0.5800571052053591, + "grad_norm": 2.2645959854125977, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7280235290527344, + "num_tokens": 135959299.0, + "step": 5282 + }, + { + "epoch": 0.5801669229079728, + "grad_norm": 2.5746827125549316, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7091844081878662, + "num_tokens": 135979692.0, + "step": 5283 + }, + { + "epoch": 0.5802767406105864, + "grad_norm": 2.2806129455566406, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7075517177581787, + "num_tokens": 136003931.0, + "step": 5284 + }, + { + "epoch": 0.5803865583132001, + "grad_norm": 2.1594038009643555, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6922118067741394, + "num_tokens": 136031140.0, + "step": 5285 + }, + { + "epoch": 0.5804963760158137, + "grad_norm": 2.1358909606933594, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.711085855960846, + "num_tokens": 136056776.0, + "step": 5286 + }, + { + "epoch": 0.5806061937184274, + "grad_norm": 2.2829813957214355, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6927779912948608, + "num_tokens": 136082166.0, + "step": 5287 + }, + { + "epoch": 0.5807160114210411, + "grad_norm": 2.1196727752685547, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7018714547157288, + "num_tokens": 136109266.0, + "step": 5288 + }, + { + "epoch": 0.5808258291236548, + "grad_norm": 2.278066873550415, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7281255722045898, + "num_tokens": 136133573.0, + "step": 5289 + }, + { + "epoch": 0.5809356468262684, + "grad_norm": 2.057854413986206, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6726193428039551, + "num_tokens": 136162199.0, + "step": 5290 + }, + { + "epoch": 0.5810454645288821, + "grad_norm": 2.1099636554718018, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6964447498321533, + "num_tokens": 136189375.0, + "step": 5291 + }, + { + "epoch": 0.5811552822314957, + "grad_norm": 1.9987367391586304, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.732168436050415, + "num_tokens": 136218255.0, + "step": 5292 + }, + { + "epoch": 0.5812650999341094, + "grad_norm": 2.350383758544922, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.709116518497467, + "num_tokens": 136241183.0, + "step": 5293 + }, + { + "epoch": 0.581374917636723, + "grad_norm": 2.160886287689209, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7226578593254089, + "num_tokens": 136267866.0, + "step": 5294 + }, + { + "epoch": 0.5814847353393368, + "grad_norm": 2.1385416984558105, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6987581253051758, + "num_tokens": 136296502.0, + "step": 5295 + }, + { + "epoch": 0.5815945530419504, + "grad_norm": 2.0934479236602783, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7041693329811096, + "num_tokens": 136322733.0, + "step": 5296 + }, + { + "epoch": 0.581704370744564, + "grad_norm": 2.067849636077881, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7117340564727783, + "num_tokens": 136350900.0, + "step": 5297 + }, + { + "epoch": 0.5818141884471777, + "grad_norm": 2.3737173080444336, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7161388397216797, + "num_tokens": 136372789.0, + "step": 5298 + }, + { + "epoch": 0.5819240061497913, + "grad_norm": 2.2968075275421143, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6947469115257263, + "num_tokens": 136398406.0, + "step": 5299 + }, + { + "epoch": 0.582033823852405, + "grad_norm": 2.123352289199829, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6931778192520142, + "num_tokens": 136426678.0, + "step": 5300 + }, + { + "epoch": 0.5821436415550186, + "grad_norm": 1.9317986965179443, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7313314080238342, + "num_tokens": 136455783.0, + "step": 5301 + }, + { + "epoch": 0.5822534592576323, + "grad_norm": 1.9857879877090454, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6969115138053894, + "num_tokens": 136484847.0, + "step": 5302 + }, + { + "epoch": 0.582363276960246, + "grad_norm": 2.1502795219421387, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6966577768325806, + "num_tokens": 136510688.0, + "step": 5303 + }, + { + "epoch": 0.5824730946628597, + "grad_norm": 2.117825746536255, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7288480401039124, + "num_tokens": 136536398.0, + "step": 5304 + }, + { + "epoch": 0.5825829123654733, + "grad_norm": 2.160108804702759, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6857748627662659, + "num_tokens": 136564734.0, + "step": 5305 + }, + { + "epoch": 0.582692730068087, + "grad_norm": 2.1637535095214844, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7005780935287476, + "num_tokens": 136589893.0, + "step": 5306 + }, + { + "epoch": 0.5828025477707006, + "grad_norm": 2.115215301513672, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.6789267063140869, + "num_tokens": 136616361.0, + "step": 5307 + }, + { + "epoch": 0.5829123654733143, + "grad_norm": 2.0384562015533447, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.6757165789604187, + "num_tokens": 136646733.0, + "step": 5308 + }, + { + "epoch": 0.5830221831759279, + "grad_norm": 2.1139163970947266, + "learning_rate": 1e-06, + "loss": 1.0832, + "mean_token_accuracy": 0.6788966059684753, + "num_tokens": 136675223.0, + "step": 5309 + }, + { + "epoch": 0.5831320008785417, + "grad_norm": 2.4098737239837646, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7158300280570984, + "num_tokens": 136695573.0, + "step": 5310 + }, + { + "epoch": 0.5832418185811553, + "grad_norm": 2.121541976928711, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7016571760177612, + "num_tokens": 136721922.0, + "step": 5311 + }, + { + "epoch": 0.583351636283769, + "grad_norm": 2.0495448112487793, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7047685980796814, + "num_tokens": 136750931.0, + "step": 5312 + }, + { + "epoch": 0.5834614539863826, + "grad_norm": 2.3513965606689453, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7088062763214111, + "num_tokens": 136773669.0, + "step": 5313 + }, + { + "epoch": 0.5835712716889963, + "grad_norm": 2.1964433193206787, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.7001955509185791, + "num_tokens": 136799370.0, + "step": 5314 + }, + { + "epoch": 0.5836810893916099, + "grad_norm": 2.0383529663085938, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.688642144203186, + "num_tokens": 136827319.0, + "step": 5315 + }, + { + "epoch": 0.5837909070942235, + "grad_norm": 2.460329294204712, + "learning_rate": 1e-06, + "loss": 1.063, + "mean_token_accuracy": 0.6814080476760864, + "num_tokens": 136848595.0, + "step": 5316 + }, + { + "epoch": 0.5839007247968373, + "grad_norm": 2.266793966293335, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.708909809589386, + "num_tokens": 136871372.0, + "step": 5317 + }, + { + "epoch": 0.584010542499451, + "grad_norm": 2.2075326442718506, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7051554918289185, + "num_tokens": 136896433.0, + "step": 5318 + }, + { + "epoch": 0.5841203602020646, + "grad_norm": 2.093926429748535, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7235754728317261, + "num_tokens": 136922367.0, + "step": 5319 + }, + { + "epoch": 0.5842301779046782, + "grad_norm": 2.2370831966400146, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7031154632568359, + "num_tokens": 136946442.0, + "step": 5320 + }, + { + "epoch": 0.5843399956072919, + "grad_norm": 2.0254557132720947, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6966606378555298, + "num_tokens": 136975787.0, + "step": 5321 + }, + { + "epoch": 0.5844498133099055, + "grad_norm": 2.3171286582946777, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6925961971282959, + "num_tokens": 137001631.0, + "step": 5322 + }, + { + "epoch": 0.5845596310125192, + "grad_norm": 1.8195178508758545, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.680820107460022, + "num_tokens": 137037065.0, + "step": 5323 + }, + { + "epoch": 0.5846694487151329, + "grad_norm": 1.96135675907135, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6866723299026489, + "num_tokens": 137069238.0, + "step": 5324 + }, + { + "epoch": 0.5847792664177466, + "grad_norm": 1.96342134475708, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6928817629814148, + "num_tokens": 137101121.0, + "step": 5325 + }, + { + "epoch": 0.5848890841203602, + "grad_norm": 2.3235442638397217, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7155277729034424, + "num_tokens": 137123480.0, + "step": 5326 + }, + { + "epoch": 0.5849989018229739, + "grad_norm": 2.3418190479278564, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7097792625427246, + "num_tokens": 137146949.0, + "step": 5327 + }, + { + "epoch": 0.5851087195255875, + "grad_norm": 2.2577052116394043, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7097369432449341, + "num_tokens": 137170679.0, + "step": 5328 + }, + { + "epoch": 0.5852185372282012, + "grad_norm": 2.41196608543396, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7262088060379028, + "num_tokens": 137191222.0, + "step": 5329 + }, + { + "epoch": 0.5853283549308148, + "grad_norm": 2.412323474884033, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6869047284126282, + "num_tokens": 137214075.0, + "step": 5330 + }, + { + "epoch": 0.5854381726334285, + "grad_norm": 2.2350504398345947, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7264227867126465, + "num_tokens": 137236956.0, + "step": 5331 + }, + { + "epoch": 0.5855479903360422, + "grad_norm": 2.294851303100586, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7372338771820068, + "num_tokens": 137260422.0, + "step": 5332 + }, + { + "epoch": 0.5856578080386559, + "grad_norm": 1.9785239696502686, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6985089778900146, + "num_tokens": 137290371.0, + "step": 5333 + }, + { + "epoch": 0.5857676257412695, + "grad_norm": 2.361194372177124, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.717978835105896, + "num_tokens": 137311228.0, + "step": 5334 + }, + { + "epoch": 0.5858774434438832, + "grad_norm": 2.2082037925720215, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.694398045539856, + "num_tokens": 137338733.0, + "step": 5335 + }, + { + "epoch": 0.5859872611464968, + "grad_norm": 2.2724292278289795, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6837853193283081, + "num_tokens": 137365267.0, + "step": 5336 + }, + { + "epoch": 0.5860970788491104, + "grad_norm": 2.022643566131592, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6791330575942993, + "num_tokens": 137395757.0, + "step": 5337 + }, + { + "epoch": 0.5862068965517241, + "grad_norm": 2.2987587451934814, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7341330051422119, + "num_tokens": 137419094.0, + "step": 5338 + }, + { + "epoch": 0.5863167142543378, + "grad_norm": 2.1753969192504883, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6970250606536865, + "num_tokens": 137446163.0, + "step": 5339 + }, + { + "epoch": 0.5864265319569515, + "grad_norm": 1.942915439605713, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7072550058364868, + "num_tokens": 137475396.0, + "step": 5340 + }, + { + "epoch": 0.5865363496595651, + "grad_norm": 2.2190604209899902, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7040928602218628, + "num_tokens": 137501158.0, + "step": 5341 + }, + { + "epoch": 0.5866461673621788, + "grad_norm": 2.3194191455841064, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6905766725540161, + "num_tokens": 137523681.0, + "step": 5342 + }, + { + "epoch": 0.5867559850647924, + "grad_norm": 2.1498119831085205, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6997305750846863, + "num_tokens": 137552549.0, + "step": 5343 + }, + { + "epoch": 0.5868658027674061, + "grad_norm": 2.008928060531616, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6937044262886047, + "num_tokens": 137583806.0, + "step": 5344 + }, + { + "epoch": 0.5869756204700197, + "grad_norm": 2.1660783290863037, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.731738805770874, + "num_tokens": 137609007.0, + "step": 5345 + }, + { + "epoch": 0.5870854381726335, + "grad_norm": 2.3205597400665283, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7230682373046875, + "num_tokens": 137630760.0, + "step": 5346 + }, + { + "epoch": 0.5871952558752471, + "grad_norm": 2.4796605110168457, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.7071648836135864, + "num_tokens": 137652789.0, + "step": 5347 + }, + { + "epoch": 0.5873050735778608, + "grad_norm": 2.08811354637146, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6784963011741638, + "num_tokens": 137683496.0, + "step": 5348 + }, + { + "epoch": 0.5874148912804744, + "grad_norm": 2.0297961235046387, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6971031427383423, + "num_tokens": 137712593.0, + "step": 5349 + }, + { + "epoch": 0.5875247089830881, + "grad_norm": 1.9956598281860352, + "learning_rate": 1e-06, + "loss": 1.1095, + "mean_token_accuracy": 0.6668800115585327, + "num_tokens": 137742867.0, + "step": 5350 + }, + { + "epoch": 0.5876345266857017, + "grad_norm": 2.033719062805176, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7128481268882751, + "num_tokens": 137771189.0, + "step": 5351 + }, + { + "epoch": 0.5877443443883154, + "grad_norm": 2.3396596908569336, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7220246195793152, + "num_tokens": 137792790.0, + "step": 5352 + }, + { + "epoch": 0.5878541620909291, + "grad_norm": 2.140986680984497, + "learning_rate": 1e-06, + "loss": 1.0899, + "mean_token_accuracy": 0.6937487721443176, + "num_tokens": 137818682.0, + "step": 5353 + }, + { + "epoch": 0.5879639797935428, + "grad_norm": 2.082979202270508, + "learning_rate": 1e-06, + "loss": 1.0702, + "mean_token_accuracy": 0.6757440567016602, + "num_tokens": 137846750.0, + "step": 5354 + }, + { + "epoch": 0.5880737974961564, + "grad_norm": 2.1515841484069824, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7038067579269409, + "num_tokens": 137872043.0, + "step": 5355 + }, + { + "epoch": 0.58818361519877, + "grad_norm": 2.199004888534546, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6970498561859131, + "num_tokens": 137896696.0, + "step": 5356 + }, + { + "epoch": 0.5882934329013837, + "grad_norm": 2.6063895225524902, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7193692922592163, + "num_tokens": 137915972.0, + "step": 5357 + }, + { + "epoch": 0.5884032506039973, + "grad_norm": 2.354140281677246, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.7041553854942322, + "num_tokens": 137940168.0, + "step": 5358 + }, + { + "epoch": 0.588513068306611, + "grad_norm": 2.5930845737457275, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.703303337097168, + "num_tokens": 137960517.0, + "step": 5359 + }, + { + "epoch": 0.5886228860092246, + "grad_norm": 2.066624402999878, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6954680681228638, + "num_tokens": 137989350.0, + "step": 5360 + }, + { + "epoch": 0.5887327037118384, + "grad_norm": 1.9680794477462769, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7466623187065125, + "num_tokens": 138018674.0, + "step": 5361 + }, + { + "epoch": 0.588842521414452, + "grad_norm": 2.3545055389404297, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7035399079322815, + "num_tokens": 138040959.0, + "step": 5362 + }, + { + "epoch": 0.5889523391170657, + "grad_norm": 2.053934097290039, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7165836095809937, + "num_tokens": 138068738.0, + "step": 5363 + }, + { + "epoch": 0.5890621568196793, + "grad_norm": 2.2589733600616455, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7243616580963135, + "num_tokens": 138090300.0, + "step": 5364 + }, + { + "epoch": 0.589171974522293, + "grad_norm": 2.3892314434051514, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7274482250213623, + "num_tokens": 138111478.0, + "step": 5365 + }, + { + "epoch": 0.5892817922249066, + "grad_norm": 2.399627685546875, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7374736070632935, + "num_tokens": 138132654.0, + "step": 5366 + }, + { + "epoch": 0.5893916099275203, + "grad_norm": 2.496438503265381, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7151108384132385, + "num_tokens": 138152736.0, + "step": 5367 + }, + { + "epoch": 0.589501427630134, + "grad_norm": 2.1848247051239014, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.711807131767273, + "num_tokens": 138177992.0, + "step": 5368 + }, + { + "epoch": 0.5896112453327477, + "grad_norm": 2.1935935020446777, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6904313564300537, + "num_tokens": 138203049.0, + "step": 5369 + }, + { + "epoch": 0.5897210630353613, + "grad_norm": 2.487518310546875, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7100975513458252, + "num_tokens": 138225046.0, + "step": 5370 + }, + { + "epoch": 0.589830880737975, + "grad_norm": 1.9386636018753052, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7014490365982056, + "num_tokens": 138255885.0, + "step": 5371 + }, + { + "epoch": 0.5899406984405886, + "grad_norm": 2.8777921199798584, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7308691740036011, + "num_tokens": 138271169.0, + "step": 5372 + }, + { + "epoch": 0.5900505161432023, + "grad_norm": 2.0825464725494385, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6970821619033813, + "num_tokens": 138301025.0, + "step": 5373 + }, + { + "epoch": 0.5901603338458159, + "grad_norm": 2.058997392654419, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7247777581214905, + "num_tokens": 138327118.0, + "step": 5374 + }, + { + "epoch": 0.5902701515484297, + "grad_norm": 2.139359474182129, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6953048706054688, + "num_tokens": 138355161.0, + "step": 5375 + }, + { + "epoch": 0.5903799692510433, + "grad_norm": 2.313934087753296, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.696478545665741, + "num_tokens": 138380954.0, + "step": 5376 + }, + { + "epoch": 0.590489786953657, + "grad_norm": 2.0454840660095215, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7068077325820923, + "num_tokens": 138408356.0, + "step": 5377 + }, + { + "epoch": 0.5905996046562706, + "grad_norm": 2.083336591720581, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7045008540153503, + "num_tokens": 138435676.0, + "step": 5378 + }, + { + "epoch": 0.5907094223588842, + "grad_norm": 2.1203179359436035, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.701743483543396, + "num_tokens": 138463412.0, + "step": 5379 + }, + { + "epoch": 0.5908192400614979, + "grad_norm": 2.0338120460510254, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7156304121017456, + "num_tokens": 138492513.0, + "step": 5380 + }, + { + "epoch": 0.5909290577641115, + "grad_norm": 1.8887745141983032, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.702030599117279, + "num_tokens": 138524352.0, + "step": 5381 + }, + { + "epoch": 0.5910388754667253, + "grad_norm": 2.1170647144317627, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7005959749221802, + "num_tokens": 138551873.0, + "step": 5382 + }, + { + "epoch": 0.5911486931693389, + "grad_norm": 2.1692614555358887, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6956414580345154, + "num_tokens": 138579196.0, + "step": 5383 + }, + { + "epoch": 0.5912585108719526, + "grad_norm": 2.407569646835327, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7097968459129333, + "num_tokens": 138600243.0, + "step": 5384 + }, + { + "epoch": 0.5913683285745662, + "grad_norm": 2.124177932739258, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6866886615753174, + "num_tokens": 138629215.0, + "step": 5385 + }, + { + "epoch": 0.5914781462771799, + "grad_norm": 2.1689467430114746, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.696135938167572, + "num_tokens": 138656009.0, + "step": 5386 + }, + { + "epoch": 0.5915879639797935, + "grad_norm": 2.226635456085205, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.679201066493988, + "num_tokens": 138679864.0, + "step": 5387 + }, + { + "epoch": 0.5916977816824072, + "grad_norm": 2.026318311691284, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6951813697814941, + "num_tokens": 138709761.0, + "step": 5388 + }, + { + "epoch": 0.5918075993850208, + "grad_norm": 2.371941089630127, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6970239877700806, + "num_tokens": 138733582.0, + "step": 5389 + }, + { + "epoch": 0.5919174170876346, + "grad_norm": 2.3480987548828125, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7065790891647339, + "num_tokens": 138757353.0, + "step": 5390 + }, + { + "epoch": 0.5920272347902482, + "grad_norm": 2.2050294876098633, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7155795693397522, + "num_tokens": 138782307.0, + "step": 5391 + }, + { + "epoch": 0.5921370524928619, + "grad_norm": 2.145496368408203, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7102212905883789, + "num_tokens": 138809121.0, + "step": 5392 + }, + { + "epoch": 0.5922468701954755, + "grad_norm": 1.9172985553741455, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6865330338478088, + "num_tokens": 138841562.0, + "step": 5393 + }, + { + "epoch": 0.5923566878980892, + "grad_norm": 2.105372428894043, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7034549713134766, + "num_tokens": 138867428.0, + "step": 5394 + }, + { + "epoch": 0.5924665056007028, + "grad_norm": 2.0810606479644775, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.707544207572937, + "num_tokens": 138895651.0, + "step": 5395 + }, + { + "epoch": 0.5925763233033164, + "grad_norm": 2.475501537322998, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6911964416503906, + "num_tokens": 138918108.0, + "step": 5396 + }, + { + "epoch": 0.5926861410059302, + "grad_norm": 2.3460965156555176, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.716007649898529, + "num_tokens": 138941330.0, + "step": 5397 + }, + { + "epoch": 0.5927959587085438, + "grad_norm": 2.1805341243743896, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6869828701019287, + "num_tokens": 138966800.0, + "step": 5398 + }, + { + "epoch": 0.5929057764111575, + "grad_norm": 2.1267611980438232, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7125257849693298, + "num_tokens": 138993749.0, + "step": 5399 + }, + { + "epoch": 0.5930155941137711, + "grad_norm": 2.3854002952575684, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.695305347442627, + "num_tokens": 139015169.0, + "step": 5400 + }, + { + "epoch": 0.5931254118163848, + "grad_norm": 2.379317283630371, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7189632654190063, + "num_tokens": 139038103.0, + "step": 5401 + }, + { + "epoch": 0.5932352295189984, + "grad_norm": 2.0864548683166504, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7060689926147461, + "num_tokens": 139065083.0, + "step": 5402 + }, + { + "epoch": 0.5933450472216121, + "grad_norm": 2.3355472087860107, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.711101770401001, + "num_tokens": 139087730.0, + "step": 5403 + }, + { + "epoch": 0.5934548649242258, + "grad_norm": 2.0042576789855957, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7219759225845337, + "num_tokens": 139117444.0, + "step": 5404 + }, + { + "epoch": 0.5935646826268395, + "grad_norm": 2.178546667098999, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6980332136154175, + "num_tokens": 139144382.0, + "step": 5405 + }, + { + "epoch": 0.5936745003294531, + "grad_norm": 2.3984785079956055, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7013558149337769, + "num_tokens": 139166914.0, + "step": 5406 + }, + { + "epoch": 0.5937843180320668, + "grad_norm": 2.3161845207214355, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7122682332992554, + "num_tokens": 139189696.0, + "step": 5407 + }, + { + "epoch": 0.5938941357346804, + "grad_norm": 2.2390875816345215, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7075129151344299, + "num_tokens": 139213578.0, + "step": 5408 + }, + { + "epoch": 0.5940039534372941, + "grad_norm": 2.093395948410034, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7017063498497009, + "num_tokens": 139240706.0, + "step": 5409 + }, + { + "epoch": 0.5941137711399077, + "grad_norm": 2.2463231086730957, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6809152364730835, + "num_tokens": 139264844.0, + "step": 5410 + }, + { + "epoch": 0.5942235888425215, + "grad_norm": 2.3625001907348633, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7233172655105591, + "num_tokens": 139283954.0, + "step": 5411 + }, + { + "epoch": 0.5943334065451351, + "grad_norm": 2.0052285194396973, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7081843614578247, + "num_tokens": 139311106.0, + "step": 5412 + }, + { + "epoch": 0.5944432242477488, + "grad_norm": 2.1036527156829834, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6869479417800903, + "num_tokens": 139338626.0, + "step": 5413 + }, + { + "epoch": 0.5945530419503624, + "grad_norm": 1.9730461835861206, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6909639835357666, + "num_tokens": 139369141.0, + "step": 5414 + }, + { + "epoch": 0.594662859652976, + "grad_norm": 2.339214563369751, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7143838405609131, + "num_tokens": 139392255.0, + "step": 5415 + }, + { + "epoch": 0.5947726773555897, + "grad_norm": 2.3877010345458984, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6929175853729248, + "num_tokens": 139416705.0, + "step": 5416 + }, + { + "epoch": 0.5948824950582033, + "grad_norm": 2.1053669452667236, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7185138463973999, + "num_tokens": 139444879.0, + "step": 5417 + }, + { + "epoch": 0.5949923127608171, + "grad_norm": 2.2376763820648193, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7086681127548218, + "num_tokens": 139468172.0, + "step": 5418 + }, + { + "epoch": 0.5951021304634307, + "grad_norm": 1.9524986743927002, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.740363597869873, + "num_tokens": 139497411.0, + "step": 5419 + }, + { + "epoch": 0.5952119481660444, + "grad_norm": 2.105961561203003, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.708338737487793, + "num_tokens": 139523330.0, + "step": 5420 + }, + { + "epoch": 0.595321765868658, + "grad_norm": 1.9844019412994385, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.7010912895202637, + "num_tokens": 139552882.0, + "step": 5421 + }, + { + "epoch": 0.5954315835712717, + "grad_norm": 2.2017173767089844, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6839405298233032, + "num_tokens": 139578994.0, + "step": 5422 + }, + { + "epoch": 0.5955414012738853, + "grad_norm": 2.438736915588379, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7028607130050659, + "num_tokens": 139599576.0, + "step": 5423 + }, + { + "epoch": 0.595651218976499, + "grad_norm": 2.17228102684021, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7162025570869446, + "num_tokens": 139624725.0, + "step": 5424 + }, + { + "epoch": 0.5957610366791126, + "grad_norm": 2.4658327102661133, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6988799571990967, + "num_tokens": 139647081.0, + "step": 5425 + }, + { + "epoch": 0.5958708543817264, + "grad_norm": 2.029344320297241, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6937432289123535, + "num_tokens": 139676708.0, + "step": 5426 + }, + { + "epoch": 0.59598067208434, + "grad_norm": 2.2365610599517822, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7136905193328857, + "num_tokens": 139701109.0, + "step": 5427 + }, + { + "epoch": 0.5960904897869537, + "grad_norm": 2.137690305709839, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7049379944801331, + "num_tokens": 139730215.0, + "step": 5428 + }, + { + "epoch": 0.5962003074895673, + "grad_norm": 2.055769443511963, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6919598579406738, + "num_tokens": 139759396.0, + "step": 5429 + }, + { + "epoch": 0.596310125192181, + "grad_norm": 2.442548990249634, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7132528424263, + "num_tokens": 139782144.0, + "step": 5430 + }, + { + "epoch": 0.5964199428947946, + "grad_norm": 2.362751007080078, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7048777341842651, + "num_tokens": 139804241.0, + "step": 5431 + }, + { + "epoch": 0.5965297605974083, + "grad_norm": 2.147951126098633, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7211787104606628, + "num_tokens": 139832417.0, + "step": 5432 + }, + { + "epoch": 0.596639578300022, + "grad_norm": 2.231588363647461, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7007128000259399, + "num_tokens": 139857330.0, + "step": 5433 + }, + { + "epoch": 0.5967493960026357, + "grad_norm": 2.1479239463806152, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7173830270767212, + "num_tokens": 139884117.0, + "step": 5434 + }, + { + "epoch": 0.5968592137052493, + "grad_norm": 2.0674726963043213, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7234981656074524, + "num_tokens": 139910634.0, + "step": 5435 + }, + { + "epoch": 0.596969031407863, + "grad_norm": 2.3007593154907227, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6846430897712708, + "num_tokens": 139936056.0, + "step": 5436 + }, + { + "epoch": 0.5970788491104766, + "grad_norm": 2.2639319896698, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7102471590042114, + "num_tokens": 139960472.0, + "step": 5437 + }, + { + "epoch": 0.5971886668130902, + "grad_norm": 2.1884891986846924, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6920590400695801, + "num_tokens": 139985477.0, + "step": 5438 + }, + { + "epoch": 0.5972984845157039, + "grad_norm": 2.1913950443267822, + "learning_rate": 1e-06, + "loss": 1.1148, + "mean_token_accuracy": 0.6661325693130493, + "num_tokens": 140013044.0, + "step": 5439 + }, + { + "epoch": 0.5974083022183176, + "grad_norm": 2.223731517791748, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6978908181190491, + "num_tokens": 140038454.0, + "step": 5440 + }, + { + "epoch": 0.5975181199209313, + "grad_norm": 2.155489444732666, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7083426713943481, + "num_tokens": 140064232.0, + "step": 5441 + }, + { + "epoch": 0.5976279376235449, + "grad_norm": 2.2787282466888428, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6885802745819092, + "num_tokens": 140088177.0, + "step": 5442 + }, + { + "epoch": 0.5977377553261586, + "grad_norm": 2.06845760345459, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6949034929275513, + "num_tokens": 140114512.0, + "step": 5443 + }, + { + "epoch": 0.5978475730287722, + "grad_norm": 2.4259588718414307, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7010600566864014, + "num_tokens": 140136726.0, + "step": 5444 + }, + { + "epoch": 0.5979573907313859, + "grad_norm": 2.2046709060668945, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.685267448425293, + "num_tokens": 140162334.0, + "step": 5445 + }, + { + "epoch": 0.5980672084339995, + "grad_norm": 1.925433874130249, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6905105113983154, + "num_tokens": 140196211.0, + "step": 5446 + }, + { + "epoch": 0.5981770261366133, + "grad_norm": 2.2367537021636963, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6783791184425354, + "num_tokens": 140221485.0, + "step": 5447 + }, + { + "epoch": 0.5982868438392269, + "grad_norm": 2.1761059761047363, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7011045813560486, + "num_tokens": 140245979.0, + "step": 5448 + }, + { + "epoch": 0.5983966615418406, + "grad_norm": 2.100475549697876, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6924446225166321, + "num_tokens": 140274610.0, + "step": 5449 + }, + { + "epoch": 0.5985064792444542, + "grad_norm": 2.4583516120910645, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7155536413192749, + "num_tokens": 140295127.0, + "step": 5450 + }, + { + "epoch": 0.5986162969470679, + "grad_norm": 2.387873411178589, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6967875361442566, + "num_tokens": 140317119.0, + "step": 5451 + }, + { + "epoch": 0.5987261146496815, + "grad_norm": 2.3894271850585938, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7198862433433533, + "num_tokens": 140337602.0, + "step": 5452 + }, + { + "epoch": 0.5988359323522952, + "grad_norm": 2.015653371810913, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.699677586555481, + "num_tokens": 140369167.0, + "step": 5453 + }, + { + "epoch": 0.5989457500549088, + "grad_norm": 1.950548529624939, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6971917152404785, + "num_tokens": 140399704.0, + "step": 5454 + }, + { + "epoch": 0.5990555677575226, + "grad_norm": 2.1364946365356445, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6886358857154846, + "num_tokens": 140426984.0, + "step": 5455 + }, + { + "epoch": 0.5991653854601362, + "grad_norm": 2.083252429962158, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.6800728440284729, + "num_tokens": 140454291.0, + "step": 5456 + }, + { + "epoch": 0.5992752031627498, + "grad_norm": 1.907707929611206, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7044368982315063, + "num_tokens": 140485501.0, + "step": 5457 + }, + { + "epoch": 0.5993850208653635, + "grad_norm": 2.360769271850586, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7072993516921997, + "num_tokens": 140508869.0, + "step": 5458 + }, + { + "epoch": 0.5994948385679771, + "grad_norm": 1.9770125150680542, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7015427350997925, + "num_tokens": 140538213.0, + "step": 5459 + }, + { + "epoch": 0.5996046562705908, + "grad_norm": 2.1372694969177246, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7272689342498779, + "num_tokens": 140562465.0, + "step": 5460 + }, + { + "epoch": 0.5997144739732044, + "grad_norm": 1.870388150215149, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6972029209136963, + "num_tokens": 140593642.0, + "step": 5461 + }, + { + "epoch": 0.5998242916758182, + "grad_norm": 2.076321601867676, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6855310201644897, + "num_tokens": 140624860.0, + "step": 5462 + }, + { + "epoch": 0.5999341093784318, + "grad_norm": 2.405484914779663, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7130697965621948, + "num_tokens": 140647788.0, + "step": 5463 + }, + { + "epoch": 0.6000439270810455, + "grad_norm": 2.250593423843384, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6955980062484741, + "num_tokens": 140670787.0, + "step": 5464 + }, + { + "epoch": 0.6001537447836591, + "grad_norm": 2.2778539657592773, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6941965818405151, + "num_tokens": 140696790.0, + "step": 5465 + }, + { + "epoch": 0.6002635624862728, + "grad_norm": 2.0357041358947754, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7152079939842224, + "num_tokens": 140726125.0, + "step": 5466 + }, + { + "epoch": 0.6003733801888864, + "grad_norm": 1.9177738428115845, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7228529453277588, + "num_tokens": 140758532.0, + "step": 5467 + }, + { + "epoch": 0.6004831978915001, + "grad_norm": 2.56416916847229, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7223901748657227, + "num_tokens": 140777700.0, + "step": 5468 + }, + { + "epoch": 0.6005930155941138, + "grad_norm": 2.3575146198272705, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6868908405303955, + "num_tokens": 140802791.0, + "step": 5469 + }, + { + "epoch": 0.6007028332967275, + "grad_norm": 2.113635778427124, + "learning_rate": 1e-06, + "loss": 1.094, + "mean_token_accuracy": 0.6719286441802979, + "num_tokens": 140829965.0, + "step": 5470 + }, + { + "epoch": 0.6008126509993411, + "grad_norm": 2.323582410812378, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6992931365966797, + "num_tokens": 140854755.0, + "step": 5471 + }, + { + "epoch": 0.6009224687019548, + "grad_norm": 2.12166166305542, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7113686203956604, + "num_tokens": 140880678.0, + "step": 5472 + }, + { + "epoch": 0.6010322864045684, + "grad_norm": 2.3254714012145996, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6828630566596985, + "num_tokens": 140907544.0, + "step": 5473 + }, + { + "epoch": 0.601142104107182, + "grad_norm": 2.1053128242492676, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7105059623718262, + "num_tokens": 140932719.0, + "step": 5474 + }, + { + "epoch": 0.6012519218097957, + "grad_norm": 2.4373369216918945, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6947250366210938, + "num_tokens": 140954690.0, + "step": 5475 + }, + { + "epoch": 0.6013617395124095, + "grad_norm": 2.416846990585327, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7174506187438965, + "num_tokens": 140976663.0, + "step": 5476 + }, + { + "epoch": 0.6014715572150231, + "grad_norm": 2.3046553134918213, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7079954147338867, + "num_tokens": 140999837.0, + "step": 5477 + }, + { + "epoch": 0.6015813749176367, + "grad_norm": 2.2890377044677734, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6831470131874084, + "num_tokens": 141022717.0, + "step": 5478 + }, + { + "epoch": 0.6016911926202504, + "grad_norm": 2.0566320419311523, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6937268972396851, + "num_tokens": 141051661.0, + "step": 5479 + }, + { + "epoch": 0.601801010322864, + "grad_norm": 2.3917901515960693, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6918889880180359, + "num_tokens": 141072698.0, + "step": 5480 + }, + { + "epoch": 0.6019108280254777, + "grad_norm": 2.187246322631836, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7102338075637817, + "num_tokens": 141097349.0, + "step": 5481 + }, + { + "epoch": 0.6020206457280913, + "grad_norm": 2.294299364089966, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.719419002532959, + "num_tokens": 141119333.0, + "step": 5482 + }, + { + "epoch": 0.602130463430705, + "grad_norm": 2.431873083114624, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7135331630706787, + "num_tokens": 141140885.0, + "step": 5483 + }, + { + "epoch": 0.6022402811333187, + "grad_norm": 2.646894693374634, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7184348702430725, + "num_tokens": 141159910.0, + "step": 5484 + }, + { + "epoch": 0.6023500988359324, + "grad_norm": 2.2697224617004395, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6949145197868347, + "num_tokens": 141186856.0, + "step": 5485 + }, + { + "epoch": 0.602459916538546, + "grad_norm": 2.212191581726074, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.701066792011261, + "num_tokens": 141211158.0, + "step": 5486 + }, + { + "epoch": 0.6025697342411597, + "grad_norm": 2.335495710372925, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7145484685897827, + "num_tokens": 141233223.0, + "step": 5487 + }, + { + "epoch": 0.6026795519437733, + "grad_norm": 2.680920362472534, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7384583950042725, + "num_tokens": 141250643.0, + "step": 5488 + }, + { + "epoch": 0.602789369646387, + "grad_norm": 2.1502158641815186, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6979119777679443, + "num_tokens": 141277318.0, + "step": 5489 + }, + { + "epoch": 0.6028991873490006, + "grad_norm": 2.320557117462158, + "learning_rate": 1e-06, + "loss": 1.0753, + "mean_token_accuracy": 0.682003378868103, + "num_tokens": 141304899.0, + "step": 5490 + }, + { + "epoch": 0.6030090050516144, + "grad_norm": 2.314683437347412, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7294472455978394, + "num_tokens": 141327206.0, + "step": 5491 + }, + { + "epoch": 0.603118822754228, + "grad_norm": 2.1971542835235596, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7050963640213013, + "num_tokens": 141353289.0, + "step": 5492 + }, + { + "epoch": 0.6032286404568417, + "grad_norm": 2.243492364883423, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7065334320068359, + "num_tokens": 141377032.0, + "step": 5493 + }, + { + "epoch": 0.6033384581594553, + "grad_norm": 2.070922374725342, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7183670997619629, + "num_tokens": 141404794.0, + "step": 5494 + }, + { + "epoch": 0.603448275862069, + "grad_norm": 1.9825572967529297, + "learning_rate": 1e-06, + "loss": 1.0783, + "mean_token_accuracy": 0.6851515173912048, + "num_tokens": 141437494.0, + "step": 5495 + }, + { + "epoch": 0.6035580935646826, + "grad_norm": 2.182988166809082, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6862902045249939, + "num_tokens": 141462860.0, + "step": 5496 + }, + { + "epoch": 0.6036679112672962, + "grad_norm": 2.3075969219207764, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7152727842330933, + "num_tokens": 141485367.0, + "step": 5497 + }, + { + "epoch": 0.60377772896991, + "grad_norm": 2.3966546058654785, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7230390310287476, + "num_tokens": 141506242.0, + "step": 5498 + }, + { + "epoch": 0.6038875466725236, + "grad_norm": 2.2057535648345947, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6999926567077637, + "num_tokens": 141532395.0, + "step": 5499 + }, + { + "epoch": 0.6039973643751373, + "grad_norm": 2.1863222122192383, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6871020793914795, + "num_tokens": 141558664.0, + "step": 5500 + }, + { + "epoch": 0.6041071820777509, + "grad_norm": 2.0342679023742676, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.711763322353363, + "num_tokens": 141585516.0, + "step": 5501 + }, + { + "epoch": 0.6042169997803646, + "grad_norm": 1.9668405055999756, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7124226689338684, + "num_tokens": 141614784.0, + "step": 5502 + }, + { + "epoch": 0.6043268174829782, + "grad_norm": 2.2247838973999023, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.717583417892456, + "num_tokens": 141638641.0, + "step": 5503 + }, + { + "epoch": 0.6044366351855919, + "grad_norm": 2.183992862701416, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7077271342277527, + "num_tokens": 141663261.0, + "step": 5504 + }, + { + "epoch": 0.6045464528882056, + "grad_norm": 1.9672027826309204, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6911789178848267, + "num_tokens": 141697230.0, + "step": 5505 + }, + { + "epoch": 0.6046562705908193, + "grad_norm": 2.165651798248291, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6921176910400391, + "num_tokens": 141725914.0, + "step": 5506 + }, + { + "epoch": 0.6047660882934329, + "grad_norm": 2.0673389434814453, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7058315873146057, + "num_tokens": 141753531.0, + "step": 5507 + }, + { + "epoch": 0.6048759059960466, + "grad_norm": 2.202962875366211, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7197459936141968, + "num_tokens": 141778535.0, + "step": 5508 + }, + { + "epoch": 0.6049857236986602, + "grad_norm": 2.25372314453125, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7170352935791016, + "num_tokens": 141802711.0, + "step": 5509 + }, + { + "epoch": 0.6050955414012739, + "grad_norm": 2.0823469161987305, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6968214511871338, + "num_tokens": 141830464.0, + "step": 5510 + }, + { + "epoch": 0.6052053591038875, + "grad_norm": 2.0166079998016357, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7245376706123352, + "num_tokens": 141860382.0, + "step": 5511 + }, + { + "epoch": 0.6053151768065012, + "grad_norm": 2.333045482635498, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7068729996681213, + "num_tokens": 141881697.0, + "step": 5512 + }, + { + "epoch": 0.6054249945091149, + "grad_norm": 2.0705747604370117, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6968880891799927, + "num_tokens": 141910997.0, + "step": 5513 + }, + { + "epoch": 0.6055348122117286, + "grad_norm": 2.620884895324707, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7122631072998047, + "num_tokens": 141931260.0, + "step": 5514 + }, + { + "epoch": 0.6056446299143422, + "grad_norm": 2.1100494861602783, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6849205493927002, + "num_tokens": 141960443.0, + "step": 5515 + }, + { + "epoch": 0.6057544476169558, + "grad_norm": 2.246990203857422, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.711036741733551, + "num_tokens": 141984499.0, + "step": 5516 + }, + { + "epoch": 0.6058642653195695, + "grad_norm": 2.1511967182159424, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7043679356575012, + "num_tokens": 142012208.0, + "step": 5517 + }, + { + "epoch": 0.6059740830221831, + "grad_norm": 2.170342445373535, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7064211964607239, + "num_tokens": 142040318.0, + "step": 5518 + }, + { + "epoch": 0.6060839007247968, + "grad_norm": 2.1085784435272217, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6906925439834595, + "num_tokens": 142068506.0, + "step": 5519 + }, + { + "epoch": 0.6061937184274105, + "grad_norm": 2.244922637939453, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6984190940856934, + "num_tokens": 142097048.0, + "step": 5520 + }, + { + "epoch": 0.6063035361300242, + "grad_norm": 2.3558549880981445, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6995896100997925, + "num_tokens": 142117796.0, + "step": 5521 + }, + { + "epoch": 0.6064133538326378, + "grad_norm": 2.217660665512085, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6926960349082947, + "num_tokens": 142142267.0, + "step": 5522 + }, + { + "epoch": 0.6065231715352515, + "grad_norm": 2.5455427169799805, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7065897583961487, + "num_tokens": 142162209.0, + "step": 5523 + }, + { + "epoch": 0.6066329892378651, + "grad_norm": 1.9933104515075684, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7068055868148804, + "num_tokens": 142191547.0, + "step": 5524 + }, + { + "epoch": 0.6067428069404788, + "grad_norm": 1.9590867757797241, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7056702375411987, + "num_tokens": 142223253.0, + "step": 5525 + }, + { + "epoch": 0.6068526246430924, + "grad_norm": 2.2012009620666504, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6883671283721924, + "num_tokens": 142249479.0, + "step": 5526 + }, + { + "epoch": 0.6069624423457062, + "grad_norm": 2.189812660217285, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7080329060554504, + "num_tokens": 142274929.0, + "step": 5527 + }, + { + "epoch": 0.6070722600483198, + "grad_norm": 2.4398083686828613, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6974037885665894, + "num_tokens": 142297983.0, + "step": 5528 + }, + { + "epoch": 0.6071820777509335, + "grad_norm": 2.398433208465576, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7158379554748535, + "num_tokens": 142318805.0, + "step": 5529 + }, + { + "epoch": 0.6072918954535471, + "grad_norm": 1.964294195175171, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6989336013793945, + "num_tokens": 142349536.0, + "step": 5530 + }, + { + "epoch": 0.6074017131561608, + "grad_norm": 2.311032295227051, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.710225522518158, + "num_tokens": 142370963.0, + "step": 5531 + }, + { + "epoch": 0.6075115308587744, + "grad_norm": 2.1758103370666504, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.7002862691879272, + "num_tokens": 142395447.0, + "step": 5532 + }, + { + "epoch": 0.607621348561388, + "grad_norm": 2.152240514755249, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6980368494987488, + "num_tokens": 142422649.0, + "step": 5533 + }, + { + "epoch": 0.6077311662640018, + "grad_norm": 2.283395528793335, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7091730833053589, + "num_tokens": 142444910.0, + "step": 5534 + }, + { + "epoch": 0.6078409839666155, + "grad_norm": 2.1522419452667236, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.701116681098938, + "num_tokens": 142471183.0, + "step": 5535 + }, + { + "epoch": 0.6079508016692291, + "grad_norm": 2.186432361602783, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7091240286827087, + "num_tokens": 142499143.0, + "step": 5536 + }, + { + "epoch": 0.6080606193718427, + "grad_norm": 1.9989821910858154, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6912871599197388, + "num_tokens": 142527556.0, + "step": 5537 + }, + { + "epoch": 0.6081704370744564, + "grad_norm": 2.2989964485168457, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7250366806983948, + "num_tokens": 142549842.0, + "step": 5538 + }, + { + "epoch": 0.60828025477707, + "grad_norm": 2.1286962032318115, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7199302911758423, + "num_tokens": 142579553.0, + "step": 5539 + }, + { + "epoch": 0.6083900724796837, + "grad_norm": 2.40272855758667, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6783612966537476, + "num_tokens": 142603165.0, + "step": 5540 + }, + { + "epoch": 0.6084998901822973, + "grad_norm": 1.8928143978118896, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7025921940803528, + "num_tokens": 142634081.0, + "step": 5541 + }, + { + "epoch": 0.6086097078849111, + "grad_norm": 2.0436789989471436, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6932588815689087, + "num_tokens": 142664349.0, + "step": 5542 + }, + { + "epoch": 0.6087195255875247, + "grad_norm": 2.094447135925293, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.706626296043396, + "num_tokens": 142693332.0, + "step": 5543 + }, + { + "epoch": 0.6088293432901384, + "grad_norm": 2.3842825889587402, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.713858962059021, + "num_tokens": 142716072.0, + "step": 5544 + }, + { + "epoch": 0.608939160992752, + "grad_norm": 1.9106886386871338, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7027761936187744, + "num_tokens": 142746884.0, + "step": 5545 + }, + { + "epoch": 0.6090489786953657, + "grad_norm": 2.143845319747925, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6887023448944092, + "num_tokens": 142773837.0, + "step": 5546 + }, + { + "epoch": 0.6091587963979793, + "grad_norm": 2.085008144378662, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7027218341827393, + "num_tokens": 142801939.0, + "step": 5547 + }, + { + "epoch": 0.609268614100593, + "grad_norm": 2.2385425567626953, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7219832539558411, + "num_tokens": 142825082.0, + "step": 5548 + }, + { + "epoch": 0.6093784318032067, + "grad_norm": 2.6228883266448975, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7195618748664856, + "num_tokens": 142843119.0, + "step": 5549 + }, + { + "epoch": 0.6094882495058204, + "grad_norm": 2.254390239715576, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.6867196559906006, + "num_tokens": 142869146.0, + "step": 5550 + }, + { + "epoch": 0.609598067208434, + "grad_norm": 1.9855024814605713, + "learning_rate": 1e-06, + "loss": 1.1004, + "mean_token_accuracy": 0.6700899600982666, + "num_tokens": 142901345.0, + "step": 5551 + }, + { + "epoch": 0.6097078849110477, + "grad_norm": 1.8942946195602417, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6963304281234741, + "num_tokens": 142935341.0, + "step": 5552 + }, + { + "epoch": 0.6098177026136613, + "grad_norm": 2.3372416496276855, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7011623382568359, + "num_tokens": 142959035.0, + "step": 5553 + }, + { + "epoch": 0.609927520316275, + "grad_norm": 2.0206847190856934, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.694139301776886, + "num_tokens": 142988875.0, + "step": 5554 + }, + { + "epoch": 0.6100373380188886, + "grad_norm": 2.229872226715088, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6878275871276855, + "num_tokens": 143015227.0, + "step": 5555 + }, + { + "epoch": 0.6101471557215024, + "grad_norm": 2.374760866165161, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7077049016952515, + "num_tokens": 143037996.0, + "step": 5556 + }, + { + "epoch": 0.610256973424116, + "grad_norm": 2.492473602294922, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7374423742294312, + "num_tokens": 143056293.0, + "step": 5557 + }, + { + "epoch": 0.6103667911267296, + "grad_norm": 2.2572309970855713, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6956163644790649, + "num_tokens": 143083265.0, + "step": 5558 + }, + { + "epoch": 0.6104766088293433, + "grad_norm": 2.070547580718994, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6921122074127197, + "num_tokens": 143112890.0, + "step": 5559 + }, + { + "epoch": 0.6105864265319569, + "grad_norm": 2.4777281284332275, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.7094738483428955, + "num_tokens": 143135793.0, + "step": 5560 + }, + { + "epoch": 0.6106962442345706, + "grad_norm": 2.078270196914673, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.708922803401947, + "num_tokens": 143163388.0, + "step": 5561 + }, + { + "epoch": 0.6108060619371842, + "grad_norm": 2.2480883598327637, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6971573829650879, + "num_tokens": 143189078.0, + "step": 5562 + }, + { + "epoch": 0.610915879639798, + "grad_norm": 2.225518226623535, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6812540292739868, + "num_tokens": 143215368.0, + "step": 5563 + }, + { + "epoch": 0.6110256973424116, + "grad_norm": 2.393420696258545, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6945215463638306, + "num_tokens": 143238691.0, + "step": 5564 + }, + { + "epoch": 0.6111355150450253, + "grad_norm": 2.1854958534240723, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6906744241714478, + "num_tokens": 143264190.0, + "step": 5565 + }, + { + "epoch": 0.6112453327476389, + "grad_norm": 2.2058959007263184, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6816955208778381, + "num_tokens": 143290726.0, + "step": 5566 + }, + { + "epoch": 0.6113551504502526, + "grad_norm": 2.052371025085449, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7209017276763916, + "num_tokens": 143317683.0, + "step": 5567 + }, + { + "epoch": 0.6114649681528662, + "grad_norm": 2.2837774753570557, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7165191173553467, + "num_tokens": 143341129.0, + "step": 5568 + }, + { + "epoch": 0.6115747858554799, + "grad_norm": 2.3340585231781006, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7134583592414856, + "num_tokens": 143362889.0, + "step": 5569 + }, + { + "epoch": 0.6116846035580935, + "grad_norm": 1.9181690216064453, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7055912613868713, + "num_tokens": 143395501.0, + "step": 5570 + }, + { + "epoch": 0.6117944212607073, + "grad_norm": 1.8301857709884644, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6975746750831604, + "num_tokens": 143430036.0, + "step": 5571 + }, + { + "epoch": 0.6119042389633209, + "grad_norm": 2.043212652206421, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7300019264221191, + "num_tokens": 143458191.0, + "step": 5572 + }, + { + "epoch": 0.6120140566659346, + "grad_norm": 2.2482244968414307, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7133020162582397, + "num_tokens": 143482718.0, + "step": 5573 + }, + { + "epoch": 0.6121238743685482, + "grad_norm": 2.153006076812744, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7032998204231262, + "num_tokens": 143509019.0, + "step": 5574 + }, + { + "epoch": 0.6122336920711619, + "grad_norm": 2.3744730949401855, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6912750005722046, + "num_tokens": 143532970.0, + "step": 5575 + }, + { + "epoch": 0.6123435097737755, + "grad_norm": 2.5353572368621826, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7162235379219055, + "num_tokens": 143553306.0, + "step": 5576 + }, + { + "epoch": 0.6124533274763891, + "grad_norm": 2.5077545642852783, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7000815868377686, + "num_tokens": 143573786.0, + "step": 5577 + }, + { + "epoch": 0.6125631451790029, + "grad_norm": 2.0683515071868896, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7249493598937988, + "num_tokens": 143600111.0, + "step": 5578 + }, + { + "epoch": 0.6126729628816165, + "grad_norm": 2.051720142364502, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7148789167404175, + "num_tokens": 143629514.0, + "step": 5579 + }, + { + "epoch": 0.6127827805842302, + "grad_norm": 2.1698789596557617, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.697892427444458, + "num_tokens": 143656932.0, + "step": 5580 + }, + { + "epoch": 0.6128925982868438, + "grad_norm": 1.9083020687103271, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6876890659332275, + "num_tokens": 143689437.0, + "step": 5581 + }, + { + "epoch": 0.6130024159894575, + "grad_norm": 2.1497888565063477, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6979895830154419, + "num_tokens": 143715185.0, + "step": 5582 + }, + { + "epoch": 0.6131122336920711, + "grad_norm": 2.217646360397339, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7210035920143127, + "num_tokens": 143739562.0, + "step": 5583 + }, + { + "epoch": 0.6132220513946848, + "grad_norm": 2.336606025695801, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6908566355705261, + "num_tokens": 143762393.0, + "step": 5584 + }, + { + "epoch": 0.6133318690972985, + "grad_norm": 2.158484935760498, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.691765308380127, + "num_tokens": 143789937.0, + "step": 5585 + }, + { + "epoch": 0.6134416867999122, + "grad_norm": 2.1478538513183594, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6858232021331787, + "num_tokens": 143816198.0, + "step": 5586 + }, + { + "epoch": 0.6135515045025258, + "grad_norm": 2.0351076126098633, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7124426960945129, + "num_tokens": 143845286.0, + "step": 5587 + }, + { + "epoch": 0.6136613222051395, + "grad_norm": 2.00644588470459, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7051265239715576, + "num_tokens": 143874598.0, + "step": 5588 + }, + { + "epoch": 0.6137711399077531, + "grad_norm": 2.136648178100586, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7069611549377441, + "num_tokens": 143899885.0, + "step": 5589 + }, + { + "epoch": 0.6138809576103668, + "grad_norm": 2.261960029602051, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7302890419960022, + "num_tokens": 143924009.0, + "step": 5590 + }, + { + "epoch": 0.6139907753129804, + "grad_norm": 2.333202600479126, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7124239206314087, + "num_tokens": 143945477.0, + "step": 5591 + }, + { + "epoch": 0.6141005930155942, + "grad_norm": 2.188875675201416, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7104387283325195, + "num_tokens": 143970127.0, + "step": 5592 + }, + { + "epoch": 0.6142104107182078, + "grad_norm": 2.0591840744018555, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6946786642074585, + "num_tokens": 143997812.0, + "step": 5593 + }, + { + "epoch": 0.6143202284208215, + "grad_norm": 2.4069132804870605, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7196158170700073, + "num_tokens": 144018974.0, + "step": 5594 + }, + { + "epoch": 0.6144300461234351, + "grad_norm": 2.3545496463775635, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6968221068382263, + "num_tokens": 144043566.0, + "step": 5595 + }, + { + "epoch": 0.6145398638260487, + "grad_norm": 2.1839635372161865, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6891649961471558, + "num_tokens": 144069951.0, + "step": 5596 + }, + { + "epoch": 0.6146496815286624, + "grad_norm": 2.266758680343628, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7194681167602539, + "num_tokens": 144092222.0, + "step": 5597 + }, + { + "epoch": 0.614759499231276, + "grad_norm": 2.253506898880005, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7339977025985718, + "num_tokens": 144117338.0, + "step": 5598 + }, + { + "epoch": 0.6148693169338898, + "grad_norm": 2.594862937927246, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7107449769973755, + "num_tokens": 144135381.0, + "step": 5599 + }, + { + "epoch": 0.6149791346365034, + "grad_norm": 2.025320529937744, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6898843050003052, + "num_tokens": 144164934.0, + "step": 5600 + }, + { + "epoch": 0.6150889523391171, + "grad_norm": 2.0211400985717773, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7209425568580627, + "num_tokens": 144192314.0, + "step": 5601 + }, + { + "epoch": 0.6151987700417307, + "grad_norm": 2.402160882949829, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7166340351104736, + "num_tokens": 144215533.0, + "step": 5602 + }, + { + "epoch": 0.6153085877443444, + "grad_norm": 2.6068828105926514, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7183858752250671, + "num_tokens": 144233936.0, + "step": 5603 + }, + { + "epoch": 0.615418405446958, + "grad_norm": 2.477588176727295, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6923924684524536, + "num_tokens": 144256088.0, + "step": 5604 + }, + { + "epoch": 0.6155282231495717, + "grad_norm": 2.0915775299072266, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6823843121528625, + "num_tokens": 144283971.0, + "step": 5605 + }, + { + "epoch": 0.6156380408521853, + "grad_norm": 2.0076241493225098, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6869555711746216, + "num_tokens": 144313817.0, + "step": 5606 + }, + { + "epoch": 0.6157478585547991, + "grad_norm": 2.282602310180664, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6927716732025146, + "num_tokens": 144338050.0, + "step": 5607 + }, + { + "epoch": 0.6158576762574127, + "grad_norm": 2.4059224128723145, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7106494903564453, + "num_tokens": 144359834.0, + "step": 5608 + }, + { + "epoch": 0.6159674939600264, + "grad_norm": 2.469054698944092, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7058336138725281, + "num_tokens": 144379164.0, + "step": 5609 + }, + { + "epoch": 0.61607731166264, + "grad_norm": 2.155595541000366, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7194296717643738, + "num_tokens": 144404963.0, + "step": 5610 + }, + { + "epoch": 0.6161871293652537, + "grad_norm": 2.2917582988739014, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7045578956604004, + "num_tokens": 144430130.0, + "step": 5611 + }, + { + "epoch": 0.6162969470678673, + "grad_norm": 2.2912638187408447, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.695976972579956, + "num_tokens": 144453373.0, + "step": 5612 + }, + { + "epoch": 0.616406764770481, + "grad_norm": 2.295083999633789, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7301012873649597, + "num_tokens": 144475712.0, + "step": 5613 + }, + { + "epoch": 0.6165165824730947, + "grad_norm": 2.1432032585144043, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7233384847640991, + "num_tokens": 144500205.0, + "step": 5614 + }, + { + "epoch": 0.6166264001757084, + "grad_norm": 2.2849442958831787, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7045661211013794, + "num_tokens": 144525435.0, + "step": 5615 + }, + { + "epoch": 0.616736217878322, + "grad_norm": 2.142714738845825, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.690411388874054, + "num_tokens": 144551946.0, + "step": 5616 + }, + { + "epoch": 0.6168460355809356, + "grad_norm": 2.068875551223755, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6770964860916138, + "num_tokens": 144581377.0, + "step": 5617 + }, + { + "epoch": 0.6169558532835493, + "grad_norm": 2.2104244232177734, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7103294134140015, + "num_tokens": 144605407.0, + "step": 5618 + }, + { + "epoch": 0.6170656709861629, + "grad_norm": 1.9513229131698608, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.694976270198822, + "num_tokens": 144638478.0, + "step": 5619 + }, + { + "epoch": 0.6171754886887766, + "grad_norm": 2.362201452255249, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6878698468208313, + "num_tokens": 144664259.0, + "step": 5620 + }, + { + "epoch": 0.6172853063913903, + "grad_norm": 1.964622139930725, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7087559700012207, + "num_tokens": 144693892.0, + "step": 5621 + }, + { + "epoch": 0.617395124094004, + "grad_norm": 2.0561764240264893, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7024843692779541, + "num_tokens": 144722285.0, + "step": 5622 + }, + { + "epoch": 0.6175049417966176, + "grad_norm": 2.0923268795013428, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7091777324676514, + "num_tokens": 144749434.0, + "step": 5623 + }, + { + "epoch": 0.6176147594992313, + "grad_norm": 2.1948046684265137, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7121505737304688, + "num_tokens": 144774448.0, + "step": 5624 + }, + { + "epoch": 0.6177245772018449, + "grad_norm": 2.0200302600860596, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.676436185836792, + "num_tokens": 144804250.0, + "step": 5625 + }, + { + "epoch": 0.6178343949044586, + "grad_norm": 2.381385564804077, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7046735286712646, + "num_tokens": 144824150.0, + "step": 5626 + }, + { + "epoch": 0.6179442126070722, + "grad_norm": 2.1952965259552, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6965450048446655, + "num_tokens": 144850763.0, + "step": 5627 + }, + { + "epoch": 0.618054030309686, + "grad_norm": 2.2991275787353516, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7262209057807922, + "num_tokens": 144875202.0, + "step": 5628 + }, + { + "epoch": 0.6181638480122996, + "grad_norm": 2.3207292556762695, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7189021110534668, + "num_tokens": 144900657.0, + "step": 5629 + }, + { + "epoch": 0.6182736657149133, + "grad_norm": 2.125965118408203, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7053059935569763, + "num_tokens": 144927191.0, + "step": 5630 + }, + { + "epoch": 0.6183834834175269, + "grad_norm": 2.237678050994873, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7070866227149963, + "num_tokens": 144951309.0, + "step": 5631 + }, + { + "epoch": 0.6184933011201406, + "grad_norm": 1.9889349937438965, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6901780366897583, + "num_tokens": 144982436.0, + "step": 5632 + }, + { + "epoch": 0.6186031188227542, + "grad_norm": 1.890545129776001, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7091492414474487, + "num_tokens": 145016289.0, + "step": 5633 + }, + { + "epoch": 0.6187129365253679, + "grad_norm": 2.226254940032959, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.7027561068534851, + "num_tokens": 145040120.0, + "step": 5634 + }, + { + "epoch": 0.6188227542279815, + "grad_norm": 2.151841163635254, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6888306140899658, + "num_tokens": 145066446.0, + "step": 5635 + }, + { + "epoch": 0.6189325719305953, + "grad_norm": 2.3626842498779297, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7118874788284302, + "num_tokens": 145088773.0, + "step": 5636 + }, + { + "epoch": 0.6190423896332089, + "grad_norm": 2.0891106128692627, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.721081018447876, + "num_tokens": 145115454.0, + "step": 5637 + }, + { + "epoch": 0.6191522073358225, + "grad_norm": 2.1855716705322266, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7000759840011597, + "num_tokens": 145141211.0, + "step": 5638 + }, + { + "epoch": 0.6192620250384362, + "grad_norm": 2.2380497455596924, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6837725639343262, + "num_tokens": 145170564.0, + "step": 5639 + }, + { + "epoch": 0.6193718427410498, + "grad_norm": 2.2571380138397217, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7043482065200806, + "num_tokens": 145195896.0, + "step": 5640 + }, + { + "epoch": 0.6194816604436635, + "grad_norm": 2.16500186920166, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.704952597618103, + "num_tokens": 145222280.0, + "step": 5641 + }, + { + "epoch": 0.6195914781462771, + "grad_norm": 2.5083706378936768, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6970844268798828, + "num_tokens": 145245900.0, + "step": 5642 + }, + { + "epoch": 0.6197012958488909, + "grad_norm": 2.6981425285339355, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7145004272460938, + "num_tokens": 145264354.0, + "step": 5643 + }, + { + "epoch": 0.6198111135515045, + "grad_norm": 2.1723077297210693, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6965438723564148, + "num_tokens": 145291502.0, + "step": 5644 + }, + { + "epoch": 0.6199209312541182, + "grad_norm": 2.2836878299713135, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6874467134475708, + "num_tokens": 145315820.0, + "step": 5645 + }, + { + "epoch": 0.6200307489567318, + "grad_norm": 2.33931565284729, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7028190493583679, + "num_tokens": 145339632.0, + "step": 5646 + }, + { + "epoch": 0.6201405666593455, + "grad_norm": 2.6197352409362793, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7307153344154358, + "num_tokens": 145358020.0, + "step": 5647 + }, + { + "epoch": 0.6202503843619591, + "grad_norm": 2.134523868560791, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.7006711363792419, + "num_tokens": 145384653.0, + "step": 5648 + }, + { + "epoch": 0.6203602020645728, + "grad_norm": 2.329498767852783, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7088847756385803, + "num_tokens": 145408444.0, + "step": 5649 + }, + { + "epoch": 0.6204700197671865, + "grad_norm": 2.416905641555786, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7357460260391235, + "num_tokens": 145429472.0, + "step": 5650 + }, + { + "epoch": 0.6205798374698002, + "grad_norm": 2.2648582458496094, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7229171991348267, + "num_tokens": 145452979.0, + "step": 5651 + }, + { + "epoch": 0.6206896551724138, + "grad_norm": 2.1535160541534424, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7127246856689453, + "num_tokens": 145478028.0, + "step": 5652 + }, + { + "epoch": 0.6207994728750275, + "grad_norm": 2.0911171436309814, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7010141611099243, + "num_tokens": 145507081.0, + "step": 5653 + }, + { + "epoch": 0.6209092905776411, + "grad_norm": 2.1990134716033936, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6876100897789001, + "num_tokens": 145533208.0, + "step": 5654 + }, + { + "epoch": 0.6210191082802548, + "grad_norm": 2.035484790802002, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7065893411636353, + "num_tokens": 145561222.0, + "step": 5655 + }, + { + "epoch": 0.6211289259828684, + "grad_norm": 2.1869239807128906, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7063705921173096, + "num_tokens": 145585369.0, + "step": 5656 + }, + { + "epoch": 0.6212387436854822, + "grad_norm": 2.3645379543304443, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6959149837493896, + "num_tokens": 145608482.0, + "step": 5657 + }, + { + "epoch": 0.6213485613880958, + "grad_norm": 2.4582431316375732, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7127864360809326, + "num_tokens": 145629108.0, + "step": 5658 + }, + { + "epoch": 0.6214583790907094, + "grad_norm": 2.072380781173706, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7155699729919434, + "num_tokens": 145656605.0, + "step": 5659 + }, + { + "epoch": 0.6215681967933231, + "grad_norm": 2.467366933822632, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7469925880432129, + "num_tokens": 145675691.0, + "step": 5660 + }, + { + "epoch": 0.6216780144959367, + "grad_norm": 2.3153092861175537, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7290899753570557, + "num_tokens": 145698504.0, + "step": 5661 + }, + { + "epoch": 0.6217878321985504, + "grad_norm": 2.6358251571655273, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7230287790298462, + "num_tokens": 145716550.0, + "step": 5662 + }, + { + "epoch": 0.621897649901164, + "grad_norm": 2.226670265197754, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7133192420005798, + "num_tokens": 145739962.0, + "step": 5663 + }, + { + "epoch": 0.6220074676037777, + "grad_norm": 1.8828394412994385, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.688643217086792, + "num_tokens": 145773301.0, + "step": 5664 + }, + { + "epoch": 0.6221172853063914, + "grad_norm": 2.4296975135803223, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.695580244064331, + "num_tokens": 145796251.0, + "step": 5665 + }, + { + "epoch": 0.6222271030090051, + "grad_norm": 2.2838847637176514, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.7086288928985596, + "num_tokens": 145820777.0, + "step": 5666 + }, + { + "epoch": 0.6223369207116187, + "grad_norm": 2.035250186920166, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7220175862312317, + "num_tokens": 145848525.0, + "step": 5667 + }, + { + "epoch": 0.6224467384142324, + "grad_norm": 2.4108078479766846, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7120382785797119, + "num_tokens": 145869814.0, + "step": 5668 + }, + { + "epoch": 0.622556556116846, + "grad_norm": 2.071188449859619, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6861553192138672, + "num_tokens": 145899239.0, + "step": 5669 + }, + { + "epoch": 0.6226663738194597, + "grad_norm": 2.1448264122009277, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7233209013938904, + "num_tokens": 145924657.0, + "step": 5670 + }, + { + "epoch": 0.6227761915220733, + "grad_norm": 2.3907525539398193, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7259299159049988, + "num_tokens": 145948361.0, + "step": 5671 + }, + { + "epoch": 0.6228860092246871, + "grad_norm": 2.227147102355957, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7092664241790771, + "num_tokens": 145971180.0, + "step": 5672 + }, + { + "epoch": 0.6229958269273007, + "grad_norm": 2.374497413635254, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6922922134399414, + "num_tokens": 145994876.0, + "step": 5673 + }, + { + "epoch": 0.6231056446299144, + "grad_norm": 2.6333131790161133, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7307783365249634, + "num_tokens": 146013570.0, + "step": 5674 + }, + { + "epoch": 0.623215462332528, + "grad_norm": 2.165931224822998, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6998127698898315, + "num_tokens": 146038234.0, + "step": 5675 + }, + { + "epoch": 0.6233252800351416, + "grad_norm": 2.2912445068359375, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6824313998222351, + "num_tokens": 146064052.0, + "step": 5676 + }, + { + "epoch": 0.6234350977377553, + "grad_norm": 1.9862157106399536, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7076736092567444, + "num_tokens": 146095360.0, + "step": 5677 + }, + { + "epoch": 0.6235449154403689, + "grad_norm": 2.0679705142974854, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7146774530410767, + "num_tokens": 146122678.0, + "step": 5678 + }, + { + "epoch": 0.6236547331429827, + "grad_norm": 2.322385549545288, + "learning_rate": 1e-06, + "loss": 1.0492, + "mean_token_accuracy": 0.698348879814148, + "num_tokens": 146147493.0, + "step": 5679 + }, + { + "epoch": 0.6237645508455963, + "grad_norm": 2.2119669914245605, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7011330127716064, + "num_tokens": 146173276.0, + "step": 5680 + }, + { + "epoch": 0.62387436854821, + "grad_norm": 2.1368372440338135, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7075433135032654, + "num_tokens": 146202489.0, + "step": 5681 + }, + { + "epoch": 0.6239841862508236, + "grad_norm": 2.580665349960327, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7199264764785767, + "num_tokens": 146223199.0, + "step": 5682 + }, + { + "epoch": 0.6240940039534373, + "grad_norm": 2.025235652923584, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7071088552474976, + "num_tokens": 146252056.0, + "step": 5683 + }, + { + "epoch": 0.6242038216560509, + "grad_norm": 1.9516035318374634, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7052913904190063, + "num_tokens": 146283565.0, + "step": 5684 + }, + { + "epoch": 0.6243136393586646, + "grad_norm": 1.9951893091201782, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7151507139205933, + "num_tokens": 146311508.0, + "step": 5685 + }, + { + "epoch": 0.6244234570612783, + "grad_norm": 2.270646810531616, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7197519540786743, + "num_tokens": 146334055.0, + "step": 5686 + }, + { + "epoch": 0.624533274763892, + "grad_norm": 2.3073368072509766, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7072198987007141, + "num_tokens": 146357546.0, + "step": 5687 + }, + { + "epoch": 0.6246430924665056, + "grad_norm": 2.385309934616089, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6837517023086548, + "num_tokens": 146380579.0, + "step": 5688 + }, + { + "epoch": 0.6247529101691193, + "grad_norm": 2.385284423828125, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6795077323913574, + "num_tokens": 146406128.0, + "step": 5689 + }, + { + "epoch": 0.6248627278717329, + "grad_norm": 2.1531779766082764, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7026304006576538, + "num_tokens": 146430671.0, + "step": 5690 + }, + { + "epoch": 0.6249725455743466, + "grad_norm": 2.1894328594207764, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6825697422027588, + "num_tokens": 146457968.0, + "step": 5691 + }, + { + "epoch": 0.6250823632769602, + "grad_norm": 2.195794105529785, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7085469961166382, + "num_tokens": 146483487.0, + "step": 5692 + }, + { + "epoch": 0.6251921809795739, + "grad_norm": 2.0423548221588135, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6986067295074463, + "num_tokens": 146513988.0, + "step": 5693 + }, + { + "epoch": 0.6253019986821876, + "grad_norm": 2.0216927528381348, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6941391825675964, + "num_tokens": 146544180.0, + "step": 5694 + }, + { + "epoch": 0.6254118163848013, + "grad_norm": 2.2780022621154785, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7274904251098633, + "num_tokens": 146568775.0, + "step": 5695 + }, + { + "epoch": 0.6255216340874149, + "grad_norm": 2.2709882259368896, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7049768567085266, + "num_tokens": 146591237.0, + "step": 5696 + }, + { + "epoch": 0.6256314517900285, + "grad_norm": 2.1956872940063477, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7073005437850952, + "num_tokens": 146616964.0, + "step": 5697 + }, + { + "epoch": 0.6257412694926422, + "grad_norm": 2.236682176589966, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.6996893882751465, + "num_tokens": 146641136.0, + "step": 5698 + }, + { + "epoch": 0.6258510871952558, + "grad_norm": 2.549917459487915, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7145304083824158, + "num_tokens": 146661368.0, + "step": 5699 + }, + { + "epoch": 0.6259609048978695, + "grad_norm": 2.091430425643921, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7037694454193115, + "num_tokens": 146688951.0, + "step": 5700 + }, + { + "epoch": 0.6260707226004832, + "grad_norm": 2.162285089492798, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.696344792842865, + "num_tokens": 146715883.0, + "step": 5701 + }, + { + "epoch": 0.6261805403030969, + "grad_norm": 2.232724189758301, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.688651442527771, + "num_tokens": 146743421.0, + "step": 5702 + }, + { + "epoch": 0.6262903580057105, + "grad_norm": 1.8698770999908447, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6848113536834717, + "num_tokens": 146778747.0, + "step": 5703 + }, + { + "epoch": 0.6264001757083242, + "grad_norm": 2.160024881362915, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6917498111724854, + "num_tokens": 146809978.0, + "step": 5704 + }, + { + "epoch": 0.6265099934109378, + "grad_norm": 2.2497901916503906, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7046761512756348, + "num_tokens": 146836445.0, + "step": 5705 + }, + { + "epoch": 0.6266198111135515, + "grad_norm": 2.3155789375305176, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7139399647712708, + "num_tokens": 146859193.0, + "step": 5706 + }, + { + "epoch": 0.6267296288161651, + "grad_norm": 2.1103618144989014, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7354748249053955, + "num_tokens": 146883971.0, + "step": 5707 + }, + { + "epoch": 0.6268394465187789, + "grad_norm": 2.226945638656616, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6978013515472412, + "num_tokens": 146909119.0, + "step": 5708 + }, + { + "epoch": 0.6269492642213925, + "grad_norm": 2.141444444656372, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.693800687789917, + "num_tokens": 146934958.0, + "step": 5709 + }, + { + "epoch": 0.6270590819240062, + "grad_norm": 2.198017120361328, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.707942008972168, + "num_tokens": 146960386.0, + "step": 5710 + }, + { + "epoch": 0.6271688996266198, + "grad_norm": 2.2091026306152344, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6978346109390259, + "num_tokens": 146983441.0, + "step": 5711 + }, + { + "epoch": 0.6272787173292335, + "grad_norm": 2.105158805847168, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7028123140335083, + "num_tokens": 147010013.0, + "step": 5712 + }, + { + "epoch": 0.6273885350318471, + "grad_norm": 2.260009527206421, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7079576849937439, + "num_tokens": 147033186.0, + "step": 5713 + }, + { + "epoch": 0.6274983527344608, + "grad_norm": 2.0951032638549805, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7282013297080994, + "num_tokens": 147060495.0, + "step": 5714 + }, + { + "epoch": 0.6276081704370745, + "grad_norm": 2.3777225017547607, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.6976646184921265, + "num_tokens": 147082393.0, + "step": 5715 + }, + { + "epoch": 0.6277179881396882, + "grad_norm": 2.1188998222351074, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7219401597976685, + "num_tokens": 147107070.0, + "step": 5716 + }, + { + "epoch": 0.6278278058423018, + "grad_norm": 2.01174259185791, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6958998441696167, + "num_tokens": 147137296.0, + "step": 5717 + }, + { + "epoch": 0.6279376235449154, + "grad_norm": 2.1680192947387695, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.708991289138794, + "num_tokens": 147163367.0, + "step": 5718 + }, + { + "epoch": 0.6280474412475291, + "grad_norm": 2.503629446029663, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7000263929367065, + "num_tokens": 147184518.0, + "step": 5719 + }, + { + "epoch": 0.6281572589501427, + "grad_norm": 2.1702592372894287, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6933903694152832, + "num_tokens": 147210953.0, + "step": 5720 + }, + { + "epoch": 0.6282670766527564, + "grad_norm": 2.3006160259246826, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6933343410491943, + "num_tokens": 147233927.0, + "step": 5721 + }, + { + "epoch": 0.62837689435537, + "grad_norm": 2.093686103820801, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7304826974868774, + "num_tokens": 147257179.0, + "step": 5722 + }, + { + "epoch": 0.6284867120579838, + "grad_norm": 2.3580024242401123, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7001702785491943, + "num_tokens": 147280593.0, + "step": 5723 + }, + { + "epoch": 0.6285965297605974, + "grad_norm": 2.001549482345581, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6916457414627075, + "num_tokens": 147311116.0, + "step": 5724 + }, + { + "epoch": 0.6287063474632111, + "grad_norm": 2.451545476913452, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7212440371513367, + "num_tokens": 147334954.0, + "step": 5725 + }, + { + "epoch": 0.6288161651658247, + "grad_norm": 2.3809964656829834, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.695768415927887, + "num_tokens": 147359924.0, + "step": 5726 + }, + { + "epoch": 0.6289259828684384, + "grad_norm": 2.316084623336792, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7128268480300903, + "num_tokens": 147381960.0, + "step": 5727 + }, + { + "epoch": 0.629035800571052, + "grad_norm": 1.970587134361267, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6744714975357056, + "num_tokens": 147414023.0, + "step": 5728 + }, + { + "epoch": 0.6291456182736657, + "grad_norm": 2.133958578109741, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6896766424179077, + "num_tokens": 147441943.0, + "step": 5729 + }, + { + "epoch": 0.6292554359762794, + "grad_norm": 2.3678412437438965, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7227351665496826, + "num_tokens": 147463145.0, + "step": 5730 + }, + { + "epoch": 0.6293652536788931, + "grad_norm": 2.2489993572235107, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6825259923934937, + "num_tokens": 147490270.0, + "step": 5731 + }, + { + "epoch": 0.6294750713815067, + "grad_norm": 2.251228094100952, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7165350914001465, + "num_tokens": 147514009.0, + "step": 5732 + }, + { + "epoch": 0.6295848890841204, + "grad_norm": 2.5267446041107178, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7107871174812317, + "num_tokens": 147535269.0, + "step": 5733 + }, + { + "epoch": 0.629694706786734, + "grad_norm": 2.1803579330444336, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7066094279289246, + "num_tokens": 147562082.0, + "step": 5734 + }, + { + "epoch": 0.6298045244893477, + "grad_norm": 2.2293522357940674, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7071404457092285, + "num_tokens": 147587261.0, + "step": 5735 + }, + { + "epoch": 0.6299143421919613, + "grad_norm": 1.9344168901443481, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6950956583023071, + "num_tokens": 147618753.0, + "step": 5736 + }, + { + "epoch": 0.630024159894575, + "grad_norm": 2.375600814819336, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7058625221252441, + "num_tokens": 147640085.0, + "step": 5737 + }, + { + "epoch": 0.6301339775971887, + "grad_norm": 2.36403751373291, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7162841558456421, + "num_tokens": 147661681.0, + "step": 5738 + }, + { + "epoch": 0.6302437952998023, + "grad_norm": 2.03130841255188, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7130879759788513, + "num_tokens": 147688628.0, + "step": 5739 + }, + { + "epoch": 0.630353613002416, + "grad_norm": 2.3740713596343994, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7348067164421082, + "num_tokens": 147708963.0, + "step": 5740 + }, + { + "epoch": 0.6304634307050296, + "grad_norm": 2.1523756980895996, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7126191854476929, + "num_tokens": 147734847.0, + "step": 5741 + }, + { + "epoch": 0.6305732484076433, + "grad_norm": 2.2361602783203125, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6976184844970703, + "num_tokens": 147758253.0, + "step": 5742 + }, + { + "epoch": 0.6306830661102569, + "grad_norm": 2.721024513244629, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7116522789001465, + "num_tokens": 147775075.0, + "step": 5743 + }, + { + "epoch": 0.6307928838128707, + "grad_norm": 1.8616315126419067, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6848264336585999, + "num_tokens": 147806974.0, + "step": 5744 + }, + { + "epoch": 0.6309027015154843, + "grad_norm": 2.1918704509735107, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7029716372489929, + "num_tokens": 147833318.0, + "step": 5745 + }, + { + "epoch": 0.631012519218098, + "grad_norm": 2.1325955390930176, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.7042468190193176, + "num_tokens": 147858607.0, + "step": 5746 + }, + { + "epoch": 0.6311223369207116, + "grad_norm": 1.9757895469665527, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7124457955360413, + "num_tokens": 147887276.0, + "step": 5747 + }, + { + "epoch": 0.6312321546233253, + "grad_norm": 2.379601240158081, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6979295015335083, + "num_tokens": 147909646.0, + "step": 5748 + }, + { + "epoch": 0.6313419723259389, + "grad_norm": 2.102532148361206, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6828067302703857, + "num_tokens": 147938873.0, + "step": 5749 + }, + { + "epoch": 0.6314517900285526, + "grad_norm": 2.1264989376068115, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7151653170585632, + "num_tokens": 147965201.0, + "step": 5750 + }, + { + "epoch": 0.6315616077311663, + "grad_norm": 2.015164852142334, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7080869078636169, + "num_tokens": 147993227.0, + "step": 5751 + }, + { + "epoch": 0.63167142543378, + "grad_norm": 2.0210933685302734, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7167549729347229, + "num_tokens": 148023186.0, + "step": 5752 + }, + { + "epoch": 0.6317812431363936, + "grad_norm": 2.570444107055664, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7672242522239685, + "num_tokens": 148040456.0, + "step": 5753 + }, + { + "epoch": 0.6318910608390073, + "grad_norm": 2.4136409759521484, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7115945219993591, + "num_tokens": 148063303.0, + "step": 5754 + }, + { + "epoch": 0.6320008785416209, + "grad_norm": 2.291024684906006, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6887101531028748, + "num_tokens": 148086608.0, + "step": 5755 + }, + { + "epoch": 0.6321106962442345, + "grad_norm": 2.4672160148620605, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6990878582000732, + "num_tokens": 148107314.0, + "step": 5756 + }, + { + "epoch": 0.6322205139468482, + "grad_norm": 2.167088508605957, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7052027583122253, + "num_tokens": 148132496.0, + "step": 5757 + }, + { + "epoch": 0.6323303316494618, + "grad_norm": 2.290332317352295, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7068315744400024, + "num_tokens": 148157498.0, + "step": 5758 + }, + { + "epoch": 0.6324401493520756, + "grad_norm": 2.162381172180176, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7137930393218994, + "num_tokens": 148180354.0, + "step": 5759 + }, + { + "epoch": 0.6325499670546892, + "grad_norm": 2.1256561279296875, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.7003017067909241, + "num_tokens": 148206601.0, + "step": 5760 + }, + { + "epoch": 0.6326597847573029, + "grad_norm": 2.0211479663848877, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7066882848739624, + "num_tokens": 148234048.0, + "step": 5761 + }, + { + "epoch": 0.6327696024599165, + "grad_norm": 2.2920827865600586, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7314332723617554, + "num_tokens": 148255610.0, + "step": 5762 + }, + { + "epoch": 0.6328794201625302, + "grad_norm": 2.212233543395996, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7057175636291504, + "num_tokens": 148279427.0, + "step": 5763 + }, + { + "epoch": 0.6329892378651438, + "grad_norm": 2.4208824634552, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7325703501701355, + "num_tokens": 148299658.0, + "step": 5764 + }, + { + "epoch": 0.6330990555677575, + "grad_norm": 2.0002214908599854, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7122840881347656, + "num_tokens": 148327760.0, + "step": 5765 + }, + { + "epoch": 0.6332088732703712, + "grad_norm": 2.028174877166748, + "learning_rate": 1e-06, + "loss": 1.076, + "mean_token_accuracy": 0.6779780387878418, + "num_tokens": 148355821.0, + "step": 5766 + }, + { + "epoch": 0.6333186909729849, + "grad_norm": 2.154975175857544, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.679884672164917, + "num_tokens": 148382918.0, + "step": 5767 + }, + { + "epoch": 0.6334285086755985, + "grad_norm": 1.9670010805130005, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.708035945892334, + "num_tokens": 148411507.0, + "step": 5768 + }, + { + "epoch": 0.6335383263782122, + "grad_norm": 1.977850317955017, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7044254541397095, + "num_tokens": 148441176.0, + "step": 5769 + }, + { + "epoch": 0.6336481440808258, + "grad_norm": 2.0825705528259277, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6949862837791443, + "num_tokens": 148468186.0, + "step": 5770 + }, + { + "epoch": 0.6337579617834395, + "grad_norm": 2.314922571182251, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.705780565738678, + "num_tokens": 148491527.0, + "step": 5771 + }, + { + "epoch": 0.6338677794860531, + "grad_norm": 2.1952967643737793, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6840908527374268, + "num_tokens": 148516369.0, + "step": 5772 + }, + { + "epoch": 0.6339775971886669, + "grad_norm": 2.438323974609375, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.696364164352417, + "num_tokens": 148537384.0, + "step": 5773 + }, + { + "epoch": 0.6340874148912805, + "grad_norm": 2.2065587043762207, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7184701561927795, + "num_tokens": 148560602.0, + "step": 5774 + }, + { + "epoch": 0.6341972325938942, + "grad_norm": 2.286616563796997, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7015501856803894, + "num_tokens": 148584058.0, + "step": 5775 + }, + { + "epoch": 0.6343070502965078, + "grad_norm": 2.142331123352051, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7061058282852173, + "num_tokens": 148611846.0, + "step": 5776 + }, + { + "epoch": 0.6344168679991214, + "grad_norm": 2.2740354537963867, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7064930200576782, + "num_tokens": 148634368.0, + "step": 5777 + }, + { + "epoch": 0.6345266857017351, + "grad_norm": 2.511828660964966, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7285233736038208, + "num_tokens": 148653847.0, + "step": 5778 + }, + { + "epoch": 0.6346365034043487, + "grad_norm": 2.375617504119873, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7124556303024292, + "num_tokens": 148675930.0, + "step": 5779 + }, + { + "epoch": 0.6347463211069625, + "grad_norm": 2.253922939300537, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.7010443210601807, + "num_tokens": 148700321.0, + "step": 5780 + }, + { + "epoch": 0.6348561388095761, + "grad_norm": 2.4404263496398926, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7170334458351135, + "num_tokens": 148721543.0, + "step": 5781 + }, + { + "epoch": 0.6349659565121898, + "grad_norm": 2.25136137008667, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.691209077835083, + "num_tokens": 148746538.0, + "step": 5782 + }, + { + "epoch": 0.6350757742148034, + "grad_norm": 2.4312777519226074, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6840829253196716, + "num_tokens": 148768887.0, + "step": 5783 + }, + { + "epoch": 0.6351855919174171, + "grad_norm": 2.049445390701294, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6941268444061279, + "num_tokens": 148797488.0, + "step": 5784 + }, + { + "epoch": 0.6352954096200307, + "grad_norm": 2.236527919769287, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7187563180923462, + "num_tokens": 148819938.0, + "step": 5785 + }, + { + "epoch": 0.6354052273226444, + "grad_norm": 2.0966029167175293, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6922550201416016, + "num_tokens": 148849203.0, + "step": 5786 + }, + { + "epoch": 0.635515045025258, + "grad_norm": 1.8175626993179321, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6973657011985779, + "num_tokens": 148888737.0, + "step": 5787 + }, + { + "epoch": 0.6356248627278718, + "grad_norm": 2.081312894821167, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7033007740974426, + "num_tokens": 148916342.0, + "step": 5788 + }, + { + "epoch": 0.6357346804304854, + "grad_norm": 2.2709107398986816, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7075082063674927, + "num_tokens": 148940915.0, + "step": 5789 + }, + { + "epoch": 0.6358444981330991, + "grad_norm": 2.1004552841186523, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7086868286132812, + "num_tokens": 148969504.0, + "step": 5790 + }, + { + "epoch": 0.6359543158357127, + "grad_norm": 2.3112967014312744, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7314849495887756, + "num_tokens": 148991283.0, + "step": 5791 + }, + { + "epoch": 0.6360641335383264, + "grad_norm": 2.2715799808502197, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7206290364265442, + "num_tokens": 149014920.0, + "step": 5792 + }, + { + "epoch": 0.63617395124094, + "grad_norm": 2.2155356407165527, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.709344744682312, + "num_tokens": 149039299.0, + "step": 5793 + }, + { + "epoch": 0.6362837689435537, + "grad_norm": 2.1798596382141113, + "learning_rate": 1e-06, + "loss": 1.0875, + "mean_token_accuracy": 0.6727985143661499, + "num_tokens": 149067397.0, + "step": 5794 + }, + { + "epoch": 0.6363935866461674, + "grad_norm": 2.2306015491485596, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7020416259765625, + "num_tokens": 149092087.0, + "step": 5795 + }, + { + "epoch": 0.636503404348781, + "grad_norm": 1.9620920419692993, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6970101594924927, + "num_tokens": 149126741.0, + "step": 5796 + }, + { + "epoch": 0.6366132220513947, + "grad_norm": 1.9781914949417114, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7143294811248779, + "num_tokens": 149159270.0, + "step": 5797 + }, + { + "epoch": 0.6367230397540083, + "grad_norm": 2.133246660232544, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6851273775100708, + "num_tokens": 149188820.0, + "step": 5798 + }, + { + "epoch": 0.636832857456622, + "grad_norm": 2.2223620414733887, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7066270112991333, + "num_tokens": 149216265.0, + "step": 5799 + }, + { + "epoch": 0.6369426751592356, + "grad_norm": 1.9479366540908813, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6789880990982056, + "num_tokens": 149247595.0, + "step": 5800 + }, + { + "epoch": 0.6370524928618493, + "grad_norm": 2.1294164657592773, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7101646661758423, + "num_tokens": 149273485.0, + "step": 5801 + }, + { + "epoch": 0.637162310564463, + "grad_norm": 2.1855905055999756, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7043299674987793, + "num_tokens": 149298041.0, + "step": 5802 + }, + { + "epoch": 0.6372721282670767, + "grad_norm": 2.1558725833892822, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7050368189811707, + "num_tokens": 149323088.0, + "step": 5803 + }, + { + "epoch": 0.6373819459696903, + "grad_norm": 2.46804141998291, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7193841934204102, + "num_tokens": 149342005.0, + "step": 5804 + }, + { + "epoch": 0.637491763672304, + "grad_norm": 2.232081651687622, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.696857750415802, + "num_tokens": 149367569.0, + "step": 5805 + }, + { + "epoch": 0.6376015813749176, + "grad_norm": 2.066293239593506, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6896818280220032, + "num_tokens": 149397108.0, + "step": 5806 + }, + { + "epoch": 0.6377113990775313, + "grad_norm": 2.4061503410339355, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7042630314826965, + "num_tokens": 149418739.0, + "step": 5807 + }, + { + "epoch": 0.6378212167801449, + "grad_norm": 2.417844533920288, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7282771468162537, + "num_tokens": 149439391.0, + "step": 5808 + }, + { + "epoch": 0.6379310344827587, + "grad_norm": 2.4496829509735107, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7201999425888062, + "num_tokens": 149460527.0, + "step": 5809 + }, + { + "epoch": 0.6380408521853723, + "grad_norm": 2.2927002906799316, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6948865652084351, + "num_tokens": 149485423.0, + "step": 5810 + }, + { + "epoch": 0.638150669887986, + "grad_norm": 2.4406044483184814, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7055636048316956, + "num_tokens": 149506236.0, + "step": 5811 + }, + { + "epoch": 0.6382604875905996, + "grad_norm": 2.3527743816375732, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7025243639945984, + "num_tokens": 149533563.0, + "step": 5812 + }, + { + "epoch": 0.6383703052932133, + "grad_norm": 2.341615676879883, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7169139385223389, + "num_tokens": 149556739.0, + "step": 5813 + }, + { + "epoch": 0.6384801229958269, + "grad_norm": 2.040290117263794, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6975715756416321, + "num_tokens": 149586367.0, + "step": 5814 + }, + { + "epoch": 0.6385899406984406, + "grad_norm": 2.099796772003174, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7031739354133606, + "num_tokens": 149614515.0, + "step": 5815 + }, + { + "epoch": 0.6386997584010542, + "grad_norm": 2.1696789264678955, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7011111378669739, + "num_tokens": 149640865.0, + "step": 5816 + }, + { + "epoch": 0.638809576103668, + "grad_norm": 2.21919584274292, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7160707116127014, + "num_tokens": 149664605.0, + "step": 5817 + }, + { + "epoch": 0.6389193938062816, + "grad_norm": 2.3191728591918945, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.6873449087142944, + "num_tokens": 149688183.0, + "step": 5818 + }, + { + "epoch": 0.6390292115088952, + "grad_norm": 1.9773516654968262, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6841757297515869, + "num_tokens": 149720238.0, + "step": 5819 + }, + { + "epoch": 0.6391390292115089, + "grad_norm": 2.183063268661499, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7061982750892639, + "num_tokens": 149747433.0, + "step": 5820 + }, + { + "epoch": 0.6392488469141225, + "grad_norm": 2.2153878211975098, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7075732946395874, + "num_tokens": 149771032.0, + "step": 5821 + }, + { + "epoch": 0.6393586646167362, + "grad_norm": 2.2563459873199463, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6965633630752563, + "num_tokens": 149796167.0, + "step": 5822 + }, + { + "epoch": 0.6394684823193498, + "grad_norm": 2.4372613430023193, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6884199976921082, + "num_tokens": 149818239.0, + "step": 5823 + }, + { + "epoch": 0.6395783000219636, + "grad_norm": 2.252197742462158, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6938668489456177, + "num_tokens": 149843839.0, + "step": 5824 + }, + { + "epoch": 0.6396881177245772, + "grad_norm": 2.683664321899414, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6924952268600464, + "num_tokens": 149870718.0, + "step": 5825 + }, + { + "epoch": 0.6397979354271909, + "grad_norm": 2.230879545211792, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.709658145904541, + "num_tokens": 149895455.0, + "step": 5826 + }, + { + "epoch": 0.6399077531298045, + "grad_norm": 2.1797945499420166, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6853435039520264, + "num_tokens": 149923109.0, + "step": 5827 + }, + { + "epoch": 0.6400175708324182, + "grad_norm": 2.328129529953003, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7188049554824829, + "num_tokens": 149947561.0, + "step": 5828 + }, + { + "epoch": 0.6401273885350318, + "grad_norm": 2.3649168014526367, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7157799005508423, + "num_tokens": 149968697.0, + "step": 5829 + }, + { + "epoch": 0.6402372062376455, + "grad_norm": 2.3783411979675293, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7181850671768188, + "num_tokens": 149993329.0, + "step": 5830 + }, + { + "epoch": 0.6403470239402592, + "grad_norm": 2.1743662357330322, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6986130475997925, + "num_tokens": 150019562.0, + "step": 5831 + }, + { + "epoch": 0.6404568416428729, + "grad_norm": 2.06365704536438, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7495601773262024, + "num_tokens": 150044692.0, + "step": 5832 + }, + { + "epoch": 0.6405666593454865, + "grad_norm": 2.2104742527008057, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6846432685852051, + "num_tokens": 150073423.0, + "step": 5833 + }, + { + "epoch": 0.6406764770481002, + "grad_norm": 2.3885934352874756, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6955896615982056, + "num_tokens": 150098225.0, + "step": 5834 + }, + { + "epoch": 0.6407862947507138, + "grad_norm": 2.365666151046753, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6841174364089966, + "num_tokens": 150121980.0, + "step": 5835 + }, + { + "epoch": 0.6408961124533274, + "grad_norm": 2.2937064170837402, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7124522924423218, + "num_tokens": 150144846.0, + "step": 5836 + }, + { + "epoch": 0.6410059301559411, + "grad_norm": 2.0476551055908203, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7202131748199463, + "num_tokens": 150171993.0, + "step": 5837 + }, + { + "epoch": 0.6411157478585549, + "grad_norm": 1.8948719501495361, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6870977282524109, + "num_tokens": 150205704.0, + "step": 5838 + }, + { + "epoch": 0.6412255655611685, + "grad_norm": 2.354541063308716, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7026329040527344, + "num_tokens": 150227645.0, + "step": 5839 + }, + { + "epoch": 0.6413353832637821, + "grad_norm": 2.1539506912231445, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7079535126686096, + "num_tokens": 150252898.0, + "step": 5840 + }, + { + "epoch": 0.6414452009663958, + "grad_norm": 2.0291857719421387, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7262338399887085, + "num_tokens": 150281124.0, + "step": 5841 + }, + { + "epoch": 0.6415550186690094, + "grad_norm": 2.006777763366699, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6948183178901672, + "num_tokens": 150310360.0, + "step": 5842 + }, + { + "epoch": 0.6416648363716231, + "grad_norm": 2.4512736797332764, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.6807532906532288, + "num_tokens": 150334060.0, + "step": 5843 + }, + { + "epoch": 0.6417746540742367, + "grad_norm": 2.18454909324646, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6980183124542236, + "num_tokens": 150360429.0, + "step": 5844 + }, + { + "epoch": 0.6418844717768504, + "grad_norm": 2.1150693893432617, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6995676159858704, + "num_tokens": 150387864.0, + "step": 5845 + }, + { + "epoch": 0.6419942894794641, + "grad_norm": 2.4358298778533936, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6978521347045898, + "num_tokens": 150411185.0, + "step": 5846 + }, + { + "epoch": 0.6421041071820778, + "grad_norm": 2.055316209793091, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7316791415214539, + "num_tokens": 150437437.0, + "step": 5847 + }, + { + "epoch": 0.6422139248846914, + "grad_norm": 2.360412120819092, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7054575681686401, + "num_tokens": 150461108.0, + "step": 5848 + }, + { + "epoch": 0.6423237425873051, + "grad_norm": 2.10628604888916, + "learning_rate": 1e-06, + "loss": 1.0841, + "mean_token_accuracy": 0.6743613481521606, + "num_tokens": 150495693.0, + "step": 5849 + }, + { + "epoch": 0.6424335602899187, + "grad_norm": 2.054267644882202, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6920551061630249, + "num_tokens": 150523858.0, + "step": 5850 + }, + { + "epoch": 0.6425433779925324, + "grad_norm": 2.3753159046173096, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7142702341079712, + "num_tokens": 150545854.0, + "step": 5851 + }, + { + "epoch": 0.642653195695146, + "grad_norm": 1.9470974206924438, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.719035804271698, + "num_tokens": 150573897.0, + "step": 5852 + }, + { + "epoch": 0.6427630133977598, + "grad_norm": 2.0841751098632812, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7058942317962646, + "num_tokens": 150601051.0, + "step": 5853 + }, + { + "epoch": 0.6428728311003734, + "grad_norm": 2.2980427742004395, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7145782709121704, + "num_tokens": 150623293.0, + "step": 5854 + }, + { + "epoch": 0.6429826488029871, + "grad_norm": 2.108959197998047, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7139226198196411, + "num_tokens": 150649856.0, + "step": 5855 + }, + { + "epoch": 0.6430924665056007, + "grad_norm": 2.0947442054748535, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.700631856918335, + "num_tokens": 150678351.0, + "step": 5856 + }, + { + "epoch": 0.6432022842082143, + "grad_norm": 1.9921636581420898, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.711475670337677, + "num_tokens": 150705373.0, + "step": 5857 + }, + { + "epoch": 0.643312101910828, + "grad_norm": 2.007399320602417, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6876354217529297, + "num_tokens": 150734245.0, + "step": 5858 + }, + { + "epoch": 0.6434219196134416, + "grad_norm": 2.082448720932007, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.6928812265396118, + "num_tokens": 150762384.0, + "step": 5859 + }, + { + "epoch": 0.6435317373160554, + "grad_norm": 2.5324454307556152, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6913349032402039, + "num_tokens": 150783599.0, + "step": 5860 + }, + { + "epoch": 0.643641555018669, + "grad_norm": 2.2613632678985596, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6910743713378906, + "num_tokens": 150808717.0, + "step": 5861 + }, + { + "epoch": 0.6437513727212827, + "grad_norm": 2.0716307163238525, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6919230222702026, + "num_tokens": 150839225.0, + "step": 5862 + }, + { + "epoch": 0.6438611904238963, + "grad_norm": 2.2677083015441895, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7101421356201172, + "num_tokens": 150864425.0, + "step": 5863 + }, + { + "epoch": 0.64397100812651, + "grad_norm": 2.091116428375244, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7092190980911255, + "num_tokens": 150892580.0, + "step": 5864 + }, + { + "epoch": 0.6440808258291236, + "grad_norm": 1.9498281478881836, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7140674591064453, + "num_tokens": 150922898.0, + "step": 5865 + }, + { + "epoch": 0.6441906435317373, + "grad_norm": 2.2064473628997803, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6950351595878601, + "num_tokens": 150949014.0, + "step": 5866 + }, + { + "epoch": 0.644300461234351, + "grad_norm": 1.9944250583648682, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7042417526245117, + "num_tokens": 150979429.0, + "step": 5867 + }, + { + "epoch": 0.6444102789369647, + "grad_norm": 2.366642951965332, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7236518859863281, + "num_tokens": 151001992.0, + "step": 5868 + }, + { + "epoch": 0.6445200966395783, + "grad_norm": 2.26436185836792, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7058459520339966, + "num_tokens": 151024883.0, + "step": 5869 + }, + { + "epoch": 0.644629914342192, + "grad_norm": 2.246380090713501, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6936140060424805, + "num_tokens": 151050205.0, + "step": 5870 + }, + { + "epoch": 0.6447397320448056, + "grad_norm": 2.208587646484375, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7099469900131226, + "num_tokens": 151075514.0, + "step": 5871 + }, + { + "epoch": 0.6448495497474193, + "grad_norm": 2.2894318103790283, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7156575918197632, + "num_tokens": 151097801.0, + "step": 5872 + }, + { + "epoch": 0.6449593674500329, + "grad_norm": 2.3669028282165527, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7203360199928284, + "num_tokens": 151120457.0, + "step": 5873 + }, + { + "epoch": 0.6450691851526466, + "grad_norm": 2.2486939430236816, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7095782160758972, + "num_tokens": 151146148.0, + "step": 5874 + }, + { + "epoch": 0.6451790028552603, + "grad_norm": 2.1266400814056396, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7180761694908142, + "num_tokens": 151173550.0, + "step": 5875 + }, + { + "epoch": 0.645288820557874, + "grad_norm": 2.201282501220703, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7296357154846191, + "num_tokens": 151196911.0, + "step": 5876 + }, + { + "epoch": 0.6453986382604876, + "grad_norm": 2.10048508644104, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7029508352279663, + "num_tokens": 151223738.0, + "step": 5877 + }, + { + "epoch": 0.6455084559631012, + "grad_norm": 2.4969542026519775, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7392148375511169, + "num_tokens": 151243024.0, + "step": 5878 + }, + { + "epoch": 0.6456182736657149, + "grad_norm": 1.8834378719329834, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6935774683952332, + "num_tokens": 151274486.0, + "step": 5879 + }, + { + "epoch": 0.6457280913683285, + "grad_norm": 2.4277701377868652, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7418888807296753, + "num_tokens": 151294342.0, + "step": 5880 + }, + { + "epoch": 0.6458379090709422, + "grad_norm": 2.0563337802886963, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6839990019798279, + "num_tokens": 151322432.0, + "step": 5881 + }, + { + "epoch": 0.6459477267735559, + "grad_norm": 2.1565380096435547, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6856589913368225, + "num_tokens": 151350305.0, + "step": 5882 + }, + { + "epoch": 0.6460575444761696, + "grad_norm": 2.166469097137451, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6952938437461853, + "num_tokens": 151375692.0, + "step": 5883 + }, + { + "epoch": 0.6461673621787832, + "grad_norm": 2.08488130569458, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7043468952178955, + "num_tokens": 151404523.0, + "step": 5884 + }, + { + "epoch": 0.6462771798813969, + "grad_norm": 2.208065986633301, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7339807748794556, + "num_tokens": 151428851.0, + "step": 5885 + }, + { + "epoch": 0.6463869975840105, + "grad_norm": 2.1003384590148926, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7242113947868347, + "num_tokens": 151454445.0, + "step": 5886 + }, + { + "epoch": 0.6464968152866242, + "grad_norm": 2.1520018577575684, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6980694532394409, + "num_tokens": 151482785.0, + "step": 5887 + }, + { + "epoch": 0.6466066329892378, + "grad_norm": 2.199183940887451, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7010847330093384, + "num_tokens": 151507036.0, + "step": 5888 + }, + { + "epoch": 0.6467164506918516, + "grad_norm": 2.211507558822632, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7067537307739258, + "num_tokens": 151532109.0, + "step": 5889 + }, + { + "epoch": 0.6468262683944652, + "grad_norm": 2.4457340240478516, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6895573139190674, + "num_tokens": 151555003.0, + "step": 5890 + }, + { + "epoch": 0.6469360860970789, + "grad_norm": 2.2479588985443115, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7086905241012573, + "num_tokens": 151577951.0, + "step": 5891 + }, + { + "epoch": 0.6470459037996925, + "grad_norm": 1.9584883451461792, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7057344913482666, + "num_tokens": 151608409.0, + "step": 5892 + }, + { + "epoch": 0.6471557215023062, + "grad_norm": 2.0520355701446533, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6952494382858276, + "num_tokens": 151637278.0, + "step": 5893 + }, + { + "epoch": 0.6472655392049198, + "grad_norm": 2.4690074920654297, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.718048095703125, + "num_tokens": 151656803.0, + "step": 5894 + }, + { + "epoch": 0.6473753569075335, + "grad_norm": 2.044064998626709, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7163581848144531, + "num_tokens": 151685616.0, + "step": 5895 + }, + { + "epoch": 0.6474851746101472, + "grad_norm": 2.0716023445129395, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.6813029646873474, + "num_tokens": 151713464.0, + "step": 5896 + }, + { + "epoch": 0.6475949923127609, + "grad_norm": 2.3926289081573486, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7283652424812317, + "num_tokens": 151735613.0, + "step": 5897 + }, + { + "epoch": 0.6477048100153745, + "grad_norm": 2.156273365020752, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6964626312255859, + "num_tokens": 151761473.0, + "step": 5898 + }, + { + "epoch": 0.6478146277179881, + "grad_norm": 2.0147244930267334, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6797627210617065, + "num_tokens": 151791794.0, + "step": 5899 + }, + { + "epoch": 0.6479244454206018, + "grad_norm": 1.962073564529419, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7086738348007202, + "num_tokens": 151822005.0, + "step": 5900 + }, + { + "epoch": 0.6480342631232154, + "grad_norm": 2.3134965896606445, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7101004123687744, + "num_tokens": 151845270.0, + "step": 5901 + }, + { + "epoch": 0.6481440808258291, + "grad_norm": 2.119567632675171, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.699578046798706, + "num_tokens": 151872381.0, + "step": 5902 + }, + { + "epoch": 0.6482538985284428, + "grad_norm": 2.1554627418518066, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.709617555141449, + "num_tokens": 151898814.0, + "step": 5903 + }, + { + "epoch": 0.6483637162310565, + "grad_norm": 1.824066400527954, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6783106327056885, + "num_tokens": 151935948.0, + "step": 5904 + }, + { + "epoch": 0.6484735339336701, + "grad_norm": 2.0098788738250732, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.693134069442749, + "num_tokens": 151964396.0, + "step": 5905 + }, + { + "epoch": 0.6485833516362838, + "grad_norm": 2.4649176597595215, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7213287353515625, + "num_tokens": 151983909.0, + "step": 5906 + }, + { + "epoch": 0.6486931693388974, + "grad_norm": 2.2816126346588135, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6929558515548706, + "num_tokens": 152010165.0, + "step": 5907 + }, + { + "epoch": 0.6488029870415111, + "grad_norm": 2.187596559524536, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7204579710960388, + "num_tokens": 152034426.0, + "step": 5908 + }, + { + "epoch": 0.6489128047441247, + "grad_norm": 2.141197443008423, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7091492414474487, + "num_tokens": 152061114.0, + "step": 5909 + }, + { + "epoch": 0.6490226224467384, + "grad_norm": 2.1665396690368652, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.724065899848938, + "num_tokens": 152086342.0, + "step": 5910 + }, + { + "epoch": 0.6491324401493521, + "grad_norm": 2.1783642768859863, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6820489168167114, + "num_tokens": 152113166.0, + "step": 5911 + }, + { + "epoch": 0.6492422578519658, + "grad_norm": 2.181521415710449, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6924370527267456, + "num_tokens": 152138486.0, + "step": 5912 + }, + { + "epoch": 0.6493520755545794, + "grad_norm": 1.8054717779159546, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7059677243232727, + "num_tokens": 152172266.0, + "step": 5913 + }, + { + "epoch": 0.6494618932571931, + "grad_norm": 1.9405531883239746, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7149903774261475, + "num_tokens": 152202281.0, + "step": 5914 + }, + { + "epoch": 0.6495717109598067, + "grad_norm": 2.2017979621887207, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.726649284362793, + "num_tokens": 152228392.0, + "step": 5915 + }, + { + "epoch": 0.6496815286624203, + "grad_norm": 2.196528196334839, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7041060924530029, + "num_tokens": 152252712.0, + "step": 5916 + }, + { + "epoch": 0.649791346365034, + "grad_norm": 1.9952884912490845, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6984347105026245, + "num_tokens": 152282108.0, + "step": 5917 + }, + { + "epoch": 0.6499011640676478, + "grad_norm": 2.153209924697876, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6981788873672485, + "num_tokens": 152310143.0, + "step": 5918 + }, + { + "epoch": 0.6500109817702614, + "grad_norm": 2.4109947681427, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.71466463804245, + "num_tokens": 152332778.0, + "step": 5919 + }, + { + "epoch": 0.650120799472875, + "grad_norm": 2.142735719680786, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7052609324455261, + "num_tokens": 152358532.0, + "step": 5920 + }, + { + "epoch": 0.6502306171754887, + "grad_norm": 2.290515661239624, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7118197679519653, + "num_tokens": 152383262.0, + "step": 5921 + }, + { + "epoch": 0.6503404348781023, + "grad_norm": 2.137960195541382, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6929309368133545, + "num_tokens": 152409632.0, + "step": 5922 + }, + { + "epoch": 0.650450252580716, + "grad_norm": 2.060884952545166, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7017186880111694, + "num_tokens": 152438474.0, + "step": 5923 + }, + { + "epoch": 0.6505600702833296, + "grad_norm": 2.26846981048584, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7346534132957458, + "num_tokens": 152459957.0, + "step": 5924 + }, + { + "epoch": 0.6506698879859434, + "grad_norm": 2.1632635593414307, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6984320282936096, + "num_tokens": 152485477.0, + "step": 5925 + }, + { + "epoch": 0.650779705688557, + "grad_norm": 2.298361301422119, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.695732831954956, + "num_tokens": 152510817.0, + "step": 5926 + }, + { + "epoch": 0.6508895233911707, + "grad_norm": 1.951112151145935, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7074480652809143, + "num_tokens": 152540866.0, + "step": 5927 + }, + { + "epoch": 0.6509993410937843, + "grad_norm": 2.066112518310547, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6868270635604858, + "num_tokens": 152568903.0, + "step": 5928 + }, + { + "epoch": 0.651109158796398, + "grad_norm": 2.0101280212402344, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.682828962802887, + "num_tokens": 152598542.0, + "step": 5929 + }, + { + "epoch": 0.6512189764990116, + "grad_norm": 2.0890209674835205, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6925421953201294, + "num_tokens": 152625441.0, + "step": 5930 + }, + { + "epoch": 0.6513287942016253, + "grad_norm": 2.1754255294799805, + "learning_rate": 1e-06, + "loss": 1.0974, + "mean_token_accuracy": 0.6706594228744507, + "num_tokens": 152653080.0, + "step": 5931 + }, + { + "epoch": 0.651438611904239, + "grad_norm": 1.9129520654678345, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.696392834186554, + "num_tokens": 152684496.0, + "step": 5932 + }, + { + "epoch": 0.6515484296068527, + "grad_norm": 2.392106056213379, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7012045383453369, + "num_tokens": 152705730.0, + "step": 5933 + }, + { + "epoch": 0.6516582473094663, + "grad_norm": 2.025259494781494, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7068486213684082, + "num_tokens": 152734539.0, + "step": 5934 + }, + { + "epoch": 0.65176806501208, + "grad_norm": 2.1549556255340576, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.7028465270996094, + "num_tokens": 152760409.0, + "step": 5935 + }, + { + "epoch": 0.6518778827146936, + "grad_norm": 2.526303291320801, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7315515279769897, + "num_tokens": 152778926.0, + "step": 5936 + }, + { + "epoch": 0.6519877004173072, + "grad_norm": 2.2020530700683594, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7074981927871704, + "num_tokens": 152802378.0, + "step": 5937 + }, + { + "epoch": 0.6520975181199209, + "grad_norm": 2.12524676322937, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6889530420303345, + "num_tokens": 152830332.0, + "step": 5938 + }, + { + "epoch": 0.6522073358225345, + "grad_norm": 2.6176629066467285, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7253761291503906, + "num_tokens": 152848301.0, + "step": 5939 + }, + { + "epoch": 0.6523171535251483, + "grad_norm": 2.2242610454559326, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6972380876541138, + "num_tokens": 152874463.0, + "step": 5940 + }, + { + "epoch": 0.6524269712277619, + "grad_norm": 2.247236490249634, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.709993839263916, + "num_tokens": 152901042.0, + "step": 5941 + }, + { + "epoch": 0.6525367889303756, + "grad_norm": 2.447023868560791, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.683179497718811, + "num_tokens": 152922952.0, + "step": 5942 + }, + { + "epoch": 0.6526466066329892, + "grad_norm": 2.203339099884033, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.688651442527771, + "num_tokens": 152950543.0, + "step": 5943 + }, + { + "epoch": 0.6527564243356029, + "grad_norm": 2.093522787094116, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.706904411315918, + "num_tokens": 152978551.0, + "step": 5944 + }, + { + "epoch": 0.6528662420382165, + "grad_norm": 2.019817352294922, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7076400518417358, + "num_tokens": 153005362.0, + "step": 5945 + }, + { + "epoch": 0.6529760597408302, + "grad_norm": 2.130021333694458, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6760261058807373, + "num_tokens": 153032415.0, + "step": 5946 + }, + { + "epoch": 0.6530858774434439, + "grad_norm": 2.2022790908813477, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7069940567016602, + "num_tokens": 153057324.0, + "step": 5947 + }, + { + "epoch": 0.6531956951460576, + "grad_norm": 2.376410484313965, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7148165702819824, + "num_tokens": 153079694.0, + "step": 5948 + }, + { + "epoch": 0.6533055128486712, + "grad_norm": 2.1717073917388916, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7012433409690857, + "num_tokens": 153105235.0, + "step": 5949 + }, + { + "epoch": 0.6534153305512849, + "grad_norm": 2.413003444671631, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7056198716163635, + "num_tokens": 153129747.0, + "step": 5950 + }, + { + "epoch": 0.6535251482538985, + "grad_norm": 2.500859022140503, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7202852964401245, + "num_tokens": 153150567.0, + "step": 5951 + }, + { + "epoch": 0.6536349659565122, + "grad_norm": 2.079989433288574, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7175581455230713, + "num_tokens": 153177706.0, + "step": 5952 + }, + { + "epoch": 0.6537447836591258, + "grad_norm": 2.1671974658966064, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7216739654541016, + "num_tokens": 153203231.0, + "step": 5953 + }, + { + "epoch": 0.6538546013617396, + "grad_norm": 2.163177013397217, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7054641842842102, + "num_tokens": 153229342.0, + "step": 5954 + }, + { + "epoch": 0.6539644190643532, + "grad_norm": 2.1839187145233154, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7115626931190491, + "num_tokens": 153255815.0, + "step": 5955 + }, + { + "epoch": 0.6540742367669669, + "grad_norm": 2.4969098567962646, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7147175669670105, + "num_tokens": 153276069.0, + "step": 5956 + }, + { + "epoch": 0.6541840544695805, + "grad_norm": 2.1708786487579346, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7314420938491821, + "num_tokens": 153299977.0, + "step": 5957 + }, + { + "epoch": 0.6542938721721941, + "grad_norm": 2.067575454711914, + "learning_rate": 1e-06, + "loss": 1.0891, + "mean_token_accuracy": 0.6825807094573975, + "num_tokens": 153327178.0, + "step": 5958 + }, + { + "epoch": 0.6544036898748078, + "grad_norm": 2.2345101833343506, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7037805318832397, + "num_tokens": 153352485.0, + "step": 5959 + }, + { + "epoch": 0.6545135075774214, + "grad_norm": 2.544560194015503, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7358779907226562, + "num_tokens": 153371069.0, + "step": 5960 + }, + { + "epoch": 0.6546233252800352, + "grad_norm": 2.056939125061035, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7054489254951477, + "num_tokens": 153399239.0, + "step": 5961 + }, + { + "epoch": 0.6547331429826488, + "grad_norm": 2.0221126079559326, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6924146413803101, + "num_tokens": 153428811.0, + "step": 5962 + }, + { + "epoch": 0.6548429606852625, + "grad_norm": 2.1687428951263428, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7167264223098755, + "num_tokens": 153455292.0, + "step": 5963 + }, + { + "epoch": 0.6549527783878761, + "grad_norm": 2.1976094245910645, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.690062403678894, + "num_tokens": 153479616.0, + "step": 5964 + }, + { + "epoch": 0.6550625960904898, + "grad_norm": 1.9089645147323608, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7024313807487488, + "num_tokens": 153512415.0, + "step": 5965 + }, + { + "epoch": 0.6551724137931034, + "grad_norm": 2.143409252166748, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.71628737449646, + "num_tokens": 153538495.0, + "step": 5966 + }, + { + "epoch": 0.6552822314957171, + "grad_norm": 2.024428367614746, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7069859504699707, + "num_tokens": 153565440.0, + "step": 5967 + }, + { + "epoch": 0.6553920491983307, + "grad_norm": 2.161324977874756, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.7012630701065063, + "num_tokens": 153592023.0, + "step": 5968 + }, + { + "epoch": 0.6555018669009445, + "grad_norm": 2.3270421028137207, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.731422483921051, + "num_tokens": 153614164.0, + "step": 5969 + }, + { + "epoch": 0.6556116846035581, + "grad_norm": 2.415402412414551, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7181596159934998, + "num_tokens": 153633173.0, + "step": 5970 + }, + { + "epoch": 0.6557215023061718, + "grad_norm": 2.4257943630218506, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7075750231742859, + "num_tokens": 153656152.0, + "step": 5971 + }, + { + "epoch": 0.6558313200087854, + "grad_norm": 2.074122428894043, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.731083869934082, + "num_tokens": 153682054.0, + "step": 5972 + }, + { + "epoch": 0.6559411377113991, + "grad_norm": 2.26883602142334, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.7010653018951416, + "num_tokens": 153707581.0, + "step": 5973 + }, + { + "epoch": 0.6560509554140127, + "grad_norm": 2.1924173831939697, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.697712779045105, + "num_tokens": 153732822.0, + "step": 5974 + }, + { + "epoch": 0.6561607731166264, + "grad_norm": 2.1381866931915283, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7070566415786743, + "num_tokens": 153759078.0, + "step": 5975 + }, + { + "epoch": 0.6562705908192401, + "grad_norm": 2.1756677627563477, + "learning_rate": 1e-06, + "loss": 1.0743, + "mean_token_accuracy": 0.6797687411308289, + "num_tokens": 153786258.0, + "step": 5976 + }, + { + "epoch": 0.6563804085218538, + "grad_norm": 2.0600850582122803, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7267413139343262, + "num_tokens": 153811929.0, + "step": 5977 + }, + { + "epoch": 0.6564902262244674, + "grad_norm": 2.076134204864502, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6856070756912231, + "num_tokens": 153840099.0, + "step": 5978 + }, + { + "epoch": 0.656600043927081, + "grad_norm": 2.086238145828247, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7051951885223389, + "num_tokens": 153867594.0, + "step": 5979 + }, + { + "epoch": 0.6567098616296947, + "grad_norm": 2.1497867107391357, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7248985767364502, + "num_tokens": 153892444.0, + "step": 5980 + }, + { + "epoch": 0.6568196793323083, + "grad_norm": 2.391904592514038, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6967854499816895, + "num_tokens": 153914716.0, + "step": 5981 + }, + { + "epoch": 0.656929497034922, + "grad_norm": 2.0956063270568848, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7289406061172485, + "num_tokens": 153940455.0, + "step": 5982 + }, + { + "epoch": 0.6570393147375357, + "grad_norm": 2.0637481212615967, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7080518007278442, + "num_tokens": 153968412.0, + "step": 5983 + }, + { + "epoch": 0.6571491324401494, + "grad_norm": 2.1094319820404053, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6856672167778015, + "num_tokens": 153995635.0, + "step": 5984 + }, + { + "epoch": 0.657258950142763, + "grad_norm": 2.1839354038238525, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6831120252609253, + "num_tokens": 154021927.0, + "step": 5985 + }, + { + "epoch": 0.6573687678453767, + "grad_norm": 2.644200325012207, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7332144379615784, + "num_tokens": 154040461.0, + "step": 5986 + }, + { + "epoch": 0.6574785855479903, + "grad_norm": 2.179716110229492, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7238961458206177, + "num_tokens": 154066007.0, + "step": 5987 + }, + { + "epoch": 0.657588403250604, + "grad_norm": 2.208951950073242, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7255426645278931, + "num_tokens": 154090788.0, + "step": 5988 + }, + { + "epoch": 0.6576982209532176, + "grad_norm": 2.3482940196990967, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7095494270324707, + "num_tokens": 154112525.0, + "step": 5989 + }, + { + "epoch": 0.6578080386558314, + "grad_norm": 2.3468177318573, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7096092700958252, + "num_tokens": 154135614.0, + "step": 5990 + }, + { + "epoch": 0.657917856358445, + "grad_norm": 2.417161464691162, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.699212372303009, + "num_tokens": 154159268.0, + "step": 5991 + }, + { + "epoch": 0.6580276740610587, + "grad_norm": 2.219367504119873, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6975650787353516, + "num_tokens": 154184411.0, + "step": 5992 + }, + { + "epoch": 0.6581374917636723, + "grad_norm": 2.015528917312622, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7353945970535278, + "num_tokens": 154211990.0, + "step": 5993 + }, + { + "epoch": 0.658247309466286, + "grad_norm": 2.4857170581817627, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7080093622207642, + "num_tokens": 154231435.0, + "step": 5994 + }, + { + "epoch": 0.6583571271688996, + "grad_norm": 2.0031821727752686, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6995611190795898, + "num_tokens": 154260548.0, + "step": 5995 + }, + { + "epoch": 0.6584669448715132, + "grad_norm": 2.078965663909912, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7019971013069153, + "num_tokens": 154288317.0, + "step": 5996 + }, + { + "epoch": 0.6585767625741269, + "grad_norm": 2.166918992996216, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7274824380874634, + "num_tokens": 154313729.0, + "step": 5997 + }, + { + "epoch": 0.6586865802767407, + "grad_norm": 2.2765345573425293, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.7030313611030579, + "num_tokens": 154337285.0, + "step": 5998 + }, + { + "epoch": 0.6587963979793543, + "grad_norm": 1.915308952331543, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7117683291435242, + "num_tokens": 154369319.0, + "step": 5999 + }, + { + "epoch": 0.6589062156819679, + "grad_norm": 1.9906033277511597, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7214740514755249, + "num_tokens": 154395958.0, + "step": 6000 + }, + { + "epoch": 0.6590160333845816, + "grad_norm": 2.1434624195098877, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6838506460189819, + "num_tokens": 154422068.0, + "step": 6001 + }, + { + "epoch": 0.6591258510871952, + "grad_norm": 2.284202814102173, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7077158689498901, + "num_tokens": 154444761.0, + "step": 6002 + }, + { + "epoch": 0.6592356687898089, + "grad_norm": 2.413919448852539, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7129673361778259, + "num_tokens": 154465919.0, + "step": 6003 + }, + { + "epoch": 0.6593454864924225, + "grad_norm": 2.405073642730713, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7043368816375732, + "num_tokens": 154486684.0, + "step": 6004 + }, + { + "epoch": 0.6594553041950363, + "grad_norm": 2.238217353820801, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7168304324150085, + "num_tokens": 154510048.0, + "step": 6005 + }, + { + "epoch": 0.6595651218976499, + "grad_norm": 2.318175792694092, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7272154688835144, + "num_tokens": 154531673.0, + "step": 6006 + }, + { + "epoch": 0.6596749396002636, + "grad_norm": 2.190690279006958, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7048861980438232, + "num_tokens": 154556895.0, + "step": 6007 + }, + { + "epoch": 0.6597847573028772, + "grad_norm": 2.121415376663208, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6972203850746155, + "num_tokens": 154584724.0, + "step": 6008 + }, + { + "epoch": 0.6598945750054909, + "grad_norm": 2.4818477630615234, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7091836929321289, + "num_tokens": 154605984.0, + "step": 6009 + }, + { + "epoch": 0.6600043927081045, + "grad_norm": 2.3512120246887207, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6943439245223999, + "num_tokens": 154629243.0, + "step": 6010 + }, + { + "epoch": 0.6601142104107182, + "grad_norm": 2.0692105293273926, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7244072556495667, + "num_tokens": 154654856.0, + "step": 6011 + }, + { + "epoch": 0.6602240281133319, + "grad_norm": 2.2663650512695312, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6890522241592407, + "num_tokens": 154679096.0, + "step": 6012 + }, + { + "epoch": 0.6603338458159456, + "grad_norm": 2.3951363563537598, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.70782470703125, + "num_tokens": 154702533.0, + "step": 6013 + }, + { + "epoch": 0.6604436635185592, + "grad_norm": 2.010348081588745, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.693163275718689, + "num_tokens": 154735355.0, + "step": 6014 + }, + { + "epoch": 0.6605534812211729, + "grad_norm": 2.2320685386657715, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6995200514793396, + "num_tokens": 154760966.0, + "step": 6015 + }, + { + "epoch": 0.6606632989237865, + "grad_norm": 2.27148699760437, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6954286098480225, + "num_tokens": 154786473.0, + "step": 6016 + }, + { + "epoch": 0.6607731166264001, + "grad_norm": 2.3318207263946533, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7041863799095154, + "num_tokens": 154808467.0, + "step": 6017 + }, + { + "epoch": 0.6608829343290138, + "grad_norm": 2.139805555343628, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7109603881835938, + "num_tokens": 154832740.0, + "step": 6018 + }, + { + "epoch": 0.6609927520316276, + "grad_norm": 2.337460517883301, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7189757823944092, + "num_tokens": 154854598.0, + "step": 6019 + }, + { + "epoch": 0.6611025697342412, + "grad_norm": 1.9457703828811646, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6743395328521729, + "num_tokens": 154889392.0, + "step": 6020 + }, + { + "epoch": 0.6612123874368548, + "grad_norm": 2.0629382133483887, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7061793208122253, + "num_tokens": 154917554.0, + "step": 6021 + }, + { + "epoch": 0.6613222051394685, + "grad_norm": 2.2280237674713135, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7147597074508667, + "num_tokens": 154942771.0, + "step": 6022 + }, + { + "epoch": 0.6614320228420821, + "grad_norm": 2.261695623397827, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7107374668121338, + "num_tokens": 154966727.0, + "step": 6023 + }, + { + "epoch": 0.6615418405446958, + "grad_norm": 2.1524264812469482, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7014461159706116, + "num_tokens": 154993602.0, + "step": 6024 + }, + { + "epoch": 0.6616516582473094, + "grad_norm": 2.25642466545105, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.72698974609375, + "num_tokens": 155016587.0, + "step": 6025 + }, + { + "epoch": 0.6617614759499231, + "grad_norm": 2.1396431922912598, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7194113731384277, + "num_tokens": 155042091.0, + "step": 6026 + }, + { + "epoch": 0.6618712936525368, + "grad_norm": 2.2015554904937744, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.70867520570755, + "num_tokens": 155067718.0, + "step": 6027 + }, + { + "epoch": 0.6619811113551505, + "grad_norm": 2.4154372215270996, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6953485608100891, + "num_tokens": 155090339.0, + "step": 6028 + }, + { + "epoch": 0.6620909290577641, + "grad_norm": 2.208528995513916, + "learning_rate": 1e-06, + "loss": 1.0917, + "mean_token_accuracy": 0.6755551695823669, + "num_tokens": 155117754.0, + "step": 6029 + }, + { + "epoch": 0.6622007467603778, + "grad_norm": 2.1379234790802, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.698900043964386, + "num_tokens": 155144255.0, + "step": 6030 + }, + { + "epoch": 0.6623105644629914, + "grad_norm": 2.421119213104248, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7218791246414185, + "num_tokens": 155164427.0, + "step": 6031 + }, + { + "epoch": 0.6624203821656051, + "grad_norm": 2.2181451320648193, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7163063287734985, + "num_tokens": 155188261.0, + "step": 6032 + }, + { + "epoch": 0.6625301998682187, + "grad_norm": 2.2417147159576416, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6908177137374878, + "num_tokens": 155211722.0, + "step": 6033 + }, + { + "epoch": 0.6626400175708325, + "grad_norm": 1.8520973920822144, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7116551399230957, + "num_tokens": 155245533.0, + "step": 6034 + }, + { + "epoch": 0.6627498352734461, + "grad_norm": 2.2767491340637207, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6981306076049805, + "num_tokens": 155270101.0, + "step": 6035 + }, + { + "epoch": 0.6628596529760598, + "grad_norm": 2.1019186973571777, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7283504605293274, + "num_tokens": 155296185.0, + "step": 6036 + }, + { + "epoch": 0.6629694706786734, + "grad_norm": 2.2194888591766357, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6960148811340332, + "num_tokens": 155321979.0, + "step": 6037 + }, + { + "epoch": 0.663079288381287, + "grad_norm": 1.964918851852417, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7274083495140076, + "num_tokens": 155350222.0, + "step": 6038 + }, + { + "epoch": 0.6631891060839007, + "grad_norm": 2.3291971683502197, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7364394664764404, + "num_tokens": 155371760.0, + "step": 6039 + }, + { + "epoch": 0.6632989237865143, + "grad_norm": 2.1327710151672363, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6840510368347168, + "num_tokens": 155400664.0, + "step": 6040 + }, + { + "epoch": 0.6634087414891281, + "grad_norm": 2.486917495727539, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.721219003200531, + "num_tokens": 155419513.0, + "step": 6041 + }, + { + "epoch": 0.6635185591917417, + "grad_norm": 1.994014024734497, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7042534351348877, + "num_tokens": 155452425.0, + "step": 6042 + }, + { + "epoch": 0.6636283768943554, + "grad_norm": 2.5902888774871826, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7259820103645325, + "num_tokens": 155471230.0, + "step": 6043 + }, + { + "epoch": 0.663738194596969, + "grad_norm": 2.4262351989746094, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7064720392227173, + "num_tokens": 155492601.0, + "step": 6044 + }, + { + "epoch": 0.6638480122995827, + "grad_norm": 2.4527645111083984, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7245683670043945, + "num_tokens": 155513403.0, + "step": 6045 + }, + { + "epoch": 0.6639578300021963, + "grad_norm": 2.147674083709717, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.7099733352661133, + "num_tokens": 155540618.0, + "step": 6046 + }, + { + "epoch": 0.66406764770481, + "grad_norm": 2.2978641986846924, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7187396883964539, + "num_tokens": 155563996.0, + "step": 6047 + }, + { + "epoch": 0.6641774654074237, + "grad_norm": 2.0750277042388916, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.704806923866272, + "num_tokens": 155592044.0, + "step": 6048 + }, + { + "epoch": 0.6642872831100374, + "grad_norm": 2.8521814346313477, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7256725430488586, + "num_tokens": 155609451.0, + "step": 6049 + }, + { + "epoch": 0.664397100812651, + "grad_norm": 2.1515848636627197, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6939064264297485, + "num_tokens": 155635315.0, + "step": 6050 + }, + { + "epoch": 0.6645069185152647, + "grad_norm": 2.1975250244140625, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7150854468345642, + "num_tokens": 155660860.0, + "step": 6051 + }, + { + "epoch": 0.6646167362178783, + "grad_norm": 2.0073070526123047, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7020412683486938, + "num_tokens": 155689920.0, + "step": 6052 + }, + { + "epoch": 0.664726553920492, + "grad_norm": 2.0958871841430664, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7148764133453369, + "num_tokens": 155715779.0, + "step": 6053 + }, + { + "epoch": 0.6648363716231056, + "grad_norm": 2.1706700325012207, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7217705845832825, + "num_tokens": 155738814.0, + "step": 6054 + }, + { + "epoch": 0.6649461893257194, + "grad_norm": 2.1605191230773926, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.682092547416687, + "num_tokens": 155768530.0, + "step": 6055 + }, + { + "epoch": 0.665056007028333, + "grad_norm": 1.8584234714508057, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6995763778686523, + "num_tokens": 155803819.0, + "step": 6056 + }, + { + "epoch": 0.6651658247309467, + "grad_norm": 2.4966490268707275, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7261990308761597, + "num_tokens": 155822389.0, + "step": 6057 + }, + { + "epoch": 0.6652756424335603, + "grad_norm": 2.2996277809143066, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.72479248046875, + "num_tokens": 155843425.0, + "step": 6058 + }, + { + "epoch": 0.665385460136174, + "grad_norm": 2.2482404708862305, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7098114490509033, + "num_tokens": 155866801.0, + "step": 6059 + }, + { + "epoch": 0.6654952778387876, + "grad_norm": 2.2451536655426025, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7275550365447998, + "num_tokens": 155889007.0, + "step": 6060 + }, + { + "epoch": 0.6656050955414012, + "grad_norm": 2.0561683177948, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7296585440635681, + "num_tokens": 155915941.0, + "step": 6061 + }, + { + "epoch": 0.6657149132440149, + "grad_norm": 1.9429363012313843, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6791365742683411, + "num_tokens": 155946421.0, + "step": 6062 + }, + { + "epoch": 0.6658247309466286, + "grad_norm": 2.1493427753448486, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6922880411148071, + "num_tokens": 155973643.0, + "step": 6063 + }, + { + "epoch": 0.6659345486492423, + "grad_norm": 2.3422412872314453, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7130041122436523, + "num_tokens": 155996462.0, + "step": 6064 + }, + { + "epoch": 0.6660443663518559, + "grad_norm": 2.4189579486846924, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7316716909408569, + "num_tokens": 156018487.0, + "step": 6065 + }, + { + "epoch": 0.6661541840544696, + "grad_norm": 2.1670992374420166, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.728481650352478, + "num_tokens": 156042615.0, + "step": 6066 + }, + { + "epoch": 0.6662640017570832, + "grad_norm": 2.2995388507843018, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7095268964767456, + "num_tokens": 156065935.0, + "step": 6067 + }, + { + "epoch": 0.6663738194596969, + "grad_norm": 2.417238473892212, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7396665811538696, + "num_tokens": 156086790.0, + "step": 6068 + }, + { + "epoch": 0.6664836371623105, + "grad_norm": 2.3402726650238037, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6846518516540527, + "num_tokens": 156110284.0, + "step": 6069 + }, + { + "epoch": 0.6665934548649243, + "grad_norm": 1.9222023487091064, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6983394622802734, + "num_tokens": 156139526.0, + "step": 6070 + }, + { + "epoch": 0.6667032725675379, + "grad_norm": 2.1922671794891357, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7273356914520264, + "num_tokens": 156163668.0, + "step": 6071 + }, + { + "epoch": 0.6668130902701516, + "grad_norm": 1.876138687133789, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6995519399642944, + "num_tokens": 156196825.0, + "step": 6072 + }, + { + "epoch": 0.6669229079727652, + "grad_norm": 1.9111372232437134, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6668102741241455, + "num_tokens": 156231168.0, + "step": 6073 + }, + { + "epoch": 0.6670327256753789, + "grad_norm": 2.1178975105285645, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7096297740936279, + "num_tokens": 156259403.0, + "step": 6074 + }, + { + "epoch": 0.6671425433779925, + "grad_norm": 2.137418031692505, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6896429061889648, + "num_tokens": 156286959.0, + "step": 6075 + }, + { + "epoch": 0.6672523610806061, + "grad_norm": 2.2146859169006348, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7120387554168701, + "num_tokens": 156311740.0, + "step": 6076 + }, + { + "epoch": 0.6673621787832199, + "grad_norm": 2.0625367164611816, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7143901586532593, + "num_tokens": 156339933.0, + "step": 6077 + }, + { + "epoch": 0.6674719964858336, + "grad_norm": 2.411738872528076, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6969250440597534, + "num_tokens": 156362735.0, + "step": 6078 + }, + { + "epoch": 0.6675818141884472, + "grad_norm": 2.222597599029541, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7075409293174744, + "num_tokens": 156387140.0, + "step": 6079 + }, + { + "epoch": 0.6676916318910608, + "grad_norm": 2.1312255859375, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6897606253623962, + "num_tokens": 156413623.0, + "step": 6080 + }, + { + "epoch": 0.6678014495936745, + "grad_norm": 2.216871500015259, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7056401968002319, + "num_tokens": 156437903.0, + "step": 6081 + }, + { + "epoch": 0.6679112672962881, + "grad_norm": 2.299391269683838, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6974587440490723, + "num_tokens": 156462874.0, + "step": 6082 + }, + { + "epoch": 0.6680210849989018, + "grad_norm": 2.324998378753662, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7017682194709778, + "num_tokens": 156487193.0, + "step": 6083 + }, + { + "epoch": 0.6681309027015155, + "grad_norm": 2.2217631340026855, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7143996357917786, + "num_tokens": 156512727.0, + "step": 6084 + }, + { + "epoch": 0.6682407204041292, + "grad_norm": 2.0581443309783936, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6922674179077148, + "num_tokens": 156542521.0, + "step": 6085 + }, + { + "epoch": 0.6683505381067428, + "grad_norm": 2.299264669418335, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7152203321456909, + "num_tokens": 156567211.0, + "step": 6086 + }, + { + "epoch": 0.6684603558093565, + "grad_norm": 1.8941878080368042, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7014445066452026, + "num_tokens": 156599041.0, + "step": 6087 + }, + { + "epoch": 0.6685701735119701, + "grad_norm": 2.051727533340454, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7149038910865784, + "num_tokens": 156626697.0, + "step": 6088 + }, + { + "epoch": 0.6686799912145838, + "grad_norm": 2.284029483795166, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7209103107452393, + "num_tokens": 156650054.0, + "step": 6089 + }, + { + "epoch": 0.6687898089171974, + "grad_norm": 2.0230307579040527, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6992787718772888, + "num_tokens": 156681115.0, + "step": 6090 + }, + { + "epoch": 0.6688996266198111, + "grad_norm": 2.0045347213745117, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6929566860198975, + "num_tokens": 156711977.0, + "step": 6091 + }, + { + "epoch": 0.6690094443224248, + "grad_norm": 2.184255361557007, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7123950719833374, + "num_tokens": 156737277.0, + "step": 6092 + }, + { + "epoch": 0.6691192620250385, + "grad_norm": 2.2717225551605225, + "learning_rate": 1e-06, + "loss": 1.0493, + "mean_token_accuracy": 0.6869438886642456, + "num_tokens": 156762413.0, + "step": 6093 + }, + { + "epoch": 0.6692290797276521, + "grad_norm": 2.0178141593933105, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7221997380256653, + "num_tokens": 156789941.0, + "step": 6094 + }, + { + "epoch": 0.6693388974302658, + "grad_norm": 2.202752113342285, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7359248399734497, + "num_tokens": 156813226.0, + "step": 6095 + }, + { + "epoch": 0.6694487151328794, + "grad_norm": 2.1435954570770264, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6898919343948364, + "num_tokens": 156842965.0, + "step": 6096 + }, + { + "epoch": 0.669558532835493, + "grad_norm": 2.0744996070861816, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7113344669342041, + "num_tokens": 156870562.0, + "step": 6097 + }, + { + "epoch": 0.6696683505381067, + "grad_norm": 1.9846279621124268, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6967446804046631, + "num_tokens": 156900401.0, + "step": 6098 + }, + { + "epoch": 0.6697781682407205, + "grad_norm": 1.9870790243148804, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6892921924591064, + "num_tokens": 156930694.0, + "step": 6099 + }, + { + "epoch": 0.6698879859433341, + "grad_norm": 2.205564498901367, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6682689189910889, + "num_tokens": 156957820.0, + "step": 6100 + }, + { + "epoch": 0.6699978036459477, + "grad_norm": 2.086886167526245, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6975479125976562, + "num_tokens": 156985032.0, + "step": 6101 + }, + { + "epoch": 0.6701076213485614, + "grad_norm": 2.3318796157836914, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7217059135437012, + "num_tokens": 157007423.0, + "step": 6102 + }, + { + "epoch": 0.670217439051175, + "grad_norm": 2.34012508392334, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7164024114608765, + "num_tokens": 157029991.0, + "step": 6103 + }, + { + "epoch": 0.6703272567537887, + "grad_norm": 2.1755192279815674, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7055257558822632, + "num_tokens": 157055119.0, + "step": 6104 + }, + { + "epoch": 0.6704370744564023, + "grad_norm": 2.5184998512268066, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.740805983543396, + "num_tokens": 157075308.0, + "step": 6105 + }, + { + "epoch": 0.6705468921590161, + "grad_norm": 2.4082765579223633, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7135633826255798, + "num_tokens": 157095468.0, + "step": 6106 + }, + { + "epoch": 0.6706567098616297, + "grad_norm": 2.424154758453369, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7209858894348145, + "num_tokens": 157117158.0, + "step": 6107 + }, + { + "epoch": 0.6707665275642434, + "grad_norm": 2.3344337940216064, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7032886147499084, + "num_tokens": 157139889.0, + "step": 6108 + }, + { + "epoch": 0.670876345266857, + "grad_norm": 2.2249677181243896, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6957396268844604, + "num_tokens": 157164507.0, + "step": 6109 + }, + { + "epoch": 0.6709861629694707, + "grad_norm": 2.4083914756774902, + "learning_rate": 1e-06, + "loss": 1.0702, + "mean_token_accuracy": 0.6846931576728821, + "num_tokens": 157186692.0, + "step": 6110 + }, + { + "epoch": 0.6710959806720843, + "grad_norm": 2.073537826538086, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6965595483779907, + "num_tokens": 157215144.0, + "step": 6111 + }, + { + "epoch": 0.671205798374698, + "grad_norm": 2.152214765548706, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.70180743932724, + "num_tokens": 157243687.0, + "step": 6112 + }, + { + "epoch": 0.6713156160773117, + "grad_norm": 2.186326503753662, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6938363313674927, + "num_tokens": 157270024.0, + "step": 6113 + }, + { + "epoch": 0.6714254337799254, + "grad_norm": 2.344468593597412, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7172191143035889, + "num_tokens": 157291582.0, + "step": 6114 + }, + { + "epoch": 0.671535251482539, + "grad_norm": 2.4798622131347656, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7183802723884583, + "num_tokens": 157312277.0, + "step": 6115 + }, + { + "epoch": 0.6716450691851527, + "grad_norm": 2.15151047706604, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.720097541809082, + "num_tokens": 157337233.0, + "step": 6116 + }, + { + "epoch": 0.6717548868877663, + "grad_norm": 2.1845405101776123, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7070937156677246, + "num_tokens": 157362522.0, + "step": 6117 + }, + { + "epoch": 0.67186470459038, + "grad_norm": 2.103163242340088, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6985121965408325, + "num_tokens": 157391645.0, + "step": 6118 + }, + { + "epoch": 0.6719745222929936, + "grad_norm": 2.136132001876831, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7052803039550781, + "num_tokens": 157416450.0, + "step": 6119 + }, + { + "epoch": 0.6720843399956072, + "grad_norm": 1.9013029336929321, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7109850645065308, + "num_tokens": 157448265.0, + "step": 6120 + }, + { + "epoch": 0.672194157698221, + "grad_norm": 2.0992226600646973, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7113909721374512, + "num_tokens": 157474535.0, + "step": 6121 + }, + { + "epoch": 0.6723039754008346, + "grad_norm": 2.179661273956299, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7174326777458191, + "num_tokens": 157498108.0, + "step": 6122 + }, + { + "epoch": 0.6724137931034483, + "grad_norm": 2.2201457023620605, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7004340887069702, + "num_tokens": 157524741.0, + "step": 6123 + }, + { + "epoch": 0.6725236108060619, + "grad_norm": 2.2254981994628906, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7268871068954468, + "num_tokens": 157547270.0, + "step": 6124 + }, + { + "epoch": 0.6726334285086756, + "grad_norm": 2.0940935611724854, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7029715180397034, + "num_tokens": 157574655.0, + "step": 6125 + }, + { + "epoch": 0.6727432462112892, + "grad_norm": 2.527312755584717, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7269982099533081, + "num_tokens": 157593261.0, + "step": 6126 + }, + { + "epoch": 0.6728530639139029, + "grad_norm": 2.176367998123169, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7277082204818726, + "num_tokens": 157617512.0, + "step": 6127 + }, + { + "epoch": 0.6729628816165166, + "grad_norm": 2.3426015377044678, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.72771817445755, + "num_tokens": 157640150.0, + "step": 6128 + }, + { + "epoch": 0.6730726993191303, + "grad_norm": 2.05422306060791, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7073367238044739, + "num_tokens": 157668009.0, + "step": 6129 + }, + { + "epoch": 0.6731825170217439, + "grad_norm": 2.2301201820373535, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6968602538108826, + "num_tokens": 157692980.0, + "step": 6130 + }, + { + "epoch": 0.6732923347243576, + "grad_norm": 2.1937224864959717, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7441020607948303, + "num_tokens": 157715697.0, + "step": 6131 + }, + { + "epoch": 0.6734021524269712, + "grad_norm": 2.318912982940674, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6989679336547852, + "num_tokens": 157739308.0, + "step": 6132 + }, + { + "epoch": 0.6735119701295849, + "grad_norm": 2.6109495162963867, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7197506427764893, + "num_tokens": 157758005.0, + "step": 6133 + }, + { + "epoch": 0.6736217878321985, + "grad_norm": 2.1402971744537354, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6915971040725708, + "num_tokens": 157786634.0, + "step": 6134 + }, + { + "epoch": 0.6737316055348123, + "grad_norm": 2.0722482204437256, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7195273041725159, + "num_tokens": 157814049.0, + "step": 6135 + }, + { + "epoch": 0.6738414232374259, + "grad_norm": 2.336698055267334, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7068004608154297, + "num_tokens": 157837568.0, + "step": 6136 + }, + { + "epoch": 0.6739512409400396, + "grad_norm": 2.4069466590881348, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.6831932067871094, + "num_tokens": 157860447.0, + "step": 6137 + }, + { + "epoch": 0.6740610586426532, + "grad_norm": 2.3842523097991943, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7257022857666016, + "num_tokens": 157882052.0, + "step": 6138 + }, + { + "epoch": 0.6741708763452668, + "grad_norm": 2.218501567840576, + "learning_rate": 1e-06, + "loss": 1.0747, + "mean_token_accuracy": 0.6790105104446411, + "num_tokens": 157906632.0, + "step": 6139 + }, + { + "epoch": 0.6742806940478805, + "grad_norm": 2.2343297004699707, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6935964822769165, + "num_tokens": 157931881.0, + "step": 6140 + }, + { + "epoch": 0.6743905117504941, + "grad_norm": 2.1860616207122803, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.730485737323761, + "num_tokens": 157957759.0, + "step": 6141 + }, + { + "epoch": 0.6745003294531079, + "grad_norm": 2.1081793308258057, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6839648485183716, + "num_tokens": 157988990.0, + "step": 6142 + }, + { + "epoch": 0.6746101471557215, + "grad_norm": 2.2035796642303467, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6853454113006592, + "num_tokens": 158014476.0, + "step": 6143 + }, + { + "epoch": 0.6747199648583352, + "grad_norm": 2.3709263801574707, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7080776691436768, + "num_tokens": 158037503.0, + "step": 6144 + }, + { + "epoch": 0.6748297825609488, + "grad_norm": 2.463289976119995, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7199960947036743, + "num_tokens": 158057043.0, + "step": 6145 + }, + { + "epoch": 0.6749396002635625, + "grad_norm": 2.366701602935791, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.710201621055603, + "num_tokens": 158079557.0, + "step": 6146 + }, + { + "epoch": 0.6750494179661761, + "grad_norm": 2.00342059135437, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.701246440410614, + "num_tokens": 158110314.0, + "step": 6147 + }, + { + "epoch": 0.6751592356687898, + "grad_norm": 2.277249574661255, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7068841457366943, + "num_tokens": 158134077.0, + "step": 6148 + }, + { + "epoch": 0.6752690533714034, + "grad_norm": 1.9987550973892212, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7140285968780518, + "num_tokens": 158164070.0, + "step": 6149 + }, + { + "epoch": 0.6753788710740172, + "grad_norm": 2.3616130352020264, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7302936315536499, + "num_tokens": 158186409.0, + "step": 6150 + }, + { + "epoch": 0.6754886887766308, + "grad_norm": 2.26375412940979, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6945233941078186, + "num_tokens": 158211692.0, + "step": 6151 + }, + { + "epoch": 0.6755985064792445, + "grad_norm": 1.8912029266357422, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7160463929176331, + "num_tokens": 158243216.0, + "step": 6152 + }, + { + "epoch": 0.6757083241818581, + "grad_norm": 2.2681539058685303, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7047473192214966, + "num_tokens": 158266738.0, + "step": 6153 + }, + { + "epoch": 0.6758181418844718, + "grad_norm": 1.916961431503296, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6821165084838867, + "num_tokens": 158299517.0, + "step": 6154 + }, + { + "epoch": 0.6759279595870854, + "grad_norm": 2.5425851345062256, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7150071859359741, + "num_tokens": 158318823.0, + "step": 6155 + }, + { + "epoch": 0.676037777289699, + "grad_norm": 2.1202800273895264, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7059085965156555, + "num_tokens": 158345542.0, + "step": 6156 + }, + { + "epoch": 0.6761475949923128, + "grad_norm": 2.2490265369415283, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.703389585018158, + "num_tokens": 158369871.0, + "step": 6157 + }, + { + "epoch": 0.6762574126949265, + "grad_norm": 1.980785608291626, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6879414319992065, + "num_tokens": 158402763.0, + "step": 6158 + }, + { + "epoch": 0.6763672303975401, + "grad_norm": 2.5102570056915283, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7177972793579102, + "num_tokens": 158422906.0, + "step": 6159 + }, + { + "epoch": 0.6764770481001537, + "grad_norm": 2.411846876144409, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7160004377365112, + "num_tokens": 158444125.0, + "step": 6160 + }, + { + "epoch": 0.6765868658027674, + "grad_norm": 2.1921262741088867, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7022244334220886, + "num_tokens": 158468990.0, + "step": 6161 + }, + { + "epoch": 0.676696683505381, + "grad_norm": 1.965964436531067, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7107797265052795, + "num_tokens": 158499078.0, + "step": 6162 + }, + { + "epoch": 0.6768065012079947, + "grad_norm": 2.324751853942871, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7072798013687134, + "num_tokens": 158521679.0, + "step": 6163 + }, + { + "epoch": 0.6769163189106084, + "grad_norm": 2.3611233234405518, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7168701887130737, + "num_tokens": 158544902.0, + "step": 6164 + }, + { + "epoch": 0.6770261366132221, + "grad_norm": 2.245832920074463, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6980427503585815, + "num_tokens": 158569549.0, + "step": 6165 + }, + { + "epoch": 0.6771359543158357, + "grad_norm": 2.2263646125793457, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7036623358726501, + "num_tokens": 158594496.0, + "step": 6166 + }, + { + "epoch": 0.6772457720184494, + "grad_norm": 1.8677246570587158, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7183462381362915, + "num_tokens": 158627401.0, + "step": 6167 + }, + { + "epoch": 0.677355589721063, + "grad_norm": 2.0628089904785156, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6812512874603271, + "num_tokens": 158657718.0, + "step": 6168 + }, + { + "epoch": 0.6774654074236767, + "grad_norm": 1.9390654563903809, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6899641752243042, + "num_tokens": 158690076.0, + "step": 6169 + }, + { + "epoch": 0.6775752251262903, + "grad_norm": 2.274052619934082, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7258262634277344, + "num_tokens": 158713578.0, + "step": 6170 + }, + { + "epoch": 0.6776850428289041, + "grad_norm": 2.1539173126220703, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7167870998382568, + "num_tokens": 158738161.0, + "step": 6171 + }, + { + "epoch": 0.6777948605315177, + "grad_norm": 1.9103647470474243, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7027996778488159, + "num_tokens": 158771310.0, + "step": 6172 + }, + { + "epoch": 0.6779046782341314, + "grad_norm": 2.2727224826812744, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6859316825866699, + "num_tokens": 158796149.0, + "step": 6173 + }, + { + "epoch": 0.678014495936745, + "grad_norm": 2.0172386169433594, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7016138434410095, + "num_tokens": 158826040.0, + "step": 6174 + }, + { + "epoch": 0.6781243136393587, + "grad_norm": 1.995863914489746, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.7098023891448975, + "num_tokens": 158857137.0, + "step": 6175 + }, + { + "epoch": 0.6782341313419723, + "grad_norm": 2.2133266925811768, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7246675491333008, + "num_tokens": 158880946.0, + "step": 6176 + }, + { + "epoch": 0.678343949044586, + "grad_norm": 1.9295302629470825, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.705134391784668, + "num_tokens": 158912272.0, + "step": 6177 + }, + { + "epoch": 0.6784537667471996, + "grad_norm": 2.0783491134643555, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7105813026428223, + "num_tokens": 158939532.0, + "step": 6178 + }, + { + "epoch": 0.6785635844498134, + "grad_norm": 2.2082934379577637, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6840492486953735, + "num_tokens": 158966960.0, + "step": 6179 + }, + { + "epoch": 0.678673402152427, + "grad_norm": 2.488410711288452, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7309929728507996, + "num_tokens": 158987244.0, + "step": 6180 + }, + { + "epoch": 0.6787832198550406, + "grad_norm": 1.909789800643921, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7111996412277222, + "num_tokens": 159017976.0, + "step": 6181 + }, + { + "epoch": 0.6788930375576543, + "grad_norm": 2.2185165882110596, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7124172449111938, + "num_tokens": 159042048.0, + "step": 6182 + }, + { + "epoch": 0.6790028552602679, + "grad_norm": 2.1746037006378174, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7075614929199219, + "num_tokens": 159067871.0, + "step": 6183 + }, + { + "epoch": 0.6791126729628816, + "grad_norm": 2.3489975929260254, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7258028984069824, + "num_tokens": 159089438.0, + "step": 6184 + }, + { + "epoch": 0.6792224906654952, + "grad_norm": 2.0303211212158203, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6925380229949951, + "num_tokens": 159119989.0, + "step": 6185 + }, + { + "epoch": 0.679332308368109, + "grad_norm": 2.1454739570617676, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6856024265289307, + "num_tokens": 159148780.0, + "step": 6186 + }, + { + "epoch": 0.6794421260707226, + "grad_norm": 2.1798412799835205, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7157810926437378, + "num_tokens": 159172239.0, + "step": 6187 + }, + { + "epoch": 0.6795519437733363, + "grad_norm": 2.2428181171417236, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7526736259460449, + "num_tokens": 159194105.0, + "step": 6188 + }, + { + "epoch": 0.6796617614759499, + "grad_norm": 2.1371452808380127, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7165805101394653, + "num_tokens": 159220857.0, + "step": 6189 + }, + { + "epoch": 0.6797715791785636, + "grad_norm": 2.3304646015167236, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7203541994094849, + "num_tokens": 159243869.0, + "step": 6190 + }, + { + "epoch": 0.6798813968811772, + "grad_norm": 2.442559242248535, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7231483459472656, + "num_tokens": 159265251.0, + "step": 6191 + }, + { + "epoch": 0.6799912145837909, + "grad_norm": 2.252459764480591, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7138552069664001, + "num_tokens": 159288861.0, + "step": 6192 + }, + { + "epoch": 0.6801010322864046, + "grad_norm": 2.2837541103363037, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7178639769554138, + "num_tokens": 159311988.0, + "step": 6193 + }, + { + "epoch": 0.6802108499890183, + "grad_norm": 2.16097354888916, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6952047348022461, + "num_tokens": 159337984.0, + "step": 6194 + }, + { + "epoch": 0.6803206676916319, + "grad_norm": 2.0832180976867676, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6973124742507935, + "num_tokens": 159366062.0, + "step": 6195 + }, + { + "epoch": 0.6804304853942456, + "grad_norm": 2.106106758117676, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6982748508453369, + "num_tokens": 159395127.0, + "step": 6196 + }, + { + "epoch": 0.6805403030968592, + "grad_norm": 2.1758127212524414, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.6964787244796753, + "num_tokens": 159420371.0, + "step": 6197 + }, + { + "epoch": 0.6806501207994728, + "grad_norm": 2.1337194442749023, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7106676697731018, + "num_tokens": 159446583.0, + "step": 6198 + }, + { + "epoch": 0.6807599385020865, + "grad_norm": 2.612592935562134, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.697563886642456, + "num_tokens": 159467992.0, + "step": 6199 + }, + { + "epoch": 0.6808697562047002, + "grad_norm": 2.2610836029052734, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6965278387069702, + "num_tokens": 159492742.0, + "step": 6200 + }, + { + "epoch": 0.6809795739073139, + "grad_norm": 1.956505298614502, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6888810396194458, + "num_tokens": 159523702.0, + "step": 6201 + }, + { + "epoch": 0.6810893916099275, + "grad_norm": 2.347626209259033, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7306855916976929, + "num_tokens": 159545809.0, + "step": 6202 + }, + { + "epoch": 0.6811992093125412, + "grad_norm": 2.1227381229400635, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6908413171768188, + "num_tokens": 159574395.0, + "step": 6203 + }, + { + "epoch": 0.6813090270151548, + "grad_norm": 2.199232339859009, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7098571062088013, + "num_tokens": 159598870.0, + "step": 6204 + }, + { + "epoch": 0.6814188447177685, + "grad_norm": 2.155169725418091, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7031927108764648, + "num_tokens": 159623296.0, + "step": 6205 + }, + { + "epoch": 0.6815286624203821, + "grad_norm": 1.9439926147460938, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6835960745811462, + "num_tokens": 159653541.0, + "step": 6206 + }, + { + "epoch": 0.6816384801229959, + "grad_norm": 2.201859474182129, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6855729222297668, + "num_tokens": 159679391.0, + "step": 6207 + }, + { + "epoch": 0.6817482978256095, + "grad_norm": 2.1495978832244873, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7121008634567261, + "num_tokens": 159706504.0, + "step": 6208 + }, + { + "epoch": 0.6818581155282232, + "grad_norm": 2.1748476028442383, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7375431060791016, + "num_tokens": 159730854.0, + "step": 6209 + }, + { + "epoch": 0.6819679332308368, + "grad_norm": 2.109126567840576, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6934147477149963, + "num_tokens": 159757407.0, + "step": 6210 + }, + { + "epoch": 0.6820777509334505, + "grad_norm": 2.1910059452056885, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6838786602020264, + "num_tokens": 159783683.0, + "step": 6211 + }, + { + "epoch": 0.6821875686360641, + "grad_norm": 2.2637367248535156, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6894454956054688, + "num_tokens": 159809211.0, + "step": 6212 + }, + { + "epoch": 0.6822973863386778, + "grad_norm": 2.165323495864868, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6893441677093506, + "num_tokens": 159836583.0, + "step": 6213 + }, + { + "epoch": 0.6824072040412914, + "grad_norm": 2.3172454833984375, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7067255973815918, + "num_tokens": 159859954.0, + "step": 6214 + }, + { + "epoch": 0.6825170217439052, + "grad_norm": 2.416569232940674, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7311301827430725, + "num_tokens": 159880459.0, + "step": 6215 + }, + { + "epoch": 0.6826268394465188, + "grad_norm": 2.442782402038574, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7271001935005188, + "num_tokens": 159900345.0, + "step": 6216 + }, + { + "epoch": 0.6827366571491325, + "grad_norm": 2.437040328979492, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7088888883590698, + "num_tokens": 159923257.0, + "step": 6217 + }, + { + "epoch": 0.6828464748517461, + "grad_norm": 2.01619291305542, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.6808167695999146, + "num_tokens": 159954595.0, + "step": 6218 + }, + { + "epoch": 0.6829562925543597, + "grad_norm": 2.298065423965454, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7127090692520142, + "num_tokens": 159977940.0, + "step": 6219 + }, + { + "epoch": 0.6830661102569734, + "grad_norm": 2.3569695949554443, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7038769721984863, + "num_tokens": 160000496.0, + "step": 6220 + }, + { + "epoch": 0.683175927959587, + "grad_norm": 1.930912971496582, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6961072683334351, + "num_tokens": 160032035.0, + "step": 6221 + }, + { + "epoch": 0.6832857456622008, + "grad_norm": 2.142516851425171, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7176415324211121, + "num_tokens": 160059734.0, + "step": 6222 + }, + { + "epoch": 0.6833955633648144, + "grad_norm": 2.335995674133301, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7103421092033386, + "num_tokens": 160081812.0, + "step": 6223 + }, + { + "epoch": 0.6835053810674281, + "grad_norm": 2.3207600116729736, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7077023983001709, + "num_tokens": 160107574.0, + "step": 6224 + }, + { + "epoch": 0.6836151987700417, + "grad_norm": 2.366694450378418, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7122288346290588, + "num_tokens": 160129198.0, + "step": 6225 + }, + { + "epoch": 0.6837250164726554, + "grad_norm": 2.1801130771636963, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7040823101997375, + "num_tokens": 160154565.0, + "step": 6226 + }, + { + "epoch": 0.683834834175269, + "grad_norm": 2.155627489089966, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6849865913391113, + "num_tokens": 160183165.0, + "step": 6227 + }, + { + "epoch": 0.6839446518778827, + "grad_norm": 2.2402586936950684, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7205874919891357, + "num_tokens": 160206282.0, + "step": 6228 + }, + { + "epoch": 0.6840544695804964, + "grad_norm": 2.138059139251709, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7112983465194702, + "num_tokens": 160233117.0, + "step": 6229 + }, + { + "epoch": 0.6841642872831101, + "grad_norm": 2.00608229637146, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7184141278266907, + "num_tokens": 160261618.0, + "step": 6230 + }, + { + "epoch": 0.6842741049857237, + "grad_norm": 2.1818618774414062, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7209591865539551, + "num_tokens": 160288376.0, + "step": 6231 + }, + { + "epoch": 0.6843839226883374, + "grad_norm": 1.9560651779174805, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6912045478820801, + "num_tokens": 160320933.0, + "step": 6232 + }, + { + "epoch": 0.684493740390951, + "grad_norm": 2.537022113800049, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7298998832702637, + "num_tokens": 160340262.0, + "step": 6233 + }, + { + "epoch": 0.6846035580935647, + "grad_norm": 2.254932403564453, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7183398604393005, + "num_tokens": 160363518.0, + "step": 6234 + }, + { + "epoch": 0.6847133757961783, + "grad_norm": 2.401589870452881, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7019336819648743, + "num_tokens": 160385993.0, + "step": 6235 + }, + { + "epoch": 0.6848231934987921, + "grad_norm": 2.0814006328582764, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.71575528383255, + "num_tokens": 160412081.0, + "step": 6236 + }, + { + "epoch": 0.6849330112014057, + "grad_norm": 2.310302734375, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7264140844345093, + "num_tokens": 160434326.0, + "step": 6237 + }, + { + "epoch": 0.6850428289040194, + "grad_norm": 2.207526922225952, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7272219657897949, + "num_tokens": 160458564.0, + "step": 6238 + }, + { + "epoch": 0.685152646606633, + "grad_norm": 2.470383882522583, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7101803421974182, + "num_tokens": 160480616.0, + "step": 6239 + }, + { + "epoch": 0.6852624643092466, + "grad_norm": 2.24908709526062, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7102275490760803, + "num_tokens": 160505321.0, + "step": 6240 + }, + { + "epoch": 0.6853722820118603, + "grad_norm": 2.0428690910339355, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7113171219825745, + "num_tokens": 160532921.0, + "step": 6241 + }, + { + "epoch": 0.6854820997144739, + "grad_norm": 2.1452770233154297, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7068668603897095, + "num_tokens": 160558665.0, + "step": 6242 + }, + { + "epoch": 0.6855919174170876, + "grad_norm": 1.9481748342514038, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6939783096313477, + "num_tokens": 160590062.0, + "step": 6243 + }, + { + "epoch": 0.6857017351197013, + "grad_norm": 2.190315008163452, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7272285223007202, + "num_tokens": 160614067.0, + "step": 6244 + }, + { + "epoch": 0.685811552822315, + "grad_norm": 2.1511480808258057, + "learning_rate": 1e-06, + "loss": 1.0616, + "mean_token_accuracy": 0.6781295537948608, + "num_tokens": 160640829.0, + "step": 6245 + }, + { + "epoch": 0.6859213705249286, + "grad_norm": 2.065936803817749, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7083849906921387, + "num_tokens": 160670868.0, + "step": 6246 + }, + { + "epoch": 0.6860311882275423, + "grad_norm": 2.3595237731933594, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7183282375335693, + "num_tokens": 160693399.0, + "step": 6247 + }, + { + "epoch": 0.6861410059301559, + "grad_norm": 2.1276559829711914, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7332919836044312, + "num_tokens": 160717774.0, + "step": 6248 + }, + { + "epoch": 0.6862508236327696, + "grad_norm": 2.044501781463623, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7042126655578613, + "num_tokens": 160745096.0, + "step": 6249 + }, + { + "epoch": 0.6863606413353832, + "grad_norm": 2.187438726425171, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7122173309326172, + "num_tokens": 160769476.0, + "step": 6250 + }, + { + "epoch": 0.686470459037997, + "grad_norm": 1.9594955444335938, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6789220571517944, + "num_tokens": 160803551.0, + "step": 6251 + }, + { + "epoch": 0.6865802767406106, + "grad_norm": 2.0512425899505615, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6968177556991577, + "num_tokens": 160832856.0, + "step": 6252 + }, + { + "epoch": 0.6866900944432243, + "grad_norm": 2.2296390533447266, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7016632556915283, + "num_tokens": 160857224.0, + "step": 6253 + }, + { + "epoch": 0.6867999121458379, + "grad_norm": 2.220764398574829, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6784679889678955, + "num_tokens": 160884183.0, + "step": 6254 + }, + { + "epoch": 0.6869097298484516, + "grad_norm": 2.2333340644836426, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6986794471740723, + "num_tokens": 160911007.0, + "step": 6255 + }, + { + "epoch": 0.6870195475510652, + "grad_norm": 2.2918267250061035, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7223248481750488, + "num_tokens": 160933465.0, + "step": 6256 + }, + { + "epoch": 0.6871293652536788, + "grad_norm": 2.086498975753784, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6924925446510315, + "num_tokens": 160961813.0, + "step": 6257 + }, + { + "epoch": 0.6872391829562926, + "grad_norm": 2.005094289779663, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6844210624694824, + "num_tokens": 160991471.0, + "step": 6258 + }, + { + "epoch": 0.6873490006589063, + "grad_norm": 1.9675730466842651, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6963239908218384, + "num_tokens": 161022196.0, + "step": 6259 + }, + { + "epoch": 0.6874588183615199, + "grad_norm": 2.1030807495117188, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7313352823257446, + "num_tokens": 161048163.0, + "step": 6260 + }, + { + "epoch": 0.6875686360641335, + "grad_norm": 2.0928494930267334, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6926619410514832, + "num_tokens": 161076648.0, + "step": 6261 + }, + { + "epoch": 0.6876784537667472, + "grad_norm": 2.3563623428344727, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7133275866508484, + "num_tokens": 161099345.0, + "step": 6262 + }, + { + "epoch": 0.6877882714693608, + "grad_norm": 2.1135973930358887, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7142922878265381, + "num_tokens": 161125471.0, + "step": 6263 + }, + { + "epoch": 0.6878980891719745, + "grad_norm": 2.821507215499878, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7382046580314636, + "num_tokens": 161142393.0, + "step": 6264 + }, + { + "epoch": 0.6880079068745882, + "grad_norm": 2.4387543201446533, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7089054584503174, + "num_tokens": 161163203.0, + "step": 6265 + }, + { + "epoch": 0.6881177245772019, + "grad_norm": 1.95541512966156, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6923842430114746, + "num_tokens": 161194979.0, + "step": 6266 + }, + { + "epoch": 0.6882275422798155, + "grad_norm": 2.1571669578552246, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7233899831771851, + "num_tokens": 161219714.0, + "step": 6267 + }, + { + "epoch": 0.6883373599824292, + "grad_norm": 2.2259368896484375, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.709315299987793, + "num_tokens": 161245437.0, + "step": 6268 + }, + { + "epoch": 0.6884471776850428, + "grad_norm": 2.0941433906555176, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.705496072769165, + "num_tokens": 161272995.0, + "step": 6269 + }, + { + "epoch": 0.6885569953876565, + "grad_norm": 2.198622703552246, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.6804778575897217, + "num_tokens": 161299132.0, + "step": 6270 + }, + { + "epoch": 0.6886668130902701, + "grad_norm": 2.329561710357666, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7162382006645203, + "num_tokens": 161321452.0, + "step": 6271 + }, + { + "epoch": 0.6887766307928838, + "grad_norm": 2.532033920288086, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7229304313659668, + "num_tokens": 161344191.0, + "step": 6272 + }, + { + "epoch": 0.6888864484954975, + "grad_norm": 1.8705685138702393, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6956095695495605, + "num_tokens": 161377295.0, + "step": 6273 + }, + { + "epoch": 0.6889962661981112, + "grad_norm": 2.2508747577667236, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7012661695480347, + "num_tokens": 161401781.0, + "step": 6274 + }, + { + "epoch": 0.6891060839007248, + "grad_norm": 2.1793429851531982, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7250194549560547, + "num_tokens": 161427804.0, + "step": 6275 + }, + { + "epoch": 0.6892159016033385, + "grad_norm": 1.9497528076171875, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7187492847442627, + "num_tokens": 161457135.0, + "step": 6276 + }, + { + "epoch": 0.6893257193059521, + "grad_norm": 2.07939076423645, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6937403678894043, + "num_tokens": 161485539.0, + "step": 6277 + }, + { + "epoch": 0.6894355370085657, + "grad_norm": 2.032985210418701, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7057435512542725, + "num_tokens": 161514955.0, + "step": 6278 + }, + { + "epoch": 0.6895453547111794, + "grad_norm": 2.0421061515808105, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6996124982833862, + "num_tokens": 161545684.0, + "step": 6279 + }, + { + "epoch": 0.6896551724137931, + "grad_norm": 2.2968926429748535, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7166221737861633, + "num_tokens": 161567736.0, + "step": 6280 + }, + { + "epoch": 0.6897649901164068, + "grad_norm": 2.453948736190796, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7030280828475952, + "num_tokens": 161589953.0, + "step": 6281 + }, + { + "epoch": 0.6898748078190204, + "grad_norm": 2.110522985458374, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7051382064819336, + "num_tokens": 161615126.0, + "step": 6282 + }, + { + "epoch": 0.6899846255216341, + "grad_norm": 1.781166911125183, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6830741763114929, + "num_tokens": 161654366.0, + "step": 6283 + }, + { + "epoch": 0.6900944432242477, + "grad_norm": 2.43534779548645, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7310119867324829, + "num_tokens": 161675311.0, + "step": 6284 + }, + { + "epoch": 0.6902042609268614, + "grad_norm": 2.532348394393921, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.723978579044342, + "num_tokens": 161695599.0, + "step": 6285 + }, + { + "epoch": 0.690314078629475, + "grad_norm": 2.1678574085235596, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7098270654678345, + "num_tokens": 161719566.0, + "step": 6286 + }, + { + "epoch": 0.6904238963320888, + "grad_norm": 2.0886027812957764, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7238534092903137, + "num_tokens": 161747249.0, + "step": 6287 + }, + { + "epoch": 0.6905337140347024, + "grad_norm": 2.27258563041687, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7097768783569336, + "num_tokens": 161769125.0, + "step": 6288 + }, + { + "epoch": 0.6906435317373161, + "grad_norm": 2.3175511360168457, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7128255367279053, + "num_tokens": 161793725.0, + "step": 6289 + }, + { + "epoch": 0.6907533494399297, + "grad_norm": 1.8401474952697754, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7132691740989685, + "num_tokens": 161826051.0, + "step": 6290 + }, + { + "epoch": 0.6908631671425434, + "grad_norm": 2.168544054031372, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7050535678863525, + "num_tokens": 161854652.0, + "step": 6291 + }, + { + "epoch": 0.690972984845157, + "grad_norm": 2.0065293312072754, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6861203908920288, + "num_tokens": 161886144.0, + "step": 6292 + }, + { + "epoch": 0.6910828025477707, + "grad_norm": 2.292539358139038, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7295776605606079, + "num_tokens": 161909414.0, + "step": 6293 + }, + { + "epoch": 0.6911926202503844, + "grad_norm": 2.222041368484497, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7064422369003296, + "num_tokens": 161935299.0, + "step": 6294 + }, + { + "epoch": 0.6913024379529981, + "grad_norm": 2.3615176677703857, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.719355046749115, + "num_tokens": 161956198.0, + "step": 6295 + }, + { + "epoch": 0.6914122556556117, + "grad_norm": 2.2480976581573486, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7097325325012207, + "num_tokens": 161982053.0, + "step": 6296 + }, + { + "epoch": 0.6915220733582254, + "grad_norm": 2.2412819862365723, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6997079253196716, + "num_tokens": 162006866.0, + "step": 6297 + }, + { + "epoch": 0.691631891060839, + "grad_norm": 2.236876964569092, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.7077115774154663, + "num_tokens": 162031361.0, + "step": 6298 + }, + { + "epoch": 0.6917417087634526, + "grad_norm": 2.016470193862915, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7360806465148926, + "num_tokens": 162059321.0, + "step": 6299 + }, + { + "epoch": 0.6918515264660663, + "grad_norm": 2.2102386951446533, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6984803080558777, + "num_tokens": 162085151.0, + "step": 6300 + }, + { + "epoch": 0.6919613441686799, + "grad_norm": 1.7268786430358887, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7143768668174744, + "num_tokens": 162123927.0, + "step": 6301 + }, + { + "epoch": 0.6920711618712937, + "grad_norm": 2.5283167362213135, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7337936162948608, + "num_tokens": 162144458.0, + "step": 6302 + }, + { + "epoch": 0.6921809795739073, + "grad_norm": 2.252702474594116, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.718701958656311, + "num_tokens": 162169069.0, + "step": 6303 + }, + { + "epoch": 0.692290797276521, + "grad_norm": 1.8172719478607178, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.692460834980011, + "num_tokens": 162203114.0, + "step": 6304 + }, + { + "epoch": 0.6924006149791346, + "grad_norm": 1.9872101545333862, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7219545245170593, + "num_tokens": 162231602.0, + "step": 6305 + }, + { + "epoch": 0.6925104326817483, + "grad_norm": 2.19507098197937, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7211565971374512, + "num_tokens": 162257388.0, + "step": 6306 + }, + { + "epoch": 0.6926202503843619, + "grad_norm": 2.113704204559326, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7136870622634888, + "num_tokens": 162286803.0, + "step": 6307 + }, + { + "epoch": 0.6927300680869756, + "grad_norm": 2.207643747329712, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7029333710670471, + "num_tokens": 162312574.0, + "step": 6308 + }, + { + "epoch": 0.6928398857895893, + "grad_norm": 2.2786781787872314, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7011121511459351, + "num_tokens": 162341090.0, + "step": 6309 + }, + { + "epoch": 0.692949703492203, + "grad_norm": 2.4517035484313965, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6834477782249451, + "num_tokens": 162364959.0, + "step": 6310 + }, + { + "epoch": 0.6930595211948166, + "grad_norm": 2.556419849395752, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7244191765785217, + "num_tokens": 162383817.0, + "step": 6311 + }, + { + "epoch": 0.6931693388974303, + "grad_norm": 2.2003331184387207, + "learning_rate": 1e-06, + "loss": 1.096, + "mean_token_accuracy": 0.6827754974365234, + "num_tokens": 162411370.0, + "step": 6312 + }, + { + "epoch": 0.6932791566000439, + "grad_norm": 2.258009910583496, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7058733701705933, + "num_tokens": 162435080.0, + "step": 6313 + }, + { + "epoch": 0.6933889743026576, + "grad_norm": 2.2636656761169434, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7055361270904541, + "num_tokens": 162460060.0, + "step": 6314 + }, + { + "epoch": 0.6934987920052712, + "grad_norm": 2.1721084117889404, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6973435878753662, + "num_tokens": 162485155.0, + "step": 6315 + }, + { + "epoch": 0.693608609707885, + "grad_norm": 2.215186357498169, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.7000216245651245, + "num_tokens": 162511508.0, + "step": 6316 + }, + { + "epoch": 0.6937184274104986, + "grad_norm": 2.2294952869415283, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6875253915786743, + "num_tokens": 162535812.0, + "step": 6317 + }, + { + "epoch": 0.6938282451131123, + "grad_norm": 2.0993685722351074, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.7008727192878723, + "num_tokens": 162562211.0, + "step": 6318 + }, + { + "epoch": 0.6939380628157259, + "grad_norm": 2.0347421169281006, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6929657459259033, + "num_tokens": 162592985.0, + "step": 6319 + }, + { + "epoch": 0.6940478805183395, + "grad_norm": 2.381246566772461, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6994303464889526, + "num_tokens": 162616306.0, + "step": 6320 + }, + { + "epoch": 0.6941576982209532, + "grad_norm": 2.1001076698303223, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7174271941184998, + "num_tokens": 162643933.0, + "step": 6321 + }, + { + "epoch": 0.6942675159235668, + "grad_norm": 2.1977431774139404, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.706209659576416, + "num_tokens": 162669530.0, + "step": 6322 + }, + { + "epoch": 0.6943773336261806, + "grad_norm": 2.1979823112487793, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7309901118278503, + "num_tokens": 162692720.0, + "step": 6323 + }, + { + "epoch": 0.6944871513287942, + "grad_norm": 2.4240784645080566, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7152664661407471, + "num_tokens": 162714897.0, + "step": 6324 + }, + { + "epoch": 0.6945969690314079, + "grad_norm": 2.12471604347229, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7409490942955017, + "num_tokens": 162741091.0, + "step": 6325 + }, + { + "epoch": 0.6947067867340215, + "grad_norm": 2.330083131790161, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7104870080947876, + "num_tokens": 162764895.0, + "step": 6326 + }, + { + "epoch": 0.6948166044366352, + "grad_norm": 2.1825854778289795, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7018975019454956, + "num_tokens": 162791063.0, + "step": 6327 + }, + { + "epoch": 0.6949264221392488, + "grad_norm": 2.2901976108551025, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.711487889289856, + "num_tokens": 162815375.0, + "step": 6328 + }, + { + "epoch": 0.6950362398418625, + "grad_norm": 2.1342661380767822, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6953401565551758, + "num_tokens": 162842398.0, + "step": 6329 + }, + { + "epoch": 0.6951460575444761, + "grad_norm": 2.0278844833374023, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6969920992851257, + "num_tokens": 162871246.0, + "step": 6330 + }, + { + "epoch": 0.6952558752470899, + "grad_norm": 2.0477187633514404, + "learning_rate": 1e-06, + "loss": 1.1318, + "mean_token_accuracy": 0.6665147542953491, + "num_tokens": 162902616.0, + "step": 6331 + }, + { + "epoch": 0.6953656929497035, + "grad_norm": 2.202871799468994, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7062512636184692, + "num_tokens": 162927856.0, + "step": 6332 + }, + { + "epoch": 0.6954755106523172, + "grad_norm": 2.0784335136413574, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6885238289833069, + "num_tokens": 162956691.0, + "step": 6333 + }, + { + "epoch": 0.6955853283549308, + "grad_norm": 2.122739553451538, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7149085402488708, + "num_tokens": 162984316.0, + "step": 6334 + }, + { + "epoch": 0.6956951460575445, + "grad_norm": 2.2251029014587402, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6903958320617676, + "num_tokens": 163011262.0, + "step": 6335 + }, + { + "epoch": 0.6958049637601581, + "grad_norm": 2.179051399230957, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7170232534408569, + "num_tokens": 163035164.0, + "step": 6336 + }, + { + "epoch": 0.6959147814627717, + "grad_norm": 2.1440868377685547, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6860972046852112, + "num_tokens": 163063549.0, + "step": 6337 + }, + { + "epoch": 0.6960245991653855, + "grad_norm": 2.3547375202178955, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.725750744342804, + "num_tokens": 163085563.0, + "step": 6338 + }, + { + "epoch": 0.6961344168679992, + "grad_norm": 2.2662229537963867, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7304988503456116, + "num_tokens": 163106810.0, + "step": 6339 + }, + { + "epoch": 0.6962442345706128, + "grad_norm": 2.110623359680176, + "learning_rate": 1e-06, + "loss": 1.0699, + "mean_token_accuracy": 0.6769059896469116, + "num_tokens": 163135994.0, + "step": 6340 + }, + { + "epoch": 0.6963540522732264, + "grad_norm": 2.011709451675415, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7206412553787231, + "num_tokens": 163165289.0, + "step": 6341 + }, + { + "epoch": 0.6964638699758401, + "grad_norm": 2.17276668548584, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6906126737594604, + "num_tokens": 163190732.0, + "step": 6342 + }, + { + "epoch": 0.6965736876784537, + "grad_norm": 2.179593801498413, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.710249662399292, + "num_tokens": 163216745.0, + "step": 6343 + }, + { + "epoch": 0.6966835053810674, + "grad_norm": 2.514475107192993, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7157841920852661, + "num_tokens": 163236742.0, + "step": 6344 + }, + { + "epoch": 0.6967933230836811, + "grad_norm": 2.1294100284576416, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6973034143447876, + "num_tokens": 163263465.0, + "step": 6345 + }, + { + "epoch": 0.6969031407862948, + "grad_norm": 2.4600772857666016, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7133461236953735, + "num_tokens": 163284360.0, + "step": 6346 + }, + { + "epoch": 0.6970129584889084, + "grad_norm": 2.2377195358276367, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6875853538513184, + "num_tokens": 163310068.0, + "step": 6347 + }, + { + "epoch": 0.6971227761915221, + "grad_norm": 2.248763084411621, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7066726684570312, + "num_tokens": 163335743.0, + "step": 6348 + }, + { + "epoch": 0.6972325938941357, + "grad_norm": 2.2422807216644287, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7275004386901855, + "num_tokens": 163361033.0, + "step": 6349 + }, + { + "epoch": 0.6973424115967494, + "grad_norm": 2.4144160747528076, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.6984779834747314, + "num_tokens": 163382875.0, + "step": 6350 + }, + { + "epoch": 0.697452229299363, + "grad_norm": 2.2495012283325195, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7033959627151489, + "num_tokens": 163407183.0, + "step": 6351 + }, + { + "epoch": 0.6975620470019768, + "grad_norm": 2.114680290222168, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6926042437553406, + "num_tokens": 163435438.0, + "step": 6352 + }, + { + "epoch": 0.6976718647045904, + "grad_norm": 2.183523654937744, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7282350063323975, + "num_tokens": 163460860.0, + "step": 6353 + }, + { + "epoch": 0.6977816824072041, + "grad_norm": 2.4937567710876465, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7089468240737915, + "num_tokens": 163480747.0, + "step": 6354 + }, + { + "epoch": 0.6978915001098177, + "grad_norm": 2.211284637451172, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7023216485977173, + "num_tokens": 163507954.0, + "step": 6355 + }, + { + "epoch": 0.6980013178124314, + "grad_norm": 2.315185785293579, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7186702489852905, + "num_tokens": 163530729.0, + "step": 6356 + }, + { + "epoch": 0.698111135515045, + "grad_norm": 2.2273335456848145, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.718710720539093, + "num_tokens": 163555679.0, + "step": 6357 + }, + { + "epoch": 0.6982209532176586, + "grad_norm": 2.157496213912964, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7033954858779907, + "num_tokens": 163583680.0, + "step": 6358 + }, + { + "epoch": 0.6983307709202724, + "grad_norm": 2.2540533542633057, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.697542130947113, + "num_tokens": 163609430.0, + "step": 6359 + }, + { + "epoch": 0.698440588622886, + "grad_norm": 2.2487664222717285, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7039930820465088, + "num_tokens": 163634321.0, + "step": 6360 + }, + { + "epoch": 0.6985504063254997, + "grad_norm": 2.403991222381592, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7190216779708862, + "num_tokens": 163656964.0, + "step": 6361 + }, + { + "epoch": 0.6986602240281133, + "grad_norm": 2.0915660858154297, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7084689140319824, + "num_tokens": 163686113.0, + "step": 6362 + }, + { + "epoch": 0.698770041730727, + "grad_norm": 2.2348177433013916, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7205828428268433, + "num_tokens": 163711285.0, + "step": 6363 + }, + { + "epoch": 0.6988798594333406, + "grad_norm": 2.147430658340454, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6947753429412842, + "num_tokens": 163738676.0, + "step": 6364 + }, + { + "epoch": 0.6989896771359543, + "grad_norm": 2.1271815299987793, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7087427973747253, + "num_tokens": 163763164.0, + "step": 6365 + }, + { + "epoch": 0.6990994948385679, + "grad_norm": 2.496589422225952, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7178827524185181, + "num_tokens": 163784363.0, + "step": 6366 + }, + { + "epoch": 0.6992093125411817, + "grad_norm": 2.3584980964660645, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.7004649639129639, + "num_tokens": 163808087.0, + "step": 6367 + }, + { + "epoch": 0.6993191302437953, + "grad_norm": 2.1092326641082764, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.716675877571106, + "num_tokens": 163834468.0, + "step": 6368 + }, + { + "epoch": 0.699428947946409, + "grad_norm": 2.359210968017578, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7141109704971313, + "num_tokens": 163861570.0, + "step": 6369 + }, + { + "epoch": 0.6995387656490226, + "grad_norm": 2.1486289501190186, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7102314829826355, + "num_tokens": 163889541.0, + "step": 6370 + }, + { + "epoch": 0.6996485833516363, + "grad_norm": 1.9674245119094849, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6814042925834656, + "num_tokens": 163921558.0, + "step": 6371 + }, + { + "epoch": 0.6997584010542499, + "grad_norm": 2.265287160873413, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7208421230316162, + "num_tokens": 163945396.0, + "step": 6372 + }, + { + "epoch": 0.6998682187568636, + "grad_norm": 2.6432852745056152, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7269527316093445, + "num_tokens": 163963826.0, + "step": 6373 + }, + { + "epoch": 0.6999780364594773, + "grad_norm": 2.1580886840820312, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7129214406013489, + "num_tokens": 163988514.0, + "step": 6374 + }, + { + "epoch": 0.700087854162091, + "grad_norm": 2.0545454025268555, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.7010043859481812, + "num_tokens": 164015238.0, + "step": 6375 + }, + { + "epoch": 0.7001976718647046, + "grad_norm": 2.3445804119110107, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7070509791374207, + "num_tokens": 164036861.0, + "step": 6376 + }, + { + "epoch": 0.7003074895673183, + "grad_norm": 2.7176599502563477, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7118455767631531, + "num_tokens": 164054861.0, + "step": 6377 + }, + { + "epoch": 0.7004173072699319, + "grad_norm": 2.2286763191223145, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7134314775466919, + "num_tokens": 164078296.0, + "step": 6378 + }, + { + "epoch": 0.7005271249725455, + "grad_norm": 1.997829794883728, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7076084613800049, + "num_tokens": 164111259.0, + "step": 6379 + }, + { + "epoch": 0.7006369426751592, + "grad_norm": 2.18233060836792, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6895923614501953, + "num_tokens": 164138216.0, + "step": 6380 + }, + { + "epoch": 0.700746760377773, + "grad_norm": 2.0110175609588623, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7406587600708008, + "num_tokens": 164169226.0, + "step": 6381 + }, + { + "epoch": 0.7008565780803866, + "grad_norm": 2.2225019931793213, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7127388715744019, + "num_tokens": 164194396.0, + "step": 6382 + }, + { + "epoch": 0.7009663957830002, + "grad_norm": 1.9984183311462402, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6938505172729492, + "num_tokens": 164223971.0, + "step": 6383 + }, + { + "epoch": 0.7010762134856139, + "grad_norm": 2.1364662647247314, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7014742493629456, + "num_tokens": 164251692.0, + "step": 6384 + }, + { + "epoch": 0.7011860311882275, + "grad_norm": 2.2659037113189697, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6931600570678711, + "num_tokens": 164279594.0, + "step": 6385 + }, + { + "epoch": 0.7012958488908412, + "grad_norm": 2.001995086669922, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.6994707584381104, + "num_tokens": 164308113.0, + "step": 6386 + }, + { + "epoch": 0.7014056665934548, + "grad_norm": 2.115643262863159, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7116979956626892, + "num_tokens": 164333377.0, + "step": 6387 + }, + { + "epoch": 0.7015154842960686, + "grad_norm": 2.1335055828094482, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7051830887794495, + "num_tokens": 164358939.0, + "step": 6388 + }, + { + "epoch": 0.7016253019986822, + "grad_norm": 2.45466685295105, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7144166231155396, + "num_tokens": 164380710.0, + "step": 6389 + }, + { + "epoch": 0.7017351197012959, + "grad_norm": 2.319438934326172, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7387082576751709, + "num_tokens": 164403365.0, + "step": 6390 + }, + { + "epoch": 0.7018449374039095, + "grad_norm": 2.0019478797912598, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7245700359344482, + "num_tokens": 164432311.0, + "step": 6391 + }, + { + "epoch": 0.7019547551065232, + "grad_norm": 1.9900587797164917, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6946579217910767, + "num_tokens": 164463242.0, + "step": 6392 + }, + { + "epoch": 0.7020645728091368, + "grad_norm": 2.2480876445770264, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7092826962471008, + "num_tokens": 164489572.0, + "step": 6393 + }, + { + "epoch": 0.7021743905117505, + "grad_norm": 2.134891986846924, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6990931034088135, + "num_tokens": 164517538.0, + "step": 6394 + }, + { + "epoch": 0.7022842082143641, + "grad_norm": 2.072448253631592, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7026031613349915, + "num_tokens": 164544824.0, + "step": 6395 + }, + { + "epoch": 0.7023940259169779, + "grad_norm": 2.1928658485412598, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.694394052028656, + "num_tokens": 164571722.0, + "step": 6396 + }, + { + "epoch": 0.7025038436195915, + "grad_norm": 2.056762456893921, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7090418338775635, + "num_tokens": 164598407.0, + "step": 6397 + }, + { + "epoch": 0.7026136613222052, + "grad_norm": 1.9211677312850952, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6998303532600403, + "num_tokens": 164632423.0, + "step": 6398 + }, + { + "epoch": 0.7027234790248188, + "grad_norm": 2.2942230701446533, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.708478569984436, + "num_tokens": 164655952.0, + "step": 6399 + }, + { + "epoch": 0.7028332967274324, + "grad_norm": 1.9888670444488525, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.707825779914856, + "num_tokens": 164684274.0, + "step": 6400 + }, + { + "epoch": 0.7029431144300461, + "grad_norm": 2.349822998046875, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6895830631256104, + "num_tokens": 164707346.0, + "step": 6401 + }, + { + "epoch": 0.7030529321326597, + "grad_norm": 2.102801561355591, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7075929045677185, + "num_tokens": 164734846.0, + "step": 6402 + }, + { + "epoch": 0.7031627498352735, + "grad_norm": 2.7007765769958496, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7257736325263977, + "num_tokens": 164752387.0, + "step": 6403 + }, + { + "epoch": 0.7032725675378871, + "grad_norm": 2.2583935260772705, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6951626539230347, + "num_tokens": 164778104.0, + "step": 6404 + }, + { + "epoch": 0.7033823852405008, + "grad_norm": 2.1705427169799805, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7157427668571472, + "num_tokens": 164803918.0, + "step": 6405 + }, + { + "epoch": 0.7034922029431144, + "grad_norm": 2.5111701488494873, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7091663479804993, + "num_tokens": 164822842.0, + "step": 6406 + }, + { + "epoch": 0.7036020206457281, + "grad_norm": 2.2446343898773193, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7053136229515076, + "num_tokens": 164846654.0, + "step": 6407 + }, + { + "epoch": 0.7037118383483417, + "grad_norm": 2.3658976554870605, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7140918970108032, + "num_tokens": 164868204.0, + "step": 6408 + }, + { + "epoch": 0.7038216560509554, + "grad_norm": 2.162504196166992, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6907804012298584, + "num_tokens": 164895050.0, + "step": 6409 + }, + { + "epoch": 0.7039314737535691, + "grad_norm": 2.113278388977051, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.703888475894928, + "num_tokens": 164921362.0, + "step": 6410 + }, + { + "epoch": 0.7040412914561828, + "grad_norm": 2.3967387676239014, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7119240164756775, + "num_tokens": 164944057.0, + "step": 6411 + }, + { + "epoch": 0.7041511091587964, + "grad_norm": 2.1132166385650635, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7209542393684387, + "num_tokens": 164971092.0, + "step": 6412 + }, + { + "epoch": 0.7042609268614101, + "grad_norm": 2.272881269454956, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7356725335121155, + "num_tokens": 164993554.0, + "step": 6413 + }, + { + "epoch": 0.7043707445640237, + "grad_norm": 2.137620687484741, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7184720039367676, + "num_tokens": 165018783.0, + "step": 6414 + }, + { + "epoch": 0.7044805622666374, + "grad_norm": 2.163754463195801, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7199520468711853, + "num_tokens": 165045157.0, + "step": 6415 + }, + { + "epoch": 0.704590379969251, + "grad_norm": 2.1531822681427, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.696346640586853, + "num_tokens": 165071469.0, + "step": 6416 + }, + { + "epoch": 0.7047001976718648, + "grad_norm": 1.8282623291015625, + "learning_rate": 1e-06, + "loss": 1.1118, + "mean_token_accuracy": 0.6714446544647217, + "num_tokens": 165109050.0, + "step": 6417 + }, + { + "epoch": 0.7048100153744784, + "grad_norm": 2.3068161010742188, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7050174474716187, + "num_tokens": 165132957.0, + "step": 6418 + }, + { + "epoch": 0.704919833077092, + "grad_norm": 2.3358314037323, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7151521444320679, + "num_tokens": 165153945.0, + "step": 6419 + }, + { + "epoch": 0.7050296507797057, + "grad_norm": 2.465348243713379, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6876399517059326, + "num_tokens": 165177236.0, + "step": 6420 + }, + { + "epoch": 0.7051394684823193, + "grad_norm": 2.243961811065674, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7011231184005737, + "num_tokens": 165202631.0, + "step": 6421 + }, + { + "epoch": 0.705249286184933, + "grad_norm": 2.0063936710357666, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7192718982696533, + "num_tokens": 165232366.0, + "step": 6422 + }, + { + "epoch": 0.7053591038875466, + "grad_norm": 2.1807405948638916, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7122218608856201, + "num_tokens": 165259413.0, + "step": 6423 + }, + { + "epoch": 0.7054689215901603, + "grad_norm": 2.445683479309082, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.717703104019165, + "num_tokens": 165281957.0, + "step": 6424 + }, + { + "epoch": 0.705578739292774, + "grad_norm": 2.488145589828491, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6977734565734863, + "num_tokens": 165302698.0, + "step": 6425 + }, + { + "epoch": 0.7056885569953877, + "grad_norm": 2.3395259380340576, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7209464311599731, + "num_tokens": 165323867.0, + "step": 6426 + }, + { + "epoch": 0.7057983746980013, + "grad_norm": 2.054136037826538, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6801749467849731, + "num_tokens": 165353900.0, + "step": 6427 + }, + { + "epoch": 0.705908192400615, + "grad_norm": 2.074620246887207, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7042726278305054, + "num_tokens": 165383236.0, + "step": 6428 + }, + { + "epoch": 0.7060180101032286, + "grad_norm": 2.2334134578704834, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.702688455581665, + "num_tokens": 165406892.0, + "step": 6429 + }, + { + "epoch": 0.7061278278058423, + "grad_norm": 2.1234664916992188, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7327880859375, + "num_tokens": 165432573.0, + "step": 6430 + }, + { + "epoch": 0.7062376455084559, + "grad_norm": 2.036574363708496, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7266427874565125, + "num_tokens": 165459277.0, + "step": 6431 + }, + { + "epoch": 0.7063474632110697, + "grad_norm": 2.1240739822387695, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7064460515975952, + "num_tokens": 165487841.0, + "step": 6432 + }, + { + "epoch": 0.7064572809136833, + "grad_norm": 2.2083654403686523, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6986932158470154, + "num_tokens": 165512363.0, + "step": 6433 + }, + { + "epoch": 0.706567098616297, + "grad_norm": 2.077528715133667, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.7056235671043396, + "num_tokens": 165540084.0, + "step": 6434 + }, + { + "epoch": 0.7066769163189106, + "grad_norm": 2.2948555946350098, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7051858901977539, + "num_tokens": 165562661.0, + "step": 6435 + }, + { + "epoch": 0.7067867340215243, + "grad_norm": 2.017855644226074, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6862791776657104, + "num_tokens": 165595377.0, + "step": 6436 + }, + { + "epoch": 0.7068965517241379, + "grad_norm": 2.053154230117798, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6974048614501953, + "num_tokens": 165625893.0, + "step": 6437 + }, + { + "epoch": 0.7070063694267515, + "grad_norm": 2.409951686859131, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6805899739265442, + "num_tokens": 165650967.0, + "step": 6438 + }, + { + "epoch": 0.7071161871293653, + "grad_norm": 2.2298948764801025, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6904799938201904, + "num_tokens": 165677021.0, + "step": 6439 + }, + { + "epoch": 0.707226004831979, + "grad_norm": 1.955197811126709, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7077512741088867, + "num_tokens": 165707730.0, + "step": 6440 + }, + { + "epoch": 0.7073358225345926, + "grad_norm": 2.2125401496887207, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7352684736251831, + "num_tokens": 165729364.0, + "step": 6441 + }, + { + "epoch": 0.7074456402372062, + "grad_norm": 2.097196102142334, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.713377058506012, + "num_tokens": 165755371.0, + "step": 6442 + }, + { + "epoch": 0.7075554579398199, + "grad_norm": 2.191789388656616, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.7026034593582153, + "num_tokens": 165781804.0, + "step": 6443 + }, + { + "epoch": 0.7076652756424335, + "grad_norm": 2.1404051780700684, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6976311802864075, + "num_tokens": 165808231.0, + "step": 6444 + }, + { + "epoch": 0.7077750933450472, + "grad_norm": 2.3723959922790527, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6886813640594482, + "num_tokens": 165830282.0, + "step": 6445 + }, + { + "epoch": 0.7078849110476609, + "grad_norm": 2.255113124847412, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6987615823745728, + "num_tokens": 165854357.0, + "step": 6446 + }, + { + "epoch": 0.7079947287502746, + "grad_norm": 1.9213871955871582, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6820973753929138, + "num_tokens": 165888195.0, + "step": 6447 + }, + { + "epoch": 0.7081045464528882, + "grad_norm": 1.975366473197937, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6846251487731934, + "num_tokens": 165921299.0, + "step": 6448 + }, + { + "epoch": 0.7082143641555019, + "grad_norm": 2.309126853942871, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7375398278236389, + "num_tokens": 165943534.0, + "step": 6449 + }, + { + "epoch": 0.7083241818581155, + "grad_norm": 2.146160364151001, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.7084850668907166, + "num_tokens": 165967856.0, + "step": 6450 + }, + { + "epoch": 0.7084339995607292, + "grad_norm": 2.416379928588867, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7156097888946533, + "num_tokens": 165988319.0, + "step": 6451 + }, + { + "epoch": 0.7085438172633428, + "grad_norm": 2.0074045658111572, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6845513582229614, + "num_tokens": 166020078.0, + "step": 6452 + }, + { + "epoch": 0.7086536349659565, + "grad_norm": 2.371227264404297, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7037976384162903, + "num_tokens": 166044306.0, + "step": 6453 + }, + { + "epoch": 0.7087634526685702, + "grad_norm": 2.624516725540161, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7082897424697876, + "num_tokens": 166064552.0, + "step": 6454 + }, + { + "epoch": 0.7088732703711839, + "grad_norm": 2.658784866333008, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7221647500991821, + "num_tokens": 166082476.0, + "step": 6455 + }, + { + "epoch": 0.7089830880737975, + "grad_norm": 2.2048423290252686, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7016658782958984, + "num_tokens": 166107598.0, + "step": 6456 + }, + { + "epoch": 0.7090929057764112, + "grad_norm": 2.1821134090423584, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7223467826843262, + "num_tokens": 166132578.0, + "step": 6457 + }, + { + "epoch": 0.7092027234790248, + "grad_norm": 2.1052000522613525, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7132652997970581, + "num_tokens": 166160435.0, + "step": 6458 + }, + { + "epoch": 0.7093125411816384, + "grad_norm": 2.1401641368865967, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6992158889770508, + "num_tokens": 166187170.0, + "step": 6459 + }, + { + "epoch": 0.7094223588842521, + "grad_norm": 2.1027991771698, + "learning_rate": 1e-06, + "loss": 1.0879, + "mean_token_accuracy": 0.6676697731018066, + "num_tokens": 166215534.0, + "step": 6460 + }, + { + "epoch": 0.7095321765868658, + "grad_norm": 2.1575560569763184, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7081901431083679, + "num_tokens": 166239798.0, + "step": 6461 + }, + { + "epoch": 0.7096419942894795, + "grad_norm": 2.037647008895874, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6941971778869629, + "num_tokens": 166267173.0, + "step": 6462 + }, + { + "epoch": 0.7097518119920931, + "grad_norm": 2.405141592025757, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7058855295181274, + "num_tokens": 166290260.0, + "step": 6463 + }, + { + "epoch": 0.7098616296947068, + "grad_norm": 2.0207207202911377, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6951864957809448, + "num_tokens": 166320231.0, + "step": 6464 + }, + { + "epoch": 0.7099714473973204, + "grad_norm": 2.1531691551208496, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.671112596988678, + "num_tokens": 166345637.0, + "step": 6465 + }, + { + "epoch": 0.7100812650999341, + "grad_norm": 2.0837199687957764, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6908711194992065, + "num_tokens": 166373852.0, + "step": 6466 + }, + { + "epoch": 0.7101910828025477, + "grad_norm": 2.1912877559661865, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6950618028640747, + "num_tokens": 166400832.0, + "step": 6467 + }, + { + "epoch": 0.7103009005051615, + "grad_norm": 2.0631110668182373, + "learning_rate": 1e-06, + "loss": 1.0894, + "mean_token_accuracy": 0.6782239675521851, + "num_tokens": 166431674.0, + "step": 6468 + }, + { + "epoch": 0.7104107182077751, + "grad_norm": 2.021951198577881, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6893388032913208, + "num_tokens": 166463491.0, + "step": 6469 + }, + { + "epoch": 0.7105205359103888, + "grad_norm": 1.9027332067489624, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7014631032943726, + "num_tokens": 166497230.0, + "step": 6470 + }, + { + "epoch": 0.7106303536130024, + "grad_norm": 2.220313549041748, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7034617066383362, + "num_tokens": 166522486.0, + "step": 6471 + }, + { + "epoch": 0.7107401713156161, + "grad_norm": 2.4017162322998047, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.704318106174469, + "num_tokens": 166544195.0, + "step": 6472 + }, + { + "epoch": 0.7108499890182297, + "grad_norm": 1.9889097213745117, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7214933633804321, + "num_tokens": 166571210.0, + "step": 6473 + }, + { + "epoch": 0.7109598067208434, + "grad_norm": 2.0300557613372803, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6864153742790222, + "num_tokens": 166602211.0, + "step": 6474 + }, + { + "epoch": 0.7110696244234571, + "grad_norm": 2.2295236587524414, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.709685742855072, + "num_tokens": 166625446.0, + "step": 6475 + }, + { + "epoch": 0.7111794421260708, + "grad_norm": 2.0356035232543945, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7129634618759155, + "num_tokens": 166652801.0, + "step": 6476 + }, + { + "epoch": 0.7112892598286844, + "grad_norm": 2.036442756652832, + "learning_rate": 1e-06, + "loss": 1.0872, + "mean_token_accuracy": 0.6814465522766113, + "num_tokens": 166682979.0, + "step": 6477 + }, + { + "epoch": 0.711399077531298, + "grad_norm": 2.330531358718872, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.6997089385986328, + "num_tokens": 166707946.0, + "step": 6478 + }, + { + "epoch": 0.7115088952339117, + "grad_norm": 1.8590484857559204, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6878279447555542, + "num_tokens": 166741791.0, + "step": 6479 + }, + { + "epoch": 0.7116187129365253, + "grad_norm": 2.2321887016296387, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7101813554763794, + "num_tokens": 166768063.0, + "step": 6480 + }, + { + "epoch": 0.711728530639139, + "grad_norm": 1.951748251914978, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6954944729804993, + "num_tokens": 166798849.0, + "step": 6481 + }, + { + "epoch": 0.7118383483417526, + "grad_norm": 2.091843605041504, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7001965045928955, + "num_tokens": 166826178.0, + "step": 6482 + }, + { + "epoch": 0.7119481660443664, + "grad_norm": 2.355316638946533, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.705143928527832, + "num_tokens": 166849638.0, + "step": 6483 + }, + { + "epoch": 0.71205798374698, + "grad_norm": 2.304788589477539, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.699641227722168, + "num_tokens": 166873089.0, + "step": 6484 + }, + { + "epoch": 0.7121678014495937, + "grad_norm": 2.129537343978882, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6986239552497864, + "num_tokens": 166901597.0, + "step": 6485 + }, + { + "epoch": 0.7122776191522073, + "grad_norm": 2.0721192359924316, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7012706995010376, + "num_tokens": 166927540.0, + "step": 6486 + }, + { + "epoch": 0.712387436854821, + "grad_norm": 2.287052869796753, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7141835689544678, + "num_tokens": 166950937.0, + "step": 6487 + }, + { + "epoch": 0.7124972545574346, + "grad_norm": 2.364311695098877, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.715835690498352, + "num_tokens": 166973265.0, + "step": 6488 + }, + { + "epoch": 0.7126070722600483, + "grad_norm": 2.5708987712860107, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7260794043540955, + "num_tokens": 166992573.0, + "step": 6489 + }, + { + "epoch": 0.712716889962662, + "grad_norm": 2.167961597442627, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.708861768245697, + "num_tokens": 167016273.0, + "step": 6490 + }, + { + "epoch": 0.7128267076652757, + "grad_norm": 2.2149624824523926, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7184299230575562, + "num_tokens": 167043156.0, + "step": 6491 + }, + { + "epoch": 0.7129365253678893, + "grad_norm": 2.471467971801758, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7190980315208435, + "num_tokens": 167062859.0, + "step": 6492 + }, + { + "epoch": 0.713046343070503, + "grad_norm": 2.041299819946289, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.7035753726959229, + "num_tokens": 167094763.0, + "step": 6493 + }, + { + "epoch": 0.7131561607731166, + "grad_norm": 2.166200876235962, + "learning_rate": 1e-06, + "loss": 1.0892, + "mean_token_accuracy": 0.6749078035354614, + "num_tokens": 167122038.0, + "step": 6494 + }, + { + "epoch": 0.7132659784757303, + "grad_norm": 2.080841541290283, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6856112480163574, + "num_tokens": 167150929.0, + "step": 6495 + }, + { + "epoch": 0.7133757961783439, + "grad_norm": 2.033845901489258, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6917974948883057, + "num_tokens": 167181275.0, + "step": 6496 + }, + { + "epoch": 0.7134856138809577, + "grad_norm": 2.190201759338379, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7263906002044678, + "num_tokens": 167206872.0, + "step": 6497 + }, + { + "epoch": 0.7135954315835713, + "grad_norm": 2.3302435874938965, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7057369947433472, + "num_tokens": 167232066.0, + "step": 6498 + }, + { + "epoch": 0.713705249286185, + "grad_norm": 2.2861406803131104, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6994522213935852, + "num_tokens": 167256636.0, + "step": 6499 + }, + { + "epoch": 0.7138150669887986, + "grad_norm": 1.9244756698608398, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7079497575759888, + "num_tokens": 167286953.0, + "step": 6500 + }, + { + "epoch": 0.7139248846914122, + "grad_norm": 2.0434157848358154, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6982904672622681, + "num_tokens": 167316196.0, + "step": 6501 + }, + { + "epoch": 0.7140347023940259, + "grad_norm": 2.1518099308013916, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6974124908447266, + "num_tokens": 167342953.0, + "step": 6502 + }, + { + "epoch": 0.7141445200966395, + "grad_norm": 2.1675925254821777, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7098459601402283, + "num_tokens": 167367304.0, + "step": 6503 + }, + { + "epoch": 0.7142543377992533, + "grad_norm": 2.288714647293091, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7118576765060425, + "num_tokens": 167391219.0, + "step": 6504 + }, + { + "epoch": 0.7143641555018669, + "grad_norm": 1.8866010904312134, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7080654501914978, + "num_tokens": 167422620.0, + "step": 6505 + }, + { + "epoch": 0.7144739732044806, + "grad_norm": 2.4473705291748047, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.736994743347168, + "num_tokens": 167442916.0, + "step": 6506 + }, + { + "epoch": 0.7145837909070942, + "grad_norm": 2.1497609615325928, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7063143253326416, + "num_tokens": 167469067.0, + "step": 6507 + }, + { + "epoch": 0.7146936086097079, + "grad_norm": 2.0143141746520996, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.694467306137085, + "num_tokens": 167497603.0, + "step": 6508 + }, + { + "epoch": 0.7148034263123215, + "grad_norm": 2.1066527366638184, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7006900310516357, + "num_tokens": 167524946.0, + "step": 6509 + }, + { + "epoch": 0.7149132440149352, + "grad_norm": 1.8848662376403809, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7055736780166626, + "num_tokens": 167557966.0, + "step": 6510 + }, + { + "epoch": 0.7150230617175488, + "grad_norm": 2.3581953048706055, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7023429870605469, + "num_tokens": 167582053.0, + "step": 6511 + }, + { + "epoch": 0.7151328794201626, + "grad_norm": 2.202227830886841, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6795779466629028, + "num_tokens": 167611332.0, + "step": 6512 + }, + { + "epoch": 0.7152426971227762, + "grad_norm": 2.259445905685425, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6912134885787964, + "num_tokens": 167637796.0, + "step": 6513 + }, + { + "epoch": 0.7153525148253899, + "grad_norm": 2.244464635848999, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6803486943244934, + "num_tokens": 167665109.0, + "step": 6514 + }, + { + "epoch": 0.7154623325280035, + "grad_norm": 2.090195417404175, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6904553174972534, + "num_tokens": 167693439.0, + "step": 6515 + }, + { + "epoch": 0.7155721502306172, + "grad_norm": 2.2634451389312744, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7181537747383118, + "num_tokens": 167716069.0, + "step": 6516 + }, + { + "epoch": 0.7156819679332308, + "grad_norm": 2.083004951477051, + "learning_rate": 1e-06, + "loss": 1.1209, + "mean_token_accuracy": 0.6607096195220947, + "num_tokens": 167745272.0, + "step": 6517 + }, + { + "epoch": 0.7157917856358444, + "grad_norm": 2.2157223224639893, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7105717658996582, + "num_tokens": 167769418.0, + "step": 6518 + }, + { + "epoch": 0.7159016033384582, + "grad_norm": 2.3306820392608643, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7167778015136719, + "num_tokens": 167792058.0, + "step": 6519 + }, + { + "epoch": 0.7160114210410718, + "grad_norm": 2.347541332244873, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7204725742340088, + "num_tokens": 167814742.0, + "step": 6520 + }, + { + "epoch": 0.7161212387436855, + "grad_norm": 1.9109625816345215, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7237815856933594, + "num_tokens": 167845759.0, + "step": 6521 + }, + { + "epoch": 0.7162310564462991, + "grad_norm": 2.246541976928711, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7014383673667908, + "num_tokens": 167870928.0, + "step": 6522 + }, + { + "epoch": 0.7163408741489128, + "grad_norm": 2.015381336212158, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7182531952857971, + "num_tokens": 167898652.0, + "step": 6523 + }, + { + "epoch": 0.7164506918515264, + "grad_norm": 2.2107090950012207, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7096264958381653, + "num_tokens": 167924090.0, + "step": 6524 + }, + { + "epoch": 0.7165605095541401, + "grad_norm": 2.385399103164673, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7166354656219482, + "num_tokens": 167945471.0, + "step": 6525 + }, + { + "epoch": 0.7166703272567538, + "grad_norm": 2.5818309783935547, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7223821878433228, + "num_tokens": 167964594.0, + "step": 6526 + }, + { + "epoch": 0.7167801449593675, + "grad_norm": 2.164031744003296, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7009118795394897, + "num_tokens": 167988872.0, + "step": 6527 + }, + { + "epoch": 0.7168899626619811, + "grad_norm": 2.2948949337005615, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7074484825134277, + "num_tokens": 168011508.0, + "step": 6528 + }, + { + "epoch": 0.7169997803645948, + "grad_norm": 1.943640947341919, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6833370923995972, + "num_tokens": 168044185.0, + "step": 6529 + }, + { + "epoch": 0.7171095980672084, + "grad_norm": 2.049549102783203, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6919798851013184, + "num_tokens": 168072199.0, + "step": 6530 + }, + { + "epoch": 0.7172194157698221, + "grad_norm": 2.2214906215667725, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7120514512062073, + "num_tokens": 168095864.0, + "step": 6531 + }, + { + "epoch": 0.7173292334724357, + "grad_norm": 2.256105661392212, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.69154292345047, + "num_tokens": 168121098.0, + "step": 6532 + }, + { + "epoch": 0.7174390511750495, + "grad_norm": 2.4247875213623047, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7140123248100281, + "num_tokens": 168144265.0, + "step": 6533 + }, + { + "epoch": 0.7175488688776631, + "grad_norm": 2.284515619277954, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7043852806091309, + "num_tokens": 168168655.0, + "step": 6534 + }, + { + "epoch": 0.7176586865802768, + "grad_norm": 2.203437328338623, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7396299839019775, + "num_tokens": 168193074.0, + "step": 6535 + }, + { + "epoch": 0.7177685042828904, + "grad_norm": 2.2951819896698, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7134523391723633, + "num_tokens": 168217312.0, + "step": 6536 + }, + { + "epoch": 0.717878321985504, + "grad_norm": 2.2721407413482666, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7092896699905396, + "num_tokens": 168241635.0, + "step": 6537 + }, + { + "epoch": 0.7179881396881177, + "grad_norm": 2.5557587146759033, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7168837785720825, + "num_tokens": 168259769.0, + "step": 6538 + }, + { + "epoch": 0.7180979573907313, + "grad_norm": 2.3267574310302734, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7006303071975708, + "num_tokens": 168283251.0, + "step": 6539 + }, + { + "epoch": 0.7182077750933451, + "grad_norm": 2.4868531227111816, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.726433515548706, + "num_tokens": 168301780.0, + "step": 6540 + }, + { + "epoch": 0.7183175927959587, + "grad_norm": 2.251723289489746, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6895667910575867, + "num_tokens": 168327279.0, + "step": 6541 + }, + { + "epoch": 0.7184274104985724, + "grad_norm": 2.2407033443450928, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7139260768890381, + "num_tokens": 168351519.0, + "step": 6542 + }, + { + "epoch": 0.718537228201186, + "grad_norm": 2.1144559383392334, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7349812388420105, + "num_tokens": 168375619.0, + "step": 6543 + }, + { + "epoch": 0.7186470459037997, + "grad_norm": 2.5043249130249023, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7113410234451294, + "num_tokens": 168397255.0, + "step": 6544 + }, + { + "epoch": 0.7187568636064133, + "grad_norm": 2.402676582336426, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7310198545455933, + "num_tokens": 168417988.0, + "step": 6545 + }, + { + "epoch": 0.718866681309027, + "grad_norm": 2.1414220333099365, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6814588904380798, + "num_tokens": 168443940.0, + "step": 6546 + }, + { + "epoch": 0.7189764990116406, + "grad_norm": 2.5742008686065674, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7232122421264648, + "num_tokens": 168464580.0, + "step": 6547 + }, + { + "epoch": 0.7190863167142544, + "grad_norm": 2.2794132232666016, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6907155513763428, + "num_tokens": 168492136.0, + "step": 6548 + }, + { + "epoch": 0.719196134416868, + "grad_norm": 2.37773060798645, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7204158306121826, + "num_tokens": 168514665.0, + "step": 6549 + }, + { + "epoch": 0.7193059521194817, + "grad_norm": 2.0302157402038574, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7078473567962646, + "num_tokens": 168544168.0, + "step": 6550 + }, + { + "epoch": 0.7194157698220953, + "grad_norm": 2.1468443870544434, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7136381268501282, + "num_tokens": 168570246.0, + "step": 6551 + }, + { + "epoch": 0.719525587524709, + "grad_norm": 2.2441163063049316, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7033423185348511, + "num_tokens": 168595073.0, + "step": 6552 + }, + { + "epoch": 0.7196354052273226, + "grad_norm": 2.0136544704437256, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6964253783226013, + "num_tokens": 168623359.0, + "step": 6553 + }, + { + "epoch": 0.7197452229299363, + "grad_norm": 2.199911117553711, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6940559148788452, + "num_tokens": 168649004.0, + "step": 6554 + }, + { + "epoch": 0.71985504063255, + "grad_norm": 2.418194532394409, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7008503675460815, + "num_tokens": 168671500.0, + "step": 6555 + }, + { + "epoch": 0.7199648583351637, + "grad_norm": 2.1952412128448486, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7159149646759033, + "num_tokens": 168695977.0, + "step": 6556 + }, + { + "epoch": 0.7200746760377773, + "grad_norm": 2.0617308616638184, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7177865505218506, + "num_tokens": 168723045.0, + "step": 6557 + }, + { + "epoch": 0.720184493740391, + "grad_norm": 2.2710278034210205, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7039092183113098, + "num_tokens": 168747940.0, + "step": 6558 + }, + { + "epoch": 0.7202943114430046, + "grad_norm": 2.5484230518341064, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7197015285491943, + "num_tokens": 168766778.0, + "step": 6559 + }, + { + "epoch": 0.7204041291456182, + "grad_norm": 2.295621871948242, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7189385890960693, + "num_tokens": 168791257.0, + "step": 6560 + }, + { + "epoch": 0.7205139468482319, + "grad_norm": 2.152644157409668, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7165395021438599, + "num_tokens": 168815127.0, + "step": 6561 + }, + { + "epoch": 0.7206237645508456, + "grad_norm": 2.0927956104278564, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7197070121765137, + "num_tokens": 168840048.0, + "step": 6562 + }, + { + "epoch": 0.7207335822534593, + "grad_norm": 2.398420810699463, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7032908201217651, + "num_tokens": 168861444.0, + "step": 6563 + }, + { + "epoch": 0.7208433999560729, + "grad_norm": 2.0321924686431885, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7215486764907837, + "num_tokens": 168890325.0, + "step": 6564 + }, + { + "epoch": 0.7209532176586866, + "grad_norm": 2.1276025772094727, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7259437441825867, + "num_tokens": 168917855.0, + "step": 6565 + }, + { + "epoch": 0.7210630353613002, + "grad_norm": 2.2670092582702637, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.6979682445526123, + "num_tokens": 168942788.0, + "step": 6566 + }, + { + "epoch": 0.7211728530639139, + "grad_norm": 2.0284719467163086, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7088578939437866, + "num_tokens": 168971896.0, + "step": 6567 + }, + { + "epoch": 0.7212826707665275, + "grad_norm": 2.149134635925293, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.691299319267273, + "num_tokens": 169001001.0, + "step": 6568 + }, + { + "epoch": 0.7213924884691413, + "grad_norm": 2.4002163410186768, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7210167646408081, + "num_tokens": 169022267.0, + "step": 6569 + }, + { + "epoch": 0.7215023061717549, + "grad_norm": 2.120467185974121, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6933238506317139, + "num_tokens": 169051210.0, + "step": 6570 + }, + { + "epoch": 0.7216121238743686, + "grad_norm": 2.121544599533081, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7158395051956177, + "num_tokens": 169078261.0, + "step": 6571 + }, + { + "epoch": 0.7217219415769822, + "grad_norm": 2.1807312965393066, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7077295780181885, + "num_tokens": 169103264.0, + "step": 6572 + }, + { + "epoch": 0.7218317592795959, + "grad_norm": 2.3332064151763916, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7113674879074097, + "num_tokens": 169125993.0, + "step": 6573 + }, + { + "epoch": 0.7219415769822095, + "grad_norm": 1.9480208158493042, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.709597647190094, + "num_tokens": 169156737.0, + "step": 6574 + }, + { + "epoch": 0.7220513946848232, + "grad_norm": 2.191596508026123, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6973217725753784, + "num_tokens": 169181591.0, + "step": 6575 + }, + { + "epoch": 0.7221612123874368, + "grad_norm": 2.29152250289917, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7087815403938293, + "num_tokens": 169206282.0, + "step": 6576 + }, + { + "epoch": 0.7222710300900506, + "grad_norm": 2.0984227657318115, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6996901035308838, + "num_tokens": 169234381.0, + "step": 6577 + }, + { + "epoch": 0.7223808477926642, + "grad_norm": 2.3311243057250977, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7121419906616211, + "num_tokens": 169258063.0, + "step": 6578 + }, + { + "epoch": 0.7224906654952779, + "grad_norm": 2.288630247116089, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6965348124504089, + "num_tokens": 169283403.0, + "step": 6579 + }, + { + "epoch": 0.7226004831978915, + "grad_norm": 2.096048593521118, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7147454023361206, + "num_tokens": 169311230.0, + "step": 6580 + }, + { + "epoch": 0.7227103009005051, + "grad_norm": 2.1992902755737305, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.679040789604187, + "num_tokens": 169338192.0, + "step": 6581 + }, + { + "epoch": 0.7228201186031188, + "grad_norm": 2.330336809158325, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7056491374969482, + "num_tokens": 169360842.0, + "step": 6582 + }, + { + "epoch": 0.7229299363057324, + "grad_norm": 2.4468047618865967, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7039897441864014, + "num_tokens": 169380441.0, + "step": 6583 + }, + { + "epoch": 0.7230397540083462, + "grad_norm": 2.173246383666992, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7024679183959961, + "num_tokens": 169406832.0, + "step": 6584 + }, + { + "epoch": 0.7231495717109598, + "grad_norm": 1.998022437095642, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6881352663040161, + "num_tokens": 169436496.0, + "step": 6585 + }, + { + "epoch": 0.7232593894135735, + "grad_norm": 2.1346585750579834, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6936068534851074, + "num_tokens": 169464100.0, + "step": 6586 + }, + { + "epoch": 0.7233692071161871, + "grad_norm": 2.0116207599639893, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6956833600997925, + "num_tokens": 169493213.0, + "step": 6587 + }, + { + "epoch": 0.7234790248188008, + "grad_norm": 2.269505023956299, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7219708561897278, + "num_tokens": 169517094.0, + "step": 6588 + }, + { + "epoch": 0.7235888425214144, + "grad_norm": 2.193850517272949, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7101678848266602, + "num_tokens": 169543917.0, + "step": 6589 + }, + { + "epoch": 0.7236986602240281, + "grad_norm": 2.2460379600524902, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.712225079536438, + "num_tokens": 169566460.0, + "step": 6590 + }, + { + "epoch": 0.7238084779266418, + "grad_norm": 2.362807035446167, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7018730640411377, + "num_tokens": 169589890.0, + "step": 6591 + }, + { + "epoch": 0.7239182956292555, + "grad_norm": 2.2186079025268555, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7017449140548706, + "num_tokens": 169614534.0, + "step": 6592 + }, + { + "epoch": 0.7240281133318691, + "grad_norm": 2.2860107421875, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7252807021141052, + "num_tokens": 169638279.0, + "step": 6593 + }, + { + "epoch": 0.7241379310344828, + "grad_norm": 2.0348384380340576, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7071168422698975, + "num_tokens": 169666992.0, + "step": 6594 + }, + { + "epoch": 0.7242477487370964, + "grad_norm": 2.4855382442474365, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7170844078063965, + "num_tokens": 169687046.0, + "step": 6595 + }, + { + "epoch": 0.72435756643971, + "grad_norm": 2.292996644973755, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7025566101074219, + "num_tokens": 169709710.0, + "step": 6596 + }, + { + "epoch": 0.7244673841423237, + "grad_norm": 2.1995785236358643, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7150563597679138, + "num_tokens": 169735193.0, + "step": 6597 + }, + { + "epoch": 0.7245772018449375, + "grad_norm": 2.0532000064849854, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7097380757331848, + "num_tokens": 169763109.0, + "step": 6598 + }, + { + "epoch": 0.7246870195475511, + "grad_norm": 2.0187206268310547, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6926442384719849, + "num_tokens": 169792630.0, + "step": 6599 + }, + { + "epoch": 0.7247968372501647, + "grad_norm": 2.2682273387908936, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7207173109054565, + "num_tokens": 169816742.0, + "step": 6600 + }, + { + "epoch": 0.7249066549527784, + "grad_norm": 2.325662851333618, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7005504965782166, + "num_tokens": 169842169.0, + "step": 6601 + }, + { + "epoch": 0.725016472655392, + "grad_norm": 2.2397620677948, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7075430750846863, + "num_tokens": 169865950.0, + "step": 6602 + }, + { + "epoch": 0.7251262903580057, + "grad_norm": 2.1552650928497314, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6880097389221191, + "num_tokens": 169892405.0, + "step": 6603 + }, + { + "epoch": 0.7252361080606193, + "grad_norm": 2.165421485900879, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7357674837112427, + "num_tokens": 169915035.0, + "step": 6604 + }, + { + "epoch": 0.725345925763233, + "grad_norm": 2.209652900695801, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7145673036575317, + "num_tokens": 169938837.0, + "step": 6605 + }, + { + "epoch": 0.7254557434658467, + "grad_norm": 2.0560190677642822, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7086056470870972, + "num_tokens": 169968192.0, + "step": 6606 + }, + { + "epoch": 0.7255655611684604, + "grad_norm": 2.3598146438598633, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7231165766716003, + "num_tokens": 169990394.0, + "step": 6607 + }, + { + "epoch": 0.725675378871074, + "grad_norm": 2.1637353897094727, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7193518280982971, + "num_tokens": 170016546.0, + "step": 6608 + }, + { + "epoch": 0.7257851965736877, + "grad_norm": 2.0385334491729736, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6951476335525513, + "num_tokens": 170046868.0, + "step": 6609 + }, + { + "epoch": 0.7258950142763013, + "grad_norm": 2.424718141555786, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7294899821281433, + "num_tokens": 170065496.0, + "step": 6610 + }, + { + "epoch": 0.726004831978915, + "grad_norm": 2.1264359951019287, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7252933979034424, + "num_tokens": 170090381.0, + "step": 6611 + }, + { + "epoch": 0.7261146496815286, + "grad_norm": 2.2680976390838623, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7047445774078369, + "num_tokens": 170113950.0, + "step": 6612 + }, + { + "epoch": 0.7262244673841424, + "grad_norm": 2.1971871852874756, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6939152479171753, + "num_tokens": 170140215.0, + "step": 6613 + }, + { + "epoch": 0.726334285086756, + "grad_norm": 2.132568836212158, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6803749203681946, + "num_tokens": 170166285.0, + "step": 6614 + }, + { + "epoch": 0.7264441027893697, + "grad_norm": 2.6094653606414795, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7068103551864624, + "num_tokens": 170186662.0, + "step": 6615 + }, + { + "epoch": 0.7265539204919833, + "grad_norm": 1.9944994449615479, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7225493788719177, + "num_tokens": 170216996.0, + "step": 6616 + }, + { + "epoch": 0.726663738194597, + "grad_norm": 2.4388365745544434, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7205352187156677, + "num_tokens": 170238192.0, + "step": 6617 + }, + { + "epoch": 0.7267735558972106, + "grad_norm": 2.2965972423553467, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6895405054092407, + "num_tokens": 170261419.0, + "step": 6618 + }, + { + "epoch": 0.7268833735998242, + "grad_norm": 2.3224833011627197, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6962993144989014, + "num_tokens": 170285234.0, + "step": 6619 + }, + { + "epoch": 0.726993191302438, + "grad_norm": 1.9873374700546265, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7196997404098511, + "num_tokens": 170313605.0, + "step": 6620 + }, + { + "epoch": 0.7271030090050516, + "grad_norm": 1.9973191022872925, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7184291481971741, + "num_tokens": 170344962.0, + "step": 6621 + }, + { + "epoch": 0.7272128267076653, + "grad_norm": 1.9627273082733154, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.699539303779602, + "num_tokens": 170375405.0, + "step": 6622 + }, + { + "epoch": 0.7273226444102789, + "grad_norm": 2.2210302352905273, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6961625814437866, + "num_tokens": 170403293.0, + "step": 6623 + }, + { + "epoch": 0.7274324621128926, + "grad_norm": 2.1283504962921143, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6754255294799805, + "num_tokens": 170429232.0, + "step": 6624 + }, + { + "epoch": 0.7275422798155062, + "grad_norm": 2.314162492752075, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6785274744033813, + "num_tokens": 170452892.0, + "step": 6625 + }, + { + "epoch": 0.7276520975181199, + "grad_norm": 2.358691692352295, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7295936346054077, + "num_tokens": 170473390.0, + "step": 6626 + }, + { + "epoch": 0.7277619152207336, + "grad_norm": 1.8791968822479248, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6968539953231812, + "num_tokens": 170506687.0, + "step": 6627 + }, + { + "epoch": 0.7278717329233473, + "grad_norm": 2.436889410018921, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7232121229171753, + "num_tokens": 170528856.0, + "step": 6628 + }, + { + "epoch": 0.7279815506259609, + "grad_norm": 2.0705461502075195, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.698239803314209, + "num_tokens": 170557464.0, + "step": 6629 + }, + { + "epoch": 0.7280913683285746, + "grad_norm": 2.3580422401428223, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7102295160293579, + "num_tokens": 170578918.0, + "step": 6630 + }, + { + "epoch": 0.7282011860311882, + "grad_norm": 1.9042712450027466, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7100006341934204, + "num_tokens": 170610583.0, + "step": 6631 + }, + { + "epoch": 0.7283110037338019, + "grad_norm": 2.1136972904205322, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6972094178199768, + "num_tokens": 170634764.0, + "step": 6632 + }, + { + "epoch": 0.7284208214364155, + "grad_norm": 2.293607711791992, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7454245090484619, + "num_tokens": 170655736.0, + "step": 6633 + }, + { + "epoch": 0.7285306391390292, + "grad_norm": 2.3838050365448, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7178874015808105, + "num_tokens": 170678364.0, + "step": 6634 + }, + { + "epoch": 0.7286404568416429, + "grad_norm": 2.237654685974121, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7157378196716309, + "num_tokens": 170702660.0, + "step": 6635 + }, + { + "epoch": 0.7287502745442566, + "grad_norm": 1.9452626705169678, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6942347884178162, + "num_tokens": 170734239.0, + "step": 6636 + }, + { + "epoch": 0.7288600922468702, + "grad_norm": 2.2935354709625244, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7077674865722656, + "num_tokens": 170757423.0, + "step": 6637 + }, + { + "epoch": 0.7289699099494839, + "grad_norm": 1.9761875867843628, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7150330543518066, + "num_tokens": 170786258.0, + "step": 6638 + }, + { + "epoch": 0.7290797276520975, + "grad_norm": 2.1743199825286865, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7241500020027161, + "num_tokens": 170809954.0, + "step": 6639 + }, + { + "epoch": 0.7291895453547111, + "grad_norm": 2.1940879821777344, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7001824378967285, + "num_tokens": 170835126.0, + "step": 6640 + }, + { + "epoch": 0.7292993630573248, + "grad_norm": 2.328784465789795, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7127843499183655, + "num_tokens": 170858563.0, + "step": 6641 + }, + { + "epoch": 0.7294091807599385, + "grad_norm": 2.150374412536621, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6903830170631409, + "num_tokens": 170884580.0, + "step": 6642 + }, + { + "epoch": 0.7295189984625522, + "grad_norm": 2.006392240524292, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6981645822525024, + "num_tokens": 170914196.0, + "step": 6643 + }, + { + "epoch": 0.7296288161651658, + "grad_norm": 2.0670015811920166, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7017115354537964, + "num_tokens": 170941327.0, + "step": 6644 + }, + { + "epoch": 0.7297386338677795, + "grad_norm": 2.1150317192077637, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7047854065895081, + "num_tokens": 170969913.0, + "step": 6645 + }, + { + "epoch": 0.7298484515703931, + "grad_norm": 1.8876229524612427, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7055384516716003, + "num_tokens": 171002039.0, + "step": 6646 + }, + { + "epoch": 0.7299582692730068, + "grad_norm": 2.1321160793304443, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7036066651344299, + "num_tokens": 171025629.0, + "step": 6647 + }, + { + "epoch": 0.7300680869756204, + "grad_norm": 2.4661734104156494, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7148401737213135, + "num_tokens": 171046707.0, + "step": 6648 + }, + { + "epoch": 0.7301779046782342, + "grad_norm": 2.455838203430176, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7426764369010925, + "num_tokens": 171065215.0, + "step": 6649 + }, + { + "epoch": 0.7302877223808478, + "grad_norm": 2.1166462898254395, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7252329587936401, + "num_tokens": 171089367.0, + "step": 6650 + }, + { + "epoch": 0.7303975400834615, + "grad_norm": 2.034968137741089, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.6784614324569702, + "num_tokens": 171121964.0, + "step": 6651 + }, + { + "epoch": 0.7305073577860751, + "grad_norm": 2.164963960647583, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7201824188232422, + "num_tokens": 171145055.0, + "step": 6652 + }, + { + "epoch": 0.7306171754886888, + "grad_norm": 2.3222475051879883, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7133378982543945, + "num_tokens": 171166841.0, + "step": 6653 + }, + { + "epoch": 0.7307269931913024, + "grad_norm": 2.1535909175872803, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7260220050811768, + "num_tokens": 171191234.0, + "step": 6654 + }, + { + "epoch": 0.7308368108939161, + "grad_norm": 2.32735276222229, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7011607885360718, + "num_tokens": 171214766.0, + "step": 6655 + }, + { + "epoch": 0.7309466285965298, + "grad_norm": 2.2701337337493896, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7104427814483643, + "num_tokens": 171237859.0, + "step": 6656 + }, + { + "epoch": 0.7310564462991435, + "grad_norm": 2.1727519035339355, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6884562373161316, + "num_tokens": 171266299.0, + "step": 6657 + }, + { + "epoch": 0.7311662640017571, + "grad_norm": 2.2927322387695312, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.688704252243042, + "num_tokens": 171292217.0, + "step": 6658 + }, + { + "epoch": 0.7312760817043708, + "grad_norm": 2.1055564880371094, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7031293511390686, + "num_tokens": 171321915.0, + "step": 6659 + }, + { + "epoch": 0.7313858994069844, + "grad_norm": 2.120054244995117, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6973963975906372, + "num_tokens": 171349717.0, + "step": 6660 + }, + { + "epoch": 0.731495717109598, + "grad_norm": 2.0913071632385254, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6995656490325928, + "num_tokens": 171379196.0, + "step": 6661 + }, + { + "epoch": 0.7316055348122117, + "grad_norm": 2.3923842906951904, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7100152969360352, + "num_tokens": 171401725.0, + "step": 6662 + }, + { + "epoch": 0.7317153525148253, + "grad_norm": 2.067016124725342, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6944494843482971, + "num_tokens": 171429605.0, + "step": 6663 + }, + { + "epoch": 0.7318251702174391, + "grad_norm": 2.218787431716919, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7292812466621399, + "num_tokens": 171453868.0, + "step": 6664 + }, + { + "epoch": 0.7319349879200527, + "grad_norm": 2.1969079971313477, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6989631056785583, + "num_tokens": 171482119.0, + "step": 6665 + }, + { + "epoch": 0.7320448056226664, + "grad_norm": 2.5699973106384277, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7396843433380127, + "num_tokens": 171501068.0, + "step": 6666 + }, + { + "epoch": 0.73215462332528, + "grad_norm": 2.710479497909546, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7284302115440369, + "num_tokens": 171517369.0, + "step": 6667 + }, + { + "epoch": 0.7322644410278937, + "grad_norm": 1.946561574935913, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6996320486068726, + "num_tokens": 171546962.0, + "step": 6668 + }, + { + "epoch": 0.7323742587305073, + "grad_norm": 2.136704206466675, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.691807746887207, + "num_tokens": 171574996.0, + "step": 6669 + }, + { + "epoch": 0.732484076433121, + "grad_norm": 2.2197208404541016, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7169479131698608, + "num_tokens": 171598608.0, + "step": 6670 + }, + { + "epoch": 0.7325938941357347, + "grad_norm": 2.1107001304626465, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7039291262626648, + "num_tokens": 171625076.0, + "step": 6671 + }, + { + "epoch": 0.7327037118383484, + "grad_norm": 2.266507625579834, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7150281667709351, + "num_tokens": 171648718.0, + "step": 6672 + }, + { + "epoch": 0.732813529540962, + "grad_norm": 1.9956210851669312, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7288964986801147, + "num_tokens": 171677534.0, + "step": 6673 + }, + { + "epoch": 0.7329233472435757, + "grad_norm": 1.9069823026657104, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6869653463363647, + "num_tokens": 171712028.0, + "step": 6674 + }, + { + "epoch": 0.7330331649461893, + "grad_norm": 2.440741777420044, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7222304344177246, + "num_tokens": 171733916.0, + "step": 6675 + }, + { + "epoch": 0.733142982648803, + "grad_norm": 2.33969783782959, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7163164615631104, + "num_tokens": 171756852.0, + "step": 6676 + }, + { + "epoch": 0.7332528003514166, + "grad_norm": 2.1075639724731445, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.713613748550415, + "num_tokens": 171783104.0, + "step": 6677 + }, + { + "epoch": 0.7333626180540304, + "grad_norm": 2.110595941543579, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7252331972122192, + "num_tokens": 171809018.0, + "step": 6678 + }, + { + "epoch": 0.733472435756644, + "grad_norm": 2.441213369369507, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7139004468917847, + "num_tokens": 171829128.0, + "step": 6679 + }, + { + "epoch": 0.7335822534592576, + "grad_norm": 2.114655017852783, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6823610663414001, + "num_tokens": 171856674.0, + "step": 6680 + }, + { + "epoch": 0.7336920711618713, + "grad_norm": 2.0848090648651123, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6827960014343262, + "num_tokens": 171885692.0, + "step": 6681 + }, + { + "epoch": 0.7338018888644849, + "grad_norm": 1.9061989784240723, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7155789136886597, + "num_tokens": 171917494.0, + "step": 6682 + }, + { + "epoch": 0.7339117065670986, + "grad_norm": 1.7968333959579468, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6917390823364258, + "num_tokens": 171955760.0, + "step": 6683 + }, + { + "epoch": 0.7340215242697122, + "grad_norm": 2.1711673736572266, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.693932056427002, + "num_tokens": 171982552.0, + "step": 6684 + }, + { + "epoch": 0.734131341972326, + "grad_norm": 2.235956907272339, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6974213123321533, + "num_tokens": 172007905.0, + "step": 6685 + }, + { + "epoch": 0.7342411596749396, + "grad_norm": 1.9952718019485474, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7230073809623718, + "num_tokens": 172037131.0, + "step": 6686 + }, + { + "epoch": 0.7343509773775533, + "grad_norm": 2.078545093536377, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6707277894020081, + "num_tokens": 172066517.0, + "step": 6687 + }, + { + "epoch": 0.7344607950801669, + "grad_norm": 2.296654224395752, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7142822742462158, + "num_tokens": 172092440.0, + "step": 6688 + }, + { + "epoch": 0.7345706127827806, + "grad_norm": 1.992513656616211, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6956315040588379, + "num_tokens": 172121358.0, + "step": 6689 + }, + { + "epoch": 0.7346804304853942, + "grad_norm": 2.1613545417785645, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7078932523727417, + "num_tokens": 172148057.0, + "step": 6690 + }, + { + "epoch": 0.7347902481880079, + "grad_norm": 2.3252923488616943, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7206913232803345, + "num_tokens": 172170169.0, + "step": 6691 + }, + { + "epoch": 0.7349000658906216, + "grad_norm": 2.11077880859375, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6883163452148438, + "num_tokens": 172198231.0, + "step": 6692 + }, + { + "epoch": 0.7350098835932353, + "grad_norm": 2.0193254947662354, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7045811414718628, + "num_tokens": 172226759.0, + "step": 6693 + }, + { + "epoch": 0.7351197012958489, + "grad_norm": 2.0855283737182617, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7204368114471436, + "num_tokens": 172255375.0, + "step": 6694 + }, + { + "epoch": 0.7352295189984626, + "grad_norm": 2.4270739555358887, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7100458741188049, + "num_tokens": 172275164.0, + "step": 6695 + }, + { + "epoch": 0.7353393367010762, + "grad_norm": 2.283862352371216, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6866122484207153, + "num_tokens": 172300948.0, + "step": 6696 + }, + { + "epoch": 0.7354491544036899, + "grad_norm": 2.2137451171875, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7039799094200134, + "num_tokens": 172325495.0, + "step": 6697 + }, + { + "epoch": 0.7355589721063035, + "grad_norm": 2.2024497985839844, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.696654736995697, + "num_tokens": 172352991.0, + "step": 6698 + }, + { + "epoch": 0.7356687898089171, + "grad_norm": 2.3019111156463623, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7151073217391968, + "num_tokens": 172377914.0, + "step": 6699 + }, + { + "epoch": 0.7357786075115309, + "grad_norm": 2.3447611331939697, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7232166528701782, + "num_tokens": 172398408.0, + "step": 6700 + }, + { + "epoch": 0.7358884252141445, + "grad_norm": 2.299194097518921, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7113156914710999, + "num_tokens": 172421085.0, + "step": 6701 + }, + { + "epoch": 0.7359982429167582, + "grad_norm": 1.8801037073135376, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.713135838508606, + "num_tokens": 172455128.0, + "step": 6702 + }, + { + "epoch": 0.7361080606193718, + "grad_norm": 1.8637676239013672, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7117009162902832, + "num_tokens": 172486675.0, + "step": 6703 + }, + { + "epoch": 0.7362178783219855, + "grad_norm": 2.5063483715057373, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.731131911277771, + "num_tokens": 172507233.0, + "step": 6704 + }, + { + "epoch": 0.7363276960245991, + "grad_norm": 2.231138229370117, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7097524404525757, + "num_tokens": 172530925.0, + "step": 6705 + }, + { + "epoch": 0.7364375137272128, + "grad_norm": 2.49224853515625, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7065878510475159, + "num_tokens": 172552115.0, + "step": 6706 + }, + { + "epoch": 0.7365473314298265, + "grad_norm": 2.1869330406188965, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7276716232299805, + "num_tokens": 172575784.0, + "step": 6707 + }, + { + "epoch": 0.7366571491324402, + "grad_norm": 2.3823373317718506, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.717826247215271, + "num_tokens": 172596927.0, + "step": 6708 + }, + { + "epoch": 0.7367669668350538, + "grad_norm": 2.1255173683166504, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6922869682312012, + "num_tokens": 172626106.0, + "step": 6709 + }, + { + "epoch": 0.7368767845376675, + "grad_norm": 2.3907573223114014, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6914976835250854, + "num_tokens": 172649686.0, + "step": 6710 + }, + { + "epoch": 0.7369866022402811, + "grad_norm": 2.431279420852661, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7202951908111572, + "num_tokens": 172669825.0, + "step": 6711 + }, + { + "epoch": 0.7370964199428948, + "grad_norm": 2.0349514484405518, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.690029501914978, + "num_tokens": 172699526.0, + "step": 6712 + }, + { + "epoch": 0.7372062376455084, + "grad_norm": 2.138193130493164, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.702579140663147, + "num_tokens": 172725026.0, + "step": 6713 + }, + { + "epoch": 0.7373160553481222, + "grad_norm": 2.088132858276367, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7175585031509399, + "num_tokens": 172750966.0, + "step": 6714 + }, + { + "epoch": 0.7374258730507358, + "grad_norm": 2.0458984375, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7029224038124084, + "num_tokens": 172781034.0, + "step": 6715 + }, + { + "epoch": 0.7375356907533495, + "grad_norm": 2.073803663253784, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6987825632095337, + "num_tokens": 172808153.0, + "step": 6716 + }, + { + "epoch": 0.7376455084559631, + "grad_norm": 2.478780508041382, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6901237964630127, + "num_tokens": 172830543.0, + "step": 6717 + }, + { + "epoch": 0.7377553261585768, + "grad_norm": 2.26918888092041, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7107344269752502, + "num_tokens": 172853755.0, + "step": 6718 + }, + { + "epoch": 0.7378651438611904, + "grad_norm": 2.331508159637451, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7078882455825806, + "num_tokens": 172877009.0, + "step": 6719 + }, + { + "epoch": 0.737974961563804, + "grad_norm": 2.423816204071045, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7254132032394409, + "num_tokens": 172899647.0, + "step": 6720 + }, + { + "epoch": 0.7380847792664178, + "grad_norm": 2.2446885108947754, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7103321552276611, + "num_tokens": 172923922.0, + "step": 6721 + }, + { + "epoch": 0.7381945969690314, + "grad_norm": 2.224883556365967, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.699897050857544, + "num_tokens": 172947244.0, + "step": 6722 + }, + { + "epoch": 0.7383044146716451, + "grad_norm": 2.2692220211029053, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7112480401992798, + "num_tokens": 172971509.0, + "step": 6723 + }, + { + "epoch": 0.7384142323742587, + "grad_norm": 2.3262641429901123, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7135355472564697, + "num_tokens": 172994849.0, + "step": 6724 + }, + { + "epoch": 0.7385240500768724, + "grad_norm": 2.386831760406494, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.737395167350769, + "num_tokens": 173013028.0, + "step": 6725 + }, + { + "epoch": 0.738633867779486, + "grad_norm": 2.1995835304260254, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6975228786468506, + "num_tokens": 173038771.0, + "step": 6726 + }, + { + "epoch": 0.7387436854820997, + "grad_norm": 2.2561025619506836, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7229297161102295, + "num_tokens": 173061451.0, + "step": 6727 + }, + { + "epoch": 0.7388535031847133, + "grad_norm": 1.9389197826385498, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7082676887512207, + "num_tokens": 173093557.0, + "step": 6728 + }, + { + "epoch": 0.7389633208873271, + "grad_norm": 2.584205389022827, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7212754487991333, + "num_tokens": 173112927.0, + "step": 6729 + }, + { + "epoch": 0.7390731385899407, + "grad_norm": 2.153794765472412, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7059109807014465, + "num_tokens": 173139964.0, + "step": 6730 + }, + { + "epoch": 0.7391829562925544, + "grad_norm": 2.124831438064575, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7119121551513672, + "num_tokens": 173165548.0, + "step": 6731 + }, + { + "epoch": 0.739292773995168, + "grad_norm": 2.4652698040008545, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7075300216674805, + "num_tokens": 173185703.0, + "step": 6732 + }, + { + "epoch": 0.7394025916977817, + "grad_norm": 2.3283708095550537, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7233443260192871, + "num_tokens": 173208020.0, + "step": 6733 + }, + { + "epoch": 0.7395124094003953, + "grad_norm": 2.4468939304351807, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6961151957511902, + "num_tokens": 173232373.0, + "step": 6734 + }, + { + "epoch": 0.739622227103009, + "grad_norm": 2.1125378608703613, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7277594208717346, + "num_tokens": 173259602.0, + "step": 6735 + }, + { + "epoch": 0.7397320448056227, + "grad_norm": 2.248488664627075, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6938599348068237, + "num_tokens": 173288032.0, + "step": 6736 + }, + { + "epoch": 0.7398418625082364, + "grad_norm": 2.6227638721466064, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7014856338500977, + "num_tokens": 173307154.0, + "step": 6737 + }, + { + "epoch": 0.73995168021085, + "grad_norm": 2.2265098094940186, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7049176692962646, + "num_tokens": 173332189.0, + "step": 6738 + }, + { + "epoch": 0.7400614979134637, + "grad_norm": 2.2019829750061035, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7141531705856323, + "num_tokens": 173358422.0, + "step": 6739 + }, + { + "epoch": 0.7401713156160773, + "grad_norm": 2.2055959701538086, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7213661670684814, + "num_tokens": 173384207.0, + "step": 6740 + }, + { + "epoch": 0.7402811333186909, + "grad_norm": 2.341702461242676, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7381812334060669, + "num_tokens": 173406169.0, + "step": 6741 + }, + { + "epoch": 0.7403909510213046, + "grad_norm": 2.3073673248291016, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7301158905029297, + "num_tokens": 173428618.0, + "step": 6742 + }, + { + "epoch": 0.7405007687239183, + "grad_norm": 2.0042407512664795, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7116779088973999, + "num_tokens": 173458220.0, + "step": 6743 + }, + { + "epoch": 0.740610586426532, + "grad_norm": 1.9990452527999878, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6983043551445007, + "num_tokens": 173489193.0, + "step": 6744 + }, + { + "epoch": 0.7407204041291456, + "grad_norm": 2.429617404937744, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7296351194381714, + "num_tokens": 173508311.0, + "step": 6745 + }, + { + "epoch": 0.7408302218317593, + "grad_norm": 2.287544012069702, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7123717665672302, + "num_tokens": 173533344.0, + "step": 6746 + }, + { + "epoch": 0.7409400395343729, + "grad_norm": 2.2433905601501465, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7083521485328674, + "num_tokens": 173558840.0, + "step": 6747 + }, + { + "epoch": 0.7410498572369866, + "grad_norm": 2.0122644901275635, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6850346326828003, + "num_tokens": 173590047.0, + "step": 6748 + }, + { + "epoch": 0.7411596749396002, + "grad_norm": 2.0558583736419678, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6901958584785461, + "num_tokens": 173619360.0, + "step": 6749 + }, + { + "epoch": 0.741269492642214, + "grad_norm": 2.3178772926330566, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.707181990146637, + "num_tokens": 173641853.0, + "step": 6750 + }, + { + "epoch": 0.7413793103448276, + "grad_norm": 2.320443868637085, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6944080591201782, + "num_tokens": 173665359.0, + "step": 6751 + }, + { + "epoch": 0.7414891280474413, + "grad_norm": 2.3584628105163574, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7372467517852783, + "num_tokens": 173686429.0, + "step": 6752 + }, + { + "epoch": 0.7415989457500549, + "grad_norm": 2.385612964630127, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7069573998451233, + "num_tokens": 173708001.0, + "step": 6753 + }, + { + "epoch": 0.7417087634526686, + "grad_norm": 2.4421029090881348, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6933306455612183, + "num_tokens": 173731779.0, + "step": 6754 + }, + { + "epoch": 0.7418185811552822, + "grad_norm": 2.131004571914673, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7025725245475769, + "num_tokens": 173757496.0, + "step": 6755 + }, + { + "epoch": 0.7419283988578959, + "grad_norm": 2.103527545928955, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.713568925857544, + "num_tokens": 173783748.0, + "step": 6756 + }, + { + "epoch": 0.7420382165605095, + "grad_norm": 2.096787452697754, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6978687047958374, + "num_tokens": 173811522.0, + "step": 6757 + }, + { + "epoch": 0.7421480342631233, + "grad_norm": 2.2086949348449707, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6880660057067871, + "num_tokens": 173837650.0, + "step": 6758 + }, + { + "epoch": 0.7422578519657369, + "grad_norm": 2.1675987243652344, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6892255544662476, + "num_tokens": 173863496.0, + "step": 6759 + }, + { + "epoch": 0.7423676696683505, + "grad_norm": 2.1456902027130127, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.705345630645752, + "num_tokens": 173890643.0, + "step": 6760 + }, + { + "epoch": 0.7424774873709642, + "grad_norm": 2.5035512447357178, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7021621465682983, + "num_tokens": 173911917.0, + "step": 6761 + }, + { + "epoch": 0.7425873050735778, + "grad_norm": 1.9871339797973633, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6980630159378052, + "num_tokens": 173940616.0, + "step": 6762 + }, + { + "epoch": 0.7426971227761915, + "grad_norm": 2.0544333457946777, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7032653093338013, + "num_tokens": 173967162.0, + "step": 6763 + }, + { + "epoch": 0.7428069404788051, + "grad_norm": 2.2541067600250244, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6834248900413513, + "num_tokens": 173992578.0, + "step": 6764 + }, + { + "epoch": 0.7429167581814189, + "grad_norm": 2.1277015209198, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.677399754524231, + "num_tokens": 174020878.0, + "step": 6765 + }, + { + "epoch": 0.7430265758840325, + "grad_norm": 2.150923013687134, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7062277793884277, + "num_tokens": 174047018.0, + "step": 6766 + }, + { + "epoch": 0.7431363935866462, + "grad_norm": 2.0782692432403564, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6919240355491638, + "num_tokens": 174074917.0, + "step": 6767 + }, + { + "epoch": 0.7432462112892598, + "grad_norm": 2.0909929275512695, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6962108016014099, + "num_tokens": 174102555.0, + "step": 6768 + }, + { + "epoch": 0.7433560289918735, + "grad_norm": 2.283167600631714, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7235198020935059, + "num_tokens": 174125066.0, + "step": 6769 + }, + { + "epoch": 0.7434658466944871, + "grad_norm": 2.2370433807373047, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6989438533782959, + "num_tokens": 174150014.0, + "step": 6770 + }, + { + "epoch": 0.7435756643971008, + "grad_norm": 2.2362797260284424, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7263047695159912, + "num_tokens": 174173371.0, + "step": 6771 + }, + { + "epoch": 0.7436854820997145, + "grad_norm": 2.0598435401916504, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7294620275497437, + "num_tokens": 174199415.0, + "step": 6772 + }, + { + "epoch": 0.7437952998023282, + "grad_norm": 2.228273630142212, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7249688506126404, + "num_tokens": 174222858.0, + "step": 6773 + }, + { + "epoch": 0.7439051175049418, + "grad_norm": 1.9848508834838867, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7103996276855469, + "num_tokens": 174252127.0, + "step": 6774 + }, + { + "epoch": 0.7440149352075555, + "grad_norm": 2.1660799980163574, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7095454335212708, + "num_tokens": 174276857.0, + "step": 6775 + }, + { + "epoch": 0.7441247529101691, + "grad_norm": 2.1824631690979004, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.7008670568466187, + "num_tokens": 174302183.0, + "step": 6776 + }, + { + "epoch": 0.7442345706127828, + "grad_norm": 2.1620829105377197, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.7039505243301392, + "num_tokens": 174328101.0, + "step": 6777 + }, + { + "epoch": 0.7443443883153964, + "grad_norm": 2.5052170753479004, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.708457350730896, + "num_tokens": 174348291.0, + "step": 6778 + }, + { + "epoch": 0.7444542060180102, + "grad_norm": 2.221072196960449, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.705990731716156, + "num_tokens": 174372824.0, + "step": 6779 + }, + { + "epoch": 0.7445640237206238, + "grad_norm": 2.1985924243927, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7337957620620728, + "num_tokens": 174397257.0, + "step": 6780 + }, + { + "epoch": 0.7446738414232374, + "grad_norm": 2.1706936359405518, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.698493480682373, + "num_tokens": 174423653.0, + "step": 6781 + }, + { + "epoch": 0.7447836591258511, + "grad_norm": 2.325554132461548, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.68822181224823, + "num_tokens": 174445947.0, + "step": 6782 + }, + { + "epoch": 0.7448934768284647, + "grad_norm": 2.6185927391052246, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7309370636940002, + "num_tokens": 174464457.0, + "step": 6783 + }, + { + "epoch": 0.7450032945310784, + "grad_norm": 2.2306387424468994, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6931498646736145, + "num_tokens": 174489826.0, + "step": 6784 + }, + { + "epoch": 0.745113112233692, + "grad_norm": 2.108095169067383, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6956855058670044, + "num_tokens": 174516750.0, + "step": 6785 + }, + { + "epoch": 0.7452229299363057, + "grad_norm": 2.1696417331695557, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6911122798919678, + "num_tokens": 174543873.0, + "step": 6786 + }, + { + "epoch": 0.7453327476389194, + "grad_norm": 2.5973141193389893, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6987095475196838, + "num_tokens": 174563318.0, + "step": 6787 + }, + { + "epoch": 0.7454425653415331, + "grad_norm": 1.9221619367599487, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6946269273757935, + "num_tokens": 174594832.0, + "step": 6788 + }, + { + "epoch": 0.7455523830441467, + "grad_norm": 2.2180633544921875, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7029060125350952, + "num_tokens": 174618912.0, + "step": 6789 + }, + { + "epoch": 0.7456622007467604, + "grad_norm": 2.5321271419525146, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7219710946083069, + "num_tokens": 174637942.0, + "step": 6790 + }, + { + "epoch": 0.745772018449374, + "grad_norm": 2.555999755859375, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7269721031188965, + "num_tokens": 174657014.0, + "step": 6791 + }, + { + "epoch": 0.7458818361519877, + "grad_norm": 2.091538190841675, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7084780335426331, + "num_tokens": 174684799.0, + "step": 6792 + }, + { + "epoch": 0.7459916538546013, + "grad_norm": 1.872597098350525, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6956104040145874, + "num_tokens": 174719377.0, + "step": 6793 + }, + { + "epoch": 0.7461014715572151, + "grad_norm": 2.1688835620880127, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6913385987281799, + "num_tokens": 174746804.0, + "step": 6794 + }, + { + "epoch": 0.7462112892598287, + "grad_norm": 2.414280652999878, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6931315064430237, + "num_tokens": 174770808.0, + "step": 6795 + }, + { + "epoch": 0.7463211069624424, + "grad_norm": 2.355414867401123, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7162986993789673, + "num_tokens": 174795003.0, + "step": 6796 + }, + { + "epoch": 0.746430924665056, + "grad_norm": 1.9872561693191528, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7425011396408081, + "num_tokens": 174821912.0, + "step": 6797 + }, + { + "epoch": 0.7465407423676697, + "grad_norm": 2.2603700160980225, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7347146272659302, + "num_tokens": 174845257.0, + "step": 6798 + }, + { + "epoch": 0.7466505600702833, + "grad_norm": 1.9695100784301758, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7249077558517456, + "num_tokens": 174875229.0, + "step": 6799 + }, + { + "epoch": 0.7467603777728969, + "grad_norm": 2.2306344509124756, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7327868938446045, + "num_tokens": 174898858.0, + "step": 6800 + }, + { + "epoch": 0.7468701954755107, + "grad_norm": 2.3076133728027344, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6997421979904175, + "num_tokens": 174921381.0, + "step": 6801 + }, + { + "epoch": 0.7469800131781243, + "grad_norm": 2.4242825508117676, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7038624882698059, + "num_tokens": 174944889.0, + "step": 6802 + }, + { + "epoch": 0.747089830880738, + "grad_norm": 2.1466057300567627, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7269966006278992, + "num_tokens": 174969031.0, + "step": 6803 + }, + { + "epoch": 0.7471996485833516, + "grad_norm": 2.567812204360962, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.718122661113739, + "num_tokens": 174989516.0, + "step": 6804 + }, + { + "epoch": 0.7473094662859653, + "grad_norm": 2.2526400089263916, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6953959465026855, + "num_tokens": 175013545.0, + "step": 6805 + }, + { + "epoch": 0.7474192839885789, + "grad_norm": 2.4413135051727295, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7123699188232422, + "num_tokens": 175036001.0, + "step": 6806 + }, + { + "epoch": 0.7475291016911926, + "grad_norm": 2.1244399547576904, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6992287635803223, + "num_tokens": 175064323.0, + "step": 6807 + }, + { + "epoch": 0.7476389193938063, + "grad_norm": 2.209625244140625, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6798363924026489, + "num_tokens": 175091760.0, + "step": 6808 + }, + { + "epoch": 0.74774873709642, + "grad_norm": 2.2115695476531982, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7262466549873352, + "num_tokens": 175116555.0, + "step": 6809 + }, + { + "epoch": 0.7478585547990336, + "grad_norm": 1.992586374282837, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7106075286865234, + "num_tokens": 175146077.0, + "step": 6810 + }, + { + "epoch": 0.7479683725016473, + "grad_norm": 2.462334156036377, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7265691757202148, + "num_tokens": 175166181.0, + "step": 6811 + }, + { + "epoch": 0.7480781902042609, + "grad_norm": 2.280698299407959, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7228597402572632, + "num_tokens": 175188274.0, + "step": 6812 + }, + { + "epoch": 0.7481880079068746, + "grad_norm": 2.220628261566162, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7063673734664917, + "num_tokens": 175215445.0, + "step": 6813 + }, + { + "epoch": 0.7482978256094882, + "grad_norm": 2.0952162742614746, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7178121209144592, + "num_tokens": 175241706.0, + "step": 6814 + }, + { + "epoch": 0.7484076433121019, + "grad_norm": 2.059512138366699, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7103282809257507, + "num_tokens": 175269627.0, + "step": 6815 + }, + { + "epoch": 0.7485174610147156, + "grad_norm": 2.418768882751465, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7028063535690308, + "num_tokens": 175291530.0, + "step": 6816 + }, + { + "epoch": 0.7486272787173293, + "grad_norm": 2.4299228191375732, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6968794465065002, + "num_tokens": 175313430.0, + "step": 6817 + }, + { + "epoch": 0.7487370964199429, + "grad_norm": 1.9790560007095337, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.694045901298523, + "num_tokens": 175341824.0, + "step": 6818 + }, + { + "epoch": 0.7488469141225566, + "grad_norm": 2.2507450580596924, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7200736999511719, + "num_tokens": 175364117.0, + "step": 6819 + }, + { + "epoch": 0.7489567318251702, + "grad_norm": 2.3821229934692383, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7074239253997803, + "num_tokens": 175386150.0, + "step": 6820 + }, + { + "epoch": 0.7490665495277838, + "grad_norm": 2.3663666248321533, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6932926177978516, + "num_tokens": 175410098.0, + "step": 6821 + }, + { + "epoch": 0.7491763672303975, + "grad_norm": 2.1492180824279785, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6900707483291626, + "num_tokens": 175440049.0, + "step": 6822 + }, + { + "epoch": 0.7492861849330112, + "grad_norm": 2.2970974445343018, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6949465870857239, + "num_tokens": 175464464.0, + "step": 6823 + }, + { + "epoch": 0.7493960026356249, + "grad_norm": 2.0630722045898438, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7171239852905273, + "num_tokens": 175491690.0, + "step": 6824 + }, + { + "epoch": 0.7495058203382385, + "grad_norm": 2.3624110221862793, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7169827818870544, + "num_tokens": 175515298.0, + "step": 6825 + }, + { + "epoch": 0.7496156380408522, + "grad_norm": 2.1892402172088623, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7057431936264038, + "num_tokens": 175542314.0, + "step": 6826 + }, + { + "epoch": 0.7497254557434658, + "grad_norm": 2.1683337688446045, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7097455263137817, + "num_tokens": 175568712.0, + "step": 6827 + }, + { + "epoch": 0.7498352734460795, + "grad_norm": 2.144195079803467, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7025618553161621, + "num_tokens": 175595151.0, + "step": 6828 + }, + { + "epoch": 0.7499450911486931, + "grad_norm": 2.0262279510498047, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.717078447341919, + "num_tokens": 175622746.0, + "step": 6829 + }, + { + "epoch": 0.7500549088513069, + "grad_norm": 2.027827501296997, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7265273332595825, + "num_tokens": 175649681.0, + "step": 6830 + }, + { + "epoch": 0.7501647265539205, + "grad_norm": 2.19039249420166, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6862760782241821, + "num_tokens": 175675734.0, + "step": 6831 + }, + { + "epoch": 0.7502745442565342, + "grad_norm": 1.9689804315567017, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7089938521385193, + "num_tokens": 175703638.0, + "step": 6832 + }, + { + "epoch": 0.7503843619591478, + "grad_norm": 2.29970645904541, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7088531255722046, + "num_tokens": 175726530.0, + "step": 6833 + }, + { + "epoch": 0.7504941796617615, + "grad_norm": 2.209230422973633, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7352522611618042, + "num_tokens": 175750135.0, + "step": 6834 + }, + { + "epoch": 0.7506039973643751, + "grad_norm": 2.3339009284973145, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7078136205673218, + "num_tokens": 175772212.0, + "step": 6835 + }, + { + "epoch": 0.7507138150669888, + "grad_norm": 2.1336636543273926, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.702074408531189, + "num_tokens": 175801843.0, + "step": 6836 + }, + { + "epoch": 0.7508236327696025, + "grad_norm": 2.3035824298858643, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6912978887557983, + "num_tokens": 175826801.0, + "step": 6837 + }, + { + "epoch": 0.7509334504722162, + "grad_norm": 2.0869109630584717, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7200044393539429, + "num_tokens": 175855664.0, + "step": 6838 + }, + { + "epoch": 0.7510432681748298, + "grad_norm": 1.9500693082809448, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7105376720428467, + "num_tokens": 175888115.0, + "step": 6839 + }, + { + "epoch": 0.7511530858774434, + "grad_norm": 2.121964693069458, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7115776538848877, + "num_tokens": 175912510.0, + "step": 6840 + }, + { + "epoch": 0.7512629035800571, + "grad_norm": 2.2672297954559326, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7139505743980408, + "num_tokens": 175935060.0, + "step": 6841 + }, + { + "epoch": 0.7513727212826707, + "grad_norm": 2.290696382522583, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7304510474205017, + "num_tokens": 175957914.0, + "step": 6842 + }, + { + "epoch": 0.7514825389852844, + "grad_norm": 1.8392916917800903, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.706053614616394, + "num_tokens": 175991896.0, + "step": 6843 + }, + { + "epoch": 0.7515923566878981, + "grad_norm": 2.290426015853882, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.7018488645553589, + "num_tokens": 176016374.0, + "step": 6844 + }, + { + "epoch": 0.7517021743905118, + "grad_norm": 2.1412220001220703, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7078467011451721, + "num_tokens": 176042849.0, + "step": 6845 + }, + { + "epoch": 0.7518119920931254, + "grad_norm": 1.7628272771835327, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7081205248832703, + "num_tokens": 176078761.0, + "step": 6846 + }, + { + "epoch": 0.7519218097957391, + "grad_norm": 2.0429344177246094, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6954535841941833, + "num_tokens": 176107785.0, + "step": 6847 + }, + { + "epoch": 0.7520316274983527, + "grad_norm": 2.04947566986084, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7020154595375061, + "num_tokens": 176135553.0, + "step": 6848 + }, + { + "epoch": 0.7521414452009664, + "grad_norm": 1.9091095924377441, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6831227540969849, + "num_tokens": 176167561.0, + "step": 6849 + }, + { + "epoch": 0.75225126290358, + "grad_norm": 2.0127477645874023, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.702269434928894, + "num_tokens": 176196199.0, + "step": 6850 + }, + { + "epoch": 0.7523610806061937, + "grad_norm": 2.0340123176574707, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6848115921020508, + "num_tokens": 176225572.0, + "step": 6851 + }, + { + "epoch": 0.7524708983088074, + "grad_norm": 2.1893279552459717, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.69095379114151, + "num_tokens": 176251730.0, + "step": 6852 + }, + { + "epoch": 0.7525807160114211, + "grad_norm": 2.3511714935302734, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.7108187675476074, + "num_tokens": 176276599.0, + "step": 6853 + }, + { + "epoch": 0.7526905337140347, + "grad_norm": 2.1242523193359375, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.7035521268844604, + "num_tokens": 176302098.0, + "step": 6854 + }, + { + "epoch": 0.7528003514166484, + "grad_norm": 2.0526180267333984, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.724413275718689, + "num_tokens": 176328743.0, + "step": 6855 + }, + { + "epoch": 0.752910169119262, + "grad_norm": 2.004887819290161, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7026278972625732, + "num_tokens": 176360137.0, + "step": 6856 + }, + { + "epoch": 0.7530199868218757, + "grad_norm": 2.131023406982422, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6928784847259521, + "num_tokens": 176385626.0, + "step": 6857 + }, + { + "epoch": 0.7531298045244893, + "grad_norm": 2.1472933292388916, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6777825951576233, + "num_tokens": 176412769.0, + "step": 6858 + }, + { + "epoch": 0.7532396222271031, + "grad_norm": 1.9129328727722168, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7079648971557617, + "num_tokens": 176442903.0, + "step": 6859 + }, + { + "epoch": 0.7533494399297167, + "grad_norm": 2.294874429702759, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.710509181022644, + "num_tokens": 176465777.0, + "step": 6860 + }, + { + "epoch": 0.7534592576323303, + "grad_norm": 2.3161544799804688, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6928623914718628, + "num_tokens": 176491017.0, + "step": 6861 + }, + { + "epoch": 0.753569075334944, + "grad_norm": 2.100034236907959, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7114306688308716, + "num_tokens": 176518940.0, + "step": 6862 + }, + { + "epoch": 0.7536788930375576, + "grad_norm": 1.947011113166809, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7237653732299805, + "num_tokens": 176550099.0, + "step": 6863 + }, + { + "epoch": 0.7537887107401713, + "grad_norm": 2.167515516281128, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.695929765701294, + "num_tokens": 176575942.0, + "step": 6864 + }, + { + "epoch": 0.7538985284427849, + "grad_norm": 2.0727264881134033, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.7104660272598267, + "num_tokens": 176602816.0, + "step": 6865 + }, + { + "epoch": 0.7540083461453987, + "grad_norm": 2.1003551483154297, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7185670137405396, + "num_tokens": 176630101.0, + "step": 6866 + }, + { + "epoch": 0.7541181638480123, + "grad_norm": 2.4458134174346924, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7015889883041382, + "num_tokens": 176651404.0, + "step": 6867 + }, + { + "epoch": 0.754227981550626, + "grad_norm": 2.1546695232391357, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7262318730354309, + "num_tokens": 176675991.0, + "step": 6868 + }, + { + "epoch": 0.7543377992532396, + "grad_norm": 2.0308687686920166, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7037106156349182, + "num_tokens": 176705792.0, + "step": 6869 + }, + { + "epoch": 0.7544476169558533, + "grad_norm": 1.9572898149490356, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7072583436965942, + "num_tokens": 176735122.0, + "step": 6870 + }, + { + "epoch": 0.7545574346584669, + "grad_norm": 2.4215047359466553, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7170685529708862, + "num_tokens": 176755496.0, + "step": 6871 + }, + { + "epoch": 0.7546672523610806, + "grad_norm": 2.273326873779297, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7110674381256104, + "num_tokens": 176781468.0, + "step": 6872 + }, + { + "epoch": 0.7547770700636943, + "grad_norm": 2.169562816619873, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7046537399291992, + "num_tokens": 176806751.0, + "step": 6873 + }, + { + "epoch": 0.754886887766308, + "grad_norm": 2.010277271270752, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7145174741744995, + "num_tokens": 176836814.0, + "step": 6874 + }, + { + "epoch": 0.7549967054689216, + "grad_norm": 2.265725612640381, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7033398151397705, + "num_tokens": 176860370.0, + "step": 6875 + }, + { + "epoch": 0.7551065231715353, + "grad_norm": 2.3080942630767822, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6988078355789185, + "num_tokens": 176886052.0, + "step": 6876 + }, + { + "epoch": 0.7552163408741489, + "grad_norm": 2.425135850906372, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.701386570930481, + "num_tokens": 176906714.0, + "step": 6877 + }, + { + "epoch": 0.7553261585767626, + "grad_norm": 2.270551919937134, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6877566576004028, + "num_tokens": 176931567.0, + "step": 6878 + }, + { + "epoch": 0.7554359762793762, + "grad_norm": 1.890663981437683, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7188991904258728, + "num_tokens": 176965222.0, + "step": 6879 + }, + { + "epoch": 0.7555457939819898, + "grad_norm": 2.6815104484558105, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7180532217025757, + "num_tokens": 176981866.0, + "step": 6880 + }, + { + "epoch": 0.7556556116846036, + "grad_norm": 2.120378255844116, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7039920687675476, + "num_tokens": 177009712.0, + "step": 6881 + }, + { + "epoch": 0.7557654293872172, + "grad_norm": 2.016315221786499, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7157154083251953, + "num_tokens": 177037021.0, + "step": 6882 + }, + { + "epoch": 0.7558752470898309, + "grad_norm": 2.1971092224121094, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7036833763122559, + "num_tokens": 177063312.0, + "step": 6883 + }, + { + "epoch": 0.7559850647924445, + "grad_norm": 2.109032154083252, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7022242546081543, + "num_tokens": 177090038.0, + "step": 6884 + }, + { + "epoch": 0.7560948824950582, + "grad_norm": 2.451542377471924, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7203242778778076, + "num_tokens": 177110358.0, + "step": 6885 + }, + { + "epoch": 0.7562047001976718, + "grad_norm": 1.9271117448806763, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6968661546707153, + "num_tokens": 177142846.0, + "step": 6886 + }, + { + "epoch": 0.7563145179002855, + "grad_norm": 2.194183826446533, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7200928330421448, + "num_tokens": 177166483.0, + "step": 6887 + }, + { + "epoch": 0.7564243356028992, + "grad_norm": 2.0568437576293945, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6993193030357361, + "num_tokens": 177196280.0, + "step": 6888 + }, + { + "epoch": 0.7565341533055129, + "grad_norm": 2.042865037918091, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6922749280929565, + "num_tokens": 177224460.0, + "step": 6889 + }, + { + "epoch": 0.7566439710081265, + "grad_norm": 2.3374056816101074, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.708428144454956, + "num_tokens": 177248892.0, + "step": 6890 + }, + { + "epoch": 0.7567537887107402, + "grad_norm": 1.9585999250411987, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7290968298912048, + "num_tokens": 177277439.0, + "step": 6891 + }, + { + "epoch": 0.7568636064133538, + "grad_norm": 2.4494738578796387, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7058132886886597, + "num_tokens": 177297262.0, + "step": 6892 + }, + { + "epoch": 0.7569734241159675, + "grad_norm": 2.3158071041107178, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.696754515171051, + "num_tokens": 177321468.0, + "step": 6893 + }, + { + "epoch": 0.7570832418185811, + "grad_norm": 2.4888803958892822, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7198442220687866, + "num_tokens": 177340244.0, + "step": 6894 + }, + { + "epoch": 0.7571930595211949, + "grad_norm": 2.1261789798736572, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7213997840881348, + "num_tokens": 177363961.0, + "step": 6895 + }, + { + "epoch": 0.7573028772238085, + "grad_norm": 2.3945751190185547, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7212350368499756, + "num_tokens": 177385160.0, + "step": 6896 + }, + { + "epoch": 0.7574126949264222, + "grad_norm": 2.2031021118164062, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6984272599220276, + "num_tokens": 177409328.0, + "step": 6897 + }, + { + "epoch": 0.7575225126290358, + "grad_norm": 2.3198094367980957, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7154948711395264, + "num_tokens": 177430924.0, + "step": 6898 + }, + { + "epoch": 0.7576323303316495, + "grad_norm": 1.9323270320892334, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6865862607955933, + "num_tokens": 177463032.0, + "step": 6899 + }, + { + "epoch": 0.7577421480342631, + "grad_norm": 2.0890896320343018, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.698079526424408, + "num_tokens": 177490418.0, + "step": 6900 + }, + { + "epoch": 0.7578519657368767, + "grad_norm": 1.9343534708023071, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7084869146347046, + "num_tokens": 177523422.0, + "step": 6901 + }, + { + "epoch": 0.7579617834394905, + "grad_norm": 2.4847376346588135, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6983109712600708, + "num_tokens": 177542358.0, + "step": 6902 + }, + { + "epoch": 0.7580716011421041, + "grad_norm": 2.09855055809021, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7146921157836914, + "num_tokens": 177568684.0, + "step": 6903 + }, + { + "epoch": 0.7581814188447178, + "grad_norm": 2.102203130722046, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7089530825614929, + "num_tokens": 177595802.0, + "step": 6904 + }, + { + "epoch": 0.7582912365473314, + "grad_norm": 1.99532151222229, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7166446447372437, + "num_tokens": 177624379.0, + "step": 6905 + }, + { + "epoch": 0.7584010542499451, + "grad_norm": 2.1166720390319824, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.671282172203064, + "num_tokens": 177651149.0, + "step": 6906 + }, + { + "epoch": 0.7585108719525587, + "grad_norm": 2.3986308574676514, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7358517646789551, + "num_tokens": 177672385.0, + "step": 6907 + }, + { + "epoch": 0.7586206896551724, + "grad_norm": 2.1314010620117188, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6936308145523071, + "num_tokens": 177700101.0, + "step": 6908 + }, + { + "epoch": 0.758730507357786, + "grad_norm": 2.1849796772003174, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7112319469451904, + "num_tokens": 177724713.0, + "step": 6909 + }, + { + "epoch": 0.7588403250603998, + "grad_norm": 2.242795705795288, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7260640859603882, + "num_tokens": 177747673.0, + "step": 6910 + }, + { + "epoch": 0.7589501427630134, + "grad_norm": 1.9730814695358276, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7404889464378357, + "num_tokens": 177774250.0, + "step": 6911 + }, + { + "epoch": 0.7590599604656271, + "grad_norm": 1.990263819694519, + "learning_rate": 1e-06, + "loss": 1.1037, + "mean_token_accuracy": 0.6673504710197449, + "num_tokens": 177807814.0, + "step": 6912 + }, + { + "epoch": 0.7591697781682407, + "grad_norm": 2.4803404808044434, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7108141183853149, + "num_tokens": 177828946.0, + "step": 6913 + }, + { + "epoch": 0.7592795958708544, + "grad_norm": 2.09173321723938, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7100237607955933, + "num_tokens": 177855843.0, + "step": 6914 + }, + { + "epoch": 0.759389413573468, + "grad_norm": 2.1787197589874268, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7054690718650818, + "num_tokens": 177881428.0, + "step": 6915 + }, + { + "epoch": 0.7594992312760817, + "grad_norm": 2.3084709644317627, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.696992039680481, + "num_tokens": 177904668.0, + "step": 6916 + }, + { + "epoch": 0.7596090489786954, + "grad_norm": 1.956477403640747, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.701026439666748, + "num_tokens": 177935665.0, + "step": 6917 + }, + { + "epoch": 0.7597188666813091, + "grad_norm": 2.12496018409729, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.689665675163269, + "num_tokens": 177963199.0, + "step": 6918 + }, + { + "epoch": 0.7598286843839227, + "grad_norm": 2.1677610874176025, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7300318479537964, + "num_tokens": 177988405.0, + "step": 6919 + }, + { + "epoch": 0.7599385020865363, + "grad_norm": 2.068166732788086, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7017574906349182, + "num_tokens": 178015821.0, + "step": 6920 + }, + { + "epoch": 0.76004831978915, + "grad_norm": 1.9507299661636353, + "learning_rate": 1e-06, + "loss": 1.0977, + "mean_token_accuracy": 0.671359121799469, + "num_tokens": 178049110.0, + "step": 6921 + }, + { + "epoch": 0.7601581374917636, + "grad_norm": 2.249756097793579, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.72989821434021, + "num_tokens": 178073863.0, + "step": 6922 + }, + { + "epoch": 0.7602679551943773, + "grad_norm": 2.250670909881592, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7325514554977417, + "num_tokens": 178097462.0, + "step": 6923 + }, + { + "epoch": 0.760377772896991, + "grad_norm": 2.190317153930664, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6958277225494385, + "num_tokens": 178124856.0, + "step": 6924 + }, + { + "epoch": 0.7604875905996047, + "grad_norm": 2.669361114501953, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7169902324676514, + "num_tokens": 178144260.0, + "step": 6925 + }, + { + "epoch": 0.7605974083022183, + "grad_norm": 2.4342174530029297, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7252256274223328, + "num_tokens": 178164620.0, + "step": 6926 + }, + { + "epoch": 0.760707226004832, + "grad_norm": 2.0887765884399414, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7104322910308838, + "num_tokens": 178191416.0, + "step": 6927 + }, + { + "epoch": 0.7608170437074456, + "grad_norm": 1.8861117362976074, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.697516143321991, + "num_tokens": 178225549.0, + "step": 6928 + }, + { + "epoch": 0.7609268614100593, + "grad_norm": 2.5899507999420166, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.723612904548645, + "num_tokens": 178243188.0, + "step": 6929 + }, + { + "epoch": 0.7610366791126729, + "grad_norm": 2.581785202026367, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6859942078590393, + "num_tokens": 178265651.0, + "step": 6930 + }, + { + "epoch": 0.7611464968152867, + "grad_norm": 2.292464017868042, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7097769975662231, + "num_tokens": 178289820.0, + "step": 6931 + }, + { + "epoch": 0.7612563145179003, + "grad_norm": 2.1909897327423096, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6940058469772339, + "num_tokens": 178316126.0, + "step": 6932 + }, + { + "epoch": 0.761366132220514, + "grad_norm": 2.1427230834960938, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7094514966011047, + "num_tokens": 178341132.0, + "step": 6933 + }, + { + "epoch": 0.7614759499231276, + "grad_norm": 2.1799590587615967, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.722858726978302, + "num_tokens": 178365239.0, + "step": 6934 + }, + { + "epoch": 0.7615857676257413, + "grad_norm": 2.1815690994262695, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7080675959587097, + "num_tokens": 178390432.0, + "step": 6935 + }, + { + "epoch": 0.7616955853283549, + "grad_norm": 1.807737112045288, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6961379647254944, + "num_tokens": 178426245.0, + "step": 6936 + }, + { + "epoch": 0.7618054030309686, + "grad_norm": 2.459519863128662, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7142691016197205, + "num_tokens": 178446750.0, + "step": 6937 + }, + { + "epoch": 0.7619152207335822, + "grad_norm": 2.3081328868865967, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7330189943313599, + "num_tokens": 178468359.0, + "step": 6938 + }, + { + "epoch": 0.762025038436196, + "grad_norm": 2.049093246459961, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7278202772140503, + "num_tokens": 178496165.0, + "step": 6939 + }, + { + "epoch": 0.7621348561388096, + "grad_norm": 2.211754322052002, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7060264945030212, + "num_tokens": 178522144.0, + "step": 6940 + }, + { + "epoch": 0.7622446738414232, + "grad_norm": 2.306436538696289, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7263031005859375, + "num_tokens": 178543914.0, + "step": 6941 + }, + { + "epoch": 0.7623544915440369, + "grad_norm": 2.06969952583313, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7351036667823792, + "num_tokens": 178569909.0, + "step": 6942 + }, + { + "epoch": 0.7624643092466505, + "grad_norm": 2.085517644882202, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7429314255714417, + "num_tokens": 178595470.0, + "step": 6943 + }, + { + "epoch": 0.7625741269492642, + "grad_norm": 2.2968156337738037, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6940082907676697, + "num_tokens": 178619744.0, + "step": 6944 + }, + { + "epoch": 0.7626839446518778, + "grad_norm": 2.097017288208008, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7128230929374695, + "num_tokens": 178646043.0, + "step": 6945 + }, + { + "epoch": 0.7627937623544916, + "grad_norm": 2.116455078125, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7123046517372131, + "num_tokens": 178674317.0, + "step": 6946 + }, + { + "epoch": 0.7629035800571052, + "grad_norm": 2.522073984146118, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7153583765029907, + "num_tokens": 178693326.0, + "step": 6947 + }, + { + "epoch": 0.7630133977597189, + "grad_norm": 2.1436426639556885, + "learning_rate": 1e-06, + "loss": 1.0934, + "mean_token_accuracy": 0.6771591901779175, + "num_tokens": 178721913.0, + "step": 6948 + }, + { + "epoch": 0.7631232154623325, + "grad_norm": 2.215965747833252, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.724947452545166, + "num_tokens": 178745153.0, + "step": 6949 + }, + { + "epoch": 0.7632330331649462, + "grad_norm": 2.295057773590088, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7128440141677856, + "num_tokens": 178767797.0, + "step": 6950 + }, + { + "epoch": 0.7633428508675598, + "grad_norm": 2.0296549797058105, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6896512508392334, + "num_tokens": 178799152.0, + "step": 6951 + }, + { + "epoch": 0.7634526685701735, + "grad_norm": 2.077906608581543, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6987674832344055, + "num_tokens": 178827617.0, + "step": 6952 + }, + { + "epoch": 0.7635624862727872, + "grad_norm": 2.190993309020996, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7346551418304443, + "num_tokens": 178852040.0, + "step": 6953 + }, + { + "epoch": 0.7636723039754009, + "grad_norm": 2.211764097213745, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6877530217170715, + "num_tokens": 178877754.0, + "step": 6954 + }, + { + "epoch": 0.7637821216780145, + "grad_norm": 2.040724992752075, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.737373948097229, + "num_tokens": 178903071.0, + "step": 6955 + }, + { + "epoch": 0.7638919393806282, + "grad_norm": 2.012651205062866, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6979930400848389, + "num_tokens": 178931150.0, + "step": 6956 + }, + { + "epoch": 0.7640017570832418, + "grad_norm": 2.3770086765289307, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7019919753074646, + "num_tokens": 178952134.0, + "step": 6957 + }, + { + "epoch": 0.7641115747858555, + "grad_norm": 2.0798635482788086, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6929094791412354, + "num_tokens": 178979899.0, + "step": 6958 + }, + { + "epoch": 0.7642213924884691, + "grad_norm": 2.157613754272461, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7259520292282104, + "num_tokens": 179003893.0, + "step": 6959 + }, + { + "epoch": 0.7643312101910829, + "grad_norm": 2.3322818279266357, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7193723917007446, + "num_tokens": 179024899.0, + "step": 6960 + }, + { + "epoch": 0.7644410278936965, + "grad_norm": 2.7647716999053955, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7126410007476807, + "num_tokens": 179042921.0, + "step": 6961 + }, + { + "epoch": 0.7645508455963101, + "grad_norm": 2.1045117378234863, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6942113041877747, + "num_tokens": 179069668.0, + "step": 6962 + }, + { + "epoch": 0.7646606632989238, + "grad_norm": 2.378385543823242, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6999649405479431, + "num_tokens": 179103443.0, + "step": 6963 + }, + { + "epoch": 0.7647704810015374, + "grad_norm": 2.189678430557251, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6987247467041016, + "num_tokens": 179129078.0, + "step": 6964 + }, + { + "epoch": 0.7648802987041511, + "grad_norm": 2.1090738773345947, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7223566770553589, + "num_tokens": 179155610.0, + "step": 6965 + }, + { + "epoch": 0.7649901164067647, + "grad_norm": 2.1237027645111084, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7065994143486023, + "num_tokens": 179185250.0, + "step": 6966 + }, + { + "epoch": 0.7650999341093784, + "grad_norm": 2.4517111778259277, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7293306589126587, + "num_tokens": 179206553.0, + "step": 6967 + }, + { + "epoch": 0.7652097518119921, + "grad_norm": 2.2071661949157715, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6976795792579651, + "num_tokens": 179232521.0, + "step": 6968 + }, + { + "epoch": 0.7653195695146058, + "grad_norm": 2.0719051361083984, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7212638854980469, + "num_tokens": 179259682.0, + "step": 6969 + }, + { + "epoch": 0.7654293872172194, + "grad_norm": 1.9297693967819214, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6878448128700256, + "num_tokens": 179289707.0, + "step": 6970 + }, + { + "epoch": 0.7655392049198331, + "grad_norm": 2.272934913635254, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6985509395599365, + "num_tokens": 179313366.0, + "step": 6971 + }, + { + "epoch": 0.7656490226224467, + "grad_norm": 2.2151997089385986, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7151864767074585, + "num_tokens": 179337257.0, + "step": 6972 + }, + { + "epoch": 0.7657588403250604, + "grad_norm": 2.2357327938079834, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7287467122077942, + "num_tokens": 179359729.0, + "step": 6973 + }, + { + "epoch": 0.765868658027674, + "grad_norm": 2.3027398586273193, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7037678956985474, + "num_tokens": 179382599.0, + "step": 6974 + }, + { + "epoch": 0.7659784757302878, + "grad_norm": 1.9923051595687866, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7097965478897095, + "num_tokens": 179413673.0, + "step": 6975 + }, + { + "epoch": 0.7660882934329014, + "grad_norm": 2.2403976917266846, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6955466270446777, + "num_tokens": 179438117.0, + "step": 6976 + }, + { + "epoch": 0.7661981111355151, + "grad_norm": 2.24955415725708, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7049790620803833, + "num_tokens": 179461620.0, + "step": 6977 + }, + { + "epoch": 0.7663079288381287, + "grad_norm": 2.455631732940674, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7244217991828918, + "num_tokens": 179480282.0, + "step": 6978 + }, + { + "epoch": 0.7664177465407424, + "grad_norm": 2.4898738861083984, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6980969905853271, + "num_tokens": 179501081.0, + "step": 6979 + }, + { + "epoch": 0.766527564243356, + "grad_norm": 2.054262161254883, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6870467066764832, + "num_tokens": 179529967.0, + "step": 6980 + }, + { + "epoch": 0.7666373819459696, + "grad_norm": 2.10365629196167, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7159532308578491, + "num_tokens": 179556966.0, + "step": 6981 + }, + { + "epoch": 0.7667471996485834, + "grad_norm": 2.132560968399048, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6759036779403687, + "num_tokens": 179584927.0, + "step": 6982 + }, + { + "epoch": 0.766857017351197, + "grad_norm": 2.4843480587005615, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6891685128211975, + "num_tokens": 179606109.0, + "step": 6983 + }, + { + "epoch": 0.7669668350538107, + "grad_norm": 2.0164339542388916, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7064549922943115, + "num_tokens": 179636200.0, + "step": 6984 + }, + { + "epoch": 0.7670766527564243, + "grad_norm": 2.0800669193267822, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7308292388916016, + "num_tokens": 179662333.0, + "step": 6985 + }, + { + "epoch": 0.767186470459038, + "grad_norm": 2.3419082164764404, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7150183916091919, + "num_tokens": 179683759.0, + "step": 6986 + }, + { + "epoch": 0.7672962881616516, + "grad_norm": 2.3217809200286865, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7121404409408569, + "num_tokens": 179706553.0, + "step": 6987 + }, + { + "epoch": 0.7674061058642653, + "grad_norm": 1.918562650680542, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6959936618804932, + "num_tokens": 179739181.0, + "step": 6988 + }, + { + "epoch": 0.767515923566879, + "grad_norm": 2.1980693340301514, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7054612040519714, + "num_tokens": 179763426.0, + "step": 6989 + }, + { + "epoch": 0.7676257412694927, + "grad_norm": 2.071274757385254, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7012970447540283, + "num_tokens": 179794635.0, + "step": 6990 + }, + { + "epoch": 0.7677355589721063, + "grad_norm": 2.240135908126831, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7005792856216431, + "num_tokens": 179819021.0, + "step": 6991 + }, + { + "epoch": 0.76784537667472, + "grad_norm": 1.9782134294509888, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6820046305656433, + "num_tokens": 179852876.0, + "step": 6992 + }, + { + "epoch": 0.7679551943773336, + "grad_norm": 2.107316732406616, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.713523805141449, + "num_tokens": 179880865.0, + "step": 6993 + }, + { + "epoch": 0.7680650120799473, + "grad_norm": 2.2542049884796143, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7325128316879272, + "num_tokens": 179903347.0, + "step": 6994 + }, + { + "epoch": 0.7681748297825609, + "grad_norm": 2.109816789627075, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7049468755722046, + "num_tokens": 179930109.0, + "step": 6995 + }, + { + "epoch": 0.7682846474851747, + "grad_norm": 2.0249080657958984, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6912378072738647, + "num_tokens": 179959874.0, + "step": 6996 + }, + { + "epoch": 0.7683944651877883, + "grad_norm": 2.299445390701294, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6972379684448242, + "num_tokens": 179983936.0, + "step": 6997 + }, + { + "epoch": 0.768504282890402, + "grad_norm": 2.2115352153778076, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6886590719223022, + "num_tokens": 180009084.0, + "step": 6998 + }, + { + "epoch": 0.7686141005930156, + "grad_norm": 2.1127641201019287, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7136588096618652, + "num_tokens": 180037190.0, + "step": 6999 + }, + { + "epoch": 0.7687239182956292, + "grad_norm": 2.5392611026763916, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7285685539245605, + "num_tokens": 180058538.0, + "step": 7000 + }, + { + "epoch": 0.7688337359982429, + "grad_norm": 2.2660388946533203, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7123969197273254, + "num_tokens": 180080618.0, + "step": 7001 + }, + { + "epoch": 0.7689435537008565, + "grad_norm": 2.138284206390381, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6998785138130188, + "num_tokens": 180107780.0, + "step": 7002 + }, + { + "epoch": 0.7690533714034702, + "grad_norm": 2.0466549396514893, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7057422399520874, + "num_tokens": 180134461.0, + "step": 7003 + }, + { + "epoch": 0.7691631891060839, + "grad_norm": 2.0475010871887207, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7272585034370422, + "num_tokens": 180162921.0, + "step": 7004 + }, + { + "epoch": 0.7692730068086976, + "grad_norm": 2.339871883392334, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7124228477478027, + "num_tokens": 180184197.0, + "step": 7005 + }, + { + "epoch": 0.7693828245113112, + "grad_norm": 2.4370005130767822, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7162940502166748, + "num_tokens": 180204457.0, + "step": 7006 + }, + { + "epoch": 0.7694926422139249, + "grad_norm": 2.0534775257110596, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7220681309700012, + "num_tokens": 180231337.0, + "step": 7007 + }, + { + "epoch": 0.7696024599165385, + "grad_norm": 2.256706953048706, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7028003931045532, + "num_tokens": 180256534.0, + "step": 7008 + }, + { + "epoch": 0.7697122776191522, + "grad_norm": 2.2079594135284424, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7194898128509521, + "num_tokens": 180281079.0, + "step": 7009 + }, + { + "epoch": 0.7698220953217658, + "grad_norm": 1.951513409614563, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6952570676803589, + "num_tokens": 180311818.0, + "step": 7010 + }, + { + "epoch": 0.7699319130243796, + "grad_norm": 2.162122964859009, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7029227614402771, + "num_tokens": 180336699.0, + "step": 7011 + }, + { + "epoch": 0.7700417307269932, + "grad_norm": 2.2383873462677, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.7015494108200073, + "num_tokens": 180362097.0, + "step": 7012 + }, + { + "epoch": 0.7701515484296069, + "grad_norm": 2.448174238204956, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7294194102287292, + "num_tokens": 180381628.0, + "step": 7013 + }, + { + "epoch": 0.7702613661322205, + "grad_norm": 2.3989129066467285, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7158890962600708, + "num_tokens": 180402915.0, + "step": 7014 + }, + { + "epoch": 0.7703711838348342, + "grad_norm": 2.0611019134521484, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6943111419677734, + "num_tokens": 180430338.0, + "step": 7015 + }, + { + "epoch": 0.7704810015374478, + "grad_norm": 2.127009868621826, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7108799815177917, + "num_tokens": 180457317.0, + "step": 7016 + }, + { + "epoch": 0.7705908192400615, + "grad_norm": 2.340418815612793, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7107195854187012, + "num_tokens": 180479347.0, + "step": 7017 + }, + { + "epoch": 0.7707006369426752, + "grad_norm": 2.3875720500946045, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7124430537223816, + "num_tokens": 180499310.0, + "step": 7018 + }, + { + "epoch": 0.7708104546452889, + "grad_norm": 2.0788393020629883, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6912089586257935, + "num_tokens": 180527523.0, + "step": 7019 + }, + { + "epoch": 0.7709202723479025, + "grad_norm": 2.0314013957977295, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6843407154083252, + "num_tokens": 180556858.0, + "step": 7020 + }, + { + "epoch": 0.7710300900505161, + "grad_norm": 2.055407762527466, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7196868062019348, + "num_tokens": 180582038.0, + "step": 7021 + }, + { + "epoch": 0.7711399077531298, + "grad_norm": 2.1785364151000977, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6904499530792236, + "num_tokens": 180609972.0, + "step": 7022 + }, + { + "epoch": 0.7712497254557434, + "grad_norm": 2.3890109062194824, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.7117800712585449, + "num_tokens": 180631540.0, + "step": 7023 + }, + { + "epoch": 0.7713595431583571, + "grad_norm": 2.260936975479126, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7196857333183289, + "num_tokens": 180655888.0, + "step": 7024 + }, + { + "epoch": 0.7714693608609708, + "grad_norm": 2.1273908615112305, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6968732476234436, + "num_tokens": 180684290.0, + "step": 7025 + }, + { + "epoch": 0.7715791785635845, + "grad_norm": 2.215701103210449, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7017902135848999, + "num_tokens": 180709346.0, + "step": 7026 + }, + { + "epoch": 0.7716889962661981, + "grad_norm": 2.349496841430664, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7095592021942139, + "num_tokens": 180735238.0, + "step": 7027 + }, + { + "epoch": 0.7717988139688118, + "grad_norm": 2.2837891578674316, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7106475830078125, + "num_tokens": 180758580.0, + "step": 7028 + }, + { + "epoch": 0.7719086316714254, + "grad_norm": 2.3537302017211914, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7219659090042114, + "num_tokens": 180783192.0, + "step": 7029 + }, + { + "epoch": 0.7720184493740391, + "grad_norm": 2.5548880100250244, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7101640701293945, + "num_tokens": 180803675.0, + "step": 7030 + }, + { + "epoch": 0.7721282670766527, + "grad_norm": 2.230693817138672, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7275890111923218, + "num_tokens": 180827837.0, + "step": 7031 + }, + { + "epoch": 0.7722380847792664, + "grad_norm": 2.3410799503326416, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6905110478401184, + "num_tokens": 180852103.0, + "step": 7032 + }, + { + "epoch": 0.7723479024818801, + "grad_norm": 2.4279024600982666, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7070643305778503, + "num_tokens": 180874483.0, + "step": 7033 + }, + { + "epoch": 0.7724577201844938, + "grad_norm": 2.1546037197113037, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6914332509040833, + "num_tokens": 180900919.0, + "step": 7034 + }, + { + "epoch": 0.7725675378871074, + "grad_norm": 2.417004108428955, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7277429103851318, + "num_tokens": 180921871.0, + "step": 7035 + }, + { + "epoch": 0.7726773555897211, + "grad_norm": 2.0126607418060303, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7225740551948547, + "num_tokens": 180949346.0, + "step": 7036 + }, + { + "epoch": 0.7727871732923347, + "grad_norm": 2.241706609725952, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6947797536849976, + "num_tokens": 180976287.0, + "step": 7037 + }, + { + "epoch": 0.7728969909949484, + "grad_norm": 2.526252269744873, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7116221189498901, + "num_tokens": 180997961.0, + "step": 7038 + }, + { + "epoch": 0.773006808697562, + "grad_norm": 2.2803709506988525, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7060006856918335, + "num_tokens": 181021488.0, + "step": 7039 + }, + { + "epoch": 0.7731166264001758, + "grad_norm": 2.3662943840026855, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6954845190048218, + "num_tokens": 181045470.0, + "step": 7040 + }, + { + "epoch": 0.7732264441027894, + "grad_norm": 2.243058204650879, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7017720937728882, + "num_tokens": 181070732.0, + "step": 7041 + }, + { + "epoch": 0.773336261805403, + "grad_norm": 2.0147314071655273, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7161253690719604, + "num_tokens": 181099338.0, + "step": 7042 + }, + { + "epoch": 0.7734460795080167, + "grad_norm": 2.208763599395752, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7205085158348083, + "num_tokens": 181122359.0, + "step": 7043 + }, + { + "epoch": 0.7735558972106303, + "grad_norm": 2.164630889892578, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7187790870666504, + "num_tokens": 181148813.0, + "step": 7044 + }, + { + "epoch": 0.773665714913244, + "grad_norm": 2.045922040939331, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6873716115951538, + "num_tokens": 181179333.0, + "step": 7045 + }, + { + "epoch": 0.7737755326158576, + "grad_norm": 2.0212409496307373, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7030255794525146, + "num_tokens": 181205932.0, + "step": 7046 + }, + { + "epoch": 0.7738853503184714, + "grad_norm": 2.0751330852508545, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7128584384918213, + "num_tokens": 181233497.0, + "step": 7047 + }, + { + "epoch": 0.773995168021085, + "grad_norm": 2.290341377258301, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6870057582855225, + "num_tokens": 181259218.0, + "step": 7048 + }, + { + "epoch": 0.7741049857236987, + "grad_norm": 2.2683379650115967, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7104820609092712, + "num_tokens": 181283958.0, + "step": 7049 + }, + { + "epoch": 0.7742148034263123, + "grad_norm": 2.2013161182403564, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7399063110351562, + "num_tokens": 181307699.0, + "step": 7050 + }, + { + "epoch": 0.774324621128926, + "grad_norm": 2.2412877082824707, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7109699249267578, + "num_tokens": 181332246.0, + "step": 7051 + }, + { + "epoch": 0.7744344388315396, + "grad_norm": 2.2175698280334473, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.718358039855957, + "num_tokens": 181357281.0, + "step": 7052 + }, + { + "epoch": 0.7745442565341533, + "grad_norm": 2.6680431365966797, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7165709137916565, + "num_tokens": 181375900.0, + "step": 7053 + }, + { + "epoch": 0.774654074236767, + "grad_norm": 2.279994010925293, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7128928303718567, + "num_tokens": 181400941.0, + "step": 7054 + }, + { + "epoch": 0.7747638919393807, + "grad_norm": 2.1782515048980713, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7086848020553589, + "num_tokens": 181427537.0, + "step": 7055 + }, + { + "epoch": 0.7748737096419943, + "grad_norm": 2.2192862033843994, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7091090083122253, + "num_tokens": 181453078.0, + "step": 7056 + }, + { + "epoch": 0.774983527344608, + "grad_norm": 2.3739452362060547, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7111879587173462, + "num_tokens": 181474850.0, + "step": 7057 + }, + { + "epoch": 0.7750933450472216, + "grad_norm": 2.238507032394409, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7039797306060791, + "num_tokens": 181500295.0, + "step": 7058 + }, + { + "epoch": 0.7752031627498353, + "grad_norm": 2.021798610687256, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6941030621528625, + "num_tokens": 181530832.0, + "step": 7059 + }, + { + "epoch": 0.7753129804524489, + "grad_norm": 2.534764051437378, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7156165838241577, + "num_tokens": 181551027.0, + "step": 7060 + }, + { + "epoch": 0.7754227981550625, + "grad_norm": 2.8279333114624023, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.731231689453125, + "num_tokens": 181569909.0, + "step": 7061 + }, + { + "epoch": 0.7755326158576763, + "grad_norm": 2.322171211242676, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7209310531616211, + "num_tokens": 181592091.0, + "step": 7062 + }, + { + "epoch": 0.77564243356029, + "grad_norm": 2.4821701049804688, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7354503870010376, + "num_tokens": 181610903.0, + "step": 7063 + }, + { + "epoch": 0.7757522512629036, + "grad_norm": 2.0164999961853027, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7088231444358826, + "num_tokens": 181638896.0, + "step": 7064 + }, + { + "epoch": 0.7758620689655172, + "grad_norm": 2.2179229259490967, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7048217058181763, + "num_tokens": 181663001.0, + "step": 7065 + }, + { + "epoch": 0.7759718866681309, + "grad_norm": 2.145738124847412, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7016001343727112, + "num_tokens": 181687288.0, + "step": 7066 + }, + { + "epoch": 0.7760817043707445, + "grad_norm": 2.341862201690674, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7336390614509583, + "num_tokens": 181709298.0, + "step": 7067 + }, + { + "epoch": 0.7761915220733582, + "grad_norm": 1.9770176410675049, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7004265189170837, + "num_tokens": 181739114.0, + "step": 7068 + }, + { + "epoch": 0.7763013397759719, + "grad_norm": 2.26912522315979, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7085011601448059, + "num_tokens": 181763680.0, + "step": 7069 + }, + { + "epoch": 0.7764111574785856, + "grad_norm": 2.42714786529541, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7202995419502258, + "num_tokens": 181783907.0, + "step": 7070 + }, + { + "epoch": 0.7765209751811992, + "grad_norm": 2.1691057682037354, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.712376058101654, + "num_tokens": 181808729.0, + "step": 7071 + }, + { + "epoch": 0.7766307928838129, + "grad_norm": 2.5915071964263916, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7078238129615784, + "num_tokens": 181829342.0, + "step": 7072 + }, + { + "epoch": 0.7767406105864265, + "grad_norm": 2.3688113689422607, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6899809837341309, + "num_tokens": 181852709.0, + "step": 7073 + }, + { + "epoch": 0.7768504282890402, + "grad_norm": 2.182539224624634, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7196642756462097, + "num_tokens": 181877296.0, + "step": 7074 + }, + { + "epoch": 0.7769602459916538, + "grad_norm": 2.3026368618011475, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7166576981544495, + "num_tokens": 181899831.0, + "step": 7075 + }, + { + "epoch": 0.7770700636942676, + "grad_norm": 2.023632526397705, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7225231528282166, + "num_tokens": 181927798.0, + "step": 7076 + }, + { + "epoch": 0.7771798813968812, + "grad_norm": 2.116267442703247, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6955029964447021, + "num_tokens": 181953387.0, + "step": 7077 + }, + { + "epoch": 0.7772896990994949, + "grad_norm": 1.941392183303833, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6895512342453003, + "num_tokens": 181985939.0, + "step": 7078 + }, + { + "epoch": 0.7773995168021085, + "grad_norm": 2.2320828437805176, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7233783006668091, + "num_tokens": 182008423.0, + "step": 7079 + }, + { + "epoch": 0.7775093345047221, + "grad_norm": 1.9795526266098022, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6924988031387329, + "num_tokens": 182038627.0, + "step": 7080 + }, + { + "epoch": 0.7776191522073358, + "grad_norm": 2.0433802604675293, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7104390859603882, + "num_tokens": 182065996.0, + "step": 7081 + }, + { + "epoch": 0.7777289699099494, + "grad_norm": 2.445894241333008, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7030302882194519, + "num_tokens": 182087091.0, + "step": 7082 + }, + { + "epoch": 0.7778387876125632, + "grad_norm": 1.8692914247512817, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7078907489776611, + "num_tokens": 182118505.0, + "step": 7083 + }, + { + "epoch": 0.7779486053151768, + "grad_norm": 2.139469623565674, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6923707723617554, + "num_tokens": 182145755.0, + "step": 7084 + }, + { + "epoch": 0.7780584230177905, + "grad_norm": 2.738396167755127, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7203001976013184, + "num_tokens": 182162777.0, + "step": 7085 + }, + { + "epoch": 0.7781682407204041, + "grad_norm": 2.268404245376587, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7115199565887451, + "num_tokens": 182186931.0, + "step": 7086 + }, + { + "epoch": 0.7782780584230178, + "grad_norm": 2.3127379417419434, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7026844024658203, + "num_tokens": 182211158.0, + "step": 7087 + }, + { + "epoch": 0.7783878761256314, + "grad_norm": 2.008673906326294, + "learning_rate": 1e-06, + "loss": 1.0935, + "mean_token_accuracy": 0.6793354749679565, + "num_tokens": 182242126.0, + "step": 7088 + }, + { + "epoch": 0.7784976938282451, + "grad_norm": 2.4208590984344482, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7109073400497437, + "num_tokens": 182263766.0, + "step": 7089 + }, + { + "epoch": 0.7786075115308587, + "grad_norm": 2.461423397064209, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7316581010818481, + "num_tokens": 182282183.0, + "step": 7090 + }, + { + "epoch": 0.7787173292334725, + "grad_norm": 2.4529924392700195, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7009273767471313, + "num_tokens": 182304675.0, + "step": 7091 + }, + { + "epoch": 0.7788271469360861, + "grad_norm": 2.1047582626342773, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6894984245300293, + "num_tokens": 182333393.0, + "step": 7092 + }, + { + "epoch": 0.7789369646386998, + "grad_norm": 1.9180517196655273, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6827998757362366, + "num_tokens": 182367549.0, + "step": 7093 + }, + { + "epoch": 0.7790467823413134, + "grad_norm": 2.4425833225250244, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.688457727432251, + "num_tokens": 182388611.0, + "step": 7094 + }, + { + "epoch": 0.7791566000439271, + "grad_norm": 2.4273734092712402, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7175690531730652, + "num_tokens": 182410341.0, + "step": 7095 + }, + { + "epoch": 0.7792664177465407, + "grad_norm": 2.3536477088928223, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.69502854347229, + "num_tokens": 182435528.0, + "step": 7096 + }, + { + "epoch": 0.7793762354491544, + "grad_norm": 2.3221895694732666, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.707554817199707, + "num_tokens": 182457437.0, + "step": 7097 + }, + { + "epoch": 0.7794860531517681, + "grad_norm": 2.2206711769104004, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7211216688156128, + "num_tokens": 182481889.0, + "step": 7098 + }, + { + "epoch": 0.7795958708543818, + "grad_norm": 2.004892587661743, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7028869390487671, + "num_tokens": 182512404.0, + "step": 7099 + }, + { + "epoch": 0.7797056885569954, + "grad_norm": 2.1474533081054688, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7130699157714844, + "num_tokens": 182539287.0, + "step": 7100 + }, + { + "epoch": 0.779815506259609, + "grad_norm": 2.279237747192383, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6924469470977783, + "num_tokens": 182566221.0, + "step": 7101 + }, + { + "epoch": 0.7799253239622227, + "grad_norm": 2.082270383834839, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7137011289596558, + "num_tokens": 182594185.0, + "step": 7102 + }, + { + "epoch": 0.7800351416648363, + "grad_norm": 2.297257661819458, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7122998833656311, + "num_tokens": 182617049.0, + "step": 7103 + }, + { + "epoch": 0.78014495936745, + "grad_norm": 2.1301467418670654, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7025097608566284, + "num_tokens": 182643193.0, + "step": 7104 + }, + { + "epoch": 0.7802547770700637, + "grad_norm": 2.063589096069336, + "learning_rate": 1e-06, + "loss": 1.0778, + "mean_token_accuracy": 0.6755269765853882, + "num_tokens": 182673507.0, + "step": 7105 + }, + { + "epoch": 0.7803645947726774, + "grad_norm": 1.9021893739700317, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7073430418968201, + "num_tokens": 182705142.0, + "step": 7106 + }, + { + "epoch": 0.780474412475291, + "grad_norm": 2.2258172035217285, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7029760479927063, + "num_tokens": 182730129.0, + "step": 7107 + }, + { + "epoch": 0.7805842301779047, + "grad_norm": 1.9834308624267578, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7142657041549683, + "num_tokens": 182755496.0, + "step": 7108 + }, + { + "epoch": 0.7806940478805183, + "grad_norm": 2.0437140464782715, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.715152382850647, + "num_tokens": 182781911.0, + "step": 7109 + }, + { + "epoch": 0.780803865583132, + "grad_norm": 2.291827917098999, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7129870653152466, + "num_tokens": 182804657.0, + "step": 7110 + }, + { + "epoch": 0.7809136832857456, + "grad_norm": 2.185878276824951, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6929377317428589, + "num_tokens": 182830263.0, + "step": 7111 + }, + { + "epoch": 0.7810235009883594, + "grad_norm": 1.9610528945922852, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7017468214035034, + "num_tokens": 182860168.0, + "step": 7112 + }, + { + "epoch": 0.781133318690973, + "grad_norm": 2.1847052574157715, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7179431319236755, + "num_tokens": 182886171.0, + "step": 7113 + }, + { + "epoch": 0.7812431363935867, + "grad_norm": 2.259385347366333, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7194772958755493, + "num_tokens": 182909204.0, + "step": 7114 + }, + { + "epoch": 0.7813529540962003, + "grad_norm": 2.282554864883423, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7097432017326355, + "num_tokens": 182932572.0, + "step": 7115 + }, + { + "epoch": 0.781462771798814, + "grad_norm": 2.1720690727233887, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7097282409667969, + "num_tokens": 182957230.0, + "step": 7116 + }, + { + "epoch": 0.7815725895014276, + "grad_norm": 2.144679546356201, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6997635364532471, + "num_tokens": 182984363.0, + "step": 7117 + }, + { + "epoch": 0.7816824072040413, + "grad_norm": 2.416501998901367, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7169049978256226, + "num_tokens": 183005604.0, + "step": 7118 + }, + { + "epoch": 0.7817922249066549, + "grad_norm": 2.0869522094726562, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.705409824848175, + "num_tokens": 183034151.0, + "step": 7119 + }, + { + "epoch": 0.7819020426092687, + "grad_norm": 2.1282436847686768, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7078964710235596, + "num_tokens": 183060686.0, + "step": 7120 + }, + { + "epoch": 0.7820118603118823, + "grad_norm": 2.2283387184143066, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.700657844543457, + "num_tokens": 183085891.0, + "step": 7121 + }, + { + "epoch": 0.782121678014496, + "grad_norm": 2.3792552947998047, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6973791122436523, + "num_tokens": 183109447.0, + "step": 7122 + }, + { + "epoch": 0.7822314957171096, + "grad_norm": 2.0537478923797607, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6997584104537964, + "num_tokens": 183139948.0, + "step": 7123 + }, + { + "epoch": 0.7823413134197232, + "grad_norm": 2.0771329402923584, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.7030719518661499, + "num_tokens": 183168167.0, + "step": 7124 + }, + { + "epoch": 0.7824511311223369, + "grad_norm": 2.0273525714874268, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6869081258773804, + "num_tokens": 183198188.0, + "step": 7125 + }, + { + "epoch": 0.7825609488249505, + "grad_norm": 1.97701895236969, + "learning_rate": 1e-06, + "loss": 1.0756, + "mean_token_accuracy": 0.6745938658714294, + "num_tokens": 183226998.0, + "step": 7126 + }, + { + "epoch": 0.7826707665275643, + "grad_norm": 2.3080084323883057, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7316403388977051, + "num_tokens": 183248034.0, + "step": 7127 + }, + { + "epoch": 0.7827805842301779, + "grad_norm": 2.135640859603882, + "learning_rate": 1e-06, + "loss": 1.0818, + "mean_token_accuracy": 0.6735338568687439, + "num_tokens": 183275220.0, + "step": 7128 + }, + { + "epoch": 0.7828904019327916, + "grad_norm": 2.201064109802246, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6954501867294312, + "num_tokens": 183300413.0, + "step": 7129 + }, + { + "epoch": 0.7830002196354052, + "grad_norm": 2.229762315750122, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7341428995132446, + "num_tokens": 183322235.0, + "step": 7130 + }, + { + "epoch": 0.7831100373380189, + "grad_norm": 2.150493860244751, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6930263042449951, + "num_tokens": 183349578.0, + "step": 7131 + }, + { + "epoch": 0.7832198550406325, + "grad_norm": 2.1996726989746094, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7048448324203491, + "num_tokens": 183375990.0, + "step": 7132 + }, + { + "epoch": 0.7833296727432462, + "grad_norm": 2.056818962097168, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6981920003890991, + "num_tokens": 183404588.0, + "step": 7133 + }, + { + "epoch": 0.7834394904458599, + "grad_norm": 2.2597458362579346, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.708507776260376, + "num_tokens": 183430217.0, + "step": 7134 + }, + { + "epoch": 0.7835493081484736, + "grad_norm": 2.3539886474609375, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7163830399513245, + "num_tokens": 183451357.0, + "step": 7135 + }, + { + "epoch": 0.7836591258510872, + "grad_norm": 2.3468399047851562, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6874474287033081, + "num_tokens": 183472731.0, + "step": 7136 + }, + { + "epoch": 0.7837689435537009, + "grad_norm": 2.0345559120178223, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6890519857406616, + "num_tokens": 183502888.0, + "step": 7137 + }, + { + "epoch": 0.7838787612563145, + "grad_norm": 2.616058349609375, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.707301676273346, + "num_tokens": 183525535.0, + "step": 7138 + }, + { + "epoch": 0.7839885789589282, + "grad_norm": 2.3502085208892822, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7375450134277344, + "num_tokens": 183545791.0, + "step": 7139 + }, + { + "epoch": 0.7840983966615418, + "grad_norm": 2.1282896995544434, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7304161787033081, + "num_tokens": 183571316.0, + "step": 7140 + }, + { + "epoch": 0.7842082143641556, + "grad_norm": 1.937251091003418, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6906388401985168, + "num_tokens": 183603811.0, + "step": 7141 + }, + { + "epoch": 0.7843180320667692, + "grad_norm": 2.0889153480529785, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7074426412582397, + "num_tokens": 183630098.0, + "step": 7142 + }, + { + "epoch": 0.7844278497693828, + "grad_norm": 2.136430263519287, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6955417990684509, + "num_tokens": 183657746.0, + "step": 7143 + }, + { + "epoch": 0.7845376674719965, + "grad_norm": 2.1797988414764404, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7107528448104858, + "num_tokens": 183682437.0, + "step": 7144 + }, + { + "epoch": 0.7846474851746101, + "grad_norm": 2.016045331954956, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7051569819450378, + "num_tokens": 183709765.0, + "step": 7145 + }, + { + "epoch": 0.7847573028772238, + "grad_norm": 2.3942556381225586, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7038856744766235, + "num_tokens": 183732461.0, + "step": 7146 + }, + { + "epoch": 0.7848671205798374, + "grad_norm": 2.4265947341918945, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7133049964904785, + "num_tokens": 183757137.0, + "step": 7147 + }, + { + "epoch": 0.7849769382824512, + "grad_norm": 2.311391830444336, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7093566060066223, + "num_tokens": 183781199.0, + "step": 7148 + }, + { + "epoch": 0.7850867559850648, + "grad_norm": 1.843151330947876, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7038068771362305, + "num_tokens": 183813495.0, + "step": 7149 + }, + { + "epoch": 0.7851965736876785, + "grad_norm": 2.120692014694214, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7023043632507324, + "num_tokens": 183841031.0, + "step": 7150 + }, + { + "epoch": 0.7853063913902921, + "grad_norm": 2.193685531616211, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6850841641426086, + "num_tokens": 183866755.0, + "step": 7151 + }, + { + "epoch": 0.7854162090929058, + "grad_norm": 2.3352067470550537, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7201905846595764, + "num_tokens": 183888869.0, + "step": 7152 + }, + { + "epoch": 0.7855260267955194, + "grad_norm": 1.939044713973999, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6987204551696777, + "num_tokens": 183920734.0, + "step": 7153 + }, + { + "epoch": 0.7856358444981331, + "grad_norm": 2.1414356231689453, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7037827968597412, + "num_tokens": 183948576.0, + "step": 7154 + }, + { + "epoch": 0.7857456622007467, + "grad_norm": 2.442891836166382, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6913650631904602, + "num_tokens": 183971162.0, + "step": 7155 + }, + { + "epoch": 0.7858554799033605, + "grad_norm": 2.214066505432129, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7149550318717957, + "num_tokens": 183996055.0, + "step": 7156 + }, + { + "epoch": 0.7859652976059741, + "grad_norm": 2.21732497215271, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7017017006874084, + "num_tokens": 184022351.0, + "step": 7157 + }, + { + "epoch": 0.7860751153085878, + "grad_norm": 2.1261403560638428, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7421293258666992, + "num_tokens": 184045207.0, + "step": 7158 + }, + { + "epoch": 0.7861849330112014, + "grad_norm": 2.622124671936035, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7018223404884338, + "num_tokens": 184065143.0, + "step": 7159 + }, + { + "epoch": 0.786294750713815, + "grad_norm": 2.3964345455169678, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7223889827728271, + "num_tokens": 184087928.0, + "step": 7160 + }, + { + "epoch": 0.7864045684164287, + "grad_norm": 2.1465904712677, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6957336664199829, + "num_tokens": 184113442.0, + "step": 7161 + }, + { + "epoch": 0.7865143861190423, + "grad_norm": 2.165255069732666, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.73419189453125, + "num_tokens": 184137408.0, + "step": 7162 + }, + { + "epoch": 0.7866242038216561, + "grad_norm": 2.313453435897827, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.683946967124939, + "num_tokens": 184162833.0, + "step": 7163 + }, + { + "epoch": 0.7867340215242697, + "grad_norm": 2.085197925567627, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6984291076660156, + "num_tokens": 184192727.0, + "step": 7164 + }, + { + "epoch": 0.7868438392268834, + "grad_norm": 2.0718729496002197, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7098808288574219, + "num_tokens": 184220403.0, + "step": 7165 + }, + { + "epoch": 0.786953656929497, + "grad_norm": 2.0506017208099365, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7243912220001221, + "num_tokens": 184245340.0, + "step": 7166 + }, + { + "epoch": 0.7870634746321107, + "grad_norm": 2.457367181777954, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7061748504638672, + "num_tokens": 184268670.0, + "step": 7167 + }, + { + "epoch": 0.7871732923347243, + "grad_norm": 2.36716890335083, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7279932498931885, + "num_tokens": 184290039.0, + "step": 7168 + }, + { + "epoch": 0.787283110037338, + "grad_norm": 1.8840417861938477, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7031957507133484, + "num_tokens": 184321944.0, + "step": 7169 + }, + { + "epoch": 0.7873929277399517, + "grad_norm": 2.1486504077911377, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7088119983673096, + "num_tokens": 184348866.0, + "step": 7170 + }, + { + "epoch": 0.7875027454425654, + "grad_norm": 2.247213125228882, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7128639221191406, + "num_tokens": 184372002.0, + "step": 7171 + }, + { + "epoch": 0.787612563145179, + "grad_norm": 2.4010426998138428, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7122917175292969, + "num_tokens": 184393056.0, + "step": 7172 + }, + { + "epoch": 0.7877223808477927, + "grad_norm": 2.0653836727142334, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.702485203742981, + "num_tokens": 184420945.0, + "step": 7173 + }, + { + "epoch": 0.7878321985504063, + "grad_norm": 2.2274510860443115, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7032911777496338, + "num_tokens": 184446874.0, + "step": 7174 + }, + { + "epoch": 0.78794201625302, + "grad_norm": 2.576972246170044, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7001975178718567, + "num_tokens": 184468130.0, + "step": 7175 + }, + { + "epoch": 0.7880518339556336, + "grad_norm": 2.4387052059173584, + "learning_rate": 1e-06, + "loss": 1.0865, + "mean_token_accuracy": 0.6702371835708618, + "num_tokens": 184492219.0, + "step": 7176 + }, + { + "epoch": 0.7881616516582474, + "grad_norm": 2.247030735015869, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.7006390690803528, + "num_tokens": 184516201.0, + "step": 7177 + }, + { + "epoch": 0.788271469360861, + "grad_norm": 2.1866848468780518, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7119460105895996, + "num_tokens": 184541905.0, + "step": 7178 + }, + { + "epoch": 0.7883812870634747, + "grad_norm": 2.2324280738830566, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6965477466583252, + "num_tokens": 184569073.0, + "step": 7179 + }, + { + "epoch": 0.7884911047660883, + "grad_norm": 2.372628927230835, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7103474140167236, + "num_tokens": 184591955.0, + "step": 7180 + }, + { + "epoch": 0.788600922468702, + "grad_norm": 2.088775634765625, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7299795746803284, + "num_tokens": 184616601.0, + "step": 7181 + }, + { + "epoch": 0.7887107401713156, + "grad_norm": 2.2082080841064453, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7128621935844421, + "num_tokens": 184640950.0, + "step": 7182 + }, + { + "epoch": 0.7888205578739292, + "grad_norm": 2.321232318878174, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7040334343910217, + "num_tokens": 184663534.0, + "step": 7183 + }, + { + "epoch": 0.7889303755765429, + "grad_norm": 2.4634010791778564, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7125831842422485, + "num_tokens": 184685870.0, + "step": 7184 + }, + { + "epoch": 0.7890401932791566, + "grad_norm": 2.324191093444824, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6959915161132812, + "num_tokens": 184710369.0, + "step": 7185 + }, + { + "epoch": 0.7891500109817703, + "grad_norm": 1.9059869050979614, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.701607346534729, + "num_tokens": 184741618.0, + "step": 7186 + }, + { + "epoch": 0.7892598286843839, + "grad_norm": 2.054791212081909, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.711759090423584, + "num_tokens": 184768856.0, + "step": 7187 + }, + { + "epoch": 0.7893696463869976, + "grad_norm": 2.4818272590637207, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.713044285774231, + "num_tokens": 184789445.0, + "step": 7188 + }, + { + "epoch": 0.7894794640896112, + "grad_norm": 2.29764461517334, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7154929637908936, + "num_tokens": 184811974.0, + "step": 7189 + }, + { + "epoch": 0.7895892817922249, + "grad_norm": 2.226576805114746, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6932268738746643, + "num_tokens": 184837196.0, + "step": 7190 + }, + { + "epoch": 0.7896990994948385, + "grad_norm": 2.38010573387146, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7061589360237122, + "num_tokens": 184859635.0, + "step": 7191 + }, + { + "epoch": 0.7898089171974523, + "grad_norm": 1.9912099838256836, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7088020443916321, + "num_tokens": 184890408.0, + "step": 7192 + }, + { + "epoch": 0.7899187349000659, + "grad_norm": 1.9920109510421753, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7055027484893799, + "num_tokens": 184919118.0, + "step": 7193 + }, + { + "epoch": 0.7900285526026796, + "grad_norm": 2.185788631439209, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7257888317108154, + "num_tokens": 184943048.0, + "step": 7194 + }, + { + "epoch": 0.7901383703052932, + "grad_norm": 1.9746333360671997, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7013214826583862, + "num_tokens": 184974665.0, + "step": 7195 + }, + { + "epoch": 0.7902481880079069, + "grad_norm": 2.4031105041503906, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7254120707511902, + "num_tokens": 184996139.0, + "step": 7196 + }, + { + "epoch": 0.7903580057105205, + "grad_norm": 1.9521796703338623, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7110324501991272, + "num_tokens": 185027083.0, + "step": 7197 + }, + { + "epoch": 0.7904678234131342, + "grad_norm": 2.301851749420166, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7177817821502686, + "num_tokens": 185049507.0, + "step": 7198 + }, + { + "epoch": 0.7905776411157479, + "grad_norm": 2.1462795734405518, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7178442478179932, + "num_tokens": 185074873.0, + "step": 7199 + }, + { + "epoch": 0.7906874588183616, + "grad_norm": 2.2254672050476074, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7252294421195984, + "num_tokens": 185098802.0, + "step": 7200 + }, + { + "epoch": 0.7907972765209752, + "grad_norm": 2.321606397628784, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7279053926467896, + "num_tokens": 185121838.0, + "step": 7201 + }, + { + "epoch": 0.7909070942235888, + "grad_norm": 2.066895008087158, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7133654356002808, + "num_tokens": 185147527.0, + "step": 7202 + }, + { + "epoch": 0.7910169119262025, + "grad_norm": 2.0521199703216553, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.719586193561554, + "num_tokens": 185173173.0, + "step": 7203 + }, + { + "epoch": 0.7911267296288161, + "grad_norm": 2.2967419624328613, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7109373807907104, + "num_tokens": 185196929.0, + "step": 7204 + }, + { + "epoch": 0.7912365473314298, + "grad_norm": 2.284212112426758, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6928017139434814, + "num_tokens": 185220704.0, + "step": 7205 + }, + { + "epoch": 0.7913463650340435, + "grad_norm": 2.068190336227417, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7231195569038391, + "num_tokens": 185245866.0, + "step": 7206 + }, + { + "epoch": 0.7914561827366572, + "grad_norm": 2.214318037033081, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.688831090927124, + "num_tokens": 185272144.0, + "step": 7207 + }, + { + "epoch": 0.7915660004392708, + "grad_norm": 2.2631006240844727, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7135570645332336, + "num_tokens": 185297156.0, + "step": 7208 + }, + { + "epoch": 0.7916758181418845, + "grad_norm": 2.251265525817871, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.69222092628479, + "num_tokens": 185322568.0, + "step": 7209 + }, + { + "epoch": 0.7917856358444981, + "grad_norm": 2.3638112545013428, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7251702547073364, + "num_tokens": 185343885.0, + "step": 7210 + }, + { + "epoch": 0.7918954535471118, + "grad_norm": 2.212965965270996, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.72213214635849, + "num_tokens": 185368648.0, + "step": 7211 + }, + { + "epoch": 0.7920052712497254, + "grad_norm": 2.245337963104248, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7197039127349854, + "num_tokens": 185391969.0, + "step": 7212 + }, + { + "epoch": 0.7921150889523391, + "grad_norm": 2.012722969055176, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7193640470504761, + "num_tokens": 185419107.0, + "step": 7213 + }, + { + "epoch": 0.7922249066549528, + "grad_norm": 2.087859630584717, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7055706977844238, + "num_tokens": 185445389.0, + "step": 7214 + }, + { + "epoch": 0.7923347243575665, + "grad_norm": 2.405958890914917, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7148857712745667, + "num_tokens": 185466143.0, + "step": 7215 + }, + { + "epoch": 0.7924445420601801, + "grad_norm": 2.3663129806518555, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6893389225006104, + "num_tokens": 185488719.0, + "step": 7216 + }, + { + "epoch": 0.7925543597627938, + "grad_norm": 1.932633399963379, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7164143323898315, + "num_tokens": 185517154.0, + "step": 7217 + }, + { + "epoch": 0.7926641774654074, + "grad_norm": 2.2164435386657715, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6972830295562744, + "num_tokens": 185541387.0, + "step": 7218 + }, + { + "epoch": 0.792773995168021, + "grad_norm": 1.863060474395752, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6922681331634521, + "num_tokens": 185576200.0, + "step": 7219 + }, + { + "epoch": 0.7928838128706347, + "grad_norm": 2.065035581588745, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6945353746414185, + "num_tokens": 185604060.0, + "step": 7220 + }, + { + "epoch": 0.7929936305732485, + "grad_norm": 1.9857735633850098, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7113087177276611, + "num_tokens": 185633657.0, + "step": 7221 + }, + { + "epoch": 0.7931034482758621, + "grad_norm": 2.0392367839813232, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7043648362159729, + "num_tokens": 185664320.0, + "step": 7222 + }, + { + "epoch": 0.7932132659784757, + "grad_norm": 2.572394371032715, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7169057130813599, + "num_tokens": 185684715.0, + "step": 7223 + }, + { + "epoch": 0.7933230836810894, + "grad_norm": 2.059887409210205, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6858077049255371, + "num_tokens": 185715904.0, + "step": 7224 + }, + { + "epoch": 0.793432901383703, + "grad_norm": 2.307739496231079, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7177420258522034, + "num_tokens": 185738179.0, + "step": 7225 + }, + { + "epoch": 0.7935427190863167, + "grad_norm": 2.5191612243652344, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7141324281692505, + "num_tokens": 185757847.0, + "step": 7226 + }, + { + "epoch": 0.7936525367889303, + "grad_norm": 2.237204074859619, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7024219632148743, + "num_tokens": 185783368.0, + "step": 7227 + }, + { + "epoch": 0.7937623544915441, + "grad_norm": 2.287508487701416, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.705691933631897, + "num_tokens": 185809209.0, + "step": 7228 + }, + { + "epoch": 0.7938721721941577, + "grad_norm": 2.1955459117889404, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7101538181304932, + "num_tokens": 185832940.0, + "step": 7229 + }, + { + "epoch": 0.7939819898967714, + "grad_norm": 2.1756751537323, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6907780170440674, + "num_tokens": 185858805.0, + "step": 7230 + }, + { + "epoch": 0.794091807599385, + "grad_norm": 2.4305808544158936, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7281293869018555, + "num_tokens": 185879494.0, + "step": 7231 + }, + { + "epoch": 0.7942016253019987, + "grad_norm": 2.0451881885528564, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.7063392996788025, + "num_tokens": 185908712.0, + "step": 7232 + }, + { + "epoch": 0.7943114430046123, + "grad_norm": 2.260956048965454, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7248740196228027, + "num_tokens": 185932874.0, + "step": 7233 + }, + { + "epoch": 0.794421260707226, + "grad_norm": 2.2626917362213135, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.6987649202346802, + "num_tokens": 185958626.0, + "step": 7234 + }, + { + "epoch": 0.7945310784098397, + "grad_norm": 2.1340742111206055, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7003028392791748, + "num_tokens": 185985750.0, + "step": 7235 + }, + { + "epoch": 0.7946408961124534, + "grad_norm": 2.2024576663970947, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7230775356292725, + "num_tokens": 186014094.0, + "step": 7236 + }, + { + "epoch": 0.794750713815067, + "grad_norm": 2.4443507194519043, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7171415090560913, + "num_tokens": 186037303.0, + "step": 7237 + }, + { + "epoch": 0.7948605315176807, + "grad_norm": 1.9569158554077148, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7032477259635925, + "num_tokens": 186067861.0, + "step": 7238 + }, + { + "epoch": 0.7949703492202943, + "grad_norm": 1.9981640577316284, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7187085151672363, + "num_tokens": 186097518.0, + "step": 7239 + }, + { + "epoch": 0.795080166922908, + "grad_norm": 2.1479578018188477, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7088624835014343, + "num_tokens": 186125000.0, + "step": 7240 + }, + { + "epoch": 0.7951899846255216, + "grad_norm": 2.5947470664978027, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7358949184417725, + "num_tokens": 186142575.0, + "step": 7241 + }, + { + "epoch": 0.7952998023281352, + "grad_norm": 2.250822067260742, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.722578227519989, + "num_tokens": 186164441.0, + "step": 7242 + }, + { + "epoch": 0.795409620030749, + "grad_norm": 2.2828152179718018, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7004574537277222, + "num_tokens": 186187939.0, + "step": 7243 + }, + { + "epoch": 0.7955194377333626, + "grad_norm": 2.379526138305664, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6753755807876587, + "num_tokens": 186210776.0, + "step": 7244 + }, + { + "epoch": 0.7956292554359763, + "grad_norm": 2.225189447402954, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7150337100028992, + "num_tokens": 186236374.0, + "step": 7245 + }, + { + "epoch": 0.7957390731385899, + "grad_norm": 2.133652448654175, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7117272615432739, + "num_tokens": 186261192.0, + "step": 7246 + }, + { + "epoch": 0.7958488908412036, + "grad_norm": 2.00862455368042, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6981383562088013, + "num_tokens": 186289945.0, + "step": 7247 + }, + { + "epoch": 0.7959587085438172, + "grad_norm": 2.1647090911865234, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7269383668899536, + "num_tokens": 186314277.0, + "step": 7248 + }, + { + "epoch": 0.7960685262464309, + "grad_norm": 2.3070826530456543, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7119622826576233, + "num_tokens": 186337148.0, + "step": 7249 + }, + { + "epoch": 0.7961783439490446, + "grad_norm": 2.0620899200439453, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.701947033405304, + "num_tokens": 186364304.0, + "step": 7250 + }, + { + "epoch": 0.7962881616516583, + "grad_norm": 2.0126490592956543, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6953666806221008, + "num_tokens": 186394720.0, + "step": 7251 + }, + { + "epoch": 0.7963979793542719, + "grad_norm": 2.1110081672668457, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6995370388031006, + "num_tokens": 186421899.0, + "step": 7252 + }, + { + "epoch": 0.7965077970568856, + "grad_norm": 1.9604822397232056, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7017067670822144, + "num_tokens": 186451862.0, + "step": 7253 + }, + { + "epoch": 0.7966176147594992, + "grad_norm": 2.218780755996704, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6859791278839111, + "num_tokens": 186476861.0, + "step": 7254 + }, + { + "epoch": 0.7967274324621129, + "grad_norm": 2.5639097690582275, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7067720293998718, + "num_tokens": 186496103.0, + "step": 7255 + }, + { + "epoch": 0.7968372501647265, + "grad_norm": 2.0850470066070557, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7039971351623535, + "num_tokens": 186523769.0, + "step": 7256 + }, + { + "epoch": 0.7969470678673403, + "grad_norm": 2.1653966903686523, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7417522072792053, + "num_tokens": 186547365.0, + "step": 7257 + }, + { + "epoch": 0.7970568855699539, + "grad_norm": 1.79552161693573, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6933192014694214, + "num_tokens": 186582953.0, + "step": 7258 + }, + { + "epoch": 0.7971667032725676, + "grad_norm": 2.080350160598755, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6955248117446899, + "num_tokens": 186611433.0, + "step": 7259 + }, + { + "epoch": 0.7972765209751812, + "grad_norm": 2.1782655715942383, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7070143818855286, + "num_tokens": 186637237.0, + "step": 7260 + }, + { + "epoch": 0.7973863386777948, + "grad_norm": 2.2233219146728516, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7292431592941284, + "num_tokens": 186662484.0, + "step": 7261 + }, + { + "epoch": 0.7974961563804085, + "grad_norm": 2.1449196338653564, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6976750493049622, + "num_tokens": 186688607.0, + "step": 7262 + }, + { + "epoch": 0.7976059740830221, + "grad_norm": 2.190849781036377, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6778554320335388, + "num_tokens": 186715847.0, + "step": 7263 + }, + { + "epoch": 0.7977157917856359, + "grad_norm": 1.9576133489608765, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7087711095809937, + "num_tokens": 186746250.0, + "step": 7264 + }, + { + "epoch": 0.7978256094882495, + "grad_norm": 1.9505388736724854, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7147095203399658, + "num_tokens": 186775590.0, + "step": 7265 + }, + { + "epoch": 0.7979354271908632, + "grad_norm": 2.346329927444458, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7322256565093994, + "num_tokens": 186796982.0, + "step": 7266 + }, + { + "epoch": 0.7980452448934768, + "grad_norm": 2.1086339950561523, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6972631812095642, + "num_tokens": 186823359.0, + "step": 7267 + }, + { + "epoch": 0.7981550625960905, + "grad_norm": 2.236065149307251, + "learning_rate": 1e-06, + "loss": 1.1044, + "mean_token_accuracy": 0.6639857888221741, + "num_tokens": 186850306.0, + "step": 7268 + }, + { + "epoch": 0.7982648802987041, + "grad_norm": 2.6376736164093018, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7163131833076477, + "num_tokens": 186869022.0, + "step": 7269 + }, + { + "epoch": 0.7983746980013178, + "grad_norm": 2.1141674518585205, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.6974043250083923, + "num_tokens": 186894943.0, + "step": 7270 + }, + { + "epoch": 0.7984845157039314, + "grad_norm": 2.5438570976257324, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7258715033531189, + "num_tokens": 186913922.0, + "step": 7271 + }, + { + "epoch": 0.7985943334065452, + "grad_norm": 2.489048957824707, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7390581369400024, + "num_tokens": 186933285.0, + "step": 7272 + }, + { + "epoch": 0.7987041511091588, + "grad_norm": 2.089641571044922, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7184876203536987, + "num_tokens": 186960468.0, + "step": 7273 + }, + { + "epoch": 0.7988139688117725, + "grad_norm": 2.336663007736206, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7283931970596313, + "num_tokens": 186981810.0, + "step": 7274 + }, + { + "epoch": 0.7989237865143861, + "grad_norm": 2.39005446434021, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7275623083114624, + "num_tokens": 187002292.0, + "step": 7275 + }, + { + "epoch": 0.7990336042169998, + "grad_norm": 2.0444271564483643, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6851160526275635, + "num_tokens": 187031761.0, + "step": 7276 + }, + { + "epoch": 0.7991434219196134, + "grad_norm": 2.0519254207611084, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7121241092681885, + "num_tokens": 187058076.0, + "step": 7277 + }, + { + "epoch": 0.799253239622227, + "grad_norm": 2.237947940826416, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7171852588653564, + "num_tokens": 187081374.0, + "step": 7278 + }, + { + "epoch": 0.7993630573248408, + "grad_norm": 2.1888976097106934, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6915218234062195, + "num_tokens": 187107989.0, + "step": 7279 + }, + { + "epoch": 0.7994728750274545, + "grad_norm": 2.257044792175293, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7014855146408081, + "num_tokens": 187130671.0, + "step": 7280 + }, + { + "epoch": 0.7995826927300681, + "grad_norm": 2.231813907623291, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7242788672447205, + "num_tokens": 187152700.0, + "step": 7281 + }, + { + "epoch": 0.7996925104326817, + "grad_norm": 2.0915043354034424, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.732393741607666, + "num_tokens": 187179404.0, + "step": 7282 + }, + { + "epoch": 0.7998023281352954, + "grad_norm": 2.0855960845947266, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7189706563949585, + "num_tokens": 187207064.0, + "step": 7283 + }, + { + "epoch": 0.799912145837909, + "grad_norm": 2.118461847305298, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7061396241188049, + "num_tokens": 187233716.0, + "step": 7284 + }, + { + "epoch": 0.8000219635405227, + "grad_norm": 1.9850257635116577, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6960536241531372, + "num_tokens": 187263135.0, + "step": 7285 + }, + { + "epoch": 0.8001317812431364, + "grad_norm": 2.264625310897827, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7001581192016602, + "num_tokens": 187287227.0, + "step": 7286 + }, + { + "epoch": 0.8002415989457501, + "grad_norm": 2.159989833831787, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.7010880708694458, + "num_tokens": 187312252.0, + "step": 7287 + }, + { + "epoch": 0.8003514166483637, + "grad_norm": 2.533604860305786, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7116785049438477, + "num_tokens": 187331474.0, + "step": 7288 + }, + { + "epoch": 0.8004612343509774, + "grad_norm": 2.155665636062622, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7290805578231812, + "num_tokens": 187355852.0, + "step": 7289 + }, + { + "epoch": 0.800571052053591, + "grad_norm": 2.2028584480285645, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7000092267990112, + "num_tokens": 187380368.0, + "step": 7290 + }, + { + "epoch": 0.8006808697562047, + "grad_norm": 2.108567476272583, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7180653214454651, + "num_tokens": 187407823.0, + "step": 7291 + }, + { + "epoch": 0.8007906874588183, + "grad_norm": 1.9167286157608032, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6919410824775696, + "num_tokens": 187440078.0, + "step": 7292 + }, + { + "epoch": 0.8009005051614321, + "grad_norm": 2.339520215988159, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7166687846183777, + "num_tokens": 187463028.0, + "step": 7293 + }, + { + "epoch": 0.8010103228640457, + "grad_norm": 1.975993037223816, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7046303153038025, + "num_tokens": 187495664.0, + "step": 7294 + }, + { + "epoch": 0.8011201405666594, + "grad_norm": 1.8618338108062744, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6986560821533203, + "num_tokens": 187529514.0, + "step": 7295 + }, + { + "epoch": 0.801229958269273, + "grad_norm": 1.9999011754989624, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7385209202766418, + "num_tokens": 187556819.0, + "step": 7296 + }, + { + "epoch": 0.8013397759718867, + "grad_norm": 2.1210885047912598, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6846190690994263, + "num_tokens": 187585697.0, + "step": 7297 + }, + { + "epoch": 0.8014495936745003, + "grad_norm": 2.2709102630615234, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7363235354423523, + "num_tokens": 187608406.0, + "step": 7298 + }, + { + "epoch": 0.801559411377114, + "grad_norm": 2.0984439849853516, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7058103680610657, + "num_tokens": 187634078.0, + "step": 7299 + }, + { + "epoch": 0.8016692290797276, + "grad_norm": 1.9259692430496216, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6875721216201782, + "num_tokens": 187666150.0, + "step": 7300 + }, + { + "epoch": 0.8017790467823414, + "grad_norm": 2.0580568313598633, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6995686292648315, + "num_tokens": 187695121.0, + "step": 7301 + }, + { + "epoch": 0.801888864484955, + "grad_norm": 2.160137891769409, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6816393136978149, + "num_tokens": 187722325.0, + "step": 7302 + }, + { + "epoch": 0.8019986821875686, + "grad_norm": 2.3301610946655273, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7046692371368408, + "num_tokens": 187746138.0, + "step": 7303 + }, + { + "epoch": 0.8021084998901823, + "grad_norm": 2.2547266483306885, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6879321336746216, + "num_tokens": 187770513.0, + "step": 7304 + }, + { + "epoch": 0.8022183175927959, + "grad_norm": 2.526348114013672, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7313599586486816, + "num_tokens": 187789099.0, + "step": 7305 + }, + { + "epoch": 0.8023281352954096, + "grad_norm": 2.2365734577178955, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6879974603652954, + "num_tokens": 187813151.0, + "step": 7306 + }, + { + "epoch": 0.8024379529980232, + "grad_norm": 2.3270370960235596, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7328683137893677, + "num_tokens": 187832849.0, + "step": 7307 + }, + { + "epoch": 0.802547770700637, + "grad_norm": 2.1511824131011963, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7120591998100281, + "num_tokens": 187857118.0, + "step": 7308 + }, + { + "epoch": 0.8026575884032506, + "grad_norm": 2.557624578475952, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7119610905647278, + "num_tokens": 187876791.0, + "step": 7309 + }, + { + "epoch": 0.8027674061058643, + "grad_norm": 2.364316940307617, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.722361147403717, + "num_tokens": 187896899.0, + "step": 7310 + }, + { + "epoch": 0.8028772238084779, + "grad_norm": 2.300016164779663, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7178696393966675, + "num_tokens": 187920168.0, + "step": 7311 + }, + { + "epoch": 0.8029870415110916, + "grad_norm": 2.0541200637817383, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7162522077560425, + "num_tokens": 187947831.0, + "step": 7312 + }, + { + "epoch": 0.8030968592137052, + "grad_norm": 2.2128536701202393, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7107473015785217, + "num_tokens": 187971348.0, + "step": 7313 + }, + { + "epoch": 0.8032066769163189, + "grad_norm": 2.380906343460083, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.713321328163147, + "num_tokens": 187992624.0, + "step": 7314 + }, + { + "epoch": 0.8033164946189326, + "grad_norm": 2.065549373626709, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7343894243240356, + "num_tokens": 188018301.0, + "step": 7315 + }, + { + "epoch": 0.8034263123215463, + "grad_norm": 2.3880133628845215, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.71677565574646, + "num_tokens": 188038044.0, + "step": 7316 + }, + { + "epoch": 0.8035361300241599, + "grad_norm": 2.1204285621643066, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6917222738265991, + "num_tokens": 188064017.0, + "step": 7317 + }, + { + "epoch": 0.8036459477267736, + "grad_norm": 2.207846164703369, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7197929620742798, + "num_tokens": 188087880.0, + "step": 7318 + }, + { + "epoch": 0.8037557654293872, + "grad_norm": 2.076495885848999, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7169232964515686, + "num_tokens": 188114476.0, + "step": 7319 + }, + { + "epoch": 0.8038655831320008, + "grad_norm": 2.4295618534088135, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7024206519126892, + "num_tokens": 188135770.0, + "step": 7320 + }, + { + "epoch": 0.8039754008346145, + "grad_norm": 2.22013521194458, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7302712202072144, + "num_tokens": 188159598.0, + "step": 7321 + }, + { + "epoch": 0.8040852185372283, + "grad_norm": 2.0183372497558594, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7044857740402222, + "num_tokens": 188189259.0, + "step": 7322 + }, + { + "epoch": 0.8041950362398419, + "grad_norm": 2.1925554275512695, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7299236059188843, + "num_tokens": 188213435.0, + "step": 7323 + }, + { + "epoch": 0.8043048539424555, + "grad_norm": 2.20406436920166, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.699628472328186, + "num_tokens": 188239104.0, + "step": 7324 + }, + { + "epoch": 0.8044146716450692, + "grad_norm": 1.857434868812561, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6935765743255615, + "num_tokens": 188273860.0, + "step": 7325 + }, + { + "epoch": 0.8045244893476828, + "grad_norm": 2.0409038066864014, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7230679392814636, + "num_tokens": 188301920.0, + "step": 7326 + }, + { + "epoch": 0.8046343070502965, + "grad_norm": 2.005695104598999, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7063608169555664, + "num_tokens": 188329545.0, + "step": 7327 + }, + { + "epoch": 0.8047441247529101, + "grad_norm": 2.081866979598999, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7035374641418457, + "num_tokens": 188355988.0, + "step": 7328 + }, + { + "epoch": 0.8048539424555239, + "grad_norm": 2.1337943077087402, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7001655697822571, + "num_tokens": 188382668.0, + "step": 7329 + }, + { + "epoch": 0.8049637601581375, + "grad_norm": 2.5066418647766113, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7214540839195251, + "num_tokens": 188402748.0, + "step": 7330 + }, + { + "epoch": 0.8050735778607512, + "grad_norm": 2.153947353363037, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7074823379516602, + "num_tokens": 188429966.0, + "step": 7331 + }, + { + "epoch": 0.8051833955633648, + "grad_norm": 2.264368772506714, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7115632891654968, + "num_tokens": 188455775.0, + "step": 7332 + }, + { + "epoch": 0.8052932132659785, + "grad_norm": 1.9179846048355103, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7057524919509888, + "num_tokens": 188486870.0, + "step": 7333 + }, + { + "epoch": 0.8054030309685921, + "grad_norm": 2.1729722023010254, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7026818990707397, + "num_tokens": 188515227.0, + "step": 7334 + }, + { + "epoch": 0.8055128486712058, + "grad_norm": 1.9891735315322876, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6828652620315552, + "num_tokens": 188546382.0, + "step": 7335 + }, + { + "epoch": 0.8056226663738194, + "grad_norm": 1.9565426111221313, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.6999975442886353, + "num_tokens": 188574933.0, + "step": 7336 + }, + { + "epoch": 0.8057324840764332, + "grad_norm": 2.430910348892212, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7349578738212585, + "num_tokens": 188594202.0, + "step": 7337 + }, + { + "epoch": 0.8058423017790468, + "grad_norm": 1.896136999130249, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7006241679191589, + "num_tokens": 188624712.0, + "step": 7338 + }, + { + "epoch": 0.8059521194816605, + "grad_norm": 2.1807262897491455, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7167097926139832, + "num_tokens": 188648841.0, + "step": 7339 + }, + { + "epoch": 0.8060619371842741, + "grad_norm": 2.227200984954834, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.7028803825378418, + "num_tokens": 188672807.0, + "step": 7340 + }, + { + "epoch": 0.8061717548868877, + "grad_norm": 2.2481741905212402, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7071876525878906, + "num_tokens": 188698009.0, + "step": 7341 + }, + { + "epoch": 0.8062815725895014, + "grad_norm": 2.1327614784240723, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7132412791252136, + "num_tokens": 188724500.0, + "step": 7342 + }, + { + "epoch": 0.806391390292115, + "grad_norm": 2.165194272994995, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7093042135238647, + "num_tokens": 188749339.0, + "step": 7343 + }, + { + "epoch": 0.8065012079947288, + "grad_norm": 2.306406021118164, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.717528223991394, + "num_tokens": 188772089.0, + "step": 7344 + }, + { + "epoch": 0.8066110256973424, + "grad_norm": 2.234250068664551, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6850603818893433, + "num_tokens": 188798625.0, + "step": 7345 + }, + { + "epoch": 0.8067208433999561, + "grad_norm": 2.0780093669891357, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7163437008857727, + "num_tokens": 188825766.0, + "step": 7346 + }, + { + "epoch": 0.8068306611025697, + "grad_norm": 2.079681396484375, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6940953731536865, + "num_tokens": 188854265.0, + "step": 7347 + }, + { + "epoch": 0.8069404788051834, + "grad_norm": 1.8958054780960083, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6891260743141174, + "num_tokens": 188888097.0, + "step": 7348 + }, + { + "epoch": 0.807050296507797, + "grad_norm": 2.2718112468719482, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7040997743606567, + "num_tokens": 188911760.0, + "step": 7349 + }, + { + "epoch": 0.8071601142104107, + "grad_norm": 1.9469878673553467, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7003411054611206, + "num_tokens": 188940375.0, + "step": 7350 + }, + { + "epoch": 0.8072699319130244, + "grad_norm": 2.288921594619751, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6882274150848389, + "num_tokens": 188965731.0, + "step": 7351 + }, + { + "epoch": 0.8073797496156381, + "grad_norm": 2.165295124053955, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.691947340965271, + "num_tokens": 188992878.0, + "step": 7352 + }, + { + "epoch": 0.8074895673182517, + "grad_norm": 2.105166435241699, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7187071442604065, + "num_tokens": 189019228.0, + "step": 7353 + }, + { + "epoch": 0.8075993850208654, + "grad_norm": 2.232367992401123, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6837753057479858, + "num_tokens": 189044425.0, + "step": 7354 + }, + { + "epoch": 0.807709202723479, + "grad_norm": 2.2288544178009033, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7196671962738037, + "num_tokens": 189067685.0, + "step": 7355 + }, + { + "epoch": 0.8078190204260927, + "grad_norm": 2.102029800415039, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6928758025169373, + "num_tokens": 189095828.0, + "step": 7356 + }, + { + "epoch": 0.8079288381287063, + "grad_norm": 2.0598316192626953, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7290879487991333, + "num_tokens": 189122166.0, + "step": 7357 + }, + { + "epoch": 0.8080386558313201, + "grad_norm": 2.218200445175171, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7066935300827026, + "num_tokens": 189145240.0, + "step": 7358 + }, + { + "epoch": 0.8081484735339337, + "grad_norm": 2.4360404014587402, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7145050764083862, + "num_tokens": 189166855.0, + "step": 7359 + }, + { + "epoch": 0.8082582912365474, + "grad_norm": 2.20868182182312, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.728093147277832, + "num_tokens": 189189694.0, + "step": 7360 + }, + { + "epoch": 0.808368108939161, + "grad_norm": 1.9472219944000244, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7094958424568176, + "num_tokens": 189220666.0, + "step": 7361 + }, + { + "epoch": 0.8084779266417746, + "grad_norm": 2.2881181240081787, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.712524950504303, + "num_tokens": 189243058.0, + "step": 7362 + }, + { + "epoch": 0.8085877443443883, + "grad_norm": 2.248969793319702, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7086688280105591, + "num_tokens": 189268108.0, + "step": 7363 + }, + { + "epoch": 0.8086975620470019, + "grad_norm": 2.2978787422180176, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7065796256065369, + "num_tokens": 189291872.0, + "step": 7364 + }, + { + "epoch": 0.8088073797496156, + "grad_norm": 2.243034601211548, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7208085656166077, + "num_tokens": 189315662.0, + "step": 7365 + }, + { + "epoch": 0.8089171974522293, + "grad_norm": 1.99271559715271, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6901670694351196, + "num_tokens": 189346960.0, + "step": 7366 + }, + { + "epoch": 0.809027015154843, + "grad_norm": 2.387636423110962, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7214246988296509, + "num_tokens": 189368730.0, + "step": 7367 + }, + { + "epoch": 0.8091368328574566, + "grad_norm": 2.4370720386505127, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7035021781921387, + "num_tokens": 189391017.0, + "step": 7368 + }, + { + "epoch": 0.8092466505600703, + "grad_norm": 1.9392083883285522, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.702061116695404, + "num_tokens": 189422506.0, + "step": 7369 + }, + { + "epoch": 0.8093564682626839, + "grad_norm": 2.3222339153289795, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7134488821029663, + "num_tokens": 189448375.0, + "step": 7370 + }, + { + "epoch": 0.8094662859652976, + "grad_norm": 2.1443703174591064, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6984813213348389, + "num_tokens": 189477534.0, + "step": 7371 + }, + { + "epoch": 0.8095761036679112, + "grad_norm": 2.1962502002716064, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7226210832595825, + "num_tokens": 189502863.0, + "step": 7372 + }, + { + "epoch": 0.809685921370525, + "grad_norm": 2.152430295944214, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6944265365600586, + "num_tokens": 189529922.0, + "step": 7373 + }, + { + "epoch": 0.8097957390731386, + "grad_norm": 2.4039676189422607, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7084171772003174, + "num_tokens": 189552106.0, + "step": 7374 + }, + { + "epoch": 0.8099055567757523, + "grad_norm": 2.10160493850708, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6951621770858765, + "num_tokens": 189580786.0, + "step": 7375 + }, + { + "epoch": 0.8100153744783659, + "grad_norm": 2.2122011184692383, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.716984212398529, + "num_tokens": 189606550.0, + "step": 7376 + }, + { + "epoch": 0.8101251921809796, + "grad_norm": 2.061220645904541, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.7078665494918823, + "num_tokens": 189633159.0, + "step": 7377 + }, + { + "epoch": 0.8102350098835932, + "grad_norm": 2.237368106842041, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7090696692466736, + "num_tokens": 189656845.0, + "step": 7378 + }, + { + "epoch": 0.8103448275862069, + "grad_norm": 2.2373766899108887, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6940463781356812, + "num_tokens": 189681529.0, + "step": 7379 + }, + { + "epoch": 0.8104546452888206, + "grad_norm": 2.27942156791687, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.697717547416687, + "num_tokens": 189705129.0, + "step": 7380 + }, + { + "epoch": 0.8105644629914343, + "grad_norm": 2.0996763706207275, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6941566467285156, + "num_tokens": 189731218.0, + "step": 7381 + }, + { + "epoch": 0.8106742806940479, + "grad_norm": 2.214704990386963, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6959669589996338, + "num_tokens": 189755533.0, + "step": 7382 + }, + { + "epoch": 0.8107840983966615, + "grad_norm": 2.4654250144958496, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7174205780029297, + "num_tokens": 189775352.0, + "step": 7383 + }, + { + "epoch": 0.8108939160992752, + "grad_norm": 2.09256911277771, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.707407534122467, + "num_tokens": 189802387.0, + "step": 7384 + }, + { + "epoch": 0.8110037338018888, + "grad_norm": 2.230482578277588, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7448381781578064, + "num_tokens": 189826560.0, + "step": 7385 + }, + { + "epoch": 0.8111135515045025, + "grad_norm": 2.056879997253418, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.716610312461853, + "num_tokens": 189854477.0, + "step": 7386 + }, + { + "epoch": 0.8112233692071162, + "grad_norm": 2.2233920097351074, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6883169412612915, + "num_tokens": 189880858.0, + "step": 7387 + }, + { + "epoch": 0.8113331869097299, + "grad_norm": 2.0977671146392822, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7119574546813965, + "num_tokens": 189907259.0, + "step": 7388 + }, + { + "epoch": 0.8114430046123435, + "grad_norm": 2.0305542945861816, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7068939208984375, + "num_tokens": 189937608.0, + "step": 7389 + }, + { + "epoch": 0.8115528223149572, + "grad_norm": 2.0752501487731934, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7000463604927063, + "num_tokens": 189966050.0, + "step": 7390 + }, + { + "epoch": 0.8116626400175708, + "grad_norm": 2.28564453125, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7206652164459229, + "num_tokens": 189988221.0, + "step": 7391 + }, + { + "epoch": 0.8117724577201845, + "grad_norm": 2.018507242202759, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7028985619544983, + "num_tokens": 190016958.0, + "step": 7392 + }, + { + "epoch": 0.8118822754227981, + "grad_norm": 2.1858203411102295, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7133747935295105, + "num_tokens": 190040902.0, + "step": 7393 + }, + { + "epoch": 0.8119920931254118, + "grad_norm": 2.534087657928467, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7187181115150452, + "num_tokens": 190060023.0, + "step": 7394 + }, + { + "epoch": 0.8121019108280255, + "grad_norm": 2.0740604400634766, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7085589170455933, + "num_tokens": 190087324.0, + "step": 7395 + }, + { + "epoch": 0.8122117285306392, + "grad_norm": 2.2351856231689453, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7106540203094482, + "num_tokens": 190109850.0, + "step": 7396 + }, + { + "epoch": 0.8123215462332528, + "grad_norm": 2.245927333831787, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7157727479934692, + "num_tokens": 190133366.0, + "step": 7397 + }, + { + "epoch": 0.8124313639358665, + "grad_norm": 2.068011522293091, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7040137648582458, + "num_tokens": 190160392.0, + "step": 7398 + }, + { + "epoch": 0.8125411816384801, + "grad_norm": 2.433833599090576, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7306978702545166, + "num_tokens": 190180920.0, + "step": 7399 + }, + { + "epoch": 0.8126509993410937, + "grad_norm": 2.507434368133545, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6886117458343506, + "num_tokens": 190202136.0, + "step": 7400 + }, + { + "epoch": 0.8127608170437074, + "grad_norm": 2.4333598613739014, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7271449565887451, + "num_tokens": 190222159.0, + "step": 7401 + }, + { + "epoch": 0.8128706347463212, + "grad_norm": 2.168041229248047, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7149900794029236, + "num_tokens": 190248143.0, + "step": 7402 + }, + { + "epoch": 0.8129804524489348, + "grad_norm": 2.1061031818389893, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7159764766693115, + "num_tokens": 190275239.0, + "step": 7403 + }, + { + "epoch": 0.8130902701515484, + "grad_norm": 2.243177890777588, + "learning_rate": 1e-06, + "loss": 1.0917, + "mean_token_accuracy": 0.6846762895584106, + "num_tokens": 190300164.0, + "step": 7404 + }, + { + "epoch": 0.8132000878541621, + "grad_norm": 1.9841817617416382, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7109728455543518, + "num_tokens": 190330560.0, + "step": 7405 + }, + { + "epoch": 0.8133099055567757, + "grad_norm": 2.026975154876709, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7093452215194702, + "num_tokens": 190357699.0, + "step": 7406 + }, + { + "epoch": 0.8134197232593894, + "grad_norm": 1.974400520324707, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7030895948410034, + "num_tokens": 190389726.0, + "step": 7407 + }, + { + "epoch": 0.813529540962003, + "grad_norm": 2.2994275093078613, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.683076024055481, + "num_tokens": 190414884.0, + "step": 7408 + }, + { + "epoch": 0.8136393586646168, + "grad_norm": 2.186561107635498, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.704265832901001, + "num_tokens": 190440283.0, + "step": 7409 + }, + { + "epoch": 0.8137491763672304, + "grad_norm": 2.1384196281433105, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7224125862121582, + "num_tokens": 190466216.0, + "step": 7410 + }, + { + "epoch": 0.8138589940698441, + "grad_norm": 2.451784610748291, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.712578296661377, + "num_tokens": 190486632.0, + "step": 7411 + }, + { + "epoch": 0.8139688117724577, + "grad_norm": 2.2204489707946777, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7230819463729858, + "num_tokens": 190510909.0, + "step": 7412 + }, + { + "epoch": 0.8140786294750714, + "grad_norm": 2.276848316192627, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7025208473205566, + "num_tokens": 190536451.0, + "step": 7413 + }, + { + "epoch": 0.814188447177685, + "grad_norm": 2.1619677543640137, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7255493998527527, + "num_tokens": 190561901.0, + "step": 7414 + }, + { + "epoch": 0.8142982648802987, + "grad_norm": 2.348893642425537, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.704492449760437, + "num_tokens": 190585171.0, + "step": 7415 + }, + { + "epoch": 0.8144080825829124, + "grad_norm": 2.2430918216705322, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7147400379180908, + "num_tokens": 190608329.0, + "step": 7416 + }, + { + "epoch": 0.8145179002855261, + "grad_norm": 2.4870917797088623, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6947534084320068, + "num_tokens": 190629944.0, + "step": 7417 + }, + { + "epoch": 0.8146277179881397, + "grad_norm": 2.5940487384796143, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7060550451278687, + "num_tokens": 190650319.0, + "step": 7418 + }, + { + "epoch": 0.8147375356907534, + "grad_norm": 2.1716363430023193, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7409154176712036, + "num_tokens": 190673774.0, + "step": 7419 + }, + { + "epoch": 0.814847353393367, + "grad_norm": 1.9424517154693604, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6910997033119202, + "num_tokens": 190705817.0, + "step": 7420 + }, + { + "epoch": 0.8149571710959806, + "grad_norm": 2.38320255279541, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7362207174301147, + "num_tokens": 190727581.0, + "step": 7421 + }, + { + "epoch": 0.8150669887985943, + "grad_norm": 2.039116382598877, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6852144002914429, + "num_tokens": 190757352.0, + "step": 7422 + }, + { + "epoch": 0.8151768065012079, + "grad_norm": 2.046121120452881, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6784505248069763, + "num_tokens": 190787269.0, + "step": 7423 + }, + { + "epoch": 0.8152866242038217, + "grad_norm": 2.430715322494507, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7257453799247742, + "num_tokens": 190806631.0, + "step": 7424 + }, + { + "epoch": 0.8153964419064353, + "grad_norm": 1.9226993322372437, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6879650950431824, + "num_tokens": 190837925.0, + "step": 7425 + }, + { + "epoch": 0.815506259609049, + "grad_norm": 2.5120091438293457, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7309969067573547, + "num_tokens": 190857326.0, + "step": 7426 + }, + { + "epoch": 0.8156160773116626, + "grad_norm": 2.1263363361358643, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7024495601654053, + "num_tokens": 190884818.0, + "step": 7427 + }, + { + "epoch": 0.8157258950142763, + "grad_norm": 2.067790985107422, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.697762668132782, + "num_tokens": 190917024.0, + "step": 7428 + }, + { + "epoch": 0.8158357127168899, + "grad_norm": 2.575690507888794, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.720449686050415, + "num_tokens": 190936077.0, + "step": 7429 + }, + { + "epoch": 0.8159455304195036, + "grad_norm": 2.1978018283843994, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6937851905822754, + "num_tokens": 190961019.0, + "step": 7430 + }, + { + "epoch": 0.8160553481221173, + "grad_norm": 2.113046884536743, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7119247913360596, + "num_tokens": 190987813.0, + "step": 7431 + }, + { + "epoch": 0.816165165824731, + "grad_norm": 2.1469106674194336, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7104436159133911, + "num_tokens": 191013139.0, + "step": 7432 + }, + { + "epoch": 0.8162749835273446, + "grad_norm": 2.4111809730529785, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7141388058662415, + "num_tokens": 191035310.0, + "step": 7433 + }, + { + "epoch": 0.8163848012299583, + "grad_norm": 2.097707748413086, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6843724250793457, + "num_tokens": 191062942.0, + "step": 7434 + }, + { + "epoch": 0.8164946189325719, + "grad_norm": 2.0346648693084717, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7110148072242737, + "num_tokens": 191093310.0, + "step": 7435 + }, + { + "epoch": 0.8166044366351856, + "grad_norm": 1.960949420928955, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6954755783081055, + "num_tokens": 191125694.0, + "step": 7436 + }, + { + "epoch": 0.8167142543377992, + "grad_norm": 2.1190972328186035, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.707172155380249, + "num_tokens": 191149949.0, + "step": 7437 + }, + { + "epoch": 0.816824072040413, + "grad_norm": 2.0932958126068115, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6922667026519775, + "num_tokens": 191179895.0, + "step": 7438 + }, + { + "epoch": 0.8169338897430266, + "grad_norm": 2.380986213684082, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7231009602546692, + "num_tokens": 191200625.0, + "step": 7439 + }, + { + "epoch": 0.8170437074456403, + "grad_norm": 2.2198586463928223, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7109163999557495, + "num_tokens": 191226523.0, + "step": 7440 + }, + { + "epoch": 0.8171535251482539, + "grad_norm": 1.9291411638259888, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6975448131561279, + "num_tokens": 191258121.0, + "step": 7441 + }, + { + "epoch": 0.8172633428508675, + "grad_norm": 2.1297686100006104, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7256331443786621, + "num_tokens": 191284028.0, + "step": 7442 + }, + { + "epoch": 0.8173731605534812, + "grad_norm": 2.361342191696167, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7070659399032593, + "num_tokens": 191305565.0, + "step": 7443 + }, + { + "epoch": 0.8174829782560948, + "grad_norm": 2.525183916091919, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7132630348205566, + "num_tokens": 191326511.0, + "step": 7444 + }, + { + "epoch": 0.8175927959587086, + "grad_norm": 2.0353353023529053, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6994506120681763, + "num_tokens": 191355661.0, + "step": 7445 + }, + { + "epoch": 0.8177026136613222, + "grad_norm": 2.288900852203369, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7095610499382019, + "num_tokens": 191379778.0, + "step": 7446 + }, + { + "epoch": 0.8178124313639359, + "grad_norm": 2.1466116905212402, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7357863783836365, + "num_tokens": 191406200.0, + "step": 7447 + }, + { + "epoch": 0.8179222490665495, + "grad_norm": 2.234375476837158, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6822015047073364, + "num_tokens": 191431249.0, + "step": 7448 + }, + { + "epoch": 0.8180320667691632, + "grad_norm": 2.084564208984375, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7071104645729065, + "num_tokens": 191460637.0, + "step": 7449 + }, + { + "epoch": 0.8181418844717768, + "grad_norm": 2.391432285308838, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6912503838539124, + "num_tokens": 191484496.0, + "step": 7450 + }, + { + "epoch": 0.8182517021743905, + "grad_norm": 2.1698296070098877, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.726181149482727, + "num_tokens": 191509065.0, + "step": 7451 + }, + { + "epoch": 0.8183615198770041, + "grad_norm": 1.9301193952560425, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6965291500091553, + "num_tokens": 191540455.0, + "step": 7452 + }, + { + "epoch": 0.8184713375796179, + "grad_norm": 2.260545015335083, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6968635320663452, + "num_tokens": 191567098.0, + "step": 7453 + }, + { + "epoch": 0.8185811552822315, + "grad_norm": 2.2301487922668457, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7355155944824219, + "num_tokens": 191590685.0, + "step": 7454 + }, + { + "epoch": 0.8186909729848452, + "grad_norm": 2.3574953079223633, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7011606693267822, + "num_tokens": 191612732.0, + "step": 7455 + }, + { + "epoch": 0.8188007906874588, + "grad_norm": 2.0116381645202637, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7073975801467896, + "num_tokens": 191640418.0, + "step": 7456 + }, + { + "epoch": 0.8189106083900725, + "grad_norm": 2.0320332050323486, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6860891580581665, + "num_tokens": 191671007.0, + "step": 7457 + }, + { + "epoch": 0.8190204260926861, + "grad_norm": 2.2266440391540527, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.7029694318771362, + "num_tokens": 191698466.0, + "step": 7458 + }, + { + "epoch": 0.8191302437952998, + "grad_norm": 2.0100395679473877, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7148073315620422, + "num_tokens": 191726930.0, + "step": 7459 + }, + { + "epoch": 0.8192400614979135, + "grad_norm": 2.2596850395202637, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7207092642784119, + "num_tokens": 191748266.0, + "step": 7460 + }, + { + "epoch": 0.8193498792005272, + "grad_norm": 2.1839001178741455, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6911036372184753, + "num_tokens": 191773142.0, + "step": 7461 + }, + { + "epoch": 0.8194596969031408, + "grad_norm": 2.25234055519104, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6884036660194397, + "num_tokens": 191798342.0, + "step": 7462 + }, + { + "epoch": 0.8195695146057544, + "grad_norm": 1.9690842628479004, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7001805305480957, + "num_tokens": 191827389.0, + "step": 7463 + }, + { + "epoch": 0.8196793323083681, + "grad_norm": 2.227961540222168, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.674354076385498, + "num_tokens": 191853530.0, + "step": 7464 + }, + { + "epoch": 0.8197891500109817, + "grad_norm": 2.182431936264038, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7117093801498413, + "num_tokens": 191876317.0, + "step": 7465 + }, + { + "epoch": 0.8198989677135954, + "grad_norm": 2.222262382507324, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7079187631607056, + "num_tokens": 191899561.0, + "step": 7466 + }, + { + "epoch": 0.8200087854162091, + "grad_norm": 2.38039493560791, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7098749876022339, + "num_tokens": 191921878.0, + "step": 7467 + }, + { + "epoch": 0.8201186031188228, + "grad_norm": 2.2139575481414795, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.717600405216217, + "num_tokens": 191946017.0, + "step": 7468 + }, + { + "epoch": 0.8202284208214364, + "grad_norm": 2.061103343963623, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6953322291374207, + "num_tokens": 191975199.0, + "step": 7469 + }, + { + "epoch": 0.8203382385240501, + "grad_norm": 2.0779905319213867, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6967207193374634, + "num_tokens": 192003686.0, + "step": 7470 + }, + { + "epoch": 0.8204480562266637, + "grad_norm": 2.360682249069214, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7135424017906189, + "num_tokens": 192026786.0, + "step": 7471 + }, + { + "epoch": 0.8205578739292774, + "grad_norm": 2.1600499153137207, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.72243332862854, + "num_tokens": 192052549.0, + "step": 7472 + }, + { + "epoch": 0.820667691631891, + "grad_norm": 2.2608070373535156, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7200899720191956, + "num_tokens": 192075689.0, + "step": 7473 + }, + { + "epoch": 0.8207775093345048, + "grad_norm": 2.3423380851745605, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7078060507774353, + "num_tokens": 192097905.0, + "step": 7474 + }, + { + "epoch": 0.8208873270371184, + "grad_norm": 2.499945878982544, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7356071472167969, + "num_tokens": 192118761.0, + "step": 7475 + }, + { + "epoch": 0.8209971447397321, + "grad_norm": 2.1659796237945557, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7309659719467163, + "num_tokens": 192143745.0, + "step": 7476 + }, + { + "epoch": 0.8211069624423457, + "grad_norm": 2.269596815109253, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7028622031211853, + "num_tokens": 192168602.0, + "step": 7477 + }, + { + "epoch": 0.8212167801449594, + "grad_norm": 2.328298330307007, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7011516094207764, + "num_tokens": 192192416.0, + "step": 7478 + }, + { + "epoch": 0.821326597847573, + "grad_norm": 2.093470573425293, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7022101879119873, + "num_tokens": 192220400.0, + "step": 7479 + }, + { + "epoch": 0.8214364155501866, + "grad_norm": 2.066162586212158, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6831451654434204, + "num_tokens": 192250119.0, + "step": 7480 + }, + { + "epoch": 0.8215462332528004, + "grad_norm": 1.990911602973938, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6993876695632935, + "num_tokens": 192279481.0, + "step": 7481 + }, + { + "epoch": 0.821656050955414, + "grad_norm": 2.0758440494537354, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7104646563529968, + "num_tokens": 192306891.0, + "step": 7482 + }, + { + "epoch": 0.8217658686580277, + "grad_norm": 2.1307528018951416, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6804933547973633, + "num_tokens": 192336501.0, + "step": 7483 + }, + { + "epoch": 0.8218756863606413, + "grad_norm": 2.0743649005889893, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7086267471313477, + "num_tokens": 192364090.0, + "step": 7484 + }, + { + "epoch": 0.821985504063255, + "grad_norm": 2.1281988620758057, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7073934674263, + "num_tokens": 192390989.0, + "step": 7485 + }, + { + "epoch": 0.8220953217658686, + "grad_norm": 2.0982062816619873, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6954627633094788, + "num_tokens": 192416605.0, + "step": 7486 + }, + { + "epoch": 0.8222051394684823, + "grad_norm": 2.5219781398773193, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7183349132537842, + "num_tokens": 192437063.0, + "step": 7487 + }, + { + "epoch": 0.8223149571710959, + "grad_norm": 2.283720016479492, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7341063022613525, + "num_tokens": 192460262.0, + "step": 7488 + }, + { + "epoch": 0.8224247748737097, + "grad_norm": 2.2752678394317627, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7011330127716064, + "num_tokens": 192485296.0, + "step": 7489 + }, + { + "epoch": 0.8225345925763233, + "grad_norm": 1.895551323890686, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.691980242729187, + "num_tokens": 192517725.0, + "step": 7490 + }, + { + "epoch": 0.822644410278937, + "grad_norm": 2.027989149093628, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7235379815101624, + "num_tokens": 192546601.0, + "step": 7491 + }, + { + "epoch": 0.8227542279815506, + "grad_norm": 2.2610573768615723, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.701315701007843, + "num_tokens": 192570123.0, + "step": 7492 + }, + { + "epoch": 0.8228640456841643, + "grad_norm": 2.0735557079315186, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6962566375732422, + "num_tokens": 192599523.0, + "step": 7493 + }, + { + "epoch": 0.8229738633867779, + "grad_norm": 2.237715482711792, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7048559784889221, + "num_tokens": 192622746.0, + "step": 7494 + }, + { + "epoch": 0.8230836810893916, + "grad_norm": 2.294530153274536, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6887901425361633, + "num_tokens": 192648196.0, + "step": 7495 + }, + { + "epoch": 0.8231934987920053, + "grad_norm": 2.150862216949463, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.7001299262046814, + "num_tokens": 192673848.0, + "step": 7496 + }, + { + "epoch": 0.823303316494619, + "grad_norm": 2.2904090881347656, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7087187170982361, + "num_tokens": 192695707.0, + "step": 7497 + }, + { + "epoch": 0.8234131341972326, + "grad_norm": 2.1848556995391846, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7138997316360474, + "num_tokens": 192721488.0, + "step": 7498 + }, + { + "epoch": 0.8235229518998463, + "grad_norm": 2.256700038909912, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7245498895645142, + "num_tokens": 192742561.0, + "step": 7499 + }, + { + "epoch": 0.8236327696024599, + "grad_norm": 2.0975253582000732, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7096401453018188, + "num_tokens": 192769382.0, + "step": 7500 + }, + { + "epoch": 0.8237425873050735, + "grad_norm": 2.295269727706909, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7224714159965515, + "num_tokens": 192791530.0, + "step": 7501 + }, + { + "epoch": 0.8238524050076872, + "grad_norm": 2.376077890396118, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7300786375999451, + "num_tokens": 192811918.0, + "step": 7502 + }, + { + "epoch": 0.823962222710301, + "grad_norm": 2.1918210983276367, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7054092884063721, + "num_tokens": 192836832.0, + "step": 7503 + }, + { + "epoch": 0.8240720404129146, + "grad_norm": 2.510537624359131, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7427301406860352, + "num_tokens": 192855866.0, + "step": 7504 + }, + { + "epoch": 0.8241818581155282, + "grad_norm": 2.26865816116333, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7070742845535278, + "num_tokens": 192879080.0, + "step": 7505 + }, + { + "epoch": 0.8242916758181419, + "grad_norm": 2.13874888420105, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7276502251625061, + "num_tokens": 192903071.0, + "step": 7506 + }, + { + "epoch": 0.8244014935207555, + "grad_norm": 2.4370744228363037, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7181468605995178, + "num_tokens": 192923801.0, + "step": 7507 + }, + { + "epoch": 0.8245113112233692, + "grad_norm": 2.065171718597412, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7328211665153503, + "num_tokens": 192950832.0, + "step": 7508 + }, + { + "epoch": 0.8246211289259828, + "grad_norm": 2.340994358062744, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7071372270584106, + "num_tokens": 192976136.0, + "step": 7509 + }, + { + "epoch": 0.8247309466285966, + "grad_norm": 2.3565287590026855, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7072699069976807, + "num_tokens": 192999449.0, + "step": 7510 + }, + { + "epoch": 0.8248407643312102, + "grad_norm": 2.130525827407837, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.693924069404602, + "num_tokens": 193027107.0, + "step": 7511 + }, + { + "epoch": 0.8249505820338239, + "grad_norm": 2.254413366317749, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6932221055030823, + "num_tokens": 193051702.0, + "step": 7512 + }, + { + "epoch": 0.8250603997364375, + "grad_norm": 2.5154166221618652, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7224059104919434, + "num_tokens": 193071069.0, + "step": 7513 + }, + { + "epoch": 0.8251702174390512, + "grad_norm": 2.1307406425476074, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6763103604316711, + "num_tokens": 193099629.0, + "step": 7514 + }, + { + "epoch": 0.8252800351416648, + "grad_norm": 2.4312829971313477, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7075217366218567, + "num_tokens": 193121297.0, + "step": 7515 + }, + { + "epoch": 0.8253898528442785, + "grad_norm": 2.1876137256622314, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7099130153656006, + "num_tokens": 193150111.0, + "step": 7516 + }, + { + "epoch": 0.8254996705468921, + "grad_norm": 2.024587869644165, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7066004276275635, + "num_tokens": 193177842.0, + "step": 7517 + }, + { + "epoch": 0.8256094882495059, + "grad_norm": 2.1771655082702637, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6968672275543213, + "num_tokens": 193202790.0, + "step": 7518 + }, + { + "epoch": 0.8257193059521195, + "grad_norm": 2.131234645843506, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6925879120826721, + "num_tokens": 193229715.0, + "step": 7519 + }, + { + "epoch": 0.8258291236547332, + "grad_norm": 2.2267887592315674, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6942667961120605, + "num_tokens": 193253651.0, + "step": 7520 + }, + { + "epoch": 0.8259389413573468, + "grad_norm": 2.147160530090332, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7041311264038086, + "num_tokens": 193280241.0, + "step": 7521 + }, + { + "epoch": 0.8260487590599604, + "grad_norm": 2.2741312980651855, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6944112777709961, + "num_tokens": 193304291.0, + "step": 7522 + }, + { + "epoch": 0.8261585767625741, + "grad_norm": 2.2415897846221924, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7117340564727783, + "num_tokens": 193328321.0, + "step": 7523 + }, + { + "epoch": 0.8262683944651877, + "grad_norm": 2.240560531616211, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7284449338912964, + "num_tokens": 193351627.0, + "step": 7524 + }, + { + "epoch": 0.8263782121678015, + "grad_norm": 2.3986222743988037, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7284024953842163, + "num_tokens": 193372980.0, + "step": 7525 + }, + { + "epoch": 0.8264880298704151, + "grad_norm": 2.176522731781006, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7098320126533508, + "num_tokens": 193398727.0, + "step": 7526 + }, + { + "epoch": 0.8265978475730288, + "grad_norm": 2.1354432106018066, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7046988606452942, + "num_tokens": 193425932.0, + "step": 7527 + }, + { + "epoch": 0.8267076652756424, + "grad_norm": 2.3304834365844727, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7307814359664917, + "num_tokens": 193450959.0, + "step": 7528 + }, + { + "epoch": 0.8268174829782561, + "grad_norm": 1.792029619216919, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6787313222885132, + "num_tokens": 193489849.0, + "step": 7529 + }, + { + "epoch": 0.8269273006808697, + "grad_norm": 2.1026313304901123, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6942680478096008, + "num_tokens": 193516892.0, + "step": 7530 + }, + { + "epoch": 0.8270371183834834, + "grad_norm": 1.9473559856414795, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7061336636543274, + "num_tokens": 193548302.0, + "step": 7531 + }, + { + "epoch": 0.8271469360860971, + "grad_norm": 2.038813591003418, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7182879447937012, + "num_tokens": 193575119.0, + "step": 7532 + }, + { + "epoch": 0.8272567537887108, + "grad_norm": 2.159555435180664, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6956019401550293, + "num_tokens": 193601614.0, + "step": 7533 + }, + { + "epoch": 0.8273665714913244, + "grad_norm": 2.1759536266326904, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6974675059318542, + "num_tokens": 193626960.0, + "step": 7534 + }, + { + "epoch": 0.8274763891939381, + "grad_norm": 2.4823391437530518, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7208570837974548, + "num_tokens": 193646892.0, + "step": 7535 + }, + { + "epoch": 0.8275862068965517, + "grad_norm": 2.135288715362549, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7171590328216553, + "num_tokens": 193672093.0, + "step": 7536 + }, + { + "epoch": 0.8276960245991654, + "grad_norm": 2.526750087738037, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7195947170257568, + "num_tokens": 193693408.0, + "step": 7537 + }, + { + "epoch": 0.827805842301779, + "grad_norm": 1.9975807666778564, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7143323421478271, + "num_tokens": 193724184.0, + "step": 7538 + }, + { + "epoch": 0.8279156600043928, + "grad_norm": 2.0100762844085693, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7210643291473389, + "num_tokens": 193751255.0, + "step": 7539 + }, + { + "epoch": 0.8280254777070064, + "grad_norm": 2.0642311573028564, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7032954096794128, + "num_tokens": 193777667.0, + "step": 7540 + }, + { + "epoch": 0.82813529540962, + "grad_norm": 2.4538702964782715, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7262120246887207, + "num_tokens": 193798175.0, + "step": 7541 + }, + { + "epoch": 0.8282451131122337, + "grad_norm": 2.0869202613830566, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7211782932281494, + "num_tokens": 193823157.0, + "step": 7542 + }, + { + "epoch": 0.8283549308148473, + "grad_norm": 1.910595178604126, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.717176079750061, + "num_tokens": 193854596.0, + "step": 7543 + }, + { + "epoch": 0.828464748517461, + "grad_norm": 2.3159451484680176, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7060770392417908, + "num_tokens": 193879074.0, + "step": 7544 + }, + { + "epoch": 0.8285745662200746, + "grad_norm": 2.213003635406494, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7101655602455139, + "num_tokens": 193903007.0, + "step": 7545 + }, + { + "epoch": 0.8286843839226883, + "grad_norm": 2.1356966495513916, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6850833892822266, + "num_tokens": 193930287.0, + "step": 7546 + }, + { + "epoch": 0.828794201625302, + "grad_norm": 1.9591480493545532, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.721118152141571, + "num_tokens": 193959790.0, + "step": 7547 + }, + { + "epoch": 0.8289040193279157, + "grad_norm": 2.1910507678985596, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6948394775390625, + "num_tokens": 193985816.0, + "step": 7548 + }, + { + "epoch": 0.8290138370305293, + "grad_norm": 2.0048673152923584, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6982579231262207, + "num_tokens": 194016096.0, + "step": 7549 + }, + { + "epoch": 0.829123654733143, + "grad_norm": 2.357027053833008, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7008885741233826, + "num_tokens": 194038310.0, + "step": 7550 + }, + { + "epoch": 0.8292334724357566, + "grad_norm": 2.2864320278167725, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7148409485816956, + "num_tokens": 194062441.0, + "step": 7551 + }, + { + "epoch": 0.8293432901383703, + "grad_norm": 2.2684104442596436, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7061843872070312, + "num_tokens": 194086661.0, + "step": 7552 + }, + { + "epoch": 0.8294531078409839, + "grad_norm": 2.007791519165039, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7384499311447144, + "num_tokens": 194115028.0, + "step": 7553 + }, + { + "epoch": 0.8295629255435977, + "grad_norm": 1.9740103483200073, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7220190763473511, + "num_tokens": 194145806.0, + "step": 7554 + }, + { + "epoch": 0.8296727432462113, + "grad_norm": 2.007856845855713, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.701126754283905, + "num_tokens": 194176259.0, + "step": 7555 + }, + { + "epoch": 0.829782560948825, + "grad_norm": 2.2680187225341797, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7452495098114014, + "num_tokens": 194199208.0, + "step": 7556 + }, + { + "epoch": 0.8298923786514386, + "grad_norm": 1.926713466644287, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6965214014053345, + "num_tokens": 194231955.0, + "step": 7557 + }, + { + "epoch": 0.8300021963540523, + "grad_norm": 2.4343512058258057, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7253907918930054, + "num_tokens": 194251621.0, + "step": 7558 + }, + { + "epoch": 0.8301120140566659, + "grad_norm": 2.0948879718780518, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.70085209608078, + "num_tokens": 194279385.0, + "step": 7559 + }, + { + "epoch": 0.8302218317592795, + "grad_norm": 2.081216335296631, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7104201316833496, + "num_tokens": 194309288.0, + "step": 7560 + }, + { + "epoch": 0.8303316494618933, + "grad_norm": 2.3562264442443848, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.714595377445221, + "num_tokens": 194332394.0, + "step": 7561 + }, + { + "epoch": 0.830441467164507, + "grad_norm": 2.380352735519409, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.6778613924980164, + "num_tokens": 194356260.0, + "step": 7562 + }, + { + "epoch": 0.8305512848671206, + "grad_norm": 2.1183629035949707, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6882573366165161, + "num_tokens": 194384685.0, + "step": 7563 + }, + { + "epoch": 0.8306611025697342, + "grad_norm": 2.2330315113067627, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.70833420753479, + "num_tokens": 194407534.0, + "step": 7564 + }, + { + "epoch": 0.8307709202723479, + "grad_norm": 2.1658670902252197, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7187347412109375, + "num_tokens": 194433098.0, + "step": 7565 + }, + { + "epoch": 0.8308807379749615, + "grad_norm": 2.2582995891571045, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.708983838558197, + "num_tokens": 194456869.0, + "step": 7566 + }, + { + "epoch": 0.8309905556775752, + "grad_norm": 1.9364851713180542, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6897356510162354, + "num_tokens": 194489777.0, + "step": 7567 + }, + { + "epoch": 0.8311003733801889, + "grad_norm": 2.282993793487549, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7460315227508545, + "num_tokens": 194512709.0, + "step": 7568 + }, + { + "epoch": 0.8312101910828026, + "grad_norm": 2.144285202026367, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6907442808151245, + "num_tokens": 194539721.0, + "step": 7569 + }, + { + "epoch": 0.8313200087854162, + "grad_norm": 2.5583853721618652, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7193971872329712, + "num_tokens": 194557431.0, + "step": 7570 + }, + { + "epoch": 0.8314298264880299, + "grad_norm": 2.0609004497528076, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6903220415115356, + "num_tokens": 194586945.0, + "step": 7571 + }, + { + "epoch": 0.8315396441906435, + "grad_norm": 2.1016933917999268, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6973134279251099, + "num_tokens": 194613143.0, + "step": 7572 + }, + { + "epoch": 0.8316494618932572, + "grad_norm": 2.339078903198242, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6926301717758179, + "num_tokens": 194637706.0, + "step": 7573 + }, + { + "epoch": 0.8317592795958708, + "grad_norm": 2.237464666366577, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.718214213848114, + "num_tokens": 194662545.0, + "step": 7574 + }, + { + "epoch": 0.8318690972984845, + "grad_norm": 2.160360813140869, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.684065580368042, + "num_tokens": 194689895.0, + "step": 7575 + }, + { + "epoch": 0.8319789150010982, + "grad_norm": 2.264141321182251, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7160912156105042, + "num_tokens": 194716440.0, + "step": 7576 + }, + { + "epoch": 0.8320887327037119, + "grad_norm": 2.209296226501465, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6977611780166626, + "num_tokens": 194740550.0, + "step": 7577 + }, + { + "epoch": 0.8321985504063255, + "grad_norm": 2.169031858444214, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7038521766662598, + "num_tokens": 194766906.0, + "step": 7578 + }, + { + "epoch": 0.8323083681089392, + "grad_norm": 2.1865413188934326, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7126525044441223, + "num_tokens": 194793298.0, + "step": 7579 + }, + { + "epoch": 0.8324181858115528, + "grad_norm": 2.0105526447296143, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7119179368019104, + "num_tokens": 194820902.0, + "step": 7580 + }, + { + "epoch": 0.8325280035141664, + "grad_norm": 2.321178674697876, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7091953754425049, + "num_tokens": 194842522.0, + "step": 7581 + }, + { + "epoch": 0.8326378212167801, + "grad_norm": 2.2146427631378174, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7019661664962769, + "num_tokens": 194868061.0, + "step": 7582 + }, + { + "epoch": 0.8327476389193939, + "grad_norm": 2.324000597000122, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7282595634460449, + "num_tokens": 194889942.0, + "step": 7583 + }, + { + "epoch": 0.8328574566220075, + "grad_norm": 1.9899970293045044, + "learning_rate": 1e-06, + "loss": 1.1043, + "mean_token_accuracy": 0.6675571203231812, + "num_tokens": 194921423.0, + "step": 7584 + }, + { + "epoch": 0.8329672743246211, + "grad_norm": 2.2724311351776123, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7261296510696411, + "num_tokens": 194945128.0, + "step": 7585 + }, + { + "epoch": 0.8330770920272348, + "grad_norm": 2.2076668739318848, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.685941219329834, + "num_tokens": 194969012.0, + "step": 7586 + }, + { + "epoch": 0.8331869097298484, + "grad_norm": 2.154562473297119, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6902867555618286, + "num_tokens": 194995259.0, + "step": 7587 + }, + { + "epoch": 0.8332967274324621, + "grad_norm": 2.01342511177063, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6954100131988525, + "num_tokens": 195023738.0, + "step": 7588 + }, + { + "epoch": 0.8334065451350757, + "grad_norm": 2.0445303916931152, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6983531713485718, + "num_tokens": 195052076.0, + "step": 7589 + }, + { + "epoch": 0.8335163628376895, + "grad_norm": 2.032761335372925, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7050133943557739, + "num_tokens": 195080514.0, + "step": 7590 + }, + { + "epoch": 0.8336261805403031, + "grad_norm": 2.1998560428619385, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6974223852157593, + "num_tokens": 195106327.0, + "step": 7591 + }, + { + "epoch": 0.8337359982429168, + "grad_norm": 2.0780844688415527, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7024037837982178, + "num_tokens": 195134493.0, + "step": 7592 + }, + { + "epoch": 0.8338458159455304, + "grad_norm": 2.255640745162964, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7092421054840088, + "num_tokens": 195159766.0, + "step": 7593 + }, + { + "epoch": 0.8339556336481441, + "grad_norm": 2.1785013675689697, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7210675477981567, + "num_tokens": 195183286.0, + "step": 7594 + }, + { + "epoch": 0.8340654513507577, + "grad_norm": 2.267507314682007, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7101199626922607, + "num_tokens": 195207324.0, + "step": 7595 + }, + { + "epoch": 0.8341752690533714, + "grad_norm": 2.000100612640381, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7096948623657227, + "num_tokens": 195235978.0, + "step": 7596 + }, + { + "epoch": 0.8342850867559851, + "grad_norm": 1.848638892173767, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7090928554534912, + "num_tokens": 195267877.0, + "step": 7597 + }, + { + "epoch": 0.8343949044585988, + "grad_norm": 2.023376703262329, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6953227519989014, + "num_tokens": 195296425.0, + "step": 7598 + }, + { + "epoch": 0.8345047221612124, + "grad_norm": 2.502821683883667, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7316829562187195, + "num_tokens": 195316390.0, + "step": 7599 + }, + { + "epoch": 0.834614539863826, + "grad_norm": 2.137516498565674, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7069591879844666, + "num_tokens": 195343458.0, + "step": 7600 + }, + { + "epoch": 0.8347243575664397, + "grad_norm": 2.238645076751709, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.688805878162384, + "num_tokens": 195368684.0, + "step": 7601 + }, + { + "epoch": 0.8348341752690533, + "grad_norm": 2.2949182987213135, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7218669652938843, + "num_tokens": 195391551.0, + "step": 7602 + }, + { + "epoch": 0.834943992971667, + "grad_norm": 2.263355016708374, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7245291471481323, + "num_tokens": 195415519.0, + "step": 7603 + }, + { + "epoch": 0.8350538106742806, + "grad_norm": 2.0280063152313232, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6982178688049316, + "num_tokens": 195444503.0, + "step": 7604 + }, + { + "epoch": 0.8351636283768944, + "grad_norm": 2.210099697113037, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7080409526824951, + "num_tokens": 195468534.0, + "step": 7605 + }, + { + "epoch": 0.835273446079508, + "grad_norm": 2.128732919692993, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7059968709945679, + "num_tokens": 195493672.0, + "step": 7606 + }, + { + "epoch": 0.8353832637821217, + "grad_norm": 2.1667473316192627, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6877265572547913, + "num_tokens": 195521999.0, + "step": 7607 + }, + { + "epoch": 0.8354930814847353, + "grad_norm": 2.269291639328003, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7028437852859497, + "num_tokens": 195547134.0, + "step": 7608 + }, + { + "epoch": 0.835602899187349, + "grad_norm": 2.180659770965576, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6917669773101807, + "num_tokens": 195572598.0, + "step": 7609 + }, + { + "epoch": 0.8357127168899626, + "grad_norm": 2.1564810276031494, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7097229361534119, + "num_tokens": 195599307.0, + "step": 7610 + }, + { + "epoch": 0.8358225345925763, + "grad_norm": 2.011991262435913, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7128205895423889, + "num_tokens": 195626698.0, + "step": 7611 + }, + { + "epoch": 0.83593235229519, + "grad_norm": 2.429821491241455, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7004366517066956, + "num_tokens": 195650178.0, + "step": 7612 + }, + { + "epoch": 0.8360421699978037, + "grad_norm": 2.248690366744995, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6917311549186707, + "num_tokens": 195675463.0, + "step": 7613 + }, + { + "epoch": 0.8361519877004173, + "grad_norm": 2.4984514713287354, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7112991213798523, + "num_tokens": 195696350.0, + "step": 7614 + }, + { + "epoch": 0.836261805403031, + "grad_norm": 2.117145299911499, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7079490423202515, + "num_tokens": 195723923.0, + "step": 7615 + }, + { + "epoch": 0.8363716231056446, + "grad_norm": 2.0081825256347656, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7175569534301758, + "num_tokens": 195753478.0, + "step": 7616 + }, + { + "epoch": 0.8364814408082583, + "grad_norm": 2.2681872844696045, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7075004577636719, + "num_tokens": 195778358.0, + "step": 7617 + }, + { + "epoch": 0.8365912585108719, + "grad_norm": 2.2776315212249756, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7483881711959839, + "num_tokens": 195800560.0, + "step": 7618 + }, + { + "epoch": 0.8367010762134857, + "grad_norm": 2.2263708114624023, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7194390296936035, + "num_tokens": 195823831.0, + "step": 7619 + }, + { + "epoch": 0.8368108939160993, + "grad_norm": 2.109593391418457, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.711857259273529, + "num_tokens": 195850920.0, + "step": 7620 + }, + { + "epoch": 0.836920711618713, + "grad_norm": 2.4204468727111816, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7001296877861023, + "num_tokens": 195873738.0, + "step": 7621 + }, + { + "epoch": 0.8370305293213266, + "grad_norm": 2.160473346710205, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7159169912338257, + "num_tokens": 195899296.0, + "step": 7622 + }, + { + "epoch": 0.8371403470239402, + "grad_norm": 2.3539178371429443, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6985088586807251, + "num_tokens": 195922432.0, + "step": 7623 + }, + { + "epoch": 0.8372501647265539, + "grad_norm": 2.5140531063079834, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7139756083488464, + "num_tokens": 195943706.0, + "step": 7624 + }, + { + "epoch": 0.8373599824291675, + "grad_norm": 2.342310667037964, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7208799719810486, + "num_tokens": 195966445.0, + "step": 7625 + }, + { + "epoch": 0.8374698001317813, + "grad_norm": 2.277372360229492, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7308531403541565, + "num_tokens": 195990005.0, + "step": 7626 + }, + { + "epoch": 0.8375796178343949, + "grad_norm": 2.37180757522583, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.735518217086792, + "num_tokens": 196010109.0, + "step": 7627 + }, + { + "epoch": 0.8376894355370086, + "grad_norm": 2.2118144035339355, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7027122974395752, + "num_tokens": 196035535.0, + "step": 7628 + }, + { + "epoch": 0.8377992532396222, + "grad_norm": 2.0051565170288086, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7387250661849976, + "num_tokens": 196064069.0, + "step": 7629 + }, + { + "epoch": 0.8379090709422359, + "grad_norm": 2.3061940670013428, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7159932851791382, + "num_tokens": 196086736.0, + "step": 7630 + }, + { + "epoch": 0.8380188886448495, + "grad_norm": 2.1794629096984863, + "learning_rate": 1e-06, + "loss": 1.0831, + "mean_token_accuracy": 0.6723348498344421, + "num_tokens": 196113339.0, + "step": 7631 + }, + { + "epoch": 0.8381287063474632, + "grad_norm": 2.2099244594573975, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7244064211845398, + "num_tokens": 196137180.0, + "step": 7632 + }, + { + "epoch": 0.8382385240500769, + "grad_norm": 2.069974899291992, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7190754413604736, + "num_tokens": 196164641.0, + "step": 7633 + }, + { + "epoch": 0.8383483417526906, + "grad_norm": 2.3491854667663574, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7077534794807434, + "num_tokens": 196188288.0, + "step": 7634 + }, + { + "epoch": 0.8384581594553042, + "grad_norm": 2.194300413131714, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7090840339660645, + "num_tokens": 196214918.0, + "step": 7635 + }, + { + "epoch": 0.8385679771579179, + "grad_norm": 1.9588152170181274, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7072458267211914, + "num_tokens": 196245889.0, + "step": 7636 + }, + { + "epoch": 0.8386777948605315, + "grad_norm": 1.9956012964248657, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.705866277217865, + "num_tokens": 196275702.0, + "step": 7637 + }, + { + "epoch": 0.8387876125631452, + "grad_norm": 1.999518632888794, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7160391807556152, + "num_tokens": 196305198.0, + "step": 7638 + }, + { + "epoch": 0.8388974302657588, + "grad_norm": 2.1604692935943604, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7200998067855835, + "num_tokens": 196330411.0, + "step": 7639 + }, + { + "epoch": 0.8390072479683724, + "grad_norm": 2.316911220550537, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.692256510257721, + "num_tokens": 196353302.0, + "step": 7640 + }, + { + "epoch": 0.8391170656709862, + "grad_norm": 2.184464693069458, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7348778247833252, + "num_tokens": 196379234.0, + "step": 7641 + }, + { + "epoch": 0.8392268833735999, + "grad_norm": 2.162794828414917, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7168763875961304, + "num_tokens": 196407153.0, + "step": 7642 + }, + { + "epoch": 0.8393367010762135, + "grad_norm": 2.227510452270508, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7058887481689453, + "num_tokens": 196430761.0, + "step": 7643 + }, + { + "epoch": 0.8394465187788271, + "grad_norm": 2.089142084121704, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6993187665939331, + "num_tokens": 196457437.0, + "step": 7644 + }, + { + "epoch": 0.8395563364814408, + "grad_norm": 2.0591702461242676, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7109319567680359, + "num_tokens": 196486206.0, + "step": 7645 + }, + { + "epoch": 0.8396661541840544, + "grad_norm": 2.292578935623169, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7098990678787231, + "num_tokens": 196509094.0, + "step": 7646 + }, + { + "epoch": 0.8397759718866681, + "grad_norm": 2.071345329284668, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6981199383735657, + "num_tokens": 196537669.0, + "step": 7647 + }, + { + "epoch": 0.8398857895892818, + "grad_norm": 2.1899397373199463, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7082760334014893, + "num_tokens": 196564206.0, + "step": 7648 + }, + { + "epoch": 0.8399956072918955, + "grad_norm": 1.9920132160186768, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6952391862869263, + "num_tokens": 196594236.0, + "step": 7649 + }, + { + "epoch": 0.8401054249945091, + "grad_norm": 2.370814800262451, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7120541334152222, + "num_tokens": 196616872.0, + "step": 7650 + }, + { + "epoch": 0.8402152426971228, + "grad_norm": 2.201498031616211, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7151544094085693, + "num_tokens": 196642953.0, + "step": 7651 + }, + { + "epoch": 0.8403250603997364, + "grad_norm": 2.1744258403778076, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6932365298271179, + "num_tokens": 196670177.0, + "step": 7652 + }, + { + "epoch": 0.8404348781023501, + "grad_norm": 2.2604739665985107, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6991703510284424, + "num_tokens": 196692807.0, + "step": 7653 + }, + { + "epoch": 0.8405446958049637, + "grad_norm": 2.031141757965088, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6933718323707581, + "num_tokens": 196720431.0, + "step": 7654 + }, + { + "epoch": 0.8406545135075775, + "grad_norm": 2.246445894241333, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7499138712882996, + "num_tokens": 196743403.0, + "step": 7655 + }, + { + "epoch": 0.8407643312101911, + "grad_norm": 2.5020437240600586, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7002630233764648, + "num_tokens": 196764418.0, + "step": 7656 + }, + { + "epoch": 0.8408741489128048, + "grad_norm": 2.168321132659912, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6867218017578125, + "num_tokens": 196789903.0, + "step": 7657 + }, + { + "epoch": 0.8409839666154184, + "grad_norm": 2.2249698638916016, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7234505414962769, + "num_tokens": 196812746.0, + "step": 7658 + }, + { + "epoch": 0.8410937843180321, + "grad_norm": 2.1953837871551514, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6933379173278809, + "num_tokens": 196839891.0, + "step": 7659 + }, + { + "epoch": 0.8412036020206457, + "grad_norm": 2.089484930038452, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6951446533203125, + "num_tokens": 196867993.0, + "step": 7660 + }, + { + "epoch": 0.8413134197232593, + "grad_norm": 2.1331639289855957, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6837996244430542, + "num_tokens": 196894717.0, + "step": 7661 + }, + { + "epoch": 0.8414232374258731, + "grad_norm": 2.663141965866089, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7075459361076355, + "num_tokens": 196912608.0, + "step": 7662 + }, + { + "epoch": 0.8415330551284868, + "grad_norm": 2.1424520015716553, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.719946563243866, + "num_tokens": 196940802.0, + "step": 7663 + }, + { + "epoch": 0.8416428728311004, + "grad_norm": 2.3169267177581787, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.711737871170044, + "num_tokens": 196962873.0, + "step": 7664 + }, + { + "epoch": 0.841752690533714, + "grad_norm": 2.100067377090454, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7080207467079163, + "num_tokens": 196989842.0, + "step": 7665 + }, + { + "epoch": 0.8418625082363277, + "grad_norm": 2.2069013118743896, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6981053948402405, + "num_tokens": 197015976.0, + "step": 7666 + }, + { + "epoch": 0.8419723259389413, + "grad_norm": 2.227487325668335, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7198028564453125, + "num_tokens": 197040292.0, + "step": 7667 + }, + { + "epoch": 0.842082143641555, + "grad_norm": 2.4074325561523438, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.692983865737915, + "num_tokens": 197063652.0, + "step": 7668 + }, + { + "epoch": 0.8421919613441686, + "grad_norm": 1.9876813888549805, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7058785557746887, + "num_tokens": 197092955.0, + "step": 7669 + }, + { + "epoch": 0.8423017790467824, + "grad_norm": 2.37101411819458, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.705280065536499, + "num_tokens": 197115656.0, + "step": 7670 + }, + { + "epoch": 0.842411596749396, + "grad_norm": 2.1192526817321777, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7200342416763306, + "num_tokens": 197142108.0, + "step": 7671 + }, + { + "epoch": 0.8425214144520097, + "grad_norm": 2.3372182846069336, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6966691613197327, + "num_tokens": 197166798.0, + "step": 7672 + }, + { + "epoch": 0.8426312321546233, + "grad_norm": 2.0599262714385986, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6980684399604797, + "num_tokens": 197196788.0, + "step": 7673 + }, + { + "epoch": 0.842741049857237, + "grad_norm": 2.433471441268921, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7210423350334167, + "num_tokens": 197219999.0, + "step": 7674 + }, + { + "epoch": 0.8428508675598506, + "grad_norm": 2.1487138271331787, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7404024600982666, + "num_tokens": 197244690.0, + "step": 7675 + }, + { + "epoch": 0.8429606852624643, + "grad_norm": 2.262925386428833, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6940646171569824, + "num_tokens": 197270306.0, + "step": 7676 + }, + { + "epoch": 0.843070502965078, + "grad_norm": 2.481884241104126, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6895289421081543, + "num_tokens": 197291705.0, + "step": 7677 + }, + { + "epoch": 0.8431803206676917, + "grad_norm": 2.183586835861206, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7166755795478821, + "num_tokens": 197315955.0, + "step": 7678 + }, + { + "epoch": 0.8432901383703053, + "grad_norm": 2.0975565910339355, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7108579874038696, + "num_tokens": 197341725.0, + "step": 7679 + }, + { + "epoch": 0.843399956072919, + "grad_norm": 2.093273401260376, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7237039804458618, + "num_tokens": 197366988.0, + "step": 7680 + }, + { + "epoch": 0.8435097737755326, + "grad_norm": 2.397479772567749, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7075353860855103, + "num_tokens": 197387160.0, + "step": 7681 + }, + { + "epoch": 0.8436195914781462, + "grad_norm": 2.2505083084106445, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.736744225025177, + "num_tokens": 197408503.0, + "step": 7682 + }, + { + "epoch": 0.8437294091807599, + "grad_norm": 2.2198398113250732, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7235568165779114, + "num_tokens": 197431485.0, + "step": 7683 + }, + { + "epoch": 0.8438392268833736, + "grad_norm": 1.9483405351638794, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7086295485496521, + "num_tokens": 197461949.0, + "step": 7684 + }, + { + "epoch": 0.8439490445859873, + "grad_norm": 2.2947709560394287, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6950943470001221, + "num_tokens": 197486130.0, + "step": 7685 + }, + { + "epoch": 0.8440588622886009, + "grad_norm": 2.4920880794525146, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7076563835144043, + "num_tokens": 197506587.0, + "step": 7686 + }, + { + "epoch": 0.8441686799912146, + "grad_norm": 2.0764386653900146, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6957021355628967, + "num_tokens": 197534095.0, + "step": 7687 + }, + { + "epoch": 0.8442784976938282, + "grad_norm": 2.28558087348938, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7095004320144653, + "num_tokens": 197557928.0, + "step": 7688 + }, + { + "epoch": 0.8443883153964419, + "grad_norm": 2.183098077774048, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7323123216629028, + "num_tokens": 197583446.0, + "step": 7689 + }, + { + "epoch": 0.8444981330990555, + "grad_norm": 2.418785333633423, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7115369439125061, + "num_tokens": 197603740.0, + "step": 7690 + }, + { + "epoch": 0.8446079508016693, + "grad_norm": 2.3174033164978027, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.683052659034729, + "num_tokens": 197628036.0, + "step": 7691 + }, + { + "epoch": 0.8447177685042829, + "grad_norm": 1.8787243366241455, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6930561065673828, + "num_tokens": 197661270.0, + "step": 7692 + }, + { + "epoch": 0.8448275862068966, + "grad_norm": 1.998430609703064, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.69746994972229, + "num_tokens": 197691305.0, + "step": 7693 + }, + { + "epoch": 0.8449374039095102, + "grad_norm": 2.075922966003418, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.722809374332428, + "num_tokens": 197719102.0, + "step": 7694 + }, + { + "epoch": 0.8450472216121239, + "grad_norm": 2.610147714614868, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7276797294616699, + "num_tokens": 197736523.0, + "step": 7695 + }, + { + "epoch": 0.8451570393147375, + "grad_norm": 2.1110289096832275, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7313844561576843, + "num_tokens": 197762096.0, + "step": 7696 + }, + { + "epoch": 0.8452668570173512, + "grad_norm": 2.375821828842163, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7171638011932373, + "num_tokens": 197784723.0, + "step": 7697 + }, + { + "epoch": 0.8453766747199648, + "grad_norm": 2.1456079483032227, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7156268954277039, + "num_tokens": 197810797.0, + "step": 7698 + }, + { + "epoch": 0.8454864924225786, + "grad_norm": 2.0240283012390137, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6858978271484375, + "num_tokens": 197841786.0, + "step": 7699 + }, + { + "epoch": 0.8455963101251922, + "grad_norm": 2.034787893295288, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7104507684707642, + "num_tokens": 197869434.0, + "step": 7700 + }, + { + "epoch": 0.8457061278278059, + "grad_norm": 2.008890390396118, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6989758014678955, + "num_tokens": 197898412.0, + "step": 7701 + }, + { + "epoch": 0.8458159455304195, + "grad_norm": 2.2059719562530518, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7239983081817627, + "num_tokens": 197922225.0, + "step": 7702 + }, + { + "epoch": 0.8459257632330331, + "grad_norm": 2.0766568183898926, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7131836414337158, + "num_tokens": 197949782.0, + "step": 7703 + }, + { + "epoch": 0.8460355809356468, + "grad_norm": 2.2953877449035645, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7007453441619873, + "num_tokens": 197973380.0, + "step": 7704 + }, + { + "epoch": 0.8461453986382604, + "grad_norm": 1.8638142347335815, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7071382999420166, + "num_tokens": 198007007.0, + "step": 7705 + }, + { + "epoch": 0.8462552163408742, + "grad_norm": 1.7628108263015747, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6914545297622681, + "num_tokens": 198044606.0, + "step": 7706 + }, + { + "epoch": 0.8463650340434878, + "grad_norm": 2.1769723892211914, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7064305543899536, + "num_tokens": 198067772.0, + "step": 7707 + }, + { + "epoch": 0.8464748517461015, + "grad_norm": 1.962612509727478, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.708838701248169, + "num_tokens": 198096717.0, + "step": 7708 + }, + { + "epoch": 0.8465846694487151, + "grad_norm": 2.2669332027435303, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.717402458190918, + "num_tokens": 198119396.0, + "step": 7709 + }, + { + "epoch": 0.8466944871513288, + "grad_norm": 2.164869785308838, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7085446715354919, + "num_tokens": 198145344.0, + "step": 7710 + }, + { + "epoch": 0.8468043048539424, + "grad_norm": 2.1835994720458984, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7157436609268188, + "num_tokens": 198170007.0, + "step": 7711 + }, + { + "epoch": 0.8469141225565561, + "grad_norm": 1.844395399093628, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6975098848342896, + "num_tokens": 198205181.0, + "step": 7712 + }, + { + "epoch": 0.8470239402591698, + "grad_norm": 2.2074341773986816, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.7000396847724915, + "num_tokens": 198230425.0, + "step": 7713 + }, + { + "epoch": 0.8471337579617835, + "grad_norm": 2.141627073287964, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7153900861740112, + "num_tokens": 198254927.0, + "step": 7714 + }, + { + "epoch": 0.8472435756643971, + "grad_norm": 2.271679162979126, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6939687728881836, + "num_tokens": 198280460.0, + "step": 7715 + }, + { + "epoch": 0.8473533933670108, + "grad_norm": 1.995688557624817, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7072153091430664, + "num_tokens": 198310108.0, + "step": 7716 + }, + { + "epoch": 0.8474632110696244, + "grad_norm": 2.347665786743164, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6975253224372864, + "num_tokens": 198335349.0, + "step": 7717 + }, + { + "epoch": 0.8475730287722381, + "grad_norm": 2.131389856338501, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.688583254814148, + "num_tokens": 198362053.0, + "step": 7718 + }, + { + "epoch": 0.8476828464748517, + "grad_norm": 2.025813579559326, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.724326491355896, + "num_tokens": 198390270.0, + "step": 7719 + }, + { + "epoch": 0.8477926641774655, + "grad_norm": 2.110917806625366, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7229434251785278, + "num_tokens": 198416086.0, + "step": 7720 + }, + { + "epoch": 0.8479024818800791, + "grad_norm": 1.9917372465133667, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6883107423782349, + "num_tokens": 198446777.0, + "step": 7721 + }, + { + "epoch": 0.8480122995826928, + "grad_norm": 2.305565595626831, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7224557399749756, + "num_tokens": 198468112.0, + "step": 7722 + }, + { + "epoch": 0.8481221172853064, + "grad_norm": 2.195803165435791, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6926955580711365, + "num_tokens": 198495083.0, + "step": 7723 + }, + { + "epoch": 0.84823193498792, + "grad_norm": 2.0826687812805176, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7022897601127625, + "num_tokens": 198520765.0, + "step": 7724 + }, + { + "epoch": 0.8483417526905337, + "grad_norm": 2.1411869525909424, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7031705379486084, + "num_tokens": 198547154.0, + "step": 7725 + }, + { + "epoch": 0.8484515703931473, + "grad_norm": 2.3601725101470947, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7015394568443298, + "num_tokens": 198570225.0, + "step": 7726 + }, + { + "epoch": 0.848561388095761, + "grad_norm": 2.1537058353424072, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6792299747467041, + "num_tokens": 198598465.0, + "step": 7727 + }, + { + "epoch": 0.8486712057983747, + "grad_norm": 2.0364038944244385, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7025284171104431, + "num_tokens": 198628156.0, + "step": 7728 + }, + { + "epoch": 0.8487810235009884, + "grad_norm": 2.1417899131774902, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6997413039207458, + "num_tokens": 198654781.0, + "step": 7729 + }, + { + "epoch": 0.848890841203602, + "grad_norm": 2.0385501384735107, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7080525159835815, + "num_tokens": 198684309.0, + "step": 7730 + }, + { + "epoch": 0.8490006589062157, + "grad_norm": 2.0409796237945557, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6881054639816284, + "num_tokens": 198713552.0, + "step": 7731 + }, + { + "epoch": 0.8491104766088293, + "grad_norm": 2.262261152267456, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7201616764068604, + "num_tokens": 198736239.0, + "step": 7732 + }, + { + "epoch": 0.849220294311443, + "grad_norm": 2.498762845993042, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7063965201377869, + "num_tokens": 198756469.0, + "step": 7733 + }, + { + "epoch": 0.8493301120140566, + "grad_norm": 2.4873900413513184, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.716387927532196, + "num_tokens": 198775780.0, + "step": 7734 + }, + { + "epoch": 0.8494399297166704, + "grad_norm": 2.1439077854156494, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6802675127983093, + "num_tokens": 198803268.0, + "step": 7735 + }, + { + "epoch": 0.849549747419284, + "grad_norm": 2.093101739883423, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.696906566619873, + "num_tokens": 198830223.0, + "step": 7736 + }, + { + "epoch": 0.8496595651218977, + "grad_norm": 2.17682147026062, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6937484741210938, + "num_tokens": 198855994.0, + "step": 7737 + }, + { + "epoch": 0.8497693828245113, + "grad_norm": 2.441270112991333, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7325936555862427, + "num_tokens": 198875422.0, + "step": 7738 + }, + { + "epoch": 0.849879200527125, + "grad_norm": 2.231658458709717, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7424218654632568, + "num_tokens": 198898011.0, + "step": 7739 + }, + { + "epoch": 0.8499890182297386, + "grad_norm": 2.002084493637085, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6944063305854797, + "num_tokens": 198927467.0, + "step": 7740 + }, + { + "epoch": 0.8500988359323522, + "grad_norm": 2.325183629989624, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7056611180305481, + "num_tokens": 198950633.0, + "step": 7741 + }, + { + "epoch": 0.850208653634966, + "grad_norm": 2.2292842864990234, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7304606437683105, + "num_tokens": 198972157.0, + "step": 7742 + }, + { + "epoch": 0.8503184713375797, + "grad_norm": 2.0002079010009766, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7185877561569214, + "num_tokens": 199001268.0, + "step": 7743 + }, + { + "epoch": 0.8504282890401933, + "grad_norm": 2.0444815158843994, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7252287864685059, + "num_tokens": 199032260.0, + "step": 7744 + }, + { + "epoch": 0.8505381067428069, + "grad_norm": 2.3097784519195557, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6860321164131165, + "num_tokens": 199056214.0, + "step": 7745 + }, + { + "epoch": 0.8506479244454206, + "grad_norm": 1.9033751487731934, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6901484727859497, + "num_tokens": 199087562.0, + "step": 7746 + }, + { + "epoch": 0.8507577421480342, + "grad_norm": 2.3849380016326904, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.706757128238678, + "num_tokens": 199108439.0, + "step": 7747 + }, + { + "epoch": 0.8508675598506479, + "grad_norm": 2.2065088748931885, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7270777225494385, + "num_tokens": 199130743.0, + "step": 7748 + }, + { + "epoch": 0.8509773775532616, + "grad_norm": 2.576932430267334, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7096686959266663, + "num_tokens": 199148952.0, + "step": 7749 + }, + { + "epoch": 0.8510871952558753, + "grad_norm": 1.9665802717208862, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7123714685440063, + "num_tokens": 199178659.0, + "step": 7750 + }, + { + "epoch": 0.8511970129584889, + "grad_norm": 2.181386947631836, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.69551682472229, + "num_tokens": 199206413.0, + "step": 7751 + }, + { + "epoch": 0.8513068306611026, + "grad_norm": 2.2893199920654297, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.6997241973876953, + "num_tokens": 199230459.0, + "step": 7752 + }, + { + "epoch": 0.8514166483637162, + "grad_norm": 1.9309016466140747, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.680824875831604, + "num_tokens": 199263823.0, + "step": 7753 + }, + { + "epoch": 0.8515264660663299, + "grad_norm": 2.229433536529541, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7023478150367737, + "num_tokens": 199289460.0, + "step": 7754 + }, + { + "epoch": 0.8516362837689435, + "grad_norm": 2.6163599491119385, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7273625135421753, + "num_tokens": 199308106.0, + "step": 7755 + }, + { + "epoch": 0.8517461014715572, + "grad_norm": 2.3619847297668457, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7230183482170105, + "num_tokens": 199328866.0, + "step": 7756 + }, + { + "epoch": 0.8518559191741709, + "grad_norm": 2.1367948055267334, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7245228290557861, + "num_tokens": 199354657.0, + "step": 7757 + }, + { + "epoch": 0.8519657368767846, + "grad_norm": 2.096975088119507, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.703538179397583, + "num_tokens": 199382856.0, + "step": 7758 + }, + { + "epoch": 0.8520755545793982, + "grad_norm": 2.5188560485839844, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6977978348731995, + "num_tokens": 199405071.0, + "step": 7759 + }, + { + "epoch": 0.8521853722820119, + "grad_norm": 2.1525018215179443, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7031704187393188, + "num_tokens": 199434812.0, + "step": 7760 + }, + { + "epoch": 0.8522951899846255, + "grad_norm": 2.1672418117523193, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6912757158279419, + "num_tokens": 199461202.0, + "step": 7761 + }, + { + "epoch": 0.8524050076872391, + "grad_norm": 2.2966506481170654, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6750196218490601, + "num_tokens": 199486547.0, + "step": 7762 + }, + { + "epoch": 0.8525148253898528, + "grad_norm": 2.143639087677002, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.707868218421936, + "num_tokens": 199511105.0, + "step": 7763 + }, + { + "epoch": 0.8526246430924665, + "grad_norm": 2.20878005027771, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7018598318099976, + "num_tokens": 199535688.0, + "step": 7764 + }, + { + "epoch": 0.8527344607950802, + "grad_norm": 2.160727024078369, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.703108012676239, + "num_tokens": 199560518.0, + "step": 7765 + }, + { + "epoch": 0.8528442784976938, + "grad_norm": 2.276550054550171, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.683964192867279, + "num_tokens": 199586667.0, + "step": 7766 + }, + { + "epoch": 0.8529540962003075, + "grad_norm": 2.5070419311523438, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.694747805595398, + "num_tokens": 199607975.0, + "step": 7767 + }, + { + "epoch": 0.8530639139029211, + "grad_norm": 2.2208738327026367, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6762310266494751, + "num_tokens": 199636329.0, + "step": 7768 + }, + { + "epoch": 0.8531737316055348, + "grad_norm": 2.4242358207702637, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7178540229797363, + "num_tokens": 199657991.0, + "step": 7769 + }, + { + "epoch": 0.8532835493081484, + "grad_norm": 2.1670877933502197, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.720282793045044, + "num_tokens": 199682143.0, + "step": 7770 + }, + { + "epoch": 0.8533933670107622, + "grad_norm": 2.265143632888794, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7190074920654297, + "num_tokens": 199705190.0, + "step": 7771 + }, + { + "epoch": 0.8535031847133758, + "grad_norm": 2.0632495880126953, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7099683880805969, + "num_tokens": 199735218.0, + "step": 7772 + }, + { + "epoch": 0.8536130024159895, + "grad_norm": 2.0949509143829346, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7327625751495361, + "num_tokens": 199762252.0, + "step": 7773 + }, + { + "epoch": 0.8537228201186031, + "grad_norm": 2.3028972148895264, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7110320329666138, + "num_tokens": 199786367.0, + "step": 7774 + }, + { + "epoch": 0.8538326378212168, + "grad_norm": 2.0548946857452393, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7221949696540833, + "num_tokens": 199815536.0, + "step": 7775 + }, + { + "epoch": 0.8539424555238304, + "grad_norm": 2.277162551879883, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6976681351661682, + "num_tokens": 199839093.0, + "step": 7776 + }, + { + "epoch": 0.8540522732264441, + "grad_norm": 2.1592156887054443, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7142370939254761, + "num_tokens": 199864532.0, + "step": 7777 + }, + { + "epoch": 0.8541620909290578, + "grad_norm": 2.2446022033691406, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.749059796333313, + "num_tokens": 199886158.0, + "step": 7778 + }, + { + "epoch": 0.8542719086316715, + "grad_norm": 2.230660915374756, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6883914470672607, + "num_tokens": 199910959.0, + "step": 7779 + }, + { + "epoch": 0.8543817263342851, + "grad_norm": 1.9650752544403076, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7035119533538818, + "num_tokens": 199942793.0, + "step": 7780 + }, + { + "epoch": 0.8544915440368988, + "grad_norm": 2.0857081413269043, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6966330409049988, + "num_tokens": 199971918.0, + "step": 7781 + }, + { + "epoch": 0.8546013617395124, + "grad_norm": 1.981433629989624, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6987623572349548, + "num_tokens": 199999128.0, + "step": 7782 + }, + { + "epoch": 0.854711179442126, + "grad_norm": 2.0232303142547607, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7134666442871094, + "num_tokens": 200026400.0, + "step": 7783 + }, + { + "epoch": 0.8548209971447397, + "grad_norm": 2.0543313026428223, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6918138265609741, + "num_tokens": 200054295.0, + "step": 7784 + }, + { + "epoch": 0.8549308148473534, + "grad_norm": 2.3783533573150635, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7220498323440552, + "num_tokens": 200074975.0, + "step": 7785 + }, + { + "epoch": 0.8550406325499671, + "grad_norm": 2.491548538208008, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7007383108139038, + "num_tokens": 200096380.0, + "step": 7786 + }, + { + "epoch": 0.8551504502525807, + "grad_norm": 2.1095573902130127, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.7055727243423462, + "num_tokens": 200122535.0, + "step": 7787 + }, + { + "epoch": 0.8552602679551944, + "grad_norm": 2.011837959289551, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7059186697006226, + "num_tokens": 200152758.0, + "step": 7788 + }, + { + "epoch": 0.855370085657808, + "grad_norm": 2.13901424407959, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7113968729972839, + "num_tokens": 200179132.0, + "step": 7789 + }, + { + "epoch": 0.8554799033604217, + "grad_norm": 2.09954571723938, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7098318338394165, + "num_tokens": 200204720.0, + "step": 7790 + }, + { + "epoch": 0.8555897210630353, + "grad_norm": 2.1344943046569824, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7175649404525757, + "num_tokens": 200228409.0, + "step": 7791 + }, + { + "epoch": 0.855699538765649, + "grad_norm": 2.327188730239868, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7206928730010986, + "num_tokens": 200250436.0, + "step": 7792 + }, + { + "epoch": 0.8558093564682627, + "grad_norm": 2.1004552841186523, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7073712348937988, + "num_tokens": 200277097.0, + "step": 7793 + }, + { + "epoch": 0.8559191741708764, + "grad_norm": 2.163724899291992, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.691452145576477, + "num_tokens": 200300771.0, + "step": 7794 + }, + { + "epoch": 0.85602899187349, + "grad_norm": 1.8908699750900269, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7086427211761475, + "num_tokens": 200333313.0, + "step": 7795 + }, + { + "epoch": 0.8561388095761037, + "grad_norm": 2.4152259826660156, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7002118825912476, + "num_tokens": 200356648.0, + "step": 7796 + }, + { + "epoch": 0.8562486272787173, + "grad_norm": 2.4403140544891357, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6980330944061279, + "num_tokens": 200378508.0, + "step": 7797 + }, + { + "epoch": 0.856358444981331, + "grad_norm": 2.3925182819366455, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.719223141670227, + "num_tokens": 200399174.0, + "step": 7798 + }, + { + "epoch": 0.8564682626839446, + "grad_norm": 2.621828317642212, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.721957802772522, + "num_tokens": 200417510.0, + "step": 7799 + }, + { + "epoch": 0.8565780803865584, + "grad_norm": 2.5510141849517822, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7304420471191406, + "num_tokens": 200434947.0, + "step": 7800 + }, + { + "epoch": 0.856687898089172, + "grad_norm": 2.372410297393799, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7236883044242859, + "num_tokens": 200457329.0, + "step": 7801 + }, + { + "epoch": 0.8567977157917857, + "grad_norm": 2.075232744216919, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6809201240539551, + "num_tokens": 200485775.0, + "step": 7802 + }, + { + "epoch": 0.8569075334943993, + "grad_norm": 2.3827755451202393, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7088801860809326, + "num_tokens": 200508076.0, + "step": 7803 + }, + { + "epoch": 0.8570173511970129, + "grad_norm": 2.179347276687622, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7161334753036499, + "num_tokens": 200534742.0, + "step": 7804 + }, + { + "epoch": 0.8571271688996266, + "grad_norm": 2.205956220626831, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6905887722969055, + "num_tokens": 200559077.0, + "step": 7805 + }, + { + "epoch": 0.8572369866022402, + "grad_norm": 2.0037145614624023, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7032755613327026, + "num_tokens": 200587700.0, + "step": 7806 + }, + { + "epoch": 0.857346804304854, + "grad_norm": 2.123744010925293, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.700699508190155, + "num_tokens": 200613616.0, + "step": 7807 + }, + { + "epoch": 0.8574566220074676, + "grad_norm": 2.1525471210479736, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7037352919578552, + "num_tokens": 200641053.0, + "step": 7808 + }, + { + "epoch": 0.8575664397100813, + "grad_norm": 1.9565049409866333, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6950327157974243, + "num_tokens": 200671586.0, + "step": 7809 + }, + { + "epoch": 0.8576762574126949, + "grad_norm": 2.1183054447174072, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.724791944026947, + "num_tokens": 200698041.0, + "step": 7810 + }, + { + "epoch": 0.8577860751153086, + "grad_norm": 2.366424798965454, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7163835763931274, + "num_tokens": 200719620.0, + "step": 7811 + }, + { + "epoch": 0.8578958928179222, + "grad_norm": 2.2524969577789307, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.711582601070404, + "num_tokens": 200743515.0, + "step": 7812 + }, + { + "epoch": 0.8580057105205359, + "grad_norm": 2.26012921333313, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7236377000808716, + "num_tokens": 200766346.0, + "step": 7813 + }, + { + "epoch": 0.8581155282231496, + "grad_norm": 2.2758703231811523, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.687889039516449, + "num_tokens": 200794763.0, + "step": 7814 + }, + { + "epoch": 0.8582253459257633, + "grad_norm": 2.2648208141326904, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6982074975967407, + "num_tokens": 200818034.0, + "step": 7815 + }, + { + "epoch": 0.8583351636283769, + "grad_norm": 2.0263264179229736, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.698525071144104, + "num_tokens": 200848956.0, + "step": 7816 + }, + { + "epoch": 0.8584449813309906, + "grad_norm": 2.15970516204834, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6996995806694031, + "num_tokens": 200874100.0, + "step": 7817 + }, + { + "epoch": 0.8585547990336042, + "grad_norm": 2.0897397994995117, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7457467913627625, + "num_tokens": 200898969.0, + "step": 7818 + }, + { + "epoch": 0.8586646167362179, + "grad_norm": 2.2047781944274902, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7145297527313232, + "num_tokens": 200925066.0, + "step": 7819 + }, + { + "epoch": 0.8587744344388315, + "grad_norm": 2.1418590545654297, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7162693738937378, + "num_tokens": 200951901.0, + "step": 7820 + }, + { + "epoch": 0.8588842521414451, + "grad_norm": 2.287895679473877, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7055614590644836, + "num_tokens": 200975245.0, + "step": 7821 + }, + { + "epoch": 0.8589940698440589, + "grad_norm": 2.103748083114624, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6854003071784973, + "num_tokens": 201002112.0, + "step": 7822 + }, + { + "epoch": 0.8591038875466726, + "grad_norm": 2.1758780479431152, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7177525758743286, + "num_tokens": 201027125.0, + "step": 7823 + }, + { + "epoch": 0.8592137052492862, + "grad_norm": 2.265428066253662, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7023561000823975, + "num_tokens": 201050466.0, + "step": 7824 + }, + { + "epoch": 0.8593235229518998, + "grad_norm": 2.24688458442688, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7057801485061646, + "num_tokens": 201075834.0, + "step": 7825 + }, + { + "epoch": 0.8594333406545135, + "grad_norm": 2.2677087783813477, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6999152898788452, + "num_tokens": 201100803.0, + "step": 7826 + }, + { + "epoch": 0.8595431583571271, + "grad_norm": 1.9952589273452759, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7151799201965332, + "num_tokens": 201128186.0, + "step": 7827 + }, + { + "epoch": 0.8596529760597408, + "grad_norm": 2.0555195808410645, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.710269570350647, + "num_tokens": 201154254.0, + "step": 7828 + }, + { + "epoch": 0.8597627937623545, + "grad_norm": 2.347064733505249, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7479766607284546, + "num_tokens": 201174622.0, + "step": 7829 + }, + { + "epoch": 0.8598726114649682, + "grad_norm": 2.2246432304382324, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7036628723144531, + "num_tokens": 201199252.0, + "step": 7830 + }, + { + "epoch": 0.8599824291675818, + "grad_norm": 2.1180269718170166, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7230669260025024, + "num_tokens": 201225051.0, + "step": 7831 + }, + { + "epoch": 0.8600922468701955, + "grad_norm": 2.0406312942504883, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6912339329719543, + "num_tokens": 201254984.0, + "step": 7832 + }, + { + "epoch": 0.8602020645728091, + "grad_norm": 2.10491943359375, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6908402442932129, + "num_tokens": 201282032.0, + "step": 7833 + }, + { + "epoch": 0.8603118822754228, + "grad_norm": 2.2959022521972656, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7067651748657227, + "num_tokens": 201305484.0, + "step": 7834 + }, + { + "epoch": 0.8604216999780364, + "grad_norm": 2.2332394123077393, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.700636088848114, + "num_tokens": 201329114.0, + "step": 7835 + }, + { + "epoch": 0.8605315176806502, + "grad_norm": 2.178287982940674, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7083550691604614, + "num_tokens": 201355190.0, + "step": 7836 + }, + { + "epoch": 0.8606413353832638, + "grad_norm": 2.1747400760650635, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7030797600746155, + "num_tokens": 201380838.0, + "step": 7837 + }, + { + "epoch": 0.8607511530858775, + "grad_norm": 2.182720184326172, + "learning_rate": 1e-06, + "loss": 1.0647, + "mean_token_accuracy": 0.6765069961547852, + "num_tokens": 201410214.0, + "step": 7838 + }, + { + "epoch": 0.8608609707884911, + "grad_norm": 1.957485556602478, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7028745412826538, + "num_tokens": 201440711.0, + "step": 7839 + }, + { + "epoch": 0.8609707884911048, + "grad_norm": 1.9257653951644897, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6791088581085205, + "num_tokens": 201473301.0, + "step": 7840 + }, + { + "epoch": 0.8610806061937184, + "grad_norm": 2.0736913681030273, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6945376992225647, + "num_tokens": 201501924.0, + "step": 7841 + }, + { + "epoch": 0.861190423896332, + "grad_norm": 2.1440021991729736, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7120164036750793, + "num_tokens": 201529421.0, + "step": 7842 + }, + { + "epoch": 0.8613002415989458, + "grad_norm": 2.368312358856201, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7401308417320251, + "num_tokens": 201550242.0, + "step": 7843 + }, + { + "epoch": 0.8614100593015594, + "grad_norm": 2.0183050632476807, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7022990584373474, + "num_tokens": 201578456.0, + "step": 7844 + }, + { + "epoch": 0.8615198770041731, + "grad_norm": 2.329646587371826, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7102164626121521, + "num_tokens": 201600752.0, + "step": 7845 + }, + { + "epoch": 0.8616296947067867, + "grad_norm": 2.3930246829986572, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7207154035568237, + "num_tokens": 201622079.0, + "step": 7846 + }, + { + "epoch": 0.8617395124094004, + "grad_norm": 2.4568612575531006, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7142676115036011, + "num_tokens": 201642509.0, + "step": 7847 + }, + { + "epoch": 0.861849330112014, + "grad_norm": 2.141437292098999, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6981197595596313, + "num_tokens": 201670149.0, + "step": 7848 + }, + { + "epoch": 0.8619591478146277, + "grad_norm": 2.36607027053833, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7086343169212341, + "num_tokens": 201691974.0, + "step": 7849 + }, + { + "epoch": 0.8620689655172413, + "grad_norm": 1.9622095823287964, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.700654149055481, + "num_tokens": 201723247.0, + "step": 7850 + }, + { + "epoch": 0.8621787832198551, + "grad_norm": 1.9217671155929565, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6972278356552124, + "num_tokens": 201754160.0, + "step": 7851 + }, + { + "epoch": 0.8622886009224687, + "grad_norm": 2.3495075702667236, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7198563814163208, + "num_tokens": 201776501.0, + "step": 7852 + }, + { + "epoch": 0.8623984186250824, + "grad_norm": 2.2239322662353516, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7201838493347168, + "num_tokens": 201799920.0, + "step": 7853 + }, + { + "epoch": 0.862508236327696, + "grad_norm": 2.2151901721954346, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7055448293685913, + "num_tokens": 201824998.0, + "step": 7854 + }, + { + "epoch": 0.8626180540303097, + "grad_norm": 2.3335461616516113, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7227472066879272, + "num_tokens": 201846908.0, + "step": 7855 + }, + { + "epoch": 0.8627278717329233, + "grad_norm": 2.2953672409057617, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7143827676773071, + "num_tokens": 201869410.0, + "step": 7856 + }, + { + "epoch": 0.862837689435537, + "grad_norm": 2.145063877105713, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7103170156478882, + "num_tokens": 201895473.0, + "step": 7857 + }, + { + "epoch": 0.8629475071381507, + "grad_norm": 2.176525831222534, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7107054591178894, + "num_tokens": 201919229.0, + "step": 7858 + }, + { + "epoch": 0.8630573248407644, + "grad_norm": 2.481297492980957, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7093454599380493, + "num_tokens": 201941552.0, + "step": 7859 + }, + { + "epoch": 0.863167142543378, + "grad_norm": 1.946250557899475, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7164772748947144, + "num_tokens": 201969025.0, + "step": 7860 + }, + { + "epoch": 0.8632769602459917, + "grad_norm": 2.458580732345581, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7287362813949585, + "num_tokens": 201989523.0, + "step": 7861 + }, + { + "epoch": 0.8633867779486053, + "grad_norm": 2.505002975463867, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6961300373077393, + "num_tokens": 202010397.0, + "step": 7862 + }, + { + "epoch": 0.863496595651219, + "grad_norm": 2.3983302116394043, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7247927188873291, + "num_tokens": 202032680.0, + "step": 7863 + }, + { + "epoch": 0.8636064133538326, + "grad_norm": 2.3263652324676514, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7284577488899231, + "num_tokens": 202054513.0, + "step": 7864 + }, + { + "epoch": 0.8637162310564463, + "grad_norm": 2.1754677295684814, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6947367787361145, + "num_tokens": 202080512.0, + "step": 7865 + }, + { + "epoch": 0.86382604875906, + "grad_norm": 2.1547811031341553, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.695502758026123, + "num_tokens": 202106990.0, + "step": 7866 + }, + { + "epoch": 0.8639358664616736, + "grad_norm": 2.2318620681762695, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7161036133766174, + "num_tokens": 202130635.0, + "step": 7867 + }, + { + "epoch": 0.8640456841642873, + "grad_norm": 2.6458890438079834, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6927136182785034, + "num_tokens": 202149583.0, + "step": 7868 + }, + { + "epoch": 0.8641555018669009, + "grad_norm": 2.2142527103424072, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6943740844726562, + "num_tokens": 202176608.0, + "step": 7869 + }, + { + "epoch": 0.8642653195695146, + "grad_norm": 2.141676425933838, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6909699440002441, + "num_tokens": 202203218.0, + "step": 7870 + }, + { + "epoch": 0.8643751372721282, + "grad_norm": 2.1006336212158203, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.73243248462677, + "num_tokens": 202228422.0, + "step": 7871 + }, + { + "epoch": 0.864484954974742, + "grad_norm": 2.1777637004852295, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7476460933685303, + "num_tokens": 202251175.0, + "step": 7872 + }, + { + "epoch": 0.8645947726773556, + "grad_norm": 2.0696327686309814, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6926213502883911, + "num_tokens": 202278271.0, + "step": 7873 + }, + { + "epoch": 0.8647045903799693, + "grad_norm": 2.0329952239990234, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7212706804275513, + "num_tokens": 202304175.0, + "step": 7874 + }, + { + "epoch": 0.8648144080825829, + "grad_norm": 2.4428441524505615, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.697832465171814, + "num_tokens": 202323664.0, + "step": 7875 + }, + { + "epoch": 0.8649242257851966, + "grad_norm": 2.283810615539551, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7139464616775513, + "num_tokens": 202346898.0, + "step": 7876 + }, + { + "epoch": 0.8650340434878102, + "grad_norm": 2.159130573272705, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7251126766204834, + "num_tokens": 202371808.0, + "step": 7877 + }, + { + "epoch": 0.8651438611904239, + "grad_norm": 2.2377264499664307, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7265706658363342, + "num_tokens": 202394980.0, + "step": 7878 + }, + { + "epoch": 0.8652536788930375, + "grad_norm": 2.195016860961914, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6875675320625305, + "num_tokens": 202421850.0, + "step": 7879 + }, + { + "epoch": 0.8653634965956513, + "grad_norm": 2.019782543182373, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6917755007743835, + "num_tokens": 202452962.0, + "step": 7880 + }, + { + "epoch": 0.8654733142982649, + "grad_norm": 2.267488956451416, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7031601071357727, + "num_tokens": 202477114.0, + "step": 7881 + }, + { + "epoch": 0.8655831320008786, + "grad_norm": 2.0054287910461426, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7023116946220398, + "num_tokens": 202505569.0, + "step": 7882 + }, + { + "epoch": 0.8656929497034922, + "grad_norm": 2.1321470737457275, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.699592113494873, + "num_tokens": 202531924.0, + "step": 7883 + }, + { + "epoch": 0.8658027674061058, + "grad_norm": 2.4266791343688965, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6860352754592896, + "num_tokens": 202553976.0, + "step": 7884 + }, + { + "epoch": 0.8659125851087195, + "grad_norm": 1.7782390117645264, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7071628570556641, + "num_tokens": 202588607.0, + "step": 7885 + }, + { + "epoch": 0.8660224028113331, + "grad_norm": 2.0403287410736084, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6955877542495728, + "num_tokens": 202617434.0, + "step": 7886 + }, + { + "epoch": 0.8661322205139469, + "grad_norm": 2.1199798583984375, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6795167326927185, + "num_tokens": 202647024.0, + "step": 7887 + }, + { + "epoch": 0.8662420382165605, + "grad_norm": 1.9919910430908203, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6991820335388184, + "num_tokens": 202679238.0, + "step": 7888 + }, + { + "epoch": 0.8663518559191742, + "grad_norm": 1.9694454669952393, + "learning_rate": 1e-06, + "loss": 1.0866, + "mean_token_accuracy": 0.6792430281639099, + "num_tokens": 202709725.0, + "step": 7889 + }, + { + "epoch": 0.8664616736217878, + "grad_norm": 2.2601590156555176, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7269446849822998, + "num_tokens": 202732564.0, + "step": 7890 + }, + { + "epoch": 0.8665714913244015, + "grad_norm": 2.2510128021240234, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.694198727607727, + "num_tokens": 202757552.0, + "step": 7891 + }, + { + "epoch": 0.8666813090270151, + "grad_norm": 1.9555929899215698, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7304073572158813, + "num_tokens": 202785507.0, + "step": 7892 + }, + { + "epoch": 0.8667911267296288, + "grad_norm": 2.2233426570892334, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7096633911132812, + "num_tokens": 202808537.0, + "step": 7893 + }, + { + "epoch": 0.8669009444322425, + "grad_norm": 2.0297138690948486, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.7016646862030029, + "num_tokens": 202838618.0, + "step": 7894 + }, + { + "epoch": 0.8670107621348562, + "grad_norm": 2.245765209197998, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7243525981903076, + "num_tokens": 202862828.0, + "step": 7895 + }, + { + "epoch": 0.8671205798374698, + "grad_norm": 2.280513286590576, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6940150856971741, + "num_tokens": 202887457.0, + "step": 7896 + }, + { + "epoch": 0.8672303975400835, + "grad_norm": 1.9665732383728027, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6939279437065125, + "num_tokens": 202917912.0, + "step": 7897 + }, + { + "epoch": 0.8673402152426971, + "grad_norm": 1.9884785413742065, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6962780356407166, + "num_tokens": 202952090.0, + "step": 7898 + }, + { + "epoch": 0.8674500329453108, + "grad_norm": 2.254706859588623, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6815625429153442, + "num_tokens": 202976938.0, + "step": 7899 + }, + { + "epoch": 0.8675598506479244, + "grad_norm": 2.1781005859375, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7183499336242676, + "num_tokens": 203001297.0, + "step": 7900 + }, + { + "epoch": 0.8676696683505382, + "grad_norm": 2.2605724334716797, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7211672067642212, + "num_tokens": 203025621.0, + "step": 7901 + }, + { + "epoch": 0.8677794860531518, + "grad_norm": 2.3495428562164307, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.7030742764472961, + "num_tokens": 203048771.0, + "step": 7902 + }, + { + "epoch": 0.8678893037557655, + "grad_norm": 2.2855637073516846, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7235113978385925, + "num_tokens": 203072205.0, + "step": 7903 + }, + { + "epoch": 0.8679991214583791, + "grad_norm": 1.9326980113983154, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6953033804893494, + "num_tokens": 203103741.0, + "step": 7904 + }, + { + "epoch": 0.8681089391609927, + "grad_norm": 2.34366774559021, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6931566596031189, + "num_tokens": 203127573.0, + "step": 7905 + }, + { + "epoch": 0.8682187568636064, + "grad_norm": 2.3812921047210693, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7009543776512146, + "num_tokens": 203150021.0, + "step": 7906 + }, + { + "epoch": 0.86832857456622, + "grad_norm": 2.0213754177093506, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6864396333694458, + "num_tokens": 203180159.0, + "step": 7907 + }, + { + "epoch": 0.8684383922688337, + "grad_norm": 1.8703131675720215, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7116420865058899, + "num_tokens": 203212252.0, + "step": 7908 + }, + { + "epoch": 0.8685482099714474, + "grad_norm": 2.129899263381958, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6898923516273499, + "num_tokens": 203238884.0, + "step": 7909 + }, + { + "epoch": 0.8686580276740611, + "grad_norm": 2.7325403690338135, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7042794823646545, + "num_tokens": 203259431.0, + "step": 7910 + }, + { + "epoch": 0.8687678453766747, + "grad_norm": 2.1024253368377686, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7138001918792725, + "num_tokens": 203285204.0, + "step": 7911 + }, + { + "epoch": 0.8688776630792884, + "grad_norm": 2.4237313270568848, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.712007999420166, + "num_tokens": 203305833.0, + "step": 7912 + }, + { + "epoch": 0.868987480781902, + "grad_norm": 1.9798986911773682, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7012818455696106, + "num_tokens": 203336241.0, + "step": 7913 + }, + { + "epoch": 0.8690972984845157, + "grad_norm": 2.2592360973358154, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.706509530544281, + "num_tokens": 203361027.0, + "step": 7914 + }, + { + "epoch": 0.8692071161871293, + "grad_norm": 2.2943692207336426, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7240259647369385, + "num_tokens": 203385642.0, + "step": 7915 + }, + { + "epoch": 0.8693169338897431, + "grad_norm": 1.9747893810272217, + "learning_rate": 1e-06, + "loss": 1.075, + "mean_token_accuracy": 0.6747447848320007, + "num_tokens": 203416518.0, + "step": 7916 + }, + { + "epoch": 0.8694267515923567, + "grad_norm": 2.4083619117736816, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7014431953430176, + "num_tokens": 203441488.0, + "step": 7917 + }, + { + "epoch": 0.8695365692949704, + "grad_norm": 1.957121729850769, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6979157328605652, + "num_tokens": 203472639.0, + "step": 7918 + }, + { + "epoch": 0.869646386997584, + "grad_norm": 2.1165425777435303, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7048764228820801, + "num_tokens": 203500469.0, + "step": 7919 + }, + { + "epoch": 0.8697562047001977, + "grad_norm": 2.1527953147888184, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6948563456535339, + "num_tokens": 203527035.0, + "step": 7920 + }, + { + "epoch": 0.8698660224028113, + "grad_norm": 2.077975273132324, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7267820835113525, + "num_tokens": 203553509.0, + "step": 7921 + }, + { + "epoch": 0.869975840105425, + "grad_norm": 2.0594544410705566, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.7041972875595093, + "num_tokens": 203580788.0, + "step": 7922 + }, + { + "epoch": 0.8700856578080387, + "grad_norm": 2.2340991497039795, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7116620540618896, + "num_tokens": 203604013.0, + "step": 7923 + }, + { + "epoch": 0.8701954755106523, + "grad_norm": 2.2805702686309814, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7232377529144287, + "num_tokens": 203627357.0, + "step": 7924 + }, + { + "epoch": 0.870305293213266, + "grad_norm": 2.116483449935913, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6867231726646423, + "num_tokens": 203654163.0, + "step": 7925 + }, + { + "epoch": 0.8704151109158796, + "grad_norm": 2.35290789604187, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7266295552253723, + "num_tokens": 203677750.0, + "step": 7926 + }, + { + "epoch": 0.8705249286184933, + "grad_norm": 2.376476526260376, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.7035989761352539, + "num_tokens": 203701870.0, + "step": 7927 + }, + { + "epoch": 0.8706347463211069, + "grad_norm": 1.9920806884765625, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6971957087516785, + "num_tokens": 203731336.0, + "step": 7928 + }, + { + "epoch": 0.8707445640237206, + "grad_norm": 2.3578219413757324, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7063586711883545, + "num_tokens": 203752779.0, + "step": 7929 + }, + { + "epoch": 0.8708543817263343, + "grad_norm": 2.328057289123535, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7238884568214417, + "num_tokens": 203776193.0, + "step": 7930 + }, + { + "epoch": 0.870964199428948, + "grad_norm": 2.076037883758545, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7115273475646973, + "num_tokens": 203803587.0, + "step": 7931 + }, + { + "epoch": 0.8710740171315616, + "grad_norm": 2.485494375228882, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7223722338676453, + "num_tokens": 203823603.0, + "step": 7932 + }, + { + "epoch": 0.8711838348341753, + "grad_norm": 1.9871680736541748, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7099822163581848, + "num_tokens": 203856423.0, + "step": 7933 + }, + { + "epoch": 0.8712936525367889, + "grad_norm": 2.418355941772461, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7259665727615356, + "num_tokens": 203877109.0, + "step": 7934 + }, + { + "epoch": 0.8714034702394026, + "grad_norm": 2.1279752254486084, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7154814600944519, + "num_tokens": 203904232.0, + "step": 7935 + }, + { + "epoch": 0.8715132879420162, + "grad_norm": 2.0642154216766357, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.697575032711029, + "num_tokens": 203934343.0, + "step": 7936 + }, + { + "epoch": 0.87162310564463, + "grad_norm": 2.34679913520813, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7150081396102905, + "num_tokens": 203956463.0, + "step": 7937 + }, + { + "epoch": 0.8717329233472436, + "grad_norm": 2.6015982627868652, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7190755009651184, + "num_tokens": 203975596.0, + "step": 7938 + }, + { + "epoch": 0.8718427410498573, + "grad_norm": 2.0054492950439453, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7164915800094604, + "num_tokens": 204005299.0, + "step": 7939 + }, + { + "epoch": 0.8719525587524709, + "grad_norm": 2.1225433349609375, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7224936485290527, + "num_tokens": 204030678.0, + "step": 7940 + }, + { + "epoch": 0.8720623764550846, + "grad_norm": 2.1874680519104004, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6962281465530396, + "num_tokens": 204056619.0, + "step": 7941 + }, + { + "epoch": 0.8721721941576982, + "grad_norm": 2.325328826904297, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6957054138183594, + "num_tokens": 204081898.0, + "step": 7942 + }, + { + "epoch": 0.8722820118603118, + "grad_norm": 2.3164546489715576, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7103211283683777, + "num_tokens": 204102928.0, + "step": 7943 + }, + { + "epoch": 0.8723918295629255, + "grad_norm": 2.057812452316284, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7221599817276001, + "num_tokens": 204128187.0, + "step": 7944 + }, + { + "epoch": 0.8725016472655392, + "grad_norm": 2.0825390815734863, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6982651948928833, + "num_tokens": 204157037.0, + "step": 7945 + }, + { + "epoch": 0.8726114649681529, + "grad_norm": 2.028134822845459, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6984232068061829, + "num_tokens": 204186733.0, + "step": 7946 + }, + { + "epoch": 0.8727212826707665, + "grad_norm": 2.102905750274658, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6861993074417114, + "num_tokens": 204214878.0, + "step": 7947 + }, + { + "epoch": 0.8728311003733802, + "grad_norm": 2.3187127113342285, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.716465950012207, + "num_tokens": 204236887.0, + "step": 7948 + }, + { + "epoch": 0.8729409180759938, + "grad_norm": 2.1659293174743652, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7130810022354126, + "num_tokens": 204263244.0, + "step": 7949 + }, + { + "epoch": 0.8730507357786075, + "grad_norm": 2.137542963027954, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.699283242225647, + "num_tokens": 204292292.0, + "step": 7950 + }, + { + "epoch": 0.8731605534812211, + "grad_norm": 2.4078173637390137, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7141343355178833, + "num_tokens": 204313427.0, + "step": 7951 + }, + { + "epoch": 0.8732703711838349, + "grad_norm": 1.977373480796814, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.705119252204895, + "num_tokens": 204343002.0, + "step": 7952 + }, + { + "epoch": 0.8733801888864485, + "grad_norm": 1.9981049299240112, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.700726330280304, + "num_tokens": 204374221.0, + "step": 7953 + }, + { + "epoch": 0.8734900065890622, + "grad_norm": 2.276543140411377, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6913625001907349, + "num_tokens": 204398663.0, + "step": 7954 + }, + { + "epoch": 0.8735998242916758, + "grad_norm": 2.2111544609069824, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.701632022857666, + "num_tokens": 204423558.0, + "step": 7955 + }, + { + "epoch": 0.8737096419942895, + "grad_norm": 2.0365641117095947, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6849209666252136, + "num_tokens": 204454288.0, + "step": 7956 + }, + { + "epoch": 0.8738194596969031, + "grad_norm": 2.3600990772247314, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6925067901611328, + "num_tokens": 204478474.0, + "step": 7957 + }, + { + "epoch": 0.8739292773995168, + "grad_norm": 1.9925645589828491, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7143573760986328, + "num_tokens": 204506402.0, + "step": 7958 + }, + { + "epoch": 0.8740390951021305, + "grad_norm": 1.9170093536376953, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7092150449752808, + "num_tokens": 204537344.0, + "step": 7959 + }, + { + "epoch": 0.8741489128047442, + "grad_norm": 2.0961320400238037, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.6778432130813599, + "num_tokens": 204567212.0, + "step": 7960 + }, + { + "epoch": 0.8742587305073578, + "grad_norm": 2.3406858444213867, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7254420518875122, + "num_tokens": 204590553.0, + "step": 7961 + }, + { + "epoch": 0.8743685482099715, + "grad_norm": 2.0893685817718506, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7096766829490662, + "num_tokens": 204618678.0, + "step": 7962 + }, + { + "epoch": 0.8744783659125851, + "grad_norm": 2.038153648376465, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7032105922698975, + "num_tokens": 204645542.0, + "step": 7963 + }, + { + "epoch": 0.8745881836151987, + "grad_norm": 2.607797622680664, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7197821736335754, + "num_tokens": 204666328.0, + "step": 7964 + }, + { + "epoch": 0.8746980013178124, + "grad_norm": 2.1579415798187256, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7240360379219055, + "num_tokens": 204691166.0, + "step": 7965 + }, + { + "epoch": 0.8748078190204261, + "grad_norm": 2.085695743560791, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7123039960861206, + "num_tokens": 204718522.0, + "step": 7966 + }, + { + "epoch": 0.8749176367230398, + "grad_norm": 2.316011667251587, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.704498291015625, + "num_tokens": 204743747.0, + "step": 7967 + }, + { + "epoch": 0.8750274544256534, + "grad_norm": 1.958790898323059, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7123650312423706, + "num_tokens": 204776540.0, + "step": 7968 + }, + { + "epoch": 0.8751372721282671, + "grad_norm": 2.3708202838897705, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7185080051422119, + "num_tokens": 204797693.0, + "step": 7969 + }, + { + "epoch": 0.8752470898308807, + "grad_norm": 2.450608015060425, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7235015630722046, + "num_tokens": 204817972.0, + "step": 7970 + }, + { + "epoch": 0.8753569075334944, + "grad_norm": 2.1393487453460693, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7225243449211121, + "num_tokens": 204845676.0, + "step": 7971 + }, + { + "epoch": 0.875466725236108, + "grad_norm": 2.240267753601074, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6966577172279358, + "num_tokens": 204873174.0, + "step": 7972 + }, + { + "epoch": 0.8755765429387217, + "grad_norm": 1.9546838998794556, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7063461542129517, + "num_tokens": 204905775.0, + "step": 7973 + }, + { + "epoch": 0.8756863606413354, + "grad_norm": 2.156763792037964, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.704973578453064, + "num_tokens": 204933515.0, + "step": 7974 + }, + { + "epoch": 0.8757961783439491, + "grad_norm": 2.1931090354919434, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6901074051856995, + "num_tokens": 204958342.0, + "step": 7975 + }, + { + "epoch": 0.8759059960465627, + "grad_norm": 2.0644993782043457, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7216043472290039, + "num_tokens": 204986351.0, + "step": 7976 + }, + { + "epoch": 0.8760158137491764, + "grad_norm": 2.328794240951538, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7087374329566956, + "num_tokens": 205010469.0, + "step": 7977 + }, + { + "epoch": 0.87612563145179, + "grad_norm": 2.4061853885650635, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7188262939453125, + "num_tokens": 205030678.0, + "step": 7978 + }, + { + "epoch": 0.8762354491544037, + "grad_norm": 2.2864739894866943, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6920290589332581, + "num_tokens": 205054727.0, + "step": 7979 + }, + { + "epoch": 0.8763452668570173, + "grad_norm": 2.295962333679199, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7198690176010132, + "num_tokens": 205081462.0, + "step": 7980 + }, + { + "epoch": 0.8764550845596311, + "grad_norm": 2.0355417728424072, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6862404346466064, + "num_tokens": 205110422.0, + "step": 7981 + }, + { + "epoch": 0.8765649022622447, + "grad_norm": 2.1670336723327637, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7246019840240479, + "num_tokens": 205134839.0, + "step": 7982 + }, + { + "epoch": 0.8766747199648584, + "grad_norm": 2.1310439109802246, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7119022607803345, + "num_tokens": 205163815.0, + "step": 7983 + }, + { + "epoch": 0.876784537667472, + "grad_norm": 2.2933342456817627, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7005602121353149, + "num_tokens": 205186526.0, + "step": 7984 + }, + { + "epoch": 0.8768943553700856, + "grad_norm": 2.0277857780456543, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6942852735519409, + "num_tokens": 205217011.0, + "step": 7985 + }, + { + "epoch": 0.8770041730726993, + "grad_norm": 2.065673589706421, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7074812650680542, + "num_tokens": 205242712.0, + "step": 7986 + }, + { + "epoch": 0.8771139907753129, + "grad_norm": 2.3645644187927246, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6904959678649902, + "num_tokens": 205266742.0, + "step": 7987 + }, + { + "epoch": 0.8772238084779267, + "grad_norm": 1.993209958076477, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6939508318901062, + "num_tokens": 205296789.0, + "step": 7988 + }, + { + "epoch": 0.8773336261805403, + "grad_norm": 1.955939531326294, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6828238368034363, + "num_tokens": 205328200.0, + "step": 7989 + }, + { + "epoch": 0.877443443883154, + "grad_norm": 2.305792808532715, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7162967920303345, + "num_tokens": 205350229.0, + "step": 7990 + }, + { + "epoch": 0.8775532615857676, + "grad_norm": 2.2894060611724854, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6972047090530396, + "num_tokens": 205372919.0, + "step": 7991 + }, + { + "epoch": 0.8776630792883813, + "grad_norm": 2.2426748275756836, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7117948532104492, + "num_tokens": 205398317.0, + "step": 7992 + }, + { + "epoch": 0.8777728969909949, + "grad_norm": 2.074681282043457, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7339577674865723, + "num_tokens": 205423505.0, + "step": 7993 + }, + { + "epoch": 0.8778827146936086, + "grad_norm": 1.9593592882156372, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6790874600410461, + "num_tokens": 205458427.0, + "step": 7994 + }, + { + "epoch": 0.8779925323962223, + "grad_norm": 2.2475922107696533, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6958168745040894, + "num_tokens": 205483490.0, + "step": 7995 + }, + { + "epoch": 0.878102350098836, + "grad_norm": 1.9416652917861938, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.690609335899353, + "num_tokens": 205517309.0, + "step": 7996 + }, + { + "epoch": 0.8782121678014496, + "grad_norm": 2.2466821670532227, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7173725366592407, + "num_tokens": 205541510.0, + "step": 7997 + }, + { + "epoch": 0.8783219855040633, + "grad_norm": 2.132087230682373, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6962151527404785, + "num_tokens": 205566873.0, + "step": 7998 + }, + { + "epoch": 0.8784318032066769, + "grad_norm": 2.086101770401001, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6907796263694763, + "num_tokens": 205596176.0, + "step": 7999 + }, + { + "epoch": 0.8785416209092906, + "grad_norm": 2.4211862087249756, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.72280353307724, + "num_tokens": 205616276.0, + "step": 8000 + }, + { + "epoch": 0.8786514386119042, + "grad_norm": 2.002284049987793, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.6985895037651062, + "num_tokens": 205646053.0, + "step": 8001 + }, + { + "epoch": 0.8787612563145178, + "grad_norm": 2.2805776596069336, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7170892357826233, + "num_tokens": 205668900.0, + "step": 8002 + }, + { + "epoch": 0.8788710740171316, + "grad_norm": 2.5126497745513916, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7148103713989258, + "num_tokens": 205688823.0, + "step": 8003 + }, + { + "epoch": 0.8789808917197452, + "grad_norm": 2.070709228515625, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6845421195030212, + "num_tokens": 205718329.0, + "step": 8004 + }, + { + "epoch": 0.8790907094223589, + "grad_norm": 2.1450934410095215, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7209093570709229, + "num_tokens": 205743138.0, + "step": 8005 + }, + { + "epoch": 0.8792005271249725, + "grad_norm": 2.57696795463562, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7364688515663147, + "num_tokens": 205760950.0, + "step": 8006 + }, + { + "epoch": 0.8793103448275862, + "grad_norm": 2.268885374069214, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6967321634292603, + "num_tokens": 205784868.0, + "step": 8007 + }, + { + "epoch": 0.8794201625301998, + "grad_norm": 2.0610194206237793, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7193198204040527, + "num_tokens": 205812457.0, + "step": 8008 + }, + { + "epoch": 0.8795299802328135, + "grad_norm": 2.126518726348877, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7376750707626343, + "num_tokens": 205836787.0, + "step": 8009 + }, + { + "epoch": 0.8796397979354272, + "grad_norm": 2.0854079723358154, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7106828689575195, + "num_tokens": 205862627.0, + "step": 8010 + }, + { + "epoch": 0.8797496156380409, + "grad_norm": 2.2972240447998047, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7339909672737122, + "num_tokens": 205885586.0, + "step": 8011 + }, + { + "epoch": 0.8798594333406545, + "grad_norm": 2.1642417907714844, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7242705821990967, + "num_tokens": 205912578.0, + "step": 8012 + }, + { + "epoch": 0.8799692510432682, + "grad_norm": 2.225370407104492, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7095540761947632, + "num_tokens": 205936792.0, + "step": 8013 + }, + { + "epoch": 0.8800790687458818, + "grad_norm": 2.125430107116699, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7096002101898193, + "num_tokens": 205965039.0, + "step": 8014 + }, + { + "epoch": 0.8801888864484955, + "grad_norm": 2.2146894931793213, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7000570297241211, + "num_tokens": 205989704.0, + "step": 8015 + }, + { + "epoch": 0.8802987041511091, + "grad_norm": 2.0175700187683105, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7032808065414429, + "num_tokens": 206018363.0, + "step": 8016 + }, + { + "epoch": 0.8804085218537229, + "grad_norm": 2.535052537918091, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7163383364677429, + "num_tokens": 206039373.0, + "step": 8017 + }, + { + "epoch": 0.8805183395563365, + "grad_norm": 2.1784555912017822, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6926357746124268, + "num_tokens": 206064274.0, + "step": 8018 + }, + { + "epoch": 0.8806281572589502, + "grad_norm": 2.198551654815674, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7064437866210938, + "num_tokens": 206089656.0, + "step": 8019 + }, + { + "epoch": 0.8807379749615638, + "grad_norm": 2.025088310241699, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7127830982208252, + "num_tokens": 206118782.0, + "step": 8020 + }, + { + "epoch": 0.8808477926641775, + "grad_norm": 2.1743838787078857, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7097610235214233, + "num_tokens": 206142302.0, + "step": 8021 + }, + { + "epoch": 0.8809576103667911, + "grad_norm": 2.4388859272003174, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7220361232757568, + "num_tokens": 206163894.0, + "step": 8022 + }, + { + "epoch": 0.8810674280694047, + "grad_norm": 2.1468374729156494, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7071844339370728, + "num_tokens": 206191416.0, + "step": 8023 + }, + { + "epoch": 0.8811772457720185, + "grad_norm": 2.0138087272644043, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7011916637420654, + "num_tokens": 206221042.0, + "step": 8024 + }, + { + "epoch": 0.8812870634746321, + "grad_norm": 2.0352838039398193, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7171890139579773, + "num_tokens": 206250134.0, + "step": 8025 + }, + { + "epoch": 0.8813968811772458, + "grad_norm": 2.2556657791137695, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7139193415641785, + "num_tokens": 206274356.0, + "step": 8026 + }, + { + "epoch": 0.8815066988798594, + "grad_norm": 1.8815046548843384, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6980013847351074, + "num_tokens": 206304376.0, + "step": 8027 + }, + { + "epoch": 0.8816165165824731, + "grad_norm": 2.0676183700561523, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7195613980293274, + "num_tokens": 206330572.0, + "step": 8028 + }, + { + "epoch": 0.8817263342850867, + "grad_norm": 2.222008228302002, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7115601897239685, + "num_tokens": 206354910.0, + "step": 8029 + }, + { + "epoch": 0.8818361519877004, + "grad_norm": 1.9447903633117676, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.71540367603302, + "num_tokens": 206384451.0, + "step": 8030 + }, + { + "epoch": 0.881945969690314, + "grad_norm": 2.286146879196167, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7205878496170044, + "num_tokens": 206406411.0, + "step": 8031 + }, + { + "epoch": 0.8820557873929278, + "grad_norm": 2.6753785610198975, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7207283973693848, + "num_tokens": 206424906.0, + "step": 8032 + }, + { + "epoch": 0.8821656050955414, + "grad_norm": 2.077766180038452, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7236769199371338, + "num_tokens": 206452151.0, + "step": 8033 + }, + { + "epoch": 0.8822754227981551, + "grad_norm": 2.2882022857666016, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7213530540466309, + "num_tokens": 206475333.0, + "step": 8034 + }, + { + "epoch": 0.8823852405007687, + "grad_norm": 2.2146739959716797, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.719440221786499, + "num_tokens": 206498885.0, + "step": 8035 + }, + { + "epoch": 0.8824950582033824, + "grad_norm": 2.362001657485962, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7028281688690186, + "num_tokens": 206520305.0, + "step": 8036 + }, + { + "epoch": 0.882604875905996, + "grad_norm": 1.8765634298324585, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.706146240234375, + "num_tokens": 206551710.0, + "step": 8037 + }, + { + "epoch": 0.8827146936086097, + "grad_norm": 2.2517647743225098, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7058908343315125, + "num_tokens": 206574186.0, + "step": 8038 + }, + { + "epoch": 0.8828245113112234, + "grad_norm": 2.011747360229492, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7022481560707092, + "num_tokens": 206603874.0, + "step": 8039 + }, + { + "epoch": 0.8829343290138371, + "grad_norm": 2.0903027057647705, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7123780846595764, + "num_tokens": 206631300.0, + "step": 8040 + }, + { + "epoch": 0.8830441467164507, + "grad_norm": 2.0317258834838867, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7131401300430298, + "num_tokens": 206658764.0, + "step": 8041 + }, + { + "epoch": 0.8831539644190644, + "grad_norm": 2.1046955585479736, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.692290186882019, + "num_tokens": 206687397.0, + "step": 8042 + }, + { + "epoch": 0.883263782121678, + "grad_norm": 2.3475639820098877, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7212613821029663, + "num_tokens": 206709046.0, + "step": 8043 + }, + { + "epoch": 0.8833735998242916, + "grad_norm": 2.07841157913208, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7229712605476379, + "num_tokens": 206734535.0, + "step": 8044 + }, + { + "epoch": 0.8834834175269053, + "grad_norm": 2.2714202404022217, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6940404772758484, + "num_tokens": 206758514.0, + "step": 8045 + }, + { + "epoch": 0.883593235229519, + "grad_norm": 2.4237756729125977, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7092279195785522, + "num_tokens": 206782316.0, + "step": 8046 + }, + { + "epoch": 0.8837030529321327, + "grad_norm": 2.1066834926605225, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7033000588417053, + "num_tokens": 206806885.0, + "step": 8047 + }, + { + "epoch": 0.8838128706347463, + "grad_norm": 1.9321821928024292, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7258806824684143, + "num_tokens": 206837157.0, + "step": 8048 + }, + { + "epoch": 0.88392268833736, + "grad_norm": 2.3529703617095947, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7004731893539429, + "num_tokens": 206861340.0, + "step": 8049 + }, + { + "epoch": 0.8840325060399736, + "grad_norm": 2.080979585647583, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7093852758407593, + "num_tokens": 206887658.0, + "step": 8050 + }, + { + "epoch": 0.8841423237425873, + "grad_norm": 2.424243450164795, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7434751391410828, + "num_tokens": 206907078.0, + "step": 8051 + }, + { + "epoch": 0.8842521414452009, + "grad_norm": 2.0699710845947266, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7189438343048096, + "num_tokens": 206933479.0, + "step": 8052 + }, + { + "epoch": 0.8843619591478147, + "grad_norm": 2.183206081390381, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6997670531272888, + "num_tokens": 206959341.0, + "step": 8053 + }, + { + "epoch": 0.8844717768504283, + "grad_norm": 2.282026529312134, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.710368275642395, + "num_tokens": 206984295.0, + "step": 8054 + }, + { + "epoch": 0.884581594553042, + "grad_norm": 2.3045687675476074, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7052592039108276, + "num_tokens": 207007419.0, + "step": 8055 + }, + { + "epoch": 0.8846914122556556, + "grad_norm": 2.2075560092926025, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7169294357299805, + "num_tokens": 207031959.0, + "step": 8056 + }, + { + "epoch": 0.8848012299582693, + "grad_norm": 2.165673017501831, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7055872082710266, + "num_tokens": 207057847.0, + "step": 8057 + }, + { + "epoch": 0.8849110476608829, + "grad_norm": 1.7926636934280396, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6991829872131348, + "num_tokens": 207094034.0, + "step": 8058 + }, + { + "epoch": 0.8850208653634966, + "grad_norm": 2.2248518466949463, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7094345092773438, + "num_tokens": 207118946.0, + "step": 8059 + }, + { + "epoch": 0.8851306830661102, + "grad_norm": 1.967787742614746, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6899011731147766, + "num_tokens": 207149655.0, + "step": 8060 + }, + { + "epoch": 0.885240500768724, + "grad_norm": 2.3164143562316895, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6935381889343262, + "num_tokens": 207174561.0, + "step": 8061 + }, + { + "epoch": 0.8853503184713376, + "grad_norm": 2.075077533721924, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7058130502700806, + "num_tokens": 207202337.0, + "step": 8062 + }, + { + "epoch": 0.8854601361739513, + "grad_norm": 1.9324971437454224, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7145657539367676, + "num_tokens": 207231209.0, + "step": 8063 + }, + { + "epoch": 0.8855699538765649, + "grad_norm": 2.1837732791900635, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7123152017593384, + "num_tokens": 207257362.0, + "step": 8064 + }, + { + "epoch": 0.8856797715791785, + "grad_norm": 2.162051200866699, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6859387159347534, + "num_tokens": 207284413.0, + "step": 8065 + }, + { + "epoch": 0.8857895892817922, + "grad_norm": 2.0996029376983643, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7120097279548645, + "num_tokens": 207310958.0, + "step": 8066 + }, + { + "epoch": 0.8858994069844058, + "grad_norm": 2.0403308868408203, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.702420711517334, + "num_tokens": 207339536.0, + "step": 8067 + }, + { + "epoch": 0.8860092246870196, + "grad_norm": 2.263075351715088, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.736227810382843, + "num_tokens": 207363264.0, + "step": 8068 + }, + { + "epoch": 0.8861190423896332, + "grad_norm": 2.5754382610321045, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7027457356452942, + "num_tokens": 207384180.0, + "step": 8069 + }, + { + "epoch": 0.8862288600922469, + "grad_norm": 2.1760008335113525, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7216640710830688, + "num_tokens": 207408963.0, + "step": 8070 + }, + { + "epoch": 0.8863386777948605, + "grad_norm": 2.1753408908843994, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7369840145111084, + "num_tokens": 207435523.0, + "step": 8071 + }, + { + "epoch": 0.8864484954974742, + "grad_norm": 2.510270118713379, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.711870014667511, + "num_tokens": 207456133.0, + "step": 8072 + }, + { + "epoch": 0.8865583132000878, + "grad_norm": 2.060995578765869, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7212066650390625, + "num_tokens": 207483607.0, + "step": 8073 + }, + { + "epoch": 0.8866681309027015, + "grad_norm": 2.147545576095581, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7320047616958618, + "num_tokens": 207510460.0, + "step": 8074 + }, + { + "epoch": 0.8867779486053152, + "grad_norm": 2.1310458183288574, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7029738426208496, + "num_tokens": 207537466.0, + "step": 8075 + }, + { + "epoch": 0.8868877663079289, + "grad_norm": 2.089550733566284, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7280598282814026, + "num_tokens": 207562637.0, + "step": 8076 + }, + { + "epoch": 0.8869975840105425, + "grad_norm": 2.313345193862915, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7035870552062988, + "num_tokens": 207587024.0, + "step": 8077 + }, + { + "epoch": 0.8871074017131562, + "grad_norm": 1.9828319549560547, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6949714422225952, + "num_tokens": 207617799.0, + "step": 8078 + }, + { + "epoch": 0.8872172194157698, + "grad_norm": 2.439100742340088, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7071989178657532, + "num_tokens": 207639064.0, + "step": 8079 + }, + { + "epoch": 0.8873270371183835, + "grad_norm": 2.379195213317871, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.700599193572998, + "num_tokens": 207659513.0, + "step": 8080 + }, + { + "epoch": 0.8874368548209971, + "grad_norm": 2.2386322021484375, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6823241710662842, + "num_tokens": 207683577.0, + "step": 8081 + }, + { + "epoch": 0.8875466725236109, + "grad_norm": 2.4771323204040527, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7117694616317749, + "num_tokens": 207703852.0, + "step": 8082 + }, + { + "epoch": 0.8876564902262245, + "grad_norm": 2.292353868484497, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7144010066986084, + "num_tokens": 207727211.0, + "step": 8083 + }, + { + "epoch": 0.8877663079288381, + "grad_norm": 2.1237058639526367, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7062137126922607, + "num_tokens": 207753970.0, + "step": 8084 + }, + { + "epoch": 0.8878761256314518, + "grad_norm": 2.27095103263855, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7072769403457642, + "num_tokens": 207777373.0, + "step": 8085 + }, + { + "epoch": 0.8879859433340654, + "grad_norm": 2.266599655151367, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7111438512802124, + "num_tokens": 207803071.0, + "step": 8086 + }, + { + "epoch": 0.8880957610366791, + "grad_norm": 2.107880115509033, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6916226148605347, + "num_tokens": 207831531.0, + "step": 8087 + }, + { + "epoch": 0.8882055787392927, + "grad_norm": 2.2829933166503906, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6997261047363281, + "num_tokens": 207858265.0, + "step": 8088 + }, + { + "epoch": 0.8883153964419065, + "grad_norm": 2.347975015640259, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7138868570327759, + "num_tokens": 207880776.0, + "step": 8089 + }, + { + "epoch": 0.8884252141445201, + "grad_norm": 1.9803540706634521, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6987553834915161, + "num_tokens": 207912333.0, + "step": 8090 + }, + { + "epoch": 0.8885350318471338, + "grad_norm": 1.7948194742202759, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6845422983169556, + "num_tokens": 207947806.0, + "step": 8091 + }, + { + "epoch": 0.8886448495497474, + "grad_norm": 2.1291158199310303, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7038246989250183, + "num_tokens": 207974205.0, + "step": 8092 + }, + { + "epoch": 0.8887546672523611, + "grad_norm": 2.1564061641693115, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7024452090263367, + "num_tokens": 208002131.0, + "step": 8093 + }, + { + "epoch": 0.8888644849549747, + "grad_norm": 1.958864450454712, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7137960195541382, + "num_tokens": 208033054.0, + "step": 8094 + }, + { + "epoch": 0.8889743026575884, + "grad_norm": 2.049668073654175, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7042266130447388, + "num_tokens": 208062067.0, + "step": 8095 + }, + { + "epoch": 0.889084120360202, + "grad_norm": 2.2523884773254395, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.714425802230835, + "num_tokens": 208086551.0, + "step": 8096 + }, + { + "epoch": 0.8891939380628158, + "grad_norm": 2.121457099914551, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7182759046554565, + "num_tokens": 208112781.0, + "step": 8097 + }, + { + "epoch": 0.8893037557654294, + "grad_norm": 2.195239543914795, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6918649077415466, + "num_tokens": 208139190.0, + "step": 8098 + }, + { + "epoch": 0.8894135734680431, + "grad_norm": 2.1440117359161377, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7465496063232422, + "num_tokens": 208163188.0, + "step": 8099 + }, + { + "epoch": 0.8895233911706567, + "grad_norm": 2.0091121196746826, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6828673481941223, + "num_tokens": 208197539.0, + "step": 8100 + }, + { + "epoch": 0.8896332088732704, + "grad_norm": 2.0364012718200684, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7010375261306763, + "num_tokens": 208225417.0, + "step": 8101 + }, + { + "epoch": 0.889743026575884, + "grad_norm": 2.3443362712860107, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6939363479614258, + "num_tokens": 208249217.0, + "step": 8102 + }, + { + "epoch": 0.8898528442784976, + "grad_norm": 1.9033668041229248, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6958293318748474, + "num_tokens": 208283120.0, + "step": 8103 + }, + { + "epoch": 0.8899626619811114, + "grad_norm": 2.2323243618011475, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.703222930431366, + "num_tokens": 208307596.0, + "step": 8104 + }, + { + "epoch": 0.890072479683725, + "grad_norm": 2.06668758392334, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6765546202659607, + "num_tokens": 208338336.0, + "step": 8105 + }, + { + "epoch": 0.8901822973863387, + "grad_norm": 1.9485960006713867, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7020986080169678, + "num_tokens": 208367414.0, + "step": 8106 + }, + { + "epoch": 0.8902921150889523, + "grad_norm": 2.196427822113037, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7130967974662781, + "num_tokens": 208392726.0, + "step": 8107 + }, + { + "epoch": 0.890401932791566, + "grad_norm": 2.5678961277008057, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7239643335342407, + "num_tokens": 208411073.0, + "step": 8108 + }, + { + "epoch": 0.8905117504941796, + "grad_norm": 1.9103957414627075, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6902674436569214, + "num_tokens": 208441630.0, + "step": 8109 + }, + { + "epoch": 0.8906215681967933, + "grad_norm": 2.1249983310699463, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7078372240066528, + "num_tokens": 208467989.0, + "step": 8110 + }, + { + "epoch": 0.890731385899407, + "grad_norm": 2.331674337387085, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7031621336936951, + "num_tokens": 208493170.0, + "step": 8111 + }, + { + "epoch": 0.8908412036020207, + "grad_norm": 2.0671985149383545, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7262448072433472, + "num_tokens": 208519561.0, + "step": 8112 + }, + { + "epoch": 0.8909510213046343, + "grad_norm": 2.5291390419006348, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.71830153465271, + "num_tokens": 208539244.0, + "step": 8113 + }, + { + "epoch": 0.891060839007248, + "grad_norm": 2.5609171390533447, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7174160480499268, + "num_tokens": 208558709.0, + "step": 8114 + }, + { + "epoch": 0.8911706567098616, + "grad_norm": 1.8537086248397827, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6965499520301819, + "num_tokens": 208590827.0, + "step": 8115 + }, + { + "epoch": 0.8912804744124753, + "grad_norm": 1.9727001190185547, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7335708141326904, + "num_tokens": 208618944.0, + "step": 8116 + }, + { + "epoch": 0.8913902921150889, + "grad_norm": 2.3176093101501465, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7297757863998413, + "num_tokens": 208640731.0, + "step": 8117 + }, + { + "epoch": 0.8915001098177027, + "grad_norm": 2.2419564723968506, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6966805458068848, + "num_tokens": 208664540.0, + "step": 8118 + }, + { + "epoch": 0.8916099275203163, + "grad_norm": 2.195655345916748, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7162883281707764, + "num_tokens": 208690855.0, + "step": 8119 + }, + { + "epoch": 0.89171974522293, + "grad_norm": 2.1691784858703613, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7169619202613831, + "num_tokens": 208718323.0, + "step": 8120 + }, + { + "epoch": 0.8918295629255436, + "grad_norm": 2.0028538703918457, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7013849020004272, + "num_tokens": 208748459.0, + "step": 8121 + }, + { + "epoch": 0.8919393806281573, + "grad_norm": 2.0895581245422363, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.7048629522323608, + "num_tokens": 208775936.0, + "step": 8122 + }, + { + "epoch": 0.8920491983307709, + "grad_norm": 2.2116429805755615, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.700657069683075, + "num_tokens": 208799955.0, + "step": 8123 + }, + { + "epoch": 0.8921590160333845, + "grad_norm": 4.2835259437561035, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7143614888191223, + "num_tokens": 208818361.0, + "step": 8124 + }, + { + "epoch": 0.8922688337359982, + "grad_norm": 2.142622470855713, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7284837961196899, + "num_tokens": 208844358.0, + "step": 8125 + }, + { + "epoch": 0.892378651438612, + "grad_norm": 2.311728000640869, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7120671272277832, + "num_tokens": 208866400.0, + "step": 8126 + }, + { + "epoch": 0.8924884691412256, + "grad_norm": 2.1723480224609375, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7203701138496399, + "num_tokens": 208889717.0, + "step": 8127 + }, + { + "epoch": 0.8925982868438392, + "grad_norm": 2.4441890716552734, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7168710231781006, + "num_tokens": 208911215.0, + "step": 8128 + }, + { + "epoch": 0.8927081045464529, + "grad_norm": 2.483142614364624, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7394996881484985, + "num_tokens": 208930729.0, + "step": 8129 + }, + { + "epoch": 0.8928179222490665, + "grad_norm": 2.03944730758667, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7133661508560181, + "num_tokens": 208957799.0, + "step": 8130 + }, + { + "epoch": 0.8929277399516802, + "grad_norm": 2.5385262966156006, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7007492780685425, + "num_tokens": 208978514.0, + "step": 8131 + }, + { + "epoch": 0.8930375576542938, + "grad_norm": 1.7407822608947754, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7034242749214172, + "num_tokens": 209017461.0, + "step": 8132 + }, + { + "epoch": 0.8931473753569076, + "grad_norm": 2.4059219360351562, + "learning_rate": 1e-06, + "loss": 1.0953, + "mean_token_accuracy": 0.6757004857063293, + "num_tokens": 209043945.0, + "step": 8133 + }, + { + "epoch": 0.8932571930595212, + "grad_norm": 2.0555129051208496, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6830628514289856, + "num_tokens": 209073175.0, + "step": 8134 + }, + { + "epoch": 0.8933670107621349, + "grad_norm": 2.089076519012451, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6840308904647827, + "num_tokens": 209101052.0, + "step": 8135 + }, + { + "epoch": 0.8934768284647485, + "grad_norm": 2.261723518371582, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7174816131591797, + "num_tokens": 209126358.0, + "step": 8136 + }, + { + "epoch": 0.8935866461673622, + "grad_norm": 2.0510950088500977, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7139917016029358, + "num_tokens": 209154995.0, + "step": 8137 + }, + { + "epoch": 0.8936964638699758, + "grad_norm": 1.9757080078125, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.719335675239563, + "num_tokens": 209184665.0, + "step": 8138 + }, + { + "epoch": 0.8938062815725895, + "grad_norm": 2.067079782485962, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7035700678825378, + "num_tokens": 209212634.0, + "step": 8139 + }, + { + "epoch": 0.8939160992752032, + "grad_norm": 2.352074384689331, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7116284370422363, + "num_tokens": 209235687.0, + "step": 8140 + }, + { + "epoch": 0.8940259169778169, + "grad_norm": 2.268709897994995, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6989097595214844, + "num_tokens": 209259205.0, + "step": 8141 + }, + { + "epoch": 0.8941357346804305, + "grad_norm": 2.373962163925171, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.707615852355957, + "num_tokens": 209280850.0, + "step": 8142 + }, + { + "epoch": 0.8942455523830442, + "grad_norm": 2.4014856815338135, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7066248655319214, + "num_tokens": 209303405.0, + "step": 8143 + }, + { + "epoch": 0.8943553700856578, + "grad_norm": 1.8314024209976196, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6914000511169434, + "num_tokens": 209339682.0, + "step": 8144 + }, + { + "epoch": 0.8944651877882714, + "grad_norm": 1.8896647691726685, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6961504817008972, + "num_tokens": 209372114.0, + "step": 8145 + }, + { + "epoch": 0.8945750054908851, + "grad_norm": 1.9458528757095337, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7170504331588745, + "num_tokens": 209400850.0, + "step": 8146 + }, + { + "epoch": 0.8946848231934988, + "grad_norm": 2.137194871902466, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6811909675598145, + "num_tokens": 209431267.0, + "step": 8147 + }, + { + "epoch": 0.8947946408961125, + "grad_norm": 2.4937524795532227, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7153317332267761, + "num_tokens": 209452412.0, + "step": 8148 + }, + { + "epoch": 0.8949044585987261, + "grad_norm": 2.664116621017456, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7012912034988403, + "num_tokens": 209470628.0, + "step": 8149 + }, + { + "epoch": 0.8950142763013398, + "grad_norm": 2.265643835067749, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7137008309364319, + "num_tokens": 209495156.0, + "step": 8150 + }, + { + "epoch": 0.8951240940039534, + "grad_norm": 2.0669007301330566, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.705964207649231, + "num_tokens": 209522732.0, + "step": 8151 + }, + { + "epoch": 0.8952339117065671, + "grad_norm": 2.154616355895996, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7090542316436768, + "num_tokens": 209549972.0, + "step": 8152 + }, + { + "epoch": 0.8953437294091807, + "grad_norm": 2.0695338249206543, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7110322713851929, + "num_tokens": 209578671.0, + "step": 8153 + }, + { + "epoch": 0.8954535471117944, + "grad_norm": 2.0686309337615967, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7122929096221924, + "num_tokens": 209605334.0, + "step": 8154 + }, + { + "epoch": 0.8955633648144081, + "grad_norm": 2.0686988830566406, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7107740044593811, + "num_tokens": 209630750.0, + "step": 8155 + }, + { + "epoch": 0.8956731825170218, + "grad_norm": 2.1049323081970215, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.724588930606842, + "num_tokens": 209655058.0, + "step": 8156 + }, + { + "epoch": 0.8957830002196354, + "grad_norm": 2.078789234161377, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7156720161437988, + "num_tokens": 209681518.0, + "step": 8157 + }, + { + "epoch": 0.8958928179222491, + "grad_norm": 2.0664119720458984, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7145994901657104, + "num_tokens": 209706752.0, + "step": 8158 + }, + { + "epoch": 0.8960026356248627, + "grad_norm": 2.344114303588867, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7162866592407227, + "num_tokens": 209730582.0, + "step": 8159 + }, + { + "epoch": 0.8961124533274764, + "grad_norm": 2.11734676361084, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6998264193534851, + "num_tokens": 209757608.0, + "step": 8160 + }, + { + "epoch": 0.89622227103009, + "grad_norm": 2.3391218185424805, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7063815593719482, + "num_tokens": 209778921.0, + "step": 8161 + }, + { + "epoch": 0.8963320887327038, + "grad_norm": 2.027151584625244, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7085602283477783, + "num_tokens": 209807418.0, + "step": 8162 + }, + { + "epoch": 0.8964419064353174, + "grad_norm": 2.5841751098632812, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7228217720985413, + "num_tokens": 209826323.0, + "step": 8163 + }, + { + "epoch": 0.896551724137931, + "grad_norm": 2.2241852283477783, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7078462839126587, + "num_tokens": 209850191.0, + "step": 8164 + }, + { + "epoch": 0.8966615418405447, + "grad_norm": 2.0050556659698486, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6983513832092285, + "num_tokens": 209878946.0, + "step": 8165 + }, + { + "epoch": 0.8967713595431583, + "grad_norm": 2.2171502113342285, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.725189745426178, + "num_tokens": 209904569.0, + "step": 8166 + }, + { + "epoch": 0.896881177245772, + "grad_norm": 2.0094363689422607, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7114839553833008, + "num_tokens": 209932826.0, + "step": 8167 + }, + { + "epoch": 0.8969909949483856, + "grad_norm": 1.9382661581039429, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7023953795433044, + "num_tokens": 209964716.0, + "step": 8168 + }, + { + "epoch": 0.8971008126509994, + "grad_norm": 2.239088296890259, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7121370434761047, + "num_tokens": 209988145.0, + "step": 8169 + }, + { + "epoch": 0.897210630353613, + "grad_norm": 2.084444761276245, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7136020064353943, + "num_tokens": 210012996.0, + "step": 8170 + }, + { + "epoch": 0.8973204480562267, + "grad_norm": 2.111088991165161, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6883996725082397, + "num_tokens": 210042948.0, + "step": 8171 + }, + { + "epoch": 0.8974302657588403, + "grad_norm": 2.2724082469940186, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7309544086456299, + "num_tokens": 210065967.0, + "step": 8172 + }, + { + "epoch": 0.897540083461454, + "grad_norm": 2.436936616897583, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7000364065170288, + "num_tokens": 210086813.0, + "step": 8173 + }, + { + "epoch": 0.8976499011640676, + "grad_norm": 2.231419086456299, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7164145112037659, + "num_tokens": 210111575.0, + "step": 8174 + }, + { + "epoch": 0.8977597188666813, + "grad_norm": 2.166684865951538, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6909201145172119, + "num_tokens": 210138427.0, + "step": 8175 + }, + { + "epoch": 0.897869536569295, + "grad_norm": 2.29826021194458, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.699498176574707, + "num_tokens": 210161505.0, + "step": 8176 + }, + { + "epoch": 0.8979793542719087, + "grad_norm": 2.087254524230957, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7073040008544922, + "num_tokens": 210187899.0, + "step": 8177 + }, + { + "epoch": 0.8980891719745223, + "grad_norm": 2.199582815170288, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6890950798988342, + "num_tokens": 210214869.0, + "step": 8178 + }, + { + "epoch": 0.898198989677136, + "grad_norm": 2.3812174797058105, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7136929035186768, + "num_tokens": 210236265.0, + "step": 8179 + }, + { + "epoch": 0.8983088073797496, + "grad_norm": 2.0826878547668457, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7043184638023376, + "num_tokens": 210263519.0, + "step": 8180 + }, + { + "epoch": 0.8984186250823633, + "grad_norm": 2.322709560394287, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7277333736419678, + "num_tokens": 210287123.0, + "step": 8181 + }, + { + "epoch": 0.8985284427849769, + "grad_norm": 2.2224526405334473, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6980067491531372, + "num_tokens": 210312517.0, + "step": 8182 + }, + { + "epoch": 0.8986382604875905, + "grad_norm": 2.0404155254364014, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6965149641036987, + "num_tokens": 210342229.0, + "step": 8183 + }, + { + "epoch": 0.8987480781902043, + "grad_norm": 1.9940141439437866, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6914774775505066, + "num_tokens": 210372948.0, + "step": 8184 + }, + { + "epoch": 0.898857895892818, + "grad_norm": 2.387503147125244, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7053260803222656, + "num_tokens": 210395987.0, + "step": 8185 + }, + { + "epoch": 0.8989677135954316, + "grad_norm": 1.9311147928237915, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6880415678024292, + "num_tokens": 210430203.0, + "step": 8186 + }, + { + "epoch": 0.8990775312980452, + "grad_norm": 2.437657117843628, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7088793516159058, + "num_tokens": 210452284.0, + "step": 8187 + }, + { + "epoch": 0.8991873490006589, + "grad_norm": 2.0201680660247803, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.7008280754089355, + "num_tokens": 210482400.0, + "step": 8188 + }, + { + "epoch": 0.8992971667032725, + "grad_norm": 2.1258695125579834, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7007334232330322, + "num_tokens": 210510728.0, + "step": 8189 + }, + { + "epoch": 0.8994069844058862, + "grad_norm": 2.1542906761169434, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.698004424571991, + "num_tokens": 210538537.0, + "step": 8190 + }, + { + "epoch": 0.8995168021084999, + "grad_norm": 2.2566964626312256, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7299044728279114, + "num_tokens": 210561886.0, + "step": 8191 + }, + { + "epoch": 0.8996266198111136, + "grad_norm": 2.0856752395629883, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6932038068771362, + "num_tokens": 210589770.0, + "step": 8192 + }, + { + "epoch": 0.8997364375137272, + "grad_norm": 2.106583595275879, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7391077280044556, + "num_tokens": 210614788.0, + "step": 8193 + }, + { + "epoch": 0.8998462552163409, + "grad_norm": 2.0882537364959717, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6980555057525635, + "num_tokens": 210643135.0, + "step": 8194 + }, + { + "epoch": 0.8999560729189545, + "grad_norm": 2.0990495681762695, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7095792293548584, + "num_tokens": 210670007.0, + "step": 8195 + }, + { + "epoch": 0.9000658906215682, + "grad_norm": 2.2942328453063965, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6961520314216614, + "num_tokens": 210697020.0, + "step": 8196 + }, + { + "epoch": 0.9001757083241818, + "grad_norm": 2.0820038318634033, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7152837514877319, + "num_tokens": 210723945.0, + "step": 8197 + }, + { + "epoch": 0.9002855260267956, + "grad_norm": 2.1692466735839844, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7013692259788513, + "num_tokens": 210750081.0, + "step": 8198 + }, + { + "epoch": 0.9003953437294092, + "grad_norm": 1.9274394512176514, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7077473998069763, + "num_tokens": 210779869.0, + "step": 8199 + }, + { + "epoch": 0.9005051614320229, + "grad_norm": 2.136946678161621, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6963755488395691, + "num_tokens": 210807320.0, + "step": 8200 + }, + { + "epoch": 0.9006149791346365, + "grad_norm": 2.274449586868286, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.700379490852356, + "num_tokens": 210831571.0, + "step": 8201 + }, + { + "epoch": 0.9007247968372502, + "grad_norm": 2.2986907958984375, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7093581557273865, + "num_tokens": 210855714.0, + "step": 8202 + }, + { + "epoch": 0.9008346145398638, + "grad_norm": 2.09562087059021, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.694503664970398, + "num_tokens": 210883827.0, + "step": 8203 + }, + { + "epoch": 0.9009444322424774, + "grad_norm": 2.0803005695343018, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7141900658607483, + "num_tokens": 210911838.0, + "step": 8204 + }, + { + "epoch": 0.9010542499450912, + "grad_norm": 2.444737434387207, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.732658863067627, + "num_tokens": 210931297.0, + "step": 8205 + }, + { + "epoch": 0.9011640676477048, + "grad_norm": 2.1492421627044678, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7048871517181396, + "num_tokens": 210956069.0, + "step": 8206 + }, + { + "epoch": 0.9012738853503185, + "grad_norm": 2.135080099105835, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7262762784957886, + "num_tokens": 210983565.0, + "step": 8207 + }, + { + "epoch": 0.9013837030529321, + "grad_norm": 2.3249566555023193, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7201055288314819, + "num_tokens": 211006014.0, + "step": 8208 + }, + { + "epoch": 0.9014935207555458, + "grad_norm": 1.9498947858810425, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6855666041374207, + "num_tokens": 211038509.0, + "step": 8209 + }, + { + "epoch": 0.9016033384581594, + "grad_norm": 1.9643619060516357, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.696815013885498, + "num_tokens": 211069437.0, + "step": 8210 + }, + { + "epoch": 0.9017131561607731, + "grad_norm": 1.963452696800232, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6944172382354736, + "num_tokens": 211099140.0, + "step": 8211 + }, + { + "epoch": 0.9018229738633867, + "grad_norm": 2.4085004329681396, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7214834690093994, + "num_tokens": 211119427.0, + "step": 8212 + }, + { + "epoch": 0.9019327915660005, + "grad_norm": 3.049720287322998, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7436779737472534, + "num_tokens": 211133405.0, + "step": 8213 + }, + { + "epoch": 0.9020426092686141, + "grad_norm": 2.4960482120513916, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6879539489746094, + "num_tokens": 211155900.0, + "step": 8214 + }, + { + "epoch": 0.9021524269712278, + "grad_norm": 2.4539377689361572, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7234711647033691, + "num_tokens": 211175515.0, + "step": 8215 + }, + { + "epoch": 0.9022622446738414, + "grad_norm": 2.1696765422821045, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7126792669296265, + "num_tokens": 211200921.0, + "step": 8216 + }, + { + "epoch": 0.9023720623764551, + "grad_norm": 2.1595563888549805, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6909862756729126, + "num_tokens": 211227508.0, + "step": 8217 + }, + { + "epoch": 0.9024818800790687, + "grad_norm": 2.0905396938323975, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.691827654838562, + "num_tokens": 211255082.0, + "step": 8218 + }, + { + "epoch": 0.9025916977816824, + "grad_norm": 2.403794050216675, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7256710529327393, + "num_tokens": 211275437.0, + "step": 8219 + }, + { + "epoch": 0.9027015154842961, + "grad_norm": 2.117981195449829, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.685949444770813, + "num_tokens": 211302156.0, + "step": 8220 + }, + { + "epoch": 0.9028113331869098, + "grad_norm": 2.1956472396850586, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.731903076171875, + "num_tokens": 211327130.0, + "step": 8221 + }, + { + "epoch": 0.9029211508895234, + "grad_norm": 2.397810220718384, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7244334816932678, + "num_tokens": 211347776.0, + "step": 8222 + }, + { + "epoch": 0.903030968592137, + "grad_norm": 2.006816864013672, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7002105712890625, + "num_tokens": 211376436.0, + "step": 8223 + }, + { + "epoch": 0.9031407862947507, + "grad_norm": 2.1920325756073, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7266726493835449, + "num_tokens": 211400339.0, + "step": 8224 + }, + { + "epoch": 0.9032506039973643, + "grad_norm": 2.5695226192474365, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7172291278839111, + "num_tokens": 211421691.0, + "step": 8225 + }, + { + "epoch": 0.903360421699978, + "grad_norm": 2.3051135540008545, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7051122188568115, + "num_tokens": 211445082.0, + "step": 8226 + }, + { + "epoch": 0.9034702394025917, + "grad_norm": 2.0484578609466553, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6937394142150879, + "num_tokens": 211474700.0, + "step": 8227 + }, + { + "epoch": 0.9035800571052054, + "grad_norm": 2.1938040256500244, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7048765420913696, + "num_tokens": 211501111.0, + "step": 8228 + }, + { + "epoch": 0.903689874807819, + "grad_norm": 2.261899948120117, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7116094827651978, + "num_tokens": 211525093.0, + "step": 8229 + }, + { + "epoch": 0.9037996925104327, + "grad_norm": 2.111711025238037, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7054046392440796, + "num_tokens": 211553385.0, + "step": 8230 + }, + { + "epoch": 0.9039095102130463, + "grad_norm": 2.3645377159118652, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7033118605613708, + "num_tokens": 211575737.0, + "step": 8231 + }, + { + "epoch": 0.90401932791566, + "grad_norm": 1.9718505144119263, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6820225119590759, + "num_tokens": 211606552.0, + "step": 8232 + }, + { + "epoch": 0.9041291456182736, + "grad_norm": 2.0891542434692383, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7089473009109497, + "num_tokens": 211632720.0, + "step": 8233 + }, + { + "epoch": 0.9042389633208874, + "grad_norm": 2.305668592453003, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7333923578262329, + "num_tokens": 211654267.0, + "step": 8234 + }, + { + "epoch": 0.904348781023501, + "grad_norm": 2.118967056274414, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6967982649803162, + "num_tokens": 211679056.0, + "step": 8235 + }, + { + "epoch": 0.9044585987261147, + "grad_norm": 2.253521203994751, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7174508571624756, + "num_tokens": 211702485.0, + "step": 8236 + }, + { + "epoch": 0.9045684164287283, + "grad_norm": 2.377251625061035, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7096524834632874, + "num_tokens": 211726132.0, + "step": 8237 + }, + { + "epoch": 0.904678234131342, + "grad_norm": 2.0896809101104736, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6974401473999023, + "num_tokens": 211753994.0, + "step": 8238 + }, + { + "epoch": 0.9047880518339556, + "grad_norm": 2.254915475845337, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7113233208656311, + "num_tokens": 211778265.0, + "step": 8239 + }, + { + "epoch": 0.9048978695365693, + "grad_norm": 1.9305927753448486, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7021974325180054, + "num_tokens": 211811904.0, + "step": 8240 + }, + { + "epoch": 0.9050076872391829, + "grad_norm": 2.1637964248657227, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7037091255187988, + "num_tokens": 211840323.0, + "step": 8241 + }, + { + "epoch": 0.9051175049417967, + "grad_norm": 2.271913528442383, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.692044734954834, + "num_tokens": 211867443.0, + "step": 8242 + }, + { + "epoch": 0.9052273226444103, + "grad_norm": 2.281294345855713, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7183661460876465, + "num_tokens": 211891955.0, + "step": 8243 + }, + { + "epoch": 0.905337140347024, + "grad_norm": 2.541278839111328, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.718873918056488, + "num_tokens": 211911668.0, + "step": 8244 + }, + { + "epoch": 0.9054469580496376, + "grad_norm": 2.5464107990264893, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6983505487442017, + "num_tokens": 211932750.0, + "step": 8245 + }, + { + "epoch": 0.9055567757522512, + "grad_norm": 2.0965723991394043, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6949227452278137, + "num_tokens": 211959709.0, + "step": 8246 + }, + { + "epoch": 0.9056665934548649, + "grad_norm": 2.1142187118530273, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7421149611473083, + "num_tokens": 211985791.0, + "step": 8247 + }, + { + "epoch": 0.9057764111574785, + "grad_norm": 2.173346996307373, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7162259817123413, + "num_tokens": 212011147.0, + "step": 8248 + }, + { + "epoch": 0.9058862288600923, + "grad_norm": 2.1698174476623535, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7212380766868591, + "num_tokens": 212034859.0, + "step": 8249 + }, + { + "epoch": 0.9059960465627059, + "grad_norm": 2.329406976699829, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7028002738952637, + "num_tokens": 212057076.0, + "step": 8250 + }, + { + "epoch": 0.9061058642653196, + "grad_norm": 1.9961051940917969, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6985375881195068, + "num_tokens": 212087740.0, + "step": 8251 + }, + { + "epoch": 0.9062156819679332, + "grad_norm": 2.2301738262176514, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7111760973930359, + "num_tokens": 212111277.0, + "step": 8252 + }, + { + "epoch": 0.9063254996705469, + "grad_norm": 2.066232919692993, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7005960941314697, + "num_tokens": 212140905.0, + "step": 8253 + }, + { + "epoch": 0.9064353173731605, + "grad_norm": 2.203542709350586, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6931720972061157, + "num_tokens": 212167048.0, + "step": 8254 + }, + { + "epoch": 0.9065451350757742, + "grad_norm": 1.9774675369262695, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6857796311378479, + "num_tokens": 212196685.0, + "step": 8255 + }, + { + "epoch": 0.9066549527783879, + "grad_norm": 2.0751004219055176, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7105268836021423, + "num_tokens": 212221610.0, + "step": 8256 + }, + { + "epoch": 0.9067647704810016, + "grad_norm": 2.4029123783111572, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7207713723182678, + "num_tokens": 212243189.0, + "step": 8257 + }, + { + "epoch": 0.9068745881836152, + "grad_norm": 1.9525854587554932, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7070456743240356, + "num_tokens": 212272687.0, + "step": 8258 + }, + { + "epoch": 0.9069844058862289, + "grad_norm": 2.2802093029022217, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7006658315658569, + "num_tokens": 212297670.0, + "step": 8259 + }, + { + "epoch": 0.9070942235888425, + "grad_norm": 2.5471014976501465, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7301175594329834, + "num_tokens": 212317361.0, + "step": 8260 + }, + { + "epoch": 0.9072040412914562, + "grad_norm": 2.1565427780151367, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7211380004882812, + "num_tokens": 212340267.0, + "step": 8261 + }, + { + "epoch": 0.9073138589940698, + "grad_norm": 2.343792676925659, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7352595329284668, + "num_tokens": 212362310.0, + "step": 8262 + }, + { + "epoch": 0.9074236766966836, + "grad_norm": 2.095698833465576, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7097231149673462, + "num_tokens": 212388235.0, + "step": 8263 + }, + { + "epoch": 0.9075334943992972, + "grad_norm": 2.2577927112579346, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7483574748039246, + "num_tokens": 212410756.0, + "step": 8264 + }, + { + "epoch": 0.9076433121019108, + "grad_norm": 1.9794665575027466, + "learning_rate": 1e-06, + "loss": 1.1355, + "mean_token_accuracy": 0.6714187860488892, + "num_tokens": 212441367.0, + "step": 8265 + }, + { + "epoch": 0.9077531298045245, + "grad_norm": 2.2885069847106934, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7011643648147583, + "num_tokens": 212465042.0, + "step": 8266 + }, + { + "epoch": 0.9078629475071381, + "grad_norm": 1.846644639968872, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6928919553756714, + "num_tokens": 212499227.0, + "step": 8267 + }, + { + "epoch": 0.9079727652097518, + "grad_norm": 2.331904649734497, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7046176791191101, + "num_tokens": 212521396.0, + "step": 8268 + }, + { + "epoch": 0.9080825829123654, + "grad_norm": 2.2644271850585938, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7180883884429932, + "num_tokens": 212545498.0, + "step": 8269 + }, + { + "epoch": 0.9081924006149792, + "grad_norm": 2.316256284713745, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7120830416679382, + "num_tokens": 212568087.0, + "step": 8270 + }, + { + "epoch": 0.9083022183175928, + "grad_norm": 2.074183940887451, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.7013686895370483, + "num_tokens": 212596327.0, + "step": 8271 + }, + { + "epoch": 0.9084120360202065, + "grad_norm": 2.1292343139648438, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6957077980041504, + "num_tokens": 212623136.0, + "step": 8272 + }, + { + "epoch": 0.9085218537228201, + "grad_norm": 2.0185415744781494, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.710676908493042, + "num_tokens": 212651332.0, + "step": 8273 + }, + { + "epoch": 0.9086316714254338, + "grad_norm": 2.192608594894409, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7245516777038574, + "num_tokens": 212675474.0, + "step": 8274 + }, + { + "epoch": 0.9087414891280474, + "grad_norm": 2.04242205619812, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7117664813995361, + "num_tokens": 212704425.0, + "step": 8275 + }, + { + "epoch": 0.9088513068306611, + "grad_norm": 2.0244994163513184, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6968567371368408, + "num_tokens": 212735895.0, + "step": 8276 + }, + { + "epoch": 0.9089611245332747, + "grad_norm": 2.340322256088257, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7210736274719238, + "num_tokens": 212758905.0, + "step": 8277 + }, + { + "epoch": 0.9090709422358885, + "grad_norm": 2.0158371925354004, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.705311119556427, + "num_tokens": 212787111.0, + "step": 8278 + }, + { + "epoch": 0.9091807599385021, + "grad_norm": 2.531761646270752, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7120300531387329, + "num_tokens": 212808798.0, + "step": 8279 + }, + { + "epoch": 0.9092905776411158, + "grad_norm": 2.2992146015167236, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7065499424934387, + "num_tokens": 212832277.0, + "step": 8280 + }, + { + "epoch": 0.9094003953437294, + "grad_norm": 2.12963604927063, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.704451858997345, + "num_tokens": 212860004.0, + "step": 8281 + }, + { + "epoch": 0.909510213046343, + "grad_norm": 1.949717402458191, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7042927742004395, + "num_tokens": 212888705.0, + "step": 8282 + }, + { + "epoch": 0.9096200307489567, + "grad_norm": 2.4185068607330322, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7112904787063599, + "num_tokens": 212910753.0, + "step": 8283 + }, + { + "epoch": 0.9097298484515703, + "grad_norm": 2.1602652072906494, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6947358846664429, + "num_tokens": 212936952.0, + "step": 8284 + }, + { + "epoch": 0.9098396661541841, + "grad_norm": 2.4964046478271484, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7249820232391357, + "num_tokens": 212957491.0, + "step": 8285 + }, + { + "epoch": 0.9099494838567977, + "grad_norm": 2.509824275970459, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7524269223213196, + "num_tokens": 212975884.0, + "step": 8286 + }, + { + "epoch": 0.9100593015594114, + "grad_norm": 1.8205747604370117, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6868029832839966, + "num_tokens": 213012395.0, + "step": 8287 + }, + { + "epoch": 0.910169119262025, + "grad_norm": 2.0655558109283447, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6944735050201416, + "num_tokens": 213039517.0, + "step": 8288 + }, + { + "epoch": 0.9102789369646387, + "grad_norm": 2.261610269546509, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7225848436355591, + "num_tokens": 213062545.0, + "step": 8289 + }, + { + "epoch": 0.9103887546672523, + "grad_norm": 2.0841214656829834, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.719883143901825, + "num_tokens": 213089738.0, + "step": 8290 + }, + { + "epoch": 0.910498572369866, + "grad_norm": 2.3765335083007812, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7151247262954712, + "num_tokens": 213111831.0, + "step": 8291 + }, + { + "epoch": 0.9106083900724797, + "grad_norm": 2.056978464126587, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7156081795692444, + "num_tokens": 213138605.0, + "step": 8292 + }, + { + "epoch": 0.9107182077750934, + "grad_norm": 2.236175537109375, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6745674014091492, + "num_tokens": 213165812.0, + "step": 8293 + }, + { + "epoch": 0.910828025477707, + "grad_norm": 2.3125483989715576, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7124105095863342, + "num_tokens": 213190246.0, + "step": 8294 + }, + { + "epoch": 0.9109378431803207, + "grad_norm": 1.9880974292755127, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6948229074478149, + "num_tokens": 213218865.0, + "step": 8295 + }, + { + "epoch": 0.9110476608829343, + "grad_norm": 2.1090471744537354, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.704757571220398, + "num_tokens": 213244488.0, + "step": 8296 + }, + { + "epoch": 0.911157478585548, + "grad_norm": 2.43499493598938, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.720367431640625, + "num_tokens": 213264608.0, + "step": 8297 + }, + { + "epoch": 0.9112672962881616, + "grad_norm": 3.0417473316192627, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7139562368392944, + "num_tokens": 213279781.0, + "step": 8298 + }, + { + "epoch": 0.9113771139907754, + "grad_norm": 2.159545660018921, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7209940552711487, + "num_tokens": 213303191.0, + "step": 8299 + }, + { + "epoch": 0.911486931693389, + "grad_norm": 2.69319224357605, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7118210196495056, + "num_tokens": 213323209.0, + "step": 8300 + }, + { + "epoch": 0.9115967493960027, + "grad_norm": 2.143672227859497, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7096110582351685, + "num_tokens": 213351187.0, + "step": 8301 + }, + { + "epoch": 0.9117065670986163, + "grad_norm": 2.13132381439209, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6985393166542053, + "num_tokens": 213377655.0, + "step": 8302 + }, + { + "epoch": 0.91181638480123, + "grad_norm": 2.304875373840332, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7153414487838745, + "num_tokens": 213399446.0, + "step": 8303 + }, + { + "epoch": 0.9119262025038436, + "grad_norm": 2.011479616165161, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6823179125785828, + "num_tokens": 213430371.0, + "step": 8304 + }, + { + "epoch": 0.9120360202064572, + "grad_norm": 2.132113456726074, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7040050029754639, + "num_tokens": 213458535.0, + "step": 8305 + }, + { + "epoch": 0.9121458379090709, + "grad_norm": 2.1300222873687744, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7198495864868164, + "num_tokens": 213484323.0, + "step": 8306 + }, + { + "epoch": 0.9122556556116846, + "grad_norm": 2.1678481101989746, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7244921326637268, + "num_tokens": 213507951.0, + "step": 8307 + }, + { + "epoch": 0.9123654733142983, + "grad_norm": 2.0772693157196045, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.694351315498352, + "num_tokens": 213536419.0, + "step": 8308 + }, + { + "epoch": 0.9124752910169119, + "grad_norm": 1.9264755249023438, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7157555818557739, + "num_tokens": 213567372.0, + "step": 8309 + }, + { + "epoch": 0.9125851087195256, + "grad_norm": 2.531851053237915, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7100301384925842, + "num_tokens": 213588238.0, + "step": 8310 + }, + { + "epoch": 0.9126949264221392, + "grad_norm": 2.061821222305298, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6967490911483765, + "num_tokens": 213617307.0, + "step": 8311 + }, + { + "epoch": 0.9128047441247529, + "grad_norm": 2.0147175788879395, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6885299682617188, + "num_tokens": 213648911.0, + "step": 8312 + }, + { + "epoch": 0.9129145618273665, + "grad_norm": 2.308413505554199, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7264279723167419, + "num_tokens": 213671003.0, + "step": 8313 + }, + { + "epoch": 0.9130243795299803, + "grad_norm": 2.096628427505493, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7067844271659851, + "num_tokens": 213697383.0, + "step": 8314 + }, + { + "epoch": 0.9131341972325939, + "grad_norm": 2.150287389755249, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7085852026939392, + "num_tokens": 213724128.0, + "step": 8315 + }, + { + "epoch": 0.9132440149352076, + "grad_norm": 2.242511749267578, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6865026950836182, + "num_tokens": 213751680.0, + "step": 8316 + }, + { + "epoch": 0.9133538326378212, + "grad_norm": 2.12017560005188, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.68948894739151, + "num_tokens": 213779629.0, + "step": 8317 + }, + { + "epoch": 0.9134636503404349, + "grad_norm": 2.007803201675415, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7118238806724548, + "num_tokens": 213807330.0, + "step": 8318 + }, + { + "epoch": 0.9135734680430485, + "grad_norm": 2.0524630546569824, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6809054017066956, + "num_tokens": 213835790.0, + "step": 8319 + }, + { + "epoch": 0.9136832857456622, + "grad_norm": 2.3220603466033936, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7102670669555664, + "num_tokens": 213859040.0, + "step": 8320 + }, + { + "epoch": 0.9137931034482759, + "grad_norm": 2.1982967853546143, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7112900018692017, + "num_tokens": 213883517.0, + "step": 8321 + }, + { + "epoch": 0.9139029211508896, + "grad_norm": 1.979711651802063, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.71952223777771, + "num_tokens": 213912139.0, + "step": 8322 + }, + { + "epoch": 0.9140127388535032, + "grad_norm": 2.245798110961914, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7314935326576233, + "num_tokens": 213936239.0, + "step": 8323 + }, + { + "epoch": 0.9141225565561168, + "grad_norm": 2.273329019546509, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7360839247703552, + "num_tokens": 213958706.0, + "step": 8324 + }, + { + "epoch": 0.9142323742587305, + "grad_norm": 2.085770606994629, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.719907283782959, + "num_tokens": 213986877.0, + "step": 8325 + }, + { + "epoch": 0.9143421919613441, + "grad_norm": 2.119297981262207, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6904423832893372, + "num_tokens": 214015092.0, + "step": 8326 + }, + { + "epoch": 0.9144520096639578, + "grad_norm": 2.0932726860046387, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6825634837150574, + "num_tokens": 214042589.0, + "step": 8327 + }, + { + "epoch": 0.9145618273665715, + "grad_norm": 2.497117280960083, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7164259552955627, + "num_tokens": 214063195.0, + "step": 8328 + }, + { + "epoch": 0.9146716450691852, + "grad_norm": 2.0149691104888916, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6902143955230713, + "num_tokens": 214092643.0, + "step": 8329 + }, + { + "epoch": 0.9147814627717988, + "grad_norm": 2.0971932411193848, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7052672505378723, + "num_tokens": 214119740.0, + "step": 8330 + }, + { + "epoch": 0.9148912804744125, + "grad_norm": 2.2355754375457764, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7027550339698792, + "num_tokens": 214145826.0, + "step": 8331 + }, + { + "epoch": 0.9150010981770261, + "grad_norm": 1.9203122854232788, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7020269632339478, + "num_tokens": 214176923.0, + "step": 8332 + }, + { + "epoch": 0.9151109158796398, + "grad_norm": 1.913191318511963, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.712607741355896, + "num_tokens": 214206421.0, + "step": 8333 + }, + { + "epoch": 0.9152207335822534, + "grad_norm": 2.1481804847717285, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7192834615707397, + "num_tokens": 214230717.0, + "step": 8334 + }, + { + "epoch": 0.9153305512848671, + "grad_norm": 1.9198089838027954, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6860466599464417, + "num_tokens": 214263151.0, + "step": 8335 + }, + { + "epoch": 0.9154403689874808, + "grad_norm": 2.2496695518493652, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6972880363464355, + "num_tokens": 214287234.0, + "step": 8336 + }, + { + "epoch": 0.9155501866900945, + "grad_norm": 2.476386785507202, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7025083303451538, + "num_tokens": 214308849.0, + "step": 8337 + }, + { + "epoch": 0.9156600043927081, + "grad_norm": 1.9197874069213867, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7040122747421265, + "num_tokens": 214340928.0, + "step": 8338 + }, + { + "epoch": 0.9157698220953218, + "grad_norm": 2.3861846923828125, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7021591663360596, + "num_tokens": 214363598.0, + "step": 8339 + }, + { + "epoch": 0.9158796397979354, + "grad_norm": 1.7963389158248901, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.699411153793335, + "num_tokens": 214400771.0, + "step": 8340 + }, + { + "epoch": 0.915989457500549, + "grad_norm": 2.0205533504486084, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7048776149749756, + "num_tokens": 214430588.0, + "step": 8341 + }, + { + "epoch": 0.9160992752031627, + "grad_norm": 1.992499828338623, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.693311870098114, + "num_tokens": 214459892.0, + "step": 8342 + }, + { + "epoch": 0.9162090929057765, + "grad_norm": 2.1171391010284424, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6866146326065063, + "num_tokens": 214488026.0, + "step": 8343 + }, + { + "epoch": 0.9163189106083901, + "grad_norm": 2.0513532161712646, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7055523991584778, + "num_tokens": 214517212.0, + "step": 8344 + }, + { + "epoch": 0.9164287283110037, + "grad_norm": 2.1674327850341797, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7018769979476929, + "num_tokens": 214541900.0, + "step": 8345 + }, + { + "epoch": 0.9165385460136174, + "grad_norm": 2.3293843269348145, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.721925675868988, + "num_tokens": 214563878.0, + "step": 8346 + }, + { + "epoch": 0.916648363716231, + "grad_norm": 2.292476177215576, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.690438985824585, + "num_tokens": 214589308.0, + "step": 8347 + }, + { + "epoch": 0.9167581814188447, + "grad_norm": 2.12431263923645, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7124401330947876, + "num_tokens": 214614772.0, + "step": 8348 + }, + { + "epoch": 0.9168679991214583, + "grad_norm": 2.137587547302246, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7139742374420166, + "num_tokens": 214642140.0, + "step": 8349 + }, + { + "epoch": 0.9169778168240721, + "grad_norm": 2.3643906116485596, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7059347629547119, + "num_tokens": 214664121.0, + "step": 8350 + }, + { + "epoch": 0.9170876345266857, + "grad_norm": 2.022242784500122, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7131986618041992, + "num_tokens": 214695499.0, + "step": 8351 + }, + { + "epoch": 0.9171974522292994, + "grad_norm": 2.4082674980163574, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6958017945289612, + "num_tokens": 214717077.0, + "step": 8352 + }, + { + "epoch": 0.917307269931913, + "grad_norm": 2.533097267150879, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7199835181236267, + "num_tokens": 214737534.0, + "step": 8353 + }, + { + "epoch": 0.9174170876345267, + "grad_norm": 2.2167859077453613, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6954537630081177, + "num_tokens": 214763923.0, + "step": 8354 + }, + { + "epoch": 0.9175269053371403, + "grad_norm": 2.428630828857422, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7379547357559204, + "num_tokens": 214783468.0, + "step": 8355 + }, + { + "epoch": 0.917636723039754, + "grad_norm": 2.182166814804077, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7154189944267273, + "num_tokens": 214809971.0, + "step": 8356 + }, + { + "epoch": 0.9177465407423677, + "grad_norm": 1.9006949663162231, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7077759504318237, + "num_tokens": 214841658.0, + "step": 8357 + }, + { + "epoch": 0.9178563584449814, + "grad_norm": 2.1214513778686523, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7116459012031555, + "num_tokens": 214868361.0, + "step": 8358 + }, + { + "epoch": 0.917966176147595, + "grad_norm": 1.9413985013961792, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6990935802459717, + "num_tokens": 214897324.0, + "step": 8359 + }, + { + "epoch": 0.9180759938502087, + "grad_norm": 2.440335750579834, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7213596105575562, + "num_tokens": 214919287.0, + "step": 8360 + }, + { + "epoch": 0.9181858115528223, + "grad_norm": 2.1961541175842285, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7075858116149902, + "num_tokens": 214945011.0, + "step": 8361 + }, + { + "epoch": 0.918295629255436, + "grad_norm": 2.2214901447296143, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7033571004867554, + "num_tokens": 214968704.0, + "step": 8362 + }, + { + "epoch": 0.9184054469580496, + "grad_norm": 2.368527889251709, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7326206564903259, + "num_tokens": 214989136.0, + "step": 8363 + }, + { + "epoch": 0.9185152646606632, + "grad_norm": 2.460714817047119, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.714106559753418, + "num_tokens": 215009285.0, + "step": 8364 + }, + { + "epoch": 0.918625082363277, + "grad_norm": 2.274820327758789, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7098778486251831, + "num_tokens": 215031506.0, + "step": 8365 + }, + { + "epoch": 0.9187349000658906, + "grad_norm": 2.1820712089538574, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7111893892288208, + "num_tokens": 215056062.0, + "step": 8366 + }, + { + "epoch": 0.9188447177685043, + "grad_norm": 2.414086103439331, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7172268033027649, + "num_tokens": 215077361.0, + "step": 8367 + }, + { + "epoch": 0.9189545354711179, + "grad_norm": 1.9775961637496948, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7008147239685059, + "num_tokens": 215106294.0, + "step": 8368 + }, + { + "epoch": 0.9190643531737316, + "grad_norm": 2.1777262687683105, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7342054843902588, + "num_tokens": 215130349.0, + "step": 8369 + }, + { + "epoch": 0.9191741708763452, + "grad_norm": 1.8728902339935303, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6783667206764221, + "num_tokens": 215162201.0, + "step": 8370 + }, + { + "epoch": 0.9192839885789589, + "grad_norm": 2.3024299144744873, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7472981214523315, + "num_tokens": 215182628.0, + "step": 8371 + }, + { + "epoch": 0.9193938062815726, + "grad_norm": 2.1504056453704834, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7071633338928223, + "num_tokens": 215207595.0, + "step": 8372 + }, + { + "epoch": 0.9195036239841863, + "grad_norm": 2.0842392444610596, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.734795868396759, + "num_tokens": 215233840.0, + "step": 8373 + }, + { + "epoch": 0.9196134416867999, + "grad_norm": 2.4088213443756104, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7138421535491943, + "num_tokens": 215254369.0, + "step": 8374 + }, + { + "epoch": 0.9197232593894136, + "grad_norm": 2.2183234691619873, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7256885766983032, + "num_tokens": 215278629.0, + "step": 8375 + }, + { + "epoch": 0.9198330770920272, + "grad_norm": 1.9229681491851807, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6930050253868103, + "num_tokens": 215311228.0, + "step": 8376 + }, + { + "epoch": 0.9199428947946409, + "grad_norm": 2.308717966079712, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6927573680877686, + "num_tokens": 215334977.0, + "step": 8377 + }, + { + "epoch": 0.9200527124972545, + "grad_norm": 1.9606491327285767, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6891201734542847, + "num_tokens": 215367835.0, + "step": 8378 + }, + { + "epoch": 0.9201625301998683, + "grad_norm": 2.524156332015991, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7276442050933838, + "num_tokens": 215387315.0, + "step": 8379 + }, + { + "epoch": 0.9202723479024819, + "grad_norm": 2.08541202545166, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6952181458473206, + "num_tokens": 215413208.0, + "step": 8380 + }, + { + "epoch": 0.9203821656050956, + "grad_norm": 2.0712835788726807, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.722267210483551, + "num_tokens": 215438624.0, + "step": 8381 + }, + { + "epoch": 0.9204919833077092, + "grad_norm": 2.370256185531616, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7227755188941956, + "num_tokens": 215461933.0, + "step": 8382 + }, + { + "epoch": 0.9206018010103229, + "grad_norm": 2.367371082305908, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7036762237548828, + "num_tokens": 215484935.0, + "step": 8383 + }, + { + "epoch": 0.9207116187129365, + "grad_norm": 2.168139696121216, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7224792242050171, + "num_tokens": 215509451.0, + "step": 8384 + }, + { + "epoch": 0.9208214364155501, + "grad_norm": 2.031982183456421, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7298744320869446, + "num_tokens": 215537696.0, + "step": 8385 + }, + { + "epoch": 0.9209312541181639, + "grad_norm": 2.2325899600982666, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7074462175369263, + "num_tokens": 215562210.0, + "step": 8386 + }, + { + "epoch": 0.9210410718207775, + "grad_norm": 2.099308729171753, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7168657779693604, + "num_tokens": 215589905.0, + "step": 8387 + }, + { + "epoch": 0.9211508895233912, + "grad_norm": 2.216193437576294, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6897382736206055, + "num_tokens": 215614973.0, + "step": 8388 + }, + { + "epoch": 0.9212607072260048, + "grad_norm": 1.9610099792480469, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7365683317184448, + "num_tokens": 215643136.0, + "step": 8389 + }, + { + "epoch": 0.9213705249286185, + "grad_norm": 2.227630138397217, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7265738248825073, + "num_tokens": 215667251.0, + "step": 8390 + }, + { + "epoch": 0.9214803426312321, + "grad_norm": 2.0816597938537598, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7107174396514893, + "num_tokens": 215694652.0, + "step": 8391 + }, + { + "epoch": 0.9215901603338458, + "grad_norm": 2.184782028198242, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7334347367286682, + "num_tokens": 215718732.0, + "step": 8392 + }, + { + "epoch": 0.9216999780364594, + "grad_norm": 2.165806770324707, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7286921739578247, + "num_tokens": 215742981.0, + "step": 8393 + }, + { + "epoch": 0.9218097957390732, + "grad_norm": 2.143170118331909, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7094752192497253, + "num_tokens": 215771232.0, + "step": 8394 + }, + { + "epoch": 0.9219196134416868, + "grad_norm": 2.2578372955322266, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7200725078582764, + "num_tokens": 215794486.0, + "step": 8395 + }, + { + "epoch": 0.9220294311443005, + "grad_norm": 2.156761884689331, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7106283903121948, + "num_tokens": 215822137.0, + "step": 8396 + }, + { + "epoch": 0.9221392488469141, + "grad_norm": 2.0650758743286133, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7077020406723022, + "num_tokens": 215850365.0, + "step": 8397 + }, + { + "epoch": 0.9222490665495278, + "grad_norm": 2.294349193572998, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7260148525238037, + "num_tokens": 215871936.0, + "step": 8398 + }, + { + "epoch": 0.9223588842521414, + "grad_norm": 2.2329261302948, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6956008672714233, + "num_tokens": 215897615.0, + "step": 8399 + }, + { + "epoch": 0.922468701954755, + "grad_norm": 1.9050557613372803, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6933389902114868, + "num_tokens": 215930149.0, + "step": 8400 + }, + { + "epoch": 0.9225785196573688, + "grad_norm": 2.0086166858673096, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6899458765983582, + "num_tokens": 215959234.0, + "step": 8401 + }, + { + "epoch": 0.9226883373599825, + "grad_norm": 2.0988619327545166, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6996840238571167, + "num_tokens": 215989413.0, + "step": 8402 + }, + { + "epoch": 0.9227981550625961, + "grad_norm": 2.108347177505493, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7174710631370544, + "num_tokens": 216013642.0, + "step": 8403 + }, + { + "epoch": 0.9229079727652097, + "grad_norm": 2.0919389724731445, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7036552429199219, + "num_tokens": 216041439.0, + "step": 8404 + }, + { + "epoch": 0.9230177904678234, + "grad_norm": 2.283341884613037, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7045585513114929, + "num_tokens": 216066464.0, + "step": 8405 + }, + { + "epoch": 0.923127608170437, + "grad_norm": 2.131852149963379, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.719780683517456, + "num_tokens": 216092530.0, + "step": 8406 + }, + { + "epoch": 0.9232374258730507, + "grad_norm": 2.0649797916412354, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7035713195800781, + "num_tokens": 216118167.0, + "step": 8407 + }, + { + "epoch": 0.9233472435756644, + "grad_norm": 2.22316837310791, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7152845859527588, + "num_tokens": 216142963.0, + "step": 8408 + }, + { + "epoch": 0.9234570612782781, + "grad_norm": 2.391052007675171, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7298691272735596, + "num_tokens": 216164794.0, + "step": 8409 + }, + { + "epoch": 0.9235668789808917, + "grad_norm": 1.931858777999878, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6986850500106812, + "num_tokens": 216196632.0, + "step": 8410 + }, + { + "epoch": 0.9236766966835054, + "grad_norm": 2.10286021232605, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6928687691688538, + "num_tokens": 216224808.0, + "step": 8411 + }, + { + "epoch": 0.923786514386119, + "grad_norm": 2.0614099502563477, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6850491762161255, + "num_tokens": 216257299.0, + "step": 8412 + }, + { + "epoch": 0.9238963320887327, + "grad_norm": 2.2486181259155273, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7259840965270996, + "num_tokens": 216280563.0, + "step": 8413 + }, + { + "epoch": 0.9240061497913463, + "grad_norm": 1.9687323570251465, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7146512269973755, + "num_tokens": 216309571.0, + "step": 8414 + }, + { + "epoch": 0.9241159674939601, + "grad_norm": 1.9056683778762817, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6809463500976562, + "num_tokens": 216342750.0, + "step": 8415 + }, + { + "epoch": 0.9242257851965737, + "grad_norm": 1.9253957271575928, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.707148015499115, + "num_tokens": 216374422.0, + "step": 8416 + }, + { + "epoch": 0.9243356028991874, + "grad_norm": 2.3346681594848633, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7216496467590332, + "num_tokens": 216396917.0, + "step": 8417 + }, + { + "epoch": 0.924445420601801, + "grad_norm": 2.220475912094116, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6864224076271057, + "num_tokens": 216422986.0, + "step": 8418 + }, + { + "epoch": 0.9245552383044147, + "grad_norm": 2.3955795764923096, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7139639258384705, + "num_tokens": 216444947.0, + "step": 8419 + }, + { + "epoch": 0.9246650560070283, + "grad_norm": 2.3269758224487305, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7111852765083313, + "num_tokens": 216468322.0, + "step": 8420 + }, + { + "epoch": 0.924774873709642, + "grad_norm": 2.7153313159942627, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7190584540367126, + "num_tokens": 216486183.0, + "step": 8421 + }, + { + "epoch": 0.9248846914122557, + "grad_norm": 2.2201178073883057, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.705244243144989, + "num_tokens": 216511320.0, + "step": 8422 + }, + { + "epoch": 0.9249945091148694, + "grad_norm": 2.0914626121520996, + "learning_rate": 1e-06, + "loss": 1.0973, + "mean_token_accuracy": 0.6823536157608032, + "num_tokens": 216542574.0, + "step": 8423 + }, + { + "epoch": 0.925104326817483, + "grad_norm": 2.082169771194458, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6917101144790649, + "num_tokens": 216573888.0, + "step": 8424 + }, + { + "epoch": 0.9252141445200966, + "grad_norm": 2.397380828857422, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7066606283187866, + "num_tokens": 216596668.0, + "step": 8425 + }, + { + "epoch": 0.9253239622227103, + "grad_norm": 2.2419562339782715, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7058588266372681, + "num_tokens": 216620318.0, + "step": 8426 + }, + { + "epoch": 0.9254337799253239, + "grad_norm": 2.087570905685425, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7050769925117493, + "num_tokens": 216645955.0, + "step": 8427 + }, + { + "epoch": 0.9255435976279376, + "grad_norm": 1.9634592533111572, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7128459215164185, + "num_tokens": 216675784.0, + "step": 8428 + }, + { + "epoch": 0.9256534153305512, + "grad_norm": 2.24484920501709, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6992394924163818, + "num_tokens": 216700214.0, + "step": 8429 + }, + { + "epoch": 0.925763233033165, + "grad_norm": 2.0159895420074463, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7068806886672974, + "num_tokens": 216730570.0, + "step": 8430 + }, + { + "epoch": 0.9258730507357786, + "grad_norm": 2.143767833709717, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6996203064918518, + "num_tokens": 216758409.0, + "step": 8431 + }, + { + "epoch": 0.9259828684383923, + "grad_norm": 2.020873785018921, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6946964263916016, + "num_tokens": 216791393.0, + "step": 8432 + }, + { + "epoch": 0.9260926861410059, + "grad_norm": 2.2181386947631836, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.726443886756897, + "num_tokens": 216814980.0, + "step": 8433 + }, + { + "epoch": 0.9262025038436196, + "grad_norm": 2.051140069961548, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7236367464065552, + "num_tokens": 216841896.0, + "step": 8434 + }, + { + "epoch": 0.9263123215462332, + "grad_norm": 2.0198354721069336, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7047815918922424, + "num_tokens": 216871181.0, + "step": 8435 + }, + { + "epoch": 0.9264221392488469, + "grad_norm": 2.143747568130493, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6913154125213623, + "num_tokens": 216896466.0, + "step": 8436 + }, + { + "epoch": 0.9265319569514606, + "grad_norm": 2.215872287750244, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7097603678703308, + "num_tokens": 216922552.0, + "step": 8437 + }, + { + "epoch": 0.9266417746540743, + "grad_norm": 1.8390765190124512, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6930171847343445, + "num_tokens": 216958607.0, + "step": 8438 + }, + { + "epoch": 0.9267515923566879, + "grad_norm": 2.209120988845825, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6956880688667297, + "num_tokens": 216982885.0, + "step": 8439 + }, + { + "epoch": 0.9268614100593016, + "grad_norm": 2.322981357574463, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7410702705383301, + "num_tokens": 217005332.0, + "step": 8440 + }, + { + "epoch": 0.9269712277619152, + "grad_norm": 2.0270884037017822, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6908552646636963, + "num_tokens": 217034353.0, + "step": 8441 + }, + { + "epoch": 0.9270810454645289, + "grad_norm": 2.208939790725708, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7051712870597839, + "num_tokens": 217058129.0, + "step": 8442 + }, + { + "epoch": 0.9271908631671425, + "grad_norm": 2.0784614086151123, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7287776470184326, + "num_tokens": 217081324.0, + "step": 8443 + }, + { + "epoch": 0.9273006808697563, + "grad_norm": 2.170100450515747, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6900680065155029, + "num_tokens": 217110077.0, + "step": 8444 + }, + { + "epoch": 0.9274104985723699, + "grad_norm": 1.873729944229126, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6928731203079224, + "num_tokens": 217142617.0, + "step": 8445 + }, + { + "epoch": 0.9275203162749835, + "grad_norm": 2.5368165969848633, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7142189145088196, + "num_tokens": 217161774.0, + "step": 8446 + }, + { + "epoch": 0.9276301339775972, + "grad_norm": 2.188755750656128, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7021427750587463, + "num_tokens": 217189554.0, + "step": 8447 + }, + { + "epoch": 0.9277399516802108, + "grad_norm": 2.4263408184051514, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7270144820213318, + "num_tokens": 217209287.0, + "step": 8448 + }, + { + "epoch": 0.9278497693828245, + "grad_norm": 2.014883279800415, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6891096234321594, + "num_tokens": 217238339.0, + "step": 8449 + }, + { + "epoch": 0.9279595870854381, + "grad_norm": 2.199869155883789, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7287666201591492, + "num_tokens": 217262696.0, + "step": 8450 + }, + { + "epoch": 0.9280694047880519, + "grad_norm": 2.2481141090393066, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6867728233337402, + "num_tokens": 217287247.0, + "step": 8451 + }, + { + "epoch": 0.9281792224906655, + "grad_norm": 2.0610787868499756, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7037181854248047, + "num_tokens": 217315537.0, + "step": 8452 + }, + { + "epoch": 0.9282890401932792, + "grad_norm": 2.1184163093566895, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.707654595375061, + "num_tokens": 217340524.0, + "step": 8453 + }, + { + "epoch": 0.9283988578958928, + "grad_norm": 2.255336046218872, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7198237180709839, + "num_tokens": 217364531.0, + "step": 8454 + }, + { + "epoch": 0.9285086755985065, + "grad_norm": 2.064638376235962, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7007024884223938, + "num_tokens": 217391235.0, + "step": 8455 + }, + { + "epoch": 0.9286184933011201, + "grad_norm": 1.7539420127868652, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.691572904586792, + "num_tokens": 217429170.0, + "step": 8456 + }, + { + "epoch": 0.9287283110037338, + "grad_norm": 2.103703498840332, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7098860740661621, + "num_tokens": 217456775.0, + "step": 8457 + }, + { + "epoch": 0.9288381287063474, + "grad_norm": 2.450819969177246, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.717544436454773, + "num_tokens": 217478454.0, + "step": 8458 + }, + { + "epoch": 0.9289479464089612, + "grad_norm": 1.944103479385376, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.677147626876831, + "num_tokens": 217509125.0, + "step": 8459 + }, + { + "epoch": 0.9290577641115748, + "grad_norm": 2.255527973175049, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7205413579940796, + "num_tokens": 217533307.0, + "step": 8460 + }, + { + "epoch": 0.9291675818141885, + "grad_norm": 1.9675688743591309, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7214137315750122, + "num_tokens": 217561669.0, + "step": 8461 + }, + { + "epoch": 0.9292773995168021, + "grad_norm": 2.0646255016326904, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7214980125427246, + "num_tokens": 217590542.0, + "step": 8462 + }, + { + "epoch": 0.9293872172194158, + "grad_norm": 2.3778464794158936, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7239781022071838, + "num_tokens": 217611907.0, + "step": 8463 + }, + { + "epoch": 0.9294970349220294, + "grad_norm": 2.5646424293518066, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7020329833030701, + "num_tokens": 217631659.0, + "step": 8464 + }, + { + "epoch": 0.929606852624643, + "grad_norm": 1.8355721235275269, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7267564535140991, + "num_tokens": 217666275.0, + "step": 8465 + }, + { + "epoch": 0.9297166703272568, + "grad_norm": 2.0609171390533447, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7100731730461121, + "num_tokens": 217696231.0, + "step": 8466 + }, + { + "epoch": 0.9298264880298704, + "grad_norm": 2.1729793548583984, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7157825827598572, + "num_tokens": 217720567.0, + "step": 8467 + }, + { + "epoch": 0.9299363057324841, + "grad_norm": 2.308483123779297, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7003006339073181, + "num_tokens": 217742841.0, + "step": 8468 + }, + { + "epoch": 0.9300461234350977, + "grad_norm": 2.1897969245910645, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7242761254310608, + "num_tokens": 217766888.0, + "step": 8469 + }, + { + "epoch": 0.9301559411377114, + "grad_norm": 2.2040417194366455, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7039810419082642, + "num_tokens": 217793074.0, + "step": 8470 + }, + { + "epoch": 0.930265758840325, + "grad_norm": 2.0100855827331543, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7167137861251831, + "num_tokens": 217820864.0, + "step": 8471 + }, + { + "epoch": 0.9303755765429387, + "grad_norm": 2.2946038246154785, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7051263451576233, + "num_tokens": 217845344.0, + "step": 8472 + }, + { + "epoch": 0.9304853942455524, + "grad_norm": 1.9666105508804321, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6944514513015747, + "num_tokens": 217876339.0, + "step": 8473 + }, + { + "epoch": 0.9305952119481661, + "grad_norm": 2.451368808746338, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7364314794540405, + "num_tokens": 217895873.0, + "step": 8474 + }, + { + "epoch": 0.9307050296507797, + "grad_norm": 2.1571474075317383, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.708991289138794, + "num_tokens": 217921150.0, + "step": 8475 + }, + { + "epoch": 0.9308148473533934, + "grad_norm": 2.4252099990844727, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7223683595657349, + "num_tokens": 217941830.0, + "step": 8476 + }, + { + "epoch": 0.930924665056007, + "grad_norm": 2.5912821292877197, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7118635177612305, + "num_tokens": 217960963.0, + "step": 8477 + }, + { + "epoch": 0.9310344827586207, + "grad_norm": 2.678300619125366, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7168737053871155, + "num_tokens": 217978818.0, + "step": 8478 + }, + { + "epoch": 0.9311443004612343, + "grad_norm": 2.1258580684661865, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7085764408111572, + "num_tokens": 218006441.0, + "step": 8479 + }, + { + "epoch": 0.9312541181638481, + "grad_norm": 2.563338279724121, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7152344584465027, + "num_tokens": 218026633.0, + "step": 8480 + }, + { + "epoch": 0.9313639358664617, + "grad_norm": 2.205264091491699, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6955558061599731, + "num_tokens": 218053316.0, + "step": 8481 + }, + { + "epoch": 0.9314737535690754, + "grad_norm": 2.307720899581909, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7044256329536438, + "num_tokens": 218080799.0, + "step": 8482 + }, + { + "epoch": 0.931583571271689, + "grad_norm": 2.301548957824707, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6909610629081726, + "num_tokens": 218104453.0, + "step": 8483 + }, + { + "epoch": 0.9316933889743026, + "grad_norm": 2.017120361328125, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7117844820022583, + "num_tokens": 218132203.0, + "step": 8484 + }, + { + "epoch": 0.9318032066769163, + "grad_norm": 2.050818681716919, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7141531705856323, + "num_tokens": 218159400.0, + "step": 8485 + }, + { + "epoch": 0.9319130243795299, + "grad_norm": 2.146146774291992, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6822344064712524, + "num_tokens": 218186787.0, + "step": 8486 + }, + { + "epoch": 0.9320228420821436, + "grad_norm": 2.136537790298462, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7004494667053223, + "num_tokens": 218213909.0, + "step": 8487 + }, + { + "epoch": 0.9321326597847573, + "grad_norm": 2.13909912109375, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6983347535133362, + "num_tokens": 218239549.0, + "step": 8488 + }, + { + "epoch": 0.932242477487371, + "grad_norm": 2.264016628265381, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7054537534713745, + "num_tokens": 218262964.0, + "step": 8489 + }, + { + "epoch": 0.9323522951899846, + "grad_norm": 2.4245991706848145, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6963341236114502, + "num_tokens": 218284464.0, + "step": 8490 + }, + { + "epoch": 0.9324621128925983, + "grad_norm": 2.1343863010406494, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7106311321258545, + "num_tokens": 218308755.0, + "step": 8491 + }, + { + "epoch": 0.9325719305952119, + "grad_norm": 2.2573301792144775, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7076802253723145, + "num_tokens": 218333234.0, + "step": 8492 + }, + { + "epoch": 0.9326817482978256, + "grad_norm": 2.0576772689819336, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7118090391159058, + "num_tokens": 218360720.0, + "step": 8493 + }, + { + "epoch": 0.9327915660004392, + "grad_norm": 2.1654186248779297, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6967995762825012, + "num_tokens": 218387791.0, + "step": 8494 + }, + { + "epoch": 0.932901383703053, + "grad_norm": 2.05614972114563, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.730944037437439, + "num_tokens": 218415668.0, + "step": 8495 + }, + { + "epoch": 0.9330112014056666, + "grad_norm": 2.306562900543213, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7009313106536865, + "num_tokens": 218439338.0, + "step": 8496 + }, + { + "epoch": 0.9331210191082803, + "grad_norm": 2.174607992172241, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7187037467956543, + "num_tokens": 218464395.0, + "step": 8497 + }, + { + "epoch": 0.9332308368108939, + "grad_norm": 2.342813491821289, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7281240820884705, + "num_tokens": 218486749.0, + "step": 8498 + }, + { + "epoch": 0.9333406545135076, + "grad_norm": 2.3062942028045654, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7240802049636841, + "num_tokens": 218509294.0, + "step": 8499 + }, + { + "epoch": 0.9334504722161212, + "grad_norm": 2.2619831562042236, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7048070430755615, + "num_tokens": 218534589.0, + "step": 8500 + }, + { + "epoch": 0.9335602899187349, + "grad_norm": 1.9251443147659302, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6974058151245117, + "num_tokens": 218564991.0, + "step": 8501 + }, + { + "epoch": 0.9336701076213486, + "grad_norm": 2.2443768978118896, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7010107040405273, + "num_tokens": 218588149.0, + "step": 8502 + }, + { + "epoch": 0.9337799253239623, + "grad_norm": 2.076859712600708, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7193825244903564, + "num_tokens": 218614682.0, + "step": 8503 + }, + { + "epoch": 0.9338897430265759, + "grad_norm": 2.177361011505127, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7078160643577576, + "num_tokens": 218641695.0, + "step": 8504 + }, + { + "epoch": 0.9339995607291895, + "grad_norm": 2.013093948364258, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7052050232887268, + "num_tokens": 218669366.0, + "step": 8505 + }, + { + "epoch": 0.9341093784318032, + "grad_norm": 2.348686456680298, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7065480351448059, + "num_tokens": 218691807.0, + "step": 8506 + }, + { + "epoch": 0.9342191961344168, + "grad_norm": 2.118044376373291, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.711716890335083, + "num_tokens": 218718142.0, + "step": 8507 + }, + { + "epoch": 0.9343290138370305, + "grad_norm": 2.560030937194824, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7121275663375854, + "num_tokens": 218737673.0, + "step": 8508 + }, + { + "epoch": 0.9344388315396442, + "grad_norm": 2.3949618339538574, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7245216965675354, + "num_tokens": 218760129.0, + "step": 8509 + }, + { + "epoch": 0.9345486492422579, + "grad_norm": 2.404554605484009, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7153738737106323, + "num_tokens": 218780834.0, + "step": 8510 + }, + { + "epoch": 0.9346584669448715, + "grad_norm": 2.216344118118286, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7183426022529602, + "num_tokens": 218803839.0, + "step": 8511 + }, + { + "epoch": 0.9347682846474852, + "grad_norm": 2.2697949409484863, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7078807950019836, + "num_tokens": 218827498.0, + "step": 8512 + }, + { + "epoch": 0.9348781023500988, + "grad_norm": 2.166752815246582, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6971107721328735, + "num_tokens": 218852052.0, + "step": 8513 + }, + { + "epoch": 0.9349879200527125, + "grad_norm": 2.069005012512207, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7041276693344116, + "num_tokens": 218880640.0, + "step": 8514 + }, + { + "epoch": 0.9350977377553261, + "grad_norm": 2.237151861190796, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7250587940216064, + "num_tokens": 218901135.0, + "step": 8515 + }, + { + "epoch": 0.9352075554579398, + "grad_norm": 1.917986512184143, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6879377961158752, + "num_tokens": 218933234.0, + "step": 8516 + }, + { + "epoch": 0.9353173731605535, + "grad_norm": 2.1393442153930664, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7028091549873352, + "num_tokens": 218958999.0, + "step": 8517 + }, + { + "epoch": 0.9354271908631672, + "grad_norm": 2.2508931159973145, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7068687081336975, + "num_tokens": 218984318.0, + "step": 8518 + }, + { + "epoch": 0.9355370085657808, + "grad_norm": 2.3113887310028076, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.715524435043335, + "num_tokens": 219008710.0, + "step": 8519 + }, + { + "epoch": 0.9356468262683945, + "grad_norm": 2.215163230895996, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7068625688552856, + "num_tokens": 219033566.0, + "step": 8520 + }, + { + "epoch": 0.9357566439710081, + "grad_norm": 2.1893715858459473, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7071095705032349, + "num_tokens": 219058081.0, + "step": 8521 + }, + { + "epoch": 0.9358664616736218, + "grad_norm": 2.2414920330047607, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7142055034637451, + "num_tokens": 219082675.0, + "step": 8522 + }, + { + "epoch": 0.9359762793762354, + "grad_norm": 2.0593416690826416, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7030400037765503, + "num_tokens": 219110692.0, + "step": 8523 + }, + { + "epoch": 0.9360860970788492, + "grad_norm": 2.5744245052337646, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7232125997543335, + "num_tokens": 219127822.0, + "step": 8524 + }, + { + "epoch": 0.9361959147814628, + "grad_norm": 1.9121127128601074, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7223252654075623, + "num_tokens": 219158183.0, + "step": 8525 + }, + { + "epoch": 0.9363057324840764, + "grad_norm": 2.092790126800537, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7081524729728699, + "num_tokens": 219184546.0, + "step": 8526 + }, + { + "epoch": 0.9364155501866901, + "grad_norm": 2.1556015014648438, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7271318435668945, + "num_tokens": 219207111.0, + "step": 8527 + }, + { + "epoch": 0.9365253678893037, + "grad_norm": 2.2567849159240723, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7359952926635742, + "num_tokens": 219229697.0, + "step": 8528 + }, + { + "epoch": 0.9366351855919174, + "grad_norm": 2.1912477016448975, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.737659215927124, + "num_tokens": 219251802.0, + "step": 8529 + }, + { + "epoch": 0.936745003294531, + "grad_norm": 2.3030762672424316, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.719985842704773, + "num_tokens": 219274742.0, + "step": 8530 + }, + { + "epoch": 0.9368548209971448, + "grad_norm": 2.288315773010254, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.690362811088562, + "num_tokens": 219303329.0, + "step": 8531 + }, + { + "epoch": 0.9369646386997584, + "grad_norm": 2.0975990295410156, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7007498741149902, + "num_tokens": 219331378.0, + "step": 8532 + }, + { + "epoch": 0.9370744564023721, + "grad_norm": 2.623849868774414, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7104700803756714, + "num_tokens": 219350043.0, + "step": 8533 + }, + { + "epoch": 0.9371842741049857, + "grad_norm": 2.302616834640503, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7116107940673828, + "num_tokens": 219373297.0, + "step": 8534 + }, + { + "epoch": 0.9372940918075994, + "grad_norm": 2.185983419418335, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7067683935165405, + "num_tokens": 219399273.0, + "step": 8535 + }, + { + "epoch": 0.937403909510213, + "grad_norm": 1.9342997074127197, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7204632759094238, + "num_tokens": 219429476.0, + "step": 8536 + }, + { + "epoch": 0.9375137272128267, + "grad_norm": 2.0333361625671387, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6951496005058289, + "num_tokens": 219458228.0, + "step": 8537 + }, + { + "epoch": 0.9376235449154404, + "grad_norm": 2.0521256923675537, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7039435505867004, + "num_tokens": 219486864.0, + "step": 8538 + }, + { + "epoch": 0.9377333626180541, + "grad_norm": 2.1336491107940674, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7007308006286621, + "num_tokens": 219513775.0, + "step": 8539 + }, + { + "epoch": 0.9378431803206677, + "grad_norm": 1.9868029356002808, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7100726962089539, + "num_tokens": 219542517.0, + "step": 8540 + }, + { + "epoch": 0.9379529980232814, + "grad_norm": 2.212308645248413, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.713020920753479, + "num_tokens": 219566447.0, + "step": 8541 + }, + { + "epoch": 0.938062815725895, + "grad_norm": 2.381230354309082, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7158552408218384, + "num_tokens": 219589386.0, + "step": 8542 + }, + { + "epoch": 0.9381726334285087, + "grad_norm": 2.312129497528076, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7135546803474426, + "num_tokens": 219613377.0, + "step": 8543 + }, + { + "epoch": 0.9382824511311223, + "grad_norm": 2.0410709381103516, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.701940655708313, + "num_tokens": 219642652.0, + "step": 8544 + }, + { + "epoch": 0.9383922688337359, + "grad_norm": 2.0901830196380615, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7108237147331238, + "num_tokens": 219668344.0, + "step": 8545 + }, + { + "epoch": 0.9385020865363497, + "grad_norm": 2.0968611240386963, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7085578441619873, + "num_tokens": 219694901.0, + "step": 8546 + }, + { + "epoch": 0.9386119042389633, + "grad_norm": 1.8976095914840698, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6933051943778992, + "num_tokens": 219726154.0, + "step": 8547 + }, + { + "epoch": 0.938721721941577, + "grad_norm": 2.281386613845825, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7031223773956299, + "num_tokens": 219751061.0, + "step": 8548 + }, + { + "epoch": 0.9388315396441906, + "grad_norm": 2.2842345237731934, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7115426659584045, + "num_tokens": 219774018.0, + "step": 8549 + }, + { + "epoch": 0.9389413573468043, + "grad_norm": 2.148120641708374, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6966418027877808, + "num_tokens": 219800835.0, + "step": 8550 + }, + { + "epoch": 0.9390511750494179, + "grad_norm": 2.0889339447021484, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7342368364334106, + "num_tokens": 219826888.0, + "step": 8551 + }, + { + "epoch": 0.9391609927520316, + "grad_norm": 2.328795909881592, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7258244752883911, + "num_tokens": 219849571.0, + "step": 8552 + }, + { + "epoch": 0.9392708104546453, + "grad_norm": 2.6184916496276855, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6994627714157104, + "num_tokens": 219870028.0, + "step": 8553 + }, + { + "epoch": 0.939380628157259, + "grad_norm": 2.697401285171509, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.697088897228241, + "num_tokens": 219888021.0, + "step": 8554 + }, + { + "epoch": 0.9394904458598726, + "grad_norm": 2.1919443607330322, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7078255414962769, + "num_tokens": 219913805.0, + "step": 8555 + }, + { + "epoch": 0.9396002635624863, + "grad_norm": 2.1424472332000732, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6907088756561279, + "num_tokens": 219940509.0, + "step": 8556 + }, + { + "epoch": 0.9397100812650999, + "grad_norm": 2.0732626914978027, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7100675106048584, + "num_tokens": 219969098.0, + "step": 8557 + }, + { + "epoch": 0.9398198989677136, + "grad_norm": 2.163727283477783, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6913803815841675, + "num_tokens": 219994491.0, + "step": 8558 + }, + { + "epoch": 0.9399297166703272, + "grad_norm": 2.5308167934417725, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7260586023330688, + "num_tokens": 220013532.0, + "step": 8559 + }, + { + "epoch": 0.940039534372941, + "grad_norm": 2.237863779067993, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.702260434627533, + "num_tokens": 220037518.0, + "step": 8560 + }, + { + "epoch": 0.9401493520755546, + "grad_norm": 2.145322799682617, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7186546325683594, + "num_tokens": 220063151.0, + "step": 8561 + }, + { + "epoch": 0.9402591697781683, + "grad_norm": 1.9936392307281494, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7025660276412964, + "num_tokens": 220093938.0, + "step": 8562 + }, + { + "epoch": 0.9403689874807819, + "grad_norm": 2.3147079944610596, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7202929258346558, + "num_tokens": 220115601.0, + "step": 8563 + }, + { + "epoch": 0.9404788051833955, + "grad_norm": 1.980933666229248, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7108923196792603, + "num_tokens": 220143785.0, + "step": 8564 + }, + { + "epoch": 0.9405886228860092, + "grad_norm": 2.0385217666625977, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7109130620956421, + "num_tokens": 220173822.0, + "step": 8565 + }, + { + "epoch": 0.9406984405886228, + "grad_norm": 2.315706968307495, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7239713668823242, + "num_tokens": 220195772.0, + "step": 8566 + }, + { + "epoch": 0.9408082582912366, + "grad_norm": 2.2355213165283203, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7240937352180481, + "num_tokens": 220219010.0, + "step": 8567 + }, + { + "epoch": 0.9409180759938502, + "grad_norm": 1.987402081489563, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7077885866165161, + "num_tokens": 220248896.0, + "step": 8568 + }, + { + "epoch": 0.9410278936964639, + "grad_norm": 2.0346758365631104, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7072839140892029, + "num_tokens": 220278145.0, + "step": 8569 + }, + { + "epoch": 0.9411377113990775, + "grad_norm": 2.4522781372070312, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7225813865661621, + "num_tokens": 220299298.0, + "step": 8570 + }, + { + "epoch": 0.9412475291016912, + "grad_norm": 2.1166296005249023, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7249179482460022, + "num_tokens": 220326082.0, + "step": 8571 + }, + { + "epoch": 0.9413573468043048, + "grad_norm": 2.2479944229125977, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.69431471824646, + "num_tokens": 220351484.0, + "step": 8572 + }, + { + "epoch": 0.9414671645069185, + "grad_norm": 2.211660146713257, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7336714267730713, + "num_tokens": 220373443.0, + "step": 8573 + }, + { + "epoch": 0.9415769822095322, + "grad_norm": 2.0742404460906982, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7142294645309448, + "num_tokens": 220402994.0, + "step": 8574 + }, + { + "epoch": 0.9416867999121459, + "grad_norm": 2.0888195037841797, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6990035176277161, + "num_tokens": 220430892.0, + "step": 8575 + }, + { + "epoch": 0.9417966176147595, + "grad_norm": 2.070841073989868, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7083175778388977, + "num_tokens": 220459917.0, + "step": 8576 + }, + { + "epoch": 0.9419064353173732, + "grad_norm": 2.0886428356170654, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7053358554840088, + "num_tokens": 220486858.0, + "step": 8577 + }, + { + "epoch": 0.9420162530199868, + "grad_norm": 2.5028698444366455, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7208365201950073, + "num_tokens": 220507035.0, + "step": 8578 + }, + { + "epoch": 0.9421260707226005, + "grad_norm": 2.17889142036438, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7241994738578796, + "num_tokens": 220533665.0, + "step": 8579 + }, + { + "epoch": 0.9422358884252141, + "grad_norm": 2.3824069499969482, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7144358158111572, + "num_tokens": 220556072.0, + "step": 8580 + }, + { + "epoch": 0.9423457061278278, + "grad_norm": 2.143990993499756, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7120062112808228, + "num_tokens": 220581438.0, + "step": 8581 + }, + { + "epoch": 0.9424555238304415, + "grad_norm": 2.1390390396118164, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6848454475402832, + "num_tokens": 220607967.0, + "step": 8582 + }, + { + "epoch": 0.9425653415330552, + "grad_norm": 2.6456544399261475, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7162761688232422, + "num_tokens": 220625944.0, + "step": 8583 + }, + { + "epoch": 0.9426751592356688, + "grad_norm": 2.029980421066284, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6917705535888672, + "num_tokens": 220654710.0, + "step": 8584 + }, + { + "epoch": 0.9427849769382824, + "grad_norm": 2.3555240631103516, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.750598132610321, + "num_tokens": 220675409.0, + "step": 8585 + }, + { + "epoch": 0.9428947946408961, + "grad_norm": 1.935465693473816, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7002981901168823, + "num_tokens": 220707920.0, + "step": 8586 + }, + { + "epoch": 0.9430046123435097, + "grad_norm": 2.1604208946228027, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.710795521736145, + "num_tokens": 220732975.0, + "step": 8587 + }, + { + "epoch": 0.9431144300461234, + "grad_norm": 2.3297438621520996, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7207776308059692, + "num_tokens": 220756670.0, + "step": 8588 + }, + { + "epoch": 0.9432242477487371, + "grad_norm": 2.4629929065704346, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7395402789115906, + "num_tokens": 220775565.0, + "step": 8589 + }, + { + "epoch": 0.9433340654513508, + "grad_norm": 2.409172296524048, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7184111475944519, + "num_tokens": 220797575.0, + "step": 8590 + }, + { + "epoch": 0.9434438831539644, + "grad_norm": 2.222609281539917, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7039036154747009, + "num_tokens": 220821775.0, + "step": 8591 + }, + { + "epoch": 0.9435537008565781, + "grad_norm": 2.2294623851776123, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7086001038551331, + "num_tokens": 220845721.0, + "step": 8592 + }, + { + "epoch": 0.9436635185591917, + "grad_norm": 2.1922709941864014, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7322348952293396, + "num_tokens": 220869183.0, + "step": 8593 + }, + { + "epoch": 0.9437733362618054, + "grad_norm": 1.8379758596420288, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6965897083282471, + "num_tokens": 220902916.0, + "step": 8594 + }, + { + "epoch": 0.943883153964419, + "grad_norm": 2.0256035327911377, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6908624768257141, + "num_tokens": 220930206.0, + "step": 8595 + }, + { + "epoch": 0.9439929716670328, + "grad_norm": 2.229304075241089, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7144953012466431, + "num_tokens": 220954849.0, + "step": 8596 + }, + { + "epoch": 0.9441027893696464, + "grad_norm": 1.7978013753890991, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7237933278083801, + "num_tokens": 220986172.0, + "step": 8597 + }, + { + "epoch": 0.9442126070722601, + "grad_norm": 2.219679594039917, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7312376499176025, + "num_tokens": 221012013.0, + "step": 8598 + }, + { + "epoch": 0.9443224247748737, + "grad_norm": 2.2540290355682373, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7020429372787476, + "num_tokens": 221035059.0, + "step": 8599 + }, + { + "epoch": 0.9444322424774874, + "grad_norm": 2.2030835151672363, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6965253353118896, + "num_tokens": 221062001.0, + "step": 8600 + }, + { + "epoch": 0.944542060180101, + "grad_norm": 2.2907936573028564, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7013174295425415, + "num_tokens": 221084894.0, + "step": 8601 + }, + { + "epoch": 0.9446518778827147, + "grad_norm": 2.0597007274627686, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.7081743478775024, + "num_tokens": 221112884.0, + "step": 8602 + }, + { + "epoch": 0.9447616955853284, + "grad_norm": 2.2547848224639893, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7168564796447754, + "num_tokens": 221136972.0, + "step": 8603 + }, + { + "epoch": 0.944871513287942, + "grad_norm": 2.1909217834472656, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.72246253490448, + "num_tokens": 221160962.0, + "step": 8604 + }, + { + "epoch": 0.9449813309905557, + "grad_norm": 2.062966823577881, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6990750432014465, + "num_tokens": 221187565.0, + "step": 8605 + }, + { + "epoch": 0.9450911486931693, + "grad_norm": 2.1305270195007324, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6925853490829468, + "num_tokens": 221215788.0, + "step": 8606 + }, + { + "epoch": 0.945200966395783, + "grad_norm": 2.081394910812378, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6785701513290405, + "num_tokens": 221246516.0, + "step": 8607 + }, + { + "epoch": 0.9453107840983966, + "grad_norm": 1.7687904834747314, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7009049654006958, + "num_tokens": 221279843.0, + "step": 8608 + }, + { + "epoch": 0.9454206018010103, + "grad_norm": 2.1387808322906494, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6888130903244019, + "num_tokens": 221308314.0, + "step": 8609 + }, + { + "epoch": 0.9455304195036239, + "grad_norm": 2.0967190265655518, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6846885085105896, + "num_tokens": 221335889.0, + "step": 8610 + }, + { + "epoch": 0.9456402372062377, + "grad_norm": 1.8797107934951782, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7113244533538818, + "num_tokens": 221366445.0, + "step": 8611 + }, + { + "epoch": 0.9457500549088513, + "grad_norm": 2.039109468460083, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7304781675338745, + "num_tokens": 221390870.0, + "step": 8612 + }, + { + "epoch": 0.945859872611465, + "grad_norm": 2.2367312908172607, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7033677101135254, + "num_tokens": 221414388.0, + "step": 8613 + }, + { + "epoch": 0.9459696903140786, + "grad_norm": 2.8406119346618652, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7135461568832397, + "num_tokens": 221430963.0, + "step": 8614 + }, + { + "epoch": 0.9460795080166923, + "grad_norm": 2.068927049636841, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6871728897094727, + "num_tokens": 221457183.0, + "step": 8615 + }, + { + "epoch": 0.9461893257193059, + "grad_norm": 2.2293028831481934, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.695144772529602, + "num_tokens": 221482145.0, + "step": 8616 + }, + { + "epoch": 0.9462991434219196, + "grad_norm": 2.37898325920105, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7197357416152954, + "num_tokens": 221502206.0, + "step": 8617 + }, + { + "epoch": 0.9464089611245333, + "grad_norm": 2.050556182861328, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7002296447753906, + "num_tokens": 221529571.0, + "step": 8618 + }, + { + "epoch": 0.946518778827147, + "grad_norm": 2.1964797973632812, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7228347659111023, + "num_tokens": 221553667.0, + "step": 8619 + }, + { + "epoch": 0.9466285965297606, + "grad_norm": 2.0512537956237793, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6821820735931396, + "num_tokens": 221582120.0, + "step": 8620 + }, + { + "epoch": 0.9467384142323743, + "grad_norm": 1.827484130859375, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6856748461723328, + "num_tokens": 221619610.0, + "step": 8621 + }, + { + "epoch": 0.9468482319349879, + "grad_norm": 2.5591773986816406, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7028628587722778, + "num_tokens": 221640626.0, + "step": 8622 + }, + { + "epoch": 0.9469580496376016, + "grad_norm": 2.163902997970581, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6879334449768066, + "num_tokens": 221668660.0, + "step": 8623 + }, + { + "epoch": 0.9470678673402152, + "grad_norm": 2.150352716445923, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7150160670280457, + "num_tokens": 221694949.0, + "step": 8624 + }, + { + "epoch": 0.947177685042829, + "grad_norm": 2.1835291385650635, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7187721133232117, + "num_tokens": 221720658.0, + "step": 8625 + }, + { + "epoch": 0.9472875027454426, + "grad_norm": 2.103956937789917, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7031209468841553, + "num_tokens": 221748621.0, + "step": 8626 + }, + { + "epoch": 0.9473973204480562, + "grad_norm": 2.322528123855591, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7313552498817444, + "num_tokens": 221770585.0, + "step": 8627 + }, + { + "epoch": 0.9475071381506699, + "grad_norm": 2.260793924331665, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7200395464897156, + "num_tokens": 221794666.0, + "step": 8628 + }, + { + "epoch": 0.9476169558532835, + "grad_norm": 2.206352949142456, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7153863906860352, + "num_tokens": 221820991.0, + "step": 8629 + }, + { + "epoch": 0.9477267735558972, + "grad_norm": 2.166361093521118, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7072592973709106, + "num_tokens": 221847252.0, + "step": 8630 + }, + { + "epoch": 0.9478365912585108, + "grad_norm": 2.1589038372039795, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6979143619537354, + "num_tokens": 221873419.0, + "step": 8631 + }, + { + "epoch": 0.9479464089611246, + "grad_norm": 2.604438304901123, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7125853300094604, + "num_tokens": 221892894.0, + "step": 8632 + }, + { + "epoch": 0.9480562266637382, + "grad_norm": 2.582029104232788, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7274174690246582, + "num_tokens": 221912856.0, + "step": 8633 + }, + { + "epoch": 0.9481660443663519, + "grad_norm": 2.0559427738189697, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7170224189758301, + "num_tokens": 221938671.0, + "step": 8634 + }, + { + "epoch": 0.9482758620689655, + "grad_norm": 2.2474687099456787, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7116812467575073, + "num_tokens": 221962736.0, + "step": 8635 + }, + { + "epoch": 0.9483856797715792, + "grad_norm": 2.325137138366699, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7142329812049866, + "num_tokens": 221985709.0, + "step": 8636 + }, + { + "epoch": 0.9484954974741928, + "grad_norm": 2.3205626010894775, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7352585792541504, + "num_tokens": 222007147.0, + "step": 8637 + }, + { + "epoch": 0.9486053151768065, + "grad_norm": 2.1702699661254883, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6994965672492981, + "num_tokens": 222034344.0, + "step": 8638 + }, + { + "epoch": 0.9487151328794201, + "grad_norm": 2.150446891784668, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7282412648200989, + "num_tokens": 222058129.0, + "step": 8639 + }, + { + "epoch": 0.9488249505820339, + "grad_norm": 2.6658689975738525, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7421659231185913, + "num_tokens": 222076364.0, + "step": 8640 + }, + { + "epoch": 0.9489347682846475, + "grad_norm": 2.35910964012146, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6998164057731628, + "num_tokens": 222099750.0, + "step": 8641 + }, + { + "epoch": 0.9490445859872612, + "grad_norm": 2.3753349781036377, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7093009352684021, + "num_tokens": 222120832.0, + "step": 8642 + }, + { + "epoch": 0.9491544036898748, + "grad_norm": 2.0840957164764404, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7216640710830688, + "num_tokens": 222146337.0, + "step": 8643 + }, + { + "epoch": 0.9492642213924884, + "grad_norm": 2.0833780765533447, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7061534523963928, + "num_tokens": 222176037.0, + "step": 8644 + }, + { + "epoch": 0.9493740390951021, + "grad_norm": 2.125300645828247, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6986021995544434, + "num_tokens": 222203019.0, + "step": 8645 + }, + { + "epoch": 0.9494838567977157, + "grad_norm": 2.1575675010681152, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.718831479549408, + "num_tokens": 222227105.0, + "step": 8646 + }, + { + "epoch": 0.9495936745003295, + "grad_norm": 2.182209014892578, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7054176330566406, + "num_tokens": 222252292.0, + "step": 8647 + }, + { + "epoch": 0.9497034922029431, + "grad_norm": 2.3609466552734375, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7062886953353882, + "num_tokens": 222274039.0, + "step": 8648 + }, + { + "epoch": 0.9498133099055568, + "grad_norm": 1.9959293603897095, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6867337226867676, + "num_tokens": 222302052.0, + "step": 8649 + }, + { + "epoch": 0.9499231276081704, + "grad_norm": 2.000460147857666, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.7026939392089844, + "num_tokens": 222331630.0, + "step": 8650 + }, + { + "epoch": 0.9500329453107841, + "grad_norm": 2.296212911605835, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6985582709312439, + "num_tokens": 222356378.0, + "step": 8651 + }, + { + "epoch": 0.9501427630133977, + "grad_norm": 2.315239667892456, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7162915468215942, + "num_tokens": 222376754.0, + "step": 8652 + }, + { + "epoch": 0.9502525807160114, + "grad_norm": 2.215834379196167, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7095837593078613, + "num_tokens": 222400172.0, + "step": 8653 + }, + { + "epoch": 0.9503623984186251, + "grad_norm": 1.8465255498886108, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6983721256256104, + "num_tokens": 222432733.0, + "step": 8654 + }, + { + "epoch": 0.9504722161212388, + "grad_norm": 2.084325075149536, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7072374224662781, + "num_tokens": 222460385.0, + "step": 8655 + }, + { + "epoch": 0.9505820338238524, + "grad_norm": 2.1978838443756104, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6987916231155396, + "num_tokens": 222485746.0, + "step": 8656 + }, + { + "epoch": 0.9506918515264661, + "grad_norm": 1.9168579578399658, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6985067129135132, + "num_tokens": 222517358.0, + "step": 8657 + }, + { + "epoch": 0.9508016692290797, + "grad_norm": 2.1072542667388916, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6993521451950073, + "num_tokens": 222543238.0, + "step": 8658 + }, + { + "epoch": 0.9509114869316934, + "grad_norm": 2.2856216430664062, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7306458353996277, + "num_tokens": 222565302.0, + "step": 8659 + }, + { + "epoch": 0.951021304634307, + "grad_norm": 2.2613346576690674, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7197433710098267, + "num_tokens": 222587694.0, + "step": 8660 + }, + { + "epoch": 0.9511311223369208, + "grad_norm": 2.102698564529419, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6955422163009644, + "num_tokens": 222613520.0, + "step": 8661 + }, + { + "epoch": 0.9512409400395344, + "grad_norm": 2.2008450031280518, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7115970849990845, + "num_tokens": 222636422.0, + "step": 8662 + }, + { + "epoch": 0.9513507577421481, + "grad_norm": 2.1784608364105225, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7057214379310608, + "num_tokens": 222661313.0, + "step": 8663 + }, + { + "epoch": 0.9514605754447617, + "grad_norm": 2.5494730472564697, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7309090495109558, + "num_tokens": 222679593.0, + "step": 8664 + }, + { + "epoch": 0.9515703931473753, + "grad_norm": 2.182088613510132, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6886863112449646, + "num_tokens": 222705788.0, + "step": 8665 + }, + { + "epoch": 0.951680210849989, + "grad_norm": 2.1018452644348145, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6829740405082703, + "num_tokens": 222734103.0, + "step": 8666 + }, + { + "epoch": 0.9517900285526026, + "grad_norm": 2.383591413497925, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7067028284072876, + "num_tokens": 222758629.0, + "step": 8667 + }, + { + "epoch": 0.9518998462552163, + "grad_norm": 2.1890861988067627, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7109402418136597, + "num_tokens": 222783662.0, + "step": 8668 + }, + { + "epoch": 0.95200966395783, + "grad_norm": 2.3207497596740723, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.711060643196106, + "num_tokens": 222806736.0, + "step": 8669 + }, + { + "epoch": 0.9521194816604437, + "grad_norm": 2.0955922603607178, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7102449536323547, + "num_tokens": 222834867.0, + "step": 8670 + }, + { + "epoch": 0.9522292993630573, + "grad_norm": 2.0321712493896484, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.714854896068573, + "num_tokens": 222863877.0, + "step": 8671 + }, + { + "epoch": 0.952339117065671, + "grad_norm": 2.2213592529296875, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7081968188285828, + "num_tokens": 222889550.0, + "step": 8672 + }, + { + "epoch": 0.9524489347682846, + "grad_norm": 2.176274299621582, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6998302340507507, + "num_tokens": 222916991.0, + "step": 8673 + }, + { + "epoch": 0.9525587524708983, + "grad_norm": 2.2276198863983154, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7127220034599304, + "num_tokens": 222938971.0, + "step": 8674 + }, + { + "epoch": 0.9526685701735119, + "grad_norm": 2.0387701988220215, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7324409484863281, + "num_tokens": 222963892.0, + "step": 8675 + }, + { + "epoch": 0.9527783878761257, + "grad_norm": 1.894028663635254, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6970776915550232, + "num_tokens": 222996541.0, + "step": 8676 + }, + { + "epoch": 0.9528882055787393, + "grad_norm": 2.4098243713378906, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.727267861366272, + "num_tokens": 223017770.0, + "step": 8677 + }, + { + "epoch": 0.952998023281353, + "grad_norm": 2.4615085124969482, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7286381721496582, + "num_tokens": 223038798.0, + "step": 8678 + }, + { + "epoch": 0.9531078409839666, + "grad_norm": 1.8171708583831787, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6898624897003174, + "num_tokens": 223073283.0, + "step": 8679 + }, + { + "epoch": 0.9532176586865803, + "grad_norm": 2.167652130126953, + "learning_rate": 1e-06, + "loss": 1.0829, + "mean_token_accuracy": 0.669522762298584, + "num_tokens": 223102349.0, + "step": 8680 + }, + { + "epoch": 0.9533274763891939, + "grad_norm": 1.8990617990493774, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7057029008865356, + "num_tokens": 223134041.0, + "step": 8681 + }, + { + "epoch": 0.9534372940918076, + "grad_norm": 2.247220039367676, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7388620972633362, + "num_tokens": 223159340.0, + "step": 8682 + }, + { + "epoch": 0.9535471117944213, + "grad_norm": 2.0360283851623535, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6898253560066223, + "num_tokens": 223187313.0, + "step": 8683 + }, + { + "epoch": 0.953656929497035, + "grad_norm": 2.1571457386016846, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7029162049293518, + "num_tokens": 223213925.0, + "step": 8684 + }, + { + "epoch": 0.9537667471996486, + "grad_norm": 1.9404107332229614, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.700941264629364, + "num_tokens": 223244760.0, + "step": 8685 + }, + { + "epoch": 0.9538765649022622, + "grad_norm": 2.0928165912628174, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6957845687866211, + "num_tokens": 223273283.0, + "step": 8686 + }, + { + "epoch": 0.9539863826048759, + "grad_norm": 2.1362688541412354, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7129495143890381, + "num_tokens": 223299543.0, + "step": 8687 + }, + { + "epoch": 0.9540962003074895, + "grad_norm": 2.1819190979003906, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7013149261474609, + "num_tokens": 223324783.0, + "step": 8688 + }, + { + "epoch": 0.9542060180101032, + "grad_norm": 2.062762975692749, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.693781852722168, + "num_tokens": 223352786.0, + "step": 8689 + }, + { + "epoch": 0.9543158357127169, + "grad_norm": 2.598449945449829, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7111320495605469, + "num_tokens": 223372299.0, + "step": 8690 + }, + { + "epoch": 0.9544256534153306, + "grad_norm": 2.2290868759155273, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6973552703857422, + "num_tokens": 223396028.0, + "step": 8691 + }, + { + "epoch": 0.9545354711179442, + "grad_norm": 2.681864023208618, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7218494415283203, + "num_tokens": 223414007.0, + "step": 8692 + }, + { + "epoch": 0.9546452888205579, + "grad_norm": 2.2848896980285645, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7324389815330505, + "num_tokens": 223435458.0, + "step": 8693 + }, + { + "epoch": 0.9547551065231715, + "grad_norm": 2.5448644161224365, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7111304998397827, + "num_tokens": 223456793.0, + "step": 8694 + }, + { + "epoch": 0.9548649242257852, + "grad_norm": 2.261650800704956, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7127094864845276, + "num_tokens": 223481506.0, + "step": 8695 + }, + { + "epoch": 0.9549747419283988, + "grad_norm": 2.2663373947143555, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7237618565559387, + "num_tokens": 223504064.0, + "step": 8696 + }, + { + "epoch": 0.9550845596310125, + "grad_norm": 2.3353142738342285, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7186503410339355, + "num_tokens": 223527348.0, + "step": 8697 + }, + { + "epoch": 0.9551943773336262, + "grad_norm": 2.3695571422576904, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7266262173652649, + "num_tokens": 223549778.0, + "step": 8698 + }, + { + "epoch": 0.9553041950362399, + "grad_norm": 2.1671392917633057, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6900204420089722, + "num_tokens": 223577146.0, + "step": 8699 + }, + { + "epoch": 0.9554140127388535, + "grad_norm": 2.3392982482910156, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7308821082115173, + "num_tokens": 223600370.0, + "step": 8700 + }, + { + "epoch": 0.9555238304414672, + "grad_norm": 2.0829954147338867, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7134672403335571, + "num_tokens": 223625612.0, + "step": 8701 + }, + { + "epoch": 0.9556336481440808, + "grad_norm": 2.3326380252838135, + "learning_rate": 1e-06, + "loss": 1.0766, + "mean_token_accuracy": 0.6840785145759583, + "num_tokens": 223650198.0, + "step": 8702 + }, + { + "epoch": 0.9557434658466945, + "grad_norm": 2.0758743286132812, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6860572695732117, + "num_tokens": 223677377.0, + "step": 8703 + }, + { + "epoch": 0.9558532835493081, + "grad_norm": 2.046802520751953, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6915508508682251, + "num_tokens": 223707299.0, + "step": 8704 + }, + { + "epoch": 0.9559631012519219, + "grad_norm": 2.1253509521484375, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7183387279510498, + "num_tokens": 223734947.0, + "step": 8705 + }, + { + "epoch": 0.9560729189545355, + "grad_norm": 2.052292823791504, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7173306345939636, + "num_tokens": 223761498.0, + "step": 8706 + }, + { + "epoch": 0.9561827366571491, + "grad_norm": 1.9885141849517822, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7238010764122009, + "num_tokens": 223789857.0, + "step": 8707 + }, + { + "epoch": 0.9562925543597628, + "grad_norm": 2.2011101245880127, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7195762991905212, + "num_tokens": 223813426.0, + "step": 8708 + }, + { + "epoch": 0.9564023720623764, + "grad_norm": 1.9676350355148315, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7149821519851685, + "num_tokens": 223843433.0, + "step": 8709 + }, + { + "epoch": 0.9565121897649901, + "grad_norm": 2.1518425941467285, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.717585563659668, + "num_tokens": 223868280.0, + "step": 8710 + }, + { + "epoch": 0.9566220074676037, + "grad_norm": 1.8960316181182861, + "learning_rate": 1e-06, + "loss": 1.0797, + "mean_token_accuracy": 0.6759469509124756, + "num_tokens": 223903648.0, + "step": 8711 + }, + { + "epoch": 0.9567318251702175, + "grad_norm": 2.125347852706909, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7440256476402283, + "num_tokens": 223927112.0, + "step": 8712 + }, + { + "epoch": 0.9568416428728311, + "grad_norm": 2.3819844722747803, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7117131352424622, + "num_tokens": 223949698.0, + "step": 8713 + }, + { + "epoch": 0.9569514605754448, + "grad_norm": 2.1398236751556396, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6927236318588257, + "num_tokens": 223977871.0, + "step": 8714 + }, + { + "epoch": 0.9570612782780584, + "grad_norm": 2.293195962905884, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7262786626815796, + "num_tokens": 224000930.0, + "step": 8715 + }, + { + "epoch": 0.9571710959806721, + "grad_norm": 1.985367774963379, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7416061162948608, + "num_tokens": 224028254.0, + "step": 8716 + }, + { + "epoch": 0.9572809136832857, + "grad_norm": 2.1890814304351807, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7217704057693481, + "num_tokens": 224053310.0, + "step": 8717 + }, + { + "epoch": 0.9573907313858994, + "grad_norm": 2.45721173286438, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7110351324081421, + "num_tokens": 224075887.0, + "step": 8718 + }, + { + "epoch": 0.9575005490885131, + "grad_norm": 2.268043041229248, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6917656064033508, + "num_tokens": 224100308.0, + "step": 8719 + }, + { + "epoch": 0.9576103667911268, + "grad_norm": 1.9976922273635864, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7083748579025269, + "num_tokens": 224127741.0, + "step": 8720 + }, + { + "epoch": 0.9577201844937404, + "grad_norm": 1.9667229652404785, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7098292112350464, + "num_tokens": 224155025.0, + "step": 8721 + }, + { + "epoch": 0.9578300021963541, + "grad_norm": 2.0465900897979736, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7142302989959717, + "num_tokens": 224180269.0, + "step": 8722 + }, + { + "epoch": 0.9579398198989677, + "grad_norm": 2.1606197357177734, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.7011916041374207, + "num_tokens": 224207867.0, + "step": 8723 + }, + { + "epoch": 0.9580496376015813, + "grad_norm": 1.9078978300094604, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6834335923194885, + "num_tokens": 224241571.0, + "step": 8724 + }, + { + "epoch": 0.958159455304195, + "grad_norm": 2.2637457847595215, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.724101185798645, + "num_tokens": 224263904.0, + "step": 8725 + }, + { + "epoch": 0.9582692730068088, + "grad_norm": 2.2383358478546143, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6900686025619507, + "num_tokens": 224291479.0, + "step": 8726 + }, + { + "epoch": 0.9583790907094224, + "grad_norm": 2.020617723464966, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6813519597053528, + "num_tokens": 224320853.0, + "step": 8727 + }, + { + "epoch": 0.958488908412036, + "grad_norm": 2.112386703491211, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7239344120025635, + "num_tokens": 224345948.0, + "step": 8728 + }, + { + "epoch": 0.9585987261146497, + "grad_norm": 2.0791687965393066, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7216887474060059, + "num_tokens": 224373089.0, + "step": 8729 + }, + { + "epoch": 0.9587085438172633, + "grad_norm": 2.167987585067749, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7057168483734131, + "num_tokens": 224401098.0, + "step": 8730 + }, + { + "epoch": 0.958818361519877, + "grad_norm": 2.064912796020508, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.70477694272995, + "num_tokens": 224427582.0, + "step": 8731 + }, + { + "epoch": 0.9589281792224906, + "grad_norm": 2.3426239490509033, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7082801461219788, + "num_tokens": 224452484.0, + "step": 8732 + }, + { + "epoch": 0.9590379969251043, + "grad_norm": 2.0589840412139893, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7139869332313538, + "num_tokens": 224481105.0, + "step": 8733 + }, + { + "epoch": 0.959147814627718, + "grad_norm": 2.1084225177764893, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7113549709320068, + "num_tokens": 224506925.0, + "step": 8734 + }, + { + "epoch": 0.9592576323303317, + "grad_norm": 2.1217880249023438, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6952602863311768, + "num_tokens": 224533916.0, + "step": 8735 + }, + { + "epoch": 0.9593674500329453, + "grad_norm": 1.8539209365844727, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6948648691177368, + "num_tokens": 224569357.0, + "step": 8736 + }, + { + "epoch": 0.959477267735559, + "grad_norm": 1.9296890497207642, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.6987968683242798, + "num_tokens": 224601244.0, + "step": 8737 + }, + { + "epoch": 0.9595870854381726, + "grad_norm": 1.9262535572052002, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7291973233222961, + "num_tokens": 224631013.0, + "step": 8738 + }, + { + "epoch": 0.9596969031407863, + "grad_norm": 1.8927801847457886, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6859204769134521, + "num_tokens": 224662660.0, + "step": 8739 + }, + { + "epoch": 0.9598067208433999, + "grad_norm": 2.349005699157715, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7191121578216553, + "num_tokens": 224686395.0, + "step": 8740 + }, + { + "epoch": 0.9599165385460137, + "grad_norm": 2.4032323360443115, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7205092310905457, + "num_tokens": 224707982.0, + "step": 8741 + }, + { + "epoch": 0.9600263562486273, + "grad_norm": 2.2430436611175537, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7069051861763, + "num_tokens": 224732097.0, + "step": 8742 + }, + { + "epoch": 0.960136173951241, + "grad_norm": 2.246680736541748, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7007202506065369, + "num_tokens": 224758779.0, + "step": 8743 + }, + { + "epoch": 0.9602459916538546, + "grad_norm": 2.270385980606079, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7267439365386963, + "num_tokens": 224780648.0, + "step": 8744 + }, + { + "epoch": 0.9603558093564682, + "grad_norm": 2.175165891647339, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.732709527015686, + "num_tokens": 224805071.0, + "step": 8745 + }, + { + "epoch": 0.9604656270590819, + "grad_norm": 2.427290678024292, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.727628231048584, + "num_tokens": 224823837.0, + "step": 8746 + }, + { + "epoch": 0.9605754447616955, + "grad_norm": 2.2236475944519043, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7186166048049927, + "num_tokens": 224848834.0, + "step": 8747 + }, + { + "epoch": 0.9606852624643093, + "grad_norm": 2.2001900672912598, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7022035121917725, + "num_tokens": 224874918.0, + "step": 8748 + }, + { + "epoch": 0.9607950801669229, + "grad_norm": 2.181079626083374, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7033320665359497, + "num_tokens": 224901136.0, + "step": 8749 + }, + { + "epoch": 0.9609048978695366, + "grad_norm": 2.2913553714752197, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7068732976913452, + "num_tokens": 224923785.0, + "step": 8750 + }, + { + "epoch": 0.9610147155721502, + "grad_norm": 2.2724905014038086, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7166798710823059, + "num_tokens": 224946552.0, + "step": 8751 + }, + { + "epoch": 0.9611245332747639, + "grad_norm": 2.3244917392730713, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6965551376342773, + "num_tokens": 224969829.0, + "step": 8752 + }, + { + "epoch": 0.9612343509773775, + "grad_norm": 2.29552960395813, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7267720699310303, + "num_tokens": 224992276.0, + "step": 8753 + }, + { + "epoch": 0.9613441686799912, + "grad_norm": 2.3946449756622314, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7150503396987915, + "num_tokens": 225013798.0, + "step": 8754 + }, + { + "epoch": 0.9614539863826049, + "grad_norm": 2.22896671295166, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7118537425994873, + "num_tokens": 225037265.0, + "step": 8755 + }, + { + "epoch": 0.9615638040852186, + "grad_norm": 1.9661859273910522, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7484237551689148, + "num_tokens": 225063377.0, + "step": 8756 + }, + { + "epoch": 0.9616736217878322, + "grad_norm": 1.9756382703781128, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.710407018661499, + "num_tokens": 225093600.0, + "step": 8757 + }, + { + "epoch": 0.9617834394904459, + "grad_norm": 2.115126609802246, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6965711116790771, + "num_tokens": 225118037.0, + "step": 8758 + }, + { + "epoch": 0.9618932571930595, + "grad_norm": 2.2722220420837402, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6924669146537781, + "num_tokens": 225144523.0, + "step": 8759 + }, + { + "epoch": 0.9620030748956732, + "grad_norm": 1.9171236753463745, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.724616289138794, + "num_tokens": 225172512.0, + "step": 8760 + }, + { + "epoch": 0.9621128925982868, + "grad_norm": 2.2819716930389404, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7122278809547424, + "num_tokens": 225196085.0, + "step": 8761 + }, + { + "epoch": 0.9622227103009005, + "grad_norm": 1.895540714263916, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7049363851547241, + "num_tokens": 225228784.0, + "step": 8762 + }, + { + "epoch": 0.9623325280035142, + "grad_norm": 2.1482787132263184, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6913891434669495, + "num_tokens": 225254540.0, + "step": 8763 + }, + { + "epoch": 0.9624423457061279, + "grad_norm": 2.2188923358917236, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7065063118934631, + "num_tokens": 225278760.0, + "step": 8764 + }, + { + "epoch": 0.9625521634087415, + "grad_norm": 2.1283366680145264, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7283741235733032, + "num_tokens": 225304258.0, + "step": 8765 + }, + { + "epoch": 0.9626619811113551, + "grad_norm": 2.1683216094970703, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6823316812515259, + "num_tokens": 225331155.0, + "step": 8766 + }, + { + "epoch": 0.9627717988139688, + "grad_norm": 2.347644567489624, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7189376354217529, + "num_tokens": 225353445.0, + "step": 8767 + }, + { + "epoch": 0.9628816165165824, + "grad_norm": 2.107410192489624, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7041999101638794, + "num_tokens": 225380211.0, + "step": 8768 + }, + { + "epoch": 0.9629914342191961, + "grad_norm": 1.8958364725112915, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7095444798469543, + "num_tokens": 225412476.0, + "step": 8769 + }, + { + "epoch": 0.9631012519218098, + "grad_norm": 2.092576742172241, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.716825008392334, + "num_tokens": 225439425.0, + "step": 8770 + }, + { + "epoch": 0.9632110696244235, + "grad_norm": 2.2474071979522705, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6859581470489502, + "num_tokens": 225465350.0, + "step": 8771 + }, + { + "epoch": 0.9633208873270371, + "grad_norm": 2.058151960372925, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7080376744270325, + "num_tokens": 225494606.0, + "step": 8772 + }, + { + "epoch": 0.9634307050296508, + "grad_norm": 2.2991597652435303, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7113600969314575, + "num_tokens": 225517244.0, + "step": 8773 + }, + { + "epoch": 0.9635405227322644, + "grad_norm": 2.0683794021606445, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7062554955482483, + "num_tokens": 225544529.0, + "step": 8774 + }, + { + "epoch": 0.9636503404348781, + "grad_norm": 1.9643818140029907, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7020578980445862, + "num_tokens": 225574241.0, + "step": 8775 + }, + { + "epoch": 0.9637601581374917, + "grad_norm": 2.1811463832855225, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7162426710128784, + "num_tokens": 225597044.0, + "step": 8776 + }, + { + "epoch": 0.9638699758401055, + "grad_norm": 2.3621368408203125, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7140629291534424, + "num_tokens": 225619325.0, + "step": 8777 + }, + { + "epoch": 0.9639797935427191, + "grad_norm": 2.3900811672210693, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7055171728134155, + "num_tokens": 225643142.0, + "step": 8778 + }, + { + "epoch": 0.9640896112453328, + "grad_norm": 2.3692002296447754, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7181592583656311, + "num_tokens": 225664439.0, + "step": 8779 + }, + { + "epoch": 0.9641994289479464, + "grad_norm": 1.9453240633010864, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6867619752883911, + "num_tokens": 225696714.0, + "step": 8780 + }, + { + "epoch": 0.9643092466505601, + "grad_norm": 1.8770043849945068, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.71183180809021, + "num_tokens": 225727706.0, + "step": 8781 + }, + { + "epoch": 0.9644190643531737, + "grad_norm": 2.416036605834961, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7059555649757385, + "num_tokens": 225749906.0, + "step": 8782 + }, + { + "epoch": 0.9645288820557874, + "grad_norm": 2.574314594268799, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7002600431442261, + "num_tokens": 225771347.0, + "step": 8783 + }, + { + "epoch": 0.9646386997584011, + "grad_norm": 2.4139139652252197, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7085614204406738, + "num_tokens": 225794472.0, + "step": 8784 + }, + { + "epoch": 0.9647485174610148, + "grad_norm": 2.2290921211242676, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7012849450111389, + "num_tokens": 225819020.0, + "step": 8785 + }, + { + "epoch": 0.9648583351636284, + "grad_norm": 2.132413625717163, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.702542781829834, + "num_tokens": 225846764.0, + "step": 8786 + }, + { + "epoch": 0.964968152866242, + "grad_norm": 2.3132212162017822, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7040177583694458, + "num_tokens": 225870411.0, + "step": 8787 + }, + { + "epoch": 0.9650779705688557, + "grad_norm": 2.2763922214508057, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7151355743408203, + "num_tokens": 225892325.0, + "step": 8788 + }, + { + "epoch": 0.9651877882714693, + "grad_norm": 2.1577413082122803, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7153310775756836, + "num_tokens": 225917034.0, + "step": 8789 + }, + { + "epoch": 0.965297605974083, + "grad_norm": 2.279045581817627, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7126812934875488, + "num_tokens": 225940983.0, + "step": 8790 + }, + { + "epoch": 0.9654074236766966, + "grad_norm": 2.4293053150177, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7079824209213257, + "num_tokens": 225962629.0, + "step": 8791 + }, + { + "epoch": 0.9655172413793104, + "grad_norm": 2.5125975608825684, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7477395534515381, + "num_tokens": 225981774.0, + "step": 8792 + }, + { + "epoch": 0.965627059081924, + "grad_norm": 2.3506009578704834, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7095745801925659, + "num_tokens": 226004589.0, + "step": 8793 + }, + { + "epoch": 0.9657368767845377, + "grad_norm": 2.081641674041748, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7136584520339966, + "num_tokens": 226033464.0, + "step": 8794 + }, + { + "epoch": 0.9658466944871513, + "grad_norm": 2.0564212799072266, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7106207609176636, + "num_tokens": 226061266.0, + "step": 8795 + }, + { + "epoch": 0.965956512189765, + "grad_norm": 2.186516523361206, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7035492658615112, + "num_tokens": 226088102.0, + "step": 8796 + }, + { + "epoch": 0.9660663298923786, + "grad_norm": 2.31455397605896, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7350335121154785, + "num_tokens": 226109240.0, + "step": 8797 + }, + { + "epoch": 0.9661761475949923, + "grad_norm": 2.2575747966766357, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7186664938926697, + "num_tokens": 226131604.0, + "step": 8798 + }, + { + "epoch": 0.966285965297606, + "grad_norm": 2.0854363441467285, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7046156525611877, + "num_tokens": 226158800.0, + "step": 8799 + }, + { + "epoch": 0.9663957830002197, + "grad_norm": 2.1046392917633057, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7311431765556335, + "num_tokens": 226183160.0, + "step": 8800 + }, + { + "epoch": 0.9665056007028333, + "grad_norm": 2.371502637863159, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7062685489654541, + "num_tokens": 226206510.0, + "step": 8801 + }, + { + "epoch": 0.966615418405447, + "grad_norm": 2.1699023246765137, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7113661766052246, + "num_tokens": 226230821.0, + "step": 8802 + }, + { + "epoch": 0.9667252361080606, + "grad_norm": 2.3698155879974365, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7245237827301025, + "num_tokens": 226252300.0, + "step": 8803 + }, + { + "epoch": 0.9668350538106742, + "grad_norm": 2.511309862136841, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7211892604827881, + "num_tokens": 226271539.0, + "step": 8804 + }, + { + "epoch": 0.9669448715132879, + "grad_norm": 2.2636592388153076, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7295023798942566, + "num_tokens": 226293432.0, + "step": 8805 + }, + { + "epoch": 0.9670546892159017, + "grad_norm": 2.1970012187957764, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7332468032836914, + "num_tokens": 226316519.0, + "step": 8806 + }, + { + "epoch": 0.9671645069185153, + "grad_norm": 2.1059820652008057, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7066317796707153, + "num_tokens": 226343458.0, + "step": 8807 + }, + { + "epoch": 0.967274324621129, + "grad_norm": 2.0748443603515625, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6902577877044678, + "num_tokens": 226373754.0, + "step": 8808 + }, + { + "epoch": 0.9673841423237426, + "grad_norm": 2.297562599182129, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7119045853614807, + "num_tokens": 226397217.0, + "step": 8809 + }, + { + "epoch": 0.9674939600263562, + "grad_norm": 2.1922566890716553, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7151303291320801, + "num_tokens": 226420666.0, + "step": 8810 + }, + { + "epoch": 0.9676037777289699, + "grad_norm": 1.9980863332748413, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7296408414840698, + "num_tokens": 226447944.0, + "step": 8811 + }, + { + "epoch": 0.9677135954315835, + "grad_norm": 2.519613742828369, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7186951637268066, + "num_tokens": 226468180.0, + "step": 8812 + }, + { + "epoch": 0.9678234131341973, + "grad_norm": 2.1093363761901855, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7154414057731628, + "num_tokens": 226496361.0, + "step": 8813 + }, + { + "epoch": 0.9679332308368109, + "grad_norm": 2.1627988815307617, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.729182243347168, + "num_tokens": 226522887.0, + "step": 8814 + }, + { + "epoch": 0.9680430485394246, + "grad_norm": 2.2615015506744385, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7123925685882568, + "num_tokens": 226548623.0, + "step": 8815 + }, + { + "epoch": 0.9681528662420382, + "grad_norm": 2.201169490814209, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7258764505386353, + "num_tokens": 226571650.0, + "step": 8816 + }, + { + "epoch": 0.9682626839446519, + "grad_norm": 2.243553876876831, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7044804096221924, + "num_tokens": 226595043.0, + "step": 8817 + }, + { + "epoch": 0.9683725016472655, + "grad_norm": 2.117596387863159, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7224343419075012, + "num_tokens": 226622439.0, + "step": 8818 + }, + { + "epoch": 0.9684823193498792, + "grad_norm": 2.3191215991973877, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6895425319671631, + "num_tokens": 226647020.0, + "step": 8819 + }, + { + "epoch": 0.9685921370524928, + "grad_norm": 1.994719386100769, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7171586751937866, + "num_tokens": 226674447.0, + "step": 8820 + }, + { + "epoch": 0.9687019547551066, + "grad_norm": 2.221341609954834, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6869231462478638, + "num_tokens": 226698513.0, + "step": 8821 + }, + { + "epoch": 0.9688117724577202, + "grad_norm": 2.13385009765625, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7169526219367981, + "num_tokens": 226722406.0, + "step": 8822 + }, + { + "epoch": 0.9689215901603339, + "grad_norm": 2.2564501762390137, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7315911054611206, + "num_tokens": 226745172.0, + "step": 8823 + }, + { + "epoch": 0.9690314078629475, + "grad_norm": 2.4515018463134766, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7193732857704163, + "num_tokens": 226766015.0, + "step": 8824 + }, + { + "epoch": 0.9691412255655611, + "grad_norm": 2.2132785320281982, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6942481994628906, + "num_tokens": 226791377.0, + "step": 8825 + }, + { + "epoch": 0.9692510432681748, + "grad_norm": 2.115570306777954, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6834572553634644, + "num_tokens": 226819056.0, + "step": 8826 + }, + { + "epoch": 0.9693608609707884, + "grad_norm": 2.2606966495513916, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7153805494308472, + "num_tokens": 226843510.0, + "step": 8827 + }, + { + "epoch": 0.9694706786734022, + "grad_norm": 2.089735269546509, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7445499300956726, + "num_tokens": 226868719.0, + "step": 8828 + }, + { + "epoch": 0.9695804963760158, + "grad_norm": 2.020608425140381, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.725853443145752, + "num_tokens": 226897329.0, + "step": 8829 + }, + { + "epoch": 0.9696903140786295, + "grad_norm": 2.1073129177093506, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7106968760490417, + "num_tokens": 226924111.0, + "step": 8830 + }, + { + "epoch": 0.9698001317812431, + "grad_norm": 2.127187728881836, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6952915191650391, + "num_tokens": 226952340.0, + "step": 8831 + }, + { + "epoch": 0.9699099494838568, + "grad_norm": 2.05820894241333, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6938983798027039, + "num_tokens": 226983616.0, + "step": 8832 + }, + { + "epoch": 0.9700197671864704, + "grad_norm": 2.437279224395752, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7148522138595581, + "num_tokens": 227005531.0, + "step": 8833 + }, + { + "epoch": 0.9701295848890841, + "grad_norm": 2.1417148113250732, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7171095609664917, + "num_tokens": 227030037.0, + "step": 8834 + }, + { + "epoch": 0.9702394025916978, + "grad_norm": 2.309835910797119, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7338615655899048, + "num_tokens": 227052415.0, + "step": 8835 + }, + { + "epoch": 0.9703492202943115, + "grad_norm": 2.0662410259246826, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.7034293413162231, + "num_tokens": 227081144.0, + "step": 8836 + }, + { + "epoch": 0.9704590379969251, + "grad_norm": 1.9292513132095337, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.7045825719833374, + "num_tokens": 227112215.0, + "step": 8837 + }, + { + "epoch": 0.9705688556995388, + "grad_norm": 2.071946382522583, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7253015041351318, + "num_tokens": 227138788.0, + "step": 8838 + }, + { + "epoch": 0.9706786734021524, + "grad_norm": 2.3142402172088623, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6912336349487305, + "num_tokens": 227162569.0, + "step": 8839 + }, + { + "epoch": 0.9707884911047661, + "grad_norm": 2.0574281215667725, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7082844376564026, + "num_tokens": 227189877.0, + "step": 8840 + }, + { + "epoch": 0.9708983088073797, + "grad_norm": 2.1746480464935303, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7113019227981567, + "num_tokens": 227215461.0, + "step": 8841 + }, + { + "epoch": 0.9710081265099935, + "grad_norm": 2.156660556793213, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7024720907211304, + "num_tokens": 227240434.0, + "step": 8842 + }, + { + "epoch": 0.9711179442126071, + "grad_norm": 1.8984779119491577, + "learning_rate": 1e-06, + "loss": 1.0765, + "mean_token_accuracy": 0.6886839866638184, + "num_tokens": 227275455.0, + "step": 8843 + }, + { + "epoch": 0.9712277619152208, + "grad_norm": 1.8690714836120605, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6932780146598816, + "num_tokens": 227309388.0, + "step": 8844 + }, + { + "epoch": 0.9713375796178344, + "grad_norm": 2.2487144470214844, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7110298871994019, + "num_tokens": 227332894.0, + "step": 8845 + }, + { + "epoch": 0.971447397320448, + "grad_norm": 2.2649898529052734, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.7037697434425354, + "num_tokens": 227357108.0, + "step": 8846 + }, + { + "epoch": 0.9715572150230617, + "grad_norm": 2.0724539756774902, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6941293478012085, + "num_tokens": 227384279.0, + "step": 8847 + }, + { + "epoch": 0.9716670327256753, + "grad_norm": 2.3946750164031982, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7144900560379028, + "num_tokens": 227405700.0, + "step": 8848 + }, + { + "epoch": 0.971776850428289, + "grad_norm": 1.931361198425293, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7218952178955078, + "num_tokens": 227434895.0, + "step": 8849 + }, + { + "epoch": 0.9718866681309027, + "grad_norm": 2.4084465503692627, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.720295786857605, + "num_tokens": 227454693.0, + "step": 8850 + }, + { + "epoch": 0.9719964858335164, + "grad_norm": 1.9876337051391602, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7050129175186157, + "num_tokens": 227487033.0, + "step": 8851 + }, + { + "epoch": 0.97210630353613, + "grad_norm": 2.0344901084899902, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.6949734687805176, + "num_tokens": 227512589.0, + "step": 8852 + }, + { + "epoch": 0.9722161212387437, + "grad_norm": 1.7590094804763794, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.679818868637085, + "num_tokens": 227548865.0, + "step": 8853 + }, + { + "epoch": 0.9723259389413573, + "grad_norm": 2.3854868412017822, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7176427245140076, + "num_tokens": 227569712.0, + "step": 8854 + }, + { + "epoch": 0.972435756643971, + "grad_norm": 2.2077784538269043, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7428943514823914, + "num_tokens": 227594648.0, + "step": 8855 + }, + { + "epoch": 0.9725455743465846, + "grad_norm": 2.163344144821167, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7365550994873047, + "num_tokens": 227618587.0, + "step": 8856 + }, + { + "epoch": 0.9726553920491984, + "grad_norm": 1.838360071182251, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6892881393432617, + "num_tokens": 227653082.0, + "step": 8857 + }, + { + "epoch": 0.972765209751812, + "grad_norm": 2.1821212768554688, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7073614597320557, + "num_tokens": 227679250.0, + "step": 8858 + }, + { + "epoch": 0.9728750274544257, + "grad_norm": 2.256882667541504, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7004525661468506, + "num_tokens": 227703498.0, + "step": 8859 + }, + { + "epoch": 0.9729848451570393, + "grad_norm": 2.131629467010498, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.707371175289154, + "num_tokens": 227730023.0, + "step": 8860 + }, + { + "epoch": 0.973094662859653, + "grad_norm": 2.0999724864959717, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.721971869468689, + "num_tokens": 227755531.0, + "step": 8861 + }, + { + "epoch": 0.9732044805622666, + "grad_norm": 2.126879930496216, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6785938143730164, + "num_tokens": 227783737.0, + "step": 8862 + }, + { + "epoch": 0.9733142982648803, + "grad_norm": 2.4375360012054443, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7295548319816589, + "num_tokens": 227803738.0, + "step": 8863 + }, + { + "epoch": 0.973424115967494, + "grad_norm": 2.4160871505737305, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7008495330810547, + "num_tokens": 227826835.0, + "step": 8864 + }, + { + "epoch": 0.9735339336701077, + "grad_norm": 2.1208391189575195, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7005413770675659, + "num_tokens": 227853700.0, + "step": 8865 + }, + { + "epoch": 0.9736437513727213, + "grad_norm": 2.100595235824585, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7045048475265503, + "num_tokens": 227879515.0, + "step": 8866 + }, + { + "epoch": 0.973753569075335, + "grad_norm": 2.127392530441284, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6964668035507202, + "num_tokens": 227907458.0, + "step": 8867 + }, + { + "epoch": 0.9738633867779486, + "grad_norm": 2.1664891242980957, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6855646371841431, + "num_tokens": 227933597.0, + "step": 8868 + }, + { + "epoch": 0.9739732044805622, + "grad_norm": 2.255875587463379, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7132650017738342, + "num_tokens": 227957936.0, + "step": 8869 + }, + { + "epoch": 0.9740830221831759, + "grad_norm": 2.3594233989715576, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7060868144035339, + "num_tokens": 227978528.0, + "step": 8870 + }, + { + "epoch": 0.9741928398857896, + "grad_norm": 2.308558225631714, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7070879936218262, + "num_tokens": 228002389.0, + "step": 8871 + }, + { + "epoch": 0.9743026575884033, + "grad_norm": 2.2674965858459473, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7369765043258667, + "num_tokens": 228025641.0, + "step": 8872 + }, + { + "epoch": 0.9744124752910169, + "grad_norm": 2.140181064605713, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6926140189170837, + "num_tokens": 228052472.0, + "step": 8873 + }, + { + "epoch": 0.9745222929936306, + "grad_norm": 2.24424147605896, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7185289859771729, + "num_tokens": 228075488.0, + "step": 8874 + }, + { + "epoch": 0.9746321106962442, + "grad_norm": 2.4105947017669678, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7334549427032471, + "num_tokens": 228096709.0, + "step": 8875 + }, + { + "epoch": 0.9747419283988579, + "grad_norm": 2.1802029609680176, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6979613304138184, + "num_tokens": 228122343.0, + "step": 8876 + }, + { + "epoch": 0.9748517461014715, + "grad_norm": 2.0545995235443115, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6886266469955444, + "num_tokens": 228150291.0, + "step": 8877 + }, + { + "epoch": 0.9749615638040853, + "grad_norm": 2.3498172760009766, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6989759206771851, + "num_tokens": 228172461.0, + "step": 8878 + }, + { + "epoch": 0.9750713815066989, + "grad_norm": 2.4752719402313232, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7133132815361023, + "num_tokens": 228192652.0, + "step": 8879 + }, + { + "epoch": 0.9751811992093126, + "grad_norm": 2.329106092453003, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7117923498153687, + "num_tokens": 228214242.0, + "step": 8880 + }, + { + "epoch": 0.9752910169119262, + "grad_norm": 2.023017406463623, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7146410942077637, + "num_tokens": 228242159.0, + "step": 8881 + }, + { + "epoch": 0.9754008346145399, + "grad_norm": 2.3626203536987305, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7037811875343323, + "num_tokens": 228265922.0, + "step": 8882 + }, + { + "epoch": 0.9755106523171535, + "grad_norm": 2.119199514389038, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7101318836212158, + "num_tokens": 228292908.0, + "step": 8883 + }, + { + "epoch": 0.9756204700197671, + "grad_norm": 2.4570231437683105, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7079702615737915, + "num_tokens": 228314223.0, + "step": 8884 + }, + { + "epoch": 0.9757302877223808, + "grad_norm": 2.3209807872772217, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7167519330978394, + "num_tokens": 228338402.0, + "step": 8885 + }, + { + "epoch": 0.9758401054249946, + "grad_norm": 2.19469952583313, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6938800811767578, + "num_tokens": 228362952.0, + "step": 8886 + }, + { + "epoch": 0.9759499231276082, + "grad_norm": 1.9145201444625854, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6852908134460449, + "num_tokens": 228396879.0, + "step": 8887 + }, + { + "epoch": 0.9760597408302218, + "grad_norm": 1.9901398420333862, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7157929539680481, + "num_tokens": 228427701.0, + "step": 8888 + }, + { + "epoch": 0.9761695585328355, + "grad_norm": 1.9578571319580078, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6977324485778809, + "num_tokens": 228458570.0, + "step": 8889 + }, + { + "epoch": 0.9762793762354491, + "grad_norm": 2.0419702529907227, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6985079646110535, + "num_tokens": 228489370.0, + "step": 8890 + }, + { + "epoch": 0.9763891939380628, + "grad_norm": 2.2494022846221924, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7271136045455933, + "num_tokens": 228512731.0, + "step": 8891 + }, + { + "epoch": 0.9764990116406764, + "grad_norm": 2.042588949203491, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7166403532028198, + "num_tokens": 228540553.0, + "step": 8892 + }, + { + "epoch": 0.9766088293432902, + "grad_norm": 2.4171440601348877, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7275876402854919, + "num_tokens": 228563165.0, + "step": 8893 + }, + { + "epoch": 0.9767186470459038, + "grad_norm": 2.0886857509613037, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7153620719909668, + "num_tokens": 228589510.0, + "step": 8894 + }, + { + "epoch": 0.9768284647485175, + "grad_norm": 2.2911536693573, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7298866510391235, + "num_tokens": 228611873.0, + "step": 8895 + }, + { + "epoch": 0.9769382824511311, + "grad_norm": 2.1163716316223145, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7181424498558044, + "num_tokens": 228637977.0, + "step": 8896 + }, + { + "epoch": 0.9770481001537448, + "grad_norm": 2.5883774757385254, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7250702381134033, + "num_tokens": 228666411.0, + "step": 8897 + }, + { + "epoch": 0.9771579178563584, + "grad_norm": 2.002070665359497, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7011548280715942, + "num_tokens": 228694496.0, + "step": 8898 + }, + { + "epoch": 0.9772677355589721, + "grad_norm": 2.122361421585083, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7006529569625854, + "num_tokens": 228721101.0, + "step": 8899 + }, + { + "epoch": 0.9773775532615858, + "grad_norm": 2.2384142875671387, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7125875353813171, + "num_tokens": 228745114.0, + "step": 8900 + }, + { + "epoch": 0.9774873709641995, + "grad_norm": 2.2710201740264893, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7107470035552979, + "num_tokens": 228768371.0, + "step": 8901 + }, + { + "epoch": 0.9775971886668131, + "grad_norm": 2.183196544647217, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7135418653488159, + "num_tokens": 228794642.0, + "step": 8902 + }, + { + "epoch": 0.9777070063694268, + "grad_norm": 2.283018112182617, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.703446626663208, + "num_tokens": 228818931.0, + "step": 8903 + }, + { + "epoch": 0.9778168240720404, + "grad_norm": 1.9152165651321411, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7179589867591858, + "num_tokens": 228849332.0, + "step": 8904 + }, + { + "epoch": 0.977926641774654, + "grad_norm": 2.419386386871338, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7173073291778564, + "num_tokens": 228871278.0, + "step": 8905 + }, + { + "epoch": 0.9780364594772677, + "grad_norm": 2.2834439277648926, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7078031897544861, + "num_tokens": 228894814.0, + "step": 8906 + }, + { + "epoch": 0.9781462771798815, + "grad_norm": 1.9253525733947754, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6808730959892273, + "num_tokens": 228927242.0, + "step": 8907 + }, + { + "epoch": 0.9782560948824951, + "grad_norm": 2.868398427963257, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7134251594543457, + "num_tokens": 228943069.0, + "step": 8908 + }, + { + "epoch": 0.9783659125851087, + "grad_norm": 2.0085506439208984, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7250694036483765, + "num_tokens": 228972059.0, + "step": 8909 + }, + { + "epoch": 0.9784757302877224, + "grad_norm": 2.2421672344207764, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.722176194190979, + "num_tokens": 228995664.0, + "step": 8910 + }, + { + "epoch": 0.978585547990336, + "grad_norm": 1.8739972114562988, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6947138905525208, + "num_tokens": 229029566.0, + "step": 8911 + }, + { + "epoch": 0.9786953656929497, + "grad_norm": 2.1106879711151123, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6846538782119751, + "num_tokens": 229057502.0, + "step": 8912 + }, + { + "epoch": 0.9788051833955633, + "grad_norm": 2.1841070652008057, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7143117189407349, + "num_tokens": 229083042.0, + "step": 8913 + }, + { + "epoch": 0.978915001098177, + "grad_norm": 2.163790464401245, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7364464998245239, + "num_tokens": 229106998.0, + "step": 8914 + }, + { + "epoch": 0.9790248188007907, + "grad_norm": 2.032402992248535, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.703689455986023, + "num_tokens": 229136104.0, + "step": 8915 + }, + { + "epoch": 0.9791346365034044, + "grad_norm": 2.2342565059661865, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6886887550354004, + "num_tokens": 229160663.0, + "step": 8916 + }, + { + "epoch": 0.979244454206018, + "grad_norm": 2.2229676246643066, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7014375925064087, + "num_tokens": 229186236.0, + "step": 8917 + }, + { + "epoch": 0.9793542719086317, + "grad_norm": 1.8382006883621216, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7330318689346313, + "num_tokens": 229219072.0, + "step": 8918 + }, + { + "epoch": 0.9794640896112453, + "grad_norm": 1.9066362380981445, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7040256261825562, + "num_tokens": 229251875.0, + "step": 8919 + }, + { + "epoch": 0.979573907313859, + "grad_norm": 2.0896801948547363, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7185509204864502, + "num_tokens": 229278088.0, + "step": 8920 + }, + { + "epoch": 0.9796837250164726, + "grad_norm": 2.2056288719177246, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7036204934120178, + "num_tokens": 229303933.0, + "step": 8921 + }, + { + "epoch": 0.9797935427190864, + "grad_norm": 2.155099391937256, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7020299434661865, + "num_tokens": 229329767.0, + "step": 8922 + }, + { + "epoch": 0.9799033604217, + "grad_norm": 2.073284387588501, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6996493339538574, + "num_tokens": 229358565.0, + "step": 8923 + }, + { + "epoch": 0.9800131781243137, + "grad_norm": 2.0254271030426025, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7222810983657837, + "num_tokens": 229385275.0, + "step": 8924 + }, + { + "epoch": 0.9801229958269273, + "grad_norm": 2.361454725265503, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7264130115509033, + "num_tokens": 229407406.0, + "step": 8925 + }, + { + "epoch": 0.980232813529541, + "grad_norm": 1.9524585008621216, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7112449407577515, + "num_tokens": 229433858.0, + "step": 8926 + }, + { + "epoch": 0.9803426312321546, + "grad_norm": 2.097130298614502, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7151464223861694, + "num_tokens": 229458858.0, + "step": 8927 + }, + { + "epoch": 0.9804524489347682, + "grad_norm": 1.9690766334533691, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7221938967704773, + "num_tokens": 229488276.0, + "step": 8928 + }, + { + "epoch": 0.980562266637382, + "grad_norm": 2.085479974746704, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6910701990127563, + "num_tokens": 229515421.0, + "step": 8929 + }, + { + "epoch": 0.9806720843399956, + "grad_norm": 2.2996654510498047, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6987549066543579, + "num_tokens": 229537679.0, + "step": 8930 + }, + { + "epoch": 0.9807819020426093, + "grad_norm": 2.3527581691741943, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6836619973182678, + "num_tokens": 229561102.0, + "step": 8931 + }, + { + "epoch": 0.9808917197452229, + "grad_norm": 2.109257698059082, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6970092058181763, + "num_tokens": 229586491.0, + "step": 8932 + }, + { + "epoch": 0.9810015374478366, + "grad_norm": 2.2351150512695312, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6999224424362183, + "num_tokens": 229611318.0, + "step": 8933 + }, + { + "epoch": 0.9811113551504502, + "grad_norm": 2.21203351020813, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6946834921836853, + "num_tokens": 229637669.0, + "step": 8934 + }, + { + "epoch": 0.9812211728530639, + "grad_norm": 2.2031991481781006, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7089862823486328, + "num_tokens": 229663174.0, + "step": 8935 + }, + { + "epoch": 0.9813309905556776, + "grad_norm": 2.265700578689575, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.696323037147522, + "num_tokens": 229687825.0, + "step": 8936 + }, + { + "epoch": 0.9814408082582913, + "grad_norm": 2.47721791267395, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7153527736663818, + "num_tokens": 229708171.0, + "step": 8937 + }, + { + "epoch": 0.9815506259609049, + "grad_norm": 2.4894044399261475, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7291920185089111, + "num_tokens": 229728257.0, + "step": 8938 + }, + { + "epoch": 0.9816604436635186, + "grad_norm": 2.230240821838379, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6952146887779236, + "num_tokens": 229753310.0, + "step": 8939 + }, + { + "epoch": 0.9817702613661322, + "grad_norm": 2.535799980163574, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7272305488586426, + "num_tokens": 229772068.0, + "step": 8940 + }, + { + "epoch": 0.9818800790687459, + "grad_norm": 2.131322145462036, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7290856838226318, + "num_tokens": 229796357.0, + "step": 8941 + }, + { + "epoch": 0.9819898967713595, + "grad_norm": 2.2777562141418457, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7321285009384155, + "num_tokens": 229819396.0, + "step": 8942 + }, + { + "epoch": 0.9820997144739732, + "grad_norm": 2.040558338165283, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6882192492485046, + "num_tokens": 229848106.0, + "step": 8943 + }, + { + "epoch": 0.9822095321765869, + "grad_norm": 2.5610501766204834, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7144293785095215, + "num_tokens": 229868407.0, + "step": 8944 + }, + { + "epoch": 0.9823193498792006, + "grad_norm": 2.0598816871643066, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7000242471694946, + "num_tokens": 229898033.0, + "step": 8945 + }, + { + "epoch": 0.9824291675818142, + "grad_norm": 2.3553316593170166, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6937046647071838, + "num_tokens": 229921820.0, + "step": 8946 + }, + { + "epoch": 0.9825389852844278, + "grad_norm": 2.4398014545440674, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7407156229019165, + "num_tokens": 229940412.0, + "step": 8947 + }, + { + "epoch": 0.9826488029870415, + "grad_norm": 2.0727360248565674, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7238748073577881, + "num_tokens": 229967061.0, + "step": 8948 + }, + { + "epoch": 0.9827586206896551, + "grad_norm": 2.171640634536743, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6865576505661011, + "num_tokens": 229993678.0, + "step": 8949 + }, + { + "epoch": 0.9828684383922688, + "grad_norm": 2.281352996826172, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6881040930747986, + "num_tokens": 230017835.0, + "step": 8950 + }, + { + "epoch": 0.9829782560948825, + "grad_norm": 1.9148836135864258, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.695218563079834, + "num_tokens": 230049150.0, + "step": 8951 + }, + { + "epoch": 0.9830880737974962, + "grad_norm": 2.1951406002044678, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.723932147026062, + "num_tokens": 230072300.0, + "step": 8952 + }, + { + "epoch": 0.9831978915001098, + "grad_norm": 2.4016919136047363, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.710210919380188, + "num_tokens": 230092034.0, + "step": 8953 + }, + { + "epoch": 0.9833077092027235, + "grad_norm": 2.0307772159576416, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7165946364402771, + "num_tokens": 230119312.0, + "step": 8954 + }, + { + "epoch": 0.9834175269053371, + "grad_norm": 2.5310773849487305, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.707338809967041, + "num_tokens": 230139790.0, + "step": 8955 + }, + { + "epoch": 0.9835273446079508, + "grad_norm": 2.026615619659424, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7059977054595947, + "num_tokens": 230167173.0, + "step": 8956 + }, + { + "epoch": 0.9836371623105644, + "grad_norm": 2.173964023590088, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6945977210998535, + "num_tokens": 230193104.0, + "step": 8957 + }, + { + "epoch": 0.9837469800131782, + "grad_norm": 2.173264265060425, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7177834510803223, + "num_tokens": 230219225.0, + "step": 8958 + }, + { + "epoch": 0.9838567977157918, + "grad_norm": 2.364900588989258, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6971913576126099, + "num_tokens": 230243879.0, + "step": 8959 + }, + { + "epoch": 0.9839666154184055, + "grad_norm": 2.1009421348571777, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7323387861251831, + "num_tokens": 230268514.0, + "step": 8960 + }, + { + "epoch": 0.9840764331210191, + "grad_norm": 2.112208366394043, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7136298418045044, + "num_tokens": 230294727.0, + "step": 8961 + }, + { + "epoch": 0.9841862508236328, + "grad_norm": 2.102602243423462, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7403408885002136, + "num_tokens": 230320409.0, + "step": 8962 + }, + { + "epoch": 0.9842960685262464, + "grad_norm": 2.4884815216064453, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7092078924179077, + "num_tokens": 230342763.0, + "step": 8963 + }, + { + "epoch": 0.98440588622886, + "grad_norm": 1.8919910192489624, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7060653567314148, + "num_tokens": 230375556.0, + "step": 8964 + }, + { + "epoch": 0.9845157039314738, + "grad_norm": 2.4179022312164307, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.707229733467102, + "num_tokens": 230397218.0, + "step": 8965 + }, + { + "epoch": 0.9846255216340875, + "grad_norm": 2.2300870418548584, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6968753337860107, + "num_tokens": 230421652.0, + "step": 8966 + }, + { + "epoch": 0.9847353393367011, + "grad_norm": 1.854223370552063, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7079887390136719, + "num_tokens": 230453443.0, + "step": 8967 + }, + { + "epoch": 0.9848451570393147, + "grad_norm": 1.9851830005645752, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7014081478118896, + "num_tokens": 230483184.0, + "step": 8968 + }, + { + "epoch": 0.9849549747419284, + "grad_norm": 1.956001877784729, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7095292806625366, + "num_tokens": 230515494.0, + "step": 8969 + }, + { + "epoch": 0.985064792444542, + "grad_norm": 2.281825542449951, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7135303020477295, + "num_tokens": 230539266.0, + "step": 8970 + }, + { + "epoch": 0.9851746101471557, + "grad_norm": 2.1272623538970947, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7288715243339539, + "num_tokens": 230566411.0, + "step": 8971 + }, + { + "epoch": 0.9852844278497693, + "grad_norm": 2.3122432231903076, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6941143274307251, + "num_tokens": 230589778.0, + "step": 8972 + }, + { + "epoch": 0.9853942455523831, + "grad_norm": 2.0374577045440674, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7186533212661743, + "num_tokens": 230619427.0, + "step": 8973 + }, + { + "epoch": 0.9855040632549967, + "grad_norm": 2.033419609069824, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6813452839851379, + "num_tokens": 230649059.0, + "step": 8974 + }, + { + "epoch": 0.9856138809576104, + "grad_norm": 2.159095525741577, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6964278221130371, + "num_tokens": 230676510.0, + "step": 8975 + }, + { + "epoch": 0.985723698660224, + "grad_norm": 2.207547903060913, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7299169898033142, + "num_tokens": 230701365.0, + "step": 8976 + }, + { + "epoch": 0.9858335163628377, + "grad_norm": 2.4326112270355225, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7170181274414062, + "num_tokens": 230720984.0, + "step": 8977 + }, + { + "epoch": 0.9859433340654513, + "grad_norm": 2.4263319969177246, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6979544162750244, + "num_tokens": 230742032.0, + "step": 8978 + }, + { + "epoch": 0.986053151768065, + "grad_norm": 2.0078721046447754, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6900933980941772, + "num_tokens": 230774547.0, + "step": 8979 + }, + { + "epoch": 0.9861629694706787, + "grad_norm": 2.619778871536255, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7173887491226196, + "num_tokens": 230792935.0, + "step": 8980 + }, + { + "epoch": 0.9862727871732924, + "grad_norm": 2.27548885345459, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7188332080841064, + "num_tokens": 230814895.0, + "step": 8981 + }, + { + "epoch": 0.986382604875906, + "grad_norm": 2.5304577350616455, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7169393301010132, + "num_tokens": 230834765.0, + "step": 8982 + }, + { + "epoch": 0.9864924225785197, + "grad_norm": 1.9697638750076294, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7222288846969604, + "num_tokens": 230865048.0, + "step": 8983 + }, + { + "epoch": 0.9866022402811333, + "grad_norm": 2.1986782550811768, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7086992263793945, + "num_tokens": 230889896.0, + "step": 8984 + }, + { + "epoch": 0.986712057983747, + "grad_norm": 2.3890717029571533, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7149151563644409, + "num_tokens": 230912157.0, + "step": 8985 + }, + { + "epoch": 0.9868218756863606, + "grad_norm": 2.390496015548706, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7070060968399048, + "num_tokens": 230934664.0, + "step": 8986 + }, + { + "epoch": 0.9869316933889744, + "grad_norm": 2.4159653186798096, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7121273279190063, + "num_tokens": 230956971.0, + "step": 8987 + }, + { + "epoch": 0.987041511091588, + "grad_norm": 2.2184712886810303, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7267875671386719, + "num_tokens": 230981919.0, + "step": 8988 + }, + { + "epoch": 0.9871513287942016, + "grad_norm": 2.3980326652526855, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6850254535675049, + "num_tokens": 231005483.0, + "step": 8989 + }, + { + "epoch": 0.9872611464968153, + "grad_norm": 2.752922773361206, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7137743234634399, + "num_tokens": 231023940.0, + "step": 8990 + }, + { + "epoch": 0.9873709641994289, + "grad_norm": 2.559060573577881, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7233331203460693, + "num_tokens": 231043351.0, + "step": 8991 + }, + { + "epoch": 0.9874807819020426, + "grad_norm": 2.3442883491516113, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7306026220321655, + "num_tokens": 231064908.0, + "step": 8992 + }, + { + "epoch": 0.9875905996046562, + "grad_norm": 2.102175712585449, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6815981268882751, + "num_tokens": 231093444.0, + "step": 8993 + }, + { + "epoch": 0.98770041730727, + "grad_norm": 1.939298152923584, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7185139656066895, + "num_tokens": 231122951.0, + "step": 8994 + }, + { + "epoch": 0.9878102350098836, + "grad_norm": 2.3551955223083496, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7485113143920898, + "num_tokens": 231142804.0, + "step": 8995 + }, + { + "epoch": 0.9879200527124973, + "grad_norm": 2.2549691200256348, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.70991051197052, + "num_tokens": 231168046.0, + "step": 8996 + }, + { + "epoch": 0.9880298704151109, + "grad_norm": 2.271484136581421, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6808990240097046, + "num_tokens": 231196584.0, + "step": 8997 + }, + { + "epoch": 0.9881396881177246, + "grad_norm": 2.1562163829803467, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.706996500492096, + "num_tokens": 231221871.0, + "step": 8998 + }, + { + "epoch": 0.9882495058203382, + "grad_norm": 2.539454698562622, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7435363531112671, + "num_tokens": 231242198.0, + "step": 8999 + }, + { + "epoch": 0.9883593235229519, + "grad_norm": 2.172985076904297, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7020642757415771, + "num_tokens": 231269215.0, + "step": 9000 + }, + { + "epoch": 0.9884691412255655, + "grad_norm": 2.129960536956787, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7129034996032715, + "num_tokens": 231295152.0, + "step": 9001 + }, + { + "epoch": 0.9885789589281793, + "grad_norm": 2.3137924671173096, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7031723260879517, + "num_tokens": 231319186.0, + "step": 9002 + }, + { + "epoch": 0.9886887766307929, + "grad_norm": 1.9601261615753174, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7023688554763794, + "num_tokens": 231350128.0, + "step": 9003 + }, + { + "epoch": 0.9887985943334066, + "grad_norm": 2.2760348320007324, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7361599206924438, + "num_tokens": 231371843.0, + "step": 9004 + }, + { + "epoch": 0.9889084120360202, + "grad_norm": 1.9888993501663208, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7379519939422607, + "num_tokens": 231399710.0, + "step": 9005 + }, + { + "epoch": 0.9890182297386338, + "grad_norm": 2.217878580093384, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6924661993980408, + "num_tokens": 231422979.0, + "step": 9006 + }, + { + "epoch": 0.9891280474412475, + "grad_norm": 2.3720035552978516, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6997886896133423, + "num_tokens": 231446754.0, + "step": 9007 + }, + { + "epoch": 0.9892378651438611, + "grad_norm": 2.411257266998291, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6943078637123108, + "num_tokens": 231471037.0, + "step": 9008 + }, + { + "epoch": 0.9893476828464749, + "grad_norm": 2.1545259952545166, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6990256309509277, + "num_tokens": 231496384.0, + "step": 9009 + }, + { + "epoch": 0.9894575005490885, + "grad_norm": 2.5613677501678467, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7323818802833557, + "num_tokens": 231514436.0, + "step": 9010 + }, + { + "epoch": 0.9895673182517022, + "grad_norm": 2.167012929916382, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7012096643447876, + "num_tokens": 231540797.0, + "step": 9011 + }, + { + "epoch": 0.9896771359543158, + "grad_norm": 2.2277493476867676, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7145180106163025, + "num_tokens": 231565645.0, + "step": 9012 + }, + { + "epoch": 0.9897869536569295, + "grad_norm": 2.189208507537842, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.724267840385437, + "num_tokens": 231592171.0, + "step": 9013 + }, + { + "epoch": 0.9898967713595431, + "grad_norm": 2.20863938331604, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7057167291641235, + "num_tokens": 231615596.0, + "step": 9014 + }, + { + "epoch": 0.9900065890621568, + "grad_norm": 2.1604163646698, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7409768104553223, + "num_tokens": 231640426.0, + "step": 9015 + }, + { + "epoch": 0.9901164067647705, + "grad_norm": 2.234656810760498, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7043284177780151, + "num_tokens": 231664742.0, + "step": 9016 + }, + { + "epoch": 0.9902262244673842, + "grad_norm": 1.9146944284439087, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.673731803894043, + "num_tokens": 231698969.0, + "step": 9017 + }, + { + "epoch": 0.9903360421699978, + "grad_norm": 2.085456371307373, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7232650518417358, + "num_tokens": 231726110.0, + "step": 9018 + }, + { + "epoch": 0.9904458598726115, + "grad_norm": 2.5418319702148438, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7147341966629028, + "num_tokens": 231745054.0, + "step": 9019 + }, + { + "epoch": 0.9905556775752251, + "grad_norm": 2.1894822120666504, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6833440065383911, + "num_tokens": 231772146.0, + "step": 9020 + }, + { + "epoch": 0.9906654952778388, + "grad_norm": 2.244154930114746, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7123911380767822, + "num_tokens": 231797017.0, + "step": 9021 + }, + { + "epoch": 0.9907753129804524, + "grad_norm": 2.1624841690063477, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7243248224258423, + "num_tokens": 231822745.0, + "step": 9022 + }, + { + "epoch": 0.9908851306830662, + "grad_norm": 2.0558013916015625, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7192322015762329, + "num_tokens": 231849148.0, + "step": 9023 + }, + { + "epoch": 0.9909949483856798, + "grad_norm": 2.380098581314087, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7091820240020752, + "num_tokens": 231872375.0, + "step": 9024 + }, + { + "epoch": 0.9911047660882935, + "grad_norm": 1.8243025541305542, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7143632769584656, + "num_tokens": 231905438.0, + "step": 9025 + }, + { + "epoch": 0.9912145837909071, + "grad_norm": 2.361837148666382, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7120040059089661, + "num_tokens": 231926714.0, + "step": 9026 + }, + { + "epoch": 0.9913244014935207, + "grad_norm": 2.6056180000305176, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7194882035255432, + "num_tokens": 231945203.0, + "step": 9027 + }, + { + "epoch": 0.9914342191961344, + "grad_norm": 2.055744171142578, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6993658542633057, + "num_tokens": 231974928.0, + "step": 9028 + }, + { + "epoch": 0.991544036898748, + "grad_norm": 2.043980598449707, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7046545147895813, + "num_tokens": 232003192.0, + "step": 9029 + }, + { + "epoch": 0.9916538546013618, + "grad_norm": 2.0051844120025635, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6916146278381348, + "num_tokens": 232034901.0, + "step": 9030 + }, + { + "epoch": 0.9917636723039754, + "grad_norm": 2.5315444469451904, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7214699983596802, + "num_tokens": 232054913.0, + "step": 9031 + }, + { + "epoch": 0.9918734900065891, + "grad_norm": 2.265329599380493, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7332305908203125, + "num_tokens": 232076382.0, + "step": 9032 + }, + { + "epoch": 0.9919833077092027, + "grad_norm": 2.0110175609588623, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7332801818847656, + "num_tokens": 232101959.0, + "step": 9033 + }, + { + "epoch": 0.9920931254118164, + "grad_norm": 2.063762664794922, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7301527261734009, + "num_tokens": 232131236.0, + "step": 9034 + }, + { + "epoch": 0.99220294311443, + "grad_norm": 2.234950304031372, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7141176462173462, + "num_tokens": 232155284.0, + "step": 9035 + }, + { + "epoch": 0.9923127608170437, + "grad_norm": 2.3444035053253174, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7100476026535034, + "num_tokens": 232177911.0, + "step": 9036 + }, + { + "epoch": 0.9924225785196573, + "grad_norm": 2.2781267166137695, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7239097356796265, + "num_tokens": 232201635.0, + "step": 9037 + }, + { + "epoch": 0.9925323962222711, + "grad_norm": 2.211259126663208, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7259733080863953, + "num_tokens": 232225606.0, + "step": 9038 + }, + { + "epoch": 0.9926422139248847, + "grad_norm": 2.249966621398926, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7253919839859009, + "num_tokens": 232248553.0, + "step": 9039 + }, + { + "epoch": 0.9927520316274984, + "grad_norm": 2.0457353591918945, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.715674877166748, + "num_tokens": 232276087.0, + "step": 9040 + }, + { + "epoch": 0.992861849330112, + "grad_norm": 2.122922658920288, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7062684297561646, + "num_tokens": 232303798.0, + "step": 9041 + }, + { + "epoch": 0.9929716670327257, + "grad_norm": 2.3191492557525635, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7012624740600586, + "num_tokens": 232326776.0, + "step": 9042 + }, + { + "epoch": 0.9930814847353393, + "grad_norm": 1.7917906045913696, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.691831648349762, + "num_tokens": 232363223.0, + "step": 9043 + }, + { + "epoch": 0.993191302437953, + "grad_norm": 2.273038864135742, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7371045351028442, + "num_tokens": 232386354.0, + "step": 9044 + }, + { + "epoch": 0.9933011201405667, + "grad_norm": 2.017071008682251, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6901752948760986, + "num_tokens": 232414158.0, + "step": 9045 + }, + { + "epoch": 0.9934109378431804, + "grad_norm": 1.9956625699996948, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7261101603507996, + "num_tokens": 232440792.0, + "step": 9046 + }, + { + "epoch": 0.993520755545794, + "grad_norm": 2.4928975105285645, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7370805144309998, + "num_tokens": 232461746.0, + "step": 9047 + }, + { + "epoch": 0.9936305732484076, + "grad_norm": 2.209188461303711, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7146007418632507, + "num_tokens": 232487583.0, + "step": 9048 + }, + { + "epoch": 0.9937403909510213, + "grad_norm": 2.241565465927124, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7212200164794922, + "num_tokens": 232512008.0, + "step": 9049 + }, + { + "epoch": 0.9938502086536349, + "grad_norm": 2.0616283416748047, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7294506430625916, + "num_tokens": 232540858.0, + "step": 9050 + }, + { + "epoch": 0.9939600263562486, + "grad_norm": 2.129361391067505, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7040942907333374, + "num_tokens": 232565256.0, + "step": 9051 + }, + { + "epoch": 0.9940698440588623, + "grad_norm": 2.2389957904815674, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7271401882171631, + "num_tokens": 232589097.0, + "step": 9052 + }, + { + "epoch": 0.994179661761476, + "grad_norm": 2.1747066974639893, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.705298900604248, + "num_tokens": 232617896.0, + "step": 9053 + }, + { + "epoch": 0.9942894794640896, + "grad_norm": 2.307391881942749, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7045212984085083, + "num_tokens": 232640345.0, + "step": 9054 + }, + { + "epoch": 0.9943992971667033, + "grad_norm": 2.3414549827575684, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.713496744632721, + "num_tokens": 232662878.0, + "step": 9055 + }, + { + "epoch": 0.9945091148693169, + "grad_norm": 2.280106544494629, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7069034576416016, + "num_tokens": 232686723.0, + "step": 9056 + }, + { + "epoch": 0.9946189325719306, + "grad_norm": 2.4491472244262695, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6981820464134216, + "num_tokens": 232706826.0, + "step": 9057 + }, + { + "epoch": 0.9947287502745442, + "grad_norm": 2.270155429840088, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7018952369689941, + "num_tokens": 232733492.0, + "step": 9058 + }, + { + "epoch": 0.994838567977158, + "grad_norm": 2.280216693878174, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7103391885757446, + "num_tokens": 232756836.0, + "step": 9059 + }, + { + "epoch": 0.9949483856797716, + "grad_norm": 2.2453396320343018, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7091052532196045, + "num_tokens": 232781811.0, + "step": 9060 + }, + { + "epoch": 0.9950582033823853, + "grad_norm": 2.0594725608825684, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7344262599945068, + "num_tokens": 232806637.0, + "step": 9061 + }, + { + "epoch": 0.9951680210849989, + "grad_norm": 1.8580299615859985, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7219902276992798, + "num_tokens": 232838675.0, + "step": 9062 + }, + { + "epoch": 0.9952778387876126, + "grad_norm": 2.544950246810913, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7055009603500366, + "num_tokens": 232858435.0, + "step": 9063 + }, + { + "epoch": 0.9953876564902262, + "grad_norm": 2.089909553527832, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7125867009162903, + "num_tokens": 232885089.0, + "step": 9064 + }, + { + "epoch": 0.9954974741928398, + "grad_norm": 2.2524259090423584, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7159137725830078, + "num_tokens": 232909151.0, + "step": 9065 + }, + { + "epoch": 0.9956072918954535, + "grad_norm": 2.075525999069214, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6933313608169556, + "num_tokens": 232936502.0, + "step": 9066 + }, + { + "epoch": 0.9957171095980673, + "grad_norm": 2.4612436294555664, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7319475412368774, + "num_tokens": 232955587.0, + "step": 9067 + }, + { + "epoch": 0.9958269273006809, + "grad_norm": 1.9223233461380005, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6927650570869446, + "num_tokens": 232987120.0, + "step": 9068 + }, + { + "epoch": 0.9959367450032945, + "grad_norm": 2.1916778087615967, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7265733480453491, + "num_tokens": 233010907.0, + "step": 9069 + }, + { + "epoch": 0.9960465627059082, + "grad_norm": 2.0247762203216553, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7167677879333496, + "num_tokens": 233039235.0, + "step": 9070 + }, + { + "epoch": 0.9961563804085218, + "grad_norm": 2.1381919384002686, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7065513134002686, + "num_tokens": 233065512.0, + "step": 9071 + }, + { + "epoch": 0.9962661981111355, + "grad_norm": 2.2501845359802246, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7174209356307983, + "num_tokens": 233088217.0, + "step": 9072 + }, + { + "epoch": 0.9963760158137491, + "grad_norm": 2.3456406593322754, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7213627696037292, + "num_tokens": 233109983.0, + "step": 9073 + }, + { + "epoch": 0.9964858335163629, + "grad_norm": 2.32059383392334, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7011330127716064, + "num_tokens": 233134239.0, + "step": 9074 + }, + { + "epoch": 0.9965956512189765, + "grad_norm": 2.2029781341552734, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7156313061714172, + "num_tokens": 233161251.0, + "step": 9075 + }, + { + "epoch": 0.9967054689215902, + "grad_norm": 2.650921106338501, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7172034382820129, + "num_tokens": 233181201.0, + "step": 9076 + }, + { + "epoch": 0.9968152866242038, + "grad_norm": 2.3623995780944824, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7171709537506104, + "num_tokens": 233202872.0, + "step": 9077 + }, + { + "epoch": 0.9969251043268175, + "grad_norm": 2.149693012237549, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.710631251335144, + "num_tokens": 233227154.0, + "step": 9078 + }, + { + "epoch": 0.9970349220294311, + "grad_norm": 2.141066312789917, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.700181245803833, + "num_tokens": 233254877.0, + "step": 9079 + }, + { + "epoch": 0.9971447397320448, + "grad_norm": 2.306586503982544, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7034776210784912, + "num_tokens": 233278777.0, + "step": 9080 + }, + { + "epoch": 0.9972545574346585, + "grad_norm": 2.2613325119018555, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.719494104385376, + "num_tokens": 233305553.0, + "step": 9081 + }, + { + "epoch": 0.9973643751372722, + "grad_norm": 2.1245133876800537, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7240322828292847, + "num_tokens": 233331869.0, + "step": 9082 + }, + { + "epoch": 0.9974741928398858, + "grad_norm": 2.076890707015991, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7116566896438599, + "num_tokens": 233358184.0, + "step": 9083 + }, + { + "epoch": 0.9975840105424995, + "grad_norm": 2.216287136077881, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.728009819984436, + "num_tokens": 233381770.0, + "step": 9084 + }, + { + "epoch": 0.9976938282451131, + "grad_norm": 2.0305349826812744, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7021548748016357, + "num_tokens": 233409356.0, + "step": 9085 + }, + { + "epoch": 0.9978036459477267, + "grad_norm": 2.0785071849823, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7006915807723999, + "num_tokens": 233435695.0, + "step": 9086 + }, + { + "epoch": 0.9979134636503404, + "grad_norm": 2.3766672611236572, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7495819330215454, + "num_tokens": 233455019.0, + "step": 9087 + }, + { + "epoch": 0.9980232813529542, + "grad_norm": 2.5010955333709717, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7283175587654114, + "num_tokens": 233474626.0, + "step": 9088 + }, + { + "epoch": 0.9981330990555678, + "grad_norm": 2.1992480754852295, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7044283151626587, + "num_tokens": 233498457.0, + "step": 9089 + }, + { + "epoch": 0.9982429167581814, + "grad_norm": 2.2747669219970703, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.723797082901001, + "num_tokens": 233520282.0, + "step": 9090 + }, + { + "epoch": 0.9983527344607951, + "grad_norm": 2.083765983581543, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7071506977081299, + "num_tokens": 233547184.0, + "step": 9091 + }, + { + "epoch": 0.9984625521634087, + "grad_norm": 2.474977731704712, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7107182741165161, + "num_tokens": 233568102.0, + "step": 9092 + }, + { + "epoch": 0.9985723698660224, + "grad_norm": 2.439506769180298, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7341068983078003, + "num_tokens": 233590036.0, + "step": 9093 + }, + { + "epoch": 0.998682187568636, + "grad_norm": 2.1139910221099854, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6843456029891968, + "num_tokens": 233618398.0, + "step": 9094 + }, + { + "epoch": 0.9987920052712497, + "grad_norm": 2.374715566635132, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7063220739364624, + "num_tokens": 233640667.0, + "step": 9095 + }, + { + "epoch": 0.9989018229738634, + "grad_norm": 2.1692371368408203, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7083674669265747, + "num_tokens": 233663055.0, + "step": 9096 + }, + { + "epoch": 0.9990116406764771, + "grad_norm": 1.7742443084716797, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.6982815265655518, + "num_tokens": 233698273.0, + "step": 9097 + }, + { + "epoch": 0.9991214583790907, + "grad_norm": 2.183241605758667, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6951037645339966, + "num_tokens": 233724592.0, + "step": 9098 + }, + { + "epoch": 0.9992312760817044, + "grad_norm": 2.1308000087738037, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6891722679138184, + "num_tokens": 233751032.0, + "step": 9099 + }, + { + "epoch": 0.999341093784318, + "grad_norm": 2.1786913871765137, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7202545404434204, + "num_tokens": 233778051.0, + "step": 9100 + }, + { + "epoch": 0.9994509114869317, + "grad_norm": 2.046961545944214, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6825011968612671, + "num_tokens": 233807625.0, + "step": 9101 + }, + { + "epoch": 0.9995607291895453, + "grad_norm": 2.3766028881073, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7019307017326355, + "num_tokens": 233829350.0, + "step": 9102 + }, + { + "epoch": 0.9996705468921591, + "grad_norm": 2.1181821823120117, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7116503715515137, + "num_tokens": 233856084.0, + "step": 9103 + }, + { + "epoch": 0.9997803645947727, + "grad_norm": 1.9278215169906616, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7075039744377136, + "num_tokens": 233884371.0, + "step": 9104 + }, + { + "epoch": 0.9998901822973864, + "grad_norm": 2.195775032043457, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7294065952301025, + "num_tokens": 233907551.0, + "step": 9105 + }, + { + "epoch": 1.0, + "grad_norm": 2.1225788593292236, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7068207263946533, + "num_tokens": 233933829.0, + "step": 9106 + }, + { + "epoch": 1.0001098177026138, + "grad_norm": 2.1990907192230225, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7063438296318054, + "num_tokens": 233957622.0, + "step": 9107 + }, + { + "epoch": 1.0002196354052273, + "grad_norm": 2.2733755111694336, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7019621133804321, + "num_tokens": 233980479.0, + "step": 9108 + }, + { + "epoch": 1.000329453107841, + "grad_norm": 2.1597728729248047, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7286179065704346, + "num_tokens": 234006441.0, + "step": 9109 + }, + { + "epoch": 1.0004392708104546, + "grad_norm": 2.1965291500091553, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7268452644348145, + "num_tokens": 234030215.0, + "step": 9110 + }, + { + "epoch": 1.0005490885130683, + "grad_norm": 2.365449905395508, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7192431092262268, + "num_tokens": 234051130.0, + "step": 9111 + }, + { + "epoch": 1.0006589062156819, + "grad_norm": 2.129545211791992, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7135910391807556, + "num_tokens": 234076779.0, + "step": 9112 + }, + { + "epoch": 1.0007687239182956, + "grad_norm": 2.176710844039917, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7205383777618408, + "num_tokens": 234102939.0, + "step": 9113 + }, + { + "epoch": 1.0008785416209094, + "grad_norm": 2.003108501434326, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7146218419075012, + "num_tokens": 234131382.0, + "step": 9114 + }, + { + "epoch": 1.000988359323523, + "grad_norm": 2.108440399169922, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.725737452507019, + "num_tokens": 234157674.0, + "step": 9115 + }, + { + "epoch": 1.0010981770261367, + "grad_norm": 1.7022725343704224, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7084708213806152, + "num_tokens": 234198495.0, + "step": 9116 + }, + { + "epoch": 1.0012079947287502, + "grad_norm": 2.1747665405273438, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6996158361434937, + "num_tokens": 234225593.0, + "step": 9117 + }, + { + "epoch": 1.001317812431364, + "grad_norm": 2.1329538822174072, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7318623065948486, + "num_tokens": 234251058.0, + "step": 9118 + }, + { + "epoch": 1.0014276301339775, + "grad_norm": 2.047548770904541, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7223608493804932, + "num_tokens": 234281206.0, + "step": 9119 + }, + { + "epoch": 1.0015374478365913, + "grad_norm": 2.273925542831421, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7285853624343872, + "num_tokens": 234305044.0, + "step": 9120 + }, + { + "epoch": 1.001647265539205, + "grad_norm": 2.088819980621338, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7358703017234802, + "num_tokens": 234333853.0, + "step": 9121 + }, + { + "epoch": 1.0017570832418186, + "grad_norm": 2.1552371978759766, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.723615288734436, + "num_tokens": 234360026.0, + "step": 9122 + }, + { + "epoch": 1.0018669009444323, + "grad_norm": 2.3940916061401367, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7020266056060791, + "num_tokens": 234383779.0, + "step": 9123 + }, + { + "epoch": 1.0019767186470458, + "grad_norm": 2.0945470333099365, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7208026647567749, + "num_tokens": 234410369.0, + "step": 9124 + }, + { + "epoch": 1.0020865363496596, + "grad_norm": 1.9445726871490479, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.6707324981689453, + "num_tokens": 234441875.0, + "step": 9125 + }, + { + "epoch": 1.0021963540522731, + "grad_norm": 2.4932548999786377, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7297163009643555, + "num_tokens": 234462619.0, + "step": 9126 + }, + { + "epoch": 1.002306171754887, + "grad_norm": 2.248932361602783, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7179644107818604, + "num_tokens": 234486406.0, + "step": 9127 + }, + { + "epoch": 1.0024159894575007, + "grad_norm": 2.305584669113159, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6910894513130188, + "num_tokens": 234510249.0, + "step": 9128 + }, + { + "epoch": 1.0025258071601142, + "grad_norm": 2.474003553390503, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7321516275405884, + "num_tokens": 234531601.0, + "step": 9129 + }, + { + "epoch": 1.002635624862728, + "grad_norm": 2.216670036315918, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7346478700637817, + "num_tokens": 234554799.0, + "step": 9130 + }, + { + "epoch": 1.0027454425653415, + "grad_norm": 2.1045165061950684, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7126911282539368, + "num_tokens": 234581450.0, + "step": 9131 + }, + { + "epoch": 1.0028552602679552, + "grad_norm": 2.064810037612915, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6770946979522705, + "num_tokens": 234611912.0, + "step": 9132 + }, + { + "epoch": 1.0029650779705688, + "grad_norm": 2.2084498405456543, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7321414351463318, + "num_tokens": 234636871.0, + "step": 9133 + }, + { + "epoch": 1.0030748956731825, + "grad_norm": 2.132035732269287, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7292395234107971, + "num_tokens": 234661722.0, + "step": 9134 + }, + { + "epoch": 1.0031847133757963, + "grad_norm": 2.2809391021728516, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7162165641784668, + "num_tokens": 234688022.0, + "step": 9135 + }, + { + "epoch": 1.0032945310784098, + "grad_norm": 1.946785807609558, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7090634107589722, + "num_tokens": 234719450.0, + "step": 9136 + }, + { + "epoch": 1.0034043487810236, + "grad_norm": 2.251269817352295, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7373725175857544, + "num_tokens": 234742637.0, + "step": 9137 + }, + { + "epoch": 1.0035141664836371, + "grad_norm": 2.2414872646331787, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7224439978599548, + "num_tokens": 234767057.0, + "step": 9138 + }, + { + "epoch": 1.0036239841862509, + "grad_norm": 2.1797966957092285, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7129583358764648, + "num_tokens": 234793278.0, + "step": 9139 + }, + { + "epoch": 1.0037338018888644, + "grad_norm": 2.4233756065368652, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7197563648223877, + "num_tokens": 234815876.0, + "step": 9140 + }, + { + "epoch": 1.0038436195914782, + "grad_norm": 2.2323389053344727, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.708075761795044, + "num_tokens": 234841099.0, + "step": 9141 + }, + { + "epoch": 1.0039534372940917, + "grad_norm": 2.0608866214752197, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6909244060516357, + "num_tokens": 234873586.0, + "step": 9142 + }, + { + "epoch": 1.0040632549967055, + "grad_norm": 1.9853384494781494, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7103270292282104, + "num_tokens": 234903569.0, + "step": 9143 + }, + { + "epoch": 1.0041730726993192, + "grad_norm": 2.784472703933716, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7322671413421631, + "num_tokens": 234922541.0, + "step": 9144 + }, + { + "epoch": 1.0042828904019327, + "grad_norm": 1.9473432302474976, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7154884338378906, + "num_tokens": 234954147.0, + "step": 9145 + }, + { + "epoch": 1.0043927081045465, + "grad_norm": 2.151885986328125, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7129101753234863, + "num_tokens": 234981403.0, + "step": 9146 + }, + { + "epoch": 1.00450252580716, + "grad_norm": 1.8275254964828491, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6934998631477356, + "num_tokens": 235017054.0, + "step": 9147 + }, + { + "epoch": 1.0046123435097738, + "grad_norm": 2.253586769104004, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7265474200248718, + "num_tokens": 235042570.0, + "step": 9148 + }, + { + "epoch": 1.0047221612123873, + "grad_norm": 2.191937208175659, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7170014381408691, + "num_tokens": 235067238.0, + "step": 9149 + }, + { + "epoch": 1.004831978915001, + "grad_norm": 2.693455934524536, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7222080230712891, + "num_tokens": 235085740.0, + "step": 9150 + }, + { + "epoch": 1.0049417966176148, + "grad_norm": 2.2508530616760254, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7080609202384949, + "num_tokens": 235112177.0, + "step": 9151 + }, + { + "epoch": 1.0050516143202284, + "grad_norm": 2.0529866218566895, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7001556158065796, + "num_tokens": 235142952.0, + "step": 9152 + }, + { + "epoch": 1.0051614320228421, + "grad_norm": 2.379739761352539, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7289268970489502, + "num_tokens": 235164308.0, + "step": 9153 + }, + { + "epoch": 1.0052712497254557, + "grad_norm": 2.045574903488159, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7097285985946655, + "num_tokens": 235191370.0, + "step": 9154 + }, + { + "epoch": 1.0053810674280694, + "grad_norm": 2.2556936740875244, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7222474813461304, + "num_tokens": 235218156.0, + "step": 9155 + }, + { + "epoch": 1.005490885130683, + "grad_norm": 2.033383846282959, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7256723046302795, + "num_tokens": 235247834.0, + "step": 9156 + }, + { + "epoch": 1.0056007028332967, + "grad_norm": 2.1034908294677734, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7242680788040161, + "num_tokens": 235276288.0, + "step": 9157 + }, + { + "epoch": 1.0057105205359105, + "grad_norm": 2.516955614089966, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7016151547431946, + "num_tokens": 235297330.0, + "step": 9158 + }, + { + "epoch": 1.005820338238524, + "grad_norm": 2.045609951019287, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6995210647583008, + "num_tokens": 235328414.0, + "step": 9159 + }, + { + "epoch": 1.0059301559411378, + "grad_norm": 2.3854408264160156, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.728613555431366, + "num_tokens": 235349846.0, + "step": 9160 + }, + { + "epoch": 1.0060399736437513, + "grad_norm": 2.127845525741577, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7204468846321106, + "num_tokens": 235378534.0, + "step": 9161 + }, + { + "epoch": 1.006149791346365, + "grad_norm": 2.6712419986724854, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7155798673629761, + "num_tokens": 235396869.0, + "step": 9162 + }, + { + "epoch": 1.0062596090489786, + "grad_norm": 2.676687717437744, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7176793813705444, + "num_tokens": 235416760.0, + "step": 9163 + }, + { + "epoch": 1.0063694267515924, + "grad_norm": 2.3114030361175537, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7108640670776367, + "num_tokens": 235439694.0, + "step": 9164 + }, + { + "epoch": 1.0064792444542061, + "grad_norm": 2.3788766860961914, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7079235315322876, + "num_tokens": 235463899.0, + "step": 9165 + }, + { + "epoch": 1.0065890621568196, + "grad_norm": 2.2415246963500977, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7171359062194824, + "num_tokens": 235489850.0, + "step": 9166 + }, + { + "epoch": 1.0066988798594334, + "grad_norm": 2.4870285987854004, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7292219996452332, + "num_tokens": 235510258.0, + "step": 9167 + }, + { + "epoch": 1.006808697562047, + "grad_norm": 2.3457417488098145, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7308486104011536, + "num_tokens": 235532657.0, + "step": 9168 + }, + { + "epoch": 1.0069185152646607, + "grad_norm": 2.0678248405456543, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7183889746665955, + "num_tokens": 235560660.0, + "step": 9169 + }, + { + "epoch": 1.0070283329672742, + "grad_norm": 2.3728373050689697, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7327854633331299, + "num_tokens": 235581272.0, + "step": 9170 + }, + { + "epoch": 1.007138150669888, + "grad_norm": 2.2238688468933105, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7239948511123657, + "num_tokens": 235606891.0, + "step": 9171 + }, + { + "epoch": 1.0072479683725017, + "grad_norm": 2.3035266399383545, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7125951051712036, + "num_tokens": 235629940.0, + "step": 9172 + }, + { + "epoch": 1.0073577860751153, + "grad_norm": 1.952118158340454, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7062348127365112, + "num_tokens": 235663509.0, + "step": 9173 + }, + { + "epoch": 1.007467603777729, + "grad_norm": 2.0707685947418213, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7120087146759033, + "num_tokens": 235693146.0, + "step": 9174 + }, + { + "epoch": 1.0075774214803426, + "grad_norm": 2.335207939147949, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7151545286178589, + "num_tokens": 235715999.0, + "step": 9175 + }, + { + "epoch": 1.0076872391829563, + "grad_norm": 2.21238374710083, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7217002511024475, + "num_tokens": 235739109.0, + "step": 9176 + }, + { + "epoch": 1.0077970568855699, + "grad_norm": 2.481391668319702, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.741084635257721, + "num_tokens": 235760626.0, + "step": 9177 + }, + { + "epoch": 1.0079068745881836, + "grad_norm": 2.2136480808258057, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.725104808807373, + "num_tokens": 235784183.0, + "step": 9178 + }, + { + "epoch": 1.0080166922907974, + "grad_norm": 1.926832675933838, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7040730714797974, + "num_tokens": 235815484.0, + "step": 9179 + }, + { + "epoch": 1.008126509993411, + "grad_norm": 2.1704134941101074, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7339900732040405, + "num_tokens": 235842556.0, + "step": 9180 + }, + { + "epoch": 1.0082363276960247, + "grad_norm": 2.149940013885498, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7376790046691895, + "num_tokens": 235867864.0, + "step": 9181 + }, + { + "epoch": 1.0083461453986382, + "grad_norm": 2.111448049545288, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7356653213500977, + "num_tokens": 235892652.0, + "step": 9182 + }, + { + "epoch": 1.008455963101252, + "grad_norm": 2.370852470397949, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7328461408615112, + "num_tokens": 235915447.0, + "step": 9183 + }, + { + "epoch": 1.0085657808038655, + "grad_norm": 2.4972684383392334, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7103906869888306, + "num_tokens": 235936476.0, + "step": 9184 + }, + { + "epoch": 1.0086755985064793, + "grad_norm": 2.294735908508301, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7074114084243774, + "num_tokens": 235963202.0, + "step": 9185 + }, + { + "epoch": 1.008785416209093, + "grad_norm": 2.0911035537719727, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7121756076812744, + "num_tokens": 235992871.0, + "step": 9186 + }, + { + "epoch": 1.0088952339117065, + "grad_norm": 2.2718374729156494, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6978400945663452, + "num_tokens": 236020399.0, + "step": 9187 + }, + { + "epoch": 1.0090050516143203, + "grad_norm": 2.2564878463745117, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6945068836212158, + "num_tokens": 236047289.0, + "step": 9188 + }, + { + "epoch": 1.0091148693169338, + "grad_norm": 2.1730804443359375, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7208899259567261, + "num_tokens": 236072215.0, + "step": 9189 + }, + { + "epoch": 1.0092246870195476, + "grad_norm": 2.605449914932251, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7444460988044739, + "num_tokens": 236089799.0, + "step": 9190 + }, + { + "epoch": 1.0093345047221611, + "grad_norm": 2.2054860591888428, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7216341495513916, + "num_tokens": 236116014.0, + "step": 9191 + }, + { + "epoch": 1.0094443224247749, + "grad_norm": 2.266463041305542, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7118843793869019, + "num_tokens": 236140577.0, + "step": 9192 + }, + { + "epoch": 1.0095541401273886, + "grad_norm": 2.5690226554870605, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7414020299911499, + "num_tokens": 236158444.0, + "step": 9193 + }, + { + "epoch": 1.0096639578300022, + "grad_norm": 2.435615062713623, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7151471376419067, + "num_tokens": 236179906.0, + "step": 9194 + }, + { + "epoch": 1.009773775532616, + "grad_norm": 1.9234257936477661, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7133294343948364, + "num_tokens": 236212451.0, + "step": 9195 + }, + { + "epoch": 1.0098835932352295, + "grad_norm": 2.478184461593628, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7289267778396606, + "num_tokens": 236233956.0, + "step": 9196 + }, + { + "epoch": 1.0099934109378432, + "grad_norm": 2.3579280376434326, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7110127210617065, + "num_tokens": 236256566.0, + "step": 9197 + }, + { + "epoch": 1.0101032286404568, + "grad_norm": 2.2187511920928955, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7189103960990906, + "num_tokens": 236283765.0, + "step": 9198 + }, + { + "epoch": 1.0102130463430705, + "grad_norm": 2.032247543334961, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.6991813778877258, + "num_tokens": 236313000.0, + "step": 9199 + }, + { + "epoch": 1.010322864045684, + "grad_norm": 2.1546835899353027, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7167664170265198, + "num_tokens": 236339946.0, + "step": 9200 + }, + { + "epoch": 1.0104326817482978, + "grad_norm": 2.094011068344116, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6890621781349182, + "num_tokens": 236367331.0, + "step": 9201 + }, + { + "epoch": 1.0105424994509116, + "grad_norm": 2.138639450073242, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7222203612327576, + "num_tokens": 236393440.0, + "step": 9202 + }, + { + "epoch": 1.010652317153525, + "grad_norm": 1.943577766418457, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7164074778556824, + "num_tokens": 236426315.0, + "step": 9203 + }, + { + "epoch": 1.0107621348561389, + "grad_norm": 2.198317050933838, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7128408551216125, + "num_tokens": 236453430.0, + "step": 9204 + }, + { + "epoch": 1.0108719525587524, + "grad_norm": 2.0846855640411377, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7425203323364258, + "num_tokens": 236480416.0, + "step": 9205 + }, + { + "epoch": 1.0109817702613662, + "grad_norm": 2.1905784606933594, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7237693071365356, + "num_tokens": 236507453.0, + "step": 9206 + }, + { + "epoch": 1.0110915879639797, + "grad_norm": 2.162382125854492, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7118146419525146, + "num_tokens": 236535707.0, + "step": 9207 + }, + { + "epoch": 1.0112014056665934, + "grad_norm": 2.3366119861602783, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6975525617599487, + "num_tokens": 236559631.0, + "step": 9208 + }, + { + "epoch": 1.0113112233692072, + "grad_norm": 1.923435091972351, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7213504314422607, + "num_tokens": 236593291.0, + "step": 9209 + }, + { + "epoch": 1.0114210410718207, + "grad_norm": 2.04905366897583, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6908825635910034, + "num_tokens": 236629527.0, + "step": 9210 + }, + { + "epoch": 1.0115308587744345, + "grad_norm": 2.2830777168273926, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7316220998764038, + "num_tokens": 236654771.0, + "step": 9211 + }, + { + "epoch": 1.011640676477048, + "grad_norm": 2.240607738494873, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7292703986167908, + "num_tokens": 236678306.0, + "step": 9212 + }, + { + "epoch": 1.0117504941796618, + "grad_norm": 2.423511266708374, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7351018190383911, + "num_tokens": 236700555.0, + "step": 9213 + }, + { + "epoch": 1.0118603118822753, + "grad_norm": 1.979823350906372, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7189687490463257, + "num_tokens": 236730016.0, + "step": 9214 + }, + { + "epoch": 1.011970129584889, + "grad_norm": 2.252788782119751, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7152162790298462, + "num_tokens": 236754647.0, + "step": 9215 + }, + { + "epoch": 1.0120799472875028, + "grad_norm": 2.2541239261627197, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7214036583900452, + "num_tokens": 236779969.0, + "step": 9216 + }, + { + "epoch": 1.0121897649901164, + "grad_norm": 2.343846082687378, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7165478467941284, + "num_tokens": 236803415.0, + "step": 9217 + }, + { + "epoch": 1.0122995826927301, + "grad_norm": 2.5176358222961426, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.726919412612915, + "num_tokens": 236822569.0, + "step": 9218 + }, + { + "epoch": 1.0124094003953437, + "grad_norm": 2.097368001937866, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7170354127883911, + "num_tokens": 236852816.0, + "step": 9219 + }, + { + "epoch": 1.0125192180979574, + "grad_norm": 2.2788920402526855, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7702437043190002, + "num_tokens": 236875201.0, + "step": 9220 + }, + { + "epoch": 1.012629035800571, + "grad_norm": 2.709488868713379, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.71574866771698, + "num_tokens": 236894905.0, + "step": 9221 + }, + { + "epoch": 1.0127388535031847, + "grad_norm": 2.6766834259033203, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.744917631149292, + "num_tokens": 236913525.0, + "step": 9222 + }, + { + "epoch": 1.0128486712057985, + "grad_norm": 2.5175094604492188, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7059295177459717, + "num_tokens": 236936336.0, + "step": 9223 + }, + { + "epoch": 1.012958488908412, + "grad_norm": 2.0447123050689697, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7019995450973511, + "num_tokens": 236968451.0, + "step": 9224 + }, + { + "epoch": 1.0130683066110258, + "grad_norm": 2.459529399871826, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.738391101360321, + "num_tokens": 236990525.0, + "step": 9225 + }, + { + "epoch": 1.0131781243136393, + "grad_norm": 2.150789499282837, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.720914900302887, + "num_tokens": 237019356.0, + "step": 9226 + }, + { + "epoch": 1.013287942016253, + "grad_norm": 2.4706618785858154, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7301239371299744, + "num_tokens": 237040155.0, + "step": 9227 + }, + { + "epoch": 1.0133977597188666, + "grad_norm": 2.582990884780884, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.722209632396698, + "num_tokens": 237061064.0, + "step": 9228 + }, + { + "epoch": 1.0135075774214803, + "grad_norm": 2.3281052112579346, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7439125776290894, + "num_tokens": 237084162.0, + "step": 9229 + }, + { + "epoch": 1.013617395124094, + "grad_norm": 2.112894058227539, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7213443517684937, + "num_tokens": 237112023.0, + "step": 9230 + }, + { + "epoch": 1.0137272128267076, + "grad_norm": 2.1944735050201416, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7242252826690674, + "num_tokens": 237137281.0, + "step": 9231 + }, + { + "epoch": 1.0138370305293214, + "grad_norm": 2.6019248962402344, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7289984226226807, + "num_tokens": 237157497.0, + "step": 9232 + }, + { + "epoch": 1.013946848231935, + "grad_norm": 2.061959981918335, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7181292772293091, + "num_tokens": 237186592.0, + "step": 9233 + }, + { + "epoch": 1.0140566659345487, + "grad_norm": 2.1505274772644043, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7037655115127563, + "num_tokens": 237215157.0, + "step": 9234 + }, + { + "epoch": 1.0141664836371622, + "grad_norm": 2.3377959728240967, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7215612530708313, + "num_tokens": 237239054.0, + "step": 9235 + }, + { + "epoch": 1.014276301339776, + "grad_norm": 2.4473490715026855, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7049969434738159, + "num_tokens": 237260444.0, + "step": 9236 + }, + { + "epoch": 1.0143861190423897, + "grad_norm": 2.2369439601898193, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7011118531227112, + "num_tokens": 237289525.0, + "step": 9237 + }, + { + "epoch": 1.0144959367450033, + "grad_norm": 2.54095458984375, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7153484225273132, + "num_tokens": 237310349.0, + "step": 9238 + }, + { + "epoch": 1.014605754447617, + "grad_norm": 2.2657470703125, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7337436676025391, + "num_tokens": 237335394.0, + "step": 9239 + }, + { + "epoch": 1.0147155721502306, + "grad_norm": 2.0398755073547363, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7200299501419067, + "num_tokens": 237365311.0, + "step": 9240 + }, + { + "epoch": 1.0148253898528443, + "grad_norm": 2.165651559829712, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7140040397644043, + "num_tokens": 237391497.0, + "step": 9241 + }, + { + "epoch": 1.0149352075554579, + "grad_norm": 2.3772711753845215, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7321075201034546, + "num_tokens": 237414019.0, + "step": 9242 + }, + { + "epoch": 1.0150450252580716, + "grad_norm": 2.2679929733276367, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7163766622543335, + "num_tokens": 237438057.0, + "step": 9243 + }, + { + "epoch": 1.0151548429606854, + "grad_norm": 2.402639865875244, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7351343631744385, + "num_tokens": 237459513.0, + "step": 9244 + }, + { + "epoch": 1.015264660663299, + "grad_norm": 1.9587221145629883, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.696790874004364, + "num_tokens": 237489692.0, + "step": 9245 + }, + { + "epoch": 1.0153744783659127, + "grad_norm": 2.037513494491577, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7177082300186157, + "num_tokens": 237521080.0, + "step": 9246 + }, + { + "epoch": 1.0154842960685262, + "grad_norm": 2.2649288177490234, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7230222225189209, + "num_tokens": 237547389.0, + "step": 9247 + }, + { + "epoch": 1.01559411377114, + "grad_norm": 2.0051937103271484, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7140375375747681, + "num_tokens": 237575771.0, + "step": 9248 + }, + { + "epoch": 1.0157039314737535, + "grad_norm": 2.125119209289551, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7377848625183105, + "num_tokens": 237601506.0, + "step": 9249 + }, + { + "epoch": 1.0158137491763672, + "grad_norm": 2.287524700164795, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7352524995803833, + "num_tokens": 237623363.0, + "step": 9250 + }, + { + "epoch": 1.015923566878981, + "grad_norm": 2.5148463249206543, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7100268602371216, + "num_tokens": 237644862.0, + "step": 9251 + }, + { + "epoch": 1.0160333845815945, + "grad_norm": 2.3376104831695557, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7088399529457092, + "num_tokens": 237668693.0, + "step": 9252 + }, + { + "epoch": 1.0161432022842083, + "grad_norm": 2.3051295280456543, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7470741271972656, + "num_tokens": 237691426.0, + "step": 9253 + }, + { + "epoch": 1.0162530199868218, + "grad_norm": 2.208299398422241, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7332020998001099, + "num_tokens": 237716409.0, + "step": 9254 + }, + { + "epoch": 1.0163628376894356, + "grad_norm": 2.415234327316284, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7117912769317627, + "num_tokens": 237737929.0, + "step": 9255 + }, + { + "epoch": 1.0164726553920491, + "grad_norm": 2.2753467559814453, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7124507427215576, + "num_tokens": 237762026.0, + "step": 9256 + }, + { + "epoch": 1.0165824730946629, + "grad_norm": 2.4196088314056396, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7368454933166504, + "num_tokens": 237782259.0, + "step": 9257 + }, + { + "epoch": 1.0166922907972764, + "grad_norm": 2.3054845333099365, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7029217481613159, + "num_tokens": 237805275.0, + "step": 9258 + }, + { + "epoch": 1.0168021084998902, + "grad_norm": 2.4116642475128174, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7316504716873169, + "num_tokens": 237828234.0, + "step": 9259 + }, + { + "epoch": 1.016911926202504, + "grad_norm": 1.9474066495895386, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6949419975280762, + "num_tokens": 237858227.0, + "step": 9260 + }, + { + "epoch": 1.0170217439051175, + "grad_norm": 2.3294737339019775, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7080371379852295, + "num_tokens": 237882347.0, + "step": 9261 + }, + { + "epoch": 1.0171315616077312, + "grad_norm": 2.155627727508545, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7045972943305969, + "num_tokens": 237910356.0, + "step": 9262 + }, + { + "epoch": 1.0172413793103448, + "grad_norm": 2.3477542400360107, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.74122554063797, + "num_tokens": 237931275.0, + "step": 9263 + }, + { + "epoch": 1.0173511970129585, + "grad_norm": 2.4169108867645264, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7499566078186035, + "num_tokens": 237951763.0, + "step": 9264 + }, + { + "epoch": 1.017461014715572, + "grad_norm": 2.3475916385650635, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7201546430587769, + "num_tokens": 237974939.0, + "step": 9265 + }, + { + "epoch": 1.0175708324181858, + "grad_norm": 2.221193790435791, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7199766039848328, + "num_tokens": 238000845.0, + "step": 9266 + }, + { + "epoch": 1.0176806501207996, + "grad_norm": 2.1220338344573975, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7291492223739624, + "num_tokens": 238029035.0, + "step": 9267 + }, + { + "epoch": 1.017790467823413, + "grad_norm": 2.373795509338379, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.697614312171936, + "num_tokens": 238054370.0, + "step": 9268 + }, + { + "epoch": 1.0179002855260268, + "grad_norm": 2.576113700866699, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7221558690071106, + "num_tokens": 238072974.0, + "step": 9269 + }, + { + "epoch": 1.0180101032286404, + "grad_norm": 2.311687707901001, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7106839418411255, + "num_tokens": 238098230.0, + "step": 9270 + }, + { + "epoch": 1.0181199209312541, + "grad_norm": 2.1708180904388428, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7129156589508057, + "num_tokens": 238125071.0, + "step": 9271 + }, + { + "epoch": 1.0182297386338677, + "grad_norm": 1.9297027587890625, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7181532382965088, + "num_tokens": 238156608.0, + "step": 9272 + }, + { + "epoch": 1.0183395563364814, + "grad_norm": 1.9714009761810303, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7464631795883179, + "num_tokens": 238183918.0, + "step": 9273 + }, + { + "epoch": 1.0184493740390952, + "grad_norm": 2.472764730453491, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.725679874420166, + "num_tokens": 238204785.0, + "step": 9274 + }, + { + "epoch": 1.0185591917417087, + "grad_norm": 2.1396403312683105, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6927993297576904, + "num_tokens": 238233320.0, + "step": 9275 + }, + { + "epoch": 1.0186690094443225, + "grad_norm": 2.333646535873413, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7103791832923889, + "num_tokens": 238258030.0, + "step": 9276 + }, + { + "epoch": 1.018778827146936, + "grad_norm": 2.7188901901245117, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7350106835365295, + "num_tokens": 238275510.0, + "step": 9277 + }, + { + "epoch": 1.0188886448495498, + "grad_norm": 2.1908538341522217, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7067028284072876, + "num_tokens": 238301437.0, + "step": 9278 + }, + { + "epoch": 1.0189984625521633, + "grad_norm": 2.087038278579712, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.711370587348938, + "num_tokens": 238329827.0, + "step": 9279 + }, + { + "epoch": 1.019108280254777, + "grad_norm": 2.3371694087982178, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7195420265197754, + "num_tokens": 238354713.0, + "step": 9280 + }, + { + "epoch": 1.0192180979573908, + "grad_norm": 2.5211122035980225, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7332844734191895, + "num_tokens": 238374335.0, + "step": 9281 + }, + { + "epoch": 1.0193279156600044, + "grad_norm": 2.506037473678589, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7235196828842163, + "num_tokens": 238395788.0, + "step": 9282 + }, + { + "epoch": 1.0194377333626181, + "grad_norm": 2.6761221885681152, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7228546142578125, + "num_tokens": 238416492.0, + "step": 9283 + }, + { + "epoch": 1.0195475510652316, + "grad_norm": 2.1937451362609863, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7306495904922485, + "num_tokens": 238442184.0, + "step": 9284 + }, + { + "epoch": 1.0196573687678454, + "grad_norm": 2.371833324432373, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7121224403381348, + "num_tokens": 238466917.0, + "step": 9285 + }, + { + "epoch": 1.019767186470459, + "grad_norm": 2.0314903259277344, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7170114517211914, + "num_tokens": 238496242.0, + "step": 9286 + }, + { + "epoch": 1.0198770041730727, + "grad_norm": 2.233105182647705, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7202762365341187, + "num_tokens": 238520838.0, + "step": 9287 + }, + { + "epoch": 1.0199868218756865, + "grad_norm": 1.8246866464614868, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7017787098884583, + "num_tokens": 238557820.0, + "step": 9288 + }, + { + "epoch": 1.0200966395783, + "grad_norm": 2.3661389350891113, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7220740914344788, + "num_tokens": 238580519.0, + "step": 9289 + }, + { + "epoch": 1.0202064572809137, + "grad_norm": 2.1960551738739014, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7345939874649048, + "num_tokens": 238605906.0, + "step": 9290 + }, + { + "epoch": 1.0203162749835273, + "grad_norm": 2.291381359100342, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.71295565366745, + "num_tokens": 238629694.0, + "step": 9291 + }, + { + "epoch": 1.020426092686141, + "grad_norm": 2.307373046875, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7476382851600647, + "num_tokens": 238651704.0, + "step": 9292 + }, + { + "epoch": 1.0205359103887546, + "grad_norm": 1.9880532026290894, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7445279955863953, + "num_tokens": 238682623.0, + "step": 9293 + }, + { + "epoch": 1.0206457280913683, + "grad_norm": 2.3101775646209717, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7069218754768372, + "num_tokens": 238709102.0, + "step": 9294 + }, + { + "epoch": 1.020755545793982, + "grad_norm": 2.4932804107666016, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7191444039344788, + "num_tokens": 238730143.0, + "step": 9295 + }, + { + "epoch": 1.0208653634965956, + "grad_norm": 2.2984137535095215, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.747710108757019, + "num_tokens": 238752307.0, + "step": 9296 + }, + { + "epoch": 1.0209751811992094, + "grad_norm": 2.2922141551971436, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7106983065605164, + "num_tokens": 238777831.0, + "step": 9297 + }, + { + "epoch": 1.021084998901823, + "grad_norm": 2.1265692710876465, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.699321985244751, + "num_tokens": 238806744.0, + "step": 9298 + }, + { + "epoch": 1.0211948166044367, + "grad_norm": 2.134694814682007, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7217681407928467, + "num_tokens": 238833742.0, + "step": 9299 + }, + { + "epoch": 1.0213046343070502, + "grad_norm": 2.5891733169555664, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7330179214477539, + "num_tokens": 238856180.0, + "step": 9300 + }, + { + "epoch": 1.021414452009664, + "grad_norm": 2.5137557983398438, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7113543748855591, + "num_tokens": 238877770.0, + "step": 9301 + }, + { + "epoch": 1.0215242697122777, + "grad_norm": 2.100403308868408, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7262757420539856, + "num_tokens": 238904137.0, + "step": 9302 + }, + { + "epoch": 1.0216340874148913, + "grad_norm": 2.259932518005371, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7311562895774841, + "num_tokens": 238927488.0, + "step": 9303 + }, + { + "epoch": 1.021743905117505, + "grad_norm": 2.460477113723755, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7194288372993469, + "num_tokens": 238949486.0, + "step": 9304 + }, + { + "epoch": 1.0218537228201185, + "grad_norm": 2.007007598876953, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7106554508209229, + "num_tokens": 238982711.0, + "step": 9305 + }, + { + "epoch": 1.0219635405227323, + "grad_norm": 2.373619318008423, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.727463960647583, + "num_tokens": 239009690.0, + "step": 9306 + }, + { + "epoch": 1.0220733582253458, + "grad_norm": 2.069986581802368, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7066112756729126, + "num_tokens": 239038703.0, + "step": 9307 + }, + { + "epoch": 1.0221831759279596, + "grad_norm": 2.156473398208618, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7191202044487, + "num_tokens": 239066099.0, + "step": 9308 + }, + { + "epoch": 1.0222929936305734, + "grad_norm": 1.9472010135650635, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7310181856155396, + "num_tokens": 239095229.0, + "step": 9309 + }, + { + "epoch": 1.022402811333187, + "grad_norm": 2.530931234359741, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7070634365081787, + "num_tokens": 239118610.0, + "step": 9310 + }, + { + "epoch": 1.0225126290358006, + "grad_norm": 2.322469711303711, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7169431447982788, + "num_tokens": 239145233.0, + "step": 9311 + }, + { + "epoch": 1.0226224467384142, + "grad_norm": 2.054225206375122, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7111647129058838, + "num_tokens": 239177783.0, + "step": 9312 + }, + { + "epoch": 1.022732264441028, + "grad_norm": 2.114572525024414, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7184762358665466, + "num_tokens": 239203562.0, + "step": 9313 + }, + { + "epoch": 1.0228420821436415, + "grad_norm": 2.1631622314453125, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7297242879867554, + "num_tokens": 239229402.0, + "step": 9314 + }, + { + "epoch": 1.0229518998462552, + "grad_norm": 2.386568307876587, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7228902578353882, + "num_tokens": 239253343.0, + "step": 9315 + }, + { + "epoch": 1.023061717548869, + "grad_norm": 2.289036750793457, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7439000606536865, + "num_tokens": 239276459.0, + "step": 9316 + }, + { + "epoch": 1.0231715352514825, + "grad_norm": 2.1723825931549072, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7140105366706848, + "num_tokens": 239304128.0, + "step": 9317 + }, + { + "epoch": 1.0232813529540963, + "grad_norm": 2.194079875946045, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.719426691532135, + "num_tokens": 239329970.0, + "step": 9318 + }, + { + "epoch": 1.0233911706567098, + "grad_norm": 2.2579071521759033, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7166986465454102, + "num_tokens": 239356833.0, + "step": 9319 + }, + { + "epoch": 1.0235009883593236, + "grad_norm": 2.0724475383758545, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7052537202835083, + "num_tokens": 239390578.0, + "step": 9320 + }, + { + "epoch": 1.023610806061937, + "grad_norm": 2.019456148147583, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7123926877975464, + "num_tokens": 239420605.0, + "step": 9321 + }, + { + "epoch": 1.0237206237645509, + "grad_norm": 2.5923242568969727, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.733028769493103, + "num_tokens": 239440524.0, + "step": 9322 + }, + { + "epoch": 1.0238304414671644, + "grad_norm": 2.198014974594116, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7155173420906067, + "num_tokens": 239467111.0, + "step": 9323 + }, + { + "epoch": 1.0239402591697782, + "grad_norm": 2.07843017578125, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6996413469314575, + "num_tokens": 239499036.0, + "step": 9324 + }, + { + "epoch": 1.024050076872392, + "grad_norm": 1.981813669204712, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7175249457359314, + "num_tokens": 239528708.0, + "step": 9325 + }, + { + "epoch": 1.0241598945750054, + "grad_norm": 2.245196580886841, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7253214120864868, + "num_tokens": 239553631.0, + "step": 9326 + }, + { + "epoch": 1.0242697122776192, + "grad_norm": 2.202749729156494, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7020610570907593, + "num_tokens": 239578418.0, + "step": 9327 + }, + { + "epoch": 1.0243795299802327, + "grad_norm": 2.3761932849884033, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7483536601066589, + "num_tokens": 239598818.0, + "step": 9328 + }, + { + "epoch": 1.0244893476828465, + "grad_norm": 2.043386459350586, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7133144736289978, + "num_tokens": 239626842.0, + "step": 9329 + }, + { + "epoch": 1.02459916538546, + "grad_norm": 2.3994140625, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7264664173126221, + "num_tokens": 239647724.0, + "step": 9330 + }, + { + "epoch": 1.0247089830880738, + "grad_norm": 2.33204984664917, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7345654368400574, + "num_tokens": 239669699.0, + "step": 9331 + }, + { + "epoch": 1.0248188007906875, + "grad_norm": 1.9694105386734009, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7099071145057678, + "num_tokens": 239699874.0, + "step": 9332 + }, + { + "epoch": 1.024928618493301, + "grad_norm": 2.233231544494629, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7462996244430542, + "num_tokens": 239723059.0, + "step": 9333 + }, + { + "epoch": 1.0250384361959148, + "grad_norm": 2.368203639984131, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7081547975540161, + "num_tokens": 239746189.0, + "step": 9334 + }, + { + "epoch": 1.0251482538985284, + "grad_norm": 1.9655249118804932, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7353863716125488, + "num_tokens": 239777181.0, + "step": 9335 + }, + { + "epoch": 1.0252580716011421, + "grad_norm": 2.4718878269195557, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.720128059387207, + "num_tokens": 239797610.0, + "step": 9336 + }, + { + "epoch": 1.0253678893037557, + "grad_norm": 1.9672540426254272, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7127459049224854, + "num_tokens": 239831829.0, + "step": 9337 + }, + { + "epoch": 1.0254777070063694, + "grad_norm": 2.3486781120300293, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.735165536403656, + "num_tokens": 239854916.0, + "step": 9338 + }, + { + "epoch": 1.0255875247089832, + "grad_norm": 2.3732101917266846, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7404550313949585, + "num_tokens": 239877741.0, + "step": 9339 + }, + { + "epoch": 1.0256973424115967, + "grad_norm": 1.9604535102844238, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6972452402114868, + "num_tokens": 239911001.0, + "step": 9340 + }, + { + "epoch": 1.0258071601142105, + "grad_norm": 2.2374660968780518, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7239810228347778, + "num_tokens": 239937008.0, + "step": 9341 + }, + { + "epoch": 1.025916977816824, + "grad_norm": 2.1465461254119873, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7096700668334961, + "num_tokens": 239966637.0, + "step": 9342 + }, + { + "epoch": 1.0260267955194378, + "grad_norm": 2.2033791542053223, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7387326955795288, + "num_tokens": 239990939.0, + "step": 9343 + }, + { + "epoch": 1.0261366132220513, + "grad_norm": 2.080411672592163, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7280392050743103, + "num_tokens": 240018093.0, + "step": 9344 + }, + { + "epoch": 1.026246430924665, + "grad_norm": 1.983664870262146, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7012895345687866, + "num_tokens": 240050735.0, + "step": 9345 + }, + { + "epoch": 1.0263562486272788, + "grad_norm": 2.3420512676239014, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7249075174331665, + "num_tokens": 240075489.0, + "step": 9346 + }, + { + "epoch": 1.0264660663298923, + "grad_norm": 1.9744575023651123, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7223275303840637, + "num_tokens": 240103896.0, + "step": 9347 + }, + { + "epoch": 1.026575884032506, + "grad_norm": 2.4912638664245605, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7316445112228394, + "num_tokens": 240124669.0, + "step": 9348 + }, + { + "epoch": 1.0266857017351196, + "grad_norm": 2.121782064437866, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7016558647155762, + "num_tokens": 240153121.0, + "step": 9349 + }, + { + "epoch": 1.0267955194377334, + "grad_norm": 2.3279244899749756, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7037563323974609, + "num_tokens": 240178422.0, + "step": 9350 + }, + { + "epoch": 1.026905337140347, + "grad_norm": 2.2175581455230713, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7376907467842102, + "num_tokens": 240203357.0, + "step": 9351 + }, + { + "epoch": 1.0270151548429607, + "grad_norm": 2.2133634090423584, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7123188972473145, + "num_tokens": 240231070.0, + "step": 9352 + }, + { + "epoch": 1.0271249725455744, + "grad_norm": 2.2661292552948, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7405034303665161, + "num_tokens": 240254796.0, + "step": 9353 + }, + { + "epoch": 1.027234790248188, + "grad_norm": 2.001753330230713, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7244247198104858, + "num_tokens": 240282412.0, + "step": 9354 + }, + { + "epoch": 1.0273446079508017, + "grad_norm": 2.2740848064422607, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7207549214363098, + "num_tokens": 240308371.0, + "step": 9355 + }, + { + "epoch": 1.0274544256534153, + "grad_norm": 2.4371776580810547, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7115002870559692, + "num_tokens": 240330728.0, + "step": 9356 + }, + { + "epoch": 1.027564243356029, + "grad_norm": 2.018037796020508, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.708397626876831, + "num_tokens": 240360020.0, + "step": 9357 + }, + { + "epoch": 1.0276740610586426, + "grad_norm": 2.5616986751556396, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.703886866569519, + "num_tokens": 240381789.0, + "step": 9358 + }, + { + "epoch": 1.0277838787612563, + "grad_norm": 2.1855251789093018, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7073569893836975, + "num_tokens": 240406876.0, + "step": 9359 + }, + { + "epoch": 1.02789369646387, + "grad_norm": 2.1448252201080322, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.716084897518158, + "num_tokens": 240434557.0, + "step": 9360 + }, + { + "epoch": 1.0280035141664836, + "grad_norm": 2.0426464080810547, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.701137900352478, + "num_tokens": 240468851.0, + "step": 9361 + }, + { + "epoch": 1.0281133318690974, + "grad_norm": 2.1329386234283447, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7266900539398193, + "num_tokens": 240496502.0, + "step": 9362 + }, + { + "epoch": 1.028223149571711, + "grad_norm": 2.1480565071105957, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.688725471496582, + "num_tokens": 240524800.0, + "step": 9363 + }, + { + "epoch": 1.0283329672743247, + "grad_norm": 2.3604326248168945, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7341132164001465, + "num_tokens": 240547396.0, + "step": 9364 + }, + { + "epoch": 1.0284427849769382, + "grad_norm": 2.1194469928741455, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7067341804504395, + "num_tokens": 240578389.0, + "step": 9365 + }, + { + "epoch": 1.028552602679552, + "grad_norm": 2.63213849067688, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7048318982124329, + "num_tokens": 240599542.0, + "step": 9366 + }, + { + "epoch": 1.0286624203821657, + "grad_norm": 2.5115227699279785, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7377727627754211, + "num_tokens": 240619849.0, + "step": 9367 + }, + { + "epoch": 1.0287722380847792, + "grad_norm": 2.387202739715576, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7374264597892761, + "num_tokens": 240644756.0, + "step": 9368 + }, + { + "epoch": 1.028882055787393, + "grad_norm": 2.2535853385925293, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7430803775787354, + "num_tokens": 240667488.0, + "step": 9369 + }, + { + "epoch": 1.0289918734900065, + "grad_norm": 2.3430333137512207, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7296351194381714, + "num_tokens": 240694402.0, + "step": 9370 + }, + { + "epoch": 1.0291016911926203, + "grad_norm": 2.179427146911621, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.685711145401001, + "num_tokens": 240725287.0, + "step": 9371 + }, + { + "epoch": 1.0292115088952338, + "grad_norm": 2.1300411224365234, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7191495299339294, + "num_tokens": 240754760.0, + "step": 9372 + }, + { + "epoch": 1.0293213265978476, + "grad_norm": 2.6044933795928955, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7328270673751831, + "num_tokens": 240776037.0, + "step": 9373 + }, + { + "epoch": 1.0294311443004613, + "grad_norm": 2.249363422393799, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7203989624977112, + "num_tokens": 240802045.0, + "step": 9374 + }, + { + "epoch": 1.0295409620030749, + "grad_norm": 2.0950684547424316, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7061320543289185, + "num_tokens": 240831644.0, + "step": 9375 + }, + { + "epoch": 1.0296507797056886, + "grad_norm": 2.0482349395751953, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7061741948127747, + "num_tokens": 240859935.0, + "step": 9376 + }, + { + "epoch": 1.0297605974083022, + "grad_norm": 2.4402098655700684, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7149456739425659, + "num_tokens": 240883419.0, + "step": 9377 + }, + { + "epoch": 1.029870415110916, + "grad_norm": 2.168729305267334, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6999095678329468, + "num_tokens": 240912534.0, + "step": 9378 + }, + { + "epoch": 1.0299802328135295, + "grad_norm": 2.608553409576416, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7332412600517273, + "num_tokens": 240932914.0, + "step": 9379 + }, + { + "epoch": 1.0300900505161432, + "grad_norm": 2.050058364868164, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.695672869682312, + "num_tokens": 240964580.0, + "step": 9380 + }, + { + "epoch": 1.030199868218757, + "grad_norm": 2.013448715209961, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7042131423950195, + "num_tokens": 240991963.0, + "step": 9381 + }, + { + "epoch": 1.0303096859213705, + "grad_norm": 2.3637099266052246, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.715354323387146, + "num_tokens": 241015144.0, + "step": 9382 + }, + { + "epoch": 1.0304195036239843, + "grad_norm": 2.0821502208709717, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6925749778747559, + "num_tokens": 241045082.0, + "step": 9383 + }, + { + "epoch": 1.0305293213265978, + "grad_norm": 2.1582698822021484, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7164678573608398, + "num_tokens": 241072193.0, + "step": 9384 + }, + { + "epoch": 1.0306391390292116, + "grad_norm": 2.2838375568389893, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7117223739624023, + "num_tokens": 241096985.0, + "step": 9385 + }, + { + "epoch": 1.030748956731825, + "grad_norm": 2.2775282859802246, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7101296186447144, + "num_tokens": 241122149.0, + "step": 9386 + }, + { + "epoch": 1.0308587744344389, + "grad_norm": 2.2759106159210205, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6853036880493164, + "num_tokens": 241148512.0, + "step": 9387 + }, + { + "epoch": 1.0309685921370524, + "grad_norm": 2.3255484104156494, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7120059728622437, + "num_tokens": 241172165.0, + "step": 9388 + }, + { + "epoch": 1.0310784098396661, + "grad_norm": 2.3589274883270264, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7104275822639465, + "num_tokens": 241197378.0, + "step": 9389 + }, + { + "epoch": 1.03118822754228, + "grad_norm": 2.1646385192871094, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7110360860824585, + "num_tokens": 241226455.0, + "step": 9390 + }, + { + "epoch": 1.0312980452448934, + "grad_norm": 2.207876205444336, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7080516219139099, + "num_tokens": 241252723.0, + "step": 9391 + }, + { + "epoch": 1.0314078629475072, + "grad_norm": 1.9171490669250488, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7212326526641846, + "num_tokens": 241286386.0, + "step": 9392 + }, + { + "epoch": 1.0315176806501207, + "grad_norm": 2.2860114574432373, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7312604784965515, + "num_tokens": 241307912.0, + "step": 9393 + }, + { + "epoch": 1.0316274983527345, + "grad_norm": 2.411231279373169, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7111184597015381, + "num_tokens": 241330135.0, + "step": 9394 + }, + { + "epoch": 1.031737316055348, + "grad_norm": 2.5121169090270996, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7247852683067322, + "num_tokens": 241352333.0, + "step": 9395 + }, + { + "epoch": 1.0318471337579618, + "grad_norm": 2.663655996322632, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7439275979995728, + "num_tokens": 241371725.0, + "step": 9396 + }, + { + "epoch": 1.0319569514605755, + "grad_norm": 2.2009589672088623, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7056115865707397, + "num_tokens": 241399853.0, + "step": 9397 + }, + { + "epoch": 1.032066769163189, + "grad_norm": 2.4176762104034424, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7201051115989685, + "num_tokens": 241422604.0, + "step": 9398 + }, + { + "epoch": 1.0321765868658028, + "grad_norm": 2.4053187370300293, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6931302547454834, + "num_tokens": 241447802.0, + "step": 9399 + }, + { + "epoch": 1.0322864045684164, + "grad_norm": 2.210198163986206, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7254093289375305, + "num_tokens": 241476185.0, + "step": 9400 + }, + { + "epoch": 1.0323962222710301, + "grad_norm": 2.442539691925049, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6966552734375, + "num_tokens": 241502036.0, + "step": 9401 + }, + { + "epoch": 1.0325060399736437, + "grad_norm": 2.2730157375335693, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7199559807777405, + "num_tokens": 241526294.0, + "step": 9402 + }, + { + "epoch": 1.0326158576762574, + "grad_norm": 2.304166316986084, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7252717614173889, + "num_tokens": 241551336.0, + "step": 9403 + }, + { + "epoch": 1.0327256753788712, + "grad_norm": 2.02362322807312, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7083854675292969, + "num_tokens": 241580927.0, + "step": 9404 + }, + { + "epoch": 1.0328354930814847, + "grad_norm": 2.271091938018799, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7314919233322144, + "num_tokens": 241606483.0, + "step": 9405 + }, + { + "epoch": 1.0329453107840985, + "grad_norm": 2.2960879802703857, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7127467393875122, + "num_tokens": 241630265.0, + "step": 9406 + }, + { + "epoch": 1.033055128486712, + "grad_norm": 2.2128307819366455, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7104042768478394, + "num_tokens": 241655317.0, + "step": 9407 + }, + { + "epoch": 1.0331649461893258, + "grad_norm": 2.0070443153381348, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7180231809616089, + "num_tokens": 241690443.0, + "step": 9408 + }, + { + "epoch": 1.0332747638919393, + "grad_norm": 2.340265989303589, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7110738158226013, + "num_tokens": 241715151.0, + "step": 9409 + }, + { + "epoch": 1.033384581594553, + "grad_norm": 2.2151763439178467, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7064002156257629, + "num_tokens": 241741533.0, + "step": 9410 + }, + { + "epoch": 1.0334943992971668, + "grad_norm": 2.327744722366333, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7220830321311951, + "num_tokens": 241765393.0, + "step": 9411 + }, + { + "epoch": 1.0336042169997803, + "grad_norm": 2.0251200199127197, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7140811085700989, + "num_tokens": 241793704.0, + "step": 9412 + }, + { + "epoch": 1.033714034702394, + "grad_norm": 2.304636240005493, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7033955454826355, + "num_tokens": 241818811.0, + "step": 9413 + }, + { + "epoch": 1.0338238524050076, + "grad_norm": 2.3076391220092773, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7156537771224976, + "num_tokens": 241842767.0, + "step": 9414 + }, + { + "epoch": 1.0339336701076214, + "grad_norm": 2.513701915740967, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7178661227226257, + "num_tokens": 241863000.0, + "step": 9415 + }, + { + "epoch": 1.034043487810235, + "grad_norm": 2.347498655319214, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7054569721221924, + "num_tokens": 241886641.0, + "step": 9416 + }, + { + "epoch": 1.0341533055128487, + "grad_norm": 2.4366934299468994, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7365454435348511, + "num_tokens": 241907142.0, + "step": 9417 + }, + { + "epoch": 1.0342631232154624, + "grad_norm": 2.316964864730835, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7194945812225342, + "num_tokens": 241930524.0, + "step": 9418 + }, + { + "epoch": 1.034372940918076, + "grad_norm": 2.434713363647461, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7255760431289673, + "num_tokens": 241955493.0, + "step": 9419 + }, + { + "epoch": 1.0344827586206897, + "grad_norm": 2.0632145404815674, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7297149300575256, + "num_tokens": 241983111.0, + "step": 9420 + }, + { + "epoch": 1.0345925763233033, + "grad_norm": 2.014008045196533, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.698410153388977, + "num_tokens": 242012711.0, + "step": 9421 + }, + { + "epoch": 1.034702394025917, + "grad_norm": 2.3255081176757812, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7181538343429565, + "num_tokens": 242037487.0, + "step": 9422 + }, + { + "epoch": 1.0348122117285306, + "grad_norm": 2.4507837295532227, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7124501466751099, + "num_tokens": 242059800.0, + "step": 9423 + }, + { + "epoch": 1.0349220294311443, + "grad_norm": 2.560330867767334, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7409793138504028, + "num_tokens": 242079539.0, + "step": 9424 + }, + { + "epoch": 1.035031847133758, + "grad_norm": 2.132908344268799, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6989314556121826, + "num_tokens": 242108310.0, + "step": 9425 + }, + { + "epoch": 1.0351416648363716, + "grad_norm": 2.111239194869995, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7263390421867371, + "num_tokens": 242134347.0, + "step": 9426 + }, + { + "epoch": 1.0352514825389854, + "grad_norm": 2.033104181289673, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6982904672622681, + "num_tokens": 242163306.0, + "step": 9427 + }, + { + "epoch": 1.035361300241599, + "grad_norm": 2.4791531562805176, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7177709341049194, + "num_tokens": 242185400.0, + "step": 9428 + }, + { + "epoch": 1.0354711179442126, + "grad_norm": 1.978713870048523, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7025064826011658, + "num_tokens": 242215808.0, + "step": 9429 + }, + { + "epoch": 1.0355809356468262, + "grad_norm": 2.2969915866851807, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7256067395210266, + "num_tokens": 242239797.0, + "step": 9430 + }, + { + "epoch": 1.03569075334944, + "grad_norm": 2.190669298171997, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7084152698516846, + "num_tokens": 242267216.0, + "step": 9431 + }, + { + "epoch": 1.0358005710520537, + "grad_norm": 2.0644278526306152, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7039228081703186, + "num_tokens": 242295183.0, + "step": 9432 + }, + { + "epoch": 1.0359103887546672, + "grad_norm": 2.16021728515625, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6958236694335938, + "num_tokens": 242323192.0, + "step": 9433 + }, + { + "epoch": 1.036020206457281, + "grad_norm": 2.1585745811462402, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.710681676864624, + "num_tokens": 242350992.0, + "step": 9434 + }, + { + "epoch": 1.0361300241598945, + "grad_norm": 2.064887046813965, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7077381014823914, + "num_tokens": 242384591.0, + "step": 9435 + }, + { + "epoch": 1.0362398418625083, + "grad_norm": 2.2767539024353027, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7095915675163269, + "num_tokens": 242410986.0, + "step": 9436 + }, + { + "epoch": 1.0363496595651218, + "grad_norm": 2.2060022354125977, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7248290181159973, + "num_tokens": 242437976.0, + "step": 9437 + }, + { + "epoch": 1.0364594772677356, + "grad_norm": 2.257124423980713, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7481534481048584, + "num_tokens": 242460299.0, + "step": 9438 + }, + { + "epoch": 1.036569294970349, + "grad_norm": 2.0975589752197266, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7156085968017578, + "num_tokens": 242489106.0, + "step": 9439 + }, + { + "epoch": 1.0366791126729629, + "grad_norm": 2.366084098815918, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7238181233406067, + "num_tokens": 242511007.0, + "step": 9440 + }, + { + "epoch": 1.0367889303755766, + "grad_norm": 2.1358482837677, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7132188081741333, + "num_tokens": 242539235.0, + "step": 9441 + }, + { + "epoch": 1.0368987480781902, + "grad_norm": 2.157453775405884, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7413052320480347, + "num_tokens": 242563425.0, + "step": 9442 + }, + { + "epoch": 1.037008565780804, + "grad_norm": 2.1125290393829346, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7027518153190613, + "num_tokens": 242594841.0, + "step": 9443 + }, + { + "epoch": 1.0371183834834174, + "grad_norm": 1.896971344947815, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.702904224395752, + "num_tokens": 242629989.0, + "step": 9444 + }, + { + "epoch": 1.0372282011860312, + "grad_norm": 2.189910411834717, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7220990657806396, + "num_tokens": 242657717.0, + "step": 9445 + }, + { + "epoch": 1.037338018888645, + "grad_norm": 2.423389196395874, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7147238254547119, + "num_tokens": 242680695.0, + "step": 9446 + }, + { + "epoch": 1.0374478365912585, + "grad_norm": 2.4841043949127197, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7311262488365173, + "num_tokens": 242701915.0, + "step": 9447 + }, + { + "epoch": 1.0375576542938723, + "grad_norm": 2.527700185775757, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7218018770217896, + "num_tokens": 242722484.0, + "step": 9448 + }, + { + "epoch": 1.0376674719964858, + "grad_norm": 2.316375255584717, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7025737762451172, + "num_tokens": 242746738.0, + "step": 9449 + }, + { + "epoch": 1.0377772896990995, + "grad_norm": 2.231611967086792, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.727016806602478, + "num_tokens": 242771395.0, + "step": 9450 + }, + { + "epoch": 1.037887107401713, + "grad_norm": 2.311704397201538, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7125240564346313, + "num_tokens": 242794559.0, + "step": 9451 + }, + { + "epoch": 1.0379969251043268, + "grad_norm": 2.3296432495117188, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7173137664794922, + "num_tokens": 242818269.0, + "step": 9452 + }, + { + "epoch": 1.0381067428069404, + "grad_norm": 2.1830596923828125, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7333962321281433, + "num_tokens": 242843931.0, + "step": 9453 + }, + { + "epoch": 1.0382165605095541, + "grad_norm": 2.2754499912261963, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7160784006118774, + "num_tokens": 242870314.0, + "step": 9454 + }, + { + "epoch": 1.0383263782121679, + "grad_norm": 2.5615899562835693, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7281519174575806, + "num_tokens": 242889488.0, + "step": 9455 + }, + { + "epoch": 1.0384361959147814, + "grad_norm": 2.383829355239868, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7382558584213257, + "num_tokens": 242910810.0, + "step": 9456 + }, + { + "epoch": 1.0385460136173952, + "grad_norm": 2.388746738433838, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7062692046165466, + "num_tokens": 242932923.0, + "step": 9457 + }, + { + "epoch": 1.0386558313200087, + "grad_norm": 2.3195414543151855, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7221734523773193, + "num_tokens": 242957099.0, + "step": 9458 + }, + { + "epoch": 1.0387656490226225, + "grad_norm": 2.027406930923462, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7189152240753174, + "num_tokens": 242987929.0, + "step": 9459 + }, + { + "epoch": 1.038875466725236, + "grad_norm": 2.5297696590423584, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7110995054244995, + "num_tokens": 243008415.0, + "step": 9460 + }, + { + "epoch": 1.0389852844278498, + "grad_norm": 2.4514267444610596, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7384617328643799, + "num_tokens": 243029742.0, + "step": 9461 + }, + { + "epoch": 1.0390951021304635, + "grad_norm": 2.2012603282928467, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7138398289680481, + "num_tokens": 243057120.0, + "step": 9462 + }, + { + "epoch": 1.039204919833077, + "grad_norm": 2.1875669956207275, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7116056084632874, + "num_tokens": 243082098.0, + "step": 9463 + }, + { + "epoch": 1.0393147375356908, + "grad_norm": 2.705397605895996, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7304444313049316, + "num_tokens": 243102762.0, + "step": 9464 + }, + { + "epoch": 1.0394245552383043, + "grad_norm": 2.317396402359009, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7239874601364136, + "num_tokens": 243128358.0, + "step": 9465 + }, + { + "epoch": 1.039534372940918, + "grad_norm": 2.1289572715759277, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.708735466003418, + "num_tokens": 243157450.0, + "step": 9466 + }, + { + "epoch": 1.0396441906435316, + "grad_norm": 2.287945032119751, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7255550622940063, + "num_tokens": 243182079.0, + "step": 9467 + }, + { + "epoch": 1.0397540083461454, + "grad_norm": 2.1996517181396484, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7295217514038086, + "num_tokens": 243208707.0, + "step": 9468 + }, + { + "epoch": 1.0398638260487592, + "grad_norm": 2.0523324012756348, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6959208250045776, + "num_tokens": 243238659.0, + "step": 9469 + }, + { + "epoch": 1.0399736437513727, + "grad_norm": 2.032010316848755, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7106053233146667, + "num_tokens": 243269429.0, + "step": 9470 + }, + { + "epoch": 1.0400834614539864, + "grad_norm": 2.7031595706939697, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7274028062820435, + "num_tokens": 243288723.0, + "step": 9471 + }, + { + "epoch": 1.0401932791566, + "grad_norm": 2.0839850902557373, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7036020755767822, + "num_tokens": 243318418.0, + "step": 9472 + }, + { + "epoch": 1.0403030968592137, + "grad_norm": 2.2483904361724854, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7017645835876465, + "num_tokens": 243344052.0, + "step": 9473 + }, + { + "epoch": 1.0404129145618273, + "grad_norm": 2.3267626762390137, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7340976595878601, + "num_tokens": 243366749.0, + "step": 9474 + }, + { + "epoch": 1.040522732264441, + "grad_norm": 2.376187562942505, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7266629934310913, + "num_tokens": 243389760.0, + "step": 9475 + }, + { + "epoch": 1.0406325499670548, + "grad_norm": 2.640615701675415, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7122997045516968, + "num_tokens": 243408945.0, + "step": 9476 + }, + { + "epoch": 1.0407423676696683, + "grad_norm": 2.1101620197296143, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7155083417892456, + "num_tokens": 243436467.0, + "step": 9477 + }, + { + "epoch": 1.040852185372282, + "grad_norm": 1.85551917552948, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.718327522277832, + "num_tokens": 243470728.0, + "step": 9478 + }, + { + "epoch": 1.0409620030748956, + "grad_norm": 2.2494046688079834, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7054561972618103, + "num_tokens": 243496929.0, + "step": 9479 + }, + { + "epoch": 1.0410718207775094, + "grad_norm": 2.2329113483428955, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7054389715194702, + "num_tokens": 243523157.0, + "step": 9480 + }, + { + "epoch": 1.041181638480123, + "grad_norm": 1.9980337619781494, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.726799726486206, + "num_tokens": 243553775.0, + "step": 9481 + }, + { + "epoch": 1.0412914561827367, + "grad_norm": 2.0808348655700684, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7200140953063965, + "num_tokens": 243582893.0, + "step": 9482 + }, + { + "epoch": 1.0414012738853504, + "grad_norm": 2.214904308319092, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7135103940963745, + "num_tokens": 243608386.0, + "step": 9483 + }, + { + "epoch": 1.041511091587964, + "grad_norm": 2.281050682067871, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.6995137929916382, + "num_tokens": 243632943.0, + "step": 9484 + }, + { + "epoch": 1.0416209092905777, + "grad_norm": 1.9733487367630005, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.715627133846283, + "num_tokens": 243664475.0, + "step": 9485 + }, + { + "epoch": 1.0417307269931912, + "grad_norm": 2.444584846496582, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7150291204452515, + "num_tokens": 243687433.0, + "step": 9486 + }, + { + "epoch": 1.041840544695805, + "grad_norm": 2.2393271923065186, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7454468011856079, + "num_tokens": 243710176.0, + "step": 9487 + }, + { + "epoch": 1.0419503623984185, + "grad_norm": 1.995341420173645, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7037252187728882, + "num_tokens": 243740471.0, + "step": 9488 + }, + { + "epoch": 1.0420601801010323, + "grad_norm": 2.136902332305908, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7257431745529175, + "num_tokens": 243767842.0, + "step": 9489 + }, + { + "epoch": 1.042169997803646, + "grad_norm": 2.3005964756011963, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7142254114151001, + "num_tokens": 243790936.0, + "step": 9490 + }, + { + "epoch": 1.0422798155062596, + "grad_norm": 2.304154872894287, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7191300392150879, + "num_tokens": 243816966.0, + "step": 9491 + }, + { + "epoch": 1.0423896332088733, + "grad_norm": 2.286268949508667, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7219619750976562, + "num_tokens": 243840658.0, + "step": 9492 + }, + { + "epoch": 1.0424994509114869, + "grad_norm": 2.4159951210021973, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7092556953430176, + "num_tokens": 243863994.0, + "step": 9493 + }, + { + "epoch": 1.0426092686141006, + "grad_norm": 2.2419097423553467, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7251074314117432, + "num_tokens": 243887301.0, + "step": 9494 + }, + { + "epoch": 1.0427190863167142, + "grad_norm": 1.9069368839263916, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7201488018035889, + "num_tokens": 243919086.0, + "step": 9495 + }, + { + "epoch": 1.042828904019328, + "grad_norm": 2.1176111698150635, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7316702604293823, + "num_tokens": 243945501.0, + "step": 9496 + }, + { + "epoch": 1.0429387217219417, + "grad_norm": 2.5164988040924072, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7359777688980103, + "num_tokens": 243968461.0, + "step": 9497 + }, + { + "epoch": 1.0430485394245552, + "grad_norm": 2.4277379512786865, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7372574806213379, + "num_tokens": 243989959.0, + "step": 9498 + }, + { + "epoch": 1.043158357127169, + "grad_norm": 2.0999557971954346, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7134560346603394, + "num_tokens": 244019616.0, + "step": 9499 + }, + { + "epoch": 1.0432681748297825, + "grad_norm": 2.229779005050659, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7293007373809814, + "num_tokens": 244045112.0, + "step": 9500 + }, + { + "epoch": 1.0433779925323963, + "grad_norm": 2.2617247104644775, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7055631875991821, + "num_tokens": 244069321.0, + "step": 9501 + }, + { + "epoch": 1.0434878102350098, + "grad_norm": 2.260176420211792, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7267802357673645, + "num_tokens": 244092601.0, + "step": 9502 + }, + { + "epoch": 1.0435976279376236, + "grad_norm": 2.3785107135772705, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.701449990272522, + "num_tokens": 244116241.0, + "step": 9503 + }, + { + "epoch": 1.043707445640237, + "grad_norm": 2.2181975841522217, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7337511777877808, + "num_tokens": 244140473.0, + "step": 9504 + }, + { + "epoch": 1.0438172633428509, + "grad_norm": 2.364534854888916, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7204731702804565, + "num_tokens": 244163445.0, + "step": 9505 + }, + { + "epoch": 1.0439270810454646, + "grad_norm": 2.485556125640869, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7232921123504639, + "num_tokens": 244186238.0, + "step": 9506 + }, + { + "epoch": 1.0440368987480781, + "grad_norm": 2.2527754306793213, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7198089361190796, + "num_tokens": 244211209.0, + "step": 9507 + }, + { + "epoch": 1.044146716450692, + "grad_norm": 2.420210838317871, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7198814153671265, + "num_tokens": 244234302.0, + "step": 9508 + }, + { + "epoch": 1.0442565341533054, + "grad_norm": 1.9960887432098389, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7178438305854797, + "num_tokens": 244264422.0, + "step": 9509 + }, + { + "epoch": 1.0443663518559192, + "grad_norm": 2.367997407913208, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7361547946929932, + "num_tokens": 244287210.0, + "step": 9510 + }, + { + "epoch": 1.0444761695585327, + "grad_norm": 2.2617876529693604, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7332133054733276, + "num_tokens": 244310857.0, + "step": 9511 + }, + { + "epoch": 1.0445859872611465, + "grad_norm": 2.6353275775909424, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.739764928817749, + "num_tokens": 244329609.0, + "step": 9512 + }, + { + "epoch": 1.0446958049637602, + "grad_norm": 2.216475009918213, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7279770970344543, + "num_tokens": 244354792.0, + "step": 9513 + }, + { + "epoch": 1.0448056226663738, + "grad_norm": 2.251107692718506, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7267613410949707, + "num_tokens": 244378302.0, + "step": 9514 + }, + { + "epoch": 1.0449154403689875, + "grad_norm": 2.1634058952331543, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6910465359687805, + "num_tokens": 244405119.0, + "step": 9515 + }, + { + "epoch": 1.045025258071601, + "grad_norm": 2.1662228107452393, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7353217005729675, + "num_tokens": 244429978.0, + "step": 9516 + }, + { + "epoch": 1.0451350757742148, + "grad_norm": 2.422697067260742, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7245632410049438, + "num_tokens": 244453227.0, + "step": 9517 + }, + { + "epoch": 1.0452448934768284, + "grad_norm": 2.3342020511627197, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6885600090026855, + "num_tokens": 244480497.0, + "step": 9518 + }, + { + "epoch": 1.0453547111794421, + "grad_norm": 2.3905229568481445, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7026416659355164, + "num_tokens": 244504665.0, + "step": 9519 + }, + { + "epoch": 1.0454645288820559, + "grad_norm": 2.0557937622070312, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7046317458152771, + "num_tokens": 244534429.0, + "step": 9520 + }, + { + "epoch": 1.0455743465846694, + "grad_norm": 2.1638472080230713, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7023910284042358, + "num_tokens": 244564111.0, + "step": 9521 + }, + { + "epoch": 1.0456841642872832, + "grad_norm": 2.166440725326538, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7128685712814331, + "num_tokens": 244590672.0, + "step": 9522 + }, + { + "epoch": 1.0457939819898967, + "grad_norm": 2.1137452125549316, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.698050856590271, + "num_tokens": 244619877.0, + "step": 9523 + }, + { + "epoch": 1.0459037996925105, + "grad_norm": 1.95005464553833, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7370622158050537, + "num_tokens": 244650550.0, + "step": 9524 + }, + { + "epoch": 1.046013617395124, + "grad_norm": 2.4623775482177734, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7270086407661438, + "num_tokens": 244672099.0, + "step": 9525 + }, + { + "epoch": 1.0461234350977378, + "grad_norm": 2.139293909072876, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7271116971969604, + "num_tokens": 244699055.0, + "step": 9526 + }, + { + "epoch": 1.0462332528003515, + "grad_norm": 2.3252580165863037, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7184864282608032, + "num_tokens": 244723864.0, + "step": 9527 + }, + { + "epoch": 1.046343070502965, + "grad_norm": 2.231487989425659, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7146881818771362, + "num_tokens": 244748720.0, + "step": 9528 + }, + { + "epoch": 1.0464528882055788, + "grad_norm": 2.044677972793579, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7298531532287598, + "num_tokens": 244777246.0, + "step": 9529 + }, + { + "epoch": 1.0465627059081923, + "grad_norm": 2.1018764972686768, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7128484845161438, + "num_tokens": 244805984.0, + "step": 9530 + }, + { + "epoch": 1.046672523610806, + "grad_norm": 2.0061864852905273, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7166042327880859, + "num_tokens": 244836331.0, + "step": 9531 + }, + { + "epoch": 1.0467823413134196, + "grad_norm": 2.6550471782684326, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.736793041229248, + "num_tokens": 244855269.0, + "step": 9532 + }, + { + "epoch": 1.0468921590160334, + "grad_norm": 2.4591307640075684, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7365397214889526, + "num_tokens": 244877830.0, + "step": 9533 + }, + { + "epoch": 1.0470019767186471, + "grad_norm": 2.208218812942505, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.744418740272522, + "num_tokens": 244902611.0, + "step": 9534 + }, + { + "epoch": 1.0471117944212607, + "grad_norm": 2.4432132244110107, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7293345928192139, + "num_tokens": 244924945.0, + "step": 9535 + }, + { + "epoch": 1.0472216121238744, + "grad_norm": 2.2701070308685303, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7084450721740723, + "num_tokens": 244950000.0, + "step": 9536 + }, + { + "epoch": 1.047331429826488, + "grad_norm": 2.158597946166992, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7180948257446289, + "num_tokens": 244977007.0, + "step": 9537 + }, + { + "epoch": 1.0474412475291017, + "grad_norm": 2.2856926918029785, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7214683294296265, + "num_tokens": 245004752.0, + "step": 9538 + }, + { + "epoch": 1.0475510652317153, + "grad_norm": 2.2025153636932373, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7114925980567932, + "num_tokens": 245032230.0, + "step": 9539 + }, + { + "epoch": 1.047660882934329, + "grad_norm": 2.2449533939361572, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7277084589004517, + "num_tokens": 245058934.0, + "step": 9540 + }, + { + "epoch": 1.0477707006369428, + "grad_norm": 2.4710195064544678, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7250388860702515, + "num_tokens": 245079673.0, + "step": 9541 + }, + { + "epoch": 1.0478805183395563, + "grad_norm": 2.725001573562622, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7263158559799194, + "num_tokens": 245098035.0, + "step": 9542 + }, + { + "epoch": 1.04799033604217, + "grad_norm": 2.070131540298462, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6831185817718506, + "num_tokens": 245130197.0, + "step": 9543 + }, + { + "epoch": 1.0481001537447836, + "grad_norm": 2.1929283142089844, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6903803944587708, + "num_tokens": 245158006.0, + "step": 9544 + }, + { + "epoch": 1.0482099714473974, + "grad_norm": 2.3730239868164062, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7452970743179321, + "num_tokens": 245179123.0, + "step": 9545 + }, + { + "epoch": 1.048319789150011, + "grad_norm": 2.231766700744629, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7013700008392334, + "num_tokens": 245204981.0, + "step": 9546 + }, + { + "epoch": 1.0484296068526247, + "grad_norm": 2.1754391193389893, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7042540311813354, + "num_tokens": 245233659.0, + "step": 9547 + }, + { + "epoch": 1.0485394245552384, + "grad_norm": 2.1313726902008057, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7057890892028809, + "num_tokens": 245260893.0, + "step": 9548 + }, + { + "epoch": 1.048649242257852, + "grad_norm": 2.102964401245117, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7374922037124634, + "num_tokens": 245287927.0, + "step": 9549 + }, + { + "epoch": 1.0487590599604657, + "grad_norm": 2.265463352203369, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7126907110214233, + "num_tokens": 245311010.0, + "step": 9550 + }, + { + "epoch": 1.0488688776630792, + "grad_norm": 2.3411967754364014, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7143687605857849, + "num_tokens": 245333521.0, + "step": 9551 + }, + { + "epoch": 1.048978695365693, + "grad_norm": 2.263505697250366, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7097187042236328, + "num_tokens": 245357240.0, + "step": 9552 + }, + { + "epoch": 1.0490885130683065, + "grad_norm": 2.5032432079315186, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7388737201690674, + "num_tokens": 245377589.0, + "step": 9553 + }, + { + "epoch": 1.0491983307709203, + "grad_norm": 2.2471742630004883, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7439346313476562, + "num_tokens": 245402192.0, + "step": 9554 + }, + { + "epoch": 1.0493081484735338, + "grad_norm": 2.1966660022735596, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7014771699905396, + "num_tokens": 245429955.0, + "step": 9555 + }, + { + "epoch": 1.0494179661761476, + "grad_norm": 2.3280458450317383, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.699741005897522, + "num_tokens": 245453488.0, + "step": 9556 + }, + { + "epoch": 1.0495277838787613, + "grad_norm": 1.966294527053833, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.714948296546936, + "num_tokens": 245485633.0, + "step": 9557 + }, + { + "epoch": 1.0496376015813749, + "grad_norm": 2.1283485889434814, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7000406980514526, + "num_tokens": 245513462.0, + "step": 9558 + }, + { + "epoch": 1.0497474192839886, + "grad_norm": 2.2372219562530518, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7071226835250854, + "num_tokens": 245539773.0, + "step": 9559 + }, + { + "epoch": 1.0498572369866022, + "grad_norm": 2.4425313472747803, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7164625525474548, + "num_tokens": 245561225.0, + "step": 9560 + }, + { + "epoch": 1.049967054689216, + "grad_norm": 2.4225986003875732, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7221004962921143, + "num_tokens": 245584393.0, + "step": 9561 + }, + { + "epoch": 1.0500768723918297, + "grad_norm": 1.8580048084259033, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7107841372489929, + "num_tokens": 245618780.0, + "step": 9562 + }, + { + "epoch": 1.0501866900944432, + "grad_norm": 2.417273998260498, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7291595935821533, + "num_tokens": 245640131.0, + "step": 9563 + }, + { + "epoch": 1.050296507797057, + "grad_norm": 2.4672904014587402, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7137273550033569, + "num_tokens": 245662509.0, + "step": 9564 + }, + { + "epoch": 1.0504063254996705, + "grad_norm": 2.0236313343048096, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7197902202606201, + "num_tokens": 245694934.0, + "step": 9565 + }, + { + "epoch": 1.0505161432022843, + "grad_norm": 2.0024755001068115, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7017872929573059, + "num_tokens": 245727469.0, + "step": 9566 + }, + { + "epoch": 1.0506259609048978, + "grad_norm": 2.271191358566284, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7155207395553589, + "num_tokens": 245751335.0, + "step": 9567 + }, + { + "epoch": 1.0507357786075116, + "grad_norm": 2.4845006465911865, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.733020544052124, + "num_tokens": 245773089.0, + "step": 9568 + }, + { + "epoch": 1.050845596310125, + "grad_norm": 2.374224901199341, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7092504501342773, + "num_tokens": 245796218.0, + "step": 9569 + }, + { + "epoch": 1.0509554140127388, + "grad_norm": 2.564762592315674, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7450950145721436, + "num_tokens": 245815124.0, + "step": 9570 + }, + { + "epoch": 1.0510652317153526, + "grad_norm": 2.3996593952178955, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6941462159156799, + "num_tokens": 245840586.0, + "step": 9571 + }, + { + "epoch": 1.0511750494179661, + "grad_norm": 2.3444716930389404, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7175798416137695, + "num_tokens": 245863757.0, + "step": 9572 + }, + { + "epoch": 1.05128486712058, + "grad_norm": 2.21587872505188, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7571889758110046, + "num_tokens": 245890437.0, + "step": 9573 + }, + { + "epoch": 1.0513946848231934, + "grad_norm": 2.1318359375, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7145955562591553, + "num_tokens": 245916524.0, + "step": 9574 + }, + { + "epoch": 1.0515045025258072, + "grad_norm": 2.4146993160247803, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7261455059051514, + "num_tokens": 245939269.0, + "step": 9575 + }, + { + "epoch": 1.0516143202284207, + "grad_norm": 2.1506283283233643, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7369842529296875, + "num_tokens": 245965603.0, + "step": 9576 + }, + { + "epoch": 1.0517241379310345, + "grad_norm": 2.2940831184387207, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7130599021911621, + "num_tokens": 245991441.0, + "step": 9577 + }, + { + "epoch": 1.0518339556336482, + "grad_norm": 2.1258130073547363, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7237038016319275, + "num_tokens": 246019229.0, + "step": 9578 + }, + { + "epoch": 1.0519437733362618, + "grad_norm": 2.5354228019714355, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.738596498966217, + "num_tokens": 246038091.0, + "step": 9579 + }, + { + "epoch": 1.0520535910388755, + "grad_norm": 2.3203964233398438, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7169475555419922, + "num_tokens": 246063121.0, + "step": 9580 + }, + { + "epoch": 1.052163408741489, + "grad_norm": 2.397341251373291, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7293664813041687, + "num_tokens": 246085068.0, + "step": 9581 + }, + { + "epoch": 1.0522732264441028, + "grad_norm": 2.181162118911743, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7215665578842163, + "num_tokens": 246111768.0, + "step": 9582 + }, + { + "epoch": 1.0523830441467164, + "grad_norm": 2.175595283508301, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6928626298904419, + "num_tokens": 246139087.0, + "step": 9583 + }, + { + "epoch": 1.05249286184933, + "grad_norm": 2.0729005336761475, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6968427896499634, + "num_tokens": 246169186.0, + "step": 9584 + }, + { + "epoch": 1.0526026795519439, + "grad_norm": 2.1508398056030273, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7237724661827087, + "num_tokens": 246195490.0, + "step": 9585 + }, + { + "epoch": 1.0527124972545574, + "grad_norm": 2.509451150894165, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7108659744262695, + "num_tokens": 246217764.0, + "step": 9586 + }, + { + "epoch": 1.0528223149571712, + "grad_norm": 2.430281639099121, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7222774028778076, + "num_tokens": 246240483.0, + "step": 9587 + }, + { + "epoch": 1.0529321326597847, + "grad_norm": 2.000322103500366, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7209516763687134, + "num_tokens": 246270626.0, + "step": 9588 + }, + { + "epoch": 1.0530419503623984, + "grad_norm": 2.527289390563965, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.740809440612793, + "num_tokens": 246291443.0, + "step": 9589 + }, + { + "epoch": 1.053151768065012, + "grad_norm": 1.9542236328125, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6895334720611572, + "num_tokens": 246325203.0, + "step": 9590 + }, + { + "epoch": 1.0532615857676257, + "grad_norm": 2.4718267917633057, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7245631217956543, + "num_tokens": 246348453.0, + "step": 9591 + }, + { + "epoch": 1.0533714034702395, + "grad_norm": 2.3358941078186035, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7077550292015076, + "num_tokens": 246372412.0, + "step": 9592 + }, + { + "epoch": 1.053481221172853, + "grad_norm": 2.136324167251587, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.751739501953125, + "num_tokens": 246398449.0, + "step": 9593 + }, + { + "epoch": 1.0535910388754668, + "grad_norm": 2.1370673179626465, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7015568017959595, + "num_tokens": 246426860.0, + "step": 9594 + }, + { + "epoch": 1.0537008565780803, + "grad_norm": 1.9049460887908936, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6876444220542908, + "num_tokens": 246462710.0, + "step": 9595 + }, + { + "epoch": 1.053810674280694, + "grad_norm": 2.105980157852173, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7305784225463867, + "num_tokens": 246490862.0, + "step": 9596 + }, + { + "epoch": 1.0539204919833076, + "grad_norm": 2.205522298812866, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7295519113540649, + "num_tokens": 246515549.0, + "step": 9597 + }, + { + "epoch": 1.0540303096859214, + "grad_norm": 2.2431247234344482, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6964385509490967, + "num_tokens": 246541303.0, + "step": 9598 + }, + { + "epoch": 1.0541401273885351, + "grad_norm": 1.9942082166671753, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7060424089431763, + "num_tokens": 246572115.0, + "step": 9599 + }, + { + "epoch": 1.0542499450911487, + "grad_norm": 1.768343210220337, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6859778165817261, + "num_tokens": 246609678.0, + "step": 9600 + }, + { + "epoch": 1.0543597627937624, + "grad_norm": 2.474961519241333, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7181923389434814, + "num_tokens": 246631219.0, + "step": 9601 + }, + { + "epoch": 1.054469580496376, + "grad_norm": 2.393655300140381, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7240734100341797, + "num_tokens": 246654726.0, + "step": 9602 + }, + { + "epoch": 1.0545793981989897, + "grad_norm": 2.4514036178588867, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.71812903881073, + "num_tokens": 246677652.0, + "step": 9603 + }, + { + "epoch": 1.0546892159016032, + "grad_norm": 2.2644662857055664, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7262084484100342, + "num_tokens": 246701906.0, + "step": 9604 + }, + { + "epoch": 1.054799033604217, + "grad_norm": 2.575242519378662, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.71100914478302, + "num_tokens": 246723725.0, + "step": 9605 + }, + { + "epoch": 1.0549088513068308, + "grad_norm": 2.448497772216797, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7295531034469604, + "num_tokens": 246745502.0, + "step": 9606 + }, + { + "epoch": 1.0550186690094443, + "grad_norm": 2.1852798461914062, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7311234474182129, + "num_tokens": 246773136.0, + "step": 9607 + }, + { + "epoch": 1.055128486712058, + "grad_norm": 2.133063316345215, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6995077729225159, + "num_tokens": 246802632.0, + "step": 9608 + }, + { + "epoch": 1.0552383044146716, + "grad_norm": 2.077346086502075, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7384629249572754, + "num_tokens": 246831589.0, + "step": 9609 + }, + { + "epoch": 1.0553481221172853, + "grad_norm": 2.4147822856903076, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7333036661148071, + "num_tokens": 246852592.0, + "step": 9610 + }, + { + "epoch": 1.0554579398198989, + "grad_norm": 2.291579246520996, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7184830904006958, + "num_tokens": 246878575.0, + "step": 9611 + }, + { + "epoch": 1.0555677575225126, + "grad_norm": 2.3589348793029785, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7149358987808228, + "num_tokens": 246901555.0, + "step": 9612 + }, + { + "epoch": 1.0556775752251264, + "grad_norm": 2.1440041065216064, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7364460229873657, + "num_tokens": 246927883.0, + "step": 9613 + }, + { + "epoch": 1.05578739292774, + "grad_norm": 2.168595314025879, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6956483721733093, + "num_tokens": 246955675.0, + "step": 9614 + }, + { + "epoch": 1.0558972106303537, + "grad_norm": 2.623286247253418, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.712200403213501, + "num_tokens": 246974066.0, + "step": 9615 + }, + { + "epoch": 1.0560070283329672, + "grad_norm": 2.121652603149414, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7304660081863403, + "num_tokens": 247003241.0, + "step": 9616 + }, + { + "epoch": 1.056116846035581, + "grad_norm": 2.1291253566741943, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7141421437263489, + "num_tokens": 247032745.0, + "step": 9617 + }, + { + "epoch": 1.0562266637381945, + "grad_norm": 2.141660213470459, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.707215428352356, + "num_tokens": 247062183.0, + "step": 9618 + }, + { + "epoch": 1.0563364814408083, + "grad_norm": 2.260071039199829, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7189216613769531, + "num_tokens": 247085154.0, + "step": 9619 + }, + { + "epoch": 1.0564462991434218, + "grad_norm": 1.9787951707839966, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6912498474121094, + "num_tokens": 247116329.0, + "step": 9620 + }, + { + "epoch": 1.0565561168460356, + "grad_norm": 2.2620222568511963, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7107733488082886, + "num_tokens": 247141529.0, + "step": 9621 + }, + { + "epoch": 1.0566659345486493, + "grad_norm": 2.145962715148926, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.708531379699707, + "num_tokens": 247170327.0, + "step": 9622 + }, + { + "epoch": 1.0567757522512629, + "grad_norm": 2.353978157043457, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7332854866981506, + "num_tokens": 247193238.0, + "step": 9623 + }, + { + "epoch": 1.0568855699538766, + "grad_norm": 2.2695541381835938, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7087855339050293, + "num_tokens": 247219539.0, + "step": 9624 + }, + { + "epoch": 1.0569953876564901, + "grad_norm": 1.90772545337677, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6977705955505371, + "num_tokens": 247254810.0, + "step": 9625 + }, + { + "epoch": 1.057105205359104, + "grad_norm": 2.328557014465332, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7200891971588135, + "num_tokens": 247278273.0, + "step": 9626 + }, + { + "epoch": 1.0572150230617177, + "grad_norm": 2.227903127670288, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7127841114997864, + "num_tokens": 247304537.0, + "step": 9627 + }, + { + "epoch": 1.0573248407643312, + "grad_norm": 2.4146957397460938, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7374680042266846, + "num_tokens": 247327102.0, + "step": 9628 + }, + { + "epoch": 1.057434658466945, + "grad_norm": 2.4775657653808594, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7293868064880371, + "num_tokens": 247349773.0, + "step": 9629 + }, + { + "epoch": 1.0575444761695585, + "grad_norm": 2.0964670181274414, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7404885292053223, + "num_tokens": 247377025.0, + "step": 9630 + }, + { + "epoch": 1.0576542938721722, + "grad_norm": 2.3653178215026855, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7437958717346191, + "num_tokens": 247397698.0, + "step": 9631 + }, + { + "epoch": 1.0577641115747858, + "grad_norm": 2.2218832969665527, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7111858129501343, + "num_tokens": 247423228.0, + "step": 9632 + }, + { + "epoch": 1.0578739292773995, + "grad_norm": 2.42714262008667, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7073822021484375, + "num_tokens": 247446858.0, + "step": 9633 + }, + { + "epoch": 1.057983746980013, + "grad_norm": 2.1646811962127686, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7019731402397156, + "num_tokens": 247475066.0, + "step": 9634 + }, + { + "epoch": 1.0580935646826268, + "grad_norm": 2.318152666091919, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7495577931404114, + "num_tokens": 247497096.0, + "step": 9635 + }, + { + "epoch": 1.0582033823852406, + "grad_norm": 2.3059821128845215, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6986833810806274, + "num_tokens": 247521537.0, + "step": 9636 + }, + { + "epoch": 1.0583132000878541, + "grad_norm": 2.1596508026123047, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7116225957870483, + "num_tokens": 247548111.0, + "step": 9637 + }, + { + "epoch": 1.0584230177904679, + "grad_norm": 2.258000135421753, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7240846157073975, + "num_tokens": 247571384.0, + "step": 9638 + }, + { + "epoch": 1.0585328354930814, + "grad_norm": 2.2511699199676514, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7111250758171082, + "num_tokens": 247597752.0, + "step": 9639 + }, + { + "epoch": 1.0586426531956952, + "grad_norm": 2.008392572402954, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7062282562255859, + "num_tokens": 247628972.0, + "step": 9640 + }, + { + "epoch": 1.0587524708983087, + "grad_norm": 2.434720516204834, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7384414672851562, + "num_tokens": 247650938.0, + "step": 9641 + }, + { + "epoch": 1.0588622886009225, + "grad_norm": 2.215409755706787, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7226753234863281, + "num_tokens": 247677268.0, + "step": 9642 + }, + { + "epoch": 1.0589721063035362, + "grad_norm": 2.424816608428955, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7361133694648743, + "num_tokens": 247700656.0, + "step": 9643 + }, + { + "epoch": 1.0590819240061498, + "grad_norm": 2.5776941776275635, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7365004420280457, + "num_tokens": 247721071.0, + "step": 9644 + }, + { + "epoch": 1.0591917417087635, + "grad_norm": 2.112464666366577, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7354382872581482, + "num_tokens": 247749224.0, + "step": 9645 + }, + { + "epoch": 1.059301559411377, + "grad_norm": 2.2219715118408203, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6995671987533569, + "num_tokens": 247776657.0, + "step": 9646 + }, + { + "epoch": 1.0594113771139908, + "grad_norm": 2.316737651824951, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7197308540344238, + "num_tokens": 247801136.0, + "step": 9647 + }, + { + "epoch": 1.0595211948166043, + "grad_norm": 2.132829189300537, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7171617746353149, + "num_tokens": 247827988.0, + "step": 9648 + }, + { + "epoch": 1.059631012519218, + "grad_norm": 2.1666579246520996, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6946138143539429, + "num_tokens": 247856285.0, + "step": 9649 + }, + { + "epoch": 1.0597408302218319, + "grad_norm": 2.590224027633667, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7071124911308289, + "num_tokens": 247876905.0, + "step": 9650 + }, + { + "epoch": 1.0598506479244454, + "grad_norm": 2.00905704498291, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6992720365524292, + "num_tokens": 247910482.0, + "step": 9651 + }, + { + "epoch": 1.0599604656270591, + "grad_norm": 2.291656017303467, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7210339307785034, + "num_tokens": 247936393.0, + "step": 9652 + }, + { + "epoch": 1.0600702833296727, + "grad_norm": 1.844054102897644, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7034852504730225, + "num_tokens": 247971763.0, + "step": 9653 + }, + { + "epoch": 1.0601801010322864, + "grad_norm": 2.390681266784668, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7140548825263977, + "num_tokens": 247995178.0, + "step": 9654 + }, + { + "epoch": 1.0602899187349, + "grad_norm": 2.421705722808838, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7102943658828735, + "num_tokens": 248018261.0, + "step": 9655 + }, + { + "epoch": 1.0603997364375137, + "grad_norm": 2.3379828929901123, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7123598456382751, + "num_tokens": 248043727.0, + "step": 9656 + }, + { + "epoch": 1.0605095541401275, + "grad_norm": 2.1348977088928223, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7204968333244324, + "num_tokens": 248071179.0, + "step": 9657 + }, + { + "epoch": 1.060619371842741, + "grad_norm": 2.113689422607422, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7111176252365112, + "num_tokens": 248098182.0, + "step": 9658 + }, + { + "epoch": 1.0607291895453548, + "grad_norm": 2.644899368286133, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7207267880439758, + "num_tokens": 248118077.0, + "step": 9659 + }, + { + "epoch": 1.0608390072479683, + "grad_norm": 2.176586866378784, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7038424015045166, + "num_tokens": 248146649.0, + "step": 9660 + }, + { + "epoch": 1.060948824950582, + "grad_norm": 2.441455364227295, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7275725603103638, + "num_tokens": 248167912.0, + "step": 9661 + }, + { + "epoch": 1.0610586426531956, + "grad_norm": 2.033792018890381, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.6992913484573364, + "num_tokens": 248197819.0, + "step": 9662 + }, + { + "epoch": 1.0611684603558094, + "grad_norm": 2.206857442855835, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.731702446937561, + "num_tokens": 248222021.0, + "step": 9663 + }, + { + "epoch": 1.0612782780584231, + "grad_norm": 2.2429075241088867, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7281039953231812, + "num_tokens": 248246414.0, + "step": 9664 + }, + { + "epoch": 1.0613880957610367, + "grad_norm": 1.975454568862915, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7106896638870239, + "num_tokens": 248278286.0, + "step": 9665 + }, + { + "epoch": 1.0614979134636504, + "grad_norm": 2.4762625694274902, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.713071882724762, + "num_tokens": 248301547.0, + "step": 9666 + }, + { + "epoch": 1.061607731166264, + "grad_norm": 2.123905658721924, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7329631447792053, + "num_tokens": 248328825.0, + "step": 9667 + }, + { + "epoch": 1.0617175488688777, + "grad_norm": 2.2275726795196533, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7109218835830688, + "num_tokens": 248355709.0, + "step": 9668 + }, + { + "epoch": 1.0618273665714912, + "grad_norm": 2.124504566192627, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7261254787445068, + "num_tokens": 248381778.0, + "step": 9669 + }, + { + "epoch": 1.061937184274105, + "grad_norm": 2.3470802307128906, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7221983671188354, + "num_tokens": 248404761.0, + "step": 9670 + }, + { + "epoch": 1.0620470019767188, + "grad_norm": 2.224525213241577, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7389209866523743, + "num_tokens": 248428700.0, + "step": 9671 + }, + { + "epoch": 1.0621568196793323, + "grad_norm": 2.1919467449188232, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7310601472854614, + "num_tokens": 248455100.0, + "step": 9672 + }, + { + "epoch": 1.062266637381946, + "grad_norm": 2.0661723613739014, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7161834239959717, + "num_tokens": 248483276.0, + "step": 9673 + }, + { + "epoch": 1.0623764550845596, + "grad_norm": 2.1888725757598877, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.692142128944397, + "num_tokens": 248514075.0, + "step": 9674 + }, + { + "epoch": 1.0624862727871733, + "grad_norm": 2.637648105621338, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7196735143661499, + "num_tokens": 248535373.0, + "step": 9675 + }, + { + "epoch": 1.0625960904897869, + "grad_norm": 2.5202980041503906, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7343769073486328, + "num_tokens": 248556872.0, + "step": 9676 + }, + { + "epoch": 1.0627059081924006, + "grad_norm": 2.3454418182373047, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7118043899536133, + "num_tokens": 248580738.0, + "step": 9677 + }, + { + "epoch": 1.0628157258950144, + "grad_norm": 2.222555160522461, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.710299015045166, + "num_tokens": 248605933.0, + "step": 9678 + }, + { + "epoch": 1.062925543597628, + "grad_norm": 1.888449788093567, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7142310738563538, + "num_tokens": 248640809.0, + "step": 9679 + }, + { + "epoch": 1.0630353613002417, + "grad_norm": 2.580825090408325, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.73674476146698, + "num_tokens": 248662245.0, + "step": 9680 + }, + { + "epoch": 1.0631451790028552, + "grad_norm": 2.519296407699585, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7170973420143127, + "num_tokens": 248684480.0, + "step": 9681 + }, + { + "epoch": 1.063254996705469, + "grad_norm": 2.194612979888916, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6876582503318787, + "num_tokens": 248714507.0, + "step": 9682 + }, + { + "epoch": 1.0633648144080825, + "grad_norm": 2.1261541843414307, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7088796496391296, + "num_tokens": 248744016.0, + "step": 9683 + }, + { + "epoch": 1.0634746321106963, + "grad_norm": 2.0813567638397217, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6975557804107666, + "num_tokens": 248771923.0, + "step": 9684 + }, + { + "epoch": 1.0635844498133098, + "grad_norm": 2.4315385818481445, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7176358699798584, + "num_tokens": 248796111.0, + "step": 9685 + }, + { + "epoch": 1.0636942675159236, + "grad_norm": 2.709716796875, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7272608280181885, + "num_tokens": 248816203.0, + "step": 9686 + }, + { + "epoch": 1.0638040852185373, + "grad_norm": 2.1889445781707764, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6838785409927368, + "num_tokens": 248844460.0, + "step": 9687 + }, + { + "epoch": 1.0639139029211508, + "grad_norm": 2.348602771759033, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7219783067703247, + "num_tokens": 248869096.0, + "step": 9688 + }, + { + "epoch": 1.0640237206237646, + "grad_norm": 2.0759525299072266, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7360053658485413, + "num_tokens": 248898237.0, + "step": 9689 + }, + { + "epoch": 1.0641335383263781, + "grad_norm": 2.4491026401519775, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7438027262687683, + "num_tokens": 248918896.0, + "step": 9690 + }, + { + "epoch": 1.064243356028992, + "grad_norm": 2.349682331085205, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.731845498085022, + "num_tokens": 248940493.0, + "step": 9691 + }, + { + "epoch": 1.0643531737316057, + "grad_norm": 2.4999918937683105, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7318488955497742, + "num_tokens": 248960566.0, + "step": 9692 + }, + { + "epoch": 1.0644629914342192, + "grad_norm": 2.452430009841919, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7049493193626404, + "num_tokens": 248983628.0, + "step": 9693 + }, + { + "epoch": 1.064572809136833, + "grad_norm": 2.169795036315918, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7208731174468994, + "num_tokens": 249012783.0, + "step": 9694 + }, + { + "epoch": 1.0646826268394465, + "grad_norm": 2.674382448196411, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7454711198806763, + "num_tokens": 249030898.0, + "step": 9695 + }, + { + "epoch": 1.0647924445420602, + "grad_norm": 2.443877696990967, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7248696088790894, + "num_tokens": 249052100.0, + "step": 9696 + }, + { + "epoch": 1.0649022622446738, + "grad_norm": 2.148336887359619, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7213488221168518, + "num_tokens": 249079142.0, + "step": 9697 + }, + { + "epoch": 1.0650120799472875, + "grad_norm": 2.5874898433685303, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7169989347457886, + "num_tokens": 249099261.0, + "step": 9698 + }, + { + "epoch": 1.065121897649901, + "grad_norm": 2.5616228580474854, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7137999534606934, + "num_tokens": 249119505.0, + "step": 9699 + }, + { + "epoch": 1.0652317153525148, + "grad_norm": 2.1999638080596924, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7321288585662842, + "num_tokens": 249145942.0, + "step": 9700 + }, + { + "epoch": 1.0653415330551286, + "grad_norm": 2.096984624862671, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7024641036987305, + "num_tokens": 249174959.0, + "step": 9701 + }, + { + "epoch": 1.065451350757742, + "grad_norm": 1.9815874099731445, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7352756261825562, + "num_tokens": 249205364.0, + "step": 9702 + }, + { + "epoch": 1.0655611684603559, + "grad_norm": 2.122802972793579, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7450681328773499, + "num_tokens": 249231119.0, + "step": 9703 + }, + { + "epoch": 1.0656709861629694, + "grad_norm": 1.9638190269470215, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7347686886787415, + "num_tokens": 249261653.0, + "step": 9704 + }, + { + "epoch": 1.0657808038655832, + "grad_norm": 2.5503952503204346, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7353211641311646, + "num_tokens": 249281622.0, + "step": 9705 + }, + { + "epoch": 1.0658906215681967, + "grad_norm": 2.1736674308776855, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7408317923545837, + "num_tokens": 249308572.0, + "step": 9706 + }, + { + "epoch": 1.0660004392708105, + "grad_norm": 2.7483839988708496, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7360724806785583, + "num_tokens": 249326577.0, + "step": 9707 + }, + { + "epoch": 1.0661102569734242, + "grad_norm": 1.9124373197555542, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6889669895172119, + "num_tokens": 249364265.0, + "step": 9708 + }, + { + "epoch": 1.0662200746760377, + "grad_norm": 2.0182836055755615, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.717661440372467, + "num_tokens": 249395542.0, + "step": 9709 + }, + { + "epoch": 1.0663298923786515, + "grad_norm": 2.3917124271392822, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7171372175216675, + "num_tokens": 249417702.0, + "step": 9710 + }, + { + "epoch": 1.066439710081265, + "grad_norm": 2.5792155265808105, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7152214646339417, + "num_tokens": 249439538.0, + "step": 9711 + }, + { + "epoch": 1.0665495277838788, + "grad_norm": 2.05428147315979, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7301952838897705, + "num_tokens": 249466264.0, + "step": 9712 + }, + { + "epoch": 1.0666593454864923, + "grad_norm": 2.254418134689331, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7278388738632202, + "num_tokens": 249490846.0, + "step": 9713 + }, + { + "epoch": 1.066769163189106, + "grad_norm": 2.4669015407562256, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7206330299377441, + "num_tokens": 249512474.0, + "step": 9714 + }, + { + "epoch": 1.0668789808917198, + "grad_norm": 2.483250379562378, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7015857696533203, + "num_tokens": 249536133.0, + "step": 9715 + }, + { + "epoch": 1.0669887985943334, + "grad_norm": 2.060680866241455, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7112890481948853, + "num_tokens": 249565485.0, + "step": 9716 + }, + { + "epoch": 1.0670986162969471, + "grad_norm": 2.3354504108428955, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7144489288330078, + "num_tokens": 249590370.0, + "step": 9717 + }, + { + "epoch": 1.0672084339995607, + "grad_norm": 1.9755641222000122, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7071471214294434, + "num_tokens": 249620399.0, + "step": 9718 + }, + { + "epoch": 1.0673182517021744, + "grad_norm": 2.3822243213653564, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7275266647338867, + "num_tokens": 249642060.0, + "step": 9719 + }, + { + "epoch": 1.067428069404788, + "grad_norm": 2.2312803268432617, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7055660486221313, + "num_tokens": 249666957.0, + "step": 9720 + }, + { + "epoch": 1.0675378871074017, + "grad_norm": 2.2553746700286865, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7173711657524109, + "num_tokens": 249691725.0, + "step": 9721 + }, + { + "epoch": 1.0676477048100155, + "grad_norm": 2.5286238193511963, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7155999541282654, + "num_tokens": 249714356.0, + "step": 9722 + }, + { + "epoch": 1.067757522512629, + "grad_norm": 2.4826745986938477, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7083067297935486, + "num_tokens": 249737200.0, + "step": 9723 + }, + { + "epoch": 1.0678673402152428, + "grad_norm": 2.3523757457733154, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7123880386352539, + "num_tokens": 249761172.0, + "step": 9724 + }, + { + "epoch": 1.0679771579178563, + "grad_norm": 2.0040102005004883, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7034574151039124, + "num_tokens": 249790092.0, + "step": 9725 + }, + { + "epoch": 1.06808697562047, + "grad_norm": 2.301752805709839, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.729386031627655, + "num_tokens": 249813449.0, + "step": 9726 + }, + { + "epoch": 1.0681967933230836, + "grad_norm": 2.233416795730591, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7160311341285706, + "num_tokens": 249840637.0, + "step": 9727 + }, + { + "epoch": 1.0683066110256974, + "grad_norm": 2.193598508834839, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7226003408432007, + "num_tokens": 249867298.0, + "step": 9728 + }, + { + "epoch": 1.068416428728311, + "grad_norm": 2.1260411739349365, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7035472393035889, + "num_tokens": 249895103.0, + "step": 9729 + }, + { + "epoch": 1.0685262464309246, + "grad_norm": 2.365612745285034, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.728813648223877, + "num_tokens": 249917280.0, + "step": 9730 + }, + { + "epoch": 1.0686360641335384, + "grad_norm": 2.235482931137085, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7297424674034119, + "num_tokens": 249941172.0, + "step": 9731 + }, + { + "epoch": 1.068745881836152, + "grad_norm": 2.4452145099639893, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7191137671470642, + "num_tokens": 249963876.0, + "step": 9732 + }, + { + "epoch": 1.0688556995387657, + "grad_norm": 2.346895933151245, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7015857696533203, + "num_tokens": 249989777.0, + "step": 9733 + }, + { + "epoch": 1.0689655172413792, + "grad_norm": 2.3680837154388428, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.727137565612793, + "num_tokens": 250013418.0, + "step": 9734 + }, + { + "epoch": 1.069075334943993, + "grad_norm": 2.364806890487671, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7247553467750549, + "num_tokens": 250037901.0, + "step": 9735 + }, + { + "epoch": 1.0691851526466065, + "grad_norm": 2.3621654510498047, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7048419713973999, + "num_tokens": 250061762.0, + "step": 9736 + }, + { + "epoch": 1.0692949703492203, + "grad_norm": 2.118215322494507, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7092600464820862, + "num_tokens": 250089763.0, + "step": 9737 + }, + { + "epoch": 1.069404788051834, + "grad_norm": 2.6401374340057373, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7246009111404419, + "num_tokens": 250108620.0, + "step": 9738 + }, + { + "epoch": 1.0695146057544476, + "grad_norm": 2.508436679840088, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7393682599067688, + "num_tokens": 250129824.0, + "step": 9739 + }, + { + "epoch": 1.0696244234570613, + "grad_norm": 2.298619270324707, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7240511178970337, + "num_tokens": 250157551.0, + "step": 9740 + }, + { + "epoch": 1.0697342411596749, + "grad_norm": 2.320352554321289, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7194425463676453, + "num_tokens": 250183843.0, + "step": 9741 + }, + { + "epoch": 1.0698440588622886, + "grad_norm": 2.1148345470428467, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7173572778701782, + "num_tokens": 250211873.0, + "step": 9742 + }, + { + "epoch": 1.0699538765649024, + "grad_norm": 2.0592713356018066, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7231280207633972, + "num_tokens": 250240181.0, + "step": 9743 + }, + { + "epoch": 1.070063694267516, + "grad_norm": 1.9782921075820923, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7072888612747192, + "num_tokens": 250269800.0, + "step": 9744 + }, + { + "epoch": 1.0701735119701297, + "grad_norm": 2.3662779331207275, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.712709903717041, + "num_tokens": 250291445.0, + "step": 9745 + }, + { + "epoch": 1.0702833296727432, + "grad_norm": 2.251725912094116, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7021366953849792, + "num_tokens": 250317560.0, + "step": 9746 + }, + { + "epoch": 1.070393147375357, + "grad_norm": 2.290057897567749, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7206019163131714, + "num_tokens": 250341120.0, + "step": 9747 + }, + { + "epoch": 1.0705029650779705, + "grad_norm": 2.1813442707061768, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7341681718826294, + "num_tokens": 250366068.0, + "step": 9748 + }, + { + "epoch": 1.0706127827805842, + "grad_norm": 2.3955283164978027, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7359148263931274, + "num_tokens": 250388917.0, + "step": 9749 + }, + { + "epoch": 1.0707226004831978, + "grad_norm": 2.3847618103027344, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7381860613822937, + "num_tokens": 250409338.0, + "step": 9750 + }, + { + "epoch": 1.0708324181858115, + "grad_norm": 2.268620252609253, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7283608913421631, + "num_tokens": 250437901.0, + "step": 9751 + }, + { + "epoch": 1.0709422358884253, + "grad_norm": 2.695413112640381, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7061032056808472, + "num_tokens": 250456746.0, + "step": 9752 + }, + { + "epoch": 1.0710520535910388, + "grad_norm": 1.9018257856369019, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7066195011138916, + "num_tokens": 250493211.0, + "step": 9753 + }, + { + "epoch": 1.0711618712936526, + "grad_norm": 2.3285577297210693, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7090386152267456, + "num_tokens": 250516287.0, + "step": 9754 + }, + { + "epoch": 1.0712716889962661, + "grad_norm": 2.25589919090271, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7048453688621521, + "num_tokens": 250541907.0, + "step": 9755 + }, + { + "epoch": 1.0713815066988799, + "grad_norm": 2.5917019844055176, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7384616136550903, + "num_tokens": 250561759.0, + "step": 9756 + }, + { + "epoch": 1.0714913244014934, + "grad_norm": 2.095200300216675, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7057017087936401, + "num_tokens": 250591822.0, + "step": 9757 + }, + { + "epoch": 1.0716011421041072, + "grad_norm": 2.4687323570251465, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7187929153442383, + "num_tokens": 250612393.0, + "step": 9758 + }, + { + "epoch": 1.071710959806721, + "grad_norm": 2.2845499515533447, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7242532968521118, + "num_tokens": 250635232.0, + "step": 9759 + }, + { + "epoch": 1.0718207775093345, + "grad_norm": 1.9533599615097046, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.713152289390564, + "num_tokens": 250665324.0, + "step": 9760 + }, + { + "epoch": 1.0719305952119482, + "grad_norm": 2.2463579177856445, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7373550534248352, + "num_tokens": 250688633.0, + "step": 9761 + }, + { + "epoch": 1.0720404129145618, + "grad_norm": 2.42163348197937, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7182700634002686, + "num_tokens": 250712391.0, + "step": 9762 + }, + { + "epoch": 1.0721502306171755, + "grad_norm": 2.1940090656280518, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7006322145462036, + "num_tokens": 250738348.0, + "step": 9763 + }, + { + "epoch": 1.072260048319789, + "grad_norm": 2.410604953765869, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7439502477645874, + "num_tokens": 250758977.0, + "step": 9764 + }, + { + "epoch": 1.0723698660224028, + "grad_norm": 2.4345576763153076, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7248221039772034, + "num_tokens": 250780144.0, + "step": 9765 + }, + { + "epoch": 1.0724796837250166, + "grad_norm": 1.9640158414840698, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7157107591629028, + "num_tokens": 250812648.0, + "step": 9766 + }, + { + "epoch": 1.07258950142763, + "grad_norm": 2.181626081466675, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7024354934692383, + "num_tokens": 250839062.0, + "step": 9767 + }, + { + "epoch": 1.0726993191302439, + "grad_norm": 2.1456410884857178, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7021199464797974, + "num_tokens": 250867105.0, + "step": 9768 + }, + { + "epoch": 1.0728091368328574, + "grad_norm": 2.1292335987091064, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7073825597763062, + "num_tokens": 250896657.0, + "step": 9769 + }, + { + "epoch": 1.0729189545354711, + "grad_norm": 2.408352851867676, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7447909116744995, + "num_tokens": 250916811.0, + "step": 9770 + }, + { + "epoch": 1.0730287722380847, + "grad_norm": 2.0616512298583984, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7142753005027771, + "num_tokens": 250945692.0, + "step": 9771 + }, + { + "epoch": 1.0731385899406984, + "grad_norm": 2.421367883682251, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7023198008537292, + "num_tokens": 250969765.0, + "step": 9772 + }, + { + "epoch": 1.0732484076433122, + "grad_norm": 2.23515248298645, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7349461317062378, + "num_tokens": 250994712.0, + "step": 9773 + }, + { + "epoch": 1.0733582253459257, + "grad_norm": 2.2819933891296387, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7196983695030212, + "num_tokens": 251019583.0, + "step": 9774 + }, + { + "epoch": 1.0734680430485395, + "grad_norm": 2.1900951862335205, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7288767099380493, + "num_tokens": 251044574.0, + "step": 9775 + }, + { + "epoch": 1.073577860751153, + "grad_norm": 2.200822353363037, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7229113578796387, + "num_tokens": 251070636.0, + "step": 9776 + }, + { + "epoch": 1.0736876784537668, + "grad_norm": 2.2625651359558105, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7391656041145325, + "num_tokens": 251094382.0, + "step": 9777 + }, + { + "epoch": 1.0737974961563803, + "grad_norm": 2.181703567504883, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7446516752243042, + "num_tokens": 251119744.0, + "step": 9778 + }, + { + "epoch": 1.073907313858994, + "grad_norm": 2.4254043102264404, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7560448050498962, + "num_tokens": 251141081.0, + "step": 9779 + }, + { + "epoch": 1.0740171315616078, + "grad_norm": 1.9218478202819824, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7157320976257324, + "num_tokens": 251174089.0, + "step": 9780 + }, + { + "epoch": 1.0741269492642214, + "grad_norm": 2.031614065170288, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.713707447052002, + "num_tokens": 251204976.0, + "step": 9781 + }, + { + "epoch": 1.0742367669668351, + "grad_norm": 2.1002001762390137, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7285251617431641, + "num_tokens": 251232409.0, + "step": 9782 + }, + { + "epoch": 1.0743465846694487, + "grad_norm": 2.5542373657226562, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7239457964897156, + "num_tokens": 251251999.0, + "step": 9783 + }, + { + "epoch": 1.0744564023720624, + "grad_norm": 2.2094271183013916, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6880612373352051, + "num_tokens": 251281487.0, + "step": 9784 + }, + { + "epoch": 1.074566220074676, + "grad_norm": 2.0845844745635986, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7197299003601074, + "num_tokens": 251308271.0, + "step": 9785 + }, + { + "epoch": 1.0746760377772897, + "grad_norm": 2.571957588195801, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7071871757507324, + "num_tokens": 251329144.0, + "step": 9786 + }, + { + "epoch": 1.0747858554799035, + "grad_norm": 2.2558798789978027, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7173353433609009, + "num_tokens": 251355356.0, + "step": 9787 + }, + { + "epoch": 1.074895673182517, + "grad_norm": 2.3008604049682617, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7053461074829102, + "num_tokens": 251382746.0, + "step": 9788 + }, + { + "epoch": 1.0750054908851308, + "grad_norm": 2.1537137031555176, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7081193923950195, + "num_tokens": 251410606.0, + "step": 9789 + }, + { + "epoch": 1.0751153085877443, + "grad_norm": 2.3745031356811523, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7206809520721436, + "num_tokens": 251433130.0, + "step": 9790 + }, + { + "epoch": 1.075225126290358, + "grad_norm": 2.0253231525421143, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7121535539627075, + "num_tokens": 251464784.0, + "step": 9791 + }, + { + "epoch": 1.0753349439929716, + "grad_norm": 2.087183713912964, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7085971832275391, + "num_tokens": 251493140.0, + "step": 9792 + }, + { + "epoch": 1.0754447616955853, + "grad_norm": 2.020805835723877, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7136541604995728, + "num_tokens": 251522647.0, + "step": 9793 + }, + { + "epoch": 1.075554579398199, + "grad_norm": 2.124567747116089, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7106096744537354, + "num_tokens": 251548689.0, + "step": 9794 + }, + { + "epoch": 1.0756643971008126, + "grad_norm": 2.3704869747161865, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7100986242294312, + "num_tokens": 251572426.0, + "step": 9795 + }, + { + "epoch": 1.0757742148034264, + "grad_norm": 2.34232234954834, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7069183588027954, + "num_tokens": 251598395.0, + "step": 9796 + }, + { + "epoch": 1.07588403250604, + "grad_norm": 2.437504768371582, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.717508852481842, + "num_tokens": 251620078.0, + "step": 9797 + }, + { + "epoch": 1.0759938502086537, + "grad_norm": 2.126373529434204, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7314797639846802, + "num_tokens": 251648169.0, + "step": 9798 + }, + { + "epoch": 1.0761036679112672, + "grad_norm": 2.3728296756744385, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6957019567489624, + "num_tokens": 251675074.0, + "step": 9799 + }, + { + "epoch": 1.076213485613881, + "grad_norm": 2.5991742610931396, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7368094325065613, + "num_tokens": 251693568.0, + "step": 9800 + }, + { + "epoch": 1.0763233033164945, + "grad_norm": 2.206376791000366, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.709889829158783, + "num_tokens": 251720128.0, + "step": 9801 + }, + { + "epoch": 1.0764331210191083, + "grad_norm": 2.2578508853912354, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7230315804481506, + "num_tokens": 251745994.0, + "step": 9802 + }, + { + "epoch": 1.076542938721722, + "grad_norm": 2.552619695663452, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7382889986038208, + "num_tokens": 251766569.0, + "step": 9803 + }, + { + "epoch": 1.0766527564243356, + "grad_norm": 2.071441650390625, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7082042694091797, + "num_tokens": 251796762.0, + "step": 9804 + }, + { + "epoch": 1.0767625741269493, + "grad_norm": 2.499307632446289, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7449885606765747, + "num_tokens": 251816802.0, + "step": 9805 + }, + { + "epoch": 1.0768723918295628, + "grad_norm": 2.438654899597168, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7212291359901428, + "num_tokens": 251840361.0, + "step": 9806 + }, + { + "epoch": 1.0769822095321766, + "grad_norm": 2.588683843612671, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7177024483680725, + "num_tokens": 251861535.0, + "step": 9807 + }, + { + "epoch": 1.0770920272347904, + "grad_norm": 2.017216682434082, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.74250727891922, + "num_tokens": 251889950.0, + "step": 9808 + }, + { + "epoch": 1.077201844937404, + "grad_norm": 2.005619764328003, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7251032590866089, + "num_tokens": 251921388.0, + "step": 9809 + }, + { + "epoch": 1.0773116626400177, + "grad_norm": 1.989405870437622, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7378569841384888, + "num_tokens": 251949934.0, + "step": 9810 + }, + { + "epoch": 1.0774214803426312, + "grad_norm": 1.8398487567901611, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7210819125175476, + "num_tokens": 251984004.0, + "step": 9811 + }, + { + "epoch": 1.077531298045245, + "grad_norm": 2.153856039047241, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7256420850753784, + "num_tokens": 252009580.0, + "step": 9812 + }, + { + "epoch": 1.0776411157478585, + "grad_norm": 2.1583251953125, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7308890223503113, + "num_tokens": 252036612.0, + "step": 9813 + }, + { + "epoch": 1.0777509334504722, + "grad_norm": 2.1083738803863525, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7094804048538208, + "num_tokens": 252065485.0, + "step": 9814 + }, + { + "epoch": 1.0778607511530858, + "grad_norm": 2.1345677375793457, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.727224588394165, + "num_tokens": 252090955.0, + "step": 9815 + }, + { + "epoch": 1.0779705688556995, + "grad_norm": 2.2625536918640137, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7170289754867554, + "num_tokens": 252117356.0, + "step": 9816 + }, + { + "epoch": 1.0780803865583133, + "grad_norm": 2.3362908363342285, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7154316902160645, + "num_tokens": 252141449.0, + "step": 9817 + }, + { + "epoch": 1.0781902042609268, + "grad_norm": 2.1628973484039307, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7098259925842285, + "num_tokens": 252168216.0, + "step": 9818 + }, + { + "epoch": 1.0783000219635406, + "grad_norm": 2.3022139072418213, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7335108518600464, + "num_tokens": 252192216.0, + "step": 9819 + }, + { + "epoch": 1.0784098396661541, + "grad_norm": 2.1454198360443115, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7156475782394409, + "num_tokens": 252218607.0, + "step": 9820 + }, + { + "epoch": 1.0785196573687679, + "grad_norm": 2.3542401790618896, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.72212815284729, + "num_tokens": 252242963.0, + "step": 9821 + }, + { + "epoch": 1.0786294750713814, + "grad_norm": 2.1808359622955322, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6983165144920349, + "num_tokens": 252270623.0, + "step": 9822 + }, + { + "epoch": 1.0787392927739952, + "grad_norm": 2.200418472290039, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6912164688110352, + "num_tokens": 252299802.0, + "step": 9823 + }, + { + "epoch": 1.078849110476609, + "grad_norm": 1.869344711303711, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6897186636924744, + "num_tokens": 252334315.0, + "step": 9824 + }, + { + "epoch": 1.0789589281792225, + "grad_norm": 2.2904746532440186, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.715825080871582, + "num_tokens": 252357394.0, + "step": 9825 + }, + { + "epoch": 1.0790687458818362, + "grad_norm": 2.4367220401763916, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.730466365814209, + "num_tokens": 252379472.0, + "step": 9826 + }, + { + "epoch": 1.0791785635844497, + "grad_norm": 2.6042821407318115, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7291555404663086, + "num_tokens": 252400109.0, + "step": 9827 + }, + { + "epoch": 1.0792883812870635, + "grad_norm": 1.9788483381271362, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7107944488525391, + "num_tokens": 252432269.0, + "step": 9828 + }, + { + "epoch": 1.079398198989677, + "grad_norm": 2.3586642742156982, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7386896014213562, + "num_tokens": 252458015.0, + "step": 9829 + }, + { + "epoch": 1.0795080166922908, + "grad_norm": 2.088916301727295, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.701222836971283, + "num_tokens": 252487759.0, + "step": 9830 + }, + { + "epoch": 1.0796178343949046, + "grad_norm": 2.5966618061065674, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7211482524871826, + "num_tokens": 252509132.0, + "step": 9831 + }, + { + "epoch": 1.079727652097518, + "grad_norm": 2.4316635131835938, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.727528989315033, + "num_tokens": 252531431.0, + "step": 9832 + }, + { + "epoch": 1.0798374698001318, + "grad_norm": 2.422801971435547, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7261350154876709, + "num_tokens": 252556446.0, + "step": 9833 + }, + { + "epoch": 1.0799472875027454, + "grad_norm": 2.142767906188965, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7177636623382568, + "num_tokens": 252584560.0, + "step": 9834 + }, + { + "epoch": 1.0800571052053591, + "grad_norm": 2.1898505687713623, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.725043773651123, + "num_tokens": 252611695.0, + "step": 9835 + }, + { + "epoch": 1.0801669229079727, + "grad_norm": 2.168639898300171, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7108386158943176, + "num_tokens": 252640567.0, + "step": 9836 + }, + { + "epoch": 1.0802767406105864, + "grad_norm": 2.410583257675171, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7366242408752441, + "num_tokens": 252660794.0, + "step": 9837 + }, + { + "epoch": 1.0803865583132002, + "grad_norm": 2.3420567512512207, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7273984551429749, + "num_tokens": 252684504.0, + "step": 9838 + }, + { + "epoch": 1.0804963760158137, + "grad_norm": 2.1837158203125, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6990792751312256, + "num_tokens": 252714689.0, + "step": 9839 + }, + { + "epoch": 1.0806061937184275, + "grad_norm": 2.114196538925171, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7108823657035828, + "num_tokens": 252743671.0, + "step": 9840 + }, + { + "epoch": 1.080716011421041, + "grad_norm": 2.174075126647949, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7447389364242554, + "num_tokens": 252768138.0, + "step": 9841 + }, + { + "epoch": 1.0808258291236548, + "grad_norm": 2.269691228866577, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7314167618751526, + "num_tokens": 252794332.0, + "step": 9842 + }, + { + "epoch": 1.0809356468262683, + "grad_norm": 2.091794490814209, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.729699969291687, + "num_tokens": 252822955.0, + "step": 9843 + }, + { + "epoch": 1.081045464528882, + "grad_norm": 2.3924708366394043, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7118132710456848, + "num_tokens": 252846401.0, + "step": 9844 + }, + { + "epoch": 1.0811552822314958, + "grad_norm": 2.2688193321228027, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7239140272140503, + "num_tokens": 252872414.0, + "step": 9845 + }, + { + "epoch": 1.0812650999341094, + "grad_norm": 2.5463716983795166, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7412437796592712, + "num_tokens": 252893154.0, + "step": 9846 + }, + { + "epoch": 1.081374917636723, + "grad_norm": 2.235013008117676, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7301687598228455, + "num_tokens": 252918385.0, + "step": 9847 + }, + { + "epoch": 1.0814847353393366, + "grad_norm": 2.3098106384277344, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7175583839416504, + "num_tokens": 252943990.0, + "step": 9848 + }, + { + "epoch": 1.0815945530419504, + "grad_norm": 2.3576719760894775, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7134839296340942, + "num_tokens": 252967488.0, + "step": 9849 + }, + { + "epoch": 1.081704370744564, + "grad_norm": 2.018087148666382, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7020835280418396, + "num_tokens": 252998339.0, + "step": 9850 + }, + { + "epoch": 1.0818141884471777, + "grad_norm": 2.469721555709839, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.734504222869873, + "num_tokens": 253019322.0, + "step": 9851 + }, + { + "epoch": 1.0819240061497912, + "grad_norm": 2.00486159324646, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.704637885093689, + "num_tokens": 253049502.0, + "step": 9852 + }, + { + "epoch": 1.082033823852405, + "grad_norm": 2.2587296962738037, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7462166547775269, + "num_tokens": 253078070.0, + "step": 9853 + }, + { + "epoch": 1.0821436415550187, + "grad_norm": 2.827211618423462, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7252721190452576, + "num_tokens": 253096059.0, + "step": 9854 + }, + { + "epoch": 1.0822534592576323, + "grad_norm": 2.1806280612945557, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7172853350639343, + "num_tokens": 253123238.0, + "step": 9855 + }, + { + "epoch": 1.082363276960246, + "grad_norm": 2.1032919883728027, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7124406099319458, + "num_tokens": 253155436.0, + "step": 9856 + }, + { + "epoch": 1.0824730946628596, + "grad_norm": 2.2244083881378174, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7186640501022339, + "num_tokens": 253181809.0, + "step": 9857 + }, + { + "epoch": 1.0825829123654733, + "grad_norm": 2.064319133758545, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6913689374923706, + "num_tokens": 253214531.0, + "step": 9858 + }, + { + "epoch": 1.082692730068087, + "grad_norm": 2.465745210647583, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7132567167282104, + "num_tokens": 253236004.0, + "step": 9859 + }, + { + "epoch": 1.0828025477707006, + "grad_norm": 1.946445345878601, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7385020852088928, + "num_tokens": 253266684.0, + "step": 9860 + }, + { + "epoch": 1.0829123654733144, + "grad_norm": 2.385080575942993, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.704533576965332, + "num_tokens": 253292576.0, + "step": 9861 + }, + { + "epoch": 1.083022183175928, + "grad_norm": 2.087970018386841, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7187207341194153, + "num_tokens": 253319694.0, + "step": 9862 + }, + { + "epoch": 1.0831320008785417, + "grad_norm": 2.4354805946350098, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7336599826812744, + "num_tokens": 253343358.0, + "step": 9863 + }, + { + "epoch": 1.0832418185811552, + "grad_norm": 2.1795389652252197, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7226096987724304, + "num_tokens": 253370334.0, + "step": 9864 + }, + { + "epoch": 1.083351636283769, + "grad_norm": 2.3640997409820557, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7390998005867004, + "num_tokens": 253392514.0, + "step": 9865 + }, + { + "epoch": 1.0834614539863825, + "grad_norm": 2.455702543258667, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7573834657669067, + "num_tokens": 253412040.0, + "step": 9866 + }, + { + "epoch": 1.0835712716889963, + "grad_norm": 2.2591936588287354, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7357288599014282, + "num_tokens": 253435914.0, + "step": 9867 + }, + { + "epoch": 1.08368108939161, + "grad_norm": 2.518864393234253, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7261674404144287, + "num_tokens": 253456152.0, + "step": 9868 + }, + { + "epoch": 1.0837909070942235, + "grad_norm": 2.1793155670166016, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7280550003051758, + "num_tokens": 253484421.0, + "step": 9869 + }, + { + "epoch": 1.0839007247968373, + "grad_norm": 2.2224066257476807, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7070943713188171, + "num_tokens": 253509714.0, + "step": 9870 + }, + { + "epoch": 1.0840105424994508, + "grad_norm": 1.9857821464538574, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7424206733703613, + "num_tokens": 253536965.0, + "step": 9871 + }, + { + "epoch": 1.0841203602020646, + "grad_norm": 2.581092596054077, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7039713859558105, + "num_tokens": 253557034.0, + "step": 9872 + }, + { + "epoch": 1.0842301779046783, + "grad_norm": 2.3160269260406494, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.686467170715332, + "num_tokens": 253583306.0, + "step": 9873 + }, + { + "epoch": 1.0843399956072919, + "grad_norm": 2.2358222007751465, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7165599465370178, + "num_tokens": 253610399.0, + "step": 9874 + }, + { + "epoch": 1.0844498133099056, + "grad_norm": 2.2273995876312256, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.698373019695282, + "num_tokens": 253636128.0, + "step": 9875 + }, + { + "epoch": 1.0845596310125192, + "grad_norm": 2.2907869815826416, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7039868831634521, + "num_tokens": 253662544.0, + "step": 9876 + }, + { + "epoch": 1.084669448715133, + "grad_norm": 2.0692660808563232, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6997746825218201, + "num_tokens": 253694355.0, + "step": 9877 + }, + { + "epoch": 1.0847792664177465, + "grad_norm": 2.2263989448547363, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6926087737083435, + "num_tokens": 253722256.0, + "step": 9878 + }, + { + "epoch": 1.0848890841203602, + "grad_norm": 2.3358192443847656, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7383162975311279, + "num_tokens": 253746497.0, + "step": 9879 + }, + { + "epoch": 1.0849989018229738, + "grad_norm": 2.0703818798065186, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7289392948150635, + "num_tokens": 253774389.0, + "step": 9880 + }, + { + "epoch": 1.0851087195255875, + "grad_norm": 2.435149669647217, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7171659469604492, + "num_tokens": 253796444.0, + "step": 9881 + }, + { + "epoch": 1.0852185372282013, + "grad_norm": 1.8852232694625854, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7044423818588257, + "num_tokens": 253834117.0, + "step": 9882 + }, + { + "epoch": 1.0853283549308148, + "grad_norm": 2.5510361194610596, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7209758758544922, + "num_tokens": 253856168.0, + "step": 9883 + }, + { + "epoch": 1.0854381726334286, + "grad_norm": 2.4223525524139404, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7194847464561462, + "num_tokens": 253880184.0, + "step": 9884 + }, + { + "epoch": 1.085547990336042, + "grad_norm": 2.1107747554779053, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6983112692832947, + "num_tokens": 253910043.0, + "step": 9885 + }, + { + "epoch": 1.0856578080386559, + "grad_norm": 2.183729887008667, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7136437296867371, + "num_tokens": 253936723.0, + "step": 9886 + }, + { + "epoch": 1.0857676257412694, + "grad_norm": 2.3399622440338135, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7432907819747925, + "num_tokens": 253959518.0, + "step": 9887 + }, + { + "epoch": 1.0858774434438832, + "grad_norm": 2.2750484943389893, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7113043069839478, + "num_tokens": 253984358.0, + "step": 9888 + }, + { + "epoch": 1.085987261146497, + "grad_norm": 2.452643394470215, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7410590648651123, + "num_tokens": 254006696.0, + "step": 9889 + }, + { + "epoch": 1.0860970788491104, + "grad_norm": 2.2396185398101807, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7273827791213989, + "num_tokens": 254031795.0, + "step": 9890 + }, + { + "epoch": 1.0862068965517242, + "grad_norm": 2.321730136871338, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7020912170410156, + "num_tokens": 254055229.0, + "step": 9891 + }, + { + "epoch": 1.0863167142543377, + "grad_norm": 2.0777041912078857, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6983857154846191, + "num_tokens": 254085660.0, + "step": 9892 + }, + { + "epoch": 1.0864265319569515, + "grad_norm": 2.186781406402588, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7071089148521423, + "num_tokens": 254113028.0, + "step": 9893 + }, + { + "epoch": 1.086536349659565, + "grad_norm": 2.279310464859009, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7278758883476257, + "num_tokens": 254137195.0, + "step": 9894 + }, + { + "epoch": 1.0866461673621788, + "grad_norm": 3.0145838260650635, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7415660619735718, + "num_tokens": 254153491.0, + "step": 9895 + }, + { + "epoch": 1.0867559850647925, + "grad_norm": 2.1718227863311768, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7267522811889648, + "num_tokens": 254180121.0, + "step": 9896 + }, + { + "epoch": 1.086865802767406, + "grad_norm": 2.1385366916656494, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.698823094367981, + "num_tokens": 254206753.0, + "step": 9897 + }, + { + "epoch": 1.0869756204700198, + "grad_norm": 2.1200480461120605, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7254767417907715, + "num_tokens": 254232265.0, + "step": 9898 + }, + { + "epoch": 1.0870854381726334, + "grad_norm": 2.053417205810547, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7076592445373535, + "num_tokens": 254263853.0, + "step": 9899 + }, + { + "epoch": 1.0871952558752471, + "grad_norm": 2.439891815185547, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7353129982948303, + "num_tokens": 254285700.0, + "step": 9900 + }, + { + "epoch": 1.0873050735778607, + "grad_norm": 2.181516647338867, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.711156964302063, + "num_tokens": 254314865.0, + "step": 9901 + }, + { + "epoch": 1.0874148912804744, + "grad_norm": 2.296923875808716, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.72116619348526, + "num_tokens": 254340464.0, + "step": 9902 + }, + { + "epoch": 1.0875247089830882, + "grad_norm": 2.225712776184082, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7270572781562805, + "num_tokens": 254366294.0, + "step": 9903 + }, + { + "epoch": 1.0876345266857017, + "grad_norm": 2.146322250366211, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7179803252220154, + "num_tokens": 254395860.0, + "step": 9904 + }, + { + "epoch": 1.0877443443883155, + "grad_norm": 2.156078815460205, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7172447443008423, + "num_tokens": 254423086.0, + "step": 9905 + }, + { + "epoch": 1.087854162090929, + "grad_norm": 2.2622575759887695, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7264506816864014, + "num_tokens": 254448927.0, + "step": 9906 + }, + { + "epoch": 1.0879639797935428, + "grad_norm": 2.305006265640259, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7250983715057373, + "num_tokens": 254473248.0, + "step": 9907 + }, + { + "epoch": 1.0880737974961563, + "grad_norm": 1.9634332656860352, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7264377474784851, + "num_tokens": 254505981.0, + "step": 9908 + }, + { + "epoch": 1.08818361519877, + "grad_norm": 2.080348491668701, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7367616295814514, + "num_tokens": 254534640.0, + "step": 9909 + }, + { + "epoch": 1.0882934329013838, + "grad_norm": 2.22745680809021, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7041335105895996, + "num_tokens": 254561690.0, + "step": 9910 + }, + { + "epoch": 1.0884032506039973, + "grad_norm": 2.0994668006896973, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7294833660125732, + "num_tokens": 254589593.0, + "step": 9911 + }, + { + "epoch": 1.088513068306611, + "grad_norm": 2.4174864292144775, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7292680740356445, + "num_tokens": 254611146.0, + "step": 9912 + }, + { + "epoch": 1.0886228860092246, + "grad_norm": 2.306133985519409, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7245834469795227, + "num_tokens": 254633995.0, + "step": 9913 + }, + { + "epoch": 1.0887327037118384, + "grad_norm": 2.117239475250244, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7165659070014954, + "num_tokens": 254663262.0, + "step": 9914 + }, + { + "epoch": 1.088842521414452, + "grad_norm": 2.203850746154785, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7157531976699829, + "num_tokens": 254689870.0, + "step": 9915 + }, + { + "epoch": 1.0889523391170657, + "grad_norm": 2.139397382736206, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7114659547805786, + "num_tokens": 254721232.0, + "step": 9916 + }, + { + "epoch": 1.0890621568196792, + "grad_norm": 2.1825027465820312, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7227143049240112, + "num_tokens": 254747001.0, + "step": 9917 + }, + { + "epoch": 1.089171974522293, + "grad_norm": 2.55074405670166, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.719663679599762, + "num_tokens": 254767493.0, + "step": 9918 + }, + { + "epoch": 1.0892817922249067, + "grad_norm": 2.293052911758423, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7022600173950195, + "num_tokens": 254794077.0, + "step": 9919 + }, + { + "epoch": 1.0893916099275203, + "grad_norm": 2.2573294639587402, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7101002335548401, + "num_tokens": 254819739.0, + "step": 9920 + }, + { + "epoch": 1.089501427630134, + "grad_norm": 2.1859946250915527, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.710136890411377, + "num_tokens": 254846593.0, + "step": 9921 + }, + { + "epoch": 1.0896112453327476, + "grad_norm": 1.985107421875, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7019202709197998, + "num_tokens": 254879162.0, + "step": 9922 + }, + { + "epoch": 1.0897210630353613, + "grad_norm": 2.2792811393737793, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7420179843902588, + "num_tokens": 254902644.0, + "step": 9923 + }, + { + "epoch": 1.089830880737975, + "grad_norm": 2.178189277648926, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7199865579605103, + "num_tokens": 254928274.0, + "step": 9924 + }, + { + "epoch": 1.0899406984405886, + "grad_norm": 2.4442074298858643, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7109622359275818, + "num_tokens": 254952039.0, + "step": 9925 + }, + { + "epoch": 1.0900505161432024, + "grad_norm": 2.1193721294403076, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.713524341583252, + "num_tokens": 254980770.0, + "step": 9926 + }, + { + "epoch": 1.090160333845816, + "grad_norm": 2.117093324661255, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6779869794845581, + "num_tokens": 255012229.0, + "step": 9927 + }, + { + "epoch": 1.0902701515484297, + "grad_norm": 2.206178903579712, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6983739733695984, + "num_tokens": 255039300.0, + "step": 9928 + }, + { + "epoch": 1.0903799692510432, + "grad_norm": 2.36332631111145, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.741466760635376, + "num_tokens": 255061576.0, + "step": 9929 + }, + { + "epoch": 1.090489786953657, + "grad_norm": 2.203313112258911, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7157631516456604, + "num_tokens": 255089734.0, + "step": 9930 + }, + { + "epoch": 1.0905996046562705, + "grad_norm": 2.081927537918091, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7197984457015991, + "num_tokens": 255116169.0, + "step": 9931 + }, + { + "epoch": 1.0907094223588842, + "grad_norm": 2.2845757007598877, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7044429779052734, + "num_tokens": 255143133.0, + "step": 9932 + }, + { + "epoch": 1.090819240061498, + "grad_norm": 2.1823649406433105, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7094035148620605, + "num_tokens": 255171337.0, + "step": 9933 + }, + { + "epoch": 1.0909290577641115, + "grad_norm": 2.1001603603363037, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7387025952339172, + "num_tokens": 255196883.0, + "step": 9934 + }, + { + "epoch": 1.0910388754667253, + "grad_norm": 2.041318893432617, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6945362091064453, + "num_tokens": 255227249.0, + "step": 9935 + }, + { + "epoch": 1.0911486931693388, + "grad_norm": 2.33866548538208, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7058925628662109, + "num_tokens": 255250782.0, + "step": 9936 + }, + { + "epoch": 1.0912585108719526, + "grad_norm": 2.244394540786743, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7287524342536926, + "num_tokens": 255277118.0, + "step": 9937 + }, + { + "epoch": 1.0913683285745663, + "grad_norm": 2.271752119064331, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7203232049942017, + "num_tokens": 255302531.0, + "step": 9938 + }, + { + "epoch": 1.0914781462771799, + "grad_norm": 2.276836633682251, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7143065929412842, + "num_tokens": 255328088.0, + "step": 9939 + }, + { + "epoch": 1.0915879639797936, + "grad_norm": 2.3330023288726807, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6924951672554016, + "num_tokens": 255353234.0, + "step": 9940 + }, + { + "epoch": 1.0916977816824072, + "grad_norm": 2.5379672050476074, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7153552770614624, + "num_tokens": 255374973.0, + "step": 9941 + }, + { + "epoch": 1.091807599385021, + "grad_norm": 2.071969747543335, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7054678201675415, + "num_tokens": 255403052.0, + "step": 9942 + }, + { + "epoch": 1.0919174170876345, + "grad_norm": 2.2997939586639404, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.720836877822876, + "num_tokens": 255428061.0, + "step": 9943 + }, + { + "epoch": 1.0920272347902482, + "grad_norm": 2.2508740425109863, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6955838799476624, + "num_tokens": 255455857.0, + "step": 9944 + }, + { + "epoch": 1.0921370524928617, + "grad_norm": 2.1879374980926514, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7153438925743103, + "num_tokens": 255480972.0, + "step": 9945 + }, + { + "epoch": 1.0922468701954755, + "grad_norm": 2.2078616619110107, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7222517728805542, + "num_tokens": 255507521.0, + "step": 9946 + }, + { + "epoch": 1.0923566878980893, + "grad_norm": 1.963999629020691, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7051447629928589, + "num_tokens": 255540011.0, + "step": 9947 + }, + { + "epoch": 1.0924665056007028, + "grad_norm": 2.196995973587036, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7154468297958374, + "num_tokens": 255567396.0, + "step": 9948 + }, + { + "epoch": 1.0925763233033166, + "grad_norm": 2.183093547821045, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7329397201538086, + "num_tokens": 255593143.0, + "step": 9949 + }, + { + "epoch": 1.09268614100593, + "grad_norm": 2.4815213680267334, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7244359850883484, + "num_tokens": 255614123.0, + "step": 9950 + }, + { + "epoch": 1.0927959587085438, + "grad_norm": 2.4859044551849365, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7246203422546387, + "num_tokens": 255635202.0, + "step": 9951 + }, + { + "epoch": 1.0929057764111574, + "grad_norm": 2.2665040493011475, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7250170111656189, + "num_tokens": 255658967.0, + "step": 9952 + }, + { + "epoch": 1.0930155941137711, + "grad_norm": 2.3670339584350586, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7197549343109131, + "num_tokens": 255682469.0, + "step": 9953 + }, + { + "epoch": 1.093125411816385, + "grad_norm": 2.3320839405059814, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7110313177108765, + "num_tokens": 255706368.0, + "step": 9954 + }, + { + "epoch": 1.0932352295189984, + "grad_norm": 2.702897548675537, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7443840503692627, + "num_tokens": 255723811.0, + "step": 9955 + }, + { + "epoch": 1.0933450472216122, + "grad_norm": 2.4495902061462402, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7134295701980591, + "num_tokens": 255745825.0, + "step": 9956 + }, + { + "epoch": 1.0934548649242257, + "grad_norm": 2.1659653186798096, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7091366648674011, + "num_tokens": 255774460.0, + "step": 9957 + }, + { + "epoch": 1.0935646826268395, + "grad_norm": 2.1461687088012695, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7196164131164551, + "num_tokens": 255801723.0, + "step": 9958 + }, + { + "epoch": 1.093674500329453, + "grad_norm": 2.0279581546783447, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7191718816757202, + "num_tokens": 255833219.0, + "step": 9959 + }, + { + "epoch": 1.0937843180320668, + "grad_norm": 2.4476826190948486, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7283113598823547, + "num_tokens": 255855663.0, + "step": 9960 + }, + { + "epoch": 1.0938941357346805, + "grad_norm": 2.1771514415740967, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7137424945831299, + "num_tokens": 255884121.0, + "step": 9961 + }, + { + "epoch": 1.094003953437294, + "grad_norm": 2.370649576187134, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.730859637260437, + "num_tokens": 255907287.0, + "step": 9962 + }, + { + "epoch": 1.0941137711399078, + "grad_norm": 2.240778684616089, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.726723313331604, + "num_tokens": 255931982.0, + "step": 9963 + }, + { + "epoch": 1.0942235888425214, + "grad_norm": 2.3852269649505615, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7237609028816223, + "num_tokens": 255955610.0, + "step": 9964 + }, + { + "epoch": 1.0943334065451351, + "grad_norm": 1.9910366535186768, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7088264226913452, + "num_tokens": 255985833.0, + "step": 9965 + }, + { + "epoch": 1.0944432242477486, + "grad_norm": 2.520749568939209, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7171456813812256, + "num_tokens": 256007494.0, + "step": 9966 + }, + { + "epoch": 1.0945530419503624, + "grad_norm": 2.186954975128174, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7129985094070435, + "num_tokens": 256034049.0, + "step": 9967 + }, + { + "epoch": 1.0946628596529762, + "grad_norm": 2.0328431129455566, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7135015726089478, + "num_tokens": 256063683.0, + "step": 9968 + }, + { + "epoch": 1.0947726773555897, + "grad_norm": 2.3310601711273193, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7247816324234009, + "num_tokens": 256088145.0, + "step": 9969 + }, + { + "epoch": 1.0948824950582035, + "grad_norm": 2.307133197784424, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7214613556861877, + "num_tokens": 256110863.0, + "step": 9970 + }, + { + "epoch": 1.094992312760817, + "grad_norm": 2.329632520675659, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7237473130226135, + "num_tokens": 256136757.0, + "step": 9971 + }, + { + "epoch": 1.0951021304634307, + "grad_norm": 2.5637712478637695, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7314227819442749, + "num_tokens": 256156777.0, + "step": 9972 + }, + { + "epoch": 1.0952119481660443, + "grad_norm": 2.330077648162842, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.723252534866333, + "num_tokens": 256180735.0, + "step": 9973 + }, + { + "epoch": 1.095321765868658, + "grad_norm": 2.269383668899536, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7157547473907471, + "num_tokens": 256205953.0, + "step": 9974 + }, + { + "epoch": 1.0954315835712718, + "grad_norm": 2.3118813037872314, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6964738368988037, + "num_tokens": 256233651.0, + "step": 9975 + }, + { + "epoch": 1.0955414012738853, + "grad_norm": 2.1793179512023926, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7159256339073181, + "num_tokens": 256262444.0, + "step": 9976 + }, + { + "epoch": 1.095651218976499, + "grad_norm": 2.2414438724517822, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7152135372161865, + "num_tokens": 256289708.0, + "step": 9977 + }, + { + "epoch": 1.0957610366791126, + "grad_norm": 2.4565393924713135, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7201952934265137, + "num_tokens": 256312182.0, + "step": 9978 + }, + { + "epoch": 1.0958708543817264, + "grad_norm": 2.272608757019043, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7419284582138062, + "num_tokens": 256337657.0, + "step": 9979 + }, + { + "epoch": 1.09598067208434, + "grad_norm": 2.161609649658203, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7022119760513306, + "num_tokens": 256366681.0, + "step": 9980 + }, + { + "epoch": 1.0960904897869537, + "grad_norm": 2.009770393371582, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.707303524017334, + "num_tokens": 256397141.0, + "step": 9981 + }, + { + "epoch": 1.0962003074895672, + "grad_norm": 2.2629475593566895, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.73496413230896, + "num_tokens": 256421154.0, + "step": 9982 + }, + { + "epoch": 1.096310125192181, + "grad_norm": 2.295722723007202, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7262003421783447, + "num_tokens": 256444183.0, + "step": 9983 + }, + { + "epoch": 1.0964199428947947, + "grad_norm": 2.00985050201416, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7115309238433838, + "num_tokens": 256476210.0, + "step": 9984 + }, + { + "epoch": 1.0965297605974083, + "grad_norm": 2.101553440093994, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7058116793632507, + "num_tokens": 256507750.0, + "step": 9985 + }, + { + "epoch": 1.096639578300022, + "grad_norm": 2.127394914627075, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7240317463874817, + "num_tokens": 256537414.0, + "step": 9986 + }, + { + "epoch": 1.0967493960026355, + "grad_norm": 2.4401798248291016, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.706623911857605, + "num_tokens": 256561164.0, + "step": 9987 + }, + { + "epoch": 1.0968592137052493, + "grad_norm": 1.9145210981369019, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7213014364242554, + "num_tokens": 256591884.0, + "step": 9988 + }, + { + "epoch": 1.096969031407863, + "grad_norm": 2.4321401119232178, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.718517541885376, + "num_tokens": 256614923.0, + "step": 9989 + }, + { + "epoch": 1.0970788491104766, + "grad_norm": 2.451726198196411, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.707351565361023, + "num_tokens": 256639246.0, + "step": 9990 + }, + { + "epoch": 1.0971886668130904, + "grad_norm": 2.402799129486084, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7089229822158813, + "num_tokens": 256664759.0, + "step": 9991 + }, + { + "epoch": 1.0972984845157039, + "grad_norm": 2.1151676177978516, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.747802197933197, + "num_tokens": 256692980.0, + "step": 9992 + }, + { + "epoch": 1.0974083022183176, + "grad_norm": 2.6311519145965576, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7113000154495239, + "num_tokens": 256714181.0, + "step": 9993 + }, + { + "epoch": 1.0975181199209312, + "grad_norm": 2.106903314590454, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7218124270439148, + "num_tokens": 256741379.0, + "step": 9994 + }, + { + "epoch": 1.097627937623545, + "grad_norm": 1.9403401613235474, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7273285388946533, + "num_tokens": 256773212.0, + "step": 9995 + }, + { + "epoch": 1.0977377553261585, + "grad_norm": 2.5768377780914307, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7370986938476562, + "num_tokens": 256792438.0, + "step": 9996 + }, + { + "epoch": 1.0978475730287722, + "grad_norm": 2.7071480751037598, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7231481075286865, + "num_tokens": 256813510.0, + "step": 9997 + }, + { + "epoch": 1.097957390731386, + "grad_norm": 2.213259220123291, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7161104679107666, + "num_tokens": 256839848.0, + "step": 9998 + }, + { + "epoch": 1.0980672084339995, + "grad_norm": 2.085747718811035, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7174234986305237, + "num_tokens": 256869479.0, + "step": 9999 + }, + { + "epoch": 1.0981770261366133, + "grad_norm": 2.143308401107788, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7321674823760986, + "num_tokens": 256896926.0, + "step": 10000 + }, + { + "epoch": 1.0982868438392268, + "grad_norm": 2.449030876159668, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7227247953414917, + "num_tokens": 256919109.0, + "step": 10001 + }, + { + "epoch": 1.0983966615418406, + "grad_norm": 2.3388376235961914, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6951320767402649, + "num_tokens": 256945926.0, + "step": 10002 + }, + { + "epoch": 1.098506479244454, + "grad_norm": 1.7384345531463623, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7229644060134888, + "num_tokens": 256985219.0, + "step": 10003 + }, + { + "epoch": 1.0986162969470679, + "grad_norm": 2.534470796585083, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7225184440612793, + "num_tokens": 257005902.0, + "step": 10004 + }, + { + "epoch": 1.0987261146496816, + "grad_norm": 2.1375346183776855, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7385749220848083, + "num_tokens": 257031365.0, + "step": 10005 + }, + { + "epoch": 1.0988359323522952, + "grad_norm": 2.5416953563690186, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7417764067649841, + "num_tokens": 257050270.0, + "step": 10006 + }, + { + "epoch": 1.098945750054909, + "grad_norm": 2.0656416416168213, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.714910626411438, + "num_tokens": 257078826.0, + "step": 10007 + }, + { + "epoch": 1.0990555677575224, + "grad_norm": 2.380192518234253, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7360918521881104, + "num_tokens": 257101757.0, + "step": 10008 + }, + { + "epoch": 1.0991653854601362, + "grad_norm": 2.385559320449829, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7111361026763916, + "num_tokens": 257124350.0, + "step": 10009 + }, + { + "epoch": 1.0992752031627497, + "grad_norm": 2.3316824436187744, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7049965858459473, + "num_tokens": 257151754.0, + "step": 10010 + }, + { + "epoch": 1.0993850208653635, + "grad_norm": 2.1738576889038086, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6852428317070007, + "num_tokens": 257183511.0, + "step": 10011 + }, + { + "epoch": 1.0994948385679773, + "grad_norm": 2.466552972793579, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7076016664505005, + "num_tokens": 257205960.0, + "step": 10012 + }, + { + "epoch": 1.0996046562705908, + "grad_norm": 1.9795221090316772, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7165566086769104, + "num_tokens": 257235307.0, + "step": 10013 + }, + { + "epoch": 1.0997144739732045, + "grad_norm": 2.246993064880371, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7139865756034851, + "num_tokens": 257261065.0, + "step": 10014 + }, + { + "epoch": 1.099824291675818, + "grad_norm": 2.6240997314453125, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7401724457740784, + "num_tokens": 257279781.0, + "step": 10015 + }, + { + "epoch": 1.0999341093784318, + "grad_norm": 2.2406883239746094, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7013593912124634, + "num_tokens": 257307211.0, + "step": 10016 + }, + { + "epoch": 1.1000439270810454, + "grad_norm": 2.3310749530792236, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7020884156227112, + "num_tokens": 257333490.0, + "step": 10017 + }, + { + "epoch": 1.1001537447836591, + "grad_norm": 2.2082371711730957, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7222375869750977, + "num_tokens": 257359899.0, + "step": 10018 + }, + { + "epoch": 1.1002635624862729, + "grad_norm": 2.440073251724243, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7292543649673462, + "num_tokens": 257381606.0, + "step": 10019 + }, + { + "epoch": 1.1003733801888864, + "grad_norm": 2.260385274887085, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7208189964294434, + "num_tokens": 257407575.0, + "step": 10020 + }, + { + "epoch": 1.1004831978915002, + "grad_norm": 2.4697976112365723, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7121754884719849, + "num_tokens": 257432344.0, + "step": 10021 + }, + { + "epoch": 1.1005930155941137, + "grad_norm": 2.042773485183716, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7237677574157715, + "num_tokens": 257461430.0, + "step": 10022 + }, + { + "epoch": 1.1007028332967275, + "grad_norm": 2.163318157196045, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7055556774139404, + "num_tokens": 257487880.0, + "step": 10023 + }, + { + "epoch": 1.100812650999341, + "grad_norm": 2.333510398864746, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7177394032478333, + "num_tokens": 257510677.0, + "step": 10024 + }, + { + "epoch": 1.1009224687019548, + "grad_norm": 2.6405839920043945, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7302495837211609, + "num_tokens": 257530548.0, + "step": 10025 + }, + { + "epoch": 1.1010322864045685, + "grad_norm": 2.530261278152466, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7172942161560059, + "num_tokens": 257551823.0, + "step": 10026 + }, + { + "epoch": 1.101142104107182, + "grad_norm": 2.344177007675171, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.723395824432373, + "num_tokens": 257575603.0, + "step": 10027 + }, + { + "epoch": 1.1012519218097958, + "grad_norm": 2.6345529556274414, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7139886617660522, + "num_tokens": 257595803.0, + "step": 10028 + }, + { + "epoch": 1.1013617395124093, + "grad_norm": 2.1716341972351074, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.720283567905426, + "num_tokens": 257621949.0, + "step": 10029 + }, + { + "epoch": 1.101471557215023, + "grad_norm": 2.184091806411743, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.7020355463027954, + "num_tokens": 257654583.0, + "step": 10030 + }, + { + "epoch": 1.1015813749176366, + "grad_norm": 1.8708690404891968, + "learning_rate": 1e-06, + "loss": 1.1148, + "mean_token_accuracy": 0.6639820337295532, + "num_tokens": 257689963.0, + "step": 10031 + }, + { + "epoch": 1.1016911926202504, + "grad_norm": 2.1716063022613525, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7164144515991211, + "num_tokens": 257716119.0, + "step": 10032 + }, + { + "epoch": 1.1018010103228641, + "grad_norm": 2.4393558502197266, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7179020643234253, + "num_tokens": 257738095.0, + "step": 10033 + }, + { + "epoch": 1.1019108280254777, + "grad_norm": 2.2530486583709717, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7068537473678589, + "num_tokens": 257764257.0, + "step": 10034 + }, + { + "epoch": 1.1020206457280914, + "grad_norm": 1.9726322889328003, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7080777883529663, + "num_tokens": 257795882.0, + "step": 10035 + }, + { + "epoch": 1.102130463430705, + "grad_norm": 2.7333595752716064, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7193027138710022, + "num_tokens": 257815158.0, + "step": 10036 + }, + { + "epoch": 1.1022402811333187, + "grad_norm": 2.3436334133148193, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7190136909484863, + "num_tokens": 257838045.0, + "step": 10037 + }, + { + "epoch": 1.1023500988359323, + "grad_norm": 2.385361671447754, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7327735424041748, + "num_tokens": 257860795.0, + "step": 10038 + }, + { + "epoch": 1.102459916538546, + "grad_norm": 2.0478906631469727, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7318166494369507, + "num_tokens": 257888100.0, + "step": 10039 + }, + { + "epoch": 1.1025697342411598, + "grad_norm": 2.252410650253296, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7179870009422302, + "num_tokens": 257914262.0, + "step": 10040 + }, + { + "epoch": 1.1026795519437733, + "grad_norm": 2.1696131229400635, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7106471061706543, + "num_tokens": 257942779.0, + "step": 10041 + }, + { + "epoch": 1.102789369646387, + "grad_norm": 2.1959080696105957, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7237008213996887, + "num_tokens": 257967063.0, + "step": 10042 + }, + { + "epoch": 1.1028991873490006, + "grad_norm": 2.4048943519592285, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7246167659759521, + "num_tokens": 257990654.0, + "step": 10043 + }, + { + "epoch": 1.1030090050516144, + "grad_norm": 2.138974189758301, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7593096494674683, + "num_tokens": 258014944.0, + "step": 10044 + }, + { + "epoch": 1.103118822754228, + "grad_norm": 2.33211088180542, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7299532890319824, + "num_tokens": 258037709.0, + "step": 10045 + }, + { + "epoch": 1.1032286404568417, + "grad_norm": 2.3365821838378906, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7318058013916016, + "num_tokens": 258062520.0, + "step": 10046 + }, + { + "epoch": 1.1033384581594552, + "grad_norm": 2.365748882293701, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7181864380836487, + "num_tokens": 258087357.0, + "step": 10047 + }, + { + "epoch": 1.103448275862069, + "grad_norm": 2.305910348892212, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7029606699943542, + "num_tokens": 258114496.0, + "step": 10048 + }, + { + "epoch": 1.1035580935646827, + "grad_norm": 2.2048699855804443, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7063790559768677, + "num_tokens": 258141921.0, + "step": 10049 + }, + { + "epoch": 1.1036679112672962, + "grad_norm": 2.5399956703186035, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.751245379447937, + "num_tokens": 258162553.0, + "step": 10050 + }, + { + "epoch": 1.10377772896991, + "grad_norm": 2.3051159381866455, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7316925525665283, + "num_tokens": 258185889.0, + "step": 10051 + }, + { + "epoch": 1.1038875466725235, + "grad_norm": 2.3007752895355225, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.707929253578186, + "num_tokens": 258210231.0, + "step": 10052 + }, + { + "epoch": 1.1039973643751373, + "grad_norm": 2.5191895961761475, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7078672051429749, + "num_tokens": 258231538.0, + "step": 10053 + }, + { + "epoch": 1.104107182077751, + "grad_norm": 1.9993854761123657, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7228813767433167, + "num_tokens": 258261781.0, + "step": 10054 + }, + { + "epoch": 1.1042169997803646, + "grad_norm": 2.1530964374542236, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7130978107452393, + "num_tokens": 258291020.0, + "step": 10055 + }, + { + "epoch": 1.1043268174829783, + "grad_norm": 2.1449899673461914, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7190262675285339, + "num_tokens": 258317636.0, + "step": 10056 + }, + { + "epoch": 1.1044366351855919, + "grad_norm": 2.1998610496520996, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.7015825510025024, + "num_tokens": 258345710.0, + "step": 10057 + }, + { + "epoch": 1.1045464528882056, + "grad_norm": 2.2456936836242676, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7108544707298279, + "num_tokens": 258372076.0, + "step": 10058 + }, + { + "epoch": 1.1046562705908192, + "grad_norm": 2.0136466026306152, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7094643115997314, + "num_tokens": 258403166.0, + "step": 10059 + }, + { + "epoch": 1.104766088293433, + "grad_norm": 2.352808713912964, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7292490005493164, + "num_tokens": 258424716.0, + "step": 10060 + }, + { + "epoch": 1.1048759059960465, + "grad_norm": 2.2870028018951416, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7194190621376038, + "num_tokens": 258449276.0, + "step": 10061 + }, + { + "epoch": 1.1049857236986602, + "grad_norm": 2.2687628269195557, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7205449938774109, + "num_tokens": 258472714.0, + "step": 10062 + }, + { + "epoch": 1.105095541401274, + "grad_norm": 2.0382649898529053, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7068870067596436, + "num_tokens": 258503341.0, + "step": 10063 + }, + { + "epoch": 1.1052053591038875, + "grad_norm": 2.297816276550293, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.725538969039917, + "num_tokens": 258529036.0, + "step": 10064 + }, + { + "epoch": 1.1053151768065013, + "grad_norm": 2.4409079551696777, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7465580105781555, + "num_tokens": 258549952.0, + "step": 10065 + }, + { + "epoch": 1.1054249945091148, + "grad_norm": 2.426691770553589, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.714611828327179, + "num_tokens": 258572096.0, + "step": 10066 + }, + { + "epoch": 1.1055348122117286, + "grad_norm": 2.269716262817383, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7236241698265076, + "num_tokens": 258594640.0, + "step": 10067 + }, + { + "epoch": 1.105644629914342, + "grad_norm": 1.9786465167999268, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7251182198524475, + "num_tokens": 258627061.0, + "step": 10068 + }, + { + "epoch": 1.1057544476169558, + "grad_norm": 2.1383323669433594, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7052896618843079, + "num_tokens": 258654597.0, + "step": 10069 + }, + { + "epoch": 1.1058642653195696, + "grad_norm": 2.13946270942688, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7211065292358398, + "num_tokens": 258682290.0, + "step": 10070 + }, + { + "epoch": 1.1059740830221831, + "grad_norm": 2.095567226409912, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.716358482837677, + "num_tokens": 258710586.0, + "step": 10071 + }, + { + "epoch": 1.106083900724797, + "grad_norm": 2.159113645553589, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7176492810249329, + "num_tokens": 258736453.0, + "step": 10072 + }, + { + "epoch": 1.1061937184274104, + "grad_norm": 2.0018510818481445, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7294808626174927, + "num_tokens": 258767794.0, + "step": 10073 + }, + { + "epoch": 1.1063035361300242, + "grad_norm": 2.0636208057403564, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.725568413734436, + "num_tokens": 258798105.0, + "step": 10074 + }, + { + "epoch": 1.1064133538326377, + "grad_norm": 2.1102476119995117, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.712283730506897, + "num_tokens": 258826967.0, + "step": 10075 + }, + { + "epoch": 1.1065231715352515, + "grad_norm": 2.466214895248413, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7231614589691162, + "num_tokens": 258847564.0, + "step": 10076 + }, + { + "epoch": 1.1066329892378652, + "grad_norm": 2.1808369159698486, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.714820146560669, + "num_tokens": 258875353.0, + "step": 10077 + }, + { + "epoch": 1.1067428069404788, + "grad_norm": 2.3862202167510986, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7337132692337036, + "num_tokens": 258897157.0, + "step": 10078 + }, + { + "epoch": 1.1068526246430925, + "grad_norm": 2.255295515060425, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7254958748817444, + "num_tokens": 258922103.0, + "step": 10079 + }, + { + "epoch": 1.106962442345706, + "grad_norm": 2.6004018783569336, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.711220920085907, + "num_tokens": 258941962.0, + "step": 10080 + }, + { + "epoch": 1.1070722600483198, + "grad_norm": 2.4726507663726807, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7305506467819214, + "num_tokens": 258963161.0, + "step": 10081 + }, + { + "epoch": 1.1071820777509334, + "grad_norm": 2.5137388706207275, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7305065393447876, + "num_tokens": 258984326.0, + "step": 10082 + }, + { + "epoch": 1.1072918954535471, + "grad_norm": 2.2325217723846436, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7243565320968628, + "num_tokens": 259008449.0, + "step": 10083 + }, + { + "epoch": 1.1074017131561609, + "grad_norm": 2.2840874195098877, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7381329536437988, + "num_tokens": 259032538.0, + "step": 10084 + }, + { + "epoch": 1.1075115308587744, + "grad_norm": 2.327038288116455, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7389712333679199, + "num_tokens": 259055655.0, + "step": 10085 + }, + { + "epoch": 1.1076213485613882, + "grad_norm": 2.265261173248291, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7008641958236694, + "num_tokens": 259082733.0, + "step": 10086 + }, + { + "epoch": 1.1077311662640017, + "grad_norm": 2.279313087463379, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7194545269012451, + "num_tokens": 259107451.0, + "step": 10087 + }, + { + "epoch": 1.1078409839666155, + "grad_norm": 2.141660451889038, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7397795915603638, + "num_tokens": 259133000.0, + "step": 10088 + }, + { + "epoch": 1.107950801669229, + "grad_norm": 2.161100387573242, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7191674709320068, + "num_tokens": 259157585.0, + "step": 10089 + }, + { + "epoch": 1.1080606193718427, + "grad_norm": 2.537987232208252, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7236573100090027, + "num_tokens": 259178875.0, + "step": 10090 + }, + { + "epoch": 1.1081704370744565, + "grad_norm": 2.6022632122039795, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7495365142822266, + "num_tokens": 259196966.0, + "step": 10091 + }, + { + "epoch": 1.10828025477707, + "grad_norm": 2.3826780319213867, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7387598752975464, + "num_tokens": 259218790.0, + "step": 10092 + }, + { + "epoch": 1.1083900724796838, + "grad_norm": 1.9754209518432617, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.725475013256073, + "num_tokens": 259246365.0, + "step": 10093 + }, + { + "epoch": 1.1084998901822973, + "grad_norm": 2.0959484577178955, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7014392018318176, + "num_tokens": 259275959.0, + "step": 10094 + }, + { + "epoch": 1.108609707884911, + "grad_norm": 2.3570597171783447, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.694190263748169, + "num_tokens": 259302671.0, + "step": 10095 + }, + { + "epoch": 1.1087195255875246, + "grad_norm": 2.1989457607269287, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7185173630714417, + "num_tokens": 259328809.0, + "step": 10096 + }, + { + "epoch": 1.1088293432901384, + "grad_norm": 2.3480608463287354, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7344244718551636, + "num_tokens": 259351044.0, + "step": 10097 + }, + { + "epoch": 1.108939160992752, + "grad_norm": 2.1857190132141113, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6998052000999451, + "num_tokens": 259379783.0, + "step": 10098 + }, + { + "epoch": 1.1090489786953657, + "grad_norm": 2.2238407135009766, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7266371846199036, + "num_tokens": 259406135.0, + "step": 10099 + }, + { + "epoch": 1.1091587963979794, + "grad_norm": 2.2522051334381104, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7227873802185059, + "num_tokens": 259431178.0, + "step": 10100 + }, + { + "epoch": 1.109268614100593, + "grad_norm": 2.183135986328125, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7102667093276978, + "num_tokens": 259456660.0, + "step": 10101 + }, + { + "epoch": 1.1093784318032067, + "grad_norm": 2.9972381591796875, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6880501508712769, + "num_tokens": 259488511.0, + "step": 10102 + }, + { + "epoch": 1.1094882495058203, + "grad_norm": 2.229621648788452, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7080826759338379, + "num_tokens": 259515655.0, + "step": 10103 + }, + { + "epoch": 1.109598067208434, + "grad_norm": 2.355271577835083, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7468284964561462, + "num_tokens": 259538618.0, + "step": 10104 + }, + { + "epoch": 1.1097078849110478, + "grad_norm": 2.625221014022827, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7073202133178711, + "num_tokens": 259559919.0, + "step": 10105 + }, + { + "epoch": 1.1098177026136613, + "grad_norm": 2.1751363277435303, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.690902590751648, + "num_tokens": 259587904.0, + "step": 10106 + }, + { + "epoch": 1.109927520316275, + "grad_norm": 2.229529619216919, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7279439568519592, + "num_tokens": 259611934.0, + "step": 10107 + }, + { + "epoch": 1.1100373380188886, + "grad_norm": 2.296151638031006, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7142395973205566, + "num_tokens": 259636673.0, + "step": 10108 + }, + { + "epoch": 1.1101471557215024, + "grad_norm": 2.4237184524536133, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7066864967346191, + "num_tokens": 259665149.0, + "step": 10109 + }, + { + "epoch": 1.110256973424116, + "grad_norm": 2.1886396408081055, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7102986574172974, + "num_tokens": 259694185.0, + "step": 10110 + }, + { + "epoch": 1.1103667911267296, + "grad_norm": 2.6840317249298096, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7387470006942749, + "num_tokens": 259711942.0, + "step": 10111 + }, + { + "epoch": 1.1104766088293432, + "grad_norm": 2.213282346725464, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7296531200408936, + "num_tokens": 259738579.0, + "step": 10112 + }, + { + "epoch": 1.110586426531957, + "grad_norm": 2.4388420581817627, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.723159909248352, + "num_tokens": 259761128.0, + "step": 10113 + }, + { + "epoch": 1.1106962442345707, + "grad_norm": 2.257066488265991, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7129067182540894, + "num_tokens": 259787231.0, + "step": 10114 + }, + { + "epoch": 1.1108060619371842, + "grad_norm": 2.3980886936187744, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7279877662658691, + "num_tokens": 259810530.0, + "step": 10115 + }, + { + "epoch": 1.110915879639798, + "grad_norm": 2.4372098445892334, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7199616432189941, + "num_tokens": 259832434.0, + "step": 10116 + }, + { + "epoch": 1.1110256973424115, + "grad_norm": 2.6302571296691895, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7280917167663574, + "num_tokens": 259863228.0, + "step": 10117 + }, + { + "epoch": 1.1111355150450253, + "grad_norm": 2.2499592304229736, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.732319176197052, + "num_tokens": 259889318.0, + "step": 10118 + }, + { + "epoch": 1.111245332747639, + "grad_norm": 2.1225175857543945, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7263902425765991, + "num_tokens": 259916205.0, + "step": 10119 + }, + { + "epoch": 1.1113551504502526, + "grad_norm": 2.0993714332580566, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7355802059173584, + "num_tokens": 259945940.0, + "step": 10120 + }, + { + "epoch": 1.1114649681528663, + "grad_norm": 1.9149314165115356, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7118222713470459, + "num_tokens": 259978610.0, + "step": 10121 + }, + { + "epoch": 1.1115747858554799, + "grad_norm": 2.306481122970581, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6988672018051147, + "num_tokens": 260003920.0, + "step": 10122 + }, + { + "epoch": 1.1116846035580936, + "grad_norm": 2.3319289684295654, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7226539850234985, + "num_tokens": 260026990.0, + "step": 10123 + }, + { + "epoch": 1.1117944212607072, + "grad_norm": 2.182734251022339, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6958868503570557, + "num_tokens": 260054847.0, + "step": 10124 + }, + { + "epoch": 1.111904238963321, + "grad_norm": 1.8938274383544922, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7173747420310974, + "num_tokens": 260088127.0, + "step": 10125 + }, + { + "epoch": 1.1120140566659344, + "grad_norm": 2.036179304122925, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.722087025642395, + "num_tokens": 260116602.0, + "step": 10126 + }, + { + "epoch": 1.1121238743685482, + "grad_norm": 2.233455181121826, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7300428748130798, + "num_tokens": 260142142.0, + "step": 10127 + }, + { + "epoch": 1.112233692071162, + "grad_norm": 2.404777765274048, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.725530207157135, + "num_tokens": 260166415.0, + "step": 10128 + }, + { + "epoch": 1.1123435097737755, + "grad_norm": 2.276406764984131, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6949383020401001, + "num_tokens": 260195519.0, + "step": 10129 + }, + { + "epoch": 1.1124533274763893, + "grad_norm": 2.354550361633301, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7298693656921387, + "num_tokens": 260220066.0, + "step": 10130 + }, + { + "epoch": 1.1125631451790028, + "grad_norm": 2.3102738857269287, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7202259302139282, + "num_tokens": 260244372.0, + "step": 10131 + }, + { + "epoch": 1.1126729628816165, + "grad_norm": 2.347212553024292, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7336411476135254, + "num_tokens": 260267211.0, + "step": 10132 + }, + { + "epoch": 1.11278278058423, + "grad_norm": 2.4521186351776123, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7142885327339172, + "num_tokens": 260288409.0, + "step": 10133 + }, + { + "epoch": 1.1128925982868438, + "grad_norm": 2.4551196098327637, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.71281898021698, + "num_tokens": 260311288.0, + "step": 10134 + }, + { + "epoch": 1.1130024159894576, + "grad_norm": 2.3473360538482666, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7335982322692871, + "num_tokens": 260334766.0, + "step": 10135 + }, + { + "epoch": 1.1131122336920711, + "grad_norm": 2.2701339721679688, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7022461295127869, + "num_tokens": 260359657.0, + "step": 10136 + }, + { + "epoch": 1.1132220513946849, + "grad_norm": 2.461902618408203, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7237467765808105, + "num_tokens": 260381672.0, + "step": 10137 + }, + { + "epoch": 1.1133318690972984, + "grad_norm": 2.0954606533050537, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6872594952583313, + "num_tokens": 260409478.0, + "step": 10138 + }, + { + "epoch": 1.1134416867999122, + "grad_norm": 2.4964759349823, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7245492339134216, + "num_tokens": 260432400.0, + "step": 10139 + }, + { + "epoch": 1.1135515045025257, + "grad_norm": 2.1622183322906494, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7367416620254517, + "num_tokens": 260458396.0, + "step": 10140 + }, + { + "epoch": 1.1136613222051395, + "grad_norm": 2.679436683654785, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7271066904067993, + "num_tokens": 260479807.0, + "step": 10141 + }, + { + "epoch": 1.1137711399077532, + "grad_norm": 2.3922598361968994, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7384540438652039, + "num_tokens": 260502156.0, + "step": 10142 + }, + { + "epoch": 1.1138809576103668, + "grad_norm": 2.5642857551574707, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7186914086341858, + "num_tokens": 260524185.0, + "step": 10143 + }, + { + "epoch": 1.1139907753129805, + "grad_norm": 2.363018751144409, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7261214256286621, + "num_tokens": 260548456.0, + "step": 10144 + }, + { + "epoch": 1.114100593015594, + "grad_norm": 2.7496371269226074, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.748520016670227, + "num_tokens": 260565541.0, + "step": 10145 + }, + { + "epoch": 1.1142104107182078, + "grad_norm": 2.168696165084839, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7096577882766724, + "num_tokens": 260594194.0, + "step": 10146 + }, + { + "epoch": 1.1143202284208213, + "grad_norm": 2.1985764503479004, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7151174545288086, + "num_tokens": 260618222.0, + "step": 10147 + }, + { + "epoch": 1.114430046123435, + "grad_norm": 2.1403353214263916, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7182353734970093, + "num_tokens": 260646681.0, + "step": 10148 + }, + { + "epoch": 1.1145398638260489, + "grad_norm": 2.2450473308563232, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7188398241996765, + "num_tokens": 260671842.0, + "step": 10149 + }, + { + "epoch": 1.1146496815286624, + "grad_norm": 2.2919440269470215, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7362371683120728, + "num_tokens": 260695016.0, + "step": 10150 + }, + { + "epoch": 1.1147594992312762, + "grad_norm": 2.2540104389190674, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6911330223083496, + "num_tokens": 260721333.0, + "step": 10151 + }, + { + "epoch": 1.1148693169338897, + "grad_norm": 2.2750918865203857, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7168319225311279, + "num_tokens": 260748859.0, + "step": 10152 + }, + { + "epoch": 1.1149791346365034, + "grad_norm": 2.485619306564331, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7229372262954712, + "num_tokens": 260771865.0, + "step": 10153 + }, + { + "epoch": 1.115088952339117, + "grad_norm": 2.2116622924804688, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6902912855148315, + "num_tokens": 260800803.0, + "step": 10154 + }, + { + "epoch": 1.1151987700417307, + "grad_norm": 2.257192850112915, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7134062647819519, + "num_tokens": 260826504.0, + "step": 10155 + }, + { + "epoch": 1.1153085877443445, + "grad_norm": 2.265477418899536, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7043972015380859, + "num_tokens": 260854274.0, + "step": 10156 + }, + { + "epoch": 1.115418405446958, + "grad_norm": 2.390491247177124, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7277246713638306, + "num_tokens": 260876698.0, + "step": 10157 + }, + { + "epoch": 1.1155282231495718, + "grad_norm": 2.3881642818450928, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7327229976654053, + "num_tokens": 260899025.0, + "step": 10158 + }, + { + "epoch": 1.1156380408521853, + "grad_norm": 2.3784637451171875, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7213919162750244, + "num_tokens": 260922158.0, + "step": 10159 + }, + { + "epoch": 1.115747858554799, + "grad_norm": 2.047968864440918, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7110003232955933, + "num_tokens": 260950236.0, + "step": 10160 + }, + { + "epoch": 1.1158576762574126, + "grad_norm": 2.6403110027313232, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7152348756790161, + "num_tokens": 260971277.0, + "step": 10161 + }, + { + "epoch": 1.1159674939600264, + "grad_norm": 2.54219913482666, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7218376994132996, + "num_tokens": 260993092.0, + "step": 10162 + }, + { + "epoch": 1.11607731166264, + "grad_norm": 2.2109973430633545, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7017150521278381, + "num_tokens": 261020839.0, + "step": 10163 + }, + { + "epoch": 1.1161871293652537, + "grad_norm": 2.0226755142211914, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7110037803649902, + "num_tokens": 261051678.0, + "step": 10164 + }, + { + "epoch": 1.1162969470678674, + "grad_norm": 2.2154808044433594, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7141386270523071, + "num_tokens": 261078011.0, + "step": 10165 + }, + { + "epoch": 1.116406764770481, + "grad_norm": 2.1996512413024902, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7124576568603516, + "num_tokens": 261105496.0, + "step": 10166 + }, + { + "epoch": 1.1165165824730947, + "grad_norm": 2.3777072429656982, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.702814519405365, + "num_tokens": 261127349.0, + "step": 10167 + }, + { + "epoch": 1.1166264001757082, + "grad_norm": 2.3693056106567383, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7150591611862183, + "num_tokens": 261151889.0, + "step": 10168 + }, + { + "epoch": 1.116736217878322, + "grad_norm": 2.3242249488830566, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6864972114562988, + "num_tokens": 261179419.0, + "step": 10169 + }, + { + "epoch": 1.1168460355809358, + "grad_norm": 2.267951250076294, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7288790941238403, + "num_tokens": 261207039.0, + "step": 10170 + }, + { + "epoch": 1.1169558532835493, + "grad_norm": 2.166872262954712, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7106574773788452, + "num_tokens": 261235081.0, + "step": 10171 + }, + { + "epoch": 1.117065670986163, + "grad_norm": 2.6303038597106934, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7421450614929199, + "num_tokens": 261254981.0, + "step": 10172 + }, + { + "epoch": 1.1171754886887766, + "grad_norm": 2.5402019023895264, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7183898687362671, + "num_tokens": 261277453.0, + "step": 10173 + }, + { + "epoch": 1.1172853063913903, + "grad_norm": 2.3508944511413574, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7032203674316406, + "num_tokens": 261303232.0, + "step": 10174 + }, + { + "epoch": 1.1173951240940039, + "grad_norm": 2.300347328186035, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7133635878562927, + "num_tokens": 261330018.0, + "step": 10175 + }, + { + "epoch": 1.1175049417966176, + "grad_norm": 2.1225945949554443, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7315355539321899, + "num_tokens": 261357061.0, + "step": 10176 + }, + { + "epoch": 1.1176147594992312, + "grad_norm": 2.543478012084961, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7450351715087891, + "num_tokens": 261378626.0, + "step": 10177 + }, + { + "epoch": 1.117724577201845, + "grad_norm": 2.4771928787231445, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7332339882850647, + "num_tokens": 261401022.0, + "step": 10178 + }, + { + "epoch": 1.1178343949044587, + "grad_norm": 2.3949174880981445, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7393298149108887, + "num_tokens": 261425399.0, + "step": 10179 + }, + { + "epoch": 1.1179442126070722, + "grad_norm": 2.1827995777130127, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7023154497146606, + "num_tokens": 261454293.0, + "step": 10180 + }, + { + "epoch": 1.118054030309686, + "grad_norm": 2.379793643951416, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7068029642105103, + "num_tokens": 261480231.0, + "step": 10181 + }, + { + "epoch": 1.1181638480122995, + "grad_norm": 2.402681827545166, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7143080830574036, + "num_tokens": 261502987.0, + "step": 10182 + }, + { + "epoch": 1.1182736657149133, + "grad_norm": 2.0569891929626465, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.729149341583252, + "num_tokens": 261531866.0, + "step": 10183 + }, + { + "epoch": 1.1183834834175268, + "grad_norm": 2.3955018520355225, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7027131915092468, + "num_tokens": 261555161.0, + "step": 10184 + }, + { + "epoch": 1.1184933011201406, + "grad_norm": 2.2813711166381836, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7234704494476318, + "num_tokens": 261578726.0, + "step": 10185 + }, + { + "epoch": 1.1186031188227543, + "grad_norm": 2.3901255130767822, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.713982880115509, + "num_tokens": 261603040.0, + "step": 10186 + }, + { + "epoch": 1.1187129365253679, + "grad_norm": 2.3696460723876953, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6918723583221436, + "num_tokens": 261628792.0, + "step": 10187 + }, + { + "epoch": 1.1188227542279816, + "grad_norm": 2.6172823905944824, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7321767210960388, + "num_tokens": 261649191.0, + "step": 10188 + }, + { + "epoch": 1.1189325719305951, + "grad_norm": 2.0923736095428467, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7058246731758118, + "num_tokens": 261677501.0, + "step": 10189 + }, + { + "epoch": 1.119042389633209, + "grad_norm": 2.291004180908203, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7153021097183228, + "num_tokens": 261701376.0, + "step": 10190 + }, + { + "epoch": 1.1191522073358224, + "grad_norm": 2.0551199913024902, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7277869582176208, + "num_tokens": 261729242.0, + "step": 10191 + }, + { + "epoch": 1.1192620250384362, + "grad_norm": 2.3895506858825684, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7244830131530762, + "num_tokens": 261750326.0, + "step": 10192 + }, + { + "epoch": 1.11937184274105, + "grad_norm": 2.4530248641967773, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7369739413261414, + "num_tokens": 261772382.0, + "step": 10193 + }, + { + "epoch": 1.1194816604436635, + "grad_norm": 2.337827205657959, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7242449522018433, + "num_tokens": 261795490.0, + "step": 10194 + }, + { + "epoch": 1.1195914781462772, + "grad_norm": 2.147310495376587, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.697877824306488, + "num_tokens": 261824185.0, + "step": 10195 + }, + { + "epoch": 1.1197012958488908, + "grad_norm": 2.4021124839782715, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7170572280883789, + "num_tokens": 261846480.0, + "step": 10196 + }, + { + "epoch": 1.1198111135515045, + "grad_norm": 2.454867362976074, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.733360767364502, + "num_tokens": 261870224.0, + "step": 10197 + }, + { + "epoch": 1.119920931254118, + "grad_norm": 2.286237955093384, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7271193265914917, + "num_tokens": 261896600.0, + "step": 10198 + }, + { + "epoch": 1.1200307489567318, + "grad_norm": 2.1709554195404053, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7040920257568359, + "num_tokens": 261925222.0, + "step": 10199 + }, + { + "epoch": 1.1201405666593456, + "grad_norm": 2.475040912628174, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7402850985527039, + "num_tokens": 261945980.0, + "step": 10200 + }, + { + "epoch": 1.1202503843619591, + "grad_norm": 1.9607218503952026, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6833614110946655, + "num_tokens": 261979791.0, + "step": 10201 + }, + { + "epoch": 1.1203602020645729, + "grad_norm": 2.1734652519226074, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7065926194190979, + "num_tokens": 262008295.0, + "step": 10202 + }, + { + "epoch": 1.1204700197671864, + "grad_norm": 2.3869638442993164, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.708992063999176, + "num_tokens": 262036396.0, + "step": 10203 + }, + { + "epoch": 1.1205798374698002, + "grad_norm": 2.2854020595550537, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7243764996528625, + "num_tokens": 262062240.0, + "step": 10204 + }, + { + "epoch": 1.1206896551724137, + "grad_norm": 2.12188458442688, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6989277601242065, + "num_tokens": 262091492.0, + "step": 10205 + }, + { + "epoch": 1.1207994728750275, + "grad_norm": 2.1087558269500732, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7087751030921936, + "num_tokens": 262121191.0, + "step": 10206 + }, + { + "epoch": 1.1209092905776412, + "grad_norm": 2.549851179122925, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7429033517837524, + "num_tokens": 262141420.0, + "step": 10207 + }, + { + "epoch": 1.1210191082802548, + "grad_norm": 2.388294219970703, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7018555402755737, + "num_tokens": 262165362.0, + "step": 10208 + }, + { + "epoch": 1.1211289259828685, + "grad_norm": 2.263209581375122, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7092397212982178, + "num_tokens": 262193224.0, + "step": 10209 + }, + { + "epoch": 1.121238743685482, + "grad_norm": 2.298999071121216, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.714148759841919, + "num_tokens": 262218237.0, + "step": 10210 + }, + { + "epoch": 1.1213485613880958, + "grad_norm": 2.2534260749816895, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7202121019363403, + "num_tokens": 262244143.0, + "step": 10211 + }, + { + "epoch": 1.1214583790907093, + "grad_norm": 2.1120362281799316, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7092212438583374, + "num_tokens": 262272357.0, + "step": 10212 + }, + { + "epoch": 1.121568196793323, + "grad_norm": 2.323484182357788, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7315233945846558, + "num_tokens": 262294632.0, + "step": 10213 + }, + { + "epoch": 1.1216780144959368, + "grad_norm": 2.0362188816070557, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7102910280227661, + "num_tokens": 262324299.0, + "step": 10214 + }, + { + "epoch": 1.1217878321985504, + "grad_norm": 2.403820514678955, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.721071183681488, + "num_tokens": 262347197.0, + "step": 10215 + }, + { + "epoch": 1.1218976499011641, + "grad_norm": 2.222404956817627, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7000672817230225, + "num_tokens": 262372357.0, + "step": 10216 + }, + { + "epoch": 1.1220074676037777, + "grad_norm": 2.446930170059204, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7317432165145874, + "num_tokens": 262392870.0, + "step": 10217 + }, + { + "epoch": 1.1221172853063914, + "grad_norm": 2.161221504211426, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6938637495040894, + "num_tokens": 262424360.0, + "step": 10218 + }, + { + "epoch": 1.122227103009005, + "grad_norm": 2.370802164077759, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.733233630657196, + "num_tokens": 262447951.0, + "step": 10219 + }, + { + "epoch": 1.1223369207116187, + "grad_norm": 2.152864694595337, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7051712870597839, + "num_tokens": 262473926.0, + "step": 10220 + }, + { + "epoch": 1.1224467384142325, + "grad_norm": 2.142505645751953, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7137733697891235, + "num_tokens": 262500149.0, + "step": 10221 + }, + { + "epoch": 1.122556556116846, + "grad_norm": 2.1105122566223145, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.680726170539856, + "num_tokens": 262531385.0, + "step": 10222 + }, + { + "epoch": 1.1226663738194598, + "grad_norm": 1.8248395919799805, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7178406715393066, + "num_tokens": 262565409.0, + "step": 10223 + }, + { + "epoch": 1.1227761915220733, + "grad_norm": 2.2382450103759766, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7014718055725098, + "num_tokens": 262590157.0, + "step": 10224 + }, + { + "epoch": 1.122886009224687, + "grad_norm": 2.123312473297119, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7090200185775757, + "num_tokens": 262620128.0, + "step": 10225 + }, + { + "epoch": 1.1229958269273006, + "grad_norm": 2.1674680709838867, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6979233026504517, + "num_tokens": 262650993.0, + "step": 10226 + }, + { + "epoch": 1.1231056446299144, + "grad_norm": 2.553623914718628, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7356266379356384, + "num_tokens": 262671301.0, + "step": 10227 + }, + { + "epoch": 1.123215462332528, + "grad_norm": 2.1971685886383057, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.723663330078125, + "num_tokens": 262698888.0, + "step": 10228 + }, + { + "epoch": 1.1233252800351416, + "grad_norm": 2.1605570316314697, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7179149389266968, + "num_tokens": 262726222.0, + "step": 10229 + }, + { + "epoch": 1.1234350977377554, + "grad_norm": 2.187321186065674, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7248722314834595, + "num_tokens": 262751500.0, + "step": 10230 + }, + { + "epoch": 1.123544915440369, + "grad_norm": 2.2894113063812256, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.69346022605896, + "num_tokens": 262781197.0, + "step": 10231 + }, + { + "epoch": 1.1236547331429827, + "grad_norm": 2.719191074371338, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7218971848487854, + "num_tokens": 262801486.0, + "step": 10232 + }, + { + "epoch": 1.1237645508455962, + "grad_norm": 2.1252593994140625, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7181544303894043, + "num_tokens": 262829009.0, + "step": 10233 + }, + { + "epoch": 1.12387436854821, + "grad_norm": 2.313870429992676, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7162445783615112, + "num_tokens": 262853068.0, + "step": 10234 + }, + { + "epoch": 1.1239841862508237, + "grad_norm": 2.515249490737915, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7324687242507935, + "num_tokens": 262874548.0, + "step": 10235 + }, + { + "epoch": 1.1240940039534373, + "grad_norm": 2.4944090843200684, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7353006601333618, + "num_tokens": 262894894.0, + "step": 10236 + }, + { + "epoch": 1.124203821656051, + "grad_norm": 2.346045732498169, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7312875986099243, + "num_tokens": 262918313.0, + "step": 10237 + }, + { + "epoch": 1.1243136393586646, + "grad_norm": 2.337240219116211, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7148432731628418, + "num_tokens": 262941317.0, + "step": 10238 + }, + { + "epoch": 1.1244234570612783, + "grad_norm": 2.2601499557495117, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6976878046989441, + "num_tokens": 262966819.0, + "step": 10239 + }, + { + "epoch": 1.1245332747638919, + "grad_norm": 2.138807773590088, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7033260464668274, + "num_tokens": 262996664.0, + "step": 10240 + }, + { + "epoch": 1.1246430924665056, + "grad_norm": 2.087679624557495, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7160410284996033, + "num_tokens": 263025188.0, + "step": 10241 + }, + { + "epoch": 1.1247529101691192, + "grad_norm": 2.1428334712982178, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7109936475753784, + "num_tokens": 263055541.0, + "step": 10242 + }, + { + "epoch": 1.124862727871733, + "grad_norm": 2.1584529876708984, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7302221655845642, + "num_tokens": 263082869.0, + "step": 10243 + }, + { + "epoch": 1.1249725455743467, + "grad_norm": 2.2257583141326904, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.724920928478241, + "num_tokens": 263107750.0, + "step": 10244 + }, + { + "epoch": 1.1250823632769602, + "grad_norm": 2.438209295272827, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7101278305053711, + "num_tokens": 263131076.0, + "step": 10245 + }, + { + "epoch": 1.125192180979574, + "grad_norm": 2.216768980026245, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6874983310699463, + "num_tokens": 263161115.0, + "step": 10246 + }, + { + "epoch": 1.1253019986821875, + "grad_norm": 2.486955165863037, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7254185080528259, + "num_tokens": 263182978.0, + "step": 10247 + }, + { + "epoch": 1.1254118163848013, + "grad_norm": 2.199267864227295, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7172355651855469, + "num_tokens": 263209161.0, + "step": 10248 + }, + { + "epoch": 1.125521634087415, + "grad_norm": 2.26271653175354, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7200063467025757, + "num_tokens": 263235929.0, + "step": 10249 + }, + { + "epoch": 1.1256314517900285, + "grad_norm": 2.4044461250305176, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7326543927192688, + "num_tokens": 263257653.0, + "step": 10250 + }, + { + "epoch": 1.1257412694926423, + "grad_norm": 2.239407777786255, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7000935673713684, + "num_tokens": 263284002.0, + "step": 10251 + }, + { + "epoch": 1.1258510871952558, + "grad_norm": 2.1534295082092285, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7191243171691895, + "num_tokens": 263311579.0, + "step": 10252 + }, + { + "epoch": 1.1259609048978696, + "grad_norm": 2.0914690494537354, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7301504611968994, + "num_tokens": 263339297.0, + "step": 10253 + }, + { + "epoch": 1.1260707226004831, + "grad_norm": 2.2865066528320312, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7194496989250183, + "num_tokens": 263364073.0, + "step": 10254 + }, + { + "epoch": 1.1261805403030969, + "grad_norm": 2.126059055328369, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7104194164276123, + "num_tokens": 263393396.0, + "step": 10255 + }, + { + "epoch": 1.1262903580057104, + "grad_norm": 2.3703103065490723, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7319542169570923, + "num_tokens": 263417909.0, + "step": 10256 + }, + { + "epoch": 1.1264001757083242, + "grad_norm": 2.2075045108795166, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7116271257400513, + "num_tokens": 263444438.0, + "step": 10257 + }, + { + "epoch": 1.126509993410938, + "grad_norm": 2.2322559356689453, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7148352265357971, + "num_tokens": 263470753.0, + "step": 10258 + }, + { + "epoch": 1.1266198111135515, + "grad_norm": 2.1637766361236572, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6912063360214233, + "num_tokens": 263500812.0, + "step": 10259 + }, + { + "epoch": 1.1267296288161652, + "grad_norm": 2.238858461380005, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7057678699493408, + "num_tokens": 263527201.0, + "step": 10260 + }, + { + "epoch": 1.1268394465187788, + "grad_norm": 2.1666972637176514, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7130980491638184, + "num_tokens": 263554597.0, + "step": 10261 + }, + { + "epoch": 1.1269492642213925, + "grad_norm": 2.3478879928588867, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7305142283439636, + "num_tokens": 263578562.0, + "step": 10262 + }, + { + "epoch": 1.127059081924006, + "grad_norm": 2.590724468231201, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7271246910095215, + "num_tokens": 263598251.0, + "step": 10263 + }, + { + "epoch": 1.1271688996266198, + "grad_norm": 2.511598825454712, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7238959670066833, + "num_tokens": 263620414.0, + "step": 10264 + }, + { + "epoch": 1.1272787173292333, + "grad_norm": 2.2393546104431152, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7057008743286133, + "num_tokens": 263647022.0, + "step": 10265 + }, + { + "epoch": 1.127388535031847, + "grad_norm": 2.1384799480438232, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7516977787017822, + "num_tokens": 263671795.0, + "step": 10266 + }, + { + "epoch": 1.1274983527344609, + "grad_norm": 2.208683967590332, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7097569108009338, + "num_tokens": 263699040.0, + "step": 10267 + }, + { + "epoch": 1.1276081704370744, + "grad_norm": 2.540576219558716, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7342445850372314, + "num_tokens": 263719405.0, + "step": 10268 + }, + { + "epoch": 1.1277179881396882, + "grad_norm": 2.546290159225464, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7169060707092285, + "num_tokens": 263739196.0, + "step": 10269 + }, + { + "epoch": 1.1278278058423017, + "grad_norm": 2.4085733890533447, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7217718362808228, + "num_tokens": 263762375.0, + "step": 10270 + }, + { + "epoch": 1.1279376235449154, + "grad_norm": 2.234896421432495, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.6985909342765808, + "num_tokens": 263787239.0, + "step": 10271 + }, + { + "epoch": 1.1280474412475292, + "grad_norm": 2.1415343284606934, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7163169980049133, + "num_tokens": 263817509.0, + "step": 10272 + }, + { + "epoch": 1.1281572589501427, + "grad_norm": 2.282413959503174, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7163904905319214, + "num_tokens": 263841855.0, + "step": 10273 + }, + { + "epoch": 1.1282670766527565, + "grad_norm": 2.4028477668762207, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7086760997772217, + "num_tokens": 263865796.0, + "step": 10274 + }, + { + "epoch": 1.12837689435537, + "grad_norm": 2.235706329345703, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7252163887023926, + "num_tokens": 263890725.0, + "step": 10275 + }, + { + "epoch": 1.1284867120579838, + "grad_norm": 2.0097439289093018, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.724777102470398, + "num_tokens": 263919177.0, + "step": 10276 + }, + { + "epoch": 1.1285965297605973, + "grad_norm": 2.6658427715301514, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7203906774520874, + "num_tokens": 263937989.0, + "step": 10277 + }, + { + "epoch": 1.128706347463211, + "grad_norm": 2.4988045692443848, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7084448337554932, + "num_tokens": 263959912.0, + "step": 10278 + }, + { + "epoch": 1.1288161651658246, + "grad_norm": 2.120509147644043, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6967862844467163, + "num_tokens": 263988959.0, + "step": 10279 + }, + { + "epoch": 1.1289259828684384, + "grad_norm": 2.339320659637451, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7188004851341248, + "num_tokens": 264013789.0, + "step": 10280 + }, + { + "epoch": 1.1290358005710521, + "grad_norm": 2.0707600116729736, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7165782451629639, + "num_tokens": 264041925.0, + "step": 10281 + }, + { + "epoch": 1.1291456182736657, + "grad_norm": 2.500384569168091, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7213842272758484, + "num_tokens": 264063760.0, + "step": 10282 + }, + { + "epoch": 1.1292554359762794, + "grad_norm": 2.0979771614074707, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6999258995056152, + "num_tokens": 264092161.0, + "step": 10283 + }, + { + "epoch": 1.129365253678893, + "grad_norm": 2.1041409969329834, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6940363645553589, + "num_tokens": 264122431.0, + "step": 10284 + }, + { + "epoch": 1.1294750713815067, + "grad_norm": 2.0110886096954346, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7261539697647095, + "num_tokens": 264154375.0, + "step": 10285 + }, + { + "epoch": 1.1295848890841205, + "grad_norm": 2.262941837310791, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7235425710678101, + "num_tokens": 264180606.0, + "step": 10286 + }, + { + "epoch": 1.129694706786734, + "grad_norm": 2.293769598007202, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.727950394153595, + "num_tokens": 264204952.0, + "step": 10287 + }, + { + "epoch": 1.1298045244893478, + "grad_norm": 2.255671739578247, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7318155765533447, + "num_tokens": 264229780.0, + "step": 10288 + }, + { + "epoch": 1.1299143421919613, + "grad_norm": 2.3208141326904297, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7294631004333496, + "num_tokens": 264253933.0, + "step": 10289 + }, + { + "epoch": 1.130024159894575, + "grad_norm": 2.05387806892395, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6853495240211487, + "num_tokens": 264283457.0, + "step": 10290 + }, + { + "epoch": 1.1301339775971886, + "grad_norm": 2.3588852882385254, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7196180820465088, + "num_tokens": 264305884.0, + "step": 10291 + }, + { + "epoch": 1.1302437952998023, + "grad_norm": 2.084115982055664, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7200652360916138, + "num_tokens": 264334775.0, + "step": 10292 + }, + { + "epoch": 1.1303536130024159, + "grad_norm": 2.0920462608337402, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7166141867637634, + "num_tokens": 264362160.0, + "step": 10293 + }, + { + "epoch": 1.1304634307050296, + "grad_norm": 2.684691905975342, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7056966423988342, + "num_tokens": 264382441.0, + "step": 10294 + }, + { + "epoch": 1.1305732484076434, + "grad_norm": 2.4401774406433105, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7240241765975952, + "num_tokens": 264404123.0, + "step": 10295 + }, + { + "epoch": 1.130683066110257, + "grad_norm": 2.0061538219451904, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7109906673431396, + "num_tokens": 264435988.0, + "step": 10296 + }, + { + "epoch": 1.1307928838128707, + "grad_norm": 2.2844552993774414, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7352467775344849, + "num_tokens": 264459839.0, + "step": 10297 + }, + { + "epoch": 1.1309027015154842, + "grad_norm": 2.378885507583618, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7055530548095703, + "num_tokens": 264481930.0, + "step": 10298 + }, + { + "epoch": 1.131012519218098, + "grad_norm": 2.2896738052368164, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7146013975143433, + "num_tokens": 264508345.0, + "step": 10299 + }, + { + "epoch": 1.1311223369207117, + "grad_norm": 2.3146443367004395, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7037429809570312, + "num_tokens": 264532321.0, + "step": 10300 + }, + { + "epoch": 1.1312321546233253, + "grad_norm": 2.286680221557617, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7041996717453003, + "num_tokens": 264557638.0, + "step": 10301 + }, + { + "epoch": 1.131341972325939, + "grad_norm": 2.250162363052368, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.720220148563385, + "num_tokens": 264582260.0, + "step": 10302 + }, + { + "epoch": 1.1314517900285526, + "grad_norm": 2.303623676300049, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7064452767372131, + "num_tokens": 264609225.0, + "step": 10303 + }, + { + "epoch": 1.1315616077311663, + "grad_norm": 2.5349743366241455, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7207640409469604, + "num_tokens": 264630943.0, + "step": 10304 + }, + { + "epoch": 1.1316714254337799, + "grad_norm": 2.332421064376831, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7302302122116089, + "num_tokens": 264653848.0, + "step": 10305 + }, + { + "epoch": 1.1317812431363936, + "grad_norm": 1.9593182802200317, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7207679748535156, + "num_tokens": 264687253.0, + "step": 10306 + }, + { + "epoch": 1.1318910608390071, + "grad_norm": 2.107052803039551, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7148505449295044, + "num_tokens": 264715853.0, + "step": 10307 + }, + { + "epoch": 1.132000878541621, + "grad_norm": 2.5911741256713867, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7449241876602173, + "num_tokens": 264734744.0, + "step": 10308 + }, + { + "epoch": 1.1321106962442347, + "grad_norm": 2.0989246368408203, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7076795697212219, + "num_tokens": 264762700.0, + "step": 10309 + }, + { + "epoch": 1.1322205139468482, + "grad_norm": 2.081303358078003, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7371056079864502, + "num_tokens": 264789217.0, + "step": 10310 + }, + { + "epoch": 1.132330331649462, + "grad_norm": 2.1441073417663574, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7098161578178406, + "num_tokens": 264817419.0, + "step": 10311 + }, + { + "epoch": 1.1324401493520755, + "grad_norm": 2.4666171073913574, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7252930402755737, + "num_tokens": 264839852.0, + "step": 10312 + }, + { + "epoch": 1.1325499670546892, + "grad_norm": 2.097472906112671, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7038429975509644, + "num_tokens": 264869047.0, + "step": 10313 + }, + { + "epoch": 1.132659784757303, + "grad_norm": 2.808088541030884, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7209585905075073, + "num_tokens": 264887706.0, + "step": 10314 + }, + { + "epoch": 1.1327696024599165, + "grad_norm": 2.770733594894409, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7106512784957886, + "num_tokens": 264907325.0, + "step": 10315 + }, + { + "epoch": 1.1328794201625303, + "grad_norm": 2.168313980102539, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7217532992362976, + "num_tokens": 264934281.0, + "step": 10316 + }, + { + "epoch": 1.1329892378651438, + "grad_norm": 2.28916335105896, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7129138708114624, + "num_tokens": 264958687.0, + "step": 10317 + }, + { + "epoch": 1.1330990555677576, + "grad_norm": 2.0263919830322266, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7080153226852417, + "num_tokens": 264988076.0, + "step": 10318 + }, + { + "epoch": 1.1332088732703711, + "grad_norm": 2.2389280796051025, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7124197483062744, + "num_tokens": 265014552.0, + "step": 10319 + }, + { + "epoch": 1.1333186909729849, + "grad_norm": 2.172987937927246, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7356817126274109, + "num_tokens": 265040059.0, + "step": 10320 + }, + { + "epoch": 1.1334285086755984, + "grad_norm": 2.3484108448028564, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.709223747253418, + "num_tokens": 265067049.0, + "step": 10321 + }, + { + "epoch": 1.1335383263782122, + "grad_norm": 2.739556312561035, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7292355298995972, + "num_tokens": 265084906.0, + "step": 10322 + }, + { + "epoch": 1.133648144080826, + "grad_norm": 1.9884880781173706, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7111275792121887, + "num_tokens": 265116119.0, + "step": 10323 + }, + { + "epoch": 1.1337579617834395, + "grad_norm": 2.5398364067077637, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7107350826263428, + "num_tokens": 265138586.0, + "step": 10324 + }, + { + "epoch": 1.1338677794860532, + "grad_norm": 2.094632148742676, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7001665830612183, + "num_tokens": 265172101.0, + "step": 10325 + }, + { + "epoch": 1.1339775971886668, + "grad_norm": 2.3100922107696533, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7018026113510132, + "num_tokens": 265198816.0, + "step": 10326 + }, + { + "epoch": 1.1340874148912805, + "grad_norm": 2.392246723175049, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7102457284927368, + "num_tokens": 265223166.0, + "step": 10327 + }, + { + "epoch": 1.134197232593894, + "grad_norm": 2.4237465858459473, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7239424586296082, + "num_tokens": 265245623.0, + "step": 10328 + }, + { + "epoch": 1.1343070502965078, + "grad_norm": 2.1903958320617676, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7006781101226807, + "num_tokens": 265274564.0, + "step": 10329 + }, + { + "epoch": 1.1344168679991213, + "grad_norm": 2.469630241394043, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7278826236724854, + "num_tokens": 265296339.0, + "step": 10330 + }, + { + "epoch": 1.134526685701735, + "grad_norm": 2.199148416519165, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7243646383285522, + "num_tokens": 265321204.0, + "step": 10331 + }, + { + "epoch": 1.1346365034043489, + "grad_norm": 2.135298728942871, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7103809714317322, + "num_tokens": 265349290.0, + "step": 10332 + }, + { + "epoch": 1.1347463211069624, + "grad_norm": 2.3140523433685303, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7240103483200073, + "num_tokens": 265372973.0, + "step": 10333 + }, + { + "epoch": 1.1348561388095761, + "grad_norm": 2.2723960876464844, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7035356760025024, + "num_tokens": 265399671.0, + "step": 10334 + }, + { + "epoch": 1.1349659565121897, + "grad_norm": 2.171612501144409, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7001926898956299, + "num_tokens": 265430647.0, + "step": 10335 + }, + { + "epoch": 1.1350757742148034, + "grad_norm": 2.2595527172088623, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7113783955574036, + "num_tokens": 265456938.0, + "step": 10336 + }, + { + "epoch": 1.1351855919174172, + "grad_norm": 2.13206148147583, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7348940372467041, + "num_tokens": 265484923.0, + "step": 10337 + }, + { + "epoch": 1.1352954096200307, + "grad_norm": 2.1628167629241943, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7180658578872681, + "num_tokens": 265512422.0, + "step": 10338 + }, + { + "epoch": 1.1354052273226445, + "grad_norm": 2.581268310546875, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7348650693893433, + "num_tokens": 265531680.0, + "step": 10339 + }, + { + "epoch": 1.135515045025258, + "grad_norm": 2.2616970539093018, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7023622989654541, + "num_tokens": 265557342.0, + "step": 10340 + }, + { + "epoch": 1.1356248627278718, + "grad_norm": 2.0876758098602295, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6961153745651245, + "num_tokens": 265586863.0, + "step": 10341 + }, + { + "epoch": 1.1357346804304853, + "grad_norm": 2.6989150047302246, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7309799790382385, + "num_tokens": 265607007.0, + "step": 10342 + }, + { + "epoch": 1.135844498133099, + "grad_norm": 1.9986681938171387, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7172195315361023, + "num_tokens": 265638096.0, + "step": 10343 + }, + { + "epoch": 1.1359543158357126, + "grad_norm": 2.604965925216675, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7054965496063232, + "num_tokens": 265662766.0, + "step": 10344 + }, + { + "epoch": 1.1360641335383264, + "grad_norm": 2.2386510372161865, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7168139815330505, + "num_tokens": 265688417.0, + "step": 10345 + }, + { + "epoch": 1.1361739512409401, + "grad_norm": 2.16941499710083, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7204890251159668, + "num_tokens": 265713795.0, + "step": 10346 + }, + { + "epoch": 1.1362837689435537, + "grad_norm": 2.3842837810516357, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7307007312774658, + "num_tokens": 265736408.0, + "step": 10347 + }, + { + "epoch": 1.1363935866461674, + "grad_norm": 2.156456708908081, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6952012181282043, + "num_tokens": 265767198.0, + "step": 10348 + }, + { + "epoch": 1.136503404348781, + "grad_norm": 2.373321771621704, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7522621750831604, + "num_tokens": 265789017.0, + "step": 10349 + }, + { + "epoch": 1.1366132220513947, + "grad_norm": 2.398589849472046, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7165743708610535, + "num_tokens": 265812090.0, + "step": 10350 + }, + { + "epoch": 1.1367230397540085, + "grad_norm": 2.4890356063842773, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7244019508361816, + "num_tokens": 265832510.0, + "step": 10351 + }, + { + "epoch": 1.136832857456622, + "grad_norm": 2.386077404022217, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7183130979537964, + "num_tokens": 265856362.0, + "step": 10352 + }, + { + "epoch": 1.1369426751592357, + "grad_norm": 2.2305362224578857, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7147308588027954, + "num_tokens": 265881346.0, + "step": 10353 + }, + { + "epoch": 1.1370524928618493, + "grad_norm": 2.4935717582702637, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6978328227996826, + "num_tokens": 265904610.0, + "step": 10354 + }, + { + "epoch": 1.137162310564463, + "grad_norm": 2.1889357566833496, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.737158477306366, + "num_tokens": 265928165.0, + "step": 10355 + }, + { + "epoch": 1.1372721282670766, + "grad_norm": 2.323051929473877, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7362961769104004, + "num_tokens": 265951162.0, + "step": 10356 + }, + { + "epoch": 1.1373819459696903, + "grad_norm": 1.9798328876495361, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7301741242408752, + "num_tokens": 265981384.0, + "step": 10357 + }, + { + "epoch": 1.1374917636723039, + "grad_norm": 2.2914576530456543, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7143775224685669, + "num_tokens": 266005240.0, + "step": 10358 + }, + { + "epoch": 1.1376015813749176, + "grad_norm": 2.0659680366516113, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.710757851600647, + "num_tokens": 266034782.0, + "step": 10359 + }, + { + "epoch": 1.1377113990775314, + "grad_norm": 2.1222267150878906, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7096899747848511, + "num_tokens": 266064111.0, + "step": 10360 + }, + { + "epoch": 1.137821216780145, + "grad_norm": 2.2345807552337646, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6994582414627075, + "num_tokens": 266090871.0, + "step": 10361 + }, + { + "epoch": 1.1379310344827587, + "grad_norm": 2.3678348064422607, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7385333776473999, + "num_tokens": 266114148.0, + "step": 10362 + }, + { + "epoch": 1.1380408521853722, + "grad_norm": 2.6873202323913574, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7278777360916138, + "num_tokens": 266132646.0, + "step": 10363 + }, + { + "epoch": 1.138150669887986, + "grad_norm": 2.263361692428589, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7200801372528076, + "num_tokens": 266157537.0, + "step": 10364 + }, + { + "epoch": 1.1382604875905997, + "grad_norm": 2.1203293800354004, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7097358703613281, + "num_tokens": 266187284.0, + "step": 10365 + }, + { + "epoch": 1.1383703052932133, + "grad_norm": 2.179525136947632, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7215304374694824, + "num_tokens": 266213702.0, + "step": 10366 + }, + { + "epoch": 1.138480122995827, + "grad_norm": 1.9520374536514282, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7168386578559875, + "num_tokens": 266246566.0, + "step": 10367 + }, + { + "epoch": 1.1385899406984406, + "grad_norm": 2.165834665298462, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7118035554885864, + "num_tokens": 266273407.0, + "step": 10368 + }, + { + "epoch": 1.1386997584010543, + "grad_norm": 2.2824206352233887, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7087925672531128, + "num_tokens": 266298013.0, + "step": 10369 + }, + { + "epoch": 1.1388095761036678, + "grad_norm": 2.111382484436035, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7280086278915405, + "num_tokens": 266325418.0, + "step": 10370 + }, + { + "epoch": 1.1389193938062816, + "grad_norm": 2.3155508041381836, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.708213746547699, + "num_tokens": 266349632.0, + "step": 10371 + }, + { + "epoch": 1.1390292115088951, + "grad_norm": 2.193556547164917, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7338591814041138, + "num_tokens": 266375325.0, + "step": 10372 + }, + { + "epoch": 1.139139029211509, + "grad_norm": 2.0719032287597656, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6981600522994995, + "num_tokens": 266404606.0, + "step": 10373 + }, + { + "epoch": 1.1392488469141226, + "grad_norm": 2.144054889678955, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6917462348937988, + "num_tokens": 266433611.0, + "step": 10374 + }, + { + "epoch": 1.1393586646167362, + "grad_norm": 2.1258938312530518, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7152497172355652, + "num_tokens": 266461494.0, + "step": 10375 + }, + { + "epoch": 1.13946848231935, + "grad_norm": 2.321894884109497, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7220258116722107, + "num_tokens": 266485432.0, + "step": 10376 + }, + { + "epoch": 1.1395783000219635, + "grad_norm": 2.314938545227051, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7229445576667786, + "num_tokens": 266507902.0, + "step": 10377 + }, + { + "epoch": 1.1396881177245772, + "grad_norm": 2.0516858100891113, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7041865587234497, + "num_tokens": 266538761.0, + "step": 10378 + }, + { + "epoch": 1.1397979354271908, + "grad_norm": 2.206312417984009, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7623864412307739, + "num_tokens": 266563234.0, + "step": 10379 + }, + { + "epoch": 1.1399077531298045, + "grad_norm": 2.4786322116851807, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7295984029769897, + "num_tokens": 266585360.0, + "step": 10380 + }, + { + "epoch": 1.1400175708324183, + "grad_norm": 2.128059148788452, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7135841250419617, + "num_tokens": 266612789.0, + "step": 10381 + }, + { + "epoch": 1.1401273885350318, + "grad_norm": 2.197061538696289, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7152321338653564, + "num_tokens": 266638539.0, + "step": 10382 + }, + { + "epoch": 1.1402372062376456, + "grad_norm": 2.395806074142456, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7132056951522827, + "num_tokens": 266663756.0, + "step": 10383 + }, + { + "epoch": 1.140347023940259, + "grad_norm": 2.0392088890075684, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7037523984909058, + "num_tokens": 266695592.0, + "step": 10384 + }, + { + "epoch": 1.1404568416428729, + "grad_norm": 2.510143280029297, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7254698872566223, + "num_tokens": 266716881.0, + "step": 10385 + }, + { + "epoch": 1.1405666593454864, + "grad_norm": 2.6067328453063965, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7437103986740112, + "num_tokens": 266735839.0, + "step": 10386 + }, + { + "epoch": 1.1406764770481002, + "grad_norm": 2.2484242916107178, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7232968807220459, + "num_tokens": 266762949.0, + "step": 10387 + }, + { + "epoch": 1.140786294750714, + "grad_norm": 2.198265552520752, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7133095264434814, + "num_tokens": 266789044.0, + "step": 10388 + }, + { + "epoch": 1.1408961124533274, + "grad_norm": 2.3143208026885986, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.735194742679596, + "num_tokens": 266810621.0, + "step": 10389 + }, + { + "epoch": 1.1410059301559412, + "grad_norm": 2.254849910736084, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7374622821807861, + "num_tokens": 266835203.0, + "step": 10390 + }, + { + "epoch": 1.1411157478585547, + "grad_norm": 2.004061460494995, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7058655023574829, + "num_tokens": 266868563.0, + "step": 10391 + }, + { + "epoch": 1.1412255655611685, + "grad_norm": 2.1421704292297363, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6929762959480286, + "num_tokens": 266896793.0, + "step": 10392 + }, + { + "epoch": 1.141335383263782, + "grad_norm": 2.2184512615203857, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6961545348167419, + "num_tokens": 266927315.0, + "step": 10393 + }, + { + "epoch": 1.1414452009663958, + "grad_norm": 2.4476521015167236, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7351371645927429, + "num_tokens": 266949939.0, + "step": 10394 + }, + { + "epoch": 1.1415550186690093, + "grad_norm": 2.55251145362854, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7170572280883789, + "num_tokens": 266972460.0, + "step": 10395 + }, + { + "epoch": 1.141664836371623, + "grad_norm": 2.056201696395874, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7158982157707214, + "num_tokens": 267002439.0, + "step": 10396 + }, + { + "epoch": 1.1417746540742368, + "grad_norm": 2.4442436695098877, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.763049840927124, + "num_tokens": 267023532.0, + "step": 10397 + }, + { + "epoch": 1.1418844717768504, + "grad_norm": 2.24772047996521, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.727080762386322, + "num_tokens": 267050320.0, + "step": 10398 + }, + { + "epoch": 1.1419942894794641, + "grad_norm": 2.2038424015045166, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7199453115463257, + "num_tokens": 267075615.0, + "step": 10399 + }, + { + "epoch": 1.1421041071820777, + "grad_norm": 2.2729785442352295, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7122138738632202, + "num_tokens": 267104883.0, + "step": 10400 + }, + { + "epoch": 1.1422139248846914, + "grad_norm": 2.532877206802368, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7140011787414551, + "num_tokens": 267131208.0, + "step": 10401 + }, + { + "epoch": 1.1423237425873052, + "grad_norm": 2.2273552417755127, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6884133815765381, + "num_tokens": 267161177.0, + "step": 10402 + }, + { + "epoch": 1.1424335602899187, + "grad_norm": 2.4149630069732666, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7017311453819275, + "num_tokens": 267184571.0, + "step": 10403 + }, + { + "epoch": 1.1425433779925325, + "grad_norm": 2.296654462814331, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7061425447463989, + "num_tokens": 267210354.0, + "step": 10404 + }, + { + "epoch": 1.142653195695146, + "grad_norm": 2.380380153656006, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.718096137046814, + "num_tokens": 267232958.0, + "step": 10405 + }, + { + "epoch": 1.1427630133977598, + "grad_norm": 2.1682708263397217, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7277448177337646, + "num_tokens": 267259983.0, + "step": 10406 + }, + { + "epoch": 1.1428728311003733, + "grad_norm": 2.242061138153076, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7115365862846375, + "num_tokens": 267287876.0, + "step": 10407 + }, + { + "epoch": 1.142982648802987, + "grad_norm": 2.5917422771453857, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7143412828445435, + "num_tokens": 267310397.0, + "step": 10408 + }, + { + "epoch": 1.1430924665056006, + "grad_norm": 2.1893768310546875, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7119012475013733, + "num_tokens": 267337341.0, + "step": 10409 + }, + { + "epoch": 1.1432022842082143, + "grad_norm": 2.230457305908203, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.721379280090332, + "num_tokens": 267362605.0, + "step": 10410 + }, + { + "epoch": 1.143312101910828, + "grad_norm": 2.5162460803985596, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7176154255867004, + "num_tokens": 267384610.0, + "step": 10411 + }, + { + "epoch": 1.1434219196134416, + "grad_norm": 2.4904420375823975, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7141714692115784, + "num_tokens": 267408099.0, + "step": 10412 + }, + { + "epoch": 1.1435317373160554, + "grad_norm": 2.356369733810425, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7283442616462708, + "num_tokens": 267431394.0, + "step": 10413 + }, + { + "epoch": 1.143641555018669, + "grad_norm": 2.2476589679718018, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.733022153377533, + "num_tokens": 267457992.0, + "step": 10414 + }, + { + "epoch": 1.1437513727212827, + "grad_norm": 2.288355588912964, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7227217555046082, + "num_tokens": 267482375.0, + "step": 10415 + }, + { + "epoch": 1.1438611904238964, + "grad_norm": 2.2841174602508545, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7016451358795166, + "num_tokens": 267510459.0, + "step": 10416 + }, + { + "epoch": 1.14397100812651, + "grad_norm": 2.087876796722412, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7275795936584473, + "num_tokens": 267538101.0, + "step": 10417 + }, + { + "epoch": 1.1440808258291237, + "grad_norm": 2.1945903301239014, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6971895098686218, + "num_tokens": 267563747.0, + "step": 10418 + }, + { + "epoch": 1.1441906435317373, + "grad_norm": 2.3535141944885254, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7377016544342041, + "num_tokens": 267586296.0, + "step": 10419 + }, + { + "epoch": 1.144300461234351, + "grad_norm": 2.0374910831451416, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7008774876594543, + "num_tokens": 267617616.0, + "step": 10420 + }, + { + "epoch": 1.1444102789369646, + "grad_norm": 2.368809700012207, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7228434085845947, + "num_tokens": 267640852.0, + "step": 10421 + }, + { + "epoch": 1.1445200966395783, + "grad_norm": 2.0528512001037598, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7162480354309082, + "num_tokens": 267671868.0, + "step": 10422 + }, + { + "epoch": 1.1446299143421919, + "grad_norm": 2.2913498878479004, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.720983624458313, + "num_tokens": 267696826.0, + "step": 10423 + }, + { + "epoch": 1.1447397320448056, + "grad_norm": 2.215927839279175, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7250518798828125, + "num_tokens": 267724169.0, + "step": 10424 + }, + { + "epoch": 1.1448495497474194, + "grad_norm": 2.5361647605895996, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.722237229347229, + "num_tokens": 267746458.0, + "step": 10425 + }, + { + "epoch": 1.144959367450033, + "grad_norm": 2.523484468460083, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7317992448806763, + "num_tokens": 267768853.0, + "step": 10426 + }, + { + "epoch": 1.1450691851526467, + "grad_norm": 2.3441808223724365, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7128851413726807, + "num_tokens": 267793714.0, + "step": 10427 + }, + { + "epoch": 1.1451790028552602, + "grad_norm": 2.5844886302948, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7051295042037964, + "num_tokens": 267813114.0, + "step": 10428 + }, + { + "epoch": 1.145288820557874, + "grad_norm": 2.242058753967285, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7409923672676086, + "num_tokens": 267839173.0, + "step": 10429 + }, + { + "epoch": 1.1453986382604877, + "grad_norm": 2.279118299484253, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7114713191986084, + "num_tokens": 267865247.0, + "step": 10430 + }, + { + "epoch": 1.1455084559631012, + "grad_norm": 2.216339349746704, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7179900407791138, + "num_tokens": 267891607.0, + "step": 10431 + }, + { + "epoch": 1.145618273665715, + "grad_norm": 2.2709195613861084, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7127723693847656, + "num_tokens": 267917478.0, + "step": 10432 + }, + { + "epoch": 1.1457280913683285, + "grad_norm": 2.4657771587371826, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7269307374954224, + "num_tokens": 267938275.0, + "step": 10433 + }, + { + "epoch": 1.1458379090709423, + "grad_norm": 2.4457497596740723, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7346506118774414, + "num_tokens": 267959565.0, + "step": 10434 + }, + { + "epoch": 1.1459477267735558, + "grad_norm": 2.2273566722869873, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7259113192558289, + "num_tokens": 267985874.0, + "step": 10435 + }, + { + "epoch": 1.1460575444761696, + "grad_norm": 2.0857789516448975, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7168393135070801, + "num_tokens": 268014513.0, + "step": 10436 + }, + { + "epoch": 1.1461673621787831, + "grad_norm": 2.3478498458862305, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7111207246780396, + "num_tokens": 268038651.0, + "step": 10437 + }, + { + "epoch": 1.1462771798813969, + "grad_norm": 2.4019758701324463, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7216798663139343, + "num_tokens": 268062110.0, + "step": 10438 + }, + { + "epoch": 1.1463869975840106, + "grad_norm": 2.4170873165130615, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7131532430648804, + "num_tokens": 268085101.0, + "step": 10439 + }, + { + "epoch": 1.1464968152866242, + "grad_norm": 2.4004814624786377, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7281749844551086, + "num_tokens": 268107592.0, + "step": 10440 + }, + { + "epoch": 1.146606632989238, + "grad_norm": 2.353037118911743, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7119091749191284, + "num_tokens": 268132785.0, + "step": 10441 + }, + { + "epoch": 1.1467164506918515, + "grad_norm": 2.4271302223205566, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7194366455078125, + "num_tokens": 268157984.0, + "step": 10442 + }, + { + "epoch": 1.1468262683944652, + "grad_norm": 2.3917436599731445, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7058576345443726, + "num_tokens": 268181371.0, + "step": 10443 + }, + { + "epoch": 1.1469360860970788, + "grad_norm": 2.163217544555664, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7532773017883301, + "num_tokens": 268207132.0, + "step": 10444 + }, + { + "epoch": 1.1470459037996925, + "grad_norm": 2.116452693939209, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7288564443588257, + "num_tokens": 268232326.0, + "step": 10445 + }, + { + "epoch": 1.147155721502306, + "grad_norm": 1.9767545461654663, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7035120725631714, + "num_tokens": 268265144.0, + "step": 10446 + }, + { + "epoch": 1.1472655392049198, + "grad_norm": 2.0331642627716064, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7263947129249573, + "num_tokens": 268294566.0, + "step": 10447 + }, + { + "epoch": 1.1473753569075336, + "grad_norm": 2.388981819152832, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7050443887710571, + "num_tokens": 268318573.0, + "step": 10448 + }, + { + "epoch": 1.147485174610147, + "grad_norm": 2.3222877979278564, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7283908724784851, + "num_tokens": 268341658.0, + "step": 10449 + }, + { + "epoch": 1.1475949923127609, + "grad_norm": 2.305657148361206, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7114250659942627, + "num_tokens": 268366297.0, + "step": 10450 + }, + { + "epoch": 1.1477048100153744, + "grad_norm": 2.1963391304016113, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7199872136116028, + "num_tokens": 268392698.0, + "step": 10451 + }, + { + "epoch": 1.1478146277179881, + "grad_norm": 2.089698314666748, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7046264410018921, + "num_tokens": 268421523.0, + "step": 10452 + }, + { + "epoch": 1.147924445420602, + "grad_norm": 2.6538245677948, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7194604277610779, + "num_tokens": 268441890.0, + "step": 10453 + }, + { + "epoch": 1.1480342631232154, + "grad_norm": 2.4270846843719482, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7302054762840271, + "num_tokens": 268466401.0, + "step": 10454 + }, + { + "epoch": 1.1481440808258292, + "grad_norm": 2.4114720821380615, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7397385835647583, + "num_tokens": 268489415.0, + "step": 10455 + }, + { + "epoch": 1.1482538985284427, + "grad_norm": 2.1987075805664062, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7247281670570374, + "num_tokens": 268515139.0, + "step": 10456 + }, + { + "epoch": 1.1483637162310565, + "grad_norm": 2.1449098587036133, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7065970301628113, + "num_tokens": 268545490.0, + "step": 10457 + }, + { + "epoch": 1.14847353393367, + "grad_norm": 2.3792357444763184, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6945956945419312, + "num_tokens": 268570171.0, + "step": 10458 + }, + { + "epoch": 1.1485833516362838, + "grad_norm": 2.1707489490509033, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6957895159721375, + "num_tokens": 268599124.0, + "step": 10459 + }, + { + "epoch": 1.1486931693388973, + "grad_norm": 2.434490919113159, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7276098728179932, + "num_tokens": 268622365.0, + "step": 10460 + }, + { + "epoch": 1.148802987041511, + "grad_norm": 2.6795566082000732, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7263597249984741, + "num_tokens": 268642218.0, + "step": 10461 + }, + { + "epoch": 1.1489128047441248, + "grad_norm": 2.307161569595337, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7259124517440796, + "num_tokens": 268665169.0, + "step": 10462 + }, + { + "epoch": 1.1490226224467384, + "grad_norm": 2.265970230102539, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.740477442741394, + "num_tokens": 268689040.0, + "step": 10463 + }, + { + "epoch": 1.1491324401493521, + "grad_norm": 2.353531837463379, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7183778285980225, + "num_tokens": 268711935.0, + "step": 10464 + }, + { + "epoch": 1.1492422578519657, + "grad_norm": 2.4455671310424805, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7410054802894592, + "num_tokens": 268733920.0, + "step": 10465 + }, + { + "epoch": 1.1493520755545794, + "grad_norm": 2.098237991333008, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.716743528842926, + "num_tokens": 268761915.0, + "step": 10466 + }, + { + "epoch": 1.1494618932571932, + "grad_norm": 2.332557201385498, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7320135831832886, + "num_tokens": 268788036.0, + "step": 10467 + }, + { + "epoch": 1.1495717109598067, + "grad_norm": 2.372145891189575, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7374602556228638, + "num_tokens": 268811036.0, + "step": 10468 + }, + { + "epoch": 1.1496815286624205, + "grad_norm": 2.294830799102783, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7556727528572083, + "num_tokens": 268834247.0, + "step": 10469 + }, + { + "epoch": 1.149791346365034, + "grad_norm": 2.3450822830200195, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.73643958568573, + "num_tokens": 268857972.0, + "step": 10470 + }, + { + "epoch": 1.1499011640676478, + "grad_norm": 2.3311824798583984, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7206381559371948, + "num_tokens": 268882079.0, + "step": 10471 + }, + { + "epoch": 1.1500109817702613, + "grad_norm": 2.3097293376922607, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7264189720153809, + "num_tokens": 268905344.0, + "step": 10472 + }, + { + "epoch": 1.150120799472875, + "grad_norm": 2.376920700073242, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7250545024871826, + "num_tokens": 268929915.0, + "step": 10473 + }, + { + "epoch": 1.1502306171754886, + "grad_norm": 2.0530712604522705, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.716804027557373, + "num_tokens": 268960666.0, + "step": 10474 + }, + { + "epoch": 1.1503404348781023, + "grad_norm": 2.3176608085632324, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7226471900939941, + "num_tokens": 268986510.0, + "step": 10475 + }, + { + "epoch": 1.150450252580716, + "grad_norm": 2.149094820022583, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7184202075004578, + "num_tokens": 269013004.0, + "step": 10476 + }, + { + "epoch": 1.1505600702833296, + "grad_norm": 2.178413152694702, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6867347955703735, + "num_tokens": 269040521.0, + "step": 10477 + }, + { + "epoch": 1.1506698879859434, + "grad_norm": 2.1327147483825684, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7372345924377441, + "num_tokens": 269067732.0, + "step": 10478 + }, + { + "epoch": 1.150779705688557, + "grad_norm": 2.041752815246582, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7203734517097473, + "num_tokens": 269097541.0, + "step": 10479 + }, + { + "epoch": 1.1508895233911707, + "grad_norm": 2.1994552612304688, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7095743417739868, + "num_tokens": 269123984.0, + "step": 10480 + }, + { + "epoch": 1.1509993410937844, + "grad_norm": 2.113227605819702, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7118014097213745, + "num_tokens": 269154558.0, + "step": 10481 + }, + { + "epoch": 1.151109158796398, + "grad_norm": 2.2815725803375244, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7182848453521729, + "num_tokens": 269178490.0, + "step": 10482 + }, + { + "epoch": 1.1512189764990117, + "grad_norm": 2.3353536128997803, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7327455282211304, + "num_tokens": 269201956.0, + "step": 10483 + }, + { + "epoch": 1.1513287942016253, + "grad_norm": 2.314842700958252, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7152383327484131, + "num_tokens": 269226230.0, + "step": 10484 + }, + { + "epoch": 1.151438611904239, + "grad_norm": 2.1225361824035645, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7436159253120422, + "num_tokens": 269252206.0, + "step": 10485 + }, + { + "epoch": 1.1515484296068526, + "grad_norm": 2.022618055343628, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7052403092384338, + "num_tokens": 269284526.0, + "step": 10486 + }, + { + "epoch": 1.1516582473094663, + "grad_norm": 2.7094833850860596, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.726576566696167, + "num_tokens": 269303846.0, + "step": 10487 + }, + { + "epoch": 1.1517680650120798, + "grad_norm": 2.341801881790161, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.7011391520500183, + "num_tokens": 269328817.0, + "step": 10488 + }, + { + "epoch": 1.1518778827146936, + "grad_norm": 2.105372190475464, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7244008779525757, + "num_tokens": 269355804.0, + "step": 10489 + }, + { + "epoch": 1.1519877004173074, + "grad_norm": 2.2480978965759277, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6932107210159302, + "num_tokens": 269382432.0, + "step": 10490 + }, + { + "epoch": 1.152097518119921, + "grad_norm": 2.5551278591156006, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7362879514694214, + "num_tokens": 269402356.0, + "step": 10491 + }, + { + "epoch": 1.1522073358225347, + "grad_norm": 2.1466526985168457, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7099151015281677, + "num_tokens": 269430072.0, + "step": 10492 + }, + { + "epoch": 1.1523171535251482, + "grad_norm": 2.3898720741271973, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7152770757675171, + "num_tokens": 269453318.0, + "step": 10493 + }, + { + "epoch": 1.152426971227762, + "grad_norm": 2.269242525100708, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7005733251571655, + "num_tokens": 269478446.0, + "step": 10494 + }, + { + "epoch": 1.1525367889303757, + "grad_norm": 2.2146129608154297, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7173111438751221, + "num_tokens": 269503346.0, + "step": 10495 + }, + { + "epoch": 1.1526466066329892, + "grad_norm": 2.11484694480896, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7079600095748901, + "num_tokens": 269534295.0, + "step": 10496 + }, + { + "epoch": 1.152756424335603, + "grad_norm": 2.2526566982269287, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7123934030532837, + "num_tokens": 269558079.0, + "step": 10497 + }, + { + "epoch": 1.1528662420382165, + "grad_norm": 2.2750844955444336, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.712518572807312, + "num_tokens": 269582825.0, + "step": 10498 + }, + { + "epoch": 1.1529760597408303, + "grad_norm": 2.4451448917388916, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7217856645584106, + "num_tokens": 269606229.0, + "step": 10499 + }, + { + "epoch": 1.1530858774434438, + "grad_norm": 2.262985944747925, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7181267738342285, + "num_tokens": 269630424.0, + "step": 10500 + }, + { + "epoch": 1.1531956951460576, + "grad_norm": 2.294358491897583, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7115535736083984, + "num_tokens": 269655912.0, + "step": 10501 + }, + { + "epoch": 1.153305512848671, + "grad_norm": 2.137410879135132, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7156531810760498, + "num_tokens": 269683618.0, + "step": 10502 + }, + { + "epoch": 1.1534153305512849, + "grad_norm": 2.0876553058624268, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7078678011894226, + "num_tokens": 269712849.0, + "step": 10503 + }, + { + "epoch": 1.1535251482538986, + "grad_norm": 2.4101760387420654, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.71811842918396, + "num_tokens": 269738973.0, + "step": 10504 + }, + { + "epoch": 1.1536349659565122, + "grad_norm": 2.1610796451568604, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7157266139984131, + "num_tokens": 269765510.0, + "step": 10505 + }, + { + "epoch": 1.153744783659126, + "grad_norm": 2.299145460128784, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7448821663856506, + "num_tokens": 269788169.0, + "step": 10506 + }, + { + "epoch": 1.1538546013617395, + "grad_norm": 2.35791277885437, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7116209268569946, + "num_tokens": 269810972.0, + "step": 10507 + }, + { + "epoch": 1.1539644190643532, + "grad_norm": 2.2627029418945312, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7201846837997437, + "num_tokens": 269835317.0, + "step": 10508 + }, + { + "epoch": 1.1540742367669667, + "grad_norm": 2.13321590423584, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7192682027816772, + "num_tokens": 269861835.0, + "step": 10509 + }, + { + "epoch": 1.1541840544695805, + "grad_norm": 2.555612325668335, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7215073108673096, + "num_tokens": 269882150.0, + "step": 10510 + }, + { + "epoch": 1.154293872172194, + "grad_norm": 2.1037089824676514, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7314599752426147, + "num_tokens": 269910091.0, + "step": 10511 + }, + { + "epoch": 1.1544036898748078, + "grad_norm": 2.286155939102173, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7161364555358887, + "num_tokens": 269934977.0, + "step": 10512 + }, + { + "epoch": 1.1545135075774215, + "grad_norm": 2.0143778324127197, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6909716129302979, + "num_tokens": 269966978.0, + "step": 10513 + }, + { + "epoch": 1.154623325280035, + "grad_norm": 2.5493361949920654, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7350285649299622, + "num_tokens": 269988315.0, + "step": 10514 + }, + { + "epoch": 1.1547331429826488, + "grad_norm": 2.1829121112823486, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7159653306007385, + "num_tokens": 270014929.0, + "step": 10515 + }, + { + "epoch": 1.1548429606852624, + "grad_norm": 2.0320498943328857, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.71141117811203, + "num_tokens": 270045787.0, + "step": 10516 + }, + { + "epoch": 1.1549527783878761, + "grad_norm": 2.1874566078186035, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7266259789466858, + "num_tokens": 270072560.0, + "step": 10517 + }, + { + "epoch": 1.15506259609049, + "grad_norm": 2.221299648284912, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7253422737121582, + "num_tokens": 270096823.0, + "step": 10518 + }, + { + "epoch": 1.1551724137931034, + "grad_norm": 2.122770309448242, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.7010800838470459, + "num_tokens": 270129057.0, + "step": 10519 + }, + { + "epoch": 1.1552822314957172, + "grad_norm": 2.2608225345611572, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7528124451637268, + "num_tokens": 270153340.0, + "step": 10520 + }, + { + "epoch": 1.1553920491983307, + "grad_norm": 2.552762031555176, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7280879020690918, + "num_tokens": 270173130.0, + "step": 10521 + }, + { + "epoch": 1.1555018669009445, + "grad_norm": 2.121549606323242, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.734418511390686, + "num_tokens": 270199879.0, + "step": 10522 + }, + { + "epoch": 1.155611684603558, + "grad_norm": 2.077686309814453, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.711910605430603, + "num_tokens": 270229792.0, + "step": 10523 + }, + { + "epoch": 1.1557215023061718, + "grad_norm": 2.2532968521118164, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.712145209312439, + "num_tokens": 270255054.0, + "step": 10524 + }, + { + "epoch": 1.1558313200087853, + "grad_norm": 2.3103461265563965, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7179169654846191, + "num_tokens": 270279240.0, + "step": 10525 + }, + { + "epoch": 1.155941137711399, + "grad_norm": 2.086374521255493, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7143198251724243, + "num_tokens": 270308858.0, + "step": 10526 + }, + { + "epoch": 1.1560509554140128, + "grad_norm": 2.3591346740722656, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6989537477493286, + "num_tokens": 270331682.0, + "step": 10527 + }, + { + "epoch": 1.1561607731166264, + "grad_norm": 2.553640604019165, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.720856785774231, + "num_tokens": 270353132.0, + "step": 10528 + }, + { + "epoch": 1.15627059081924, + "grad_norm": 2.3374183177948, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.704028308391571, + "num_tokens": 270376948.0, + "step": 10529 + }, + { + "epoch": 1.1563804085218536, + "grad_norm": 2.3709511756896973, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7250186204910278, + "num_tokens": 270399700.0, + "step": 10530 + }, + { + "epoch": 1.1564902262244674, + "grad_norm": 2.4009339809417725, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7198591232299805, + "num_tokens": 270422806.0, + "step": 10531 + }, + { + "epoch": 1.1566000439270812, + "grad_norm": 2.3485121726989746, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7021464705467224, + "num_tokens": 270447398.0, + "step": 10532 + }, + { + "epoch": 1.1567098616296947, + "grad_norm": 2.3408515453338623, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7378987073898315, + "num_tokens": 270470796.0, + "step": 10533 + }, + { + "epoch": 1.1568196793323084, + "grad_norm": 2.1110663414001465, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7051819562911987, + "num_tokens": 270499294.0, + "step": 10534 + }, + { + "epoch": 1.156929497034922, + "grad_norm": 2.249889612197876, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.739407479763031, + "num_tokens": 270522697.0, + "step": 10535 + }, + { + "epoch": 1.1570393147375357, + "grad_norm": 2.1753311157226562, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7068185210227966, + "num_tokens": 270551355.0, + "step": 10536 + }, + { + "epoch": 1.1571491324401493, + "grad_norm": 2.25559401512146, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7100723385810852, + "num_tokens": 270577525.0, + "step": 10537 + }, + { + "epoch": 1.157258950142763, + "grad_norm": 2.2682747840881348, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7529686689376831, + "num_tokens": 270600762.0, + "step": 10538 + }, + { + "epoch": 1.1573687678453766, + "grad_norm": 2.410365343093872, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7257871627807617, + "num_tokens": 270624759.0, + "step": 10539 + }, + { + "epoch": 1.1574785855479903, + "grad_norm": 2.453455924987793, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7401751279830933, + "num_tokens": 270646301.0, + "step": 10540 + }, + { + "epoch": 1.157588403250604, + "grad_norm": 2.453381061553955, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7406136989593506, + "num_tokens": 270667016.0, + "step": 10541 + }, + { + "epoch": 1.1576982209532176, + "grad_norm": 2.339409828186035, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7183573246002197, + "num_tokens": 270690569.0, + "step": 10542 + }, + { + "epoch": 1.1578080386558314, + "grad_norm": 2.469980001449585, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7333451509475708, + "num_tokens": 270715059.0, + "step": 10543 + }, + { + "epoch": 1.157917856358445, + "grad_norm": 2.092453718185425, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.708383321762085, + "num_tokens": 270745233.0, + "step": 10544 + }, + { + "epoch": 1.1580276740610587, + "grad_norm": 2.1230340003967285, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.6994251012802124, + "num_tokens": 270772982.0, + "step": 10545 + }, + { + "epoch": 1.1581374917636724, + "grad_norm": 2.1318488121032715, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.746758222579956, + "num_tokens": 270799602.0, + "step": 10546 + }, + { + "epoch": 1.158247309466286, + "grad_norm": 2.210392713546753, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7200678586959839, + "num_tokens": 270827702.0, + "step": 10547 + }, + { + "epoch": 1.1583571271688997, + "grad_norm": 2.306778907775879, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7095325589179993, + "num_tokens": 270852439.0, + "step": 10548 + }, + { + "epoch": 1.1584669448715132, + "grad_norm": 2.1713805198669434, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7211850881576538, + "num_tokens": 270880059.0, + "step": 10549 + }, + { + "epoch": 1.158576762574127, + "grad_norm": 2.271486759185791, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7177448272705078, + "num_tokens": 270905926.0, + "step": 10550 + }, + { + "epoch": 1.1586865802767405, + "grad_norm": 2.3790996074676514, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7271497249603271, + "num_tokens": 270930753.0, + "step": 10551 + }, + { + "epoch": 1.1587963979793543, + "grad_norm": 2.3162128925323486, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7429381608963013, + "num_tokens": 270954210.0, + "step": 10552 + }, + { + "epoch": 1.1589062156819678, + "grad_norm": 2.179292917251587, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.715419352054596, + "num_tokens": 270981525.0, + "step": 10553 + }, + { + "epoch": 1.1590160333845816, + "grad_norm": 2.209822177886963, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7215220928192139, + "num_tokens": 271006806.0, + "step": 10554 + }, + { + "epoch": 1.1591258510871953, + "grad_norm": 2.5293755531311035, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7429796457290649, + "num_tokens": 271027007.0, + "step": 10555 + }, + { + "epoch": 1.1592356687898089, + "grad_norm": 2.13677978515625, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7296637296676636, + "num_tokens": 271055017.0, + "step": 10556 + }, + { + "epoch": 1.1593454864924226, + "grad_norm": 2.3156440258026123, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.711006760597229, + "num_tokens": 271079612.0, + "step": 10557 + }, + { + "epoch": 1.1594553041950362, + "grad_norm": 2.116468667984009, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.6983659267425537, + "num_tokens": 271107265.0, + "step": 10558 + }, + { + "epoch": 1.15956512189765, + "grad_norm": 2.4050698280334473, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7457975745201111, + "num_tokens": 271130195.0, + "step": 10559 + }, + { + "epoch": 1.1596749396002635, + "grad_norm": 2.387014865875244, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7215369939804077, + "num_tokens": 271154941.0, + "step": 10560 + }, + { + "epoch": 1.1597847573028772, + "grad_norm": 2.4421494007110596, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7491776943206787, + "num_tokens": 271178064.0, + "step": 10561 + }, + { + "epoch": 1.159894575005491, + "grad_norm": 2.4618852138519287, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7442018985748291, + "num_tokens": 271201257.0, + "step": 10562 + }, + { + "epoch": 1.1600043927081045, + "grad_norm": 2.368621826171875, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7236120104789734, + "num_tokens": 271224708.0, + "step": 10563 + }, + { + "epoch": 1.1601142104107183, + "grad_norm": 2.231867551803589, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7103797793388367, + "num_tokens": 271251009.0, + "step": 10564 + }, + { + "epoch": 1.1602240281133318, + "grad_norm": 2.203808546066284, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.719645619392395, + "num_tokens": 271278238.0, + "step": 10565 + }, + { + "epoch": 1.1603338458159456, + "grad_norm": 2.1871960163116455, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6920976638793945, + "num_tokens": 271306153.0, + "step": 10566 + }, + { + "epoch": 1.160443663518559, + "grad_norm": 2.192902088165283, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7211697101593018, + "num_tokens": 271335025.0, + "step": 10567 + }, + { + "epoch": 1.1605534812211729, + "grad_norm": 2.5935235023498535, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7182009220123291, + "num_tokens": 271356350.0, + "step": 10568 + }, + { + "epoch": 1.1606632989237866, + "grad_norm": 2.1664438247680664, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7194592952728271, + "num_tokens": 271383360.0, + "step": 10569 + }, + { + "epoch": 1.1607731166264001, + "grad_norm": 2.4819066524505615, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7212948799133301, + "num_tokens": 271403419.0, + "step": 10570 + }, + { + "epoch": 1.160882934329014, + "grad_norm": 2.1123664379119873, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7467453479766846, + "num_tokens": 271430339.0, + "step": 10571 + }, + { + "epoch": 1.1609927520316274, + "grad_norm": 2.2723355293273926, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6978362202644348, + "num_tokens": 271455289.0, + "step": 10572 + }, + { + "epoch": 1.1611025697342412, + "grad_norm": 2.1817471981048584, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7210874557495117, + "num_tokens": 271481573.0, + "step": 10573 + }, + { + "epoch": 1.1612123874368547, + "grad_norm": 2.322911500930786, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.747265100479126, + "num_tokens": 271504575.0, + "step": 10574 + }, + { + "epoch": 1.1613222051394685, + "grad_norm": 2.1412103176116943, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7293117046356201, + "num_tokens": 271531630.0, + "step": 10575 + }, + { + "epoch": 1.161432022842082, + "grad_norm": 1.9617985486984253, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7118216753005981, + "num_tokens": 271565325.0, + "step": 10576 + }, + { + "epoch": 1.1615418405446958, + "grad_norm": 2.294829845428467, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.741748571395874, + "num_tokens": 271588315.0, + "step": 10577 + }, + { + "epoch": 1.1616516582473095, + "grad_norm": 2.4415693283081055, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7461757659912109, + "num_tokens": 271610356.0, + "step": 10578 + }, + { + "epoch": 1.161761475949923, + "grad_norm": 2.074005365371704, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7051704525947571, + "num_tokens": 271640771.0, + "step": 10579 + }, + { + "epoch": 1.1618712936525368, + "grad_norm": 2.2488064765930176, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.714678168296814, + "num_tokens": 271668069.0, + "step": 10580 + }, + { + "epoch": 1.1619811113551504, + "grad_norm": 2.0373787879943848, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7084596753120422, + "num_tokens": 271698932.0, + "step": 10581 + }, + { + "epoch": 1.1620909290577641, + "grad_norm": 2.1762073040008545, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7228772640228271, + "num_tokens": 271723964.0, + "step": 10582 + }, + { + "epoch": 1.1622007467603779, + "grad_norm": 2.583331346511841, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7515643239021301, + "num_tokens": 271742265.0, + "step": 10583 + }, + { + "epoch": 1.1623105644629914, + "grad_norm": 2.4897375106811523, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7272434234619141, + "num_tokens": 271764299.0, + "step": 10584 + }, + { + "epoch": 1.1624203821656052, + "grad_norm": 2.4785103797912598, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7402009963989258, + "num_tokens": 271785659.0, + "step": 10585 + }, + { + "epoch": 1.1625301998682187, + "grad_norm": 2.3224029541015625, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6911742687225342, + "num_tokens": 271813667.0, + "step": 10586 + }, + { + "epoch": 1.1626400175708325, + "grad_norm": 2.271162509918213, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7366570830345154, + "num_tokens": 271836290.0, + "step": 10587 + }, + { + "epoch": 1.162749835273446, + "grad_norm": 2.2510478496551514, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7120437622070312, + "num_tokens": 271863372.0, + "step": 10588 + }, + { + "epoch": 1.1628596529760598, + "grad_norm": 2.0708022117614746, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.728724479675293, + "num_tokens": 271892585.0, + "step": 10589 + }, + { + "epoch": 1.1629694706786733, + "grad_norm": 2.3154823780059814, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7014236450195312, + "num_tokens": 271917251.0, + "step": 10590 + }, + { + "epoch": 1.163079288381287, + "grad_norm": 2.263465642929077, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7094640731811523, + "num_tokens": 271942195.0, + "step": 10591 + }, + { + "epoch": 1.1631891060839008, + "grad_norm": 2.2104122638702393, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7192968130111694, + "num_tokens": 271967910.0, + "step": 10592 + }, + { + "epoch": 1.1632989237865143, + "grad_norm": 2.147834539413452, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7209022045135498, + "num_tokens": 271994951.0, + "step": 10593 + }, + { + "epoch": 1.163408741489128, + "grad_norm": 2.2357020378112793, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7302639484405518, + "num_tokens": 272019251.0, + "step": 10594 + }, + { + "epoch": 1.1635185591917416, + "grad_norm": 2.3626887798309326, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7350600957870483, + "num_tokens": 272040958.0, + "step": 10595 + }, + { + "epoch": 1.1636283768943554, + "grad_norm": 2.233116865158081, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.709985613822937, + "num_tokens": 272065841.0, + "step": 10596 + }, + { + "epoch": 1.1637381945969691, + "grad_norm": 2.5782153606414795, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7471216917037964, + "num_tokens": 272084969.0, + "step": 10597 + }, + { + "epoch": 1.1638480122995827, + "grad_norm": 2.4135239124298096, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7140885591506958, + "num_tokens": 272106409.0, + "step": 10598 + }, + { + "epoch": 1.1639578300021964, + "grad_norm": 2.3773083686828613, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.717205822467804, + "num_tokens": 272129785.0, + "step": 10599 + }, + { + "epoch": 1.16406764770481, + "grad_norm": 2.5037662982940674, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7340347170829773, + "num_tokens": 272151749.0, + "step": 10600 + }, + { + "epoch": 1.1641774654074237, + "grad_norm": 2.5285260677337646, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.70613694190979, + "num_tokens": 272174030.0, + "step": 10601 + }, + { + "epoch": 1.1642872831100373, + "grad_norm": 2.1997737884521484, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7441585063934326, + "num_tokens": 272199704.0, + "step": 10602 + }, + { + "epoch": 1.164397100812651, + "grad_norm": 2.098418712615967, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7068612575531006, + "num_tokens": 272228983.0, + "step": 10603 + }, + { + "epoch": 1.1645069185152646, + "grad_norm": 2.119337320327759, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7209672331809998, + "num_tokens": 272257169.0, + "step": 10604 + }, + { + "epoch": 1.1646167362178783, + "grad_norm": 2.1876869201660156, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7254073619842529, + "num_tokens": 272282757.0, + "step": 10605 + }, + { + "epoch": 1.164726553920492, + "grad_norm": 2.004868745803833, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7509717345237732, + "num_tokens": 272309532.0, + "step": 10606 + }, + { + "epoch": 1.1648363716231056, + "grad_norm": 2.5192203521728516, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7411051988601685, + "num_tokens": 272331001.0, + "step": 10607 + }, + { + "epoch": 1.1649461893257194, + "grad_norm": 2.1665689945220947, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6882411241531372, + "num_tokens": 272358927.0, + "step": 10608 + }, + { + "epoch": 1.165056007028333, + "grad_norm": 2.2517099380493164, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7053771018981934, + "num_tokens": 272383934.0, + "step": 10609 + }, + { + "epoch": 1.1651658247309467, + "grad_norm": 2.3023874759674072, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.734215259552002, + "num_tokens": 272406608.0, + "step": 10610 + }, + { + "epoch": 1.1652756424335604, + "grad_norm": 2.5285675525665283, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7459962368011475, + "num_tokens": 272427355.0, + "step": 10611 + }, + { + "epoch": 1.165385460136174, + "grad_norm": 2.3441898822784424, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7197901606559753, + "num_tokens": 272452180.0, + "step": 10612 + }, + { + "epoch": 1.1654952778387877, + "grad_norm": 2.5856311321258545, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7390875816345215, + "num_tokens": 272471894.0, + "step": 10613 + }, + { + "epoch": 1.1656050955414012, + "grad_norm": 2.177447557449341, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7349671721458435, + "num_tokens": 272498245.0, + "step": 10614 + }, + { + "epoch": 1.165714913244015, + "grad_norm": 2.4009995460510254, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7270758152008057, + "num_tokens": 272521088.0, + "step": 10615 + }, + { + "epoch": 1.1658247309466285, + "grad_norm": 2.3140761852264404, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7389097213745117, + "num_tokens": 272544557.0, + "step": 10616 + }, + { + "epoch": 1.1659345486492423, + "grad_norm": 2.422090530395508, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7340525388717651, + "num_tokens": 272565497.0, + "step": 10617 + }, + { + "epoch": 1.1660443663518558, + "grad_norm": 2.49272084236145, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7334892749786377, + "num_tokens": 272586580.0, + "step": 10618 + }, + { + "epoch": 1.1661541840544696, + "grad_norm": 2.2319791316986084, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7422869205474854, + "num_tokens": 272613466.0, + "step": 10619 + }, + { + "epoch": 1.1662640017570833, + "grad_norm": 2.5637824535369873, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7069613933563232, + "num_tokens": 272637513.0, + "step": 10620 + }, + { + "epoch": 1.1663738194596969, + "grad_norm": 2.2320923805236816, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7202088832855225, + "num_tokens": 272663150.0, + "step": 10621 + }, + { + "epoch": 1.1664836371623106, + "grad_norm": 2.0607705116271973, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7304105758666992, + "num_tokens": 272691506.0, + "step": 10622 + }, + { + "epoch": 1.1665934548649242, + "grad_norm": 2.300912857055664, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7368819117546082, + "num_tokens": 272714954.0, + "step": 10623 + }, + { + "epoch": 1.166703272567538, + "grad_norm": 2.167290210723877, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7187804579734802, + "num_tokens": 272740300.0, + "step": 10624 + }, + { + "epoch": 1.1668130902701515, + "grad_norm": 2.3416805267333984, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.730053186416626, + "num_tokens": 272763855.0, + "step": 10625 + }, + { + "epoch": 1.1669229079727652, + "grad_norm": 2.265122652053833, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6966490745544434, + "num_tokens": 272789803.0, + "step": 10626 + }, + { + "epoch": 1.167032725675379, + "grad_norm": 2.180237054824829, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7091888785362244, + "num_tokens": 272818090.0, + "step": 10627 + }, + { + "epoch": 1.1671425433779925, + "grad_norm": 2.029448986053467, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7233841419219971, + "num_tokens": 272847981.0, + "step": 10628 + }, + { + "epoch": 1.1672523610806063, + "grad_norm": 2.007185459136963, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7102905511856079, + "num_tokens": 272879526.0, + "step": 10629 + }, + { + "epoch": 1.1673621787832198, + "grad_norm": 2.6779661178588867, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.719092071056366, + "num_tokens": 272899533.0, + "step": 10630 + }, + { + "epoch": 1.1674719964858336, + "grad_norm": 2.2499709129333496, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7068873047828674, + "num_tokens": 272926250.0, + "step": 10631 + }, + { + "epoch": 1.167581814188447, + "grad_norm": 2.19683837890625, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7260429263114929, + "num_tokens": 272952530.0, + "step": 10632 + }, + { + "epoch": 1.1676916318910608, + "grad_norm": 2.100369453430176, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7179331183433533, + "num_tokens": 272980587.0, + "step": 10633 + }, + { + "epoch": 1.1678014495936746, + "grad_norm": 2.329085350036621, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7288155555725098, + "num_tokens": 273004148.0, + "step": 10634 + }, + { + "epoch": 1.1679112672962881, + "grad_norm": 2.434343099594116, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7224030494689941, + "num_tokens": 273025835.0, + "step": 10635 + }, + { + "epoch": 1.168021084998902, + "grad_norm": 2.2202320098876953, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7031832933425903, + "num_tokens": 273053067.0, + "step": 10636 + }, + { + "epoch": 1.1681309027015154, + "grad_norm": 2.0761144161224365, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.722774863243103, + "num_tokens": 273081297.0, + "step": 10637 + }, + { + "epoch": 1.1682407204041292, + "grad_norm": 2.0684053897857666, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7147319316864014, + "num_tokens": 273109180.0, + "step": 10638 + }, + { + "epoch": 1.1683505381067427, + "grad_norm": 2.45723032951355, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.743415355682373, + "num_tokens": 273129844.0, + "step": 10639 + }, + { + "epoch": 1.1684603558093565, + "grad_norm": 2.34167218208313, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7034639120101929, + "num_tokens": 273154698.0, + "step": 10640 + }, + { + "epoch": 1.16857017351197, + "grad_norm": 2.3410890102386475, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7104998230934143, + "num_tokens": 273181269.0, + "step": 10641 + }, + { + "epoch": 1.1686799912145838, + "grad_norm": 2.6650443077087402, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7171380519866943, + "num_tokens": 273200245.0, + "step": 10642 + }, + { + "epoch": 1.1687898089171975, + "grad_norm": 2.071202039718628, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7091894149780273, + "num_tokens": 273231104.0, + "step": 10643 + }, + { + "epoch": 1.168899626619811, + "grad_norm": 2.2158303260803223, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.724831223487854, + "num_tokens": 273255181.0, + "step": 10644 + }, + { + "epoch": 1.1690094443224248, + "grad_norm": 2.196718692779541, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.683489203453064, + "num_tokens": 273285651.0, + "step": 10645 + }, + { + "epoch": 1.1691192620250384, + "grad_norm": 2.446645975112915, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7275818586349487, + "num_tokens": 273307039.0, + "step": 10646 + }, + { + "epoch": 1.169229079727652, + "grad_norm": 1.9433457851409912, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6935299634933472, + "num_tokens": 273340897.0, + "step": 10647 + }, + { + "epoch": 1.1693388974302659, + "grad_norm": 2.2482175827026367, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7227993011474609, + "num_tokens": 273366306.0, + "step": 10648 + }, + { + "epoch": 1.1694487151328794, + "grad_norm": 2.411836862564087, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7142572402954102, + "num_tokens": 273390034.0, + "step": 10649 + }, + { + "epoch": 1.1695585328354932, + "grad_norm": 2.610464096069336, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.752944827079773, + "num_tokens": 273409604.0, + "step": 10650 + }, + { + "epoch": 1.1696683505381067, + "grad_norm": 2.0734920501708984, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.713873028755188, + "num_tokens": 273440163.0, + "step": 10651 + }, + { + "epoch": 1.1697781682407205, + "grad_norm": 2.0456337928771973, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.728055477142334, + "num_tokens": 273470577.0, + "step": 10652 + }, + { + "epoch": 1.169887985943334, + "grad_norm": 2.24934720993042, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7311768531799316, + "num_tokens": 273504475.0, + "step": 10653 + }, + { + "epoch": 1.1699978036459477, + "grad_norm": 2.2895524501800537, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7173291444778442, + "num_tokens": 273531468.0, + "step": 10654 + }, + { + "epoch": 1.1701076213485613, + "grad_norm": 2.4434738159179688, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.741161584854126, + "num_tokens": 273553469.0, + "step": 10655 + }, + { + "epoch": 1.170217439051175, + "grad_norm": 2.1960737705230713, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7157254219055176, + "num_tokens": 273580761.0, + "step": 10656 + }, + { + "epoch": 1.1703272567537888, + "grad_norm": 2.2299644947052, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7295525074005127, + "num_tokens": 273604853.0, + "step": 10657 + }, + { + "epoch": 1.1704370744564023, + "grad_norm": 2.357433795928955, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7385563850402832, + "num_tokens": 273627818.0, + "step": 10658 + }, + { + "epoch": 1.170546892159016, + "grad_norm": 2.682589054107666, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7009783983230591, + "num_tokens": 273650226.0, + "step": 10659 + }, + { + "epoch": 1.1706567098616296, + "grad_norm": 2.5247302055358887, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7278085947036743, + "num_tokens": 273671233.0, + "step": 10660 + }, + { + "epoch": 1.1707665275642434, + "grad_norm": 2.2251081466674805, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7291127443313599, + "num_tokens": 273696510.0, + "step": 10661 + }, + { + "epoch": 1.1708763452668571, + "grad_norm": 2.172973155975342, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.725045919418335, + "num_tokens": 273721919.0, + "step": 10662 + }, + { + "epoch": 1.1709861629694707, + "grad_norm": 2.2353861331939697, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7127254605293274, + "num_tokens": 273747149.0, + "step": 10663 + }, + { + "epoch": 1.1710959806720844, + "grad_norm": 2.715118169784546, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7247954607009888, + "num_tokens": 273766841.0, + "step": 10664 + }, + { + "epoch": 1.171205798374698, + "grad_norm": 2.4297335147857666, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7281248569488525, + "num_tokens": 273788015.0, + "step": 10665 + }, + { + "epoch": 1.1713156160773117, + "grad_norm": 2.1259138584136963, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6864996552467346, + "num_tokens": 273816849.0, + "step": 10666 + }, + { + "epoch": 1.1714254337799253, + "grad_norm": 2.035125494003296, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6924408078193665, + "num_tokens": 273851609.0, + "step": 10667 + }, + { + "epoch": 1.171535251482539, + "grad_norm": 2.1944220066070557, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7063040137290955, + "num_tokens": 273878926.0, + "step": 10668 + }, + { + "epoch": 1.1716450691851525, + "grad_norm": 2.194249391555786, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7332923412322998, + "num_tokens": 273905389.0, + "step": 10669 + }, + { + "epoch": 1.1717548868877663, + "grad_norm": 2.118774890899658, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7331342697143555, + "num_tokens": 273930876.0, + "step": 10670 + }, + { + "epoch": 1.17186470459038, + "grad_norm": 2.520282745361328, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7427000999450684, + "num_tokens": 273951447.0, + "step": 10671 + }, + { + "epoch": 1.1719745222929936, + "grad_norm": 2.3236329555511475, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6971627473831177, + "num_tokens": 273977943.0, + "step": 10672 + }, + { + "epoch": 1.1720843399956073, + "grad_norm": 2.1124463081359863, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.705649733543396, + "num_tokens": 274005741.0, + "step": 10673 + }, + { + "epoch": 1.1721941576982209, + "grad_norm": 2.010709285736084, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7274250984191895, + "num_tokens": 274036277.0, + "step": 10674 + }, + { + "epoch": 1.1723039754008346, + "grad_norm": 2.270331621170044, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7101777195930481, + "num_tokens": 274060644.0, + "step": 10675 + }, + { + "epoch": 1.1724137931034484, + "grad_norm": 2.4118571281433105, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7499274015426636, + "num_tokens": 274081630.0, + "step": 10676 + }, + { + "epoch": 1.172523610806062, + "grad_norm": 2.04252552986145, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7146022915840149, + "num_tokens": 274113670.0, + "step": 10677 + }, + { + "epoch": 1.1726334285086757, + "grad_norm": 2.3610355854034424, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7098596096038818, + "num_tokens": 274138166.0, + "step": 10678 + }, + { + "epoch": 1.1727432462112892, + "grad_norm": 2.3741581439971924, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7208090424537659, + "num_tokens": 274162030.0, + "step": 10679 + }, + { + "epoch": 1.172853063913903, + "grad_norm": 2.157550573348999, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7020789980888367, + "num_tokens": 274191228.0, + "step": 10680 + }, + { + "epoch": 1.1729628816165165, + "grad_norm": 2.39982271194458, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.731991708278656, + "num_tokens": 274216077.0, + "step": 10681 + }, + { + "epoch": 1.1730726993191303, + "grad_norm": 2.216487407684326, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7413488626480103, + "num_tokens": 274239967.0, + "step": 10682 + }, + { + "epoch": 1.1731825170217438, + "grad_norm": 2.1838250160217285, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7188013195991516, + "num_tokens": 274266512.0, + "step": 10683 + }, + { + "epoch": 1.1732923347243576, + "grad_norm": 2.2678675651550293, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.711470901966095, + "num_tokens": 274291911.0, + "step": 10684 + }, + { + "epoch": 1.1734021524269713, + "grad_norm": 2.303086757659912, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7236006259918213, + "num_tokens": 274319024.0, + "step": 10685 + }, + { + "epoch": 1.1735119701295849, + "grad_norm": 2.40095853805542, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7051159143447876, + "num_tokens": 274343208.0, + "step": 10686 + }, + { + "epoch": 1.1736217878321986, + "grad_norm": 2.2738046646118164, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7314703464508057, + "num_tokens": 274366337.0, + "step": 10687 + }, + { + "epoch": 1.1737316055348122, + "grad_norm": 2.186100482940674, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6955369710922241, + "num_tokens": 274393139.0, + "step": 10688 + }, + { + "epoch": 1.173841423237426, + "grad_norm": 2.1351537704467773, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7510759830474854, + "num_tokens": 274417722.0, + "step": 10689 + }, + { + "epoch": 1.1739512409400394, + "grad_norm": 2.324575185775757, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7310376763343811, + "num_tokens": 274440871.0, + "step": 10690 + }, + { + "epoch": 1.1740610586426532, + "grad_norm": 2.0755531787872314, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.707036018371582, + "num_tokens": 274468713.0, + "step": 10691 + }, + { + "epoch": 1.1741708763452667, + "grad_norm": 2.332498550415039, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7371379137039185, + "num_tokens": 274492775.0, + "step": 10692 + }, + { + "epoch": 1.1742806940478805, + "grad_norm": 2.4791600704193115, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7403289675712585, + "num_tokens": 274515126.0, + "step": 10693 + }, + { + "epoch": 1.1743905117504942, + "grad_norm": 2.2567150592803955, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7109358906745911, + "num_tokens": 274541254.0, + "step": 10694 + }, + { + "epoch": 1.1745003294531078, + "grad_norm": 2.2314810752868652, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7086347341537476, + "num_tokens": 274569298.0, + "step": 10695 + }, + { + "epoch": 1.1746101471557215, + "grad_norm": 2.408482789993286, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7075139284133911, + "num_tokens": 274593347.0, + "step": 10696 + }, + { + "epoch": 1.174719964858335, + "grad_norm": 1.9722908735275269, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6925920844078064, + "num_tokens": 274624938.0, + "step": 10697 + }, + { + "epoch": 1.1748297825609488, + "grad_norm": 1.902270793914795, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7175458669662476, + "num_tokens": 274660125.0, + "step": 10698 + }, + { + "epoch": 1.1749396002635626, + "grad_norm": 2.045947313308716, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.716431200504303, + "num_tokens": 274691803.0, + "step": 10699 + }, + { + "epoch": 1.1750494179661761, + "grad_norm": 1.999112844467163, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7051034569740295, + "num_tokens": 274721664.0, + "step": 10700 + }, + { + "epoch": 1.1751592356687899, + "grad_norm": 2.3383045196533203, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7277352213859558, + "num_tokens": 274745176.0, + "step": 10701 + }, + { + "epoch": 1.1752690533714034, + "grad_norm": 2.4138400554656982, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7460517883300781, + "num_tokens": 274766569.0, + "step": 10702 + }, + { + "epoch": 1.1753788710740172, + "grad_norm": 2.1164629459381104, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.729520857334137, + "num_tokens": 274793225.0, + "step": 10703 + }, + { + "epoch": 1.1754886887766307, + "grad_norm": 2.1843981742858887, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.6980831623077393, + "num_tokens": 274820306.0, + "step": 10704 + }, + { + "epoch": 1.1755985064792445, + "grad_norm": 2.4436421394348145, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7253259420394897, + "num_tokens": 274843458.0, + "step": 10705 + }, + { + "epoch": 1.175708324181858, + "grad_norm": 2.1516082286834717, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7384838461875916, + "num_tokens": 274870820.0, + "step": 10706 + }, + { + "epoch": 1.1758181418844718, + "grad_norm": 2.1827242374420166, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7205991744995117, + "num_tokens": 274898463.0, + "step": 10707 + }, + { + "epoch": 1.1759279595870855, + "grad_norm": 2.3448166847229004, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.743649959564209, + "num_tokens": 274920161.0, + "step": 10708 + }, + { + "epoch": 1.176037777289699, + "grad_norm": 2.6183316707611084, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7231010794639587, + "num_tokens": 274940135.0, + "step": 10709 + }, + { + "epoch": 1.1761475949923128, + "grad_norm": 2.130981683731079, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7023214101791382, + "num_tokens": 274968525.0, + "step": 10710 + }, + { + "epoch": 1.1762574126949263, + "grad_norm": 1.9952908754348755, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7237396836280823, + "num_tokens": 274999098.0, + "step": 10711 + }, + { + "epoch": 1.17636723039754, + "grad_norm": 2.5264673233032227, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7107826471328735, + "num_tokens": 275020333.0, + "step": 10712 + }, + { + "epoch": 1.1764770481001539, + "grad_norm": 2.0802628993988037, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7320224642753601, + "num_tokens": 275049893.0, + "step": 10713 + }, + { + "epoch": 1.1765868658027674, + "grad_norm": 2.7665603160858154, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7197559475898743, + "num_tokens": 275068599.0, + "step": 10714 + }, + { + "epoch": 1.1766966835053811, + "grad_norm": 2.0289597511291504, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7093876004219055, + "num_tokens": 275097790.0, + "step": 10715 + }, + { + "epoch": 1.1768065012079947, + "grad_norm": 2.3564414978027344, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7184383273124695, + "num_tokens": 275122679.0, + "step": 10716 + }, + { + "epoch": 1.1769163189106084, + "grad_norm": 2.1759555339813232, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7155654430389404, + "num_tokens": 275148901.0, + "step": 10717 + }, + { + "epoch": 1.177026136613222, + "grad_norm": 2.624849557876587, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7270200848579407, + "num_tokens": 275169478.0, + "step": 10718 + }, + { + "epoch": 1.1771359543158357, + "grad_norm": 2.4139556884765625, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7211886644363403, + "num_tokens": 275193075.0, + "step": 10719 + }, + { + "epoch": 1.1772457720184493, + "grad_norm": 2.268122673034668, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7078548669815063, + "num_tokens": 275218705.0, + "step": 10720 + }, + { + "epoch": 1.177355589721063, + "grad_norm": 2.35850191116333, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7220787405967712, + "num_tokens": 275240702.0, + "step": 10721 + }, + { + "epoch": 1.1774654074236768, + "grad_norm": 2.3243207931518555, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6983445882797241, + "num_tokens": 275266778.0, + "step": 10722 + }, + { + "epoch": 1.1775752251262903, + "grad_norm": 2.2735660076141357, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7404932975769043, + "num_tokens": 275293556.0, + "step": 10723 + }, + { + "epoch": 1.177685042828904, + "grad_norm": 2.1138296127319336, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7314437031745911, + "num_tokens": 275321791.0, + "step": 10724 + }, + { + "epoch": 1.1777948605315176, + "grad_norm": 2.363276481628418, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7275530099868774, + "num_tokens": 275345916.0, + "step": 10725 + }, + { + "epoch": 1.1779046782341314, + "grad_norm": 2.372596025466919, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.719131588935852, + "num_tokens": 275368934.0, + "step": 10726 + }, + { + "epoch": 1.1780144959367451, + "grad_norm": 2.4077298641204834, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7043306231498718, + "num_tokens": 275395749.0, + "step": 10727 + }, + { + "epoch": 1.1781243136393587, + "grad_norm": 2.10160231590271, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7126023769378662, + "num_tokens": 275424501.0, + "step": 10728 + }, + { + "epoch": 1.1782341313419724, + "grad_norm": 2.1963791847229004, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7617532014846802, + "num_tokens": 275449870.0, + "step": 10729 + }, + { + "epoch": 1.178343949044586, + "grad_norm": 2.127959966659546, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7532211542129517, + "num_tokens": 275474290.0, + "step": 10730 + }, + { + "epoch": 1.1784537667471997, + "grad_norm": 2.185453414916992, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7283828854560852, + "num_tokens": 275500983.0, + "step": 10731 + }, + { + "epoch": 1.1785635844498132, + "grad_norm": 2.3076517581939697, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7121245265007019, + "num_tokens": 275526170.0, + "step": 10732 + }, + { + "epoch": 1.178673402152427, + "grad_norm": 2.221209764480591, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7045434713363647, + "num_tokens": 275554542.0, + "step": 10733 + }, + { + "epoch": 1.1787832198550405, + "grad_norm": 2.2798027992248535, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7150300741195679, + "num_tokens": 275579821.0, + "step": 10734 + }, + { + "epoch": 1.1788930375576543, + "grad_norm": 2.3428988456726074, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7076635360717773, + "num_tokens": 275604834.0, + "step": 10735 + }, + { + "epoch": 1.179002855260268, + "grad_norm": 2.3844776153564453, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7285835146903992, + "num_tokens": 275630335.0, + "step": 10736 + }, + { + "epoch": 1.1791126729628816, + "grad_norm": 2.2699077129364014, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7236417531967163, + "num_tokens": 275654481.0, + "step": 10737 + }, + { + "epoch": 1.1792224906654953, + "grad_norm": 2.7427961826324463, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7198131084442139, + "num_tokens": 275673434.0, + "step": 10738 + }, + { + "epoch": 1.1793323083681089, + "grad_norm": 1.9502265453338623, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7166352868080139, + "num_tokens": 275703751.0, + "step": 10739 + }, + { + "epoch": 1.1794421260707226, + "grad_norm": 1.9469871520996094, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7178224325180054, + "num_tokens": 275733892.0, + "step": 10740 + }, + { + "epoch": 1.1795519437733364, + "grad_norm": 1.9042909145355225, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6964015960693359, + "num_tokens": 275769645.0, + "step": 10741 + }, + { + "epoch": 1.17966176147595, + "grad_norm": 2.3798892498016357, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7076945304870605, + "num_tokens": 275795763.0, + "step": 10742 + }, + { + "epoch": 1.1797715791785637, + "grad_norm": 2.35896897315979, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7425529956817627, + "num_tokens": 275818260.0, + "step": 10743 + }, + { + "epoch": 1.1798813968811772, + "grad_norm": 2.415079355239868, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7142312526702881, + "num_tokens": 275841932.0, + "step": 10744 + }, + { + "epoch": 1.179991214583791, + "grad_norm": 2.5621275901794434, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7272419333457947, + "num_tokens": 275861564.0, + "step": 10745 + }, + { + "epoch": 1.1801010322864045, + "grad_norm": 2.442694902420044, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.734229326248169, + "num_tokens": 275883458.0, + "step": 10746 + }, + { + "epoch": 1.1802108499890183, + "grad_norm": 1.9809300899505615, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7012661099433899, + "num_tokens": 275916830.0, + "step": 10747 + }, + { + "epoch": 1.1803206676916318, + "grad_norm": 1.9712575674057007, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7233834266662598, + "num_tokens": 275948382.0, + "step": 10748 + }, + { + "epoch": 1.1804304853942456, + "grad_norm": 2.497358560562134, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7528916597366333, + "num_tokens": 275966864.0, + "step": 10749 + }, + { + "epoch": 1.1805403030968593, + "grad_norm": 2.262775182723999, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7172260880470276, + "num_tokens": 275991888.0, + "step": 10750 + }, + { + "epoch": 1.1806501207994728, + "grad_norm": 1.9656745195388794, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6985940933227539, + "num_tokens": 276024464.0, + "step": 10751 + }, + { + "epoch": 1.1807599385020866, + "grad_norm": 2.387906789779663, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7359426021575928, + "num_tokens": 276047132.0, + "step": 10752 + }, + { + "epoch": 1.1808697562047001, + "grad_norm": 2.3067915439605713, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7091128826141357, + "num_tokens": 276071180.0, + "step": 10753 + }, + { + "epoch": 1.180979573907314, + "grad_norm": 2.262442111968994, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7018388509750366, + "num_tokens": 276098692.0, + "step": 10754 + }, + { + "epoch": 1.1810893916099274, + "grad_norm": 2.092101573944092, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7072354555130005, + "num_tokens": 276127644.0, + "step": 10755 + }, + { + "epoch": 1.1811992093125412, + "grad_norm": 2.280395984649658, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7320992350578308, + "num_tokens": 276154690.0, + "step": 10756 + }, + { + "epoch": 1.1813090270151547, + "grad_norm": 2.1610751152038574, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7232522964477539, + "num_tokens": 276182582.0, + "step": 10757 + }, + { + "epoch": 1.1814188447177685, + "grad_norm": 2.2888643741607666, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7373658418655396, + "num_tokens": 276207477.0, + "step": 10758 + }, + { + "epoch": 1.1815286624203822, + "grad_norm": 1.9722142219543457, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7343540191650391, + "num_tokens": 276237103.0, + "step": 10759 + }, + { + "epoch": 1.1816384801229958, + "grad_norm": 2.1676278114318848, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6940333247184753, + "num_tokens": 276264827.0, + "step": 10760 + }, + { + "epoch": 1.1817482978256095, + "grad_norm": 2.2015349864959717, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7054164409637451, + "num_tokens": 276292181.0, + "step": 10761 + }, + { + "epoch": 1.181858115528223, + "grad_norm": 1.953162670135498, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7162560820579529, + "num_tokens": 276324139.0, + "step": 10762 + }, + { + "epoch": 1.1819679332308368, + "grad_norm": 2.493281364440918, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7202785611152649, + "num_tokens": 276347058.0, + "step": 10763 + }, + { + "epoch": 1.1820777509334506, + "grad_norm": 2.2451212406158447, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7262948155403137, + "num_tokens": 276370919.0, + "step": 10764 + }, + { + "epoch": 1.1821875686360641, + "grad_norm": 2.205636739730835, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7036154270172119, + "num_tokens": 276398537.0, + "step": 10765 + }, + { + "epoch": 1.1822973863386779, + "grad_norm": 2.299916982650757, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7358479499816895, + "num_tokens": 276421793.0, + "step": 10766 + }, + { + "epoch": 1.1824072040412914, + "grad_norm": 2.524923086166382, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7322752475738525, + "num_tokens": 276442232.0, + "step": 10767 + }, + { + "epoch": 1.1825170217439052, + "grad_norm": 2.2283332347869873, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7102615833282471, + "num_tokens": 276467710.0, + "step": 10768 + }, + { + "epoch": 1.1826268394465187, + "grad_norm": 2.261772632598877, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7309925556182861, + "num_tokens": 276490871.0, + "step": 10769 + }, + { + "epoch": 1.1827366571491325, + "grad_norm": 2.178710460662842, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7176694273948669, + "num_tokens": 276517937.0, + "step": 10770 + }, + { + "epoch": 1.182846474851746, + "grad_norm": 2.281738758087158, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7323175668716431, + "num_tokens": 276542908.0, + "step": 10771 + }, + { + "epoch": 1.1829562925543597, + "grad_norm": 2.018789529800415, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.729432225227356, + "num_tokens": 276572296.0, + "step": 10772 + }, + { + "epoch": 1.1830661102569735, + "grad_norm": 2.5437519550323486, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7109106779098511, + "num_tokens": 276592016.0, + "step": 10773 + }, + { + "epoch": 1.183175927959587, + "grad_norm": 2.2812881469726562, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7151973247528076, + "num_tokens": 276620862.0, + "step": 10774 + }, + { + "epoch": 1.1832857456622008, + "grad_norm": 2.399358034133911, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.706706166267395, + "num_tokens": 276643542.0, + "step": 10775 + }, + { + "epoch": 1.1833955633648143, + "grad_norm": 1.9237346649169922, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6950883269309998, + "num_tokens": 276678748.0, + "step": 10776 + }, + { + "epoch": 1.183505381067428, + "grad_norm": 2.3606960773468018, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7106985449790955, + "num_tokens": 276702519.0, + "step": 10777 + }, + { + "epoch": 1.1836151987700418, + "grad_norm": 2.4085168838500977, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7011740207672119, + "num_tokens": 276725980.0, + "step": 10778 + }, + { + "epoch": 1.1837250164726554, + "grad_norm": 2.28558087348938, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7289503812789917, + "num_tokens": 276751726.0, + "step": 10779 + }, + { + "epoch": 1.1838348341752691, + "grad_norm": 2.0106019973754883, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7203018069267273, + "num_tokens": 276782963.0, + "step": 10780 + }, + { + "epoch": 1.1839446518778827, + "grad_norm": 2.3131322860717773, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.703575611114502, + "num_tokens": 276807956.0, + "step": 10781 + }, + { + "epoch": 1.1840544695804964, + "grad_norm": 2.005174398422241, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.689760684967041, + "num_tokens": 276841619.0, + "step": 10782 + }, + { + "epoch": 1.18416428728311, + "grad_norm": 2.3311192989349365, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6957347393035889, + "num_tokens": 276868037.0, + "step": 10783 + }, + { + "epoch": 1.1842741049857237, + "grad_norm": 2.575923204421997, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7096229791641235, + "num_tokens": 276888684.0, + "step": 10784 + }, + { + "epoch": 1.1843839226883373, + "grad_norm": 2.1459014415740967, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7236445546150208, + "num_tokens": 276915087.0, + "step": 10785 + }, + { + "epoch": 1.184493740390951, + "grad_norm": 2.407069683074951, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7209258079528809, + "num_tokens": 276937406.0, + "step": 10786 + }, + { + "epoch": 1.1846035580935648, + "grad_norm": 1.9665381908416748, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.727044939994812, + "num_tokens": 276970595.0, + "step": 10787 + }, + { + "epoch": 1.1847133757961783, + "grad_norm": 2.2545673847198486, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7132271528244019, + "num_tokens": 276996127.0, + "step": 10788 + }, + { + "epoch": 1.184823193498792, + "grad_norm": 2.370683431625366, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7044475078582764, + "num_tokens": 277021002.0, + "step": 10789 + }, + { + "epoch": 1.1849330112014056, + "grad_norm": 2.419835090637207, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7349761724472046, + "num_tokens": 277043509.0, + "step": 10790 + }, + { + "epoch": 1.1850428289040194, + "grad_norm": 2.1123852729797363, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7157348394393921, + "num_tokens": 277072662.0, + "step": 10791 + }, + { + "epoch": 1.185152646606633, + "grad_norm": 2.305802822113037, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7289323806762695, + "num_tokens": 277095888.0, + "step": 10792 + }, + { + "epoch": 1.1852624643092466, + "grad_norm": 2.24159836769104, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7074985504150391, + "num_tokens": 277121090.0, + "step": 10793 + }, + { + "epoch": 1.1853722820118604, + "grad_norm": 2.351691246032715, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7143932580947876, + "num_tokens": 277144885.0, + "step": 10794 + }, + { + "epoch": 1.185482099714474, + "grad_norm": 2.359208822250366, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7256678938865662, + "num_tokens": 277170328.0, + "step": 10795 + }, + { + "epoch": 1.1855919174170877, + "grad_norm": 2.186969757080078, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7148082256317139, + "num_tokens": 277195748.0, + "step": 10796 + }, + { + "epoch": 1.1857017351197012, + "grad_norm": 2.381669759750366, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.730258047580719, + "num_tokens": 277220242.0, + "step": 10797 + }, + { + "epoch": 1.185811552822315, + "grad_norm": 2.146059274673462, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7244012951850891, + "num_tokens": 277247384.0, + "step": 10798 + }, + { + "epoch": 1.1859213705249285, + "grad_norm": 2.2589452266693115, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7051283121109009, + "num_tokens": 277273648.0, + "step": 10799 + }, + { + "epoch": 1.1860311882275423, + "grad_norm": 2.2651658058166504, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7153834104537964, + "num_tokens": 277301443.0, + "step": 10800 + }, + { + "epoch": 1.186141005930156, + "grad_norm": 2.193737506866455, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7329082489013672, + "num_tokens": 277328331.0, + "step": 10801 + }, + { + "epoch": 1.1862508236327696, + "grad_norm": 2.2560646533966064, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6975287199020386, + "num_tokens": 277358376.0, + "step": 10802 + }, + { + "epoch": 1.1863606413353833, + "grad_norm": 2.415766954421997, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7217242121696472, + "num_tokens": 277381285.0, + "step": 10803 + }, + { + "epoch": 1.1864704590379969, + "grad_norm": 2.093013048171997, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6973323822021484, + "num_tokens": 277411709.0, + "step": 10804 + }, + { + "epoch": 1.1865802767406106, + "grad_norm": 2.3159258365631104, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7359238862991333, + "num_tokens": 277436735.0, + "step": 10805 + }, + { + "epoch": 1.1866900944432242, + "grad_norm": 2.390674114227295, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7351700067520142, + "num_tokens": 277458676.0, + "step": 10806 + }, + { + "epoch": 1.186799912145838, + "grad_norm": 2.0377368927001953, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7346034049987793, + "num_tokens": 277489523.0, + "step": 10807 + }, + { + "epoch": 1.1869097298484517, + "grad_norm": 2.1316072940826416, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7129139304161072, + "num_tokens": 277518165.0, + "step": 10808 + }, + { + "epoch": 1.1870195475510652, + "grad_norm": 2.4705488681793213, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7445294857025146, + "num_tokens": 277540858.0, + "step": 10809 + }, + { + "epoch": 1.187129365253679, + "grad_norm": 2.227128028869629, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7194149494171143, + "num_tokens": 277567285.0, + "step": 10810 + }, + { + "epoch": 1.1872391829562925, + "grad_norm": 2.126164197921753, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.718028724193573, + "num_tokens": 277595961.0, + "step": 10811 + }, + { + "epoch": 1.1873490006589063, + "grad_norm": 2.2120001316070557, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7057573795318604, + "num_tokens": 277623639.0, + "step": 10812 + }, + { + "epoch": 1.1874588183615198, + "grad_norm": 2.3730311393737793, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7335641384124756, + "num_tokens": 277647393.0, + "step": 10813 + }, + { + "epoch": 1.1875686360641335, + "grad_norm": 2.281670331954956, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7349352240562439, + "num_tokens": 277671307.0, + "step": 10814 + }, + { + "epoch": 1.1876784537667473, + "grad_norm": 2.3925015926361084, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7132595777511597, + "num_tokens": 277694177.0, + "step": 10815 + }, + { + "epoch": 1.1877882714693608, + "grad_norm": 2.438737154006958, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7427656650543213, + "num_tokens": 277716145.0, + "step": 10816 + }, + { + "epoch": 1.1878980891719746, + "grad_norm": 2.5295212268829346, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.711461067199707, + "num_tokens": 277737168.0, + "step": 10817 + }, + { + "epoch": 1.1880079068745881, + "grad_norm": 2.213724136352539, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7116986513137817, + "num_tokens": 277763636.0, + "step": 10818 + }, + { + "epoch": 1.1881177245772019, + "grad_norm": 2.4058353900909424, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7245385646820068, + "num_tokens": 277787554.0, + "step": 10819 + }, + { + "epoch": 1.1882275422798154, + "grad_norm": 2.312790870666504, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7441972494125366, + "num_tokens": 277810789.0, + "step": 10820 + }, + { + "epoch": 1.1883373599824292, + "grad_norm": 1.8975251913070679, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7103651165962219, + "num_tokens": 277845595.0, + "step": 10821 + }, + { + "epoch": 1.1884471776850427, + "grad_norm": 2.144773006439209, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7254617214202881, + "num_tokens": 277871324.0, + "step": 10822 + }, + { + "epoch": 1.1885569953876565, + "grad_norm": 2.4407126903533936, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7380551099777222, + "num_tokens": 277892881.0, + "step": 10823 + }, + { + "epoch": 1.1886668130902702, + "grad_norm": 2.051771640777588, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7091441750526428, + "num_tokens": 277923707.0, + "step": 10824 + }, + { + "epoch": 1.1887766307928838, + "grad_norm": 2.2571287155151367, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7431946992874146, + "num_tokens": 277947443.0, + "step": 10825 + }, + { + "epoch": 1.1888864484954975, + "grad_norm": 2.446007490158081, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7457553148269653, + "num_tokens": 277967188.0, + "step": 10826 + }, + { + "epoch": 1.188996266198111, + "grad_norm": 2.399709701538086, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7032725811004639, + "num_tokens": 277989812.0, + "step": 10827 + }, + { + "epoch": 1.1891060839007248, + "grad_norm": 2.6899988651275635, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7437951564788818, + "num_tokens": 278009077.0, + "step": 10828 + }, + { + "epoch": 1.1892159016033386, + "grad_norm": 2.3885650634765625, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7429107427597046, + "num_tokens": 278031368.0, + "step": 10829 + }, + { + "epoch": 1.189325719305952, + "grad_norm": 1.9892836809158325, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7045455574989319, + "num_tokens": 278063759.0, + "step": 10830 + }, + { + "epoch": 1.1894355370085659, + "grad_norm": 1.9677119255065918, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7121196985244751, + "num_tokens": 278096814.0, + "step": 10831 + }, + { + "epoch": 1.1895453547111794, + "grad_norm": 2.377232313156128, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6918150782585144, + "num_tokens": 278121058.0, + "step": 10832 + }, + { + "epoch": 1.1896551724137931, + "grad_norm": 2.012957811355591, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7157455682754517, + "num_tokens": 278154642.0, + "step": 10833 + }, + { + "epoch": 1.1897649901164067, + "grad_norm": 2.322540521621704, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7160236835479736, + "num_tokens": 278178135.0, + "step": 10834 + }, + { + "epoch": 1.1898748078190204, + "grad_norm": 2.346970558166504, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.716209888458252, + "num_tokens": 278201944.0, + "step": 10835 + }, + { + "epoch": 1.189984625521634, + "grad_norm": 2.053455114364624, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6895986795425415, + "num_tokens": 278234925.0, + "step": 10836 + }, + { + "epoch": 1.1900944432242477, + "grad_norm": 2.2971274852752686, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.723094642162323, + "num_tokens": 278261859.0, + "step": 10837 + }, + { + "epoch": 1.1902042609268615, + "grad_norm": 2.56050181388855, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7253156304359436, + "num_tokens": 278282839.0, + "step": 10838 + }, + { + "epoch": 1.190314078629475, + "grad_norm": 2.1670689582824707, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7102178335189819, + "num_tokens": 278310013.0, + "step": 10839 + }, + { + "epoch": 1.1904238963320888, + "grad_norm": 2.1603832244873047, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7079880833625793, + "num_tokens": 278339319.0, + "step": 10840 + }, + { + "epoch": 1.1905337140347023, + "grad_norm": 2.146726608276367, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7155582904815674, + "num_tokens": 278366567.0, + "step": 10841 + }, + { + "epoch": 1.190643531737316, + "grad_norm": 2.2999165058135986, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7207334637641907, + "num_tokens": 278390363.0, + "step": 10842 + }, + { + "epoch": 1.1907533494399298, + "grad_norm": 2.300023078918457, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7090963125228882, + "num_tokens": 278414613.0, + "step": 10843 + }, + { + "epoch": 1.1908631671425434, + "grad_norm": 2.1501474380493164, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7013303637504578, + "num_tokens": 278443546.0, + "step": 10844 + }, + { + "epoch": 1.1909729848451571, + "grad_norm": 2.027470588684082, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7181475758552551, + "num_tokens": 278473284.0, + "step": 10845 + }, + { + "epoch": 1.1910828025477707, + "grad_norm": 2.0893971920013428, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7444009184837341, + "num_tokens": 278499801.0, + "step": 10846 + }, + { + "epoch": 1.1911926202503844, + "grad_norm": 2.0774922370910645, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7251416444778442, + "num_tokens": 278529394.0, + "step": 10847 + }, + { + "epoch": 1.191302437952998, + "grad_norm": 2.1939427852630615, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7356597185134888, + "num_tokens": 278556006.0, + "step": 10848 + }, + { + "epoch": 1.1914122556556117, + "grad_norm": 2.339287519454956, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7146341800689697, + "num_tokens": 278579768.0, + "step": 10849 + }, + { + "epoch": 1.1915220733582252, + "grad_norm": 2.2041175365448, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7406467795372009, + "num_tokens": 278604617.0, + "step": 10850 + }, + { + "epoch": 1.191631891060839, + "grad_norm": 2.1777873039245605, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7084740400314331, + "num_tokens": 278633837.0, + "step": 10851 + }, + { + "epoch": 1.1917417087634528, + "grad_norm": 2.3012659549713135, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7087490558624268, + "num_tokens": 278659009.0, + "step": 10852 + }, + { + "epoch": 1.1918515264660663, + "grad_norm": 2.3712525367736816, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7307310700416565, + "num_tokens": 278682659.0, + "step": 10853 + }, + { + "epoch": 1.19196134416868, + "grad_norm": 2.1714630126953125, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7170403599739075, + "num_tokens": 278710461.0, + "step": 10854 + }, + { + "epoch": 1.1920711618712936, + "grad_norm": 1.9577035903930664, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7275339365005493, + "num_tokens": 278738978.0, + "step": 10855 + }, + { + "epoch": 1.1921809795739073, + "grad_norm": 2.2419071197509766, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7266030311584473, + "num_tokens": 278765870.0, + "step": 10856 + }, + { + "epoch": 1.192290797276521, + "grad_norm": 2.048612117767334, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6829288601875305, + "num_tokens": 278796670.0, + "step": 10857 + }, + { + "epoch": 1.1924006149791346, + "grad_norm": 2.378793716430664, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7143080830574036, + "num_tokens": 278819775.0, + "step": 10858 + }, + { + "epoch": 1.1925104326817484, + "grad_norm": 2.438617467880249, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7515690326690674, + "num_tokens": 278839861.0, + "step": 10859 + }, + { + "epoch": 1.192620250384362, + "grad_norm": 2.2474656105041504, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.68117356300354, + "num_tokens": 278868333.0, + "step": 10860 + }, + { + "epoch": 1.1927300680869757, + "grad_norm": 2.284649610519409, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7179630994796753, + "num_tokens": 278892784.0, + "step": 10861 + }, + { + "epoch": 1.1928398857895892, + "grad_norm": 2.114642381668091, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6996713876724243, + "num_tokens": 278919975.0, + "step": 10862 + }, + { + "epoch": 1.192949703492203, + "grad_norm": 2.3560400009155273, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7047975063323975, + "num_tokens": 278945234.0, + "step": 10863 + }, + { + "epoch": 1.1930595211948165, + "grad_norm": 2.331573963165283, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7396178245544434, + "num_tokens": 278967608.0, + "step": 10864 + }, + { + "epoch": 1.1931693388974303, + "grad_norm": 2.1159026622772217, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7285771369934082, + "num_tokens": 278994092.0, + "step": 10865 + }, + { + "epoch": 1.193279156600044, + "grad_norm": 2.0669054985046387, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7101823091506958, + "num_tokens": 279024756.0, + "step": 10866 + }, + { + "epoch": 1.1933889743026576, + "grad_norm": 2.060727119445801, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6938353180885315, + "num_tokens": 279055288.0, + "step": 10867 + }, + { + "epoch": 1.1934987920052713, + "grad_norm": 2.2047431468963623, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7118122577667236, + "num_tokens": 279084291.0, + "step": 10868 + }, + { + "epoch": 1.1936086097078848, + "grad_norm": 2.4346930980682373, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7238902449607849, + "num_tokens": 279105042.0, + "step": 10869 + }, + { + "epoch": 1.1937184274104986, + "grad_norm": 2.203859567642212, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7223942279815674, + "num_tokens": 279132231.0, + "step": 10870 + }, + { + "epoch": 1.1938282451131121, + "grad_norm": 2.2311925888061523, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6883264780044556, + "num_tokens": 279158132.0, + "step": 10871 + }, + { + "epoch": 1.193938062815726, + "grad_norm": 2.305852174758911, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.71889728307724, + "num_tokens": 279180547.0, + "step": 10872 + }, + { + "epoch": 1.1940478805183394, + "grad_norm": 2.1619324684143066, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7185909748077393, + "num_tokens": 279207628.0, + "step": 10873 + }, + { + "epoch": 1.1941576982209532, + "grad_norm": 2.0384771823883057, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7087534666061401, + "num_tokens": 279238638.0, + "step": 10874 + }, + { + "epoch": 1.194267515923567, + "grad_norm": 2.304874897003174, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7069928050041199, + "num_tokens": 279264750.0, + "step": 10875 + }, + { + "epoch": 1.1943773336261805, + "grad_norm": 2.593198776245117, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7069869637489319, + "num_tokens": 279285177.0, + "step": 10876 + }, + { + "epoch": 1.1944871513287942, + "grad_norm": 2.1888699531555176, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7233968377113342, + "num_tokens": 279310506.0, + "step": 10877 + }, + { + "epoch": 1.1945969690314078, + "grad_norm": 2.3617498874664307, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7178972959518433, + "num_tokens": 279334270.0, + "step": 10878 + }, + { + "epoch": 1.1947067867340215, + "grad_norm": 2.09773588180542, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7146145105361938, + "num_tokens": 279363430.0, + "step": 10879 + }, + { + "epoch": 1.1948166044366353, + "grad_norm": 2.3767106533050537, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7188311815261841, + "num_tokens": 279387552.0, + "step": 10880 + }, + { + "epoch": 1.1949264221392488, + "grad_norm": 2.207078218460083, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6951977610588074, + "num_tokens": 279416437.0, + "step": 10881 + }, + { + "epoch": 1.1950362398418626, + "grad_norm": 2.2423362731933594, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7331984639167786, + "num_tokens": 279439483.0, + "step": 10882 + }, + { + "epoch": 1.1951460575444761, + "grad_norm": 2.5408265590667725, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7224195003509521, + "num_tokens": 279460417.0, + "step": 10883 + }, + { + "epoch": 1.1952558752470899, + "grad_norm": 2.3188719749450684, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7176097631454468, + "num_tokens": 279483463.0, + "step": 10884 + }, + { + "epoch": 1.1953656929497034, + "grad_norm": 2.5104129314422607, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7130289077758789, + "num_tokens": 279505506.0, + "step": 10885 + }, + { + "epoch": 1.1954755106523172, + "grad_norm": 2.1140174865722656, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.726320743560791, + "num_tokens": 279532027.0, + "step": 10886 + }, + { + "epoch": 1.1955853283549307, + "grad_norm": 2.838979482650757, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7349973917007446, + "num_tokens": 279549224.0, + "step": 10887 + }, + { + "epoch": 1.1956951460575445, + "grad_norm": 2.2002949714660645, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7103827595710754, + "num_tokens": 279576750.0, + "step": 10888 + }, + { + "epoch": 1.1958049637601582, + "grad_norm": 2.140221357345581, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7206345200538635, + "num_tokens": 279602746.0, + "step": 10889 + }, + { + "epoch": 1.1959147814627717, + "grad_norm": 2.092731237411499, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7222139239311218, + "num_tokens": 279629826.0, + "step": 10890 + }, + { + "epoch": 1.1960245991653855, + "grad_norm": 2.1901471614837646, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7497917413711548, + "num_tokens": 279657528.0, + "step": 10891 + }, + { + "epoch": 1.196134416867999, + "grad_norm": 2.2662193775177, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7482422590255737, + "num_tokens": 279679965.0, + "step": 10892 + }, + { + "epoch": 1.1962442345706128, + "grad_norm": 2.3190879821777344, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7187052965164185, + "num_tokens": 279703769.0, + "step": 10893 + }, + { + "epoch": 1.1963540522732266, + "grad_norm": 2.1266331672668457, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7080216407775879, + "num_tokens": 279732087.0, + "step": 10894 + }, + { + "epoch": 1.19646386997584, + "grad_norm": 2.37973690032959, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7159873247146606, + "num_tokens": 279756306.0, + "step": 10895 + }, + { + "epoch": 1.1965736876784538, + "grad_norm": 2.172023296356201, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7167978286743164, + "num_tokens": 279780335.0, + "step": 10896 + }, + { + "epoch": 1.1966835053810674, + "grad_norm": 2.2050983905792236, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7191590070724487, + "num_tokens": 279805530.0, + "step": 10897 + }, + { + "epoch": 1.1967933230836811, + "grad_norm": 2.1956114768981934, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7275841236114502, + "num_tokens": 279829121.0, + "step": 10898 + }, + { + "epoch": 1.1969031407862947, + "grad_norm": 1.9941115379333496, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7031671404838562, + "num_tokens": 279861427.0, + "step": 10899 + }, + { + "epoch": 1.1970129584889084, + "grad_norm": 2.1731197834014893, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7271504998207092, + "num_tokens": 279888380.0, + "step": 10900 + }, + { + "epoch": 1.197122776191522, + "grad_norm": 2.1376845836639404, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7027214765548706, + "num_tokens": 279913888.0, + "step": 10901 + }, + { + "epoch": 1.1972325938941357, + "grad_norm": 2.3130645751953125, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7552642226219177, + "num_tokens": 279935552.0, + "step": 10902 + }, + { + "epoch": 1.1973424115967495, + "grad_norm": 2.3249709606170654, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7386902570724487, + "num_tokens": 279959003.0, + "step": 10903 + }, + { + "epoch": 1.197452229299363, + "grad_norm": 2.3859047889709473, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7073231935501099, + "num_tokens": 279982313.0, + "step": 10904 + }, + { + "epoch": 1.1975620470019768, + "grad_norm": 2.425511598587036, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7245627045631409, + "num_tokens": 280004224.0, + "step": 10905 + }, + { + "epoch": 1.1976718647045903, + "grad_norm": 2.6953792572021484, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.718948483467102, + "num_tokens": 280024111.0, + "step": 10906 + }, + { + "epoch": 1.197781682407204, + "grad_norm": 2.0546016693115234, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7323968410491943, + "num_tokens": 280052084.0, + "step": 10907 + }, + { + "epoch": 1.1978915001098178, + "grad_norm": 2.4767744541168213, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7253847122192383, + "num_tokens": 280073958.0, + "step": 10908 + }, + { + "epoch": 1.1980013178124314, + "grad_norm": 2.257550001144409, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7221051454544067, + "num_tokens": 280100559.0, + "step": 10909 + }, + { + "epoch": 1.1981111355150451, + "grad_norm": 2.306910753250122, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7009236216545105, + "num_tokens": 280124814.0, + "step": 10910 + }, + { + "epoch": 1.1982209532176586, + "grad_norm": 2.267695903778076, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7336755990982056, + "num_tokens": 280150756.0, + "step": 10911 + }, + { + "epoch": 1.1983307709202724, + "grad_norm": 1.9640071392059326, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7172331809997559, + "num_tokens": 280183281.0, + "step": 10912 + }, + { + "epoch": 1.198440588622886, + "grad_norm": 2.4337823390960693, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7283120155334473, + "num_tokens": 280206321.0, + "step": 10913 + }, + { + "epoch": 1.1985504063254997, + "grad_norm": 1.966317892074585, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6995911598205566, + "num_tokens": 280237805.0, + "step": 10914 + }, + { + "epoch": 1.1986602240281132, + "grad_norm": 1.9425569772720337, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7050116658210754, + "num_tokens": 280272197.0, + "step": 10915 + }, + { + "epoch": 1.198770041730727, + "grad_norm": 2.2836570739746094, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7291674613952637, + "num_tokens": 280297215.0, + "step": 10916 + }, + { + "epoch": 1.1988798594333407, + "grad_norm": 2.1664657592773438, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.726691722869873, + "num_tokens": 280323119.0, + "step": 10917 + }, + { + "epoch": 1.1989896771359543, + "grad_norm": 2.1504909992218018, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7142154574394226, + "num_tokens": 280351414.0, + "step": 10918 + }, + { + "epoch": 1.199099494838568, + "grad_norm": 2.7317960262298584, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7194278240203857, + "num_tokens": 280370427.0, + "step": 10919 + }, + { + "epoch": 1.1992093125411816, + "grad_norm": 2.1593611240386963, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7030175924301147, + "num_tokens": 280398090.0, + "step": 10920 + }, + { + "epoch": 1.1993191302437953, + "grad_norm": 2.0344045162200928, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7188828587532043, + "num_tokens": 280426865.0, + "step": 10921 + }, + { + "epoch": 1.199428947946409, + "grad_norm": 2.265576124191284, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7221400737762451, + "num_tokens": 280451673.0, + "step": 10922 + }, + { + "epoch": 1.1995387656490226, + "grad_norm": 2.1335630416870117, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7182011604309082, + "num_tokens": 280479373.0, + "step": 10923 + }, + { + "epoch": 1.1996485833516364, + "grad_norm": 2.22226619720459, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7370397448539734, + "num_tokens": 280502729.0, + "step": 10924 + }, + { + "epoch": 1.19975840105425, + "grad_norm": 2.427577018737793, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7323999404907227, + "num_tokens": 280525982.0, + "step": 10925 + }, + { + "epoch": 1.1998682187568637, + "grad_norm": 2.0587894916534424, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7363302111625671, + "num_tokens": 280554752.0, + "step": 10926 + }, + { + "epoch": 1.1999780364594772, + "grad_norm": 2.095928192138672, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7497719526290894, + "num_tokens": 280582532.0, + "step": 10927 + }, + { + "epoch": 1.200087854162091, + "grad_norm": 2.3085126876831055, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7058348655700684, + "num_tokens": 280608727.0, + "step": 10928 + }, + { + "epoch": 1.2001976718647045, + "grad_norm": 2.0639448165893555, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7094013690948486, + "num_tokens": 280638856.0, + "step": 10929 + }, + { + "epoch": 1.2003074895673183, + "grad_norm": 2.203993082046509, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7004283666610718, + "num_tokens": 280665041.0, + "step": 10930 + }, + { + "epoch": 1.200417307269932, + "grad_norm": 2.637746810913086, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7318166494369507, + "num_tokens": 280684382.0, + "step": 10931 + }, + { + "epoch": 1.2005271249725455, + "grad_norm": 2.369112253189087, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7269668579101562, + "num_tokens": 280707272.0, + "step": 10932 + }, + { + "epoch": 1.2006369426751593, + "grad_norm": 2.176241874694824, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7167366743087769, + "num_tokens": 280734444.0, + "step": 10933 + }, + { + "epoch": 1.2007467603777728, + "grad_norm": 2.20145845413208, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7128238677978516, + "num_tokens": 280761711.0, + "step": 10934 + }, + { + "epoch": 1.2008565780803866, + "grad_norm": 2.6405537128448486, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.696745753288269, + "num_tokens": 280787719.0, + "step": 10935 + }, + { + "epoch": 1.2009663957830001, + "grad_norm": 2.0963222980499268, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7013143301010132, + "num_tokens": 280817611.0, + "step": 10936 + }, + { + "epoch": 1.2010762134856139, + "grad_norm": 2.517512559890747, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7304118871688843, + "num_tokens": 280838359.0, + "step": 10937 + }, + { + "epoch": 1.2011860311882274, + "grad_norm": 2.341488838195801, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7169479727745056, + "num_tokens": 280862249.0, + "step": 10938 + }, + { + "epoch": 1.2012958488908412, + "grad_norm": 2.103055238723755, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7332026958465576, + "num_tokens": 280888861.0, + "step": 10939 + }, + { + "epoch": 1.201405666593455, + "grad_norm": 2.0584840774536133, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6980893611907959, + "num_tokens": 280920732.0, + "step": 10940 + }, + { + "epoch": 1.2015154842960685, + "grad_norm": 2.465151309967041, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7067842483520508, + "num_tokens": 280944557.0, + "step": 10941 + }, + { + "epoch": 1.2016253019986822, + "grad_norm": 2.032479763031006, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7098671793937683, + "num_tokens": 280976896.0, + "step": 10942 + }, + { + "epoch": 1.2017351197012958, + "grad_norm": 2.2254416942596436, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7228173613548279, + "num_tokens": 281001894.0, + "step": 10943 + }, + { + "epoch": 1.2018449374039095, + "grad_norm": 2.1735315322875977, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7298182845115662, + "num_tokens": 281028456.0, + "step": 10944 + }, + { + "epoch": 1.2019547551065233, + "grad_norm": 2.6912853717803955, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7176393270492554, + "num_tokens": 281048628.0, + "step": 10945 + }, + { + "epoch": 1.2020645728091368, + "grad_norm": 2.3205604553222656, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7254800796508789, + "num_tokens": 281071471.0, + "step": 10946 + }, + { + "epoch": 1.2021743905117506, + "grad_norm": 2.314333915710449, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7406100630760193, + "num_tokens": 281095348.0, + "step": 10947 + }, + { + "epoch": 1.202284208214364, + "grad_norm": 2.5510568618774414, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7239084243774414, + "num_tokens": 281117734.0, + "step": 10948 + }, + { + "epoch": 1.2023940259169779, + "grad_norm": 2.2650768756866455, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7103750705718994, + "num_tokens": 281143332.0, + "step": 10949 + }, + { + "epoch": 1.2025038436195914, + "grad_norm": 2.2370076179504395, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7310841679573059, + "num_tokens": 281169583.0, + "step": 10950 + }, + { + "epoch": 1.2026136613222052, + "grad_norm": 2.117008686065674, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7337535619735718, + "num_tokens": 281197388.0, + "step": 10951 + }, + { + "epoch": 1.2027234790248187, + "grad_norm": 2.0161213874816895, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6829112768173218, + "num_tokens": 281229232.0, + "step": 10952 + }, + { + "epoch": 1.2028332967274324, + "grad_norm": 2.314929723739624, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7204305529594421, + "num_tokens": 281254267.0, + "step": 10953 + }, + { + "epoch": 1.2029431144300462, + "grad_norm": 2.2157299518585205, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7380151152610779, + "num_tokens": 281279315.0, + "step": 10954 + }, + { + "epoch": 1.2030529321326597, + "grad_norm": 2.1873939037323, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7253091335296631, + "num_tokens": 281308288.0, + "step": 10955 + }, + { + "epoch": 1.2031627498352735, + "grad_norm": 2.1909985542297363, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7178251147270203, + "num_tokens": 281335445.0, + "step": 10956 + }, + { + "epoch": 1.203272567537887, + "grad_norm": 2.070957660675049, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7089110612869263, + "num_tokens": 281365347.0, + "step": 10957 + }, + { + "epoch": 1.2033823852405008, + "grad_norm": 2.037583589553833, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7123299241065979, + "num_tokens": 281397439.0, + "step": 10958 + }, + { + "epoch": 1.2034922029431145, + "grad_norm": 2.3264384269714355, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7335613369941711, + "num_tokens": 281421210.0, + "step": 10959 + }, + { + "epoch": 1.203602020645728, + "grad_norm": 2.087677240371704, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.705457329750061, + "num_tokens": 281449411.0, + "step": 10960 + }, + { + "epoch": 1.2037118383483418, + "grad_norm": 2.291757583618164, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7344779372215271, + "num_tokens": 281474270.0, + "step": 10961 + }, + { + "epoch": 1.2038216560509554, + "grad_norm": 2.163978099822998, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7111334800720215, + "num_tokens": 281502282.0, + "step": 10962 + }, + { + "epoch": 1.2039314737535691, + "grad_norm": 2.4427664279937744, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7168350219726562, + "num_tokens": 281523844.0, + "step": 10963 + }, + { + "epoch": 1.2040412914561827, + "grad_norm": 2.4010684490203857, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7182859182357788, + "num_tokens": 281545916.0, + "step": 10964 + }, + { + "epoch": 1.2041511091587964, + "grad_norm": 2.3912699222564697, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7149832844734192, + "num_tokens": 281568765.0, + "step": 10965 + }, + { + "epoch": 1.20426092686141, + "grad_norm": 2.6461243629455566, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7280964851379395, + "num_tokens": 281589567.0, + "step": 10966 + }, + { + "epoch": 1.2043707445640237, + "grad_norm": 2.436371088027954, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7240288853645325, + "num_tokens": 281611559.0, + "step": 10967 + }, + { + "epoch": 1.2044805622666375, + "grad_norm": 2.4134750366210938, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7284067869186401, + "num_tokens": 281634086.0, + "step": 10968 + }, + { + "epoch": 1.204590379969251, + "grad_norm": 2.193971872329712, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7190480828285217, + "num_tokens": 281661025.0, + "step": 10969 + }, + { + "epoch": 1.2047001976718648, + "grad_norm": 2.2845985889434814, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7040351033210754, + "num_tokens": 281686069.0, + "step": 10970 + }, + { + "epoch": 1.2048100153744783, + "grad_norm": 2.3747687339782715, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7333736419677734, + "num_tokens": 281709846.0, + "step": 10971 + }, + { + "epoch": 1.204919833077092, + "grad_norm": 2.3463196754455566, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7238360643386841, + "num_tokens": 281734186.0, + "step": 10972 + }, + { + "epoch": 1.2050296507797058, + "grad_norm": 2.139465570449829, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7335860133171082, + "num_tokens": 281761702.0, + "step": 10973 + }, + { + "epoch": 1.2051394684823193, + "grad_norm": 2.5587120056152344, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7516046762466431, + "num_tokens": 281783049.0, + "step": 10974 + }, + { + "epoch": 1.205249286184933, + "grad_norm": 2.5788629055023193, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7367334365844727, + "num_tokens": 281801698.0, + "step": 10975 + }, + { + "epoch": 1.2053591038875466, + "grad_norm": 2.1715376377105713, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7216945886611938, + "num_tokens": 281828373.0, + "step": 10976 + }, + { + "epoch": 1.2054689215901604, + "grad_norm": 2.2575318813323975, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7076219320297241, + "num_tokens": 281855668.0, + "step": 10977 + }, + { + "epoch": 1.205578739292774, + "grad_norm": 2.148468255996704, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.700218141078949, + "num_tokens": 281882642.0, + "step": 10978 + }, + { + "epoch": 1.2056885569953877, + "grad_norm": 2.065162181854248, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7065984010696411, + "num_tokens": 281913507.0, + "step": 10979 + }, + { + "epoch": 1.2057983746980012, + "grad_norm": 2.2133209705352783, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.723055899143219, + "num_tokens": 281940410.0, + "step": 10980 + }, + { + "epoch": 1.205908192400615, + "grad_norm": 1.9726060628890991, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7301172018051147, + "num_tokens": 281969567.0, + "step": 10981 + }, + { + "epoch": 1.2060180101032287, + "grad_norm": 2.2012226581573486, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7127066850662231, + "num_tokens": 281996254.0, + "step": 10982 + }, + { + "epoch": 1.2061278278058423, + "grad_norm": 2.228318452835083, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7170679569244385, + "num_tokens": 282022204.0, + "step": 10983 + }, + { + "epoch": 1.206237645508456, + "grad_norm": 2.078326940536499, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7117968201637268, + "num_tokens": 282049864.0, + "step": 10984 + }, + { + "epoch": 1.2063474632110696, + "grad_norm": 2.3759140968322754, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7230446934700012, + "num_tokens": 282073126.0, + "step": 10985 + }, + { + "epoch": 1.2064572809136833, + "grad_norm": 2.2741851806640625, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7132886648178101, + "num_tokens": 282097732.0, + "step": 10986 + }, + { + "epoch": 1.2065670986162969, + "grad_norm": 2.1410224437713623, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7281256318092346, + "num_tokens": 282125107.0, + "step": 10987 + }, + { + "epoch": 1.2066769163189106, + "grad_norm": 2.068413734436035, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7338432669639587, + "num_tokens": 282151955.0, + "step": 10988 + }, + { + "epoch": 1.2067867340215244, + "grad_norm": 2.309973955154419, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7350744009017944, + "num_tokens": 282173836.0, + "step": 10989 + }, + { + "epoch": 1.206896551724138, + "grad_norm": 2.503173828125, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7181857228279114, + "num_tokens": 282196094.0, + "step": 10990 + }, + { + "epoch": 1.2070063694267517, + "grad_norm": 2.463228464126587, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7173399925231934, + "num_tokens": 282218376.0, + "step": 10991 + }, + { + "epoch": 1.2071161871293652, + "grad_norm": 2.357729911804199, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7060662508010864, + "num_tokens": 282241998.0, + "step": 10992 + }, + { + "epoch": 1.207226004831979, + "grad_norm": 2.398676872253418, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7301105260848999, + "num_tokens": 282264165.0, + "step": 10993 + }, + { + "epoch": 1.2073358225345925, + "grad_norm": 2.2227447032928467, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7222886681556702, + "num_tokens": 282289959.0, + "step": 10994 + }, + { + "epoch": 1.2074456402372062, + "grad_norm": 2.469557762145996, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7299420833587646, + "num_tokens": 282313805.0, + "step": 10995 + }, + { + "epoch": 1.20755545793982, + "grad_norm": 2.5588207244873047, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7386220097541809, + "num_tokens": 282335195.0, + "step": 10996 + }, + { + "epoch": 1.2076652756424335, + "grad_norm": 2.3086705207824707, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7091217637062073, + "num_tokens": 282360433.0, + "step": 10997 + }, + { + "epoch": 1.2077750933450473, + "grad_norm": 2.215904712677002, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7207262516021729, + "num_tokens": 282386914.0, + "step": 10998 + }, + { + "epoch": 1.2078849110476608, + "grad_norm": 2.6172032356262207, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7331098914146423, + "num_tokens": 282407097.0, + "step": 10999 + }, + { + "epoch": 1.2079947287502746, + "grad_norm": 2.1027042865753174, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7509728670120239, + "num_tokens": 282432487.0, + "step": 11000 + }, + { + "epoch": 1.2081045464528881, + "grad_norm": 2.2566230297088623, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7246691584587097, + "num_tokens": 282457514.0, + "step": 11001 + }, + { + "epoch": 1.2082143641555019, + "grad_norm": 2.3037469387054443, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7251635789871216, + "num_tokens": 282480956.0, + "step": 11002 + }, + { + "epoch": 1.2083241818581154, + "grad_norm": 2.1759395599365234, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7198927402496338, + "num_tokens": 282508900.0, + "step": 11003 + }, + { + "epoch": 1.2084339995607292, + "grad_norm": 2.308098554611206, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6931684017181396, + "num_tokens": 282534234.0, + "step": 11004 + }, + { + "epoch": 1.208543817263343, + "grad_norm": 2.1574087142944336, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7083926796913147, + "num_tokens": 282564141.0, + "step": 11005 + }, + { + "epoch": 1.2086536349659565, + "grad_norm": 2.2170441150665283, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7300929427146912, + "num_tokens": 282589208.0, + "step": 11006 + }, + { + "epoch": 1.2087634526685702, + "grad_norm": 1.9729121923446655, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7082018256187439, + "num_tokens": 282619745.0, + "step": 11007 + }, + { + "epoch": 1.2088732703711838, + "grad_norm": 2.1422345638275146, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7396067380905151, + "num_tokens": 282645708.0, + "step": 11008 + }, + { + "epoch": 1.2089830880737975, + "grad_norm": 2.178290843963623, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7153609991073608, + "num_tokens": 282673705.0, + "step": 11009 + }, + { + "epoch": 1.2090929057764113, + "grad_norm": 2.4060556888580322, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7306002378463745, + "num_tokens": 282698168.0, + "step": 11010 + }, + { + "epoch": 1.2092027234790248, + "grad_norm": 2.110417127609253, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7367457747459412, + "num_tokens": 282724440.0, + "step": 11011 + }, + { + "epoch": 1.2093125411816386, + "grad_norm": 2.207615613937378, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7221970558166504, + "num_tokens": 282749027.0, + "step": 11012 + }, + { + "epoch": 1.209422358884252, + "grad_norm": 2.1089367866516113, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7114415168762207, + "num_tokens": 282777931.0, + "step": 11013 + }, + { + "epoch": 1.2095321765868658, + "grad_norm": 2.0877630710601807, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7111853361129761, + "num_tokens": 282806334.0, + "step": 11014 + }, + { + "epoch": 1.2096419942894794, + "grad_norm": 2.217456102371216, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.704210102558136, + "num_tokens": 282832064.0, + "step": 11015 + }, + { + "epoch": 1.2097518119920931, + "grad_norm": 2.183206558227539, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7296434640884399, + "num_tokens": 282857485.0, + "step": 11016 + }, + { + "epoch": 1.2098616296947067, + "grad_norm": 2.312286615371704, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7525074481964111, + "num_tokens": 282880209.0, + "step": 11017 + }, + { + "epoch": 1.2099714473973204, + "grad_norm": 2.3791003227233887, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7117191553115845, + "num_tokens": 282902146.0, + "step": 11018 + }, + { + "epoch": 1.2100812650999342, + "grad_norm": 1.9508326053619385, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7128543853759766, + "num_tokens": 282933970.0, + "step": 11019 + }, + { + "epoch": 1.2101910828025477, + "grad_norm": 2.3250138759613037, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7219247221946716, + "num_tokens": 282959515.0, + "step": 11020 + }, + { + "epoch": 1.2103009005051615, + "grad_norm": 2.246788263320923, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6903243064880371, + "num_tokens": 282987306.0, + "step": 11021 + }, + { + "epoch": 1.210410718207775, + "grad_norm": 2.165875196456909, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7444478273391724, + "num_tokens": 283013190.0, + "step": 11022 + }, + { + "epoch": 1.2105205359103888, + "grad_norm": 2.086054801940918, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6972377300262451, + "num_tokens": 283041577.0, + "step": 11023 + }, + { + "epoch": 1.2106303536130025, + "grad_norm": 2.440572738647461, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7217485904693604, + "num_tokens": 283063059.0, + "step": 11024 + }, + { + "epoch": 1.210740171315616, + "grad_norm": 2.261072874069214, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7274534106254578, + "num_tokens": 283087337.0, + "step": 11025 + }, + { + "epoch": 1.2108499890182298, + "grad_norm": 2.287116765975952, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7187985181808472, + "num_tokens": 283112114.0, + "step": 11026 + }, + { + "epoch": 1.2109598067208434, + "grad_norm": 2.1036031246185303, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7087681889533997, + "num_tokens": 283140915.0, + "step": 11027 + }, + { + "epoch": 1.2110696244234571, + "grad_norm": 2.251277446746826, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7311110496520996, + "num_tokens": 283167653.0, + "step": 11028 + }, + { + "epoch": 1.2111794421260706, + "grad_norm": 2.5013070106506348, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7326180934906006, + "num_tokens": 283187502.0, + "step": 11029 + }, + { + "epoch": 1.2112892598286844, + "grad_norm": 2.352860927581787, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7140281200408936, + "num_tokens": 283212290.0, + "step": 11030 + }, + { + "epoch": 1.211399077531298, + "grad_norm": 1.9167816638946533, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6895468831062317, + "num_tokens": 283247108.0, + "step": 11031 + }, + { + "epoch": 1.2115088952339117, + "grad_norm": 2.3602211475372314, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7139383554458618, + "num_tokens": 283271374.0, + "step": 11032 + }, + { + "epoch": 1.2116187129365255, + "grad_norm": 2.2560064792633057, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.738690972328186, + "num_tokens": 283295966.0, + "step": 11033 + }, + { + "epoch": 1.211728530639139, + "grad_norm": 2.027970552444458, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7018759250640869, + "num_tokens": 283329320.0, + "step": 11034 + }, + { + "epoch": 1.2118383483417527, + "grad_norm": 2.1469991207122803, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.704943060874939, + "num_tokens": 283357217.0, + "step": 11035 + }, + { + "epoch": 1.2119481660443663, + "grad_norm": 2.337453603744507, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7175170183181763, + "num_tokens": 283382570.0, + "step": 11036 + }, + { + "epoch": 1.21205798374698, + "grad_norm": 2.155829429626465, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7160482406616211, + "num_tokens": 283409986.0, + "step": 11037 + }, + { + "epoch": 1.2121678014495938, + "grad_norm": 2.4455063343048096, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7187072038650513, + "num_tokens": 283432959.0, + "step": 11038 + }, + { + "epoch": 1.2122776191522073, + "grad_norm": 2.6658103466033936, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7286372780799866, + "num_tokens": 283453025.0, + "step": 11039 + }, + { + "epoch": 1.212387436854821, + "grad_norm": 2.2255160808563232, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7292943000793457, + "num_tokens": 283478822.0, + "step": 11040 + }, + { + "epoch": 1.2124972545574346, + "grad_norm": 2.3432624340057373, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7233071327209473, + "num_tokens": 283501996.0, + "step": 11041 + }, + { + "epoch": 1.2126070722600484, + "grad_norm": 2.213909387588501, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.707224428653717, + "num_tokens": 283528703.0, + "step": 11042 + }, + { + "epoch": 1.212716889962662, + "grad_norm": 1.9553614854812622, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7302883863449097, + "num_tokens": 283560222.0, + "step": 11043 + }, + { + "epoch": 1.2128267076652757, + "grad_norm": 2.3629729747772217, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7096531391143799, + "num_tokens": 283585043.0, + "step": 11044 + }, + { + "epoch": 1.2129365253678892, + "grad_norm": 2.334859848022461, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7342092990875244, + "num_tokens": 283606801.0, + "step": 11045 + }, + { + "epoch": 1.213046343070503, + "grad_norm": 2.34342098236084, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7291303873062134, + "num_tokens": 283628310.0, + "step": 11046 + }, + { + "epoch": 1.2131561607731167, + "grad_norm": 2.0094263553619385, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7085295915603638, + "num_tokens": 283657755.0, + "step": 11047 + }, + { + "epoch": 1.2132659784757303, + "grad_norm": 2.188995838165283, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.701376736164093, + "num_tokens": 283684456.0, + "step": 11048 + }, + { + "epoch": 1.213375796178344, + "grad_norm": 2.2870054244995117, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7291169762611389, + "num_tokens": 283708900.0, + "step": 11049 + }, + { + "epoch": 1.2134856138809575, + "grad_norm": 2.8444433212280273, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7161567807197571, + "num_tokens": 283725912.0, + "step": 11050 + }, + { + "epoch": 1.2135954315835713, + "grad_norm": 2.4308969974517822, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6960266828536987, + "num_tokens": 283749683.0, + "step": 11051 + }, + { + "epoch": 1.2137052492861848, + "grad_norm": 2.340925693511963, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.715274453163147, + "num_tokens": 283774922.0, + "step": 11052 + }, + { + "epoch": 1.2138150669887986, + "grad_norm": 2.1959450244903564, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6883562207221985, + "num_tokens": 283803350.0, + "step": 11053 + }, + { + "epoch": 1.2139248846914121, + "grad_norm": 1.8616918325424194, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7182521820068359, + "num_tokens": 283837101.0, + "step": 11054 + }, + { + "epoch": 1.2140347023940259, + "grad_norm": 2.0815823078155518, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.712053120136261, + "num_tokens": 283866568.0, + "step": 11055 + }, + { + "epoch": 1.2141445200966396, + "grad_norm": 2.323899745941162, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7208759784698486, + "num_tokens": 283889708.0, + "step": 11056 + }, + { + "epoch": 1.2142543377992532, + "grad_norm": 2.3211631774902344, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7284725904464722, + "num_tokens": 283914226.0, + "step": 11057 + }, + { + "epoch": 1.214364155501867, + "grad_norm": 2.3585402965545654, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7044660449028015, + "num_tokens": 283937831.0, + "step": 11058 + }, + { + "epoch": 1.2144739732044805, + "grad_norm": 2.3721137046813965, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7083064317703247, + "num_tokens": 283961470.0, + "step": 11059 + }, + { + "epoch": 1.2145837909070942, + "grad_norm": 2.017975091934204, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6914238929748535, + "num_tokens": 283991760.0, + "step": 11060 + }, + { + "epoch": 1.214693608609708, + "grad_norm": 2.426095962524414, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7230863571166992, + "num_tokens": 284015604.0, + "step": 11061 + }, + { + "epoch": 1.2148034263123215, + "grad_norm": 2.3055903911590576, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7114924192428589, + "num_tokens": 284040693.0, + "step": 11062 + }, + { + "epoch": 1.2149132440149353, + "grad_norm": 2.306021213531494, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7272896766662598, + "num_tokens": 284064357.0, + "step": 11063 + }, + { + "epoch": 1.2150230617175488, + "grad_norm": 2.4983553886413574, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7327433228492737, + "num_tokens": 284086287.0, + "step": 11064 + }, + { + "epoch": 1.2151328794201626, + "grad_norm": 2.1338019371032715, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7217416763305664, + "num_tokens": 284114385.0, + "step": 11065 + }, + { + "epoch": 1.215242697122776, + "grad_norm": 2.3393657207489014, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7126054763793945, + "num_tokens": 284138812.0, + "step": 11066 + }, + { + "epoch": 1.2153525148253899, + "grad_norm": 2.4253411293029785, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7107221484184265, + "num_tokens": 284161839.0, + "step": 11067 + }, + { + "epoch": 1.2154623325280034, + "grad_norm": 2.458479166030884, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7367175221443176, + "num_tokens": 284184662.0, + "step": 11068 + }, + { + "epoch": 1.2155721502306172, + "grad_norm": 2.1741559505462646, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7072274684906006, + "num_tokens": 284213371.0, + "step": 11069 + }, + { + "epoch": 1.215681967933231, + "grad_norm": 2.7226104736328125, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7216925024986267, + "num_tokens": 284231954.0, + "step": 11070 + }, + { + "epoch": 1.2157917856358444, + "grad_norm": 2.1630589962005615, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7081520557403564, + "num_tokens": 284259952.0, + "step": 11071 + }, + { + "epoch": 1.2159016033384582, + "grad_norm": 2.0521080493927, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7143356800079346, + "num_tokens": 284288731.0, + "step": 11072 + }, + { + "epoch": 1.2160114210410717, + "grad_norm": 2.3227896690368652, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7260863780975342, + "num_tokens": 284312951.0, + "step": 11073 + }, + { + "epoch": 1.2161212387436855, + "grad_norm": 2.2229528427124023, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7172766923904419, + "num_tokens": 284339950.0, + "step": 11074 + }, + { + "epoch": 1.2162310564462993, + "grad_norm": 2.566263198852539, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7378813028335571, + "num_tokens": 284361649.0, + "step": 11075 + }, + { + "epoch": 1.2163408741489128, + "grad_norm": 2.354166030883789, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7207801342010498, + "num_tokens": 284385677.0, + "step": 11076 + }, + { + "epoch": 1.2164506918515265, + "grad_norm": 2.498845338821411, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7410441637039185, + "num_tokens": 284406654.0, + "step": 11077 + }, + { + "epoch": 1.21656050955414, + "grad_norm": 2.1425561904907227, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7326085567474365, + "num_tokens": 284433143.0, + "step": 11078 + }, + { + "epoch": 1.2166703272567538, + "grad_norm": 2.164111375808716, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7353894710540771, + "num_tokens": 284458787.0, + "step": 11079 + }, + { + "epoch": 1.2167801449593674, + "grad_norm": 2.6371915340423584, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7214934825897217, + "num_tokens": 284480009.0, + "step": 11080 + }, + { + "epoch": 1.2168899626619811, + "grad_norm": 2.355992555618286, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7248146533966064, + "num_tokens": 284505827.0, + "step": 11081 + }, + { + "epoch": 1.2169997803645947, + "grad_norm": 2.5613789558410645, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7430161833763123, + "num_tokens": 284526059.0, + "step": 11082 + }, + { + "epoch": 1.2171095980672084, + "grad_norm": 2.416865348815918, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7099651098251343, + "num_tokens": 284550472.0, + "step": 11083 + }, + { + "epoch": 1.2172194157698222, + "grad_norm": 2.2252838611602783, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7105979323387146, + "num_tokens": 284577078.0, + "step": 11084 + }, + { + "epoch": 1.2173292334724357, + "grad_norm": 2.3372366428375244, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7090303301811218, + "num_tokens": 284602289.0, + "step": 11085 + }, + { + "epoch": 1.2174390511750495, + "grad_norm": 2.3860628604888916, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7199593782424927, + "num_tokens": 284625553.0, + "step": 11086 + }, + { + "epoch": 1.217548868877663, + "grad_norm": 2.446281909942627, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.716246485710144, + "num_tokens": 284649043.0, + "step": 11087 + }, + { + "epoch": 1.2176586865802768, + "grad_norm": 2.3087902069091797, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7383503913879395, + "num_tokens": 284674332.0, + "step": 11088 + }, + { + "epoch": 1.2177685042828905, + "grad_norm": 2.5020968914031982, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7339801788330078, + "num_tokens": 284694654.0, + "step": 11089 + }, + { + "epoch": 1.217878321985504, + "grad_norm": 2.438026189804077, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7324066758155823, + "num_tokens": 284718862.0, + "step": 11090 + }, + { + "epoch": 1.2179881396881178, + "grad_norm": 2.5152204036712646, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7222816944122314, + "num_tokens": 284739463.0, + "step": 11091 + }, + { + "epoch": 1.2180979573907313, + "grad_norm": 2.358600378036499, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7336260080337524, + "num_tokens": 284761664.0, + "step": 11092 + }, + { + "epoch": 1.218207775093345, + "grad_norm": 2.3796262741088867, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7111155986785889, + "num_tokens": 284788091.0, + "step": 11093 + }, + { + "epoch": 1.2183175927959586, + "grad_norm": 1.8779563903808594, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7035454511642456, + "num_tokens": 284822481.0, + "step": 11094 + }, + { + "epoch": 1.2184274104985724, + "grad_norm": 2.550030469894409, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7278386950492859, + "num_tokens": 284842705.0, + "step": 11095 + }, + { + "epoch": 1.218537228201186, + "grad_norm": 2.4220306873321533, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7091145515441895, + "num_tokens": 284866074.0, + "step": 11096 + }, + { + "epoch": 1.2186470459037997, + "grad_norm": 2.4223382472991943, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7087392807006836, + "num_tokens": 284890108.0, + "step": 11097 + }, + { + "epoch": 1.2187568636064134, + "grad_norm": 2.3827977180480957, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7055432200431824, + "num_tokens": 284916566.0, + "step": 11098 + }, + { + "epoch": 1.218866681309027, + "grad_norm": 2.503605842590332, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7333728671073914, + "num_tokens": 284936888.0, + "step": 11099 + }, + { + "epoch": 1.2189764990116407, + "grad_norm": 2.5271151065826416, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7326529026031494, + "num_tokens": 284959070.0, + "step": 11100 + }, + { + "epoch": 1.2190863167142543, + "grad_norm": 2.431210517883301, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7118656635284424, + "num_tokens": 284982372.0, + "step": 11101 + }, + { + "epoch": 1.219196134416868, + "grad_norm": 2.3108668327331543, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7227965593338013, + "num_tokens": 285007758.0, + "step": 11102 + }, + { + "epoch": 1.2193059521194818, + "grad_norm": 2.31178879737854, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7011233568191528, + "num_tokens": 285034360.0, + "step": 11103 + }, + { + "epoch": 1.2194157698220953, + "grad_norm": 2.193535566329956, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7186185121536255, + "num_tokens": 285060110.0, + "step": 11104 + }, + { + "epoch": 1.219525587524709, + "grad_norm": 2.23093318939209, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6915680170059204, + "num_tokens": 285084338.0, + "step": 11105 + }, + { + "epoch": 1.2196354052273226, + "grad_norm": 2.4493987560272217, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7030338048934937, + "num_tokens": 285106975.0, + "step": 11106 + }, + { + "epoch": 1.2197452229299364, + "grad_norm": 2.0309269428253174, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7177813053131104, + "num_tokens": 285136058.0, + "step": 11107 + }, + { + "epoch": 1.21985504063255, + "grad_norm": 2.4033639430999756, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7102757096290588, + "num_tokens": 285160579.0, + "step": 11108 + }, + { + "epoch": 1.2199648583351637, + "grad_norm": 2.118324041366577, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7096544504165649, + "num_tokens": 285191042.0, + "step": 11109 + }, + { + "epoch": 1.2200746760377772, + "grad_norm": 2.6114656925201416, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.722218930721283, + "num_tokens": 285211046.0, + "step": 11110 + }, + { + "epoch": 1.220184493740391, + "grad_norm": 2.422978639602661, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7261650562286377, + "num_tokens": 285233274.0, + "step": 11111 + }, + { + "epoch": 1.2202943114430047, + "grad_norm": 2.5460994243621826, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6989448070526123, + "num_tokens": 285255983.0, + "step": 11112 + }, + { + "epoch": 1.2204041291456182, + "grad_norm": 2.2537906169891357, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.734792947769165, + "num_tokens": 285280935.0, + "step": 11113 + }, + { + "epoch": 1.220513946848232, + "grad_norm": 2.2024986743927, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7056853771209717, + "num_tokens": 285309432.0, + "step": 11114 + }, + { + "epoch": 1.2206237645508455, + "grad_norm": 2.1871564388275146, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7161540985107422, + "num_tokens": 285336450.0, + "step": 11115 + }, + { + "epoch": 1.2207335822534593, + "grad_norm": 2.5773754119873047, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7201762199401855, + "num_tokens": 285356113.0, + "step": 11116 + }, + { + "epoch": 1.2208433999560728, + "grad_norm": 2.219459056854248, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7283109426498413, + "num_tokens": 285382201.0, + "step": 11117 + }, + { + "epoch": 1.2209532176586866, + "grad_norm": 2.561823606491089, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7160989046096802, + "num_tokens": 285404384.0, + "step": 11118 + }, + { + "epoch": 1.2210630353613001, + "grad_norm": 2.368086576461792, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.735252320766449, + "num_tokens": 285427589.0, + "step": 11119 + }, + { + "epoch": 1.2211728530639139, + "grad_norm": 2.450613021850586, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7070425748825073, + "num_tokens": 285450397.0, + "step": 11120 + }, + { + "epoch": 1.2212826707665276, + "grad_norm": 2.050706148147583, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7219340205192566, + "num_tokens": 285481030.0, + "step": 11121 + }, + { + "epoch": 1.2213924884691412, + "grad_norm": 2.1899664402008057, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7017827033996582, + "num_tokens": 285508209.0, + "step": 11122 + }, + { + "epoch": 1.221502306171755, + "grad_norm": 2.3060622215270996, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7258633971214294, + "num_tokens": 285533714.0, + "step": 11123 + }, + { + "epoch": 1.2216121238743685, + "grad_norm": 2.271929979324341, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7103629112243652, + "num_tokens": 285559899.0, + "step": 11124 + }, + { + "epoch": 1.2217219415769822, + "grad_norm": 2.32678484916687, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6992576122283936, + "num_tokens": 285584790.0, + "step": 11125 + }, + { + "epoch": 1.221831759279596, + "grad_norm": 2.0996341705322266, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7486067414283752, + "num_tokens": 285612983.0, + "step": 11126 + }, + { + "epoch": 1.2219415769822095, + "grad_norm": 2.140314817428589, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6983550786972046, + "num_tokens": 285642751.0, + "step": 11127 + }, + { + "epoch": 1.2220513946848233, + "grad_norm": 2.3395638465881348, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7144296169281006, + "num_tokens": 285667391.0, + "step": 11128 + }, + { + "epoch": 1.2221612123874368, + "grad_norm": 2.104257345199585, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7033178806304932, + "num_tokens": 285697289.0, + "step": 11129 + }, + { + "epoch": 1.2222710300900506, + "grad_norm": 2.588169813156128, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7145147919654846, + "num_tokens": 285718814.0, + "step": 11130 + }, + { + "epoch": 1.222380847792664, + "grad_norm": 2.488084316253662, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7255738973617554, + "num_tokens": 285740665.0, + "step": 11131 + }, + { + "epoch": 1.2224906654952779, + "grad_norm": 2.267943859100342, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6903831362724304, + "num_tokens": 285767535.0, + "step": 11132 + }, + { + "epoch": 1.2226004831978914, + "grad_norm": 2.2264578342437744, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7354824542999268, + "num_tokens": 285792621.0, + "step": 11133 + }, + { + "epoch": 1.2227103009005051, + "grad_norm": 2.355170726776123, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7310583591461182, + "num_tokens": 285815415.0, + "step": 11134 + }, + { + "epoch": 1.222820118603119, + "grad_norm": 2.3672823905944824, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.729929506778717, + "num_tokens": 285838906.0, + "step": 11135 + }, + { + "epoch": 1.2229299363057324, + "grad_norm": 2.157257080078125, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7033568620681763, + "num_tokens": 285867772.0, + "step": 11136 + }, + { + "epoch": 1.2230397540083462, + "grad_norm": 1.884019374847412, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7204175591468811, + "num_tokens": 285901118.0, + "step": 11137 + }, + { + "epoch": 1.2231495717109597, + "grad_norm": 1.9137418270111084, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6893609762191772, + "num_tokens": 285935740.0, + "step": 11138 + }, + { + "epoch": 1.2232593894135735, + "grad_norm": 2.339970350265503, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7010388374328613, + "num_tokens": 285959026.0, + "step": 11139 + }, + { + "epoch": 1.2233692071161872, + "grad_norm": 2.19512939453125, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.6995592713356018, + "num_tokens": 285985352.0, + "step": 11140 + }, + { + "epoch": 1.2234790248188008, + "grad_norm": 2.3173792362213135, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.723560094833374, + "num_tokens": 286008211.0, + "step": 11141 + }, + { + "epoch": 1.2235888425214145, + "grad_norm": 2.0589425563812256, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7065534591674805, + "num_tokens": 286037305.0, + "step": 11142 + }, + { + "epoch": 1.223698660224028, + "grad_norm": 2.260606527328491, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7063323259353638, + "num_tokens": 286062097.0, + "step": 11143 + }, + { + "epoch": 1.2238084779266418, + "grad_norm": 2.1441996097564697, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.725601851940155, + "num_tokens": 286090247.0, + "step": 11144 + }, + { + "epoch": 1.2239182956292554, + "grad_norm": 2.3067636489868164, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7192390561103821, + "num_tokens": 286115928.0, + "step": 11145 + }, + { + "epoch": 1.2240281133318691, + "grad_norm": 2.5393266677856445, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.725056529045105, + "num_tokens": 286139055.0, + "step": 11146 + }, + { + "epoch": 1.2241379310344827, + "grad_norm": 2.035233736038208, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.737684965133667, + "num_tokens": 286166053.0, + "step": 11147 + }, + { + "epoch": 1.2242477487370964, + "grad_norm": 2.2518157958984375, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7443596124649048, + "num_tokens": 286189813.0, + "step": 11148 + }, + { + "epoch": 1.2243575664397102, + "grad_norm": 2.4027926921844482, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7206019759178162, + "num_tokens": 286214945.0, + "step": 11149 + }, + { + "epoch": 1.2244673841423237, + "grad_norm": 2.234484910964966, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7197849750518799, + "num_tokens": 286240726.0, + "step": 11150 + }, + { + "epoch": 1.2245772018449375, + "grad_norm": 2.2467410564422607, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7088972330093384, + "num_tokens": 286266982.0, + "step": 11151 + }, + { + "epoch": 1.224687019547551, + "grad_norm": 1.9830304384231567, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7196954488754272, + "num_tokens": 286298174.0, + "step": 11152 + }, + { + "epoch": 1.2247968372501647, + "grad_norm": 2.284121513366699, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7281601428985596, + "num_tokens": 286322101.0, + "step": 11153 + }, + { + "epoch": 1.2249066549527785, + "grad_norm": 2.1667494773864746, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7227672338485718, + "num_tokens": 286348483.0, + "step": 11154 + }, + { + "epoch": 1.225016472655392, + "grad_norm": 2.0884382724761963, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7170518636703491, + "num_tokens": 286377521.0, + "step": 11155 + }, + { + "epoch": 1.2251262903580058, + "grad_norm": 2.2742552757263184, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6975577473640442, + "num_tokens": 286402731.0, + "step": 11156 + }, + { + "epoch": 1.2252361080606193, + "grad_norm": 2.271319627761841, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6987729072570801, + "num_tokens": 286428133.0, + "step": 11157 + }, + { + "epoch": 1.225345925763233, + "grad_norm": 2.2261219024658203, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7203065156936646, + "num_tokens": 286454962.0, + "step": 11158 + }, + { + "epoch": 1.2254557434658466, + "grad_norm": 2.1905405521392822, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7108386754989624, + "num_tokens": 286480442.0, + "step": 11159 + }, + { + "epoch": 1.2255655611684604, + "grad_norm": 2.3532869815826416, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7347599267959595, + "num_tokens": 286503706.0, + "step": 11160 + }, + { + "epoch": 1.225675378871074, + "grad_norm": 2.283881425857544, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7381425499916077, + "num_tokens": 286526619.0, + "step": 11161 + }, + { + "epoch": 1.2257851965736877, + "grad_norm": 2.2040984630584717, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7168459296226501, + "num_tokens": 286551853.0, + "step": 11162 + }, + { + "epoch": 1.2258950142763014, + "grad_norm": 2.2383320331573486, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7265682816505432, + "num_tokens": 286576760.0, + "step": 11163 + }, + { + "epoch": 1.226004831978915, + "grad_norm": 2.490954875946045, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7367982268333435, + "num_tokens": 286597395.0, + "step": 11164 + }, + { + "epoch": 1.2261146496815287, + "grad_norm": 2.520984649658203, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7132058143615723, + "num_tokens": 286618197.0, + "step": 11165 + }, + { + "epoch": 1.2262244673841423, + "grad_norm": 2.1269211769104004, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7156565189361572, + "num_tokens": 286647897.0, + "step": 11166 + }, + { + "epoch": 1.226334285086756, + "grad_norm": 2.5712761878967285, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7287888526916504, + "num_tokens": 286670297.0, + "step": 11167 + }, + { + "epoch": 1.2264441027893696, + "grad_norm": 2.026599884033203, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7213190793991089, + "num_tokens": 286699165.0, + "step": 11168 + }, + { + "epoch": 1.2265539204919833, + "grad_norm": 2.032944679260254, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7047964334487915, + "num_tokens": 286728254.0, + "step": 11169 + }, + { + "epoch": 1.226663738194597, + "grad_norm": 2.2097508907318115, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.709631085395813, + "num_tokens": 286754411.0, + "step": 11170 + }, + { + "epoch": 1.2267735558972106, + "grad_norm": 2.3234150409698486, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7248867750167847, + "num_tokens": 286778811.0, + "step": 11171 + }, + { + "epoch": 1.2268833735998244, + "grad_norm": 2.2009403705596924, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7193050384521484, + "num_tokens": 286806592.0, + "step": 11172 + }, + { + "epoch": 1.226993191302438, + "grad_norm": 2.114499807357788, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6953352689743042, + "num_tokens": 286836685.0, + "step": 11173 + }, + { + "epoch": 1.2271030090050516, + "grad_norm": 2.289984941482544, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7165483832359314, + "num_tokens": 286862987.0, + "step": 11174 + }, + { + "epoch": 1.2272128267076652, + "grad_norm": 2.384546995162964, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7150272727012634, + "num_tokens": 286888268.0, + "step": 11175 + }, + { + "epoch": 1.227322644410279, + "grad_norm": 1.9802522659301758, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7269842624664307, + "num_tokens": 286918506.0, + "step": 11176 + }, + { + "epoch": 1.2274324621128927, + "grad_norm": 2.1967408657073975, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7051310539245605, + "num_tokens": 286945784.0, + "step": 11177 + }, + { + "epoch": 1.2275422798155062, + "grad_norm": 2.2584519386291504, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7048838138580322, + "num_tokens": 286975020.0, + "step": 11178 + }, + { + "epoch": 1.22765209751812, + "grad_norm": 2.150892496109009, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6985301971435547, + "num_tokens": 287005640.0, + "step": 11179 + }, + { + "epoch": 1.2277619152207335, + "grad_norm": 2.1991472244262695, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7255012392997742, + "num_tokens": 287031257.0, + "step": 11180 + }, + { + "epoch": 1.2278717329233473, + "grad_norm": 2.0885260105133057, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6993644833564758, + "num_tokens": 287060929.0, + "step": 11181 + }, + { + "epoch": 1.2279815506259608, + "grad_norm": 2.4791576862335205, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7186532020568848, + "num_tokens": 287083864.0, + "step": 11182 + }, + { + "epoch": 1.2280913683285746, + "grad_norm": 2.228971481323242, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7084183692932129, + "num_tokens": 287110076.0, + "step": 11183 + }, + { + "epoch": 1.228201186031188, + "grad_norm": 2.341869592666626, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7058646082878113, + "num_tokens": 287134856.0, + "step": 11184 + }, + { + "epoch": 1.2283110037338019, + "grad_norm": 2.6333136558532715, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.724766731262207, + "num_tokens": 287153831.0, + "step": 11185 + }, + { + "epoch": 1.2284208214364156, + "grad_norm": 2.103135108947754, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7204643487930298, + "num_tokens": 287183868.0, + "step": 11186 + }, + { + "epoch": 1.2285306391390292, + "grad_norm": 2.3329930305480957, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7119813561439514, + "num_tokens": 287207976.0, + "step": 11187 + }, + { + "epoch": 1.228640456841643, + "grad_norm": 2.4425766468048096, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7081159353256226, + "num_tokens": 287232072.0, + "step": 11188 + }, + { + "epoch": 1.2287502745442564, + "grad_norm": 2.450096845626831, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7269080281257629, + "num_tokens": 287253389.0, + "step": 11189 + }, + { + "epoch": 1.2288600922468702, + "grad_norm": 2.237330913543701, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7096552848815918, + "num_tokens": 287279397.0, + "step": 11190 + }, + { + "epoch": 1.228969909949484, + "grad_norm": 2.3877689838409424, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.714092493057251, + "num_tokens": 287304165.0, + "step": 11191 + }, + { + "epoch": 1.2290797276520975, + "grad_norm": 2.315365791320801, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7179563045501709, + "num_tokens": 287329617.0, + "step": 11192 + }, + { + "epoch": 1.2291895453547113, + "grad_norm": 2.2275147438049316, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6981897354125977, + "num_tokens": 287354855.0, + "step": 11193 + }, + { + "epoch": 1.2292993630573248, + "grad_norm": 2.165442705154419, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7342405319213867, + "num_tokens": 287380487.0, + "step": 11194 + }, + { + "epoch": 1.2294091807599385, + "grad_norm": 2.057300329208374, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7157670259475708, + "num_tokens": 287407145.0, + "step": 11195 + }, + { + "epoch": 1.229518998462552, + "grad_norm": 2.285421371459961, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7046598196029663, + "num_tokens": 287432826.0, + "step": 11196 + }, + { + "epoch": 1.2296288161651658, + "grad_norm": 2.2335262298583984, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7253192663192749, + "num_tokens": 287459319.0, + "step": 11197 + }, + { + "epoch": 1.2297386338677794, + "grad_norm": 2.20908522605896, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7088879346847534, + "num_tokens": 287485504.0, + "step": 11198 + }, + { + "epoch": 1.2298484515703931, + "grad_norm": 2.594036817550659, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7166016101837158, + "num_tokens": 287506334.0, + "step": 11199 + }, + { + "epoch": 1.2299582692730069, + "grad_norm": 2.3190884590148926, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7140119671821594, + "num_tokens": 287530335.0, + "step": 11200 + }, + { + "epoch": 1.2300680869756204, + "grad_norm": 2.171520709991455, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7165791988372803, + "num_tokens": 287559288.0, + "step": 11201 + }, + { + "epoch": 1.2301779046782342, + "grad_norm": 2.2822470664978027, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7057342529296875, + "num_tokens": 287582974.0, + "step": 11202 + }, + { + "epoch": 1.2302877223808477, + "grad_norm": 2.208137273788452, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7172013521194458, + "num_tokens": 287609345.0, + "step": 11203 + }, + { + "epoch": 1.2303975400834615, + "grad_norm": 2.331728935241699, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7104790210723877, + "num_tokens": 287633376.0, + "step": 11204 + }, + { + "epoch": 1.2305073577860752, + "grad_norm": 1.9583195447921753, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.698434591293335, + "num_tokens": 287668276.0, + "step": 11205 + }, + { + "epoch": 1.2306171754886888, + "grad_norm": 2.2626495361328125, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7327711582183838, + "num_tokens": 287693038.0, + "step": 11206 + }, + { + "epoch": 1.2307269931913025, + "grad_norm": 2.3815481662750244, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7008495330810547, + "num_tokens": 287717100.0, + "step": 11207 + }, + { + "epoch": 1.230836810893916, + "grad_norm": 2.0173370838165283, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7067731022834778, + "num_tokens": 287747601.0, + "step": 11208 + }, + { + "epoch": 1.2309466285965298, + "grad_norm": 2.114314317703247, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7369446754455566, + "num_tokens": 287773865.0, + "step": 11209 + }, + { + "epoch": 1.2310564462991433, + "grad_norm": 2.2490837574005127, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7070109844207764, + "num_tokens": 287800456.0, + "step": 11210 + }, + { + "epoch": 1.231166264001757, + "grad_norm": 2.606198310852051, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.715796709060669, + "num_tokens": 287819433.0, + "step": 11211 + }, + { + "epoch": 1.2312760817043706, + "grad_norm": 2.3040077686309814, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7406920790672302, + "num_tokens": 287843132.0, + "step": 11212 + }, + { + "epoch": 1.2313858994069844, + "grad_norm": 2.3445613384246826, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7307710647583008, + "num_tokens": 287867358.0, + "step": 11213 + }, + { + "epoch": 1.2314957171095982, + "grad_norm": 2.1069161891937256, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6982563138008118, + "num_tokens": 287895129.0, + "step": 11214 + }, + { + "epoch": 1.2316055348122117, + "grad_norm": 2.175457000732422, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6978081464767456, + "num_tokens": 287921800.0, + "step": 11215 + }, + { + "epoch": 1.2317153525148254, + "grad_norm": 2.3603885173797607, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7079111933708191, + "num_tokens": 287945416.0, + "step": 11216 + }, + { + "epoch": 1.231825170217439, + "grad_norm": 2.00913405418396, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.720405638217926, + "num_tokens": 287977456.0, + "step": 11217 + }, + { + "epoch": 1.2319349879200527, + "grad_norm": 2.3180785179138184, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7283329367637634, + "num_tokens": 288000056.0, + "step": 11218 + }, + { + "epoch": 1.2320448056226665, + "grad_norm": 2.5193188190460205, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7289154529571533, + "num_tokens": 288021353.0, + "step": 11219 + }, + { + "epoch": 1.23215462332528, + "grad_norm": 2.0486135482788086, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7309233546257019, + "num_tokens": 288050041.0, + "step": 11220 + }, + { + "epoch": 1.2322644410278938, + "grad_norm": 2.460592746734619, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7290492057800293, + "num_tokens": 288071790.0, + "step": 11221 + }, + { + "epoch": 1.2323742587305073, + "grad_norm": 2.397412061691284, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7404592037200928, + "num_tokens": 288095596.0, + "step": 11222 + }, + { + "epoch": 1.232484076433121, + "grad_norm": 2.729107141494751, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7364364862442017, + "num_tokens": 288114013.0, + "step": 11223 + }, + { + "epoch": 1.2325938941357346, + "grad_norm": 2.3213648796081543, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7070094347000122, + "num_tokens": 288139159.0, + "step": 11224 + }, + { + "epoch": 1.2327037118383484, + "grad_norm": 2.0626885890960693, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7187199592590332, + "num_tokens": 288168639.0, + "step": 11225 + }, + { + "epoch": 1.232813529540962, + "grad_norm": 2.1381568908691406, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7125846147537231, + "num_tokens": 288199342.0, + "step": 11226 + }, + { + "epoch": 1.2329233472435757, + "grad_norm": 2.306983470916748, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6929877400398254, + "num_tokens": 288225545.0, + "step": 11227 + }, + { + "epoch": 1.2330331649461894, + "grad_norm": 2.220754384994507, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7196406126022339, + "num_tokens": 288250830.0, + "step": 11228 + }, + { + "epoch": 1.233142982648803, + "grad_norm": 2.2168381214141846, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7122675180435181, + "num_tokens": 288276988.0, + "step": 11229 + }, + { + "epoch": 1.2332528003514167, + "grad_norm": 2.6764767169952393, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.737498939037323, + "num_tokens": 288296087.0, + "step": 11230 + }, + { + "epoch": 1.2333626180540302, + "grad_norm": 2.5368003845214844, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7066975831985474, + "num_tokens": 288317537.0, + "step": 11231 + }, + { + "epoch": 1.233472435756644, + "grad_norm": 2.2519776821136475, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.718218207359314, + "num_tokens": 288342931.0, + "step": 11232 + }, + { + "epoch": 1.2335822534592575, + "grad_norm": 2.0481019020080566, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7059213519096375, + "num_tokens": 288372402.0, + "step": 11233 + }, + { + "epoch": 1.2336920711618713, + "grad_norm": 2.2990915775299072, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7335804104804993, + "num_tokens": 288398325.0, + "step": 11234 + }, + { + "epoch": 1.2338018888644848, + "grad_norm": 2.192257881164551, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.727170467376709, + "num_tokens": 288424882.0, + "step": 11235 + }, + { + "epoch": 1.2339117065670986, + "grad_norm": 2.5127663612365723, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6964583396911621, + "num_tokens": 288448545.0, + "step": 11236 + }, + { + "epoch": 1.2340215242697123, + "grad_norm": 2.365769386291504, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7017731666564941, + "num_tokens": 288474183.0, + "step": 11237 + }, + { + "epoch": 1.2341313419723259, + "grad_norm": 2.1064767837524414, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.712431013584137, + "num_tokens": 288504269.0, + "step": 11238 + }, + { + "epoch": 1.2342411596749396, + "grad_norm": 2.408061981201172, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7150344848632812, + "num_tokens": 288527519.0, + "step": 11239 + }, + { + "epoch": 1.2343509773775532, + "grad_norm": 2.729602575302124, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7260069847106934, + "num_tokens": 288545235.0, + "step": 11240 + }, + { + "epoch": 1.234460795080167, + "grad_norm": 2.397035598754883, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7065609693527222, + "num_tokens": 288570215.0, + "step": 11241 + }, + { + "epoch": 1.2345706127827807, + "grad_norm": 2.084521770477295, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7243496179580688, + "num_tokens": 288597914.0, + "step": 11242 + }, + { + "epoch": 1.2346804304853942, + "grad_norm": 2.18230938911438, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7118357419967651, + "num_tokens": 288625466.0, + "step": 11243 + }, + { + "epoch": 1.234790248188008, + "grad_norm": 2.3611764907836914, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6980606317520142, + "num_tokens": 288648997.0, + "step": 11244 + }, + { + "epoch": 1.2349000658906215, + "grad_norm": 2.2947657108306885, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7299007177352905, + "num_tokens": 288671779.0, + "step": 11245 + }, + { + "epoch": 1.2350098835932353, + "grad_norm": 2.2159085273742676, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7263510227203369, + "num_tokens": 288697369.0, + "step": 11246 + }, + { + "epoch": 1.2351197012958488, + "grad_norm": 2.0551300048828125, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7084372043609619, + "num_tokens": 288728451.0, + "step": 11247 + }, + { + "epoch": 1.2352295189984626, + "grad_norm": 2.1527199745178223, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.704066812992096, + "num_tokens": 288755483.0, + "step": 11248 + }, + { + "epoch": 1.235339336701076, + "grad_norm": 2.299647092819214, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7131248712539673, + "num_tokens": 288779310.0, + "step": 11249 + }, + { + "epoch": 1.2354491544036899, + "grad_norm": 2.214414119720459, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7291145324707031, + "num_tokens": 288805298.0, + "step": 11250 + }, + { + "epoch": 1.2355589721063036, + "grad_norm": 2.060795783996582, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7294824123382568, + "num_tokens": 288836602.0, + "step": 11251 + }, + { + "epoch": 1.2356687898089171, + "grad_norm": 2.428691864013672, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6987468004226685, + "num_tokens": 288860165.0, + "step": 11252 + }, + { + "epoch": 1.235778607511531, + "grad_norm": 2.063784122467041, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7070383429527283, + "num_tokens": 288890152.0, + "step": 11253 + }, + { + "epoch": 1.2358884252141444, + "grad_norm": 2.258394479751587, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7057255506515503, + "num_tokens": 288916218.0, + "step": 11254 + }, + { + "epoch": 1.2359982429167582, + "grad_norm": 2.393505573272705, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7091101408004761, + "num_tokens": 288941026.0, + "step": 11255 + }, + { + "epoch": 1.236108060619372, + "grad_norm": 2.7130284309387207, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7304768562316895, + "num_tokens": 288958991.0, + "step": 11256 + }, + { + "epoch": 1.2362178783219855, + "grad_norm": 2.280620574951172, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7104537487030029, + "num_tokens": 288983573.0, + "step": 11257 + }, + { + "epoch": 1.2363276960245992, + "grad_norm": 2.133608341217041, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7109271883964539, + "num_tokens": 289012208.0, + "step": 11258 + }, + { + "epoch": 1.2364375137272128, + "grad_norm": 2.243476152420044, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7143785953521729, + "num_tokens": 289037906.0, + "step": 11259 + }, + { + "epoch": 1.2365473314298265, + "grad_norm": 2.176910877227783, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7091394662857056, + "num_tokens": 289063919.0, + "step": 11260 + }, + { + "epoch": 1.23665714913244, + "grad_norm": 2.2311959266662598, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7037662267684937, + "num_tokens": 289090516.0, + "step": 11261 + }, + { + "epoch": 1.2367669668350538, + "grad_norm": 2.122891902923584, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7152254581451416, + "num_tokens": 289119973.0, + "step": 11262 + }, + { + "epoch": 1.2368767845376674, + "grad_norm": 2.2145004272460938, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7094138264656067, + "num_tokens": 289148717.0, + "step": 11263 + }, + { + "epoch": 1.2369866022402811, + "grad_norm": 2.314551830291748, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7335095405578613, + "num_tokens": 289172855.0, + "step": 11264 + }, + { + "epoch": 1.2370964199428949, + "grad_norm": 2.1179401874542236, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6960639953613281, + "num_tokens": 289201142.0, + "step": 11265 + }, + { + "epoch": 1.2372062376455084, + "grad_norm": 2.3636386394500732, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7610299587249756, + "num_tokens": 289223771.0, + "step": 11266 + }, + { + "epoch": 1.2373160553481222, + "grad_norm": 2.3682405948638916, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7153292894363403, + "num_tokens": 289246566.0, + "step": 11267 + }, + { + "epoch": 1.2374258730507357, + "grad_norm": 2.3486623764038086, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7229294776916504, + "num_tokens": 289269829.0, + "step": 11268 + }, + { + "epoch": 1.2375356907533495, + "grad_norm": 2.2842390537261963, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7210227251052856, + "num_tokens": 289295699.0, + "step": 11269 + }, + { + "epoch": 1.2376455084559632, + "grad_norm": 2.248241901397705, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7192330360412598, + "num_tokens": 289320260.0, + "step": 11270 + }, + { + "epoch": 1.2377553261585768, + "grad_norm": 2.196840286254883, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7099867463111877, + "num_tokens": 289347171.0, + "step": 11271 + }, + { + "epoch": 1.2378651438611905, + "grad_norm": 2.4148662090301514, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7156021595001221, + "num_tokens": 289370629.0, + "step": 11272 + }, + { + "epoch": 1.237974961563804, + "grad_norm": 2.272658586502075, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7247508764266968, + "num_tokens": 289396177.0, + "step": 11273 + }, + { + "epoch": 1.2380847792664178, + "grad_norm": 2.436344623565674, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7311270833015442, + "num_tokens": 289418093.0, + "step": 11274 + }, + { + "epoch": 1.2381945969690313, + "grad_norm": 2.3551878929138184, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7297418713569641, + "num_tokens": 289440125.0, + "step": 11275 + }, + { + "epoch": 1.238304414671645, + "grad_norm": 2.1517322063446045, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7121061086654663, + "num_tokens": 289467406.0, + "step": 11276 + }, + { + "epoch": 1.2384142323742586, + "grad_norm": 2.2477035522460938, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7172347903251648, + "num_tokens": 289493469.0, + "step": 11277 + }, + { + "epoch": 1.2385240500768724, + "grad_norm": 2.6490538120269775, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.704430878162384, + "num_tokens": 289513528.0, + "step": 11278 + }, + { + "epoch": 1.2386338677794861, + "grad_norm": 2.611064910888672, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7445217967033386, + "num_tokens": 289533111.0, + "step": 11279 + }, + { + "epoch": 1.2387436854820997, + "grad_norm": 2.503749132156372, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7202195525169373, + "num_tokens": 289554466.0, + "step": 11280 + }, + { + "epoch": 1.2388535031847134, + "grad_norm": 2.042179584503174, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7140977382659912, + "num_tokens": 289585962.0, + "step": 11281 + }, + { + "epoch": 1.238963320887327, + "grad_norm": 1.935388207435608, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7147676944732666, + "num_tokens": 289616902.0, + "step": 11282 + }, + { + "epoch": 1.2390731385899407, + "grad_norm": 2.4634156227111816, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7104247212409973, + "num_tokens": 289637989.0, + "step": 11283 + }, + { + "epoch": 1.2391829562925545, + "grad_norm": 2.215874433517456, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7243776321411133, + "num_tokens": 289663439.0, + "step": 11284 + }, + { + "epoch": 1.239292773995168, + "grad_norm": 2.356799364089966, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7076080441474915, + "num_tokens": 289689070.0, + "step": 11285 + }, + { + "epoch": 1.2394025916977818, + "grad_norm": 2.138105630874634, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6795207262039185, + "num_tokens": 289719896.0, + "step": 11286 + }, + { + "epoch": 1.2395124094003953, + "grad_norm": 2.2638907432556152, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7074443697929382, + "num_tokens": 289745129.0, + "step": 11287 + }, + { + "epoch": 1.239622227103009, + "grad_norm": 2.4029810428619385, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7198617458343506, + "num_tokens": 289768238.0, + "step": 11288 + }, + { + "epoch": 1.2397320448056226, + "grad_norm": 2.192140579223633, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7154533863067627, + "num_tokens": 289795265.0, + "step": 11289 + }, + { + "epoch": 1.2398418625082364, + "grad_norm": 2.2284538745880127, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7253150343894958, + "num_tokens": 289824060.0, + "step": 11290 + }, + { + "epoch": 1.23995168021085, + "grad_norm": 2.072685718536377, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7364726066589355, + "num_tokens": 289852206.0, + "step": 11291 + }, + { + "epoch": 1.2400614979134637, + "grad_norm": 2.379948854446411, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7276399731636047, + "num_tokens": 289875557.0, + "step": 11292 + }, + { + "epoch": 1.2401713156160774, + "grad_norm": 2.265758991241455, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7279911041259766, + "num_tokens": 289901240.0, + "step": 11293 + }, + { + "epoch": 1.240281133318691, + "grad_norm": 2.369279384613037, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7065258026123047, + "num_tokens": 289927094.0, + "step": 11294 + }, + { + "epoch": 1.2403909510213047, + "grad_norm": 2.206684112548828, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6917161345481873, + "num_tokens": 289955576.0, + "step": 11295 + }, + { + "epoch": 1.2405007687239182, + "grad_norm": 2.427767038345337, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7300797700881958, + "num_tokens": 289978866.0, + "step": 11296 + }, + { + "epoch": 1.240610586426532, + "grad_norm": 2.3563694953918457, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7483132481575012, + "num_tokens": 290003812.0, + "step": 11297 + }, + { + "epoch": 1.2407204041291455, + "grad_norm": 2.0093071460723877, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7058106064796448, + "num_tokens": 290035239.0, + "step": 11298 + }, + { + "epoch": 1.2408302218317593, + "grad_norm": 2.2655773162841797, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.713126540184021, + "num_tokens": 290059950.0, + "step": 11299 + }, + { + "epoch": 1.2409400395343728, + "grad_norm": 2.2520711421966553, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7283040285110474, + "num_tokens": 290086262.0, + "step": 11300 + }, + { + "epoch": 1.2410498572369866, + "grad_norm": 2.6440773010253906, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7201831340789795, + "num_tokens": 290106700.0, + "step": 11301 + }, + { + "epoch": 1.2411596749396003, + "grad_norm": 2.332956075668335, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7277379035949707, + "num_tokens": 290129174.0, + "step": 11302 + }, + { + "epoch": 1.2412694926422139, + "grad_norm": 2.2348716259002686, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.7016212940216064, + "num_tokens": 290158485.0, + "step": 11303 + }, + { + "epoch": 1.2413793103448276, + "grad_norm": 2.1910059452056885, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6921910047531128, + "num_tokens": 290186725.0, + "step": 11304 + }, + { + "epoch": 1.2414891280474412, + "grad_norm": 2.3681142330169678, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7180354595184326, + "num_tokens": 290209815.0, + "step": 11305 + }, + { + "epoch": 1.241598945750055, + "grad_norm": 2.4084577560424805, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7394852638244629, + "num_tokens": 290231925.0, + "step": 11306 + }, + { + "epoch": 1.2417087634526687, + "grad_norm": 2.3979296684265137, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.717594563961029, + "num_tokens": 290254047.0, + "step": 11307 + }, + { + "epoch": 1.2418185811552822, + "grad_norm": 2.327014923095703, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.733421802520752, + "num_tokens": 290277722.0, + "step": 11308 + }, + { + "epoch": 1.241928398857896, + "grad_norm": 2.136228084564209, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7208422422409058, + "num_tokens": 290304301.0, + "step": 11309 + }, + { + "epoch": 1.2420382165605095, + "grad_norm": 2.167466402053833, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.72181236743927, + "num_tokens": 290330194.0, + "step": 11310 + }, + { + "epoch": 1.2421480342631233, + "grad_norm": 2.4766952991485596, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7019248008728027, + "num_tokens": 290352975.0, + "step": 11311 + }, + { + "epoch": 1.2422578519657368, + "grad_norm": 2.3730385303497314, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7201415300369263, + "num_tokens": 290377894.0, + "step": 11312 + }, + { + "epoch": 1.2423676696683505, + "grad_norm": 2.246171474456787, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.732583224773407, + "num_tokens": 290406240.0, + "step": 11313 + }, + { + "epoch": 1.242477487370964, + "grad_norm": 1.9810450077056885, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7060505747795105, + "num_tokens": 290434861.0, + "step": 11314 + }, + { + "epoch": 1.2425873050735778, + "grad_norm": 2.379910945892334, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7283221483230591, + "num_tokens": 290457621.0, + "step": 11315 + }, + { + "epoch": 1.2426971227761916, + "grad_norm": 2.1774396896362305, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.720766007900238, + "num_tokens": 290485037.0, + "step": 11316 + }, + { + "epoch": 1.2428069404788051, + "grad_norm": 2.2436912059783936, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7449369430541992, + "num_tokens": 290508370.0, + "step": 11317 + }, + { + "epoch": 1.242916758181419, + "grad_norm": 2.354959011077881, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7169433236122131, + "num_tokens": 290531211.0, + "step": 11318 + }, + { + "epoch": 1.2430265758840324, + "grad_norm": 2.2246289253234863, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7428828477859497, + "num_tokens": 290554838.0, + "step": 11319 + }, + { + "epoch": 1.2431363935866462, + "grad_norm": 2.0163753032684326, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6968672275543213, + "num_tokens": 290586309.0, + "step": 11320 + }, + { + "epoch": 1.24324621128926, + "grad_norm": 2.1753616333007812, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7262675166130066, + "num_tokens": 290611707.0, + "step": 11321 + }, + { + "epoch": 1.2433560289918735, + "grad_norm": 2.674755096435547, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7263514399528503, + "num_tokens": 290631651.0, + "step": 11322 + }, + { + "epoch": 1.2434658466944872, + "grad_norm": 2.0550897121429443, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.724346399307251, + "num_tokens": 290662085.0, + "step": 11323 + }, + { + "epoch": 1.2435756643971008, + "grad_norm": 2.0861754417419434, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6998917460441589, + "num_tokens": 290691944.0, + "step": 11324 + }, + { + "epoch": 1.2436854820997145, + "grad_norm": 2.402467966079712, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7234352827072144, + "num_tokens": 290715479.0, + "step": 11325 + }, + { + "epoch": 1.243795299802328, + "grad_norm": 2.1203880310058594, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7343080639839172, + "num_tokens": 290745513.0, + "step": 11326 + }, + { + "epoch": 1.2439051175049418, + "grad_norm": 2.2479407787323, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7176882028579712, + "num_tokens": 290772648.0, + "step": 11327 + }, + { + "epoch": 1.2440149352075554, + "grad_norm": 2.270944118499756, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7255553603172302, + "num_tokens": 290796494.0, + "step": 11328 + }, + { + "epoch": 1.244124752910169, + "grad_norm": 2.282586097717285, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7153689861297607, + "num_tokens": 290821218.0, + "step": 11329 + }, + { + "epoch": 1.2442345706127829, + "grad_norm": 2.2732133865356445, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.728653609752655, + "num_tokens": 290847790.0, + "step": 11330 + }, + { + "epoch": 1.2443443883153964, + "grad_norm": 2.5955941677093506, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7419999837875366, + "num_tokens": 290868797.0, + "step": 11331 + }, + { + "epoch": 1.2444542060180102, + "grad_norm": 2.1404242515563965, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7211151123046875, + "num_tokens": 290896905.0, + "step": 11332 + }, + { + "epoch": 1.2445640237206237, + "grad_norm": 2.1960108280181885, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7251192331314087, + "num_tokens": 290924380.0, + "step": 11333 + }, + { + "epoch": 1.2446738414232374, + "grad_norm": 2.254830837249756, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7262251377105713, + "num_tokens": 290947834.0, + "step": 11334 + }, + { + "epoch": 1.2447836591258512, + "grad_norm": 2.2723443508148193, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7206829190254211, + "num_tokens": 290974878.0, + "step": 11335 + }, + { + "epoch": 1.2448934768284647, + "grad_norm": 2.248595952987671, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7313084602355957, + "num_tokens": 291001487.0, + "step": 11336 + }, + { + "epoch": 1.2450032945310785, + "grad_norm": 2.1670942306518555, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7267196774482727, + "num_tokens": 291028603.0, + "step": 11337 + }, + { + "epoch": 1.245113112233692, + "grad_norm": 2.2860755920410156, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7032740116119385, + "num_tokens": 291054190.0, + "step": 11338 + }, + { + "epoch": 1.2452229299363058, + "grad_norm": 2.3450756072998047, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7260112762451172, + "num_tokens": 291075432.0, + "step": 11339 + }, + { + "epoch": 1.2453327476389193, + "grad_norm": 2.410693883895874, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7264896631240845, + "num_tokens": 291099619.0, + "step": 11340 + }, + { + "epoch": 1.245442565341533, + "grad_norm": 2.1097421646118164, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7066895365715027, + "num_tokens": 291126245.0, + "step": 11341 + }, + { + "epoch": 1.2455523830441466, + "grad_norm": 2.0265581607818604, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7008138298988342, + "num_tokens": 291155751.0, + "step": 11342 + }, + { + "epoch": 1.2456622007467604, + "grad_norm": 2.7069191932678223, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7483338713645935, + "num_tokens": 291175468.0, + "step": 11343 + }, + { + "epoch": 1.2457720184493741, + "grad_norm": 2.6410739421844482, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6984050273895264, + "num_tokens": 291198658.0, + "step": 11344 + }, + { + "epoch": 1.2458818361519877, + "grad_norm": 2.2772767543792725, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7141426801681519, + "num_tokens": 291224268.0, + "step": 11345 + }, + { + "epoch": 1.2459916538546014, + "grad_norm": 2.34263014793396, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.714522123336792, + "num_tokens": 291247198.0, + "step": 11346 + }, + { + "epoch": 1.246101471557215, + "grad_norm": 2.384843111038208, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7385774850845337, + "num_tokens": 291270769.0, + "step": 11347 + }, + { + "epoch": 1.2462112892598287, + "grad_norm": 2.1458911895751953, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.723841667175293, + "num_tokens": 291298120.0, + "step": 11348 + }, + { + "epoch": 1.2463211069624422, + "grad_norm": 2.3331642150878906, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6946260333061218, + "num_tokens": 291324237.0, + "step": 11349 + }, + { + "epoch": 1.246430924665056, + "grad_norm": 2.0467066764831543, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7026140689849854, + "num_tokens": 291354252.0, + "step": 11350 + }, + { + "epoch": 1.2465407423676698, + "grad_norm": 2.1436266899108887, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7168042063713074, + "num_tokens": 291381094.0, + "step": 11351 + }, + { + "epoch": 1.2466505600702833, + "grad_norm": 2.216895341873169, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7071268558502197, + "num_tokens": 291407553.0, + "step": 11352 + }, + { + "epoch": 1.246760377772897, + "grad_norm": 2.255183696746826, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7159667611122131, + "num_tokens": 291433142.0, + "step": 11353 + }, + { + "epoch": 1.2468701954755106, + "grad_norm": 2.356841564178467, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7153966426849365, + "num_tokens": 291457051.0, + "step": 11354 + }, + { + "epoch": 1.2469800131781243, + "grad_norm": 2.545409917831421, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7318768501281738, + "num_tokens": 291476862.0, + "step": 11355 + }, + { + "epoch": 1.2470898308807379, + "grad_norm": 2.1237432956695557, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.708757758140564, + "num_tokens": 291506218.0, + "step": 11356 + }, + { + "epoch": 1.2471996485833516, + "grad_norm": 2.0058460235595703, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6976783275604248, + "num_tokens": 291539550.0, + "step": 11357 + }, + { + "epoch": 1.2473094662859654, + "grad_norm": 2.213987112045288, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7214412689208984, + "num_tokens": 291565872.0, + "step": 11358 + }, + { + "epoch": 1.247419283988579, + "grad_norm": 1.963752269744873, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7197649478912354, + "num_tokens": 291596257.0, + "step": 11359 + }, + { + "epoch": 1.2475291016911927, + "grad_norm": 2.0835113525390625, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7035975456237793, + "num_tokens": 291626107.0, + "step": 11360 + }, + { + "epoch": 1.2476389193938062, + "grad_norm": 2.4284136295318604, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7378975749015808, + "num_tokens": 291648216.0, + "step": 11361 + }, + { + "epoch": 1.24774873709642, + "grad_norm": 2.054466485977173, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.711269736289978, + "num_tokens": 291676431.0, + "step": 11362 + }, + { + "epoch": 1.2478585547990335, + "grad_norm": 2.463693857192993, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7326979637145996, + "num_tokens": 291697796.0, + "step": 11363 + }, + { + "epoch": 1.2479683725016473, + "grad_norm": 2.3455588817596436, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7284568548202515, + "num_tokens": 291721747.0, + "step": 11364 + }, + { + "epoch": 1.2480781902042608, + "grad_norm": 2.6464505195617676, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7357441186904907, + "num_tokens": 291740388.0, + "step": 11365 + }, + { + "epoch": 1.2481880079068746, + "grad_norm": 2.2070631980895996, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7119989395141602, + "num_tokens": 291765395.0, + "step": 11366 + }, + { + "epoch": 1.2482978256094883, + "grad_norm": 2.109351396560669, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7088950276374817, + "num_tokens": 291795573.0, + "step": 11367 + }, + { + "epoch": 1.2484076433121019, + "grad_norm": 2.1204135417938232, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7322027683258057, + "num_tokens": 291822670.0, + "step": 11368 + }, + { + "epoch": 1.2485174610147156, + "grad_norm": 2.4885785579681396, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7296626567840576, + "num_tokens": 291843785.0, + "step": 11369 + }, + { + "epoch": 1.2486272787173291, + "grad_norm": 2.340378522872925, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7063493728637695, + "num_tokens": 291867937.0, + "step": 11370 + }, + { + "epoch": 1.248737096419943, + "grad_norm": 2.0958821773529053, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7081456780433655, + "num_tokens": 291897909.0, + "step": 11371 + }, + { + "epoch": 1.2488469141225567, + "grad_norm": 2.155890703201294, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.727500319480896, + "num_tokens": 291926186.0, + "step": 11372 + }, + { + "epoch": 1.2489567318251702, + "grad_norm": 2.2068326473236084, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7012301087379456, + "num_tokens": 291952378.0, + "step": 11373 + }, + { + "epoch": 1.249066549527784, + "grad_norm": 2.324831008911133, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7470874786376953, + "num_tokens": 291975557.0, + "step": 11374 + }, + { + "epoch": 1.2491763672303975, + "grad_norm": 2.4185564517974854, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7235682010650635, + "num_tokens": 291998016.0, + "step": 11375 + }, + { + "epoch": 1.2492861849330112, + "grad_norm": 2.196122646331787, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6895885467529297, + "num_tokens": 292025483.0, + "step": 11376 + }, + { + "epoch": 1.2493960026356248, + "grad_norm": 2.4451918601989746, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7464878559112549, + "num_tokens": 292046539.0, + "step": 11377 + }, + { + "epoch": 1.2495058203382385, + "grad_norm": 2.500020742416382, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7248028516769409, + "num_tokens": 292068450.0, + "step": 11378 + }, + { + "epoch": 1.249615638040852, + "grad_norm": 2.09220814704895, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7142478227615356, + "num_tokens": 292095686.0, + "step": 11379 + }, + { + "epoch": 1.2497254557434658, + "grad_norm": 2.193957805633545, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7324844598770142, + "num_tokens": 292121058.0, + "step": 11380 + }, + { + "epoch": 1.2498352734460796, + "grad_norm": 2.2209277153015137, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7082858085632324, + "num_tokens": 292146371.0, + "step": 11381 + }, + { + "epoch": 1.2499450911486931, + "grad_norm": 2.0830538272857666, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7107772827148438, + "num_tokens": 292173006.0, + "step": 11382 + }, + { + "epoch": 1.2500549088513069, + "grad_norm": 2.1442959308624268, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7042176723480225, + "num_tokens": 292201154.0, + "step": 11383 + }, + { + "epoch": 1.2501647265539204, + "grad_norm": 2.1093618869781494, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7168829441070557, + "num_tokens": 292226296.0, + "step": 11384 + }, + { + "epoch": 1.2502745442565342, + "grad_norm": 2.3206100463867188, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7203864455223083, + "num_tokens": 292250577.0, + "step": 11385 + }, + { + "epoch": 1.250384361959148, + "grad_norm": 2.660994529724121, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7186678647994995, + "num_tokens": 292271340.0, + "step": 11386 + }, + { + "epoch": 1.2504941796617615, + "grad_norm": 2.3807342052459717, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.716839075088501, + "num_tokens": 292293197.0, + "step": 11387 + }, + { + "epoch": 1.2506039973643752, + "grad_norm": 2.443305492401123, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7201198935508728, + "num_tokens": 292315173.0, + "step": 11388 + }, + { + "epoch": 1.2507138150669888, + "grad_norm": 2.3279662132263184, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.75123131275177, + "num_tokens": 292339528.0, + "step": 11389 + }, + { + "epoch": 1.2508236327696025, + "grad_norm": 1.990586757659912, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6848635673522949, + "num_tokens": 292371812.0, + "step": 11390 + }, + { + "epoch": 1.250933450472216, + "grad_norm": 2.1452748775482178, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.708680272102356, + "num_tokens": 292402327.0, + "step": 11391 + }, + { + "epoch": 1.2510432681748298, + "grad_norm": 2.2909841537475586, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7287344932556152, + "num_tokens": 292425608.0, + "step": 11392 + }, + { + "epoch": 1.2511530858774433, + "grad_norm": 2.569077253341675, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7458524107933044, + "num_tokens": 292444293.0, + "step": 11393 + }, + { + "epoch": 1.251262903580057, + "grad_norm": 2.2208969593048096, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.738888144493103, + "num_tokens": 292470053.0, + "step": 11394 + }, + { + "epoch": 1.2513727212826709, + "grad_norm": 2.5161948204040527, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7319855093955994, + "num_tokens": 292490947.0, + "step": 11395 + }, + { + "epoch": 1.2514825389852844, + "grad_norm": 2.5245532989501953, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7152112722396851, + "num_tokens": 292514778.0, + "step": 11396 + }, + { + "epoch": 1.2515923566878981, + "grad_norm": 2.484522581100464, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7434099912643433, + "num_tokens": 292536793.0, + "step": 11397 + }, + { + "epoch": 1.2517021743905117, + "grad_norm": 2.030717372894287, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7097975015640259, + "num_tokens": 292567030.0, + "step": 11398 + }, + { + "epoch": 1.2518119920931254, + "grad_norm": 2.2565412521362305, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.728965163230896, + "num_tokens": 292593754.0, + "step": 11399 + }, + { + "epoch": 1.2519218097957392, + "grad_norm": 2.397127866744995, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7309820652008057, + "num_tokens": 292617121.0, + "step": 11400 + }, + { + "epoch": 1.2520316274983527, + "grad_norm": 2.1715333461761475, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7095882892608643, + "num_tokens": 292644387.0, + "step": 11401 + }, + { + "epoch": 1.2521414452009663, + "grad_norm": 2.059749126434326, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7180594205856323, + "num_tokens": 292673811.0, + "step": 11402 + }, + { + "epoch": 1.25225126290358, + "grad_norm": 2.402820587158203, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7060667276382446, + "num_tokens": 292698618.0, + "step": 11403 + }, + { + "epoch": 1.2523610806061938, + "grad_norm": 2.2749440670013428, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7278504967689514, + "num_tokens": 292722588.0, + "step": 11404 + }, + { + "epoch": 1.2524708983088073, + "grad_norm": 2.148333787918091, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7207309007644653, + "num_tokens": 292749137.0, + "step": 11405 + }, + { + "epoch": 1.252580716011421, + "grad_norm": 2.312276601791382, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7340325713157654, + "num_tokens": 292773458.0, + "step": 11406 + }, + { + "epoch": 1.2526905337140346, + "grad_norm": 2.3886210918426514, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7323565483093262, + "num_tokens": 292797979.0, + "step": 11407 + }, + { + "epoch": 1.2528003514166484, + "grad_norm": 2.1307804584503174, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.74245285987854, + "num_tokens": 292826481.0, + "step": 11408 + }, + { + "epoch": 1.2529101691192621, + "grad_norm": 2.473588705062866, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7271912693977356, + "num_tokens": 292847276.0, + "step": 11409 + }, + { + "epoch": 1.2530199868218757, + "grad_norm": 2.3413214683532715, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7240768671035767, + "num_tokens": 292871425.0, + "step": 11410 + }, + { + "epoch": 1.2531298045244894, + "grad_norm": 2.1795427799224854, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7213858366012573, + "num_tokens": 292897525.0, + "step": 11411 + }, + { + "epoch": 1.253239622227103, + "grad_norm": 2.505000114440918, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7000001072883606, + "num_tokens": 292919353.0, + "step": 11412 + }, + { + "epoch": 1.2533494399297167, + "grad_norm": 2.205378532409668, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7324735522270203, + "num_tokens": 292945516.0, + "step": 11413 + }, + { + "epoch": 1.2534592576323305, + "grad_norm": 2.23232102394104, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.718387246131897, + "num_tokens": 292971977.0, + "step": 11414 + }, + { + "epoch": 1.253569075334944, + "grad_norm": 2.144009590148926, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7205893397331238, + "num_tokens": 293000532.0, + "step": 11415 + }, + { + "epoch": 1.2536788930375575, + "grad_norm": 2.123830556869507, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7083885669708252, + "num_tokens": 293031140.0, + "step": 11416 + }, + { + "epoch": 1.2537887107401713, + "grad_norm": 2.0596437454223633, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7151592969894409, + "num_tokens": 293059736.0, + "step": 11417 + }, + { + "epoch": 1.253898528442785, + "grad_norm": 2.19993257522583, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7283552289009094, + "num_tokens": 293085765.0, + "step": 11418 + }, + { + "epoch": 1.2540083461453986, + "grad_norm": 2.3108155727386475, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7227827310562134, + "num_tokens": 293110869.0, + "step": 11419 + }, + { + "epoch": 1.2541181638480123, + "grad_norm": 2.1217856407165527, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7171899080276489, + "num_tokens": 293138449.0, + "step": 11420 + }, + { + "epoch": 1.2542279815506259, + "grad_norm": 2.1039206981658936, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7188799381256104, + "num_tokens": 293167436.0, + "step": 11421 + }, + { + "epoch": 1.2543377992532396, + "grad_norm": 2.0616087913513184, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7139549255371094, + "num_tokens": 293197354.0, + "step": 11422 + }, + { + "epoch": 1.2544476169558534, + "grad_norm": 2.1413540840148926, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7217418551445007, + "num_tokens": 293224006.0, + "step": 11423 + }, + { + "epoch": 1.254557434658467, + "grad_norm": 2.0886919498443604, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.724430501461029, + "num_tokens": 293251434.0, + "step": 11424 + }, + { + "epoch": 1.2546672523610807, + "grad_norm": 2.392280340194702, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7335385084152222, + "num_tokens": 293273554.0, + "step": 11425 + }, + { + "epoch": 1.2547770700636942, + "grad_norm": 2.4891552925109863, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.6999703049659729, + "num_tokens": 293296637.0, + "step": 11426 + }, + { + "epoch": 1.254886887766308, + "grad_norm": 2.1285641193389893, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7142787575721741, + "num_tokens": 293324872.0, + "step": 11427 + }, + { + "epoch": 1.2549967054689217, + "grad_norm": 2.574404716491699, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7286289930343628, + "num_tokens": 293344318.0, + "step": 11428 + }, + { + "epoch": 1.2551065231715353, + "grad_norm": 2.258176565170288, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.723845362663269, + "num_tokens": 293370232.0, + "step": 11429 + }, + { + "epoch": 1.2552163408741488, + "grad_norm": 2.077256441116333, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6956050992012024, + "num_tokens": 293401656.0, + "step": 11430 + }, + { + "epoch": 1.2553261585767626, + "grad_norm": 2.6355230808258057, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7257764339447021, + "num_tokens": 293420379.0, + "step": 11431 + }, + { + "epoch": 1.2554359762793763, + "grad_norm": 2.441981554031372, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7338113188743591, + "num_tokens": 293443179.0, + "step": 11432 + }, + { + "epoch": 1.2555457939819898, + "grad_norm": 2.1706292629241943, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7254249453544617, + "num_tokens": 293470403.0, + "step": 11433 + }, + { + "epoch": 1.2556556116846036, + "grad_norm": 2.2007336616516113, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7170525193214417, + "num_tokens": 293496217.0, + "step": 11434 + }, + { + "epoch": 1.2557654293872171, + "grad_norm": 2.237184524536133, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.726172924041748, + "num_tokens": 293520869.0, + "step": 11435 + }, + { + "epoch": 1.255875247089831, + "grad_norm": 2.0134694576263428, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6922227740287781, + "num_tokens": 293555506.0, + "step": 11436 + }, + { + "epoch": 1.2559850647924446, + "grad_norm": 2.231072187423706, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7292238473892212, + "num_tokens": 293580610.0, + "step": 11437 + }, + { + "epoch": 1.2560948824950582, + "grad_norm": 2.142368793487549, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7094159126281738, + "num_tokens": 293609648.0, + "step": 11438 + }, + { + "epoch": 1.256204700197672, + "grad_norm": 2.169316530227661, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7061440944671631, + "num_tokens": 293639410.0, + "step": 11439 + }, + { + "epoch": 1.2563145179002855, + "grad_norm": 2.5542209148406982, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7066928148269653, + "num_tokens": 293660314.0, + "step": 11440 + }, + { + "epoch": 1.2564243356028992, + "grad_norm": 2.223496675491333, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7172306776046753, + "num_tokens": 293684666.0, + "step": 11441 + }, + { + "epoch": 1.2565341533055128, + "grad_norm": 2.6112427711486816, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7272539138793945, + "num_tokens": 293705268.0, + "step": 11442 + }, + { + "epoch": 1.2566439710081265, + "grad_norm": 2.0573627948760986, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7149887084960938, + "num_tokens": 293734366.0, + "step": 11443 + }, + { + "epoch": 1.25675378871074, + "grad_norm": 2.5543291568756104, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7489956617355347, + "num_tokens": 293752974.0, + "step": 11444 + }, + { + "epoch": 1.2568636064133538, + "grad_norm": 2.2287559509277344, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7321657538414001, + "num_tokens": 293777187.0, + "step": 11445 + }, + { + "epoch": 1.2569734241159676, + "grad_norm": 2.0444176197052, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7340595722198486, + "num_tokens": 293806012.0, + "step": 11446 + }, + { + "epoch": 1.257083241818581, + "grad_norm": 2.1153886318206787, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7227852940559387, + "num_tokens": 293835371.0, + "step": 11447 + }, + { + "epoch": 1.2571930595211949, + "grad_norm": 2.1419999599456787, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7024728655815125, + "num_tokens": 293863456.0, + "step": 11448 + }, + { + "epoch": 1.2573028772238084, + "grad_norm": 2.0918545722961426, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7241429686546326, + "num_tokens": 293891265.0, + "step": 11449 + }, + { + "epoch": 1.2574126949264222, + "grad_norm": 2.4170637130737305, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7217608690261841, + "num_tokens": 293912543.0, + "step": 11450 + }, + { + "epoch": 1.257522512629036, + "grad_norm": 1.9522311687469482, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7136267423629761, + "num_tokens": 293944126.0, + "step": 11451 + }, + { + "epoch": 1.2576323303316495, + "grad_norm": 2.4538536071777344, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7332526445388794, + "num_tokens": 293965922.0, + "step": 11452 + }, + { + "epoch": 1.257742148034263, + "grad_norm": 2.34014630317688, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7092311382293701, + "num_tokens": 293990749.0, + "step": 11453 + }, + { + "epoch": 1.2578519657368767, + "grad_norm": 2.84918212890625, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7411899566650391, + "num_tokens": 294008483.0, + "step": 11454 + }, + { + "epoch": 1.2579617834394905, + "grad_norm": 2.0112366676330566, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7218722105026245, + "num_tokens": 294038608.0, + "step": 11455 + }, + { + "epoch": 1.258071601142104, + "grad_norm": 2.2225680351257324, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7275387048721313, + "num_tokens": 294065278.0, + "step": 11456 + }, + { + "epoch": 1.2581814188447178, + "grad_norm": 2.3734982013702393, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7219899892807007, + "num_tokens": 294090527.0, + "step": 11457 + }, + { + "epoch": 1.2582912365473313, + "grad_norm": 2.0984079837799072, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7294171452522278, + "num_tokens": 294119298.0, + "step": 11458 + }, + { + "epoch": 1.258401054249945, + "grad_norm": 1.943918228149414, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7352038621902466, + "num_tokens": 294151284.0, + "step": 11459 + }, + { + "epoch": 1.2585108719525588, + "grad_norm": 1.9983032941818237, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7081562280654907, + "num_tokens": 294183992.0, + "step": 11460 + }, + { + "epoch": 1.2586206896551724, + "grad_norm": 2.0888471603393555, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7337750792503357, + "num_tokens": 294212605.0, + "step": 11461 + }, + { + "epoch": 1.2587305073577861, + "grad_norm": 2.3446450233459473, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7068570852279663, + "num_tokens": 294235690.0, + "step": 11462 + }, + { + "epoch": 1.2588403250603997, + "grad_norm": 2.2857847213745117, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7206990718841553, + "num_tokens": 294261579.0, + "step": 11463 + }, + { + "epoch": 1.2589501427630134, + "grad_norm": 1.9783170223236084, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6932069063186646, + "num_tokens": 294295050.0, + "step": 11464 + }, + { + "epoch": 1.2590599604656272, + "grad_norm": 2.5258607864379883, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.738469123840332, + "num_tokens": 294314802.0, + "step": 11465 + }, + { + "epoch": 1.2591697781682407, + "grad_norm": 2.29429030418396, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7350223064422607, + "num_tokens": 294338962.0, + "step": 11466 + }, + { + "epoch": 1.2592795958708543, + "grad_norm": 1.9889380931854248, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6981557607650757, + "num_tokens": 294368949.0, + "step": 11467 + }, + { + "epoch": 1.259389413573468, + "grad_norm": 2.269782781600952, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7416925430297852, + "num_tokens": 294390384.0, + "step": 11468 + }, + { + "epoch": 1.2594992312760818, + "grad_norm": 2.1347477436065674, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7169605493545532, + "num_tokens": 294415476.0, + "step": 11469 + }, + { + "epoch": 1.2596090489786953, + "grad_norm": 2.294468402862549, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6970160603523254, + "num_tokens": 294440383.0, + "step": 11470 + }, + { + "epoch": 1.259718866681309, + "grad_norm": 2.1967740058898926, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7218986749649048, + "num_tokens": 294468135.0, + "step": 11471 + }, + { + "epoch": 1.2598286843839226, + "grad_norm": 2.082259178161621, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7292082905769348, + "num_tokens": 294496863.0, + "step": 11472 + }, + { + "epoch": 1.2599385020865363, + "grad_norm": 2.087562084197998, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7122087478637695, + "num_tokens": 294523617.0, + "step": 11473 + }, + { + "epoch": 1.26004831978915, + "grad_norm": 2.0070714950561523, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7066221237182617, + "num_tokens": 294553495.0, + "step": 11474 + }, + { + "epoch": 1.2601581374917636, + "grad_norm": 2.239933729171753, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7282301187515259, + "num_tokens": 294579702.0, + "step": 11475 + }, + { + "epoch": 1.2602679551943774, + "grad_norm": 2.1294658184051514, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.697224497795105, + "num_tokens": 294610376.0, + "step": 11476 + }, + { + "epoch": 1.260377772896991, + "grad_norm": 2.4351511001586914, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7420698404312134, + "num_tokens": 294632515.0, + "step": 11477 + }, + { + "epoch": 1.2604875905996047, + "grad_norm": 2.0836384296417236, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7250658273696899, + "num_tokens": 294661216.0, + "step": 11478 + }, + { + "epoch": 1.2605974083022184, + "grad_norm": 2.4775187969207764, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7127124071121216, + "num_tokens": 294684109.0, + "step": 11479 + }, + { + "epoch": 1.260707226004832, + "grad_norm": 2.338665723800659, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.722245454788208, + "num_tokens": 294707588.0, + "step": 11480 + }, + { + "epoch": 1.2608170437074455, + "grad_norm": 2.6501705646514893, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7143982648849487, + "num_tokens": 294728671.0, + "step": 11481 + }, + { + "epoch": 1.2609268614100593, + "grad_norm": 2.159809112548828, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7159419059753418, + "num_tokens": 294756213.0, + "step": 11482 + }, + { + "epoch": 1.261036679112673, + "grad_norm": 2.23691987991333, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7016525268554688, + "num_tokens": 294783023.0, + "step": 11483 + }, + { + "epoch": 1.2611464968152866, + "grad_norm": 2.1300251483917236, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7143803834915161, + "num_tokens": 294811117.0, + "step": 11484 + }, + { + "epoch": 1.2612563145179003, + "grad_norm": 2.250159502029419, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7071313261985779, + "num_tokens": 294837516.0, + "step": 11485 + }, + { + "epoch": 1.2613661322205139, + "grad_norm": 2.4517822265625, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7289116978645325, + "num_tokens": 294860506.0, + "step": 11486 + }, + { + "epoch": 1.2614759499231276, + "grad_norm": 2.280764579772949, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7434435486793518, + "num_tokens": 294884977.0, + "step": 11487 + }, + { + "epoch": 1.2615857676257414, + "grad_norm": 1.9601620435714722, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.718376874923706, + "num_tokens": 294916953.0, + "step": 11488 + }, + { + "epoch": 1.261695585328355, + "grad_norm": 2.3432815074920654, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7488716840744019, + "num_tokens": 294936818.0, + "step": 11489 + }, + { + "epoch": 1.2618054030309687, + "grad_norm": 2.3469948768615723, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7163712978363037, + "num_tokens": 294960121.0, + "step": 11490 + }, + { + "epoch": 1.2619152207335822, + "grad_norm": 2.4704463481903076, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7322815656661987, + "num_tokens": 294980522.0, + "step": 11491 + }, + { + "epoch": 1.262025038436196, + "grad_norm": 2.42409086227417, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7624698877334595, + "num_tokens": 295001282.0, + "step": 11492 + }, + { + "epoch": 1.2621348561388097, + "grad_norm": 2.2092862129211426, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7092658281326294, + "num_tokens": 295028005.0, + "step": 11493 + }, + { + "epoch": 1.2622446738414232, + "grad_norm": 2.248509407043457, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.70343416929245, + "num_tokens": 295056711.0, + "step": 11494 + }, + { + "epoch": 1.2623544915440368, + "grad_norm": 2.2634711265563965, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.71117103099823, + "num_tokens": 295084094.0, + "step": 11495 + }, + { + "epoch": 1.2624643092466505, + "grad_norm": 2.345494508743286, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7067238688468933, + "num_tokens": 295109342.0, + "step": 11496 + }, + { + "epoch": 1.2625741269492643, + "grad_norm": 2.2251715660095215, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6898282766342163, + "num_tokens": 295134916.0, + "step": 11497 + }, + { + "epoch": 1.2626839446518778, + "grad_norm": 2.5440444946289062, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7272189855575562, + "num_tokens": 295155630.0, + "step": 11498 + }, + { + "epoch": 1.2627937623544916, + "grad_norm": 2.346193790435791, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7185496091842651, + "num_tokens": 295178366.0, + "step": 11499 + }, + { + "epoch": 1.2629035800571051, + "grad_norm": 2.3585262298583984, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7095075845718384, + "num_tokens": 295201880.0, + "step": 11500 + }, + { + "epoch": 1.2630133977597189, + "grad_norm": 2.349635124206543, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7272742986679077, + "num_tokens": 295226567.0, + "step": 11501 + }, + { + "epoch": 1.2631232154623326, + "grad_norm": 2.3376755714416504, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7299885749816895, + "num_tokens": 295251982.0, + "step": 11502 + }, + { + "epoch": 1.2632330331649462, + "grad_norm": 2.507110357284546, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7189034223556519, + "num_tokens": 295274237.0, + "step": 11503 + }, + { + "epoch": 1.26334285086756, + "grad_norm": 2.0097525119781494, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.737144947052002, + "num_tokens": 295303606.0, + "step": 11504 + }, + { + "epoch": 1.2634526685701735, + "grad_norm": 2.2673280239105225, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.716056764125824, + "num_tokens": 295328944.0, + "step": 11505 + }, + { + "epoch": 1.2635624862727872, + "grad_norm": 2.4371423721313477, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7570181488990784, + "num_tokens": 295351139.0, + "step": 11506 + }, + { + "epoch": 1.2636723039754008, + "grad_norm": 2.4477362632751465, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7321454286575317, + "num_tokens": 295372485.0, + "step": 11507 + }, + { + "epoch": 1.2637821216780145, + "grad_norm": 2.170886516571045, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7117525935173035, + "num_tokens": 295399791.0, + "step": 11508 + }, + { + "epoch": 1.263891939380628, + "grad_norm": 2.3873507976531982, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7137414216995239, + "num_tokens": 295422204.0, + "step": 11509 + }, + { + "epoch": 1.2640017570832418, + "grad_norm": 2.2133395671844482, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7055292129516602, + "num_tokens": 295450348.0, + "step": 11510 + }, + { + "epoch": 1.2641115747858556, + "grad_norm": 2.726266860961914, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7425028085708618, + "num_tokens": 295469200.0, + "step": 11511 + }, + { + "epoch": 1.264221392488469, + "grad_norm": 2.706679105758667, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7377558946609497, + "num_tokens": 295486190.0, + "step": 11512 + }, + { + "epoch": 1.2643312101910829, + "grad_norm": 1.9331574440002441, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6912097930908203, + "num_tokens": 295521113.0, + "step": 11513 + }, + { + "epoch": 1.2644410278936964, + "grad_norm": 2.363992929458618, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7232336401939392, + "num_tokens": 295545819.0, + "step": 11514 + }, + { + "epoch": 1.2645508455963101, + "grad_norm": 2.887183666229248, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7304964065551758, + "num_tokens": 295563392.0, + "step": 11515 + }, + { + "epoch": 1.264660663298924, + "grad_norm": 2.4176599979400635, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7078366279602051, + "num_tokens": 295587682.0, + "step": 11516 + }, + { + "epoch": 1.2647704810015374, + "grad_norm": 2.072216749191284, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.732985258102417, + "num_tokens": 295614677.0, + "step": 11517 + }, + { + "epoch": 1.264880298704151, + "grad_norm": 2.387199640274048, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7012073993682861, + "num_tokens": 295637981.0, + "step": 11518 + }, + { + "epoch": 1.2649901164067647, + "grad_norm": 2.3210926055908203, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7213578224182129, + "num_tokens": 295662787.0, + "step": 11519 + }, + { + "epoch": 1.2650999341093785, + "grad_norm": 2.3267979621887207, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6955872178077698, + "num_tokens": 295688510.0, + "step": 11520 + }, + { + "epoch": 1.265209751811992, + "grad_norm": 2.4533255100250244, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7343515157699585, + "num_tokens": 295710680.0, + "step": 11521 + }, + { + "epoch": 1.2653195695146058, + "grad_norm": 2.204195022583008, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7083297967910767, + "num_tokens": 295736361.0, + "step": 11522 + }, + { + "epoch": 1.2654293872172193, + "grad_norm": 1.9676848649978638, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7122745513916016, + "num_tokens": 295768869.0, + "step": 11523 + }, + { + "epoch": 1.265539204919833, + "grad_norm": 2.418733596801758, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7362247109413147, + "num_tokens": 295788816.0, + "step": 11524 + }, + { + "epoch": 1.2656490226224468, + "grad_norm": 2.2157318592071533, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7110192775726318, + "num_tokens": 295816065.0, + "step": 11525 + }, + { + "epoch": 1.2657588403250604, + "grad_norm": 2.274237871170044, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7383570671081543, + "num_tokens": 295839200.0, + "step": 11526 + }, + { + "epoch": 1.2658686580276741, + "grad_norm": 2.2042603492736816, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6922399997711182, + "num_tokens": 295865660.0, + "step": 11527 + }, + { + "epoch": 1.2659784757302877, + "grad_norm": 2.4660449028015137, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7389132976531982, + "num_tokens": 295886797.0, + "step": 11528 + }, + { + "epoch": 1.2660882934329014, + "grad_norm": 2.6139068603515625, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.751491904258728, + "num_tokens": 295905168.0, + "step": 11529 + }, + { + "epoch": 1.2661981111355152, + "grad_norm": 2.4939041137695312, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7400314807891846, + "num_tokens": 295928490.0, + "step": 11530 + }, + { + "epoch": 1.2663079288381287, + "grad_norm": 2.106787919998169, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7132338285446167, + "num_tokens": 295956713.0, + "step": 11531 + }, + { + "epoch": 1.2664177465407422, + "grad_norm": 2.050006151199341, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7117857336997986, + "num_tokens": 295986951.0, + "step": 11532 + }, + { + "epoch": 1.266527564243356, + "grad_norm": 2.2614588737487793, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7256787419319153, + "num_tokens": 296013074.0, + "step": 11533 + }, + { + "epoch": 1.2666373819459698, + "grad_norm": 2.20717716217041, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.733171820640564, + "num_tokens": 296038373.0, + "step": 11534 + }, + { + "epoch": 1.2667471996485833, + "grad_norm": 2.1466877460479736, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7117277383804321, + "num_tokens": 296066051.0, + "step": 11535 + }, + { + "epoch": 1.266857017351197, + "grad_norm": 2.105006456375122, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7377558946609497, + "num_tokens": 296097038.0, + "step": 11536 + }, + { + "epoch": 1.2669668350538106, + "grad_norm": 2.093827724456787, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.700913667678833, + "num_tokens": 296127080.0, + "step": 11537 + }, + { + "epoch": 1.2670766527564243, + "grad_norm": 2.005836009979248, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7172855138778687, + "num_tokens": 296159028.0, + "step": 11538 + }, + { + "epoch": 1.267186470459038, + "grad_norm": 2.2396328449249268, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7142868638038635, + "num_tokens": 296183828.0, + "step": 11539 + }, + { + "epoch": 1.2672962881616516, + "grad_norm": 2.288769483566284, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7304669618606567, + "num_tokens": 296206900.0, + "step": 11540 + }, + { + "epoch": 1.2674061058642654, + "grad_norm": 2.338529109954834, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7240238189697266, + "num_tokens": 296232520.0, + "step": 11541 + }, + { + "epoch": 1.267515923566879, + "grad_norm": 2.4764673709869385, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7367950081825256, + "num_tokens": 296254128.0, + "step": 11542 + }, + { + "epoch": 1.2676257412694927, + "grad_norm": 2.169069528579712, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.725904643535614, + "num_tokens": 296280493.0, + "step": 11543 + }, + { + "epoch": 1.2677355589721064, + "grad_norm": 2.2434895038604736, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7232657074928284, + "num_tokens": 296305956.0, + "step": 11544 + }, + { + "epoch": 1.26784537667472, + "grad_norm": 2.132307529449463, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7220546007156372, + "num_tokens": 296331997.0, + "step": 11545 + }, + { + "epoch": 1.2679551943773335, + "grad_norm": 2.079848527908325, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.712350606918335, + "num_tokens": 296361471.0, + "step": 11546 + }, + { + "epoch": 1.2680650120799473, + "grad_norm": 2.146270751953125, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7123949527740479, + "num_tokens": 296389705.0, + "step": 11547 + }, + { + "epoch": 1.268174829782561, + "grad_norm": 2.0617012977600098, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.70552659034729, + "num_tokens": 296423396.0, + "step": 11548 + }, + { + "epoch": 1.2682846474851746, + "grad_norm": 2.234884738922119, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7211962342262268, + "num_tokens": 296448420.0, + "step": 11549 + }, + { + "epoch": 1.2683944651877883, + "grad_norm": 2.3132832050323486, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.705484926700592, + "num_tokens": 296474927.0, + "step": 11550 + }, + { + "epoch": 1.2685042828904018, + "grad_norm": 2.1845896244049072, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7317662239074707, + "num_tokens": 296500767.0, + "step": 11551 + }, + { + "epoch": 1.2686141005930156, + "grad_norm": 2.3970727920532227, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7574445009231567, + "num_tokens": 296522093.0, + "step": 11552 + }, + { + "epoch": 1.2687239182956294, + "grad_norm": 2.3496103286743164, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.738088846206665, + "num_tokens": 296544116.0, + "step": 11553 + }, + { + "epoch": 1.268833735998243, + "grad_norm": 2.2281031608581543, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.728442370891571, + "num_tokens": 296570228.0, + "step": 11554 + }, + { + "epoch": 1.2689435537008567, + "grad_norm": 2.502434253692627, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7127221822738647, + "num_tokens": 296592574.0, + "step": 11555 + }, + { + "epoch": 1.2690533714034702, + "grad_norm": 2.095715045928955, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7164612412452698, + "num_tokens": 296619748.0, + "step": 11556 + }, + { + "epoch": 1.269163189106084, + "grad_norm": 2.265286922454834, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7087613344192505, + "num_tokens": 296645705.0, + "step": 11557 + }, + { + "epoch": 1.2692730068086975, + "grad_norm": 2.455111503601074, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.730251669883728, + "num_tokens": 296666858.0, + "step": 11558 + }, + { + "epoch": 1.2693828245113112, + "grad_norm": 2.354210138320923, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7162990570068359, + "num_tokens": 296693406.0, + "step": 11559 + }, + { + "epoch": 1.2694926422139248, + "grad_norm": 2.2205140590667725, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7388510704040527, + "num_tokens": 296719365.0, + "step": 11560 + }, + { + "epoch": 1.2696024599165385, + "grad_norm": 2.4461939334869385, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7067006826400757, + "num_tokens": 296741176.0, + "step": 11561 + }, + { + "epoch": 1.2697122776191523, + "grad_norm": 2.264110565185547, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7149258852005005, + "num_tokens": 296766302.0, + "step": 11562 + }, + { + "epoch": 1.2698220953217658, + "grad_norm": 2.1797378063201904, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7310179471969604, + "num_tokens": 296794246.0, + "step": 11563 + }, + { + "epoch": 1.2699319130243796, + "grad_norm": 2.1406362056732178, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7421725988388062, + "num_tokens": 296820780.0, + "step": 11564 + }, + { + "epoch": 1.2700417307269931, + "grad_norm": 2.3057069778442383, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7296959757804871, + "num_tokens": 296843523.0, + "step": 11565 + }, + { + "epoch": 1.2701515484296069, + "grad_norm": 2.122847080230713, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.713402509689331, + "num_tokens": 296873182.0, + "step": 11566 + }, + { + "epoch": 1.2702613661322206, + "grad_norm": 2.4951233863830566, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.721697986125946, + "num_tokens": 296896359.0, + "step": 11567 + }, + { + "epoch": 1.2703711838348342, + "grad_norm": 2.5406711101531982, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7384142875671387, + "num_tokens": 296916153.0, + "step": 11568 + }, + { + "epoch": 1.270481001537448, + "grad_norm": 2.3747730255126953, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7176721692085266, + "num_tokens": 296938498.0, + "step": 11569 + }, + { + "epoch": 1.2705908192400615, + "grad_norm": 2.1090149879455566, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7065081596374512, + "num_tokens": 296966397.0, + "step": 11570 + }, + { + "epoch": 1.2707006369426752, + "grad_norm": 2.167818307876587, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7315928936004639, + "num_tokens": 296992258.0, + "step": 11571 + }, + { + "epoch": 1.2708104546452887, + "grad_norm": 2.567704916000366, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.724198043346405, + "num_tokens": 297012642.0, + "step": 11572 + }, + { + "epoch": 1.2709202723479025, + "grad_norm": 2.265815019607544, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7369751930236816, + "num_tokens": 297037812.0, + "step": 11573 + }, + { + "epoch": 1.271030090050516, + "grad_norm": 2.329136848449707, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7236064672470093, + "num_tokens": 297061273.0, + "step": 11574 + }, + { + "epoch": 1.2711399077531298, + "grad_norm": 2.415074586868286, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7144243717193604, + "num_tokens": 297083605.0, + "step": 11575 + }, + { + "epoch": 1.2712497254557436, + "grad_norm": 1.9780133962631226, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7000420689582825, + "num_tokens": 297113888.0, + "step": 11576 + }, + { + "epoch": 1.271359543158357, + "grad_norm": 2.0556843280792236, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7135913968086243, + "num_tokens": 297144212.0, + "step": 11577 + }, + { + "epoch": 1.2714693608609708, + "grad_norm": 2.1211740970611572, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7419939637184143, + "num_tokens": 297171045.0, + "step": 11578 + }, + { + "epoch": 1.2715791785635844, + "grad_norm": 2.404503345489502, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7206293344497681, + "num_tokens": 297194429.0, + "step": 11579 + }, + { + "epoch": 1.2716889962661981, + "grad_norm": 2.3050270080566406, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7382253408432007, + "num_tokens": 297219241.0, + "step": 11580 + }, + { + "epoch": 1.271798813968812, + "grad_norm": 2.192462205886841, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7415392994880676, + "num_tokens": 297243981.0, + "step": 11581 + }, + { + "epoch": 1.2719086316714254, + "grad_norm": 2.0612263679504395, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7178411483764648, + "num_tokens": 297273686.0, + "step": 11582 + }, + { + "epoch": 1.272018449374039, + "grad_norm": 2.1686971187591553, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7249113321304321, + "num_tokens": 297301789.0, + "step": 11583 + }, + { + "epoch": 1.2721282670766527, + "grad_norm": 2.1987240314483643, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7355412840843201, + "num_tokens": 297328320.0, + "step": 11584 + }, + { + "epoch": 1.2722380847792665, + "grad_norm": 1.9825143814086914, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7189575433731079, + "num_tokens": 297357273.0, + "step": 11585 + }, + { + "epoch": 1.27234790248188, + "grad_norm": 2.0258517265319824, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7164316177368164, + "num_tokens": 297388301.0, + "step": 11586 + }, + { + "epoch": 1.2724577201844938, + "grad_norm": 2.1496474742889404, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7155221104621887, + "num_tokens": 297416410.0, + "step": 11587 + }, + { + "epoch": 1.2725675378871073, + "grad_norm": 2.171675205230713, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6879354119300842, + "num_tokens": 297442763.0, + "step": 11588 + }, + { + "epoch": 1.272677355589721, + "grad_norm": 2.2878620624542236, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7150013446807861, + "num_tokens": 297468190.0, + "step": 11589 + }, + { + "epoch": 1.2727871732923348, + "grad_norm": 2.1022613048553467, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7361934185028076, + "num_tokens": 297495340.0, + "step": 11590 + }, + { + "epoch": 1.2728969909949484, + "grad_norm": 2.2432761192321777, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.708172082901001, + "num_tokens": 297523781.0, + "step": 11591 + }, + { + "epoch": 1.273006808697562, + "grad_norm": 2.1273064613342285, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7171132564544678, + "num_tokens": 297552062.0, + "step": 11592 + }, + { + "epoch": 1.2731166264001756, + "grad_norm": 2.4064903259277344, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7153662443161011, + "num_tokens": 297575638.0, + "step": 11593 + }, + { + "epoch": 1.2732264441027894, + "grad_norm": 2.5813944339752197, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7258247137069702, + "num_tokens": 297596631.0, + "step": 11594 + }, + { + "epoch": 1.2733362618054032, + "grad_norm": 2.3086605072021484, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7135356664657593, + "num_tokens": 297620071.0, + "step": 11595 + }, + { + "epoch": 1.2734460795080167, + "grad_norm": 2.4276630878448486, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7124340534210205, + "num_tokens": 297643443.0, + "step": 11596 + }, + { + "epoch": 1.2735558972106302, + "grad_norm": 2.4529316425323486, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.71293044090271, + "num_tokens": 297665757.0, + "step": 11597 + }, + { + "epoch": 1.273665714913244, + "grad_norm": 2.0475246906280518, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7005348205566406, + "num_tokens": 297696380.0, + "step": 11598 + }, + { + "epoch": 1.2737755326158577, + "grad_norm": 2.9000232219696045, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7238298654556274, + "num_tokens": 297712268.0, + "step": 11599 + }, + { + "epoch": 1.2738853503184713, + "grad_norm": 2.397780179977417, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.712826669216156, + "num_tokens": 297735250.0, + "step": 11600 + }, + { + "epoch": 1.273995168021085, + "grad_norm": 2.354525089263916, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7238237857818604, + "num_tokens": 297758822.0, + "step": 11601 + }, + { + "epoch": 1.2741049857236986, + "grad_norm": 2.6011111736297607, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7368751168251038, + "num_tokens": 297777580.0, + "step": 11602 + }, + { + "epoch": 1.2742148034263123, + "grad_norm": 2.300377368927002, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7326046824455261, + "num_tokens": 297802700.0, + "step": 11603 + }, + { + "epoch": 1.274324621128926, + "grad_norm": 2.454697370529175, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7126481533050537, + "num_tokens": 297825465.0, + "step": 11604 + }, + { + "epoch": 1.2744344388315396, + "grad_norm": 2.1487529277801514, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7131943702697754, + "num_tokens": 297853720.0, + "step": 11605 + }, + { + "epoch": 1.2745442565341534, + "grad_norm": 2.742215394973755, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7225486636161804, + "num_tokens": 297872426.0, + "step": 11606 + }, + { + "epoch": 1.274654074236767, + "grad_norm": 2.2872114181518555, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7268948554992676, + "num_tokens": 297897794.0, + "step": 11607 + }, + { + "epoch": 1.2747638919393807, + "grad_norm": 2.237701892852783, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.705721914768219, + "num_tokens": 297924325.0, + "step": 11608 + }, + { + "epoch": 1.2748737096419944, + "grad_norm": 2.1890804767608643, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7156193256378174, + "num_tokens": 297950364.0, + "step": 11609 + }, + { + "epoch": 1.274983527344608, + "grad_norm": 2.1425750255584717, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7101597785949707, + "num_tokens": 297979834.0, + "step": 11610 + }, + { + "epoch": 1.2750933450472215, + "grad_norm": 2.261052131652832, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7287459373474121, + "num_tokens": 298004654.0, + "step": 11611 + }, + { + "epoch": 1.2752031627498353, + "grad_norm": 1.9518667459487915, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7163479328155518, + "num_tokens": 298035742.0, + "step": 11612 + }, + { + "epoch": 1.275312980452449, + "grad_norm": 2.165738821029663, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6953690648078918, + "num_tokens": 298061530.0, + "step": 11613 + }, + { + "epoch": 1.2754227981550625, + "grad_norm": 2.3441812992095947, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7044171094894409, + "num_tokens": 298085544.0, + "step": 11614 + }, + { + "epoch": 1.2755326158576763, + "grad_norm": 2.1818554401397705, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7188507914543152, + "num_tokens": 298112725.0, + "step": 11615 + }, + { + "epoch": 1.2756424335602898, + "grad_norm": 2.190372943878174, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.7085775136947632, + "num_tokens": 298140647.0, + "step": 11616 + }, + { + "epoch": 1.2757522512629036, + "grad_norm": 2.2649712562561035, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7263373732566833, + "num_tokens": 298165025.0, + "step": 11617 + }, + { + "epoch": 1.2758620689655173, + "grad_norm": 2.3333375453948975, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7108347415924072, + "num_tokens": 298187688.0, + "step": 11618 + }, + { + "epoch": 1.2759718866681309, + "grad_norm": 2.283970594406128, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6947647333145142, + "num_tokens": 298213996.0, + "step": 11619 + }, + { + "epoch": 1.2760817043707446, + "grad_norm": 2.7404587268829346, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7530723810195923, + "num_tokens": 298231381.0, + "step": 11620 + }, + { + "epoch": 1.2761915220733582, + "grad_norm": 2.1817739009857178, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7553583383560181, + "num_tokens": 298256925.0, + "step": 11621 + }, + { + "epoch": 1.276301339775972, + "grad_norm": 2.3962762355804443, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.733873724937439, + "num_tokens": 298280122.0, + "step": 11622 + }, + { + "epoch": 1.2764111574785855, + "grad_norm": 2.197237968444824, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7158442735671997, + "num_tokens": 298307354.0, + "step": 11623 + }, + { + "epoch": 1.2765209751811992, + "grad_norm": 2.293260335922241, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7237361073493958, + "num_tokens": 298332860.0, + "step": 11624 + }, + { + "epoch": 1.2766307928838128, + "grad_norm": 2.5681631565093994, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7320733666419983, + "num_tokens": 298353644.0, + "step": 11625 + }, + { + "epoch": 1.2767406105864265, + "grad_norm": 2.175809860229492, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7107805013656616, + "num_tokens": 298382411.0, + "step": 11626 + }, + { + "epoch": 1.2768504282890403, + "grad_norm": 2.318103313446045, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7013554573059082, + "num_tokens": 298405230.0, + "step": 11627 + }, + { + "epoch": 1.2769602459916538, + "grad_norm": 2.1427462100982666, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6920067667961121, + "num_tokens": 298432146.0, + "step": 11628 + }, + { + "epoch": 1.2770700636942676, + "grad_norm": 2.2768452167510986, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7190121412277222, + "num_tokens": 298457942.0, + "step": 11629 + }, + { + "epoch": 1.277179881396881, + "grad_norm": 2.3069534301757812, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7164613604545593, + "num_tokens": 298483135.0, + "step": 11630 + }, + { + "epoch": 1.2772896990994949, + "grad_norm": 2.192507266998291, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7284536361694336, + "num_tokens": 298508585.0, + "step": 11631 + }, + { + "epoch": 1.2773995168021086, + "grad_norm": 2.694972276687622, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7365448474884033, + "num_tokens": 298527842.0, + "step": 11632 + }, + { + "epoch": 1.2775093345047221, + "grad_norm": 2.2392146587371826, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6856176853179932, + "num_tokens": 298555751.0, + "step": 11633 + }, + { + "epoch": 1.2776191522073357, + "grad_norm": 2.858833074569702, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7433674335479736, + "num_tokens": 298572181.0, + "step": 11634 + }, + { + "epoch": 1.2777289699099494, + "grad_norm": 2.1549229621887207, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7227070927619934, + "num_tokens": 298597365.0, + "step": 11635 + }, + { + "epoch": 1.2778387876125632, + "grad_norm": 2.0619213581085205, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7253950834274292, + "num_tokens": 298626150.0, + "step": 11636 + }, + { + "epoch": 1.2779486053151767, + "grad_norm": 2.262763738632202, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7232299447059631, + "num_tokens": 298649689.0, + "step": 11637 + }, + { + "epoch": 1.2780584230177905, + "grad_norm": 2.1678452491760254, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7009211778640747, + "num_tokens": 298678438.0, + "step": 11638 + }, + { + "epoch": 1.278168240720404, + "grad_norm": 2.1970512866973877, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7199606895446777, + "num_tokens": 298704418.0, + "step": 11639 + }, + { + "epoch": 1.2782780584230178, + "grad_norm": 2.084291696548462, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7152284383773804, + "num_tokens": 298730438.0, + "step": 11640 + }, + { + "epoch": 1.2783878761256315, + "grad_norm": 1.96700119972229, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7227663993835449, + "num_tokens": 298761055.0, + "step": 11641 + }, + { + "epoch": 1.278497693828245, + "grad_norm": 2.160625696182251, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7102570533752441, + "num_tokens": 298788832.0, + "step": 11642 + }, + { + "epoch": 1.2786075115308588, + "grad_norm": 2.212761878967285, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7196211814880371, + "num_tokens": 298814348.0, + "step": 11643 + }, + { + "epoch": 1.2787173292334724, + "grad_norm": 2.2978081703186035, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7052079439163208, + "num_tokens": 298839655.0, + "step": 11644 + }, + { + "epoch": 1.2788271469360861, + "grad_norm": 2.0820460319519043, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7369718551635742, + "num_tokens": 298866223.0, + "step": 11645 + }, + { + "epoch": 1.2789369646386999, + "grad_norm": 2.606053590774536, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7333102226257324, + "num_tokens": 298886156.0, + "step": 11646 + }, + { + "epoch": 1.2790467823413134, + "grad_norm": 2.147804021835327, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7268936634063721, + "num_tokens": 298912476.0, + "step": 11647 + }, + { + "epoch": 1.279156600043927, + "grad_norm": 2.16302227973938, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7226455211639404, + "num_tokens": 298939269.0, + "step": 11648 + }, + { + "epoch": 1.2792664177465407, + "grad_norm": 2.576843500137329, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7220591306686401, + "num_tokens": 298958599.0, + "step": 11649 + }, + { + "epoch": 1.2793762354491545, + "grad_norm": 2.3550069332122803, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7179574966430664, + "num_tokens": 298983179.0, + "step": 11650 + }, + { + "epoch": 1.279486053151768, + "grad_norm": 2.3018100261688232, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7242224812507629, + "num_tokens": 299006625.0, + "step": 11651 + }, + { + "epoch": 1.2795958708543818, + "grad_norm": 3.2203316688537598, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7302473187446594, + "num_tokens": 299033170.0, + "step": 11652 + }, + { + "epoch": 1.2797056885569953, + "grad_norm": 2.5477211475372314, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7156014442443848, + "num_tokens": 299056057.0, + "step": 11653 + }, + { + "epoch": 1.279815506259609, + "grad_norm": 2.2310707569122314, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7215414643287659, + "num_tokens": 299079898.0, + "step": 11654 + }, + { + "epoch": 1.2799253239622228, + "grad_norm": 2.198627471923828, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6959840655326843, + "num_tokens": 299107203.0, + "step": 11655 + }, + { + "epoch": 1.2800351416648363, + "grad_norm": 2.0592284202575684, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7017827033996582, + "num_tokens": 299135776.0, + "step": 11656 + }, + { + "epoch": 1.28014495936745, + "grad_norm": 2.4638378620147705, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7427173256874084, + "num_tokens": 299156293.0, + "step": 11657 + }, + { + "epoch": 1.2802547770700636, + "grad_norm": 2.1379499435424805, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7092873454093933, + "num_tokens": 299184633.0, + "step": 11658 + }, + { + "epoch": 1.2803645947726774, + "grad_norm": 2.1342101097106934, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7181463241577148, + "num_tokens": 299213226.0, + "step": 11659 + }, + { + "epoch": 1.2804744124752911, + "grad_norm": 2.4712958335876465, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7227940559387207, + "num_tokens": 299233488.0, + "step": 11660 + }, + { + "epoch": 1.2805842301779047, + "grad_norm": 2.4844307899475098, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7220067381858826, + "num_tokens": 299256442.0, + "step": 11661 + }, + { + "epoch": 1.2806940478805182, + "grad_norm": 2.471682548522949, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7398661375045776, + "num_tokens": 299276864.0, + "step": 11662 + }, + { + "epoch": 1.280803865583132, + "grad_norm": 2.199176788330078, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7009564638137817, + "num_tokens": 299303377.0, + "step": 11663 + }, + { + "epoch": 1.2809136832857457, + "grad_norm": 2.137721300125122, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6974906325340271, + "num_tokens": 299331957.0, + "step": 11664 + }, + { + "epoch": 1.2810235009883593, + "grad_norm": 2.3587543964385986, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6913250088691711, + "num_tokens": 299357379.0, + "step": 11665 + }, + { + "epoch": 1.281133318690973, + "grad_norm": 2.0212085247039795, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7262154221534729, + "num_tokens": 299387633.0, + "step": 11666 + }, + { + "epoch": 1.2812431363935866, + "grad_norm": 2.170490264892578, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7208499908447266, + "num_tokens": 299414935.0, + "step": 11667 + }, + { + "epoch": 1.2813529540962003, + "grad_norm": 2.081791877746582, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7038213610649109, + "num_tokens": 299445324.0, + "step": 11668 + }, + { + "epoch": 1.281462771798814, + "grad_norm": 2.2246334552764893, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7225973606109619, + "num_tokens": 299471311.0, + "step": 11669 + }, + { + "epoch": 1.2815725895014276, + "grad_norm": 2.4867382049560547, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7338767647743225, + "num_tokens": 299494149.0, + "step": 11670 + }, + { + "epoch": 1.2816824072040414, + "grad_norm": 2.288987398147583, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7247768640518188, + "num_tokens": 299519731.0, + "step": 11671 + }, + { + "epoch": 1.281792224906655, + "grad_norm": 2.1153616905212402, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7211230993270874, + "num_tokens": 299546539.0, + "step": 11672 + }, + { + "epoch": 1.2819020426092687, + "grad_norm": 2.6423797607421875, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7332063913345337, + "num_tokens": 299565556.0, + "step": 11673 + }, + { + "epoch": 1.2820118603118824, + "grad_norm": 2.6869139671325684, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7435312271118164, + "num_tokens": 299583760.0, + "step": 11674 + }, + { + "epoch": 1.282121678014496, + "grad_norm": 2.371006965637207, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7416766881942749, + "num_tokens": 299605835.0, + "step": 11675 + }, + { + "epoch": 1.2822314957171095, + "grad_norm": 2.113448143005371, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.769432008266449, + "num_tokens": 299630268.0, + "step": 11676 + }, + { + "epoch": 1.2823413134197232, + "grad_norm": 2.322894334793091, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7437459230422974, + "num_tokens": 299653705.0, + "step": 11677 + }, + { + "epoch": 1.282451131122337, + "grad_norm": 2.19805908203125, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7077939510345459, + "num_tokens": 299684646.0, + "step": 11678 + }, + { + "epoch": 1.2825609488249505, + "grad_norm": 2.3680636882781982, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.719597578048706, + "num_tokens": 299709407.0, + "step": 11679 + }, + { + "epoch": 1.2826707665275643, + "grad_norm": 2.131957769393921, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7183261513710022, + "num_tokens": 299737342.0, + "step": 11680 + }, + { + "epoch": 1.2827805842301778, + "grad_norm": 2.5239531993865967, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7358524799346924, + "num_tokens": 299758839.0, + "step": 11681 + }, + { + "epoch": 1.2828904019327916, + "grad_norm": 2.14617919921875, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7199012041091919, + "num_tokens": 299787219.0, + "step": 11682 + }, + { + "epoch": 1.2830002196354053, + "grad_norm": 2.0477097034454346, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7049102187156677, + "num_tokens": 299816195.0, + "step": 11683 + }, + { + "epoch": 1.2831100373380189, + "grad_norm": 2.439194679260254, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7273777723312378, + "num_tokens": 299839128.0, + "step": 11684 + }, + { + "epoch": 1.2832198550406326, + "grad_norm": 2.4655637741088867, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7462970018386841, + "num_tokens": 299859970.0, + "step": 11685 + }, + { + "epoch": 1.2833296727432462, + "grad_norm": 2.3795323371887207, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7197596430778503, + "num_tokens": 299885579.0, + "step": 11686 + }, + { + "epoch": 1.28343949044586, + "grad_norm": 2.3665077686309814, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7104106545448303, + "num_tokens": 299909129.0, + "step": 11687 + }, + { + "epoch": 1.2835493081484735, + "grad_norm": 2.4380075931549072, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7157705426216125, + "num_tokens": 299932313.0, + "step": 11688 + }, + { + "epoch": 1.2836591258510872, + "grad_norm": 2.2017107009887695, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7088589668273926, + "num_tokens": 299958198.0, + "step": 11689 + }, + { + "epoch": 1.2837689435537007, + "grad_norm": 2.3330750465393066, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7123167514801025, + "num_tokens": 299980718.0, + "step": 11690 + }, + { + "epoch": 1.2838787612563145, + "grad_norm": 2.1145615577697754, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7439576387405396, + "num_tokens": 300008489.0, + "step": 11691 + }, + { + "epoch": 1.2839885789589283, + "grad_norm": 2.485297918319702, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7280402779579163, + "num_tokens": 300030328.0, + "step": 11692 + }, + { + "epoch": 1.2840983966615418, + "grad_norm": 1.9970524311065674, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7301136255264282, + "num_tokens": 300058612.0, + "step": 11693 + }, + { + "epoch": 1.2842082143641556, + "grad_norm": 2.2831778526306152, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7116692066192627, + "num_tokens": 300083630.0, + "step": 11694 + }, + { + "epoch": 1.284318032066769, + "grad_norm": 2.6782267093658447, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7388760447502136, + "num_tokens": 300102297.0, + "step": 11695 + }, + { + "epoch": 1.2844278497693828, + "grad_norm": 2.3275201320648193, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7293752431869507, + "num_tokens": 300127572.0, + "step": 11696 + }, + { + "epoch": 1.2845376674719966, + "grad_norm": 2.1827964782714844, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7134985327720642, + "num_tokens": 300155195.0, + "step": 11697 + }, + { + "epoch": 1.2846474851746101, + "grad_norm": 2.3421082496643066, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7168832421302795, + "num_tokens": 300180260.0, + "step": 11698 + }, + { + "epoch": 1.2847573028772237, + "grad_norm": 2.230325937271118, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7089555263519287, + "num_tokens": 300205464.0, + "step": 11699 + }, + { + "epoch": 1.2848671205798374, + "grad_norm": 2.1825339794158936, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7038958072662354, + "num_tokens": 300236478.0, + "step": 11700 + }, + { + "epoch": 1.2849769382824512, + "grad_norm": 2.414409637451172, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7260376214981079, + "num_tokens": 300260652.0, + "step": 11701 + }, + { + "epoch": 1.2850867559850647, + "grad_norm": 2.186265468597412, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7116947174072266, + "num_tokens": 300288709.0, + "step": 11702 + }, + { + "epoch": 1.2851965736876785, + "grad_norm": 2.0669071674346924, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.729195237159729, + "num_tokens": 300319006.0, + "step": 11703 + }, + { + "epoch": 1.285306391390292, + "grad_norm": 2.0545616149902344, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7027128338813782, + "num_tokens": 300347744.0, + "step": 11704 + }, + { + "epoch": 1.2854162090929058, + "grad_norm": 2.2704029083251953, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7351980209350586, + "num_tokens": 300370112.0, + "step": 11705 + }, + { + "epoch": 1.2855260267955195, + "grad_norm": 1.8871361017227173, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7316333651542664, + "num_tokens": 300402909.0, + "step": 11706 + }, + { + "epoch": 1.285635844498133, + "grad_norm": 2.4546220302581787, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7347452640533447, + "num_tokens": 300424743.0, + "step": 11707 + }, + { + "epoch": 1.2857456622007468, + "grad_norm": 2.2224366664886475, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7353165745735168, + "num_tokens": 300450782.0, + "step": 11708 + }, + { + "epoch": 1.2858554799033604, + "grad_norm": 2.124403476715088, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7265839576721191, + "num_tokens": 300477766.0, + "step": 11709 + }, + { + "epoch": 1.2859652976059741, + "grad_norm": 1.8886297941207886, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7180272340774536, + "num_tokens": 300514662.0, + "step": 11710 + }, + { + "epoch": 1.2860751153085879, + "grad_norm": 2.32234787940979, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7199290990829468, + "num_tokens": 300540773.0, + "step": 11711 + }, + { + "epoch": 1.2861849330112014, + "grad_norm": 2.3532965183258057, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7163009643554688, + "num_tokens": 300565389.0, + "step": 11712 + }, + { + "epoch": 1.286294750713815, + "grad_norm": 2.356563091278076, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7146804332733154, + "num_tokens": 300589170.0, + "step": 11713 + }, + { + "epoch": 1.2864045684164287, + "grad_norm": 2.2390966415405273, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7300690412521362, + "num_tokens": 300614323.0, + "step": 11714 + }, + { + "epoch": 1.2865143861190425, + "grad_norm": 2.143925189971924, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7020569443702698, + "num_tokens": 300643156.0, + "step": 11715 + }, + { + "epoch": 1.286624203821656, + "grad_norm": 2.5149595737457275, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7463918328285217, + "num_tokens": 300664098.0, + "step": 11716 + }, + { + "epoch": 1.2867340215242697, + "grad_norm": 2.395632743835449, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7281810641288757, + "num_tokens": 300685447.0, + "step": 11717 + }, + { + "epoch": 1.2868438392268833, + "grad_norm": 2.259131908416748, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.702092170715332, + "num_tokens": 300710842.0, + "step": 11718 + }, + { + "epoch": 1.286953656929497, + "grad_norm": 2.0397684574127197, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7083526253700256, + "num_tokens": 300740278.0, + "step": 11719 + }, + { + "epoch": 1.2870634746321108, + "grad_norm": 2.3159773349761963, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7423363327980042, + "num_tokens": 300763753.0, + "step": 11720 + }, + { + "epoch": 1.2871732923347243, + "grad_norm": 2.083007574081421, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7185705900192261, + "num_tokens": 300793241.0, + "step": 11721 + }, + { + "epoch": 1.287283110037338, + "grad_norm": 2.3515100479125977, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7140870094299316, + "num_tokens": 300817423.0, + "step": 11722 + }, + { + "epoch": 1.2873929277399516, + "grad_norm": 2.1696314811706543, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7407591342926025, + "num_tokens": 300844322.0, + "step": 11723 + }, + { + "epoch": 1.2875027454425654, + "grad_norm": 2.357858657836914, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.711105465888977, + "num_tokens": 300867386.0, + "step": 11724 + }, + { + "epoch": 1.2876125631451791, + "grad_norm": 2.012664794921875, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7077794075012207, + "num_tokens": 300898881.0, + "step": 11725 + }, + { + "epoch": 1.2877223808477927, + "grad_norm": 2.2333757877349854, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7221421003341675, + "num_tokens": 300925698.0, + "step": 11726 + }, + { + "epoch": 1.2878321985504062, + "grad_norm": 2.2017431259155273, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7403703927993774, + "num_tokens": 300951380.0, + "step": 11727 + }, + { + "epoch": 1.28794201625302, + "grad_norm": 2.2158327102661133, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7004159092903137, + "num_tokens": 300979998.0, + "step": 11728 + }, + { + "epoch": 1.2880518339556337, + "grad_norm": 2.318305253982544, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7172485589981079, + "num_tokens": 301005723.0, + "step": 11729 + }, + { + "epoch": 1.2881616516582473, + "grad_norm": 2.410947322845459, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7146564722061157, + "num_tokens": 301029352.0, + "step": 11730 + }, + { + "epoch": 1.288271469360861, + "grad_norm": 2.3115713596343994, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7064403295516968, + "num_tokens": 301052776.0, + "step": 11731 + }, + { + "epoch": 1.2883812870634745, + "grad_norm": 2.131643772125244, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.708981990814209, + "num_tokens": 301081234.0, + "step": 11732 + }, + { + "epoch": 1.2884911047660883, + "grad_norm": 2.251526117324829, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6984509229660034, + "num_tokens": 301108055.0, + "step": 11733 + }, + { + "epoch": 1.288600922468702, + "grad_norm": 2.3474013805389404, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7207430601119995, + "num_tokens": 301132375.0, + "step": 11734 + }, + { + "epoch": 1.2887107401713156, + "grad_norm": 2.3596160411834717, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.734272301197052, + "num_tokens": 301156292.0, + "step": 11735 + }, + { + "epoch": 1.2888205578739294, + "grad_norm": 1.9240548610687256, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.694650411605835, + "num_tokens": 301192199.0, + "step": 11736 + }, + { + "epoch": 1.2889303755765429, + "grad_norm": 1.9454082250595093, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7375361919403076, + "num_tokens": 301222333.0, + "step": 11737 + }, + { + "epoch": 1.2890401932791566, + "grad_norm": 2.517503023147583, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7410975694656372, + "num_tokens": 301243095.0, + "step": 11738 + }, + { + "epoch": 1.2891500109817704, + "grad_norm": 2.1691880226135254, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7294576168060303, + "num_tokens": 301270701.0, + "step": 11739 + }, + { + "epoch": 1.289259828684384, + "grad_norm": 2.1729674339294434, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6988996267318726, + "num_tokens": 301299012.0, + "step": 11740 + }, + { + "epoch": 1.2893696463869975, + "grad_norm": 2.0555880069732666, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7247682809829712, + "num_tokens": 301330613.0, + "step": 11741 + }, + { + "epoch": 1.2894794640896112, + "grad_norm": 2.2281174659729004, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.703054666519165, + "num_tokens": 301357197.0, + "step": 11742 + }, + { + "epoch": 1.289589281792225, + "grad_norm": 2.026782751083374, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7177364826202393, + "num_tokens": 301387619.0, + "step": 11743 + }, + { + "epoch": 1.2896990994948385, + "grad_norm": 2.2248494625091553, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.745097279548645, + "num_tokens": 301413239.0, + "step": 11744 + }, + { + "epoch": 1.2898089171974523, + "grad_norm": 2.1904380321502686, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7321071624755859, + "num_tokens": 301439081.0, + "step": 11745 + }, + { + "epoch": 1.2899187349000658, + "grad_norm": 2.3583757877349854, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7084707021713257, + "num_tokens": 301464875.0, + "step": 11746 + }, + { + "epoch": 1.2900285526026796, + "grad_norm": 2.385481357574463, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7258532643318176, + "num_tokens": 301487882.0, + "step": 11747 + }, + { + "epoch": 1.2901383703052933, + "grad_norm": 2.110886812210083, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7125745415687561, + "num_tokens": 301519128.0, + "step": 11748 + }, + { + "epoch": 1.2902481880079069, + "grad_norm": 2.42232346534729, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7155282497406006, + "num_tokens": 301541385.0, + "step": 11749 + }, + { + "epoch": 1.2903580057105206, + "grad_norm": 2.169909715652466, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7359558939933777, + "num_tokens": 301567724.0, + "step": 11750 + }, + { + "epoch": 1.2904678234131342, + "grad_norm": 2.2374579906463623, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7074038982391357, + "num_tokens": 301595805.0, + "step": 11751 + }, + { + "epoch": 1.290577641115748, + "grad_norm": 2.1325314044952393, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.71818608045578, + "num_tokens": 301622329.0, + "step": 11752 + }, + { + "epoch": 1.2906874588183614, + "grad_norm": 2.3356242179870605, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7140059471130371, + "num_tokens": 301646010.0, + "step": 11753 + }, + { + "epoch": 1.2907972765209752, + "grad_norm": 2.035381317138672, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7020094394683838, + "num_tokens": 301675304.0, + "step": 11754 + }, + { + "epoch": 1.2909070942235887, + "grad_norm": 2.6692593097686768, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7356953620910645, + "num_tokens": 301694309.0, + "step": 11755 + }, + { + "epoch": 1.2910169119262025, + "grad_norm": 2.6022794246673584, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7320683002471924, + "num_tokens": 301715598.0, + "step": 11756 + }, + { + "epoch": 1.2911267296288162, + "grad_norm": 2.2763168811798096, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7134007811546326, + "num_tokens": 301742271.0, + "step": 11757 + }, + { + "epoch": 1.2912365473314298, + "grad_norm": 2.321803092956543, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6983000636100769, + "num_tokens": 301766228.0, + "step": 11758 + }, + { + "epoch": 1.2913463650340435, + "grad_norm": 2.4402499198913574, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7238144874572754, + "num_tokens": 301787370.0, + "step": 11759 + }, + { + "epoch": 1.291456182736657, + "grad_norm": 2.258424758911133, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7208725214004517, + "num_tokens": 301811933.0, + "step": 11760 + }, + { + "epoch": 1.2915660004392708, + "grad_norm": 2.238844871520996, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.744691014289856, + "num_tokens": 301836885.0, + "step": 11761 + }, + { + "epoch": 1.2916758181418846, + "grad_norm": 2.526695489883423, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7373366355895996, + "num_tokens": 301857085.0, + "step": 11762 + }, + { + "epoch": 1.2917856358444981, + "grad_norm": 2.2894835472106934, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7255625128746033, + "num_tokens": 301880637.0, + "step": 11763 + }, + { + "epoch": 1.2918954535471117, + "grad_norm": 2.1686863899230957, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7265832424163818, + "num_tokens": 301906942.0, + "step": 11764 + }, + { + "epoch": 1.2920052712497254, + "grad_norm": 2.150271415710449, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7072535157203674, + "num_tokens": 301935828.0, + "step": 11765 + }, + { + "epoch": 1.2921150889523392, + "grad_norm": 2.2951295375823975, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.695986270904541, + "num_tokens": 301962002.0, + "step": 11766 + }, + { + "epoch": 1.2922249066549527, + "grad_norm": 2.0251173973083496, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6967567205429077, + "num_tokens": 301994129.0, + "step": 11767 + }, + { + "epoch": 1.2923347243575665, + "grad_norm": 2.5086288452148438, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7217261791229248, + "num_tokens": 302017366.0, + "step": 11768 + }, + { + "epoch": 1.29244454206018, + "grad_norm": 2.4698240756988525, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7090677618980408, + "num_tokens": 302041821.0, + "step": 11769 + }, + { + "epoch": 1.2925543597627938, + "grad_norm": 2.4820659160614014, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7306609153747559, + "num_tokens": 302063273.0, + "step": 11770 + }, + { + "epoch": 1.2926641774654075, + "grad_norm": 2.1253609657287598, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7209402322769165, + "num_tokens": 302092598.0, + "step": 11771 + }, + { + "epoch": 1.292773995168021, + "grad_norm": 2.1183817386627197, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7323604226112366, + "num_tokens": 302119111.0, + "step": 11772 + }, + { + "epoch": 1.2928838128706348, + "grad_norm": 2.356261730194092, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7389826774597168, + "num_tokens": 302142211.0, + "step": 11773 + }, + { + "epoch": 1.2929936305732483, + "grad_norm": 2.2817704677581787, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7002618312835693, + "num_tokens": 302167392.0, + "step": 11774 + }, + { + "epoch": 1.293103448275862, + "grad_norm": 2.47617506980896, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7048169374465942, + "num_tokens": 302191093.0, + "step": 11775 + }, + { + "epoch": 1.2932132659784759, + "grad_norm": 2.233595609664917, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7178177833557129, + "num_tokens": 302217138.0, + "step": 11776 + }, + { + "epoch": 1.2933230836810894, + "grad_norm": 2.291821241378784, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7332446575164795, + "num_tokens": 302241228.0, + "step": 11777 + }, + { + "epoch": 1.293432901383703, + "grad_norm": 1.9655160903930664, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.732317328453064, + "num_tokens": 302272979.0, + "step": 11778 + }, + { + "epoch": 1.2935427190863167, + "grad_norm": 2.0597341060638428, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.746624231338501, + "num_tokens": 302300301.0, + "step": 11779 + }, + { + "epoch": 1.2936525367889304, + "grad_norm": 2.143747329711914, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7252980470657349, + "num_tokens": 302325374.0, + "step": 11780 + }, + { + "epoch": 1.293762354491544, + "grad_norm": 2.1709165573120117, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7346635460853577, + "num_tokens": 302349910.0, + "step": 11781 + }, + { + "epoch": 1.2938721721941577, + "grad_norm": 2.433286190032959, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7072131037712097, + "num_tokens": 302371887.0, + "step": 11782 + }, + { + "epoch": 1.2939819898967713, + "grad_norm": 2.3775506019592285, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.733856737613678, + "num_tokens": 302393764.0, + "step": 11783 + }, + { + "epoch": 1.294091807599385, + "grad_norm": 2.289520263671875, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7171435356140137, + "num_tokens": 302419297.0, + "step": 11784 + }, + { + "epoch": 1.2942016253019988, + "grad_norm": 2.381732225418091, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7030574083328247, + "num_tokens": 302442922.0, + "step": 11785 + }, + { + "epoch": 1.2943114430046123, + "grad_norm": 2.280540704727173, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7326638698577881, + "num_tokens": 302467407.0, + "step": 11786 + }, + { + "epoch": 1.294421260707226, + "grad_norm": 2.026289701461792, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7107391357421875, + "num_tokens": 302496788.0, + "step": 11787 + }, + { + "epoch": 1.2945310784098396, + "grad_norm": 2.7426092624664307, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7374390363693237, + "num_tokens": 302515385.0, + "step": 11788 + }, + { + "epoch": 1.2946408961124534, + "grad_norm": 2.4741828441619873, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7421969175338745, + "num_tokens": 302537086.0, + "step": 11789 + }, + { + "epoch": 1.2947507138150671, + "grad_norm": 2.283460855484009, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7623023986816406, + "num_tokens": 302559931.0, + "step": 11790 + }, + { + "epoch": 1.2948605315176807, + "grad_norm": 2.234318494796753, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.725671112537384, + "num_tokens": 302587649.0, + "step": 11791 + }, + { + "epoch": 1.2949703492202942, + "grad_norm": 2.410135507583618, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7437318563461304, + "num_tokens": 302609358.0, + "step": 11792 + }, + { + "epoch": 1.295080166922908, + "grad_norm": 2.1165964603424072, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6963338255882263, + "num_tokens": 302638155.0, + "step": 11793 + }, + { + "epoch": 1.2951899846255217, + "grad_norm": 2.3974997997283936, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7474344372749329, + "num_tokens": 302659166.0, + "step": 11794 + }, + { + "epoch": 1.2952998023281352, + "grad_norm": 2.123440980911255, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7130461931228638, + "num_tokens": 302686710.0, + "step": 11795 + }, + { + "epoch": 1.295409620030749, + "grad_norm": 2.357673406600952, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7055178284645081, + "num_tokens": 302710603.0, + "step": 11796 + }, + { + "epoch": 1.2955194377333625, + "grad_norm": 2.4195261001586914, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7202765345573425, + "num_tokens": 302732816.0, + "step": 11797 + }, + { + "epoch": 1.2956292554359763, + "grad_norm": 2.2597882747650146, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7171173691749573, + "num_tokens": 302756500.0, + "step": 11798 + }, + { + "epoch": 1.29573907313859, + "grad_norm": 2.3856046199798584, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7415151000022888, + "num_tokens": 302778036.0, + "step": 11799 + }, + { + "epoch": 1.2958488908412036, + "grad_norm": 2.3307878971099854, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7418000102043152, + "num_tokens": 302800253.0, + "step": 11800 + }, + { + "epoch": 1.2959587085438173, + "grad_norm": 2.622835159301758, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7337925434112549, + "num_tokens": 302819197.0, + "step": 11801 + }, + { + "epoch": 1.2960685262464309, + "grad_norm": 2.203773260116577, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7110550403594971, + "num_tokens": 302846275.0, + "step": 11802 + }, + { + "epoch": 1.2961783439490446, + "grad_norm": 2.2195534706115723, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6873443126678467, + "num_tokens": 302871689.0, + "step": 11803 + }, + { + "epoch": 1.2962881616516582, + "grad_norm": 2.618657112121582, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7159119248390198, + "num_tokens": 302892004.0, + "step": 11804 + }, + { + "epoch": 1.296397979354272, + "grad_norm": 2.540257692337036, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7273435592651367, + "num_tokens": 302913792.0, + "step": 11805 + }, + { + "epoch": 1.2965077970568855, + "grad_norm": 2.193666696548462, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.710779070854187, + "num_tokens": 302939930.0, + "step": 11806 + }, + { + "epoch": 1.2966176147594992, + "grad_norm": 2.0195999145507812, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7118182182312012, + "num_tokens": 302970622.0, + "step": 11807 + }, + { + "epoch": 1.296727432462113, + "grad_norm": 1.9897316694259644, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7149727940559387, + "num_tokens": 303002674.0, + "step": 11808 + }, + { + "epoch": 1.2968372501647265, + "grad_norm": 2.315857172012329, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7228541970252991, + "num_tokens": 303025965.0, + "step": 11809 + }, + { + "epoch": 1.2969470678673403, + "grad_norm": 2.503814458847046, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7292279005050659, + "num_tokens": 303045606.0, + "step": 11810 + }, + { + "epoch": 1.2970568855699538, + "grad_norm": 2.269073963165283, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7436027526855469, + "num_tokens": 303068456.0, + "step": 11811 + }, + { + "epoch": 1.2971667032725676, + "grad_norm": 2.3699164390563965, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7299872636795044, + "num_tokens": 303090334.0, + "step": 11812 + }, + { + "epoch": 1.2972765209751813, + "grad_norm": 2.199044942855835, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7224941849708557, + "num_tokens": 303115544.0, + "step": 11813 + }, + { + "epoch": 1.2973863386777948, + "grad_norm": 2.4060158729553223, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7025153040885925, + "num_tokens": 303140394.0, + "step": 11814 + }, + { + "epoch": 1.2974961563804086, + "grad_norm": 2.1052584648132324, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.728215217590332, + "num_tokens": 303168434.0, + "step": 11815 + }, + { + "epoch": 1.2976059740830221, + "grad_norm": 2.2645177841186523, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7080574631690979, + "num_tokens": 303194640.0, + "step": 11816 + }, + { + "epoch": 1.297715791785636, + "grad_norm": 2.3993940353393555, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7140825986862183, + "num_tokens": 303217386.0, + "step": 11817 + }, + { + "epoch": 1.2978256094882494, + "grad_norm": 2.4022858142852783, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7151079177856445, + "num_tokens": 303239381.0, + "step": 11818 + }, + { + "epoch": 1.2979354271908632, + "grad_norm": 2.2083868980407715, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7101229429244995, + "num_tokens": 303266512.0, + "step": 11819 + }, + { + "epoch": 1.2980452448934767, + "grad_norm": 2.1315112113952637, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7133570313453674, + "num_tokens": 303296479.0, + "step": 11820 + }, + { + "epoch": 1.2981550625960905, + "grad_norm": 2.0211498737335205, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7464322447776794, + "num_tokens": 303325382.0, + "step": 11821 + }, + { + "epoch": 1.2982648802987042, + "grad_norm": 2.2314083576202393, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7512470483779907, + "num_tokens": 303349108.0, + "step": 11822 + }, + { + "epoch": 1.2983746980013178, + "grad_norm": 2.4895026683807373, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7249997854232788, + "num_tokens": 303373044.0, + "step": 11823 + }, + { + "epoch": 1.2984845157039315, + "grad_norm": 2.28641676902771, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7397110462188721, + "num_tokens": 303396386.0, + "step": 11824 + }, + { + "epoch": 1.298594333406545, + "grad_norm": 2.259666681289673, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7076590657234192, + "num_tokens": 303422317.0, + "step": 11825 + }, + { + "epoch": 1.2987041511091588, + "grad_norm": 2.0406932830810547, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7107254266738892, + "num_tokens": 303452510.0, + "step": 11826 + }, + { + "epoch": 1.2988139688117726, + "grad_norm": 2.292323589324951, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7286825776100159, + "num_tokens": 303478326.0, + "step": 11827 + }, + { + "epoch": 1.2989237865143861, + "grad_norm": 2.8605458736419678, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7327256202697754, + "num_tokens": 303495569.0, + "step": 11828 + }, + { + "epoch": 1.2990336042169996, + "grad_norm": 2.3965508937835693, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7238205671310425, + "num_tokens": 303518477.0, + "step": 11829 + }, + { + "epoch": 1.2991434219196134, + "grad_norm": 2.1724815368652344, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7029579877853394, + "num_tokens": 303548504.0, + "step": 11830 + }, + { + "epoch": 1.2992532396222272, + "grad_norm": 2.2417404651641846, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7307010293006897, + "num_tokens": 303574654.0, + "step": 11831 + }, + { + "epoch": 1.2993630573248407, + "grad_norm": 2.3412420749664307, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7167012691497803, + "num_tokens": 303598374.0, + "step": 11832 + }, + { + "epoch": 1.2994728750274545, + "grad_norm": 2.0974934101104736, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7056975960731506, + "num_tokens": 303628697.0, + "step": 11833 + }, + { + "epoch": 1.299582692730068, + "grad_norm": 2.19819974899292, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7293710708618164, + "num_tokens": 303654623.0, + "step": 11834 + }, + { + "epoch": 1.2996925104326817, + "grad_norm": 2.324695587158203, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7024435997009277, + "num_tokens": 303681582.0, + "step": 11835 + }, + { + "epoch": 1.2998023281352955, + "grad_norm": 2.746107816696167, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7430996298789978, + "num_tokens": 303698549.0, + "step": 11836 + }, + { + "epoch": 1.299912145837909, + "grad_norm": 2.4507954120635986, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7145608067512512, + "num_tokens": 303720369.0, + "step": 11837 + }, + { + "epoch": 1.3000219635405228, + "grad_norm": 2.5697779655456543, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7264280319213867, + "num_tokens": 303741257.0, + "step": 11838 + }, + { + "epoch": 1.3001317812431363, + "grad_norm": 2.0407702922821045, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7089241743087769, + "num_tokens": 303770756.0, + "step": 11839 + }, + { + "epoch": 1.30024159894575, + "grad_norm": 2.141725778579712, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7059611082077026, + "num_tokens": 303798001.0, + "step": 11840 + }, + { + "epoch": 1.3003514166483638, + "grad_norm": 2.2184367179870605, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7087826132774353, + "num_tokens": 303824741.0, + "step": 11841 + }, + { + "epoch": 1.3004612343509774, + "grad_norm": 2.2384448051452637, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7376782298088074, + "num_tokens": 303851020.0, + "step": 11842 + }, + { + "epoch": 1.300571052053591, + "grad_norm": 2.198423385620117, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7052299380302429, + "num_tokens": 303878450.0, + "step": 11843 + }, + { + "epoch": 1.3006808697562047, + "grad_norm": 2.2271063327789307, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7206118106842041, + "num_tokens": 303904143.0, + "step": 11844 + }, + { + "epoch": 1.3007906874588184, + "grad_norm": 2.285911798477173, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7301321029663086, + "num_tokens": 303929629.0, + "step": 11845 + }, + { + "epoch": 1.300900505161432, + "grad_norm": 2.606938123703003, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7266727685928345, + "num_tokens": 303950264.0, + "step": 11846 + }, + { + "epoch": 1.3010103228640457, + "grad_norm": 2.0145905017852783, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7061331272125244, + "num_tokens": 303981923.0, + "step": 11847 + }, + { + "epoch": 1.3011201405666593, + "grad_norm": 2.2728443145751953, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7315840721130371, + "num_tokens": 304005932.0, + "step": 11848 + }, + { + "epoch": 1.301229958269273, + "grad_norm": 2.333292007446289, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7143514156341553, + "num_tokens": 304030719.0, + "step": 11849 + }, + { + "epoch": 1.3013397759718868, + "grad_norm": 2.1000864505767822, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7172242403030396, + "num_tokens": 304059845.0, + "step": 11850 + }, + { + "epoch": 1.3014495936745003, + "grad_norm": 2.1929526329040527, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7201954126358032, + "num_tokens": 304085239.0, + "step": 11851 + }, + { + "epoch": 1.301559411377114, + "grad_norm": 2.495110511779785, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7086172103881836, + "num_tokens": 304107163.0, + "step": 11852 + }, + { + "epoch": 1.3016692290797276, + "grad_norm": 2.0617756843566895, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7328135371208191, + "num_tokens": 304135001.0, + "step": 11853 + }, + { + "epoch": 1.3017790467823414, + "grad_norm": 2.3703105449676514, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7098960876464844, + "num_tokens": 304159273.0, + "step": 11854 + }, + { + "epoch": 1.301888864484955, + "grad_norm": 2.344569444656372, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7289809584617615, + "num_tokens": 304182725.0, + "step": 11855 + }, + { + "epoch": 1.3019986821875686, + "grad_norm": 2.5561578273773193, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7432847619056702, + "num_tokens": 304202702.0, + "step": 11856 + }, + { + "epoch": 1.3021084998901822, + "grad_norm": 2.1904497146606445, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7118164300918579, + "num_tokens": 304229931.0, + "step": 11857 + }, + { + "epoch": 1.302218317592796, + "grad_norm": 2.7370400428771973, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7234424352645874, + "num_tokens": 304248585.0, + "step": 11858 + }, + { + "epoch": 1.3023281352954097, + "grad_norm": 2.2325515747070312, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7167103290557861, + "num_tokens": 304273556.0, + "step": 11859 + }, + { + "epoch": 1.3024379529980232, + "grad_norm": 2.479875326156616, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7260850667953491, + "num_tokens": 304294997.0, + "step": 11860 + }, + { + "epoch": 1.302547770700637, + "grad_norm": 2.5707414150238037, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7511286735534668, + "num_tokens": 304313491.0, + "step": 11861 + }, + { + "epoch": 1.3026575884032505, + "grad_norm": 2.2531914710998535, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6874128580093384, + "num_tokens": 304341695.0, + "step": 11862 + }, + { + "epoch": 1.3027674061058643, + "grad_norm": 2.360703468322754, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7201440930366516, + "num_tokens": 304366110.0, + "step": 11863 + }, + { + "epoch": 1.302877223808478, + "grad_norm": 2.2113075256347656, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7124456763267517, + "num_tokens": 304393368.0, + "step": 11864 + }, + { + "epoch": 1.3029870415110916, + "grad_norm": 2.3313701152801514, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7170820236206055, + "num_tokens": 304417010.0, + "step": 11865 + }, + { + "epoch": 1.3030968592137053, + "grad_norm": 2.0236055850982666, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7460671663284302, + "num_tokens": 304444731.0, + "step": 11866 + }, + { + "epoch": 1.3032066769163189, + "grad_norm": 2.1264095306396484, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7221416234970093, + "num_tokens": 304474468.0, + "step": 11867 + }, + { + "epoch": 1.3033164946189326, + "grad_norm": 2.3100521564483643, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7315949201583862, + "num_tokens": 304496318.0, + "step": 11868 + }, + { + "epoch": 1.3034263123215462, + "grad_norm": 2.4634203910827637, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7376824617385864, + "num_tokens": 304519409.0, + "step": 11869 + }, + { + "epoch": 1.30353613002416, + "grad_norm": 2.2565464973449707, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7249824404716492, + "num_tokens": 304543658.0, + "step": 11870 + }, + { + "epoch": 1.3036459477267734, + "grad_norm": 2.014742612838745, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7220668792724609, + "num_tokens": 304573688.0, + "step": 11871 + }, + { + "epoch": 1.3037557654293872, + "grad_norm": 2.158564567565918, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7232150435447693, + "num_tokens": 304601183.0, + "step": 11872 + }, + { + "epoch": 1.303865583132001, + "grad_norm": 2.1482913494110107, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7247778177261353, + "num_tokens": 304630576.0, + "step": 11873 + }, + { + "epoch": 1.3039754008346145, + "grad_norm": 2.1762187480926514, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7227668762207031, + "num_tokens": 304656949.0, + "step": 11874 + }, + { + "epoch": 1.3040852185372283, + "grad_norm": 2.1809329986572266, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7337946891784668, + "num_tokens": 304683095.0, + "step": 11875 + }, + { + "epoch": 1.3041950362398418, + "grad_norm": 2.1419155597686768, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7013473510742188, + "num_tokens": 304711517.0, + "step": 11876 + }, + { + "epoch": 1.3043048539424555, + "grad_norm": 2.236517906188965, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7196907997131348, + "num_tokens": 304738114.0, + "step": 11877 + }, + { + "epoch": 1.3044146716450693, + "grad_norm": 2.396435499191284, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7308915853500366, + "num_tokens": 304762062.0, + "step": 11878 + }, + { + "epoch": 1.3045244893476828, + "grad_norm": 2.556581974029541, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.715227484703064, + "num_tokens": 304783515.0, + "step": 11879 + }, + { + "epoch": 1.3046343070502964, + "grad_norm": 2.1633448600769043, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7017320394515991, + "num_tokens": 304812922.0, + "step": 11880 + }, + { + "epoch": 1.3047441247529101, + "grad_norm": 2.3519489765167236, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7367821931838989, + "num_tokens": 304837481.0, + "step": 11881 + }, + { + "epoch": 1.3048539424555239, + "grad_norm": 2.3360090255737305, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7099981307983398, + "num_tokens": 304860275.0, + "step": 11882 + }, + { + "epoch": 1.3049637601581374, + "grad_norm": 2.4854683876037598, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7388590574264526, + "num_tokens": 304881426.0, + "step": 11883 + }, + { + "epoch": 1.3050735778607512, + "grad_norm": 2.3807151317596436, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7261478304862976, + "num_tokens": 304904821.0, + "step": 11884 + }, + { + "epoch": 1.3051833955633647, + "grad_norm": 2.3358898162841797, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7109072208404541, + "num_tokens": 304928155.0, + "step": 11885 + }, + { + "epoch": 1.3052932132659785, + "grad_norm": 2.396087646484375, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7347581386566162, + "num_tokens": 304950881.0, + "step": 11886 + }, + { + "epoch": 1.3054030309685922, + "grad_norm": 2.1506145000457764, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7006633281707764, + "num_tokens": 304980193.0, + "step": 11887 + }, + { + "epoch": 1.3055128486712058, + "grad_norm": 2.1316328048706055, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7262289524078369, + "num_tokens": 305008785.0, + "step": 11888 + }, + { + "epoch": 1.3056226663738195, + "grad_norm": 1.9888314008712769, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7257994413375854, + "num_tokens": 305039336.0, + "step": 11889 + }, + { + "epoch": 1.305732484076433, + "grad_norm": 1.953342318534851, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7039340734481812, + "num_tokens": 305072536.0, + "step": 11890 + }, + { + "epoch": 1.3058423017790468, + "grad_norm": 2.6986796855926514, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7469081878662109, + "num_tokens": 305091783.0, + "step": 11891 + }, + { + "epoch": 1.3059521194816606, + "grad_norm": 2.1400768756866455, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7460883855819702, + "num_tokens": 305118903.0, + "step": 11892 + }, + { + "epoch": 1.306061937184274, + "grad_norm": 2.1843717098236084, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7180237770080566, + "num_tokens": 305146522.0, + "step": 11893 + }, + { + "epoch": 1.3061717548868876, + "grad_norm": 2.018448829650879, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7296338677406311, + "num_tokens": 305176789.0, + "step": 11894 + }, + { + "epoch": 1.3062815725895014, + "grad_norm": 2.1816039085388184, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7272491455078125, + "num_tokens": 305203581.0, + "step": 11895 + }, + { + "epoch": 1.3063913902921152, + "grad_norm": 2.445127248764038, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7244257926940918, + "num_tokens": 305223901.0, + "step": 11896 + }, + { + "epoch": 1.3065012079947287, + "grad_norm": 2.4695661067962646, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7390686273574829, + "num_tokens": 305245786.0, + "step": 11897 + }, + { + "epoch": 1.3066110256973424, + "grad_norm": 2.360578775405884, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7346343994140625, + "num_tokens": 305272140.0, + "step": 11898 + }, + { + "epoch": 1.306720843399956, + "grad_norm": 2.264192819595337, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6986891031265259, + "num_tokens": 305296139.0, + "step": 11899 + }, + { + "epoch": 1.3068306611025697, + "grad_norm": 2.15387225151062, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6941663026809692, + "num_tokens": 305324388.0, + "step": 11900 + }, + { + "epoch": 1.3069404788051835, + "grad_norm": 2.156205177307129, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7162826061248779, + "num_tokens": 305349880.0, + "step": 11901 + }, + { + "epoch": 1.307050296507797, + "grad_norm": 2.0495684146881104, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.709845244884491, + "num_tokens": 305380497.0, + "step": 11902 + }, + { + "epoch": 1.3071601142104108, + "grad_norm": 2.2973294258117676, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7253808975219727, + "num_tokens": 305405396.0, + "step": 11903 + }, + { + "epoch": 1.3072699319130243, + "grad_norm": 2.338134765625, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7252727150917053, + "num_tokens": 305429054.0, + "step": 11904 + }, + { + "epoch": 1.307379749615638, + "grad_norm": 2.1991779804229736, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7030329704284668, + "num_tokens": 305458928.0, + "step": 11905 + }, + { + "epoch": 1.3074895673182518, + "grad_norm": 2.4180102348327637, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.726911723613739, + "num_tokens": 305479243.0, + "step": 11906 + }, + { + "epoch": 1.3075993850208654, + "grad_norm": 2.0814077854156494, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.6995850205421448, + "num_tokens": 305509685.0, + "step": 11907 + }, + { + "epoch": 1.307709202723479, + "grad_norm": 2.006230592727661, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7076313495635986, + "num_tokens": 305539579.0, + "step": 11908 + }, + { + "epoch": 1.3078190204260927, + "grad_norm": 2.1739919185638428, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7069575786590576, + "num_tokens": 305567899.0, + "step": 11909 + }, + { + "epoch": 1.3079288381287064, + "grad_norm": 2.231210947036743, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.707205057144165, + "num_tokens": 305594154.0, + "step": 11910 + }, + { + "epoch": 1.30803865583132, + "grad_norm": 2.4573252201080322, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7280733585357666, + "num_tokens": 305616380.0, + "step": 11911 + }, + { + "epoch": 1.3081484735339337, + "grad_norm": 2.6583175659179688, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7385291457176208, + "num_tokens": 305636303.0, + "step": 11912 + }, + { + "epoch": 1.3082582912365472, + "grad_norm": 2.1780240535736084, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7074106931686401, + "num_tokens": 305664938.0, + "step": 11913 + }, + { + "epoch": 1.308368108939161, + "grad_norm": 2.5232019424438477, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7503399848937988, + "num_tokens": 305683978.0, + "step": 11914 + }, + { + "epoch": 1.3084779266417748, + "grad_norm": 1.9728193283081055, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7097444534301758, + "num_tokens": 305715179.0, + "step": 11915 + }, + { + "epoch": 1.3085877443443883, + "grad_norm": 2.319406747817993, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7052508592605591, + "num_tokens": 305739755.0, + "step": 11916 + }, + { + "epoch": 1.308697562047002, + "grad_norm": 2.3916635513305664, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7468087673187256, + "num_tokens": 305763127.0, + "step": 11917 + }, + { + "epoch": 1.3088073797496156, + "grad_norm": 1.9559129476547241, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6969954967498779, + "num_tokens": 305795549.0, + "step": 11918 + }, + { + "epoch": 1.3089171974522293, + "grad_norm": 2.298997163772583, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7228571176528931, + "num_tokens": 305820887.0, + "step": 11919 + }, + { + "epoch": 1.309027015154843, + "grad_norm": 2.2793548107147217, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7353346347808838, + "num_tokens": 305846605.0, + "step": 11920 + }, + { + "epoch": 1.3091368328574566, + "grad_norm": 2.1073949337005615, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7068300843238831, + "num_tokens": 305874459.0, + "step": 11921 + }, + { + "epoch": 1.3092466505600702, + "grad_norm": 2.3531036376953125, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.722952127456665, + "num_tokens": 305900471.0, + "step": 11922 + }, + { + "epoch": 1.309356468262684, + "grad_norm": 2.4881885051727295, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7288468480110168, + "num_tokens": 305921539.0, + "step": 11923 + }, + { + "epoch": 1.3094662859652977, + "grad_norm": 2.2633118629455566, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7472741603851318, + "num_tokens": 305947282.0, + "step": 11924 + }, + { + "epoch": 1.3095761036679112, + "grad_norm": 2.423408031463623, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7284939289093018, + "num_tokens": 305970983.0, + "step": 11925 + }, + { + "epoch": 1.309685921370525, + "grad_norm": 2.2146499156951904, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7104892134666443, + "num_tokens": 305995897.0, + "step": 11926 + }, + { + "epoch": 1.3097957390731385, + "grad_norm": 2.4617233276367188, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7360007166862488, + "num_tokens": 306018407.0, + "step": 11927 + }, + { + "epoch": 1.3099055567757523, + "grad_norm": 2.160417079925537, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7269632816314697, + "num_tokens": 306045884.0, + "step": 11928 + }, + { + "epoch": 1.310015374478366, + "grad_norm": 1.9559547901153564, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6870830059051514, + "num_tokens": 306078800.0, + "step": 11929 + }, + { + "epoch": 1.3101251921809796, + "grad_norm": 2.144428253173828, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.6991651058197021, + "num_tokens": 306104952.0, + "step": 11930 + }, + { + "epoch": 1.3102350098835933, + "grad_norm": 2.143509864807129, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7154091596603394, + "num_tokens": 306135855.0, + "step": 11931 + }, + { + "epoch": 1.3103448275862069, + "grad_norm": 1.9835630655288696, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7249307632446289, + "num_tokens": 306166398.0, + "step": 11932 + }, + { + "epoch": 1.3104546452888206, + "grad_norm": 2.1108596324920654, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7090217471122742, + "num_tokens": 306195355.0, + "step": 11933 + }, + { + "epoch": 1.3105644629914341, + "grad_norm": 2.0561463832855225, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7080991268157959, + "num_tokens": 306223833.0, + "step": 11934 + }, + { + "epoch": 1.310674280694048, + "grad_norm": 2.2033467292785645, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6881369352340698, + "num_tokens": 306253916.0, + "step": 11935 + }, + { + "epoch": 1.3107840983966614, + "grad_norm": 2.334272623062134, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7122055888175964, + "num_tokens": 306280348.0, + "step": 11936 + }, + { + "epoch": 1.3108939160992752, + "grad_norm": 2.371859312057495, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6982249021530151, + "num_tokens": 306304472.0, + "step": 11937 + }, + { + "epoch": 1.311003733801889, + "grad_norm": 2.21271014213562, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7221693992614746, + "num_tokens": 306330442.0, + "step": 11938 + }, + { + "epoch": 1.3111135515045025, + "grad_norm": 2.405271291732788, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.728219211101532, + "num_tokens": 306353915.0, + "step": 11939 + }, + { + "epoch": 1.3112233692071162, + "grad_norm": 2.224565029144287, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7372931838035583, + "num_tokens": 306380118.0, + "step": 11940 + }, + { + "epoch": 1.3113331869097298, + "grad_norm": 2.0859529972076416, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.695654034614563, + "num_tokens": 306409317.0, + "step": 11941 + }, + { + "epoch": 1.3114430046123435, + "grad_norm": 2.301711082458496, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7087409496307373, + "num_tokens": 306435111.0, + "step": 11942 + }, + { + "epoch": 1.3115528223149573, + "grad_norm": 2.449711322784424, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7174967527389526, + "num_tokens": 306458500.0, + "step": 11943 + }, + { + "epoch": 1.3116626400175708, + "grad_norm": 2.223905086517334, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6971399188041687, + "num_tokens": 306484560.0, + "step": 11944 + }, + { + "epoch": 1.3117724577201844, + "grad_norm": 2.2503983974456787, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7046052813529968, + "num_tokens": 306509707.0, + "step": 11945 + }, + { + "epoch": 1.3118822754227981, + "grad_norm": 2.2650327682495117, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7009223699569702, + "num_tokens": 306535269.0, + "step": 11946 + }, + { + "epoch": 1.3119920931254119, + "grad_norm": 2.096935749053955, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7136785387992859, + "num_tokens": 306561381.0, + "step": 11947 + }, + { + "epoch": 1.3121019108280254, + "grad_norm": 2.073166608810425, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7051528692245483, + "num_tokens": 306589204.0, + "step": 11948 + }, + { + "epoch": 1.3122117285306392, + "grad_norm": 2.199979066848755, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6955101490020752, + "num_tokens": 306614768.0, + "step": 11949 + }, + { + "epoch": 1.3123215462332527, + "grad_norm": 2.206444263458252, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7033689618110657, + "num_tokens": 306642266.0, + "step": 11950 + }, + { + "epoch": 1.3124313639358665, + "grad_norm": 2.25429105758667, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.725878119468689, + "num_tokens": 306666329.0, + "step": 11951 + }, + { + "epoch": 1.3125411816384802, + "grad_norm": 2.2940423488616943, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7117615938186646, + "num_tokens": 306690283.0, + "step": 11952 + }, + { + "epoch": 1.3126509993410937, + "grad_norm": 2.171210765838623, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7092256546020508, + "num_tokens": 306715688.0, + "step": 11953 + }, + { + "epoch": 1.3127608170437075, + "grad_norm": 2.2816226482391357, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6985515356063843, + "num_tokens": 306743626.0, + "step": 11954 + }, + { + "epoch": 1.312870634746321, + "grad_norm": 2.294987678527832, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.719515323638916, + "num_tokens": 306767773.0, + "step": 11955 + }, + { + "epoch": 1.3129804524489348, + "grad_norm": 2.675093650817871, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7284132242202759, + "num_tokens": 306786745.0, + "step": 11956 + }, + { + "epoch": 1.3130902701515486, + "grad_norm": 2.418670654296875, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7294583320617676, + "num_tokens": 306808698.0, + "step": 11957 + }, + { + "epoch": 1.313200087854162, + "grad_norm": 2.377682685852051, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7254645824432373, + "num_tokens": 306829594.0, + "step": 11958 + }, + { + "epoch": 1.3133099055567756, + "grad_norm": 2.1435787677764893, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.698296844959259, + "num_tokens": 306859411.0, + "step": 11959 + }, + { + "epoch": 1.3134197232593894, + "grad_norm": 2.575834274291992, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7208282947540283, + "num_tokens": 306879659.0, + "step": 11960 + }, + { + "epoch": 1.3135295409620031, + "grad_norm": 2.0788280963897705, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7053914666175842, + "num_tokens": 306909613.0, + "step": 11961 + }, + { + "epoch": 1.3136393586646167, + "grad_norm": 2.0943541526794434, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7111308574676514, + "num_tokens": 306941254.0, + "step": 11962 + }, + { + "epoch": 1.3137491763672304, + "grad_norm": 2.363748073577881, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7200908064842224, + "num_tokens": 306965667.0, + "step": 11963 + }, + { + "epoch": 1.313858994069844, + "grad_norm": 2.3420863151550293, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7497997879981995, + "num_tokens": 306988719.0, + "step": 11964 + }, + { + "epoch": 1.3139688117724577, + "grad_norm": 2.322955846786499, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6996711492538452, + "num_tokens": 307014550.0, + "step": 11965 + }, + { + "epoch": 1.3140786294750715, + "grad_norm": 2.0681848526000977, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7032349109649658, + "num_tokens": 307042456.0, + "step": 11966 + }, + { + "epoch": 1.314188447177685, + "grad_norm": 2.198863983154297, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.732692539691925, + "num_tokens": 307068066.0, + "step": 11967 + }, + { + "epoch": 1.3142982648802988, + "grad_norm": 2.059155225753784, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6866965889930725, + "num_tokens": 307097898.0, + "step": 11968 + }, + { + "epoch": 1.3144080825829123, + "grad_norm": 2.516791582107544, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7325960397720337, + "num_tokens": 307119145.0, + "step": 11969 + }, + { + "epoch": 1.314517900285526, + "grad_norm": 2.1148574352264404, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7129127979278564, + "num_tokens": 307148185.0, + "step": 11970 + }, + { + "epoch": 1.3146277179881398, + "grad_norm": 2.428813934326172, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7081707715988159, + "num_tokens": 307170883.0, + "step": 11971 + }, + { + "epoch": 1.3147375356907534, + "grad_norm": 2.4792592525482178, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7260012030601501, + "num_tokens": 307192436.0, + "step": 11972 + }, + { + "epoch": 1.314847353393367, + "grad_norm": 2.2217442989349365, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6959385871887207, + "num_tokens": 307220701.0, + "step": 11973 + }, + { + "epoch": 1.3149571710959806, + "grad_norm": 2.537977933883667, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7149779200553894, + "num_tokens": 307241824.0, + "step": 11974 + }, + { + "epoch": 1.3150669887985944, + "grad_norm": 2.2485451698303223, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7140414714813232, + "num_tokens": 307266137.0, + "step": 11975 + }, + { + "epoch": 1.315176806501208, + "grad_norm": 2.1213784217834473, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7086600065231323, + "num_tokens": 307294595.0, + "step": 11976 + }, + { + "epoch": 1.3152866242038217, + "grad_norm": 2.229965925216675, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7296029329299927, + "num_tokens": 307318159.0, + "step": 11977 + }, + { + "epoch": 1.3153964419064352, + "grad_norm": 2.013115644454956, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.701854944229126, + "num_tokens": 307349479.0, + "step": 11978 + }, + { + "epoch": 1.315506259609049, + "grad_norm": 2.4228713512420654, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7154610753059387, + "num_tokens": 307372289.0, + "step": 11979 + }, + { + "epoch": 1.3156160773116627, + "grad_norm": 2.15268874168396, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7129406332969666, + "num_tokens": 307397440.0, + "step": 11980 + }, + { + "epoch": 1.3157258950142763, + "grad_norm": 1.8394018411636353, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7246049642562866, + "num_tokens": 307432380.0, + "step": 11981 + }, + { + "epoch": 1.31583571271689, + "grad_norm": 2.449734687805176, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7360429763793945, + "num_tokens": 307453689.0, + "step": 11982 + }, + { + "epoch": 1.3159455304195036, + "grad_norm": 2.2385685443878174, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7248995304107666, + "num_tokens": 307480092.0, + "step": 11983 + }, + { + "epoch": 1.3160553481221173, + "grad_norm": 2.258789300918579, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7300014495849609, + "num_tokens": 307506100.0, + "step": 11984 + }, + { + "epoch": 1.3161651658247309, + "grad_norm": 2.079817533493042, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7253155708312988, + "num_tokens": 307533893.0, + "step": 11985 + }, + { + "epoch": 1.3162749835273446, + "grad_norm": 2.1846771240234375, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7085036039352417, + "num_tokens": 307562381.0, + "step": 11986 + }, + { + "epoch": 1.3163848012299582, + "grad_norm": 2.4577035903930664, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7380937933921814, + "num_tokens": 307583540.0, + "step": 11987 + }, + { + "epoch": 1.316494618932572, + "grad_norm": 2.3794467449188232, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7067010402679443, + "num_tokens": 307605905.0, + "step": 11988 + }, + { + "epoch": 1.3166044366351857, + "grad_norm": 2.146378517150879, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7421497106552124, + "num_tokens": 307634372.0, + "step": 11989 + }, + { + "epoch": 1.3167142543377992, + "grad_norm": 1.961574673652649, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7004200220108032, + "num_tokens": 307665497.0, + "step": 11990 + }, + { + "epoch": 1.316824072040413, + "grad_norm": 2.314530372619629, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7288873195648193, + "num_tokens": 307690444.0, + "step": 11991 + }, + { + "epoch": 1.3169338897430265, + "grad_norm": 2.303603410720825, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7039980888366699, + "num_tokens": 307716805.0, + "step": 11992 + }, + { + "epoch": 1.3170437074456403, + "grad_norm": 2.1518373489379883, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7109609246253967, + "num_tokens": 307745798.0, + "step": 11993 + }, + { + "epoch": 1.317153525148254, + "grad_norm": 2.379370927810669, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7329142093658447, + "num_tokens": 307769366.0, + "step": 11994 + }, + { + "epoch": 1.3172633428508675, + "grad_norm": 2.0652267932891846, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7253592014312744, + "num_tokens": 307797404.0, + "step": 11995 + }, + { + "epoch": 1.3173731605534813, + "grad_norm": 2.4402964115142822, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7228220701217651, + "num_tokens": 307819819.0, + "step": 11996 + }, + { + "epoch": 1.3174829782560948, + "grad_norm": 2.1426939964294434, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7518097758293152, + "num_tokens": 307845900.0, + "step": 11997 + }, + { + "epoch": 1.3175927959587086, + "grad_norm": 2.164198875427246, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6909170150756836, + "num_tokens": 307874547.0, + "step": 11998 + }, + { + "epoch": 1.3177026136613221, + "grad_norm": 2.113356828689575, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7067726254463196, + "num_tokens": 307902973.0, + "step": 11999 + }, + { + "epoch": 1.3178124313639359, + "grad_norm": 2.2203054428100586, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7117559909820557, + "num_tokens": 307930417.0, + "step": 12000 + }, + { + "epoch": 1.3179222490665494, + "grad_norm": 2.5824825763702393, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7276027202606201, + "num_tokens": 307949558.0, + "step": 12001 + }, + { + "epoch": 1.3180320667691632, + "grad_norm": 2.32301926612854, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7124906182289124, + "num_tokens": 307975442.0, + "step": 12002 + }, + { + "epoch": 1.318141884471777, + "grad_norm": 2.204380512237549, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7212645411491394, + "num_tokens": 307999788.0, + "step": 12003 + }, + { + "epoch": 1.3182517021743905, + "grad_norm": 2.257007360458374, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7021857500076294, + "num_tokens": 308025212.0, + "step": 12004 + }, + { + "epoch": 1.3183615198770042, + "grad_norm": 2.335020065307617, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7254934310913086, + "num_tokens": 308050185.0, + "step": 12005 + }, + { + "epoch": 1.3184713375796178, + "grad_norm": 2.3876917362213135, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7095729112625122, + "num_tokens": 308075176.0, + "step": 12006 + }, + { + "epoch": 1.3185811552822315, + "grad_norm": 2.2247819900512695, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7038052082061768, + "num_tokens": 308102621.0, + "step": 12007 + }, + { + "epoch": 1.3186909729848453, + "grad_norm": 2.3197214603424072, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7286577224731445, + "num_tokens": 308126196.0, + "step": 12008 + }, + { + "epoch": 1.3188007906874588, + "grad_norm": 2.3375346660614014, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7278949022293091, + "num_tokens": 308150224.0, + "step": 12009 + }, + { + "epoch": 1.3189106083900723, + "grad_norm": 2.3811862468719482, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7249324917793274, + "num_tokens": 308174351.0, + "step": 12010 + }, + { + "epoch": 1.319020426092686, + "grad_norm": 2.523566484451294, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7392164468765259, + "num_tokens": 308194335.0, + "step": 12011 + }, + { + "epoch": 1.3191302437952999, + "grad_norm": 2.4784092903137207, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7215257883071899, + "num_tokens": 308217536.0, + "step": 12012 + }, + { + "epoch": 1.3192400614979134, + "grad_norm": 2.088604211807251, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.718037486076355, + "num_tokens": 308244346.0, + "step": 12013 + }, + { + "epoch": 1.3193498792005272, + "grad_norm": 2.2414112091064453, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7201525568962097, + "num_tokens": 308270624.0, + "step": 12014 + }, + { + "epoch": 1.3194596969031407, + "grad_norm": 2.4468367099761963, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.734012246131897, + "num_tokens": 308292333.0, + "step": 12015 + }, + { + "epoch": 1.3195695146057544, + "grad_norm": 2.4958384037017822, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7363338470458984, + "num_tokens": 308314161.0, + "step": 12016 + }, + { + "epoch": 1.3196793323083682, + "grad_norm": 2.4051449298858643, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7359083890914917, + "num_tokens": 308336863.0, + "step": 12017 + }, + { + "epoch": 1.3197891500109817, + "grad_norm": 2.3799216747283936, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7253532409667969, + "num_tokens": 308359921.0, + "step": 12018 + }, + { + "epoch": 1.3198989677135955, + "grad_norm": 2.624584674835205, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7366252541542053, + "num_tokens": 308380021.0, + "step": 12019 + }, + { + "epoch": 1.320008785416209, + "grad_norm": 2.0817365646362305, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7391592264175415, + "num_tokens": 308408004.0, + "step": 12020 + }, + { + "epoch": 1.3201186031188228, + "grad_norm": 2.2516472339630127, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7243744730949402, + "num_tokens": 308433456.0, + "step": 12021 + }, + { + "epoch": 1.3202284208214365, + "grad_norm": 2.1955630779266357, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7353969216346741, + "num_tokens": 308460000.0, + "step": 12022 + }, + { + "epoch": 1.32033823852405, + "grad_norm": 2.4008095264434814, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7321859002113342, + "num_tokens": 308481635.0, + "step": 12023 + }, + { + "epoch": 1.3204480562266636, + "grad_norm": 2.2819600105285645, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7047932147979736, + "num_tokens": 308508370.0, + "step": 12024 + }, + { + "epoch": 1.3205578739292774, + "grad_norm": 2.2401981353759766, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.702978253364563, + "num_tokens": 308536047.0, + "step": 12025 + }, + { + "epoch": 1.3206676916318911, + "grad_norm": 2.1603879928588867, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7266362905502319, + "num_tokens": 308562662.0, + "step": 12026 + }, + { + "epoch": 1.3207775093345047, + "grad_norm": 2.1202003955841064, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7052294015884399, + "num_tokens": 308591306.0, + "step": 12027 + }, + { + "epoch": 1.3208873270371184, + "grad_norm": 2.1679816246032715, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7315715551376343, + "num_tokens": 308617857.0, + "step": 12028 + }, + { + "epoch": 1.320997144739732, + "grad_norm": 2.415562629699707, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7379282712936401, + "num_tokens": 308640050.0, + "step": 12029 + }, + { + "epoch": 1.3211069624423457, + "grad_norm": 2.4742016792297363, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7086421251296997, + "num_tokens": 308662427.0, + "step": 12030 + }, + { + "epoch": 1.3212167801449595, + "grad_norm": 2.2072112560272217, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7207618951797485, + "num_tokens": 308688610.0, + "step": 12031 + }, + { + "epoch": 1.321326597847573, + "grad_norm": 2.5596837997436523, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7235726714134216, + "num_tokens": 308709221.0, + "step": 12032 + }, + { + "epoch": 1.3214364155501868, + "grad_norm": 2.249577045440674, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.732965350151062, + "num_tokens": 308734039.0, + "step": 12033 + }, + { + "epoch": 1.3215462332528003, + "grad_norm": 2.130584478378296, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7136867046356201, + "num_tokens": 308761399.0, + "step": 12034 + }, + { + "epoch": 1.321656050955414, + "grad_norm": 2.3864223957061768, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7353343963623047, + "num_tokens": 308785706.0, + "step": 12035 + }, + { + "epoch": 1.3217658686580278, + "grad_norm": 2.0667905807495117, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7198609113693237, + "num_tokens": 308814995.0, + "step": 12036 + }, + { + "epoch": 1.3218756863606413, + "grad_norm": 2.272623300552368, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7161158323287964, + "num_tokens": 308839064.0, + "step": 12037 + }, + { + "epoch": 1.3219855040632549, + "grad_norm": 2.2586352825164795, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7379319071769714, + "num_tokens": 308864277.0, + "step": 12038 + }, + { + "epoch": 1.3220953217658686, + "grad_norm": 2.1334445476531982, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7059839367866516, + "num_tokens": 308893472.0, + "step": 12039 + }, + { + "epoch": 1.3222051394684824, + "grad_norm": 2.20869779586792, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7188969850540161, + "num_tokens": 308918436.0, + "step": 12040 + }, + { + "epoch": 1.322314957171096, + "grad_norm": 2.066469669342041, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7087090015411377, + "num_tokens": 308948262.0, + "step": 12041 + }, + { + "epoch": 1.3224247748737097, + "grad_norm": 2.0754544734954834, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7031369805335999, + "num_tokens": 308977965.0, + "step": 12042 + }, + { + "epoch": 1.3225345925763232, + "grad_norm": 2.1883544921875, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7140153050422668, + "num_tokens": 309004666.0, + "step": 12043 + }, + { + "epoch": 1.322644410278937, + "grad_norm": 2.424835205078125, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7340782880783081, + "num_tokens": 309027279.0, + "step": 12044 + }, + { + "epoch": 1.3227542279815507, + "grad_norm": 2.410037040710449, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.72662353515625, + "num_tokens": 309049187.0, + "step": 12045 + }, + { + "epoch": 1.3228640456841643, + "grad_norm": 2.2033209800720215, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7143815159797668, + "num_tokens": 309077193.0, + "step": 12046 + }, + { + "epoch": 1.322973863386778, + "grad_norm": 2.3163352012634277, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6965397596359253, + "num_tokens": 309104962.0, + "step": 12047 + }, + { + "epoch": 1.3230836810893916, + "grad_norm": 2.2466580867767334, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7346322536468506, + "num_tokens": 309129398.0, + "step": 12048 + }, + { + "epoch": 1.3231934987920053, + "grad_norm": 2.190483570098877, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7192103266716003, + "num_tokens": 309156608.0, + "step": 12049 + }, + { + "epoch": 1.3233033164946189, + "grad_norm": 2.0219316482543945, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7357126474380493, + "num_tokens": 309186502.0, + "step": 12050 + }, + { + "epoch": 1.3234131341972326, + "grad_norm": 2.2381792068481445, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7261784076690674, + "num_tokens": 309213256.0, + "step": 12051 + }, + { + "epoch": 1.3235229518998461, + "grad_norm": 2.257261276245117, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7313919067382812, + "num_tokens": 309237877.0, + "step": 12052 + }, + { + "epoch": 1.32363276960246, + "grad_norm": 2.0961790084838867, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7219018340110779, + "num_tokens": 309267740.0, + "step": 12053 + }, + { + "epoch": 1.3237425873050737, + "grad_norm": 2.0178210735321045, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7095340490341187, + "num_tokens": 309301580.0, + "step": 12054 + }, + { + "epoch": 1.3238524050076872, + "grad_norm": 1.997367024421692, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7190684080123901, + "num_tokens": 309330610.0, + "step": 12055 + }, + { + "epoch": 1.323962222710301, + "grad_norm": 2.3437459468841553, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7281481623649597, + "num_tokens": 309352781.0, + "step": 12056 + }, + { + "epoch": 1.3240720404129145, + "grad_norm": 2.2018954753875732, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7328321933746338, + "num_tokens": 309377864.0, + "step": 12057 + }, + { + "epoch": 1.3241818581155282, + "grad_norm": 2.488375186920166, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7342917919158936, + "num_tokens": 309399721.0, + "step": 12058 + }, + { + "epoch": 1.324291675818142, + "grad_norm": 2.544621706008911, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7246384620666504, + "num_tokens": 309421219.0, + "step": 12059 + }, + { + "epoch": 1.3244014935207555, + "grad_norm": 2.150933265686035, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.72520512342453, + "num_tokens": 309448093.0, + "step": 12060 + }, + { + "epoch": 1.324511311223369, + "grad_norm": 2.3293421268463135, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7190260887145996, + "num_tokens": 309473350.0, + "step": 12061 + }, + { + "epoch": 1.3246211289259828, + "grad_norm": 2.558009624481201, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7293060421943665, + "num_tokens": 309494239.0, + "step": 12062 + }, + { + "epoch": 1.3247309466285966, + "grad_norm": 2.2326505184173584, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6971994042396545, + "num_tokens": 309524325.0, + "step": 12063 + }, + { + "epoch": 1.3248407643312101, + "grad_norm": 2.4598288536071777, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7073046565055847, + "num_tokens": 309547167.0, + "step": 12064 + }, + { + "epoch": 1.3249505820338239, + "grad_norm": 1.8838646411895752, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7028443813323975, + "num_tokens": 309583598.0, + "step": 12065 + }, + { + "epoch": 1.3250603997364374, + "grad_norm": 2.401829719543457, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7201862335205078, + "num_tokens": 309605659.0, + "step": 12066 + }, + { + "epoch": 1.3251702174390512, + "grad_norm": 2.5247433185577393, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7195823192596436, + "num_tokens": 309626666.0, + "step": 12067 + }, + { + "epoch": 1.325280035141665, + "grad_norm": 2.308811664581299, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7207186818122864, + "num_tokens": 309651908.0, + "step": 12068 + }, + { + "epoch": 1.3253898528442785, + "grad_norm": 2.3196709156036377, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7024752497673035, + "num_tokens": 309677470.0, + "step": 12069 + }, + { + "epoch": 1.3254996705468922, + "grad_norm": 2.0458672046661377, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7124246954917908, + "num_tokens": 309705573.0, + "step": 12070 + }, + { + "epoch": 1.3256094882495058, + "grad_norm": 2.3659210205078125, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7183977365493774, + "num_tokens": 309728333.0, + "step": 12071 + }, + { + "epoch": 1.3257193059521195, + "grad_norm": 2.165684700012207, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7158480286598206, + "num_tokens": 309756266.0, + "step": 12072 + }, + { + "epoch": 1.3258291236547333, + "grad_norm": 2.068615436553955, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7121143341064453, + "num_tokens": 309786575.0, + "step": 12073 + }, + { + "epoch": 1.3259389413573468, + "grad_norm": 2.3381686210632324, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7461912631988525, + "num_tokens": 309808669.0, + "step": 12074 + }, + { + "epoch": 1.3260487590599603, + "grad_norm": 2.418888807296753, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7095284461975098, + "num_tokens": 309831430.0, + "step": 12075 + }, + { + "epoch": 1.326158576762574, + "grad_norm": 2.052971363067627, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7389082908630371, + "num_tokens": 309860371.0, + "step": 12076 + }, + { + "epoch": 1.3262683944651878, + "grad_norm": 2.143998384475708, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7063902020454407, + "num_tokens": 309889992.0, + "step": 12077 + }, + { + "epoch": 1.3263782121678014, + "grad_norm": 2.0714454650878906, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7284359335899353, + "num_tokens": 309918985.0, + "step": 12078 + }, + { + "epoch": 1.3264880298704151, + "grad_norm": 2.179363965988159, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7066684365272522, + "num_tokens": 309946924.0, + "step": 12079 + }, + { + "epoch": 1.3265978475730287, + "grad_norm": 2.4122843742370605, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7153012156486511, + "num_tokens": 309969449.0, + "step": 12080 + }, + { + "epoch": 1.3267076652756424, + "grad_norm": 2.322455644607544, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7290982007980347, + "num_tokens": 309994307.0, + "step": 12081 + }, + { + "epoch": 1.3268174829782562, + "grad_norm": 2.0859553813934326, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7430837750434875, + "num_tokens": 310021679.0, + "step": 12082 + }, + { + "epoch": 1.3269273006808697, + "grad_norm": 2.389859914779663, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7270138263702393, + "num_tokens": 310044191.0, + "step": 12083 + }, + { + "epoch": 1.3270371183834835, + "grad_norm": 2.415178060531616, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7117274403572083, + "num_tokens": 310067758.0, + "step": 12084 + }, + { + "epoch": 1.327146936086097, + "grad_norm": 2.1167705059051514, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7219551801681519, + "num_tokens": 310096728.0, + "step": 12085 + }, + { + "epoch": 1.3272567537887108, + "grad_norm": 2.241467237472534, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.728974461555481, + "num_tokens": 310120908.0, + "step": 12086 + }, + { + "epoch": 1.3273665714913245, + "grad_norm": 2.328857421875, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7424758672714233, + "num_tokens": 310142250.0, + "step": 12087 + }, + { + "epoch": 1.327476389193938, + "grad_norm": 2.3072359561920166, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7208517789840698, + "num_tokens": 310168559.0, + "step": 12088 + }, + { + "epoch": 1.3275862068965516, + "grad_norm": 2.141338586807251, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.72648686170578, + "num_tokens": 310196249.0, + "step": 12089 + }, + { + "epoch": 1.3276960245991654, + "grad_norm": 2.2813289165496826, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7332212924957275, + "num_tokens": 310221284.0, + "step": 12090 + }, + { + "epoch": 1.3278058423017791, + "grad_norm": 2.143388509750366, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7226859927177429, + "num_tokens": 310248040.0, + "step": 12091 + }, + { + "epoch": 1.3279156600043927, + "grad_norm": 2.3092429637908936, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.717119574546814, + "num_tokens": 310273685.0, + "step": 12092 + }, + { + "epoch": 1.3280254777070064, + "grad_norm": 2.0670409202575684, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7253752946853638, + "num_tokens": 310303960.0, + "step": 12093 + }, + { + "epoch": 1.32813529540962, + "grad_norm": 2.147103786468506, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7207744717597961, + "num_tokens": 310330214.0, + "step": 12094 + }, + { + "epoch": 1.3282451131122337, + "grad_norm": 2.3084495067596436, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7066113948822021, + "num_tokens": 310357555.0, + "step": 12095 + }, + { + "epoch": 1.3283549308148475, + "grad_norm": 2.2386324405670166, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7108924388885498, + "num_tokens": 310385661.0, + "step": 12096 + }, + { + "epoch": 1.328464748517461, + "grad_norm": 2.113684892654419, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6906077861785889, + "num_tokens": 310415574.0, + "step": 12097 + }, + { + "epoch": 1.3285745662200747, + "grad_norm": 2.4241340160369873, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7239748239517212, + "num_tokens": 310438317.0, + "step": 12098 + }, + { + "epoch": 1.3286843839226883, + "grad_norm": 2.2920515537261963, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7375495433807373, + "num_tokens": 310462138.0, + "step": 12099 + }, + { + "epoch": 1.328794201625302, + "grad_norm": 2.536433458328247, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7187917232513428, + "num_tokens": 310481889.0, + "step": 12100 + }, + { + "epoch": 1.3289040193279158, + "grad_norm": 2.271250009536743, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7123385667800903, + "num_tokens": 310506740.0, + "step": 12101 + }, + { + "epoch": 1.3290138370305293, + "grad_norm": 2.244253635406494, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7140071392059326, + "num_tokens": 310531723.0, + "step": 12102 + }, + { + "epoch": 1.3291236547331429, + "grad_norm": 2.013439416885376, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7122547030448914, + "num_tokens": 310562089.0, + "step": 12103 + }, + { + "epoch": 1.3292334724357566, + "grad_norm": 2.467827558517456, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7096034288406372, + "num_tokens": 310584937.0, + "step": 12104 + }, + { + "epoch": 1.3293432901383704, + "grad_norm": 2.177912712097168, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7249683141708374, + "num_tokens": 310612166.0, + "step": 12105 + }, + { + "epoch": 1.329453107840984, + "grad_norm": 2.3218913078308105, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7176687717437744, + "num_tokens": 310636485.0, + "step": 12106 + }, + { + "epoch": 1.3295629255435977, + "grad_norm": 2.073996067047119, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7177512049674988, + "num_tokens": 310666002.0, + "step": 12107 + }, + { + "epoch": 1.3296727432462112, + "grad_norm": 2.3911285400390625, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7174286842346191, + "num_tokens": 310689914.0, + "step": 12108 + }, + { + "epoch": 1.329782560948825, + "grad_norm": 2.161458969116211, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.715356707572937, + "num_tokens": 310717458.0, + "step": 12109 + }, + { + "epoch": 1.3298923786514387, + "grad_norm": 2.6028549671173096, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7269911170005798, + "num_tokens": 310736571.0, + "step": 12110 + }, + { + "epoch": 1.3300021963540523, + "grad_norm": 2.5072319507598877, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7235110998153687, + "num_tokens": 310757353.0, + "step": 12111 + }, + { + "epoch": 1.330112014056666, + "grad_norm": 2.343020439147949, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7267504930496216, + "num_tokens": 310780633.0, + "step": 12112 + }, + { + "epoch": 1.3302218317592795, + "grad_norm": 2.3808741569519043, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7123767137527466, + "num_tokens": 310805372.0, + "step": 12113 + }, + { + "epoch": 1.3303316494618933, + "grad_norm": 2.5394840240478516, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7039415836334229, + "num_tokens": 310828937.0, + "step": 12114 + }, + { + "epoch": 1.3304414671645068, + "grad_norm": 2.307356119155884, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7251399159431458, + "num_tokens": 310852992.0, + "step": 12115 + }, + { + "epoch": 1.3305512848671206, + "grad_norm": 2.414158344268799, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7043352723121643, + "num_tokens": 310878746.0, + "step": 12116 + }, + { + "epoch": 1.3306611025697341, + "grad_norm": 2.210080146789551, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7055039405822754, + "num_tokens": 310904541.0, + "step": 12117 + }, + { + "epoch": 1.330770920272348, + "grad_norm": 2.1942083835601807, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7155510187149048, + "num_tokens": 310932001.0, + "step": 12118 + }, + { + "epoch": 1.3308807379749616, + "grad_norm": 2.5525786876678467, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7354397773742676, + "num_tokens": 310951934.0, + "step": 12119 + }, + { + "epoch": 1.3309905556775752, + "grad_norm": 2.357142448425293, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6961851119995117, + "num_tokens": 310977496.0, + "step": 12120 + }, + { + "epoch": 1.331100373380189, + "grad_norm": 2.033576488494873, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7086039781570435, + "num_tokens": 311007635.0, + "step": 12121 + }, + { + "epoch": 1.3312101910828025, + "grad_norm": 2.354818820953369, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7197830677032471, + "num_tokens": 311030445.0, + "step": 12122 + }, + { + "epoch": 1.3313200087854162, + "grad_norm": 2.2854881286621094, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7331537008285522, + "num_tokens": 311057895.0, + "step": 12123 + }, + { + "epoch": 1.33142982648803, + "grad_norm": 2.059621810913086, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7161298990249634, + "num_tokens": 311086012.0, + "step": 12124 + }, + { + "epoch": 1.3315396441906435, + "grad_norm": 2.1399543285369873, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7015950083732605, + "num_tokens": 311115045.0, + "step": 12125 + }, + { + "epoch": 1.331649461893257, + "grad_norm": 2.5066351890563965, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7196122407913208, + "num_tokens": 311136559.0, + "step": 12126 + }, + { + "epoch": 1.3317592795958708, + "grad_norm": 1.8587839603424072, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7063875198364258, + "num_tokens": 311169648.0, + "step": 12127 + }, + { + "epoch": 1.3318690972984846, + "grad_norm": 1.9639562368392944, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7222509384155273, + "num_tokens": 311201095.0, + "step": 12128 + }, + { + "epoch": 1.331978915001098, + "grad_norm": 2.1287882328033447, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7143324613571167, + "num_tokens": 311228490.0, + "step": 12129 + }, + { + "epoch": 1.3320887327037119, + "grad_norm": 2.2616758346557617, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7143514156341553, + "num_tokens": 311255661.0, + "step": 12130 + }, + { + "epoch": 1.3321985504063254, + "grad_norm": 2.147700786590576, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7139244675636292, + "num_tokens": 311284707.0, + "step": 12131 + }, + { + "epoch": 1.3323083681089392, + "grad_norm": 1.9693406820297241, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7113521099090576, + "num_tokens": 311318483.0, + "step": 12132 + }, + { + "epoch": 1.332418185811553, + "grad_norm": 2.1624643802642822, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7085755467414856, + "num_tokens": 311346856.0, + "step": 12133 + }, + { + "epoch": 1.3325280035141664, + "grad_norm": 2.0766029357910156, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7046011090278625, + "num_tokens": 311377828.0, + "step": 12134 + }, + { + "epoch": 1.3326378212167802, + "grad_norm": 2.3671200275421143, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6923221349716187, + "num_tokens": 311403079.0, + "step": 12135 + }, + { + "epoch": 1.3327476389193937, + "grad_norm": 2.1031124591827393, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7061612606048584, + "num_tokens": 311433564.0, + "step": 12136 + }, + { + "epoch": 1.3328574566220075, + "grad_norm": 2.165564775466919, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7128874063491821, + "num_tokens": 311462198.0, + "step": 12137 + }, + { + "epoch": 1.3329672743246213, + "grad_norm": 2.3963141441345215, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7338600158691406, + "num_tokens": 311484835.0, + "step": 12138 + }, + { + "epoch": 1.3330770920272348, + "grad_norm": 2.4043772220611572, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7220691442489624, + "num_tokens": 311507216.0, + "step": 12139 + }, + { + "epoch": 1.3331869097298483, + "grad_norm": 1.984772801399231, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7388115525245667, + "num_tokens": 311537942.0, + "step": 12140 + }, + { + "epoch": 1.333296727432462, + "grad_norm": 2.042370557785034, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.709827184677124, + "num_tokens": 311567545.0, + "step": 12141 + }, + { + "epoch": 1.3334065451350758, + "grad_norm": 2.1338255405426025, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7259624600410461, + "num_tokens": 311595323.0, + "step": 12142 + }, + { + "epoch": 1.3335163628376894, + "grad_norm": 2.4331183433532715, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7032486200332642, + "num_tokens": 311617783.0, + "step": 12143 + }, + { + "epoch": 1.3336261805403031, + "grad_norm": 2.186222553253174, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7469816207885742, + "num_tokens": 311642473.0, + "step": 12144 + }, + { + "epoch": 1.3337359982429167, + "grad_norm": 2.137134075164795, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7228271961212158, + "num_tokens": 311669999.0, + "step": 12145 + }, + { + "epoch": 1.3338458159455304, + "grad_norm": 2.179210662841797, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7041362524032593, + "num_tokens": 311697295.0, + "step": 12146 + }, + { + "epoch": 1.3339556336481442, + "grad_norm": 2.1383349895477295, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7112611532211304, + "num_tokens": 311724490.0, + "step": 12147 + }, + { + "epoch": 1.3340654513507577, + "grad_norm": 2.4436280727386475, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.737378716468811, + "num_tokens": 311746969.0, + "step": 12148 + }, + { + "epoch": 1.3341752690533715, + "grad_norm": 2.0688605308532715, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6896691918373108, + "num_tokens": 311775080.0, + "step": 12149 + }, + { + "epoch": 1.334285086755985, + "grad_norm": 2.0916435718536377, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6918376684188843, + "num_tokens": 311803854.0, + "step": 12150 + }, + { + "epoch": 1.3343949044585988, + "grad_norm": 2.247122287750244, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7265246510505676, + "num_tokens": 311828728.0, + "step": 12151 + }, + { + "epoch": 1.3345047221612125, + "grad_norm": 2.07470440864563, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.7067866325378418, + "num_tokens": 311856505.0, + "step": 12152 + }, + { + "epoch": 1.334614539863826, + "grad_norm": 2.5455923080444336, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7405992746353149, + "num_tokens": 311876666.0, + "step": 12153 + }, + { + "epoch": 1.3347243575664396, + "grad_norm": 2.377809762954712, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7171501517295837, + "num_tokens": 311901697.0, + "step": 12154 + }, + { + "epoch": 1.3348341752690533, + "grad_norm": 2.4707589149475098, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.726162314414978, + "num_tokens": 311923552.0, + "step": 12155 + }, + { + "epoch": 1.334943992971667, + "grad_norm": 2.578744411468506, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7111310362815857, + "num_tokens": 311944275.0, + "step": 12156 + }, + { + "epoch": 1.3350538106742806, + "grad_norm": 2.153320550918579, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7372092008590698, + "num_tokens": 311969942.0, + "step": 12157 + }, + { + "epoch": 1.3351636283768944, + "grad_norm": 2.3453173637390137, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7313075065612793, + "num_tokens": 311992057.0, + "step": 12158 + }, + { + "epoch": 1.335273446079508, + "grad_norm": 2.3134231567382812, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6908881664276123, + "num_tokens": 312017101.0, + "step": 12159 + }, + { + "epoch": 1.3353832637821217, + "grad_norm": 2.4674391746520996, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7059770822525024, + "num_tokens": 312038951.0, + "step": 12160 + }, + { + "epoch": 1.3354930814847354, + "grad_norm": 2.227820873260498, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7201730012893677, + "num_tokens": 312064219.0, + "step": 12161 + }, + { + "epoch": 1.335602899187349, + "grad_norm": 2.1453495025634766, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7091848254203796, + "num_tokens": 312093992.0, + "step": 12162 + }, + { + "epoch": 1.3357127168899627, + "grad_norm": 2.3464224338531494, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7176584005355835, + "num_tokens": 312120220.0, + "step": 12163 + }, + { + "epoch": 1.3358225345925763, + "grad_norm": 2.272836685180664, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7130062580108643, + "num_tokens": 312146617.0, + "step": 12164 + }, + { + "epoch": 1.33593235229519, + "grad_norm": 2.251667022705078, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7166000604629517, + "num_tokens": 312172412.0, + "step": 12165 + }, + { + "epoch": 1.3360421699978036, + "grad_norm": 1.9343955516815186, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7176135778427124, + "num_tokens": 312206335.0, + "step": 12166 + }, + { + "epoch": 1.3361519877004173, + "grad_norm": 1.9400503635406494, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6986293196678162, + "num_tokens": 312238319.0, + "step": 12167 + }, + { + "epoch": 1.3362618054030309, + "grad_norm": 2.0097196102142334, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7133296132087708, + "num_tokens": 312269463.0, + "step": 12168 + }, + { + "epoch": 1.3363716231056446, + "grad_norm": 2.58699369430542, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7287477254867554, + "num_tokens": 312290835.0, + "step": 12169 + }, + { + "epoch": 1.3364814408082584, + "grad_norm": 2.0346450805664062, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6918389797210693, + "num_tokens": 312323149.0, + "step": 12170 + }, + { + "epoch": 1.336591258510872, + "grad_norm": 1.996983528137207, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6983422040939331, + "num_tokens": 312355141.0, + "step": 12171 + }, + { + "epoch": 1.3367010762134857, + "grad_norm": 2.651580810546875, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7286404371261597, + "num_tokens": 312375530.0, + "step": 12172 + }, + { + "epoch": 1.3368108939160992, + "grad_norm": 1.9705482721328735, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6978107690811157, + "num_tokens": 312408307.0, + "step": 12173 + }, + { + "epoch": 1.336920711618713, + "grad_norm": 2.100100040435791, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7117588520050049, + "num_tokens": 312437476.0, + "step": 12174 + }, + { + "epoch": 1.3370305293213267, + "grad_norm": 2.6371185779571533, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7329134345054626, + "num_tokens": 312456454.0, + "step": 12175 + }, + { + "epoch": 1.3371403470239402, + "grad_norm": 2.265552282333374, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7245951890945435, + "num_tokens": 312480340.0, + "step": 12176 + }, + { + "epoch": 1.337250164726554, + "grad_norm": 2.0215225219726562, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7374846339225769, + "num_tokens": 312506392.0, + "step": 12177 + }, + { + "epoch": 1.3373599824291675, + "grad_norm": 2.3245325088500977, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.728498637676239, + "num_tokens": 312530963.0, + "step": 12178 + }, + { + "epoch": 1.3374698001317813, + "grad_norm": 2.2293622493743896, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7151522040367126, + "num_tokens": 312555489.0, + "step": 12179 + }, + { + "epoch": 1.3375796178343948, + "grad_norm": 1.9092258214950562, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.705669105052948, + "num_tokens": 312589328.0, + "step": 12180 + }, + { + "epoch": 1.3376894355370086, + "grad_norm": 2.0763540267944336, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7340225577354431, + "num_tokens": 312616925.0, + "step": 12181 + }, + { + "epoch": 1.3377992532396221, + "grad_norm": 2.317077398300171, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7082042098045349, + "num_tokens": 312641034.0, + "step": 12182 + }, + { + "epoch": 1.3379090709422359, + "grad_norm": 2.4516589641571045, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.733060896396637, + "num_tokens": 312663275.0, + "step": 12183 + }, + { + "epoch": 1.3380188886448496, + "grad_norm": 2.2507026195526123, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7193434834480286, + "num_tokens": 312687068.0, + "step": 12184 + }, + { + "epoch": 1.3381287063474632, + "grad_norm": 2.1141088008880615, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7085158824920654, + "num_tokens": 312715201.0, + "step": 12185 + }, + { + "epoch": 1.338238524050077, + "grad_norm": 2.518372058868408, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7194869518280029, + "num_tokens": 312736092.0, + "step": 12186 + }, + { + "epoch": 1.3383483417526905, + "grad_norm": 2.2090084552764893, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.726325511932373, + "num_tokens": 312763449.0, + "step": 12187 + }, + { + "epoch": 1.3384581594553042, + "grad_norm": 2.067963123321533, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7220484018325806, + "num_tokens": 312790460.0, + "step": 12188 + }, + { + "epoch": 1.338567977157918, + "grad_norm": 2.2171268463134766, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7594159841537476, + "num_tokens": 312816090.0, + "step": 12189 + }, + { + "epoch": 1.3386777948605315, + "grad_norm": 2.2113096714019775, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7195260524749756, + "num_tokens": 312841836.0, + "step": 12190 + }, + { + "epoch": 1.338787612563145, + "grad_norm": 2.538182497024536, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7217923402786255, + "num_tokens": 312862565.0, + "step": 12191 + }, + { + "epoch": 1.3388974302657588, + "grad_norm": 2.1965110301971436, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7211219072341919, + "num_tokens": 312889190.0, + "step": 12192 + }, + { + "epoch": 1.3390072479683726, + "grad_norm": 1.9386729001998901, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7268229722976685, + "num_tokens": 312923835.0, + "step": 12193 + }, + { + "epoch": 1.339117065670986, + "grad_norm": 2.2786285877227783, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7289854288101196, + "num_tokens": 312947915.0, + "step": 12194 + }, + { + "epoch": 1.3392268833735999, + "grad_norm": 2.2099685668945312, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7224207520484924, + "num_tokens": 312973720.0, + "step": 12195 + }, + { + "epoch": 1.3393367010762134, + "grad_norm": 2.362545967102051, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.705752968788147, + "num_tokens": 312997418.0, + "step": 12196 + }, + { + "epoch": 1.3394465187788271, + "grad_norm": 2.3255388736724854, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.732180118560791, + "num_tokens": 313022738.0, + "step": 12197 + }, + { + "epoch": 1.339556336481441, + "grad_norm": 2.5710291862487793, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7314335107803345, + "num_tokens": 313042602.0, + "step": 12198 + }, + { + "epoch": 1.3396661541840544, + "grad_norm": 1.958134651184082, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6878253817558289, + "num_tokens": 313075666.0, + "step": 12199 + }, + { + "epoch": 1.3397759718866682, + "grad_norm": 2.220689296722412, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.715462863445282, + "num_tokens": 313102091.0, + "step": 12200 + }, + { + "epoch": 1.3398857895892817, + "grad_norm": 2.135970115661621, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7153923511505127, + "num_tokens": 313130471.0, + "step": 12201 + }, + { + "epoch": 1.3399956072918955, + "grad_norm": 2.048243522644043, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7076036930084229, + "num_tokens": 313162631.0, + "step": 12202 + }, + { + "epoch": 1.3401054249945092, + "grad_norm": 2.5944221019744873, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7544665336608887, + "num_tokens": 313182365.0, + "step": 12203 + }, + { + "epoch": 1.3402152426971228, + "grad_norm": 2.291102170944214, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7189009189605713, + "num_tokens": 313207725.0, + "step": 12204 + }, + { + "epoch": 1.3403250603997363, + "grad_norm": 2.2955551147460938, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7137652635574341, + "num_tokens": 313232643.0, + "step": 12205 + }, + { + "epoch": 1.34043487810235, + "grad_norm": 2.327786684036255, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7203090190887451, + "num_tokens": 313256373.0, + "step": 12206 + }, + { + "epoch": 1.3405446958049638, + "grad_norm": 2.230501890182495, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.707363486289978, + "num_tokens": 313283094.0, + "step": 12207 + }, + { + "epoch": 1.3406545135075774, + "grad_norm": 2.2106947898864746, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.745697557926178, + "num_tokens": 313310765.0, + "step": 12208 + }, + { + "epoch": 1.3407643312101911, + "grad_norm": 2.3862173557281494, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7495097517967224, + "num_tokens": 313333346.0, + "step": 12209 + }, + { + "epoch": 1.3408741489128047, + "grad_norm": 2.431791067123413, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7235785722732544, + "num_tokens": 313355752.0, + "step": 12210 + }, + { + "epoch": 1.3409839666154184, + "grad_norm": 2.0353047847747803, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7168852090835571, + "num_tokens": 313383282.0, + "step": 12211 + }, + { + "epoch": 1.3410937843180322, + "grad_norm": 2.3697497844696045, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.731229841709137, + "num_tokens": 313406484.0, + "step": 12212 + }, + { + "epoch": 1.3412036020206457, + "grad_norm": 2.328737735748291, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7371610403060913, + "num_tokens": 313429491.0, + "step": 12213 + }, + { + "epoch": 1.3413134197232595, + "grad_norm": 2.306765079498291, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7457457780838013, + "num_tokens": 313453683.0, + "step": 12214 + }, + { + "epoch": 1.341423237425873, + "grad_norm": 1.9689275026321411, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6977432370185852, + "num_tokens": 313484103.0, + "step": 12215 + }, + { + "epoch": 1.3415330551284868, + "grad_norm": 2.233771562576294, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7155017256736755, + "num_tokens": 313508898.0, + "step": 12216 + }, + { + "epoch": 1.3416428728311005, + "grad_norm": 2.103337049484253, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.704715371131897, + "num_tokens": 313537843.0, + "step": 12217 + }, + { + "epoch": 1.341752690533714, + "grad_norm": 2.7650985717773438, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7261717319488525, + "num_tokens": 313555715.0, + "step": 12218 + }, + { + "epoch": 1.3418625082363276, + "grad_norm": 2.1137313842773438, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7237651348114014, + "num_tokens": 313583291.0, + "step": 12219 + }, + { + "epoch": 1.3419723259389413, + "grad_norm": 2.3949317932128906, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7088658809661865, + "num_tokens": 313606443.0, + "step": 12220 + }, + { + "epoch": 1.342082143641555, + "grad_norm": 2.399808645248413, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7154854536056519, + "num_tokens": 313629871.0, + "step": 12221 + }, + { + "epoch": 1.3421919613441686, + "grad_norm": 1.9369081258773804, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.692992091178894, + "num_tokens": 313662681.0, + "step": 12222 + }, + { + "epoch": 1.3423017790467824, + "grad_norm": 2.253425121307373, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.741956353187561, + "num_tokens": 313686938.0, + "step": 12223 + }, + { + "epoch": 1.342411596749396, + "grad_norm": 2.1291768550872803, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7296421527862549, + "num_tokens": 313712276.0, + "step": 12224 + }, + { + "epoch": 1.3425214144520097, + "grad_norm": 2.319849729537964, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.73985755443573, + "num_tokens": 313737854.0, + "step": 12225 + }, + { + "epoch": 1.3426312321546234, + "grad_norm": 2.4068830013275146, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7190999388694763, + "num_tokens": 313762531.0, + "step": 12226 + }, + { + "epoch": 1.342741049857237, + "grad_norm": 2.2973549365997314, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7175784111022949, + "num_tokens": 313787628.0, + "step": 12227 + }, + { + "epoch": 1.3428508675598507, + "grad_norm": 2.518580675125122, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7116561532020569, + "num_tokens": 313809069.0, + "step": 12228 + }, + { + "epoch": 1.3429606852624643, + "grad_norm": 2.211324691772461, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7222583293914795, + "num_tokens": 313838309.0, + "step": 12229 + }, + { + "epoch": 1.343070502965078, + "grad_norm": 2.3268351554870605, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7239594459533691, + "num_tokens": 313861487.0, + "step": 12230 + }, + { + "epoch": 1.3431803206676916, + "grad_norm": 2.0414247512817383, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7214114665985107, + "num_tokens": 313891098.0, + "step": 12231 + }, + { + "epoch": 1.3432901383703053, + "grad_norm": 2.4092493057250977, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7324330806732178, + "num_tokens": 313913325.0, + "step": 12232 + }, + { + "epoch": 1.3433999560729188, + "grad_norm": 2.1102845668792725, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7119603753089905, + "num_tokens": 313943353.0, + "step": 12233 + }, + { + "epoch": 1.3435097737755326, + "grad_norm": 2.059523344039917, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7004407644271851, + "num_tokens": 313971655.0, + "step": 12234 + }, + { + "epoch": 1.3436195914781464, + "grad_norm": 1.9716581106185913, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7123438119888306, + "num_tokens": 314002094.0, + "step": 12235 + }, + { + "epoch": 1.34372940918076, + "grad_norm": 2.1999552249908447, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7114107608795166, + "num_tokens": 314028435.0, + "step": 12236 + }, + { + "epoch": 1.3438392268833736, + "grad_norm": 2.371229648590088, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7293378114700317, + "num_tokens": 314051728.0, + "step": 12237 + }, + { + "epoch": 1.3439490445859872, + "grad_norm": 2.2807247638702393, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6978241801261902, + "num_tokens": 314079386.0, + "step": 12238 + }, + { + "epoch": 1.344058862288601, + "grad_norm": 1.9516288042068481, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7105975151062012, + "num_tokens": 314113277.0, + "step": 12239 + }, + { + "epoch": 1.3441686799912147, + "grad_norm": 2.164271593093872, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7179391384124756, + "num_tokens": 314141117.0, + "step": 12240 + }, + { + "epoch": 1.3442784976938282, + "grad_norm": 2.1747117042541504, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6912716627120972, + "num_tokens": 314168840.0, + "step": 12241 + }, + { + "epoch": 1.3443883153964418, + "grad_norm": 2.255708694458008, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7101174592971802, + "num_tokens": 314193431.0, + "step": 12242 + }, + { + "epoch": 1.3444981330990555, + "grad_norm": 2.5536534786224365, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7515783905982971, + "num_tokens": 314213330.0, + "step": 12243 + }, + { + "epoch": 1.3446079508016693, + "grad_norm": 2.0154905319213867, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7011520862579346, + "num_tokens": 314246301.0, + "step": 12244 + }, + { + "epoch": 1.3447177685042828, + "grad_norm": 2.2190093994140625, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7165147066116333, + "num_tokens": 314272876.0, + "step": 12245 + }, + { + "epoch": 1.3448275862068966, + "grad_norm": 2.3051414489746094, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7359818816184998, + "num_tokens": 314295376.0, + "step": 12246 + }, + { + "epoch": 1.34493740390951, + "grad_norm": 2.2588517665863037, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7112685441970825, + "num_tokens": 314321520.0, + "step": 12247 + }, + { + "epoch": 1.3450472216121239, + "grad_norm": 2.376864433288574, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7001162767410278, + "num_tokens": 314347245.0, + "step": 12248 + }, + { + "epoch": 1.3451570393147376, + "grad_norm": 2.0992164611816406, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6869639158248901, + "num_tokens": 314377911.0, + "step": 12249 + }, + { + "epoch": 1.3452668570173512, + "grad_norm": 2.15507173538208, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.703951358795166, + "num_tokens": 314408866.0, + "step": 12250 + }, + { + "epoch": 1.345376674719965, + "grad_norm": 2.623623847961426, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7380494475364685, + "num_tokens": 314429635.0, + "step": 12251 + }, + { + "epoch": 1.3454864924225785, + "grad_norm": 2.399341583251953, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.717915952205658, + "num_tokens": 314454456.0, + "step": 12252 + }, + { + "epoch": 1.3455963101251922, + "grad_norm": 2.188913583755493, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7227322459220886, + "num_tokens": 314481124.0, + "step": 12253 + }, + { + "epoch": 1.345706127827806, + "grad_norm": 2.376157760620117, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7138970494270325, + "num_tokens": 314503302.0, + "step": 12254 + }, + { + "epoch": 1.3458159455304195, + "grad_norm": 2.4316985607147217, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.74461430311203, + "num_tokens": 314524240.0, + "step": 12255 + }, + { + "epoch": 1.345925763233033, + "grad_norm": 2.3575191497802734, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7311323881149292, + "num_tokens": 314547489.0, + "step": 12256 + }, + { + "epoch": 1.3460355809356468, + "grad_norm": 2.3288729190826416, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.716454267501831, + "num_tokens": 314571002.0, + "step": 12257 + }, + { + "epoch": 1.3461453986382605, + "grad_norm": 2.2236382961273193, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7144743204116821, + "num_tokens": 314599654.0, + "step": 12258 + }, + { + "epoch": 1.346255216340874, + "grad_norm": 2.1774792671203613, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7240585684776306, + "num_tokens": 314625814.0, + "step": 12259 + }, + { + "epoch": 1.3463650340434878, + "grad_norm": 2.324334144592285, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.737829327583313, + "num_tokens": 314649622.0, + "step": 12260 + }, + { + "epoch": 1.3464748517461014, + "grad_norm": 2.6386609077453613, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7268186807632446, + "num_tokens": 314670426.0, + "step": 12261 + }, + { + "epoch": 1.3465846694487151, + "grad_norm": 2.242863893508911, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.722665011882782, + "num_tokens": 314696547.0, + "step": 12262 + }, + { + "epoch": 1.346694487151329, + "grad_norm": 2.0511527061462402, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7074511051177979, + "num_tokens": 314726587.0, + "step": 12263 + }, + { + "epoch": 1.3468043048539424, + "grad_norm": 2.2416419982910156, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.702447772026062, + "num_tokens": 314754712.0, + "step": 12264 + }, + { + "epoch": 1.3469141225565562, + "grad_norm": 2.1522464752197266, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7273483872413635, + "num_tokens": 314783108.0, + "step": 12265 + }, + { + "epoch": 1.3470239402591697, + "grad_norm": 2.3172733783721924, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7180184125900269, + "num_tokens": 314806579.0, + "step": 12266 + }, + { + "epoch": 1.3471337579617835, + "grad_norm": 2.036801815032959, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6943256258964539, + "num_tokens": 314836193.0, + "step": 12267 + }, + { + "epoch": 1.3472435756643972, + "grad_norm": 2.226837635040283, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6883295178413391, + "num_tokens": 314865434.0, + "step": 12268 + }, + { + "epoch": 1.3473533933670108, + "grad_norm": 2.4924793243408203, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7016034126281738, + "num_tokens": 314886434.0, + "step": 12269 + }, + { + "epoch": 1.3474632110696243, + "grad_norm": 2.2164692878723145, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7010579109191895, + "num_tokens": 314912652.0, + "step": 12270 + }, + { + "epoch": 1.347573028772238, + "grad_norm": 1.9048144817352295, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7354521751403809, + "num_tokens": 314944786.0, + "step": 12271 + }, + { + "epoch": 1.3476828464748518, + "grad_norm": 2.3246288299560547, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7201804518699646, + "num_tokens": 314970399.0, + "step": 12272 + }, + { + "epoch": 1.3477926641774653, + "grad_norm": 2.2357261180877686, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.6992199420928955, + "num_tokens": 314996171.0, + "step": 12273 + }, + { + "epoch": 1.347902481880079, + "grad_norm": 2.357403039932251, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7034395933151245, + "num_tokens": 315020214.0, + "step": 12274 + }, + { + "epoch": 1.3480122995826926, + "grad_norm": 2.0918235778808594, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7090799808502197, + "num_tokens": 315050483.0, + "step": 12275 + }, + { + "epoch": 1.3481221172853064, + "grad_norm": 2.35772442817688, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.730435848236084, + "num_tokens": 315079853.0, + "step": 12276 + }, + { + "epoch": 1.3482319349879202, + "grad_norm": 2.304617404937744, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.721135675907135, + "num_tokens": 315104396.0, + "step": 12277 + }, + { + "epoch": 1.3483417526905337, + "grad_norm": 2.1604092121124268, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7026361227035522, + "num_tokens": 315134309.0, + "step": 12278 + }, + { + "epoch": 1.3484515703931474, + "grad_norm": 1.965044379234314, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6927914619445801, + "num_tokens": 315170544.0, + "step": 12279 + }, + { + "epoch": 1.348561388095761, + "grad_norm": 2.0601377487182617, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6980450749397278, + "num_tokens": 315201382.0, + "step": 12280 + }, + { + "epoch": 1.3486712057983747, + "grad_norm": 2.4570887088775635, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7269087433815002, + "num_tokens": 315222045.0, + "step": 12281 + }, + { + "epoch": 1.3487810235009885, + "grad_norm": 2.177762508392334, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7203593850135803, + "num_tokens": 315247464.0, + "step": 12282 + }, + { + "epoch": 1.348890841203602, + "grad_norm": 2.0311086177825928, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7369646430015564, + "num_tokens": 315276402.0, + "step": 12283 + }, + { + "epoch": 1.3490006589062156, + "grad_norm": 2.3058273792266846, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7216488122940063, + "num_tokens": 315299615.0, + "step": 12284 + }, + { + "epoch": 1.3491104766088293, + "grad_norm": 2.086693286895752, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7034233808517456, + "num_tokens": 315328024.0, + "step": 12285 + }, + { + "epoch": 1.349220294311443, + "grad_norm": 2.338259696960449, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.704416036605835, + "num_tokens": 315353104.0, + "step": 12286 + }, + { + "epoch": 1.3493301120140566, + "grad_norm": 2.4047718048095703, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7356247305870056, + "num_tokens": 315375241.0, + "step": 12287 + }, + { + "epoch": 1.3494399297166704, + "grad_norm": 2.396730422973633, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7420319318771362, + "num_tokens": 315396281.0, + "step": 12288 + }, + { + "epoch": 1.349549747419284, + "grad_norm": 2.133131980895996, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7363275289535522, + "num_tokens": 315421318.0, + "step": 12289 + }, + { + "epoch": 1.3496595651218977, + "grad_norm": 2.1867904663085938, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7076800465583801, + "num_tokens": 315447182.0, + "step": 12290 + }, + { + "epoch": 1.3497693828245114, + "grad_norm": 2.335555076599121, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7219520807266235, + "num_tokens": 315470952.0, + "step": 12291 + }, + { + "epoch": 1.349879200527125, + "grad_norm": 2.089829921722412, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7218427062034607, + "num_tokens": 315499026.0, + "step": 12292 + }, + { + "epoch": 1.3499890182297387, + "grad_norm": 2.4055097103118896, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7113907933235168, + "num_tokens": 315522719.0, + "step": 12293 + }, + { + "epoch": 1.3500988359323522, + "grad_norm": 2.3256075382232666, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7167313098907471, + "num_tokens": 315546814.0, + "step": 12294 + }, + { + "epoch": 1.350208653634966, + "grad_norm": 2.3953349590301514, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7186065912246704, + "num_tokens": 315568786.0, + "step": 12295 + }, + { + "epoch": 1.3503184713375795, + "grad_norm": 2.2598443031311035, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7302147150039673, + "num_tokens": 315595436.0, + "step": 12296 + }, + { + "epoch": 1.3504282890401933, + "grad_norm": 2.2808544635772705, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7182676792144775, + "num_tokens": 315623477.0, + "step": 12297 + }, + { + "epoch": 1.3505381067428068, + "grad_norm": 1.9659878015518188, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7219409942626953, + "num_tokens": 315656063.0, + "step": 12298 + }, + { + "epoch": 1.3506479244454206, + "grad_norm": 2.1190621852874756, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7047547698020935, + "num_tokens": 315682945.0, + "step": 12299 + }, + { + "epoch": 1.3507577421480343, + "grad_norm": 2.2897658348083496, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.732642650604248, + "num_tokens": 315707034.0, + "step": 12300 + }, + { + "epoch": 1.3508675598506479, + "grad_norm": 2.1205809116363525, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6908398866653442, + "num_tokens": 315735624.0, + "step": 12301 + }, + { + "epoch": 1.3509773775532616, + "grad_norm": 2.5509066581726074, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.725174069404602, + "num_tokens": 315755215.0, + "step": 12302 + }, + { + "epoch": 1.3510871952558752, + "grad_norm": 2.155832290649414, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7180294394493103, + "num_tokens": 315782414.0, + "step": 12303 + }, + { + "epoch": 1.351197012958489, + "grad_norm": 2.338327407836914, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7095190286636353, + "num_tokens": 315807331.0, + "step": 12304 + }, + { + "epoch": 1.3513068306611027, + "grad_norm": 2.4006264209747314, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.6999382376670837, + "num_tokens": 315831159.0, + "step": 12305 + }, + { + "epoch": 1.3514166483637162, + "grad_norm": 2.812821626663208, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7536384463310242, + "num_tokens": 315848068.0, + "step": 12306 + }, + { + "epoch": 1.3515264660663298, + "grad_norm": 2.1812806129455566, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.686618983745575, + "num_tokens": 315874817.0, + "step": 12307 + }, + { + "epoch": 1.3516362837689435, + "grad_norm": 2.507211685180664, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7337741255760193, + "num_tokens": 315894990.0, + "step": 12308 + }, + { + "epoch": 1.3517461014715573, + "grad_norm": 2.2140109539031982, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7135213017463684, + "num_tokens": 315922424.0, + "step": 12309 + }, + { + "epoch": 1.3518559191741708, + "grad_norm": 2.4279651641845703, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7125036120414734, + "num_tokens": 315945388.0, + "step": 12310 + }, + { + "epoch": 1.3519657368767846, + "grad_norm": 2.3235127925872803, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7258561849594116, + "num_tokens": 315970139.0, + "step": 12311 + }, + { + "epoch": 1.352075554579398, + "grad_norm": 2.2916369438171387, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7229584455490112, + "num_tokens": 315993895.0, + "step": 12312 + }, + { + "epoch": 1.3521853722820119, + "grad_norm": 2.2371439933776855, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7179834842681885, + "num_tokens": 316019196.0, + "step": 12313 + }, + { + "epoch": 1.3522951899846256, + "grad_norm": 2.064892292022705, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7291555404663086, + "num_tokens": 316047593.0, + "step": 12314 + }, + { + "epoch": 1.3524050076872391, + "grad_norm": 2.3802027702331543, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6987334489822388, + "num_tokens": 316074325.0, + "step": 12315 + }, + { + "epoch": 1.352514825389853, + "grad_norm": 2.2471718788146973, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7372440099716187, + "num_tokens": 316101931.0, + "step": 12316 + }, + { + "epoch": 1.3526246430924664, + "grad_norm": 2.2522292137145996, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7221872806549072, + "num_tokens": 316127973.0, + "step": 12317 + }, + { + "epoch": 1.3527344607950802, + "grad_norm": 2.202296018600464, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7134793996810913, + "num_tokens": 316155768.0, + "step": 12318 + }, + { + "epoch": 1.352844278497694, + "grad_norm": 2.2073023319244385, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7123920917510986, + "num_tokens": 316181264.0, + "step": 12319 + }, + { + "epoch": 1.3529540962003075, + "grad_norm": 2.148521900177002, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7063177824020386, + "num_tokens": 316209038.0, + "step": 12320 + }, + { + "epoch": 1.353063913902921, + "grad_norm": 2.306055784225464, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7234916687011719, + "num_tokens": 316232963.0, + "step": 12321 + }, + { + "epoch": 1.3531737316055348, + "grad_norm": 2.075002431869507, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7227635383605957, + "num_tokens": 316263072.0, + "step": 12322 + }, + { + "epoch": 1.3532835493081485, + "grad_norm": 2.380797863006592, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7389222383499146, + "num_tokens": 316285857.0, + "step": 12323 + }, + { + "epoch": 1.353393367010762, + "grad_norm": 2.311389207839966, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7310045957565308, + "num_tokens": 316308811.0, + "step": 12324 + }, + { + "epoch": 1.3535031847133758, + "grad_norm": 2.2018425464630127, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7106150388717651, + "num_tokens": 316333918.0, + "step": 12325 + }, + { + "epoch": 1.3536130024159894, + "grad_norm": 2.535466194152832, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7228033542633057, + "num_tokens": 316356990.0, + "step": 12326 + }, + { + "epoch": 1.3537228201186031, + "grad_norm": 2.1589269638061523, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7237091064453125, + "num_tokens": 316383091.0, + "step": 12327 + }, + { + "epoch": 1.3538326378212169, + "grad_norm": 2.314612865447998, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7145658731460571, + "num_tokens": 316407316.0, + "step": 12328 + }, + { + "epoch": 1.3539424555238304, + "grad_norm": 2.2381503582000732, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7211413979530334, + "num_tokens": 316433455.0, + "step": 12329 + }, + { + "epoch": 1.3540522732264442, + "grad_norm": 2.2072582244873047, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7294674515724182, + "num_tokens": 316458980.0, + "step": 12330 + }, + { + "epoch": 1.3541620909290577, + "grad_norm": 2.1530873775482178, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7143104672431946, + "num_tokens": 316487085.0, + "step": 12331 + }, + { + "epoch": 1.3542719086316715, + "grad_norm": 2.3732805252075195, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.719506025314331, + "num_tokens": 316511102.0, + "step": 12332 + }, + { + "epoch": 1.3543817263342852, + "grad_norm": 2.321573495864868, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7424816489219666, + "num_tokens": 316533845.0, + "step": 12333 + }, + { + "epoch": 1.3544915440368988, + "grad_norm": 2.2097067832946777, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7189632058143616, + "num_tokens": 316562483.0, + "step": 12334 + }, + { + "epoch": 1.3546013617395123, + "grad_norm": 2.0631394386291504, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7000319361686707, + "num_tokens": 316593243.0, + "step": 12335 + }, + { + "epoch": 1.354711179442126, + "grad_norm": 2.7481179237365723, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.734207034111023, + "num_tokens": 316611123.0, + "step": 12336 + }, + { + "epoch": 1.3548209971447398, + "grad_norm": 2.2965757846832275, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7091193199157715, + "num_tokens": 316637552.0, + "step": 12337 + }, + { + "epoch": 1.3549308148473533, + "grad_norm": 2.5355136394500732, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7204197645187378, + "num_tokens": 316657993.0, + "step": 12338 + }, + { + "epoch": 1.355040632549967, + "grad_norm": 2.092811107635498, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7260338664054871, + "num_tokens": 316687331.0, + "step": 12339 + }, + { + "epoch": 1.3551504502525806, + "grad_norm": 2.3816990852355957, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7143248319625854, + "num_tokens": 316713080.0, + "step": 12340 + }, + { + "epoch": 1.3552602679551944, + "grad_norm": 2.132235288619995, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7020056247711182, + "num_tokens": 316739949.0, + "step": 12341 + }, + { + "epoch": 1.3553700856578081, + "grad_norm": 2.4920847415924072, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7350454330444336, + "num_tokens": 316759347.0, + "step": 12342 + }, + { + "epoch": 1.3554799033604217, + "grad_norm": 2.185914993286133, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7320941686630249, + "num_tokens": 316785435.0, + "step": 12343 + }, + { + "epoch": 1.3555897210630354, + "grad_norm": 2.0719637870788574, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.715012788772583, + "num_tokens": 316815499.0, + "step": 12344 + }, + { + "epoch": 1.355699538765649, + "grad_norm": 2.4311728477478027, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7326890826225281, + "num_tokens": 316837577.0, + "step": 12345 + }, + { + "epoch": 1.3558093564682627, + "grad_norm": 2.403846025466919, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7241033911705017, + "num_tokens": 316860005.0, + "step": 12346 + }, + { + "epoch": 1.3559191741708763, + "grad_norm": 2.6411778926849365, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7428690195083618, + "num_tokens": 316878685.0, + "step": 12347 + }, + { + "epoch": 1.35602899187349, + "grad_norm": 2.3300232887268066, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7262362837791443, + "num_tokens": 316902510.0, + "step": 12348 + }, + { + "epoch": 1.3561388095761036, + "grad_norm": 2.2666618824005127, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7159359455108643, + "num_tokens": 316927706.0, + "step": 12349 + }, + { + "epoch": 1.3562486272787173, + "grad_norm": 2.245191812515259, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7334924340248108, + "num_tokens": 316954240.0, + "step": 12350 + }, + { + "epoch": 1.356358444981331, + "grad_norm": 2.145855188369751, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7130742073059082, + "num_tokens": 316981567.0, + "step": 12351 + }, + { + "epoch": 1.3564682626839446, + "grad_norm": 2.3921024799346924, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7676955461502075, + "num_tokens": 317002073.0, + "step": 12352 + }, + { + "epoch": 1.3565780803865584, + "grad_norm": 2.094822883605957, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.6962414383888245, + "num_tokens": 317031356.0, + "step": 12353 + }, + { + "epoch": 1.356687898089172, + "grad_norm": 2.1013383865356445, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7169933915138245, + "num_tokens": 317060095.0, + "step": 12354 + }, + { + "epoch": 1.3567977157917857, + "grad_norm": 2.176650285720825, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7331520318984985, + "num_tokens": 317087231.0, + "step": 12355 + }, + { + "epoch": 1.3569075334943994, + "grad_norm": 2.4462389945983887, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7050825357437134, + "num_tokens": 317111415.0, + "step": 12356 + }, + { + "epoch": 1.357017351197013, + "grad_norm": 2.4311158657073975, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7224631905555725, + "num_tokens": 317135258.0, + "step": 12357 + }, + { + "epoch": 1.3571271688996267, + "grad_norm": 2.152672290802002, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7287154197692871, + "num_tokens": 317163322.0, + "step": 12358 + }, + { + "epoch": 1.3572369866022402, + "grad_norm": 2.366518974304199, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7451767325401306, + "num_tokens": 317186583.0, + "step": 12359 + }, + { + "epoch": 1.357346804304854, + "grad_norm": 2.0642199516296387, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6936888694763184, + "num_tokens": 317214622.0, + "step": 12360 + }, + { + "epoch": 1.3574566220074675, + "grad_norm": 2.3242528438568115, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7160929441452026, + "num_tokens": 317238317.0, + "step": 12361 + }, + { + "epoch": 1.3575664397100813, + "grad_norm": 2.231860637664795, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6972224116325378, + "num_tokens": 317267685.0, + "step": 12362 + }, + { + "epoch": 1.3576762574126948, + "grad_norm": 2.356548309326172, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7464081048965454, + "num_tokens": 317290319.0, + "step": 12363 + }, + { + "epoch": 1.3577860751153086, + "grad_norm": 2.3525190353393555, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7212828397750854, + "num_tokens": 317313059.0, + "step": 12364 + }, + { + "epoch": 1.3578958928179223, + "grad_norm": 2.2253944873809814, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7531182765960693, + "num_tokens": 317338175.0, + "step": 12365 + }, + { + "epoch": 1.3580057105205359, + "grad_norm": 2.530731201171875, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7169845700263977, + "num_tokens": 317360095.0, + "step": 12366 + }, + { + "epoch": 1.3581155282231496, + "grad_norm": 2.250664234161377, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7205287218093872, + "num_tokens": 317387053.0, + "step": 12367 + }, + { + "epoch": 1.3582253459257632, + "grad_norm": 2.462313175201416, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7434338331222534, + "num_tokens": 317407467.0, + "step": 12368 + }, + { + "epoch": 1.358335163628377, + "grad_norm": 2.579211473464966, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7063513398170471, + "num_tokens": 317429322.0, + "step": 12369 + }, + { + "epoch": 1.3584449813309907, + "grad_norm": 1.9356096982955933, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6916587948799133, + "num_tokens": 317463407.0, + "step": 12370 + }, + { + "epoch": 1.3585547990336042, + "grad_norm": 2.097990036010742, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7145686149597168, + "num_tokens": 317492243.0, + "step": 12371 + }, + { + "epoch": 1.3586646167362177, + "grad_norm": 2.803673267364502, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7161905765533447, + "num_tokens": 317512381.0, + "step": 12372 + }, + { + "epoch": 1.3587744344388315, + "grad_norm": 2.2514383792877197, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6964116096496582, + "num_tokens": 317538624.0, + "step": 12373 + }, + { + "epoch": 1.3588842521414453, + "grad_norm": 2.1174545288085938, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6997696161270142, + "num_tokens": 317564664.0, + "step": 12374 + }, + { + "epoch": 1.3589940698440588, + "grad_norm": 2.343780279159546, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7354621887207031, + "num_tokens": 317587864.0, + "step": 12375 + }, + { + "epoch": 1.3591038875466726, + "grad_norm": 2.245948314666748, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7284418940544128, + "num_tokens": 317614160.0, + "step": 12376 + }, + { + "epoch": 1.359213705249286, + "grad_norm": 2.2439215183258057, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7281426787376404, + "num_tokens": 317637638.0, + "step": 12377 + }, + { + "epoch": 1.3593235229518998, + "grad_norm": 2.3962976932525635, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7286449670791626, + "num_tokens": 317661153.0, + "step": 12378 + }, + { + "epoch": 1.3594333406545136, + "grad_norm": 2.2960736751556396, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7444831728935242, + "num_tokens": 317684764.0, + "step": 12379 + }, + { + "epoch": 1.3595431583571271, + "grad_norm": 2.532775402069092, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7165448665618896, + "num_tokens": 317706119.0, + "step": 12380 + }, + { + "epoch": 1.359652976059741, + "grad_norm": 2.0664687156677246, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7265941500663757, + "num_tokens": 317734829.0, + "step": 12381 + }, + { + "epoch": 1.3597627937623544, + "grad_norm": 2.600590229034424, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7170583605766296, + "num_tokens": 317754441.0, + "step": 12382 + }, + { + "epoch": 1.3598726114649682, + "grad_norm": 2.532028913497925, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7007970809936523, + "num_tokens": 317778173.0, + "step": 12383 + }, + { + "epoch": 1.359982429167582, + "grad_norm": 2.461500883102417, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7101644277572632, + "num_tokens": 317801678.0, + "step": 12384 + }, + { + "epoch": 1.3600922468701955, + "grad_norm": 2.3855602741241455, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7062217593193054, + "num_tokens": 317826651.0, + "step": 12385 + }, + { + "epoch": 1.360202064572809, + "grad_norm": 2.1441800594329834, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6950926184654236, + "num_tokens": 317857518.0, + "step": 12386 + }, + { + "epoch": 1.3603118822754228, + "grad_norm": 2.341609239578247, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7330829501152039, + "num_tokens": 317882022.0, + "step": 12387 + }, + { + "epoch": 1.3604216999780365, + "grad_norm": 2.1093316078186035, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7111990451812744, + "num_tokens": 317911629.0, + "step": 12388 + }, + { + "epoch": 1.36053151768065, + "grad_norm": 2.3642489910125732, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.738395631313324, + "num_tokens": 317934711.0, + "step": 12389 + }, + { + "epoch": 1.3606413353832638, + "grad_norm": 1.9949926137924194, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7454590201377869, + "num_tokens": 317965935.0, + "step": 12390 + }, + { + "epoch": 1.3607511530858774, + "grad_norm": 2.355917453765869, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.716701865196228, + "num_tokens": 317988343.0, + "step": 12391 + }, + { + "epoch": 1.360860970788491, + "grad_norm": 2.182697057723999, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6955654621124268, + "num_tokens": 318014737.0, + "step": 12392 + }, + { + "epoch": 1.3609707884911049, + "grad_norm": 2.1936275959014893, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7077600955963135, + "num_tokens": 318042119.0, + "step": 12393 + }, + { + "epoch": 1.3610806061937184, + "grad_norm": 2.4793663024902344, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7287205457687378, + "num_tokens": 318064162.0, + "step": 12394 + }, + { + "epoch": 1.3611904238963322, + "grad_norm": 2.2798638343811035, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7412230968475342, + "num_tokens": 318089228.0, + "step": 12395 + }, + { + "epoch": 1.3613002415989457, + "grad_norm": 2.3632798194885254, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7231259346008301, + "num_tokens": 318113162.0, + "step": 12396 + }, + { + "epoch": 1.3614100593015594, + "grad_norm": 2.1425435543060303, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6962413787841797, + "num_tokens": 318141955.0, + "step": 12397 + }, + { + "epoch": 1.3615198770041732, + "grad_norm": 2.040924549102783, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7069869637489319, + "num_tokens": 318172577.0, + "step": 12398 + }, + { + "epoch": 1.3616296947067867, + "grad_norm": 2.1690378189086914, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6943123936653137, + "num_tokens": 318202140.0, + "step": 12399 + }, + { + "epoch": 1.3617395124094003, + "grad_norm": 2.1837971210479736, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7139481902122498, + "num_tokens": 318229709.0, + "step": 12400 + }, + { + "epoch": 1.361849330112014, + "grad_norm": 2.46524715423584, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7084511518478394, + "num_tokens": 318251516.0, + "step": 12401 + }, + { + "epoch": 1.3619591478146278, + "grad_norm": 2.2811315059661865, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7256941199302673, + "num_tokens": 318274825.0, + "step": 12402 + }, + { + "epoch": 1.3620689655172413, + "grad_norm": 2.429804801940918, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7053877711296082, + "num_tokens": 318297492.0, + "step": 12403 + }, + { + "epoch": 1.362178783219855, + "grad_norm": 2.241642475128174, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7207984924316406, + "num_tokens": 318324510.0, + "step": 12404 + }, + { + "epoch": 1.3622886009224686, + "grad_norm": 2.1990230083465576, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6956145763397217, + "num_tokens": 318353158.0, + "step": 12405 + }, + { + "epoch": 1.3623984186250824, + "grad_norm": 2.2673051357269287, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7414870262145996, + "num_tokens": 318377679.0, + "step": 12406 + }, + { + "epoch": 1.3625082363276961, + "grad_norm": 2.4817044734954834, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7166872024536133, + "num_tokens": 318399989.0, + "step": 12407 + }, + { + "epoch": 1.3626180540303097, + "grad_norm": 2.4842541217803955, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7198309898376465, + "num_tokens": 318422765.0, + "step": 12408 + }, + { + "epoch": 1.3627278717329234, + "grad_norm": 2.124133825302124, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7260454297065735, + "num_tokens": 318451162.0, + "step": 12409 + }, + { + "epoch": 1.362837689435537, + "grad_norm": 2.3118624687194824, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7225836515426636, + "num_tokens": 318479135.0, + "step": 12410 + }, + { + "epoch": 1.3629475071381507, + "grad_norm": 2.2999684810638428, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7327582836151123, + "num_tokens": 318504087.0, + "step": 12411 + }, + { + "epoch": 1.3630573248407643, + "grad_norm": 2.111628293991089, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7084728479385376, + "num_tokens": 318531023.0, + "step": 12412 + }, + { + "epoch": 1.363167142543378, + "grad_norm": 2.2491438388824463, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7283182740211487, + "num_tokens": 318554763.0, + "step": 12413 + }, + { + "epoch": 1.3632769602459915, + "grad_norm": 2.3490984439849854, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.733852207660675, + "num_tokens": 318576896.0, + "step": 12414 + }, + { + "epoch": 1.3633867779486053, + "grad_norm": 2.280306339263916, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7096590995788574, + "num_tokens": 318601943.0, + "step": 12415 + }, + { + "epoch": 1.363496595651219, + "grad_norm": 2.244809865951538, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.700208306312561, + "num_tokens": 318629728.0, + "step": 12416 + }, + { + "epoch": 1.3636064133538326, + "grad_norm": 2.17014741897583, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7140315771102905, + "num_tokens": 318655978.0, + "step": 12417 + }, + { + "epoch": 1.3637162310564463, + "grad_norm": 1.9608650207519531, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6811075210571289, + "num_tokens": 318689957.0, + "step": 12418 + }, + { + "epoch": 1.3638260487590599, + "grad_norm": 2.0069053173065186, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7126455307006836, + "num_tokens": 318720791.0, + "step": 12419 + }, + { + "epoch": 1.3639358664616736, + "grad_norm": 2.252786636352539, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7285352349281311, + "num_tokens": 318747156.0, + "step": 12420 + }, + { + "epoch": 1.3640456841642874, + "grad_norm": 2.175508737564087, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7226136326789856, + "num_tokens": 318774487.0, + "step": 12421 + }, + { + "epoch": 1.364155501866901, + "grad_norm": 1.9852657318115234, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7061286568641663, + "num_tokens": 318806178.0, + "step": 12422 + }, + { + "epoch": 1.3642653195695145, + "grad_norm": 2.3514859676361084, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7336060404777527, + "num_tokens": 318829876.0, + "step": 12423 + }, + { + "epoch": 1.3643751372721282, + "grad_norm": 2.1493430137634277, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7152166962623596, + "num_tokens": 318856939.0, + "step": 12424 + }, + { + "epoch": 1.364484954974742, + "grad_norm": 3.15643310546875, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7228502035140991, + "num_tokens": 318875250.0, + "step": 12425 + }, + { + "epoch": 1.3645947726773555, + "grad_norm": 2.3393445014953613, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.6973797678947449, + "num_tokens": 318902288.0, + "step": 12426 + }, + { + "epoch": 1.3647045903799693, + "grad_norm": 2.5517830848693848, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7101189494132996, + "num_tokens": 318925471.0, + "step": 12427 + }, + { + "epoch": 1.3648144080825828, + "grad_norm": 2.1818275451660156, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6953256130218506, + "num_tokens": 318955650.0, + "step": 12428 + }, + { + "epoch": 1.3649242257851966, + "grad_norm": 2.490187168121338, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7278722524642944, + "num_tokens": 318977739.0, + "step": 12429 + }, + { + "epoch": 1.3650340434878103, + "grad_norm": 2.0035879611968994, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7451313734054565, + "num_tokens": 319005107.0, + "step": 12430 + }, + { + "epoch": 1.3651438611904239, + "grad_norm": 2.0047709941864014, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7074092030525208, + "num_tokens": 319035530.0, + "step": 12431 + }, + { + "epoch": 1.3652536788930376, + "grad_norm": 2.200744867324829, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7136909365653992, + "num_tokens": 319060296.0, + "step": 12432 + }, + { + "epoch": 1.3653634965956511, + "grad_norm": 2.216684579849243, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7266413569450378, + "num_tokens": 319084480.0, + "step": 12433 + }, + { + "epoch": 1.365473314298265, + "grad_norm": 2.2744052410125732, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7170839905738831, + "num_tokens": 319108438.0, + "step": 12434 + }, + { + "epoch": 1.3655831320008787, + "grad_norm": 2.2770721912384033, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7051857709884644, + "num_tokens": 319133917.0, + "step": 12435 + }, + { + "epoch": 1.3656929497034922, + "grad_norm": 2.2322659492492676, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7169331908226013, + "num_tokens": 319158079.0, + "step": 12436 + }, + { + "epoch": 1.3658027674061057, + "grad_norm": 2.235161066055298, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6891498565673828, + "num_tokens": 319185385.0, + "step": 12437 + }, + { + "epoch": 1.3659125851087195, + "grad_norm": 2.3647665977478027, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7304826974868774, + "num_tokens": 319208833.0, + "step": 12438 + }, + { + "epoch": 1.3660224028113332, + "grad_norm": 2.522345781326294, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7227476835250854, + "num_tokens": 319230428.0, + "step": 12439 + }, + { + "epoch": 1.3661322205139468, + "grad_norm": 2.360790729522705, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7088061571121216, + "num_tokens": 319255791.0, + "step": 12440 + }, + { + "epoch": 1.3662420382165605, + "grad_norm": 2.376817464828491, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7154168486595154, + "num_tokens": 319280117.0, + "step": 12441 + }, + { + "epoch": 1.366351855919174, + "grad_norm": 2.121452569961548, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6995105743408203, + "num_tokens": 319309389.0, + "step": 12442 + }, + { + "epoch": 1.3664616736217878, + "grad_norm": 2.3318333625793457, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.736003041267395, + "num_tokens": 319332094.0, + "step": 12443 + }, + { + "epoch": 1.3665714913244016, + "grad_norm": 2.5558674335479736, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7214961051940918, + "num_tokens": 319352014.0, + "step": 12444 + }, + { + "epoch": 1.3666813090270151, + "grad_norm": 2.360121250152588, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7365036010742188, + "num_tokens": 319374048.0, + "step": 12445 + }, + { + "epoch": 1.3667911267296289, + "grad_norm": 2.1284706592559814, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7180145978927612, + "num_tokens": 319401916.0, + "step": 12446 + }, + { + "epoch": 1.3669009444322424, + "grad_norm": 2.7045350074768066, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7091457843780518, + "num_tokens": 319421389.0, + "step": 12447 + }, + { + "epoch": 1.3670107621348562, + "grad_norm": 2.623842477798462, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7459418773651123, + "num_tokens": 319440581.0, + "step": 12448 + }, + { + "epoch": 1.36712057983747, + "grad_norm": 2.305791139602661, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7309920191764832, + "num_tokens": 319463704.0, + "step": 12449 + }, + { + "epoch": 1.3672303975400835, + "grad_norm": 2.3597424030303955, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7315589785575867, + "num_tokens": 319485116.0, + "step": 12450 + }, + { + "epoch": 1.367340215242697, + "grad_norm": 2.254427433013916, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7270720601081848, + "num_tokens": 319510082.0, + "step": 12451 + }, + { + "epoch": 1.3674500329453108, + "grad_norm": 2.402601957321167, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.720439076423645, + "num_tokens": 319533059.0, + "step": 12452 + }, + { + "epoch": 1.3675598506479245, + "grad_norm": 2.1727099418640137, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7056005597114563, + "num_tokens": 319560669.0, + "step": 12453 + }, + { + "epoch": 1.367669668350538, + "grad_norm": 2.052462100982666, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.723391056060791, + "num_tokens": 319589790.0, + "step": 12454 + }, + { + "epoch": 1.3677794860531518, + "grad_norm": 2.3833000659942627, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7140474319458008, + "num_tokens": 319612590.0, + "step": 12455 + }, + { + "epoch": 1.3678893037557653, + "grad_norm": 2.062621593475342, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.6962370276451111, + "num_tokens": 319645172.0, + "step": 12456 + }, + { + "epoch": 1.367999121458379, + "grad_norm": 2.015706777572632, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7133660316467285, + "num_tokens": 319675402.0, + "step": 12457 + }, + { + "epoch": 1.3681089391609929, + "grad_norm": 1.8691619634628296, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7175819277763367, + "num_tokens": 319710291.0, + "step": 12458 + }, + { + "epoch": 1.3682187568636064, + "grad_norm": 2.1542701721191406, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6983702182769775, + "num_tokens": 319737270.0, + "step": 12459 + }, + { + "epoch": 1.3683285745662201, + "grad_norm": 2.2775018215179443, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.736428439617157, + "num_tokens": 319762101.0, + "step": 12460 + }, + { + "epoch": 1.3684383922688337, + "grad_norm": 2.3406152725219727, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6799767017364502, + "num_tokens": 319790028.0, + "step": 12461 + }, + { + "epoch": 1.3685482099714474, + "grad_norm": 2.381920576095581, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7143381834030151, + "num_tokens": 319813349.0, + "step": 12462 + }, + { + "epoch": 1.3686580276740612, + "grad_norm": 2.115239381790161, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7238579988479614, + "num_tokens": 319840009.0, + "step": 12463 + }, + { + "epoch": 1.3687678453766747, + "grad_norm": 2.1786270141601562, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7204455137252808, + "num_tokens": 319866961.0, + "step": 12464 + }, + { + "epoch": 1.3688776630792883, + "grad_norm": 2.261765241622925, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7105773091316223, + "num_tokens": 319893558.0, + "step": 12465 + }, + { + "epoch": 1.368987480781902, + "grad_norm": 2.682041645050049, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7291208505630493, + "num_tokens": 319912275.0, + "step": 12466 + }, + { + "epoch": 1.3690972984845158, + "grad_norm": 2.260387659072876, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.726276159286499, + "num_tokens": 319935465.0, + "step": 12467 + }, + { + "epoch": 1.3692071161871293, + "grad_norm": 2.281444549560547, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7024645805358887, + "num_tokens": 319961528.0, + "step": 12468 + }, + { + "epoch": 1.369316933889743, + "grad_norm": 2.3554158210754395, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.733921468257904, + "num_tokens": 319985000.0, + "step": 12469 + }, + { + "epoch": 1.3694267515923566, + "grad_norm": 2.0103039741516113, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7147667407989502, + "num_tokens": 320019065.0, + "step": 12470 + }, + { + "epoch": 1.3695365692949704, + "grad_norm": 2.1505234241485596, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7185259461402893, + "num_tokens": 320046957.0, + "step": 12471 + }, + { + "epoch": 1.3696463869975841, + "grad_norm": 2.145225763320923, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7221741080284119, + "num_tokens": 320072408.0, + "step": 12472 + }, + { + "epoch": 1.3697562047001977, + "grad_norm": 2.161897659301758, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7217763066291809, + "num_tokens": 320099385.0, + "step": 12473 + }, + { + "epoch": 1.3698660224028114, + "grad_norm": 2.319777011871338, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7104026079177856, + "num_tokens": 320124534.0, + "step": 12474 + }, + { + "epoch": 1.369975840105425, + "grad_norm": 2.0776124000549316, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.717517614364624, + "num_tokens": 320151369.0, + "step": 12475 + }, + { + "epoch": 1.3700856578080387, + "grad_norm": 2.367814302444458, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7303874492645264, + "num_tokens": 320174399.0, + "step": 12476 + }, + { + "epoch": 1.3701954755106522, + "grad_norm": 2.3087973594665527, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7177741527557373, + "num_tokens": 320199110.0, + "step": 12477 + }, + { + "epoch": 1.370305293213266, + "grad_norm": 1.98088538646698, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.73245769739151, + "num_tokens": 320229339.0, + "step": 12478 + }, + { + "epoch": 1.3704151109158795, + "grad_norm": 2.255902051925659, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7172201871871948, + "num_tokens": 320253596.0, + "step": 12479 + }, + { + "epoch": 1.3705249286184933, + "grad_norm": 2.29091739654541, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7367638349533081, + "num_tokens": 320276733.0, + "step": 12480 + }, + { + "epoch": 1.370634746321107, + "grad_norm": 2.321390390396118, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7220224142074585, + "num_tokens": 320299489.0, + "step": 12481 + }, + { + "epoch": 1.3707445640237206, + "grad_norm": 2.1635468006134033, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7129342555999756, + "num_tokens": 320326752.0, + "step": 12482 + }, + { + "epoch": 1.3708543817263343, + "grad_norm": 2.218212604522705, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7172995805740356, + "num_tokens": 320354006.0, + "step": 12483 + }, + { + "epoch": 1.3709641994289479, + "grad_norm": 2.0787558555603027, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7322177290916443, + "num_tokens": 320383005.0, + "step": 12484 + }, + { + "epoch": 1.3710740171315616, + "grad_norm": 2.314167022705078, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7115868330001831, + "num_tokens": 320407235.0, + "step": 12485 + }, + { + "epoch": 1.3711838348341754, + "grad_norm": 2.2063746452331543, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7425431609153748, + "num_tokens": 320432864.0, + "step": 12486 + }, + { + "epoch": 1.371293652536789, + "grad_norm": 2.1316094398498535, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7267460823059082, + "num_tokens": 320459721.0, + "step": 12487 + }, + { + "epoch": 1.3714034702394025, + "grad_norm": 2.212444543838501, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7120757102966309, + "num_tokens": 320486860.0, + "step": 12488 + }, + { + "epoch": 1.3715132879420162, + "grad_norm": 2.2783315181732178, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7340784668922424, + "num_tokens": 320511114.0, + "step": 12489 + }, + { + "epoch": 1.37162310564463, + "grad_norm": 2.420541524887085, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7195624113082886, + "num_tokens": 320533935.0, + "step": 12490 + }, + { + "epoch": 1.3717329233472435, + "grad_norm": 2.320513963699341, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.725624680519104, + "num_tokens": 320557832.0, + "step": 12491 + }, + { + "epoch": 1.3718427410498573, + "grad_norm": 2.1958446502685547, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7406036853790283, + "num_tokens": 320582839.0, + "step": 12492 + }, + { + "epoch": 1.3719525587524708, + "grad_norm": 2.445228338241577, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7287386655807495, + "num_tokens": 320606083.0, + "step": 12493 + }, + { + "epoch": 1.3720623764550846, + "grad_norm": 2.4482357501983643, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7069636583328247, + "num_tokens": 320629677.0, + "step": 12494 + }, + { + "epoch": 1.3721721941576983, + "grad_norm": 2.376878023147583, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7392151355743408, + "num_tokens": 320652238.0, + "step": 12495 + }, + { + "epoch": 1.3722820118603118, + "grad_norm": 2.299313545227051, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.720017671585083, + "num_tokens": 320675974.0, + "step": 12496 + }, + { + "epoch": 1.3723918295629256, + "grad_norm": 2.081843852996826, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7156330347061157, + "num_tokens": 320705750.0, + "step": 12497 + }, + { + "epoch": 1.3725016472655391, + "grad_norm": 2.1056580543518066, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7269080877304077, + "num_tokens": 320731526.0, + "step": 12498 + }, + { + "epoch": 1.372611464968153, + "grad_norm": 2.2036495208740234, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7342102527618408, + "num_tokens": 320756003.0, + "step": 12499 + }, + { + "epoch": 1.3727212826707667, + "grad_norm": 2.162708282470703, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7179166078567505, + "num_tokens": 320785750.0, + "step": 12500 + }, + { + "epoch": 1.3728311003733802, + "grad_norm": 2.737607002258301, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7231673002243042, + "num_tokens": 320804106.0, + "step": 12501 + }, + { + "epoch": 1.3729409180759937, + "grad_norm": 2.4065210819244385, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7213696241378784, + "num_tokens": 320825132.0, + "step": 12502 + }, + { + "epoch": 1.3730507357786075, + "grad_norm": 2.3005948066711426, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7112482786178589, + "num_tokens": 320849950.0, + "step": 12503 + }, + { + "epoch": 1.3731605534812212, + "grad_norm": 2.097304582595825, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.711381196975708, + "num_tokens": 320880190.0, + "step": 12504 + }, + { + "epoch": 1.3732703711838348, + "grad_norm": 2.224364995956421, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7078996896743774, + "num_tokens": 320907605.0, + "step": 12505 + }, + { + "epoch": 1.3733801888864485, + "grad_norm": 1.9863953590393066, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7234694957733154, + "num_tokens": 320939182.0, + "step": 12506 + }, + { + "epoch": 1.373490006589062, + "grad_norm": 2.39060115814209, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7188870310783386, + "num_tokens": 320961969.0, + "step": 12507 + }, + { + "epoch": 1.3735998242916758, + "grad_norm": 2.1137399673461914, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.714164137840271, + "num_tokens": 320988830.0, + "step": 12508 + }, + { + "epoch": 1.3737096419942896, + "grad_norm": 2.208665609359741, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7423171997070312, + "num_tokens": 321015010.0, + "step": 12509 + }, + { + "epoch": 1.3738194596969031, + "grad_norm": 2.312513828277588, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7115417122840881, + "num_tokens": 321039272.0, + "step": 12510 + }, + { + "epoch": 1.3739292773995169, + "grad_norm": 2.434890031814575, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7180286645889282, + "num_tokens": 321061191.0, + "step": 12511 + }, + { + "epoch": 1.3740390951021304, + "grad_norm": 2.084120512008667, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.703871488571167, + "num_tokens": 321090943.0, + "step": 12512 + }, + { + "epoch": 1.3741489128047442, + "grad_norm": 2.2349655628204346, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6859891414642334, + "num_tokens": 321118093.0, + "step": 12513 + }, + { + "epoch": 1.374258730507358, + "grad_norm": 2.0596015453338623, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6831358075141907, + "num_tokens": 321149510.0, + "step": 12514 + }, + { + "epoch": 1.3743685482099715, + "grad_norm": 2.3115899562835693, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7210456132888794, + "num_tokens": 321172061.0, + "step": 12515 + }, + { + "epoch": 1.374478365912585, + "grad_norm": 2.324382781982422, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7211704850196838, + "num_tokens": 321196746.0, + "step": 12516 + }, + { + "epoch": 1.3745881836151987, + "grad_norm": 2.0393152236938477, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.695117712020874, + "num_tokens": 321227761.0, + "step": 12517 + }, + { + "epoch": 1.3746980013178125, + "grad_norm": 2.431546211242676, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7176815867424011, + "num_tokens": 321250068.0, + "step": 12518 + }, + { + "epoch": 1.374807819020426, + "grad_norm": 2.1523311138153076, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.725155234336853, + "num_tokens": 321279306.0, + "step": 12519 + }, + { + "epoch": 1.3749176367230398, + "grad_norm": 2.035421133041382, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7186203002929688, + "num_tokens": 321309619.0, + "step": 12520 + }, + { + "epoch": 1.3750274544256533, + "grad_norm": 2.325152635574341, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7338247895240784, + "num_tokens": 321332370.0, + "step": 12521 + }, + { + "epoch": 1.375137272128267, + "grad_norm": 2.314518451690674, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7263974547386169, + "num_tokens": 321354947.0, + "step": 12522 + }, + { + "epoch": 1.3752470898308808, + "grad_norm": 2.227020263671875, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.7029600143432617, + "num_tokens": 321381507.0, + "step": 12523 + }, + { + "epoch": 1.3753569075334944, + "grad_norm": 2.4431557655334473, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7218921184539795, + "num_tokens": 321404660.0, + "step": 12524 + }, + { + "epoch": 1.3754667252361081, + "grad_norm": 2.2201004028320312, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7459380626678467, + "num_tokens": 321429896.0, + "step": 12525 + }, + { + "epoch": 1.3755765429387217, + "grad_norm": 2.2482473850250244, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7050797343254089, + "num_tokens": 321455363.0, + "step": 12526 + }, + { + "epoch": 1.3756863606413354, + "grad_norm": 1.9816795587539673, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7541433572769165, + "num_tokens": 321484305.0, + "step": 12527 + }, + { + "epoch": 1.3757961783439492, + "grad_norm": 2.0768983364105225, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7095341682434082, + "num_tokens": 321512713.0, + "step": 12528 + }, + { + "epoch": 1.3759059960465627, + "grad_norm": 2.356940269470215, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7203065156936646, + "num_tokens": 321537630.0, + "step": 12529 + }, + { + "epoch": 1.3760158137491763, + "grad_norm": 2.350191831588745, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7365888357162476, + "num_tokens": 321560116.0, + "step": 12530 + }, + { + "epoch": 1.37612563145179, + "grad_norm": 2.3041670322418213, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7299054861068726, + "num_tokens": 321585751.0, + "step": 12531 + }, + { + "epoch": 1.3762354491544038, + "grad_norm": 2.3273253440856934, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7402511835098267, + "num_tokens": 321609304.0, + "step": 12532 + }, + { + "epoch": 1.3763452668570173, + "grad_norm": 2.0681850910186768, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7025660276412964, + "num_tokens": 321640472.0, + "step": 12533 + }, + { + "epoch": 1.376455084559631, + "grad_norm": 2.042203903198242, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7291285991668701, + "num_tokens": 321669156.0, + "step": 12534 + }, + { + "epoch": 1.3765649022622446, + "grad_norm": 2.296807050704956, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7346606254577637, + "num_tokens": 321691973.0, + "step": 12535 + }, + { + "epoch": 1.3766747199648584, + "grad_norm": 2.2666494846343994, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.72165447473526, + "num_tokens": 321716490.0, + "step": 12536 + }, + { + "epoch": 1.376784537667472, + "grad_norm": 2.2396678924560547, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7109673023223877, + "num_tokens": 321742688.0, + "step": 12537 + }, + { + "epoch": 1.3768943553700856, + "grad_norm": 2.06915545463562, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7088715434074402, + "num_tokens": 321770969.0, + "step": 12538 + }, + { + "epoch": 1.3770041730726994, + "grad_norm": 2.2540817260742188, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7056788802146912, + "num_tokens": 321798006.0, + "step": 12539 + }, + { + "epoch": 1.377113990775313, + "grad_norm": 2.222761631011963, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7050475478172302, + "num_tokens": 321826289.0, + "step": 12540 + }, + { + "epoch": 1.3772238084779267, + "grad_norm": 2.323676824569702, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7425082921981812, + "num_tokens": 321850375.0, + "step": 12541 + }, + { + "epoch": 1.3773336261805402, + "grad_norm": 2.144526958465576, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7676974534988403, + "num_tokens": 321875308.0, + "step": 12542 + }, + { + "epoch": 1.377443443883154, + "grad_norm": 2.1957578659057617, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7191562652587891, + "num_tokens": 321903344.0, + "step": 12543 + }, + { + "epoch": 1.3775532615857675, + "grad_norm": 2.3097825050354004, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7402230501174927, + "num_tokens": 321925844.0, + "step": 12544 + }, + { + "epoch": 1.3776630792883813, + "grad_norm": 2.444786548614502, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7063695192337036, + "num_tokens": 321947988.0, + "step": 12545 + }, + { + "epoch": 1.377772896990995, + "grad_norm": 2.175344467163086, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7264974117279053, + "num_tokens": 321974992.0, + "step": 12546 + }, + { + "epoch": 1.3778827146936086, + "grad_norm": 2.0467710494995117, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6974266767501831, + "num_tokens": 322006597.0, + "step": 12547 + }, + { + "epoch": 1.3779925323962223, + "grad_norm": 2.200202703475952, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7538095116615295, + "num_tokens": 322031303.0, + "step": 12548 + }, + { + "epoch": 1.3781023500988359, + "grad_norm": 2.4330148696899414, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7229399681091309, + "num_tokens": 322051744.0, + "step": 12549 + }, + { + "epoch": 1.3782121678014496, + "grad_norm": 2.266406536102295, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.731229305267334, + "num_tokens": 322075445.0, + "step": 12550 + }, + { + "epoch": 1.3783219855040634, + "grad_norm": 2.393634796142578, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7078906297683716, + "num_tokens": 322100799.0, + "step": 12551 + }, + { + "epoch": 1.378431803206677, + "grad_norm": 2.3866896629333496, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7531262636184692, + "num_tokens": 322124796.0, + "step": 12552 + }, + { + "epoch": 1.3785416209092904, + "grad_norm": 2.4773006439208984, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7106082439422607, + "num_tokens": 322149199.0, + "step": 12553 + }, + { + "epoch": 1.3786514386119042, + "grad_norm": 2.173882007598877, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7504107356071472, + "num_tokens": 322174324.0, + "step": 12554 + }, + { + "epoch": 1.378761256314518, + "grad_norm": 2.896820306777954, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7375131249427795, + "num_tokens": 322192045.0, + "step": 12555 + }, + { + "epoch": 1.3788710740171315, + "grad_norm": 2.499873399734497, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7164177298545837, + "num_tokens": 322215061.0, + "step": 12556 + }, + { + "epoch": 1.3789808917197452, + "grad_norm": 2.274354934692383, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7351545095443726, + "num_tokens": 322239872.0, + "step": 12557 + }, + { + "epoch": 1.3790907094223588, + "grad_norm": 2.385397434234619, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7091735601425171, + "num_tokens": 322264121.0, + "step": 12558 + }, + { + "epoch": 1.3792005271249725, + "grad_norm": 2.1865651607513428, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7379975318908691, + "num_tokens": 322290210.0, + "step": 12559 + }, + { + "epoch": 1.3793103448275863, + "grad_norm": 2.356074571609497, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.72381591796875, + "num_tokens": 322315013.0, + "step": 12560 + }, + { + "epoch": 1.3794201625301998, + "grad_norm": 2.4819953441619873, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.732689619064331, + "num_tokens": 322336236.0, + "step": 12561 + }, + { + "epoch": 1.3795299802328136, + "grad_norm": 2.2901885509490967, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7241368889808655, + "num_tokens": 322360670.0, + "step": 12562 + }, + { + "epoch": 1.3796397979354271, + "grad_norm": 2.1759846210479736, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7002617120742798, + "num_tokens": 322390270.0, + "step": 12563 + }, + { + "epoch": 1.3797496156380409, + "grad_norm": 2.085918426513672, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7009423971176147, + "num_tokens": 322419989.0, + "step": 12564 + }, + { + "epoch": 1.3798594333406546, + "grad_norm": 2.013664960861206, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7502720355987549, + "num_tokens": 322449449.0, + "step": 12565 + }, + { + "epoch": 1.3799692510432682, + "grad_norm": 2.111807346343994, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.710869312286377, + "num_tokens": 322478369.0, + "step": 12566 + }, + { + "epoch": 1.3800790687458817, + "grad_norm": 2.4965713024139404, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7317682504653931, + "num_tokens": 322499244.0, + "step": 12567 + }, + { + "epoch": 1.3801888864484955, + "grad_norm": 2.1414806842803955, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6984379291534424, + "num_tokens": 322528482.0, + "step": 12568 + }, + { + "epoch": 1.3802987041511092, + "grad_norm": 2.2370288372039795, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.704524576663971, + "num_tokens": 322553730.0, + "step": 12569 + }, + { + "epoch": 1.3804085218537228, + "grad_norm": 1.936017632484436, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7393512725830078, + "num_tokens": 322586434.0, + "step": 12570 + }, + { + "epoch": 1.3805183395563365, + "grad_norm": 2.284278631210327, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7346234321594238, + "num_tokens": 322610032.0, + "step": 12571 + }, + { + "epoch": 1.38062815725895, + "grad_norm": 2.0711560249328613, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7268216609954834, + "num_tokens": 322638759.0, + "step": 12572 + }, + { + "epoch": 1.3807379749615638, + "grad_norm": 2.169341564178467, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7174588441848755, + "num_tokens": 322666852.0, + "step": 12573 + }, + { + "epoch": 1.3808477926641776, + "grad_norm": 2.272191047668457, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7178810834884644, + "num_tokens": 322692853.0, + "step": 12574 + }, + { + "epoch": 1.380957610366791, + "grad_norm": 2.2815234661102295, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7230960726737976, + "num_tokens": 322716542.0, + "step": 12575 + }, + { + "epoch": 1.3810674280694049, + "grad_norm": 2.2741494178771973, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7231554985046387, + "num_tokens": 322742587.0, + "step": 12576 + }, + { + "epoch": 1.3811772457720184, + "grad_norm": 2.145934820175171, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7322484850883484, + "num_tokens": 322770584.0, + "step": 12577 + }, + { + "epoch": 1.3812870634746321, + "grad_norm": 2.306741952896118, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7177543640136719, + "num_tokens": 322796136.0, + "step": 12578 + }, + { + "epoch": 1.381396881177246, + "grad_norm": 2.273728847503662, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7059004306793213, + "num_tokens": 322822113.0, + "step": 12579 + }, + { + "epoch": 1.3815066988798594, + "grad_norm": 2.1533524990081787, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7062121629714966, + "num_tokens": 322851424.0, + "step": 12580 + }, + { + "epoch": 1.381616516582473, + "grad_norm": 2.4951696395874023, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7379544973373413, + "num_tokens": 322873246.0, + "step": 12581 + }, + { + "epoch": 1.3817263342850867, + "grad_norm": 2.1467461585998535, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7012858986854553, + "num_tokens": 322904118.0, + "step": 12582 + }, + { + "epoch": 1.3818361519877005, + "grad_norm": 2.1587929725646973, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7348762154579163, + "num_tokens": 322932963.0, + "step": 12583 + }, + { + "epoch": 1.381945969690314, + "grad_norm": 2.0806753635406494, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7180408239364624, + "num_tokens": 322961014.0, + "step": 12584 + }, + { + "epoch": 1.3820557873929278, + "grad_norm": 2.3843164443969727, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7186457514762878, + "num_tokens": 322984052.0, + "step": 12585 + }, + { + "epoch": 1.3821656050955413, + "grad_norm": 2.222433090209961, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7254365682601929, + "num_tokens": 323009565.0, + "step": 12586 + }, + { + "epoch": 1.382275422798155, + "grad_norm": 2.1821696758270264, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7076358795166016, + "num_tokens": 323037847.0, + "step": 12587 + }, + { + "epoch": 1.3823852405007688, + "grad_norm": 2.48332142829895, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7121009826660156, + "num_tokens": 323061930.0, + "step": 12588 + }, + { + "epoch": 1.3824950582033824, + "grad_norm": 2.4123494625091553, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7294068336486816, + "num_tokens": 323084765.0, + "step": 12589 + }, + { + "epoch": 1.3826048759059961, + "grad_norm": 2.0990371704101562, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7200058102607727, + "num_tokens": 323112532.0, + "step": 12590 + }, + { + "epoch": 1.3827146936086097, + "grad_norm": 2.3154404163360596, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7058802247047424, + "num_tokens": 323137140.0, + "step": 12591 + }, + { + "epoch": 1.3828245113112234, + "grad_norm": 1.879194736480713, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7236106991767883, + "num_tokens": 323166922.0, + "step": 12592 + }, + { + "epoch": 1.382934329013837, + "grad_norm": 2.2237207889556885, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7468024492263794, + "num_tokens": 323190944.0, + "step": 12593 + }, + { + "epoch": 1.3830441467164507, + "grad_norm": 2.661996603012085, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7169089317321777, + "num_tokens": 323210809.0, + "step": 12594 + }, + { + "epoch": 1.3831539644190642, + "grad_norm": 2.548245668411255, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7240121960639954, + "num_tokens": 323231355.0, + "step": 12595 + }, + { + "epoch": 1.383263782121678, + "grad_norm": 2.5927934646606445, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7358670830726624, + "num_tokens": 323250213.0, + "step": 12596 + }, + { + "epoch": 1.3833735998242918, + "grad_norm": 2.182995080947876, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7019492983818054, + "num_tokens": 323279673.0, + "step": 12597 + }, + { + "epoch": 1.3834834175269053, + "grad_norm": 2.26139235496521, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.723770022392273, + "num_tokens": 323304737.0, + "step": 12598 + }, + { + "epoch": 1.383593235229519, + "grad_norm": 2.2243664264678955, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7076903581619263, + "num_tokens": 323329743.0, + "step": 12599 + }, + { + "epoch": 1.3837030529321326, + "grad_norm": 2.380964756011963, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7201421856880188, + "num_tokens": 323353263.0, + "step": 12600 + }, + { + "epoch": 1.3838128706347463, + "grad_norm": 2.025951385498047, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.735312283039093, + "num_tokens": 323383747.0, + "step": 12601 + }, + { + "epoch": 1.38392268833736, + "grad_norm": 2.409548759460449, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6980600953102112, + "num_tokens": 323406572.0, + "step": 12602 + }, + { + "epoch": 1.3840325060399736, + "grad_norm": 2.082165479660034, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7053168416023254, + "num_tokens": 323435342.0, + "step": 12603 + }, + { + "epoch": 1.3841423237425874, + "grad_norm": 2.1951863765716553, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.729588508605957, + "num_tokens": 323461921.0, + "step": 12604 + }, + { + "epoch": 1.384252141445201, + "grad_norm": 2.162353754043579, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7469953894615173, + "num_tokens": 323488097.0, + "step": 12605 + }, + { + "epoch": 1.3843619591478147, + "grad_norm": 2.4763379096984863, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7370483875274658, + "num_tokens": 323508698.0, + "step": 12606 + }, + { + "epoch": 1.3844717768504282, + "grad_norm": 2.0410428047180176, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7243421673774719, + "num_tokens": 323538773.0, + "step": 12607 + }, + { + "epoch": 1.384581594553042, + "grad_norm": 2.266646385192871, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6932304501533508, + "num_tokens": 323565494.0, + "step": 12608 + }, + { + "epoch": 1.3846914122556555, + "grad_norm": 1.980972170829773, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7063987851142883, + "num_tokens": 323597800.0, + "step": 12609 + }, + { + "epoch": 1.3848012299582693, + "grad_norm": 2.0921311378479004, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6862337589263916, + "num_tokens": 323627471.0, + "step": 12610 + }, + { + "epoch": 1.384911047660883, + "grad_norm": 1.981100082397461, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7170418500900269, + "num_tokens": 323656923.0, + "step": 12611 + }, + { + "epoch": 1.3850208653634966, + "grad_norm": 2.4111926555633545, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7406394481658936, + "num_tokens": 323677799.0, + "step": 12612 + }, + { + "epoch": 1.3851306830661103, + "grad_norm": 2.2064015865325928, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7073415517807007, + "num_tokens": 323706943.0, + "step": 12613 + }, + { + "epoch": 1.3852405007687238, + "grad_norm": 2.3712942600250244, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.719221830368042, + "num_tokens": 323730480.0, + "step": 12614 + }, + { + "epoch": 1.3853503184713376, + "grad_norm": 2.059530735015869, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6985172629356384, + "num_tokens": 323762482.0, + "step": 12615 + }, + { + "epoch": 1.3854601361739514, + "grad_norm": 2.254578113555908, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7152862548828125, + "num_tokens": 323788090.0, + "step": 12616 + }, + { + "epoch": 1.385569953876565, + "grad_norm": 2.4061007499694824, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7040513753890991, + "num_tokens": 323812121.0, + "step": 12617 + }, + { + "epoch": 1.3856797715791784, + "grad_norm": 2.151531219482422, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7094179391860962, + "num_tokens": 323840661.0, + "step": 12618 + }, + { + "epoch": 1.3857895892817922, + "grad_norm": 2.3394575119018555, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7219414114952087, + "num_tokens": 323863978.0, + "step": 12619 + }, + { + "epoch": 1.385899406984406, + "grad_norm": 2.2054615020751953, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7154136896133423, + "num_tokens": 323890268.0, + "step": 12620 + }, + { + "epoch": 1.3860092246870195, + "grad_norm": 1.9212889671325684, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7075793743133545, + "num_tokens": 323922914.0, + "step": 12621 + }, + { + "epoch": 1.3861190423896332, + "grad_norm": 2.4151220321655273, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7084857225418091, + "num_tokens": 323946780.0, + "step": 12622 + }, + { + "epoch": 1.3862288600922468, + "grad_norm": 2.2051737308502197, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7290773391723633, + "num_tokens": 323973890.0, + "step": 12623 + }, + { + "epoch": 1.3863386777948605, + "grad_norm": 2.25016450881958, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7012416124343872, + "num_tokens": 324000486.0, + "step": 12624 + }, + { + "epoch": 1.3864484954974743, + "grad_norm": 2.173736810684204, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7114511728286743, + "num_tokens": 324028073.0, + "step": 12625 + }, + { + "epoch": 1.3865583132000878, + "grad_norm": 2.4289679527282715, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7355378866195679, + "num_tokens": 324048402.0, + "step": 12626 + }, + { + "epoch": 1.3866681309027016, + "grad_norm": 2.1753571033477783, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6977195739746094, + "num_tokens": 324078466.0, + "step": 12627 + }, + { + "epoch": 1.3867779486053151, + "grad_norm": 2.4557459354400635, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7493677139282227, + "num_tokens": 324100710.0, + "step": 12628 + }, + { + "epoch": 1.3868877663079289, + "grad_norm": 2.255592107772827, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7344508171081543, + "num_tokens": 324125022.0, + "step": 12629 + }, + { + "epoch": 1.3869975840105426, + "grad_norm": 2.0809576511383057, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7328416109085083, + "num_tokens": 324151418.0, + "step": 12630 + }, + { + "epoch": 1.3871074017131562, + "grad_norm": 2.122880458831787, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7276563048362732, + "num_tokens": 324179176.0, + "step": 12631 + }, + { + "epoch": 1.3872172194157697, + "grad_norm": 2.0399351119995117, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.717314600944519, + "num_tokens": 324208404.0, + "step": 12632 + }, + { + "epoch": 1.3873270371183835, + "grad_norm": 2.063488006591797, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7015233039855957, + "num_tokens": 324240071.0, + "step": 12633 + }, + { + "epoch": 1.3874368548209972, + "grad_norm": 2.241298198699951, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7221720218658447, + "num_tokens": 324265063.0, + "step": 12634 + }, + { + "epoch": 1.3875466725236107, + "grad_norm": 2.4536256790161133, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.710913896560669, + "num_tokens": 324287775.0, + "step": 12635 + }, + { + "epoch": 1.3876564902262245, + "grad_norm": 2.262359142303467, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6819267272949219, + "num_tokens": 324315229.0, + "step": 12636 + }, + { + "epoch": 1.387766307928838, + "grad_norm": 2.640068769454956, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7227832078933716, + "num_tokens": 324336044.0, + "step": 12637 + }, + { + "epoch": 1.3878761256314518, + "grad_norm": 2.1819956302642822, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7132370471954346, + "num_tokens": 324365805.0, + "step": 12638 + }, + { + "epoch": 1.3879859433340656, + "grad_norm": 2.2385451793670654, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7284880876541138, + "num_tokens": 324389764.0, + "step": 12639 + }, + { + "epoch": 1.388095761036679, + "grad_norm": 2.187933921813965, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7161909341812134, + "num_tokens": 324418037.0, + "step": 12640 + }, + { + "epoch": 1.3882055787392928, + "grad_norm": 2.4438979625701904, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7486725449562073, + "num_tokens": 324438977.0, + "step": 12641 + }, + { + "epoch": 1.3883153964419064, + "grad_norm": 2.3895843029022217, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7426718473434448, + "num_tokens": 324460933.0, + "step": 12642 + }, + { + "epoch": 1.3884252141445201, + "grad_norm": 2.317807674407959, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7202771902084351, + "num_tokens": 324485057.0, + "step": 12643 + }, + { + "epoch": 1.388535031847134, + "grad_norm": 2.6237070560455322, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7236286997795105, + "num_tokens": 324506094.0, + "step": 12644 + }, + { + "epoch": 1.3886448495497474, + "grad_norm": 2.152130126953125, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7381519079208374, + "num_tokens": 324534146.0, + "step": 12645 + }, + { + "epoch": 1.388754667252361, + "grad_norm": 2.270211935043335, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7123017907142639, + "num_tokens": 324560383.0, + "step": 12646 + }, + { + "epoch": 1.3888644849549747, + "grad_norm": 2.409942388534546, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7241430878639221, + "num_tokens": 324582617.0, + "step": 12647 + }, + { + "epoch": 1.3889743026575885, + "grad_norm": 2.3453712463378906, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.723879337310791, + "num_tokens": 324606620.0, + "step": 12648 + }, + { + "epoch": 1.389084120360202, + "grad_norm": 2.6360363960266113, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7439024448394775, + "num_tokens": 324626602.0, + "step": 12649 + }, + { + "epoch": 1.3891939380628158, + "grad_norm": 2.178245782852173, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6967384219169617, + "num_tokens": 324654899.0, + "step": 12650 + }, + { + "epoch": 1.3893037557654293, + "grad_norm": 2.0031816959381104, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7163758873939514, + "num_tokens": 324686783.0, + "step": 12651 + }, + { + "epoch": 1.389413573468043, + "grad_norm": 2.2418174743652344, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7172532081604004, + "num_tokens": 324715012.0, + "step": 12652 + }, + { + "epoch": 1.3895233911706568, + "grad_norm": 2.0250654220581055, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7050951719284058, + "num_tokens": 324744911.0, + "step": 12653 + }, + { + "epoch": 1.3896332088732704, + "grad_norm": 2.336024045944214, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7158294320106506, + "num_tokens": 324770670.0, + "step": 12654 + }, + { + "epoch": 1.389743026575884, + "grad_norm": 2.0204100608825684, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7047452926635742, + "num_tokens": 324802851.0, + "step": 12655 + }, + { + "epoch": 1.3898528442784976, + "grad_norm": 2.6267096996307373, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7155367136001587, + "num_tokens": 324823039.0, + "step": 12656 + }, + { + "epoch": 1.3899626619811114, + "grad_norm": 2.0709731578826904, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7243994474411011, + "num_tokens": 324850157.0, + "step": 12657 + }, + { + "epoch": 1.390072479683725, + "grad_norm": 2.1091866493225098, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7085000276565552, + "num_tokens": 324881186.0, + "step": 12658 + }, + { + "epoch": 1.3901822973863387, + "grad_norm": 2.3108227252960205, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7215903997421265, + "num_tokens": 324904707.0, + "step": 12659 + }, + { + "epoch": 1.3902921150889522, + "grad_norm": 2.3428242206573486, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7464064359664917, + "num_tokens": 324927805.0, + "step": 12660 + }, + { + "epoch": 1.390401932791566, + "grad_norm": 2.4584949016571045, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7316352128982544, + "num_tokens": 324951214.0, + "step": 12661 + }, + { + "epoch": 1.3905117504941797, + "grad_norm": 2.252880573272705, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7431660890579224, + "num_tokens": 324974231.0, + "step": 12662 + }, + { + "epoch": 1.3906215681967933, + "grad_norm": 2.3423287868499756, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7335360646247864, + "num_tokens": 324998445.0, + "step": 12663 + }, + { + "epoch": 1.390731385899407, + "grad_norm": 2.0825486183166504, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7097162008285522, + "num_tokens": 325028099.0, + "step": 12664 + }, + { + "epoch": 1.3908412036020206, + "grad_norm": 2.4865360260009766, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7145260572433472, + "num_tokens": 325050393.0, + "step": 12665 + }, + { + "epoch": 1.3909510213046343, + "grad_norm": 2.3272743225097656, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7272006869316101, + "num_tokens": 325073658.0, + "step": 12666 + }, + { + "epoch": 1.391060839007248, + "grad_norm": 2.1467039585113525, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7262351512908936, + "num_tokens": 325102598.0, + "step": 12667 + }, + { + "epoch": 1.3911706567098616, + "grad_norm": 2.1731178760528564, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7411254048347473, + "num_tokens": 325127954.0, + "step": 12668 + }, + { + "epoch": 1.3912804744124752, + "grad_norm": 2.197150707244873, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6982602477073669, + "num_tokens": 325154143.0, + "step": 12669 + }, + { + "epoch": 1.391390292115089, + "grad_norm": 1.8329486846923828, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7022110223770142, + "num_tokens": 325186917.0, + "step": 12670 + }, + { + "epoch": 1.3915001098177027, + "grad_norm": 2.163191556930542, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7292605042457581, + "num_tokens": 325213775.0, + "step": 12671 + }, + { + "epoch": 1.3916099275203162, + "grad_norm": 2.4477005004882812, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7218774557113647, + "num_tokens": 325238290.0, + "step": 12672 + }, + { + "epoch": 1.39171974522293, + "grad_norm": 2.077008008956909, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7158787250518799, + "num_tokens": 325264883.0, + "step": 12673 + }, + { + "epoch": 1.3918295629255435, + "grad_norm": 2.398714065551758, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7431082725524902, + "num_tokens": 325285157.0, + "step": 12674 + }, + { + "epoch": 1.3919393806281573, + "grad_norm": 2.183457612991333, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7346980571746826, + "num_tokens": 325310207.0, + "step": 12675 + }, + { + "epoch": 1.392049198330771, + "grad_norm": 2.1767261028289795, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7372226715087891, + "num_tokens": 325336177.0, + "step": 12676 + }, + { + "epoch": 1.3921590160333845, + "grad_norm": 2.2757656574249268, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.723086416721344, + "num_tokens": 325361453.0, + "step": 12677 + }, + { + "epoch": 1.3922688337359983, + "grad_norm": 2.440685987472534, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7231783866882324, + "num_tokens": 325383531.0, + "step": 12678 + }, + { + "epoch": 1.3923786514386118, + "grad_norm": 2.3761467933654785, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7141600847244263, + "num_tokens": 325406863.0, + "step": 12679 + }, + { + "epoch": 1.3924884691412256, + "grad_norm": 2.2781736850738525, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7408307790756226, + "num_tokens": 325429799.0, + "step": 12680 + }, + { + "epoch": 1.3925982868438394, + "grad_norm": 2.223522424697876, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7301098108291626, + "num_tokens": 325455668.0, + "step": 12681 + }, + { + "epoch": 1.3927081045464529, + "grad_norm": 2.2917659282684326, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7410608530044556, + "num_tokens": 325479880.0, + "step": 12682 + }, + { + "epoch": 1.3928179222490664, + "grad_norm": 2.4250593185424805, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7351710796356201, + "num_tokens": 325500691.0, + "step": 12683 + }, + { + "epoch": 1.3929277399516802, + "grad_norm": 2.2288544178009033, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7239065170288086, + "num_tokens": 325525745.0, + "step": 12684 + }, + { + "epoch": 1.393037557654294, + "grad_norm": 2.301401376724243, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.752112090587616, + "num_tokens": 325547730.0, + "step": 12685 + }, + { + "epoch": 1.3931473753569075, + "grad_norm": 2.1984469890594482, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7064319849014282, + "num_tokens": 325578529.0, + "step": 12686 + }, + { + "epoch": 1.3932571930595212, + "grad_norm": 2.2859349250793457, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7274909615516663, + "num_tokens": 325605433.0, + "step": 12687 + }, + { + "epoch": 1.3933670107621348, + "grad_norm": 2.6907594203948975, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7442102432250977, + "num_tokens": 325624543.0, + "step": 12688 + }, + { + "epoch": 1.3934768284647485, + "grad_norm": 2.1825814247131348, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7117533683776855, + "num_tokens": 325653341.0, + "step": 12689 + }, + { + "epoch": 1.3935866461673623, + "grad_norm": 2.508166790008545, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7345728874206543, + "num_tokens": 325674734.0, + "step": 12690 + }, + { + "epoch": 1.3936964638699758, + "grad_norm": 2.279391050338745, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7127310037612915, + "num_tokens": 325701394.0, + "step": 12691 + }, + { + "epoch": 1.3938062815725896, + "grad_norm": 2.2357468605041504, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7262893319129944, + "num_tokens": 325725153.0, + "step": 12692 + }, + { + "epoch": 1.393916099275203, + "grad_norm": 2.27651309967041, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7397990226745605, + "num_tokens": 325750362.0, + "step": 12693 + }, + { + "epoch": 1.3940259169778169, + "grad_norm": 2.114748477935791, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7318972945213318, + "num_tokens": 325777419.0, + "step": 12694 + }, + { + "epoch": 1.3941357346804306, + "grad_norm": 2.2160604000091553, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7245250940322876, + "num_tokens": 325802674.0, + "step": 12695 + }, + { + "epoch": 1.3942455523830442, + "grad_norm": 2.1666102409362793, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7202029228210449, + "num_tokens": 325829072.0, + "step": 12696 + }, + { + "epoch": 1.3943553700856577, + "grad_norm": 2.6393604278564453, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7452148199081421, + "num_tokens": 325849481.0, + "step": 12697 + }, + { + "epoch": 1.3944651877882714, + "grad_norm": 2.16111159324646, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7169011831283569, + "num_tokens": 325876443.0, + "step": 12698 + }, + { + "epoch": 1.3945750054908852, + "grad_norm": 2.7054402828216553, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7266504764556885, + "num_tokens": 325895081.0, + "step": 12699 + }, + { + "epoch": 1.3946848231934987, + "grad_norm": 2.343851089477539, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7255100011825562, + "num_tokens": 325918273.0, + "step": 12700 + }, + { + "epoch": 1.3947946408961125, + "grad_norm": 2.7234344482421875, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7350521087646484, + "num_tokens": 325937899.0, + "step": 12701 + }, + { + "epoch": 1.394904458598726, + "grad_norm": 2.8339710235595703, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7330081462860107, + "num_tokens": 325956655.0, + "step": 12702 + }, + { + "epoch": 1.3950142763013398, + "grad_norm": 1.997789740562439, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6871284246444702, + "num_tokens": 325989276.0, + "step": 12703 + }, + { + "epoch": 1.3951240940039535, + "grad_norm": 2.0872511863708496, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7080725431442261, + "num_tokens": 326020349.0, + "step": 12704 + }, + { + "epoch": 1.395233911706567, + "grad_norm": 2.241640567779541, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7444031238555908, + "num_tokens": 326044513.0, + "step": 12705 + }, + { + "epoch": 1.3953437294091808, + "grad_norm": 1.982488989830017, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7381496429443359, + "num_tokens": 326072726.0, + "step": 12706 + }, + { + "epoch": 1.3954535471117944, + "grad_norm": 2.0898830890655518, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6998124718666077, + "num_tokens": 326101381.0, + "step": 12707 + }, + { + "epoch": 1.3955633648144081, + "grad_norm": 2.3097410202026367, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7113089561462402, + "num_tokens": 326128670.0, + "step": 12708 + }, + { + "epoch": 1.3956731825170219, + "grad_norm": 2.1379973888397217, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7025375366210938, + "num_tokens": 326157156.0, + "step": 12709 + }, + { + "epoch": 1.3957830002196354, + "grad_norm": 2.158702850341797, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7328804731369019, + "num_tokens": 326184227.0, + "step": 12710 + }, + { + "epoch": 1.395892817922249, + "grad_norm": 2.303199291229248, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7217212319374084, + "num_tokens": 326209633.0, + "step": 12711 + }, + { + "epoch": 1.3960026356248627, + "grad_norm": 2.1294586658477783, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6944702863693237, + "num_tokens": 326239122.0, + "step": 12712 + }, + { + "epoch": 1.3961124533274765, + "grad_norm": 2.030555486679077, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7037561535835266, + "num_tokens": 326268715.0, + "step": 12713 + }, + { + "epoch": 1.39622227103009, + "grad_norm": 2.0899548530578613, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.716235876083374, + "num_tokens": 326299316.0, + "step": 12714 + }, + { + "epoch": 1.3963320887327038, + "grad_norm": 2.056826591491699, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7011357545852661, + "num_tokens": 326331127.0, + "step": 12715 + }, + { + "epoch": 1.3964419064353173, + "grad_norm": 2.578413248062134, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7415750622749329, + "num_tokens": 326350958.0, + "step": 12716 + }, + { + "epoch": 1.396551724137931, + "grad_norm": 2.3231468200683594, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7156696319580078, + "num_tokens": 326376788.0, + "step": 12717 + }, + { + "epoch": 1.3966615418405448, + "grad_norm": 2.1507136821746826, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.6989130973815918, + "num_tokens": 326405755.0, + "step": 12718 + }, + { + "epoch": 1.3967713595431583, + "grad_norm": 2.1314103603363037, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7041143178939819, + "num_tokens": 326433764.0, + "step": 12719 + }, + { + "epoch": 1.396881177245772, + "grad_norm": 2.216606855392456, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7167683839797974, + "num_tokens": 326459719.0, + "step": 12720 + }, + { + "epoch": 1.3969909949483856, + "grad_norm": 1.9533013105392456, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7245861291885376, + "num_tokens": 326491059.0, + "step": 12721 + }, + { + "epoch": 1.3971008126509994, + "grad_norm": 2.293457508087158, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7354791164398193, + "num_tokens": 326516760.0, + "step": 12722 + }, + { + "epoch": 1.397210630353613, + "grad_norm": 2.274488925933838, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7232978940010071, + "num_tokens": 326542078.0, + "step": 12723 + }, + { + "epoch": 1.3973204480562267, + "grad_norm": 2.0140678882598877, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7191476225852966, + "num_tokens": 326571287.0, + "step": 12724 + }, + { + "epoch": 1.3974302657588402, + "grad_norm": 2.132368326187134, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7158447504043579, + "num_tokens": 326599393.0, + "step": 12725 + }, + { + "epoch": 1.397540083461454, + "grad_norm": 2.081921339035034, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7113441228866577, + "num_tokens": 326629813.0, + "step": 12726 + }, + { + "epoch": 1.3976499011640677, + "grad_norm": 2.278101682662964, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7149036526679993, + "num_tokens": 326656368.0, + "step": 12727 + }, + { + "epoch": 1.3977597188666813, + "grad_norm": 2.271260976791382, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6859130859375, + "num_tokens": 326681273.0, + "step": 12728 + }, + { + "epoch": 1.397869536569295, + "grad_norm": 2.0867974758148193, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6973037719726562, + "num_tokens": 326710320.0, + "step": 12729 + }, + { + "epoch": 1.3979793542719086, + "grad_norm": 2.118475914001465, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7066259384155273, + "num_tokens": 326741261.0, + "step": 12730 + }, + { + "epoch": 1.3980891719745223, + "grad_norm": 2.6118555068969727, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7386173009872437, + "num_tokens": 326761505.0, + "step": 12731 + }, + { + "epoch": 1.398198989677136, + "grad_norm": 2.33268666267395, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7368506193161011, + "num_tokens": 326783826.0, + "step": 12732 + }, + { + "epoch": 1.3983088073797496, + "grad_norm": 2.2404110431671143, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7160552144050598, + "num_tokens": 326809876.0, + "step": 12733 + }, + { + "epoch": 1.3984186250823631, + "grad_norm": 2.160029888153076, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7403156757354736, + "num_tokens": 326835725.0, + "step": 12734 + }, + { + "epoch": 1.398528442784977, + "grad_norm": 2.448333740234375, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7053585052490234, + "num_tokens": 326857526.0, + "step": 12735 + }, + { + "epoch": 1.3986382604875907, + "grad_norm": 2.1635067462921143, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6909536123275757, + "num_tokens": 326885354.0, + "step": 12736 + }, + { + "epoch": 1.3987480781902042, + "grad_norm": 2.2676479816436768, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7135199308395386, + "num_tokens": 326910650.0, + "step": 12737 + }, + { + "epoch": 1.398857895892818, + "grad_norm": 2.188302516937256, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7177711725234985, + "num_tokens": 326936592.0, + "step": 12738 + }, + { + "epoch": 1.3989677135954315, + "grad_norm": 2.079892635345459, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7086686491966248, + "num_tokens": 326964266.0, + "step": 12739 + }, + { + "epoch": 1.3990775312980452, + "grad_norm": 2.4595232009887695, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7222653031349182, + "num_tokens": 326986632.0, + "step": 12740 + }, + { + "epoch": 1.399187349000659, + "grad_norm": 2.3860363960266113, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7086927890777588, + "num_tokens": 327010665.0, + "step": 12741 + }, + { + "epoch": 1.3992971667032725, + "grad_norm": 2.2311618328094482, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7049477696418762, + "num_tokens": 327035813.0, + "step": 12742 + }, + { + "epoch": 1.3994069844058863, + "grad_norm": 2.3712961673736572, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7208805084228516, + "num_tokens": 327059869.0, + "step": 12743 + }, + { + "epoch": 1.3995168021084998, + "grad_norm": 2.3210017681121826, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7116414308547974, + "num_tokens": 327085027.0, + "step": 12744 + }, + { + "epoch": 1.3996266198111136, + "grad_norm": 2.321296453475952, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7207232117652893, + "num_tokens": 327110318.0, + "step": 12745 + }, + { + "epoch": 1.3997364375137273, + "grad_norm": 2.434027910232544, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7066637873649597, + "num_tokens": 327133020.0, + "step": 12746 + }, + { + "epoch": 1.3998462552163409, + "grad_norm": 1.9079517126083374, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7166702747344971, + "num_tokens": 327165622.0, + "step": 12747 + }, + { + "epoch": 1.3999560729189544, + "grad_norm": 2.12990403175354, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6852244138717651, + "num_tokens": 327193375.0, + "step": 12748 + }, + { + "epoch": 1.4000658906215682, + "grad_norm": 2.3284778594970703, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7241083383560181, + "num_tokens": 327217723.0, + "step": 12749 + }, + { + "epoch": 1.400175708324182, + "grad_norm": 2.327686071395874, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7226657867431641, + "num_tokens": 327240270.0, + "step": 12750 + }, + { + "epoch": 1.4002855260267955, + "grad_norm": 2.316080331802368, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7159889936447144, + "num_tokens": 327263859.0, + "step": 12751 + }, + { + "epoch": 1.4003953437294092, + "grad_norm": 2.273305654525757, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7395757436752319, + "num_tokens": 327287359.0, + "step": 12752 + }, + { + "epoch": 1.4005051614320227, + "grad_norm": 2.0952374935150146, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7339814305305481, + "num_tokens": 327315617.0, + "step": 12753 + }, + { + "epoch": 1.4006149791346365, + "grad_norm": 2.0965778827667236, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7126035690307617, + "num_tokens": 327341221.0, + "step": 12754 + }, + { + "epoch": 1.4007247968372503, + "grad_norm": 2.314124822616577, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7448558807373047, + "num_tokens": 327364190.0, + "step": 12755 + }, + { + "epoch": 1.4008346145398638, + "grad_norm": 2.0900251865386963, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.694959282875061, + "num_tokens": 327395403.0, + "step": 12756 + }, + { + "epoch": 1.4009444322424776, + "grad_norm": 2.70114803314209, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7147358059883118, + "num_tokens": 327414908.0, + "step": 12757 + }, + { + "epoch": 1.401054249945091, + "grad_norm": 2.2579917907714844, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6974620819091797, + "num_tokens": 327441895.0, + "step": 12758 + }, + { + "epoch": 1.4011640676477048, + "grad_norm": 2.3289341926574707, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7230645418167114, + "num_tokens": 327466630.0, + "step": 12759 + }, + { + "epoch": 1.4012738853503186, + "grad_norm": 2.1942169666290283, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.747252881526947, + "num_tokens": 327491816.0, + "step": 12760 + }, + { + "epoch": 1.4013837030529321, + "grad_norm": 2.1731719970703125, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7055554389953613, + "num_tokens": 327520279.0, + "step": 12761 + }, + { + "epoch": 1.4014935207555457, + "grad_norm": 2.376725196838379, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7174951434135437, + "num_tokens": 327542057.0, + "step": 12762 + }, + { + "epoch": 1.4016033384581594, + "grad_norm": 2.1842174530029297, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6994378566741943, + "num_tokens": 327569843.0, + "step": 12763 + }, + { + "epoch": 1.4017131561607732, + "grad_norm": 2.4501793384552, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7128688097000122, + "num_tokens": 327592254.0, + "step": 12764 + }, + { + "epoch": 1.4018229738633867, + "grad_norm": 2.1535604000091553, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7231456637382507, + "num_tokens": 327620969.0, + "step": 12765 + }, + { + "epoch": 1.4019327915660005, + "grad_norm": 2.5092132091522217, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7260486483573914, + "num_tokens": 327641473.0, + "step": 12766 + }, + { + "epoch": 1.402042609268614, + "grad_norm": 2.305481433868408, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7434119582176208, + "num_tokens": 327663223.0, + "step": 12767 + }, + { + "epoch": 1.4021524269712278, + "grad_norm": 2.3180432319641113, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7225503921508789, + "num_tokens": 327686544.0, + "step": 12768 + }, + { + "epoch": 1.4022622446738415, + "grad_norm": 2.150883674621582, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7215991616249084, + "num_tokens": 327712058.0, + "step": 12769 + }, + { + "epoch": 1.402372062376455, + "grad_norm": 2.278414487838745, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.72312331199646, + "num_tokens": 327736306.0, + "step": 12770 + }, + { + "epoch": 1.4024818800790688, + "grad_norm": 2.176131010055542, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7030428647994995, + "num_tokens": 327762992.0, + "step": 12771 + }, + { + "epoch": 1.4025916977816824, + "grad_norm": 2.381417989730835, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7365059852600098, + "num_tokens": 327786225.0, + "step": 12772 + }, + { + "epoch": 1.4027015154842961, + "grad_norm": 2.321899890899658, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7324900031089783, + "num_tokens": 327811008.0, + "step": 12773 + }, + { + "epoch": 1.4028113331869096, + "grad_norm": 2.0007340908050537, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7333974242210388, + "num_tokens": 327840048.0, + "step": 12774 + }, + { + "epoch": 1.4029211508895234, + "grad_norm": 2.2535929679870605, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7312138080596924, + "num_tokens": 327864756.0, + "step": 12775 + }, + { + "epoch": 1.403030968592137, + "grad_norm": 2.3232948780059814, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7102383971214294, + "num_tokens": 327889774.0, + "step": 12776 + }, + { + "epoch": 1.4031407862947507, + "grad_norm": 2.1810545921325684, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7461444139480591, + "num_tokens": 327915508.0, + "step": 12777 + }, + { + "epoch": 1.4032506039973645, + "grad_norm": 2.1587681770324707, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7165049314498901, + "num_tokens": 327942100.0, + "step": 12778 + }, + { + "epoch": 1.403360421699978, + "grad_norm": 2.0239012241363525, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7048660516738892, + "num_tokens": 327973003.0, + "step": 12779 + }, + { + "epoch": 1.4034702394025917, + "grad_norm": 1.9447766542434692, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6885615587234497, + "num_tokens": 328006336.0, + "step": 12780 + }, + { + "epoch": 1.4035800571052053, + "grad_norm": 1.9824316501617432, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7193020582199097, + "num_tokens": 328036923.0, + "step": 12781 + }, + { + "epoch": 1.403689874807819, + "grad_norm": 2.0058419704437256, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6978870034217834, + "num_tokens": 328066800.0, + "step": 12782 + }, + { + "epoch": 1.4037996925104328, + "grad_norm": 2.2871177196502686, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7137761116027832, + "num_tokens": 328092268.0, + "step": 12783 + }, + { + "epoch": 1.4039095102130463, + "grad_norm": 2.175225019454956, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6992802619934082, + "num_tokens": 328120554.0, + "step": 12784 + }, + { + "epoch": 1.40401932791566, + "grad_norm": 2.099806070327759, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7033411860466003, + "num_tokens": 328148471.0, + "step": 12785 + }, + { + "epoch": 1.4041291456182736, + "grad_norm": 2.000887155532837, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.6988791823387146, + "num_tokens": 328179703.0, + "step": 12786 + }, + { + "epoch": 1.4042389633208874, + "grad_norm": 2.266023874282837, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.716769814491272, + "num_tokens": 328204323.0, + "step": 12787 + }, + { + "epoch": 1.404348781023501, + "grad_norm": 2.4178950786590576, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7389973402023315, + "num_tokens": 328226335.0, + "step": 12788 + }, + { + "epoch": 1.4044585987261147, + "grad_norm": 2.060133934020996, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7180383801460266, + "num_tokens": 328253630.0, + "step": 12789 + }, + { + "epoch": 1.4045684164287282, + "grad_norm": 2.1048784255981445, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7296383380889893, + "num_tokens": 328282855.0, + "step": 12790 + }, + { + "epoch": 1.404678234131342, + "grad_norm": 2.306548833847046, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7119399309158325, + "num_tokens": 328306348.0, + "step": 12791 + }, + { + "epoch": 1.4047880518339557, + "grad_norm": 2.1082661151885986, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7138459086418152, + "num_tokens": 328334753.0, + "step": 12792 + }, + { + "epoch": 1.4048978695365693, + "grad_norm": 2.2818222045898438, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7225396633148193, + "num_tokens": 328359617.0, + "step": 12793 + }, + { + "epoch": 1.405007687239183, + "grad_norm": 2.3602612018585205, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7049375176429749, + "num_tokens": 328383072.0, + "step": 12794 + }, + { + "epoch": 1.4051175049417965, + "grad_norm": 2.233208417892456, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7146086692810059, + "num_tokens": 328409705.0, + "step": 12795 + }, + { + "epoch": 1.4052273226444103, + "grad_norm": 2.2359979152679443, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7214316129684448, + "num_tokens": 328435225.0, + "step": 12796 + }, + { + "epoch": 1.405337140347024, + "grad_norm": 2.539707899093628, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.728000819683075, + "num_tokens": 328455355.0, + "step": 12797 + }, + { + "epoch": 1.4054469580496376, + "grad_norm": 2.54431414604187, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7211081385612488, + "num_tokens": 328476357.0, + "step": 12798 + }, + { + "epoch": 1.4055567757522511, + "grad_norm": 2.2126498222351074, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6987239122390747, + "num_tokens": 328505008.0, + "step": 12799 + }, + { + "epoch": 1.4056665934548649, + "grad_norm": 2.4470603466033936, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.727368950843811, + "num_tokens": 328526993.0, + "step": 12800 + }, + { + "epoch": 1.4057764111574786, + "grad_norm": 2.4971470832824707, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.714961051940918, + "num_tokens": 328552522.0, + "step": 12801 + }, + { + "epoch": 1.4058862288600922, + "grad_norm": 2.1758482456207275, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7078409790992737, + "num_tokens": 328580869.0, + "step": 12802 + }, + { + "epoch": 1.405996046562706, + "grad_norm": 2.471780776977539, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7125557661056519, + "num_tokens": 328603112.0, + "step": 12803 + }, + { + "epoch": 1.4061058642653195, + "grad_norm": 2.1730964183807373, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7152546644210815, + "num_tokens": 328631038.0, + "step": 12804 + }, + { + "epoch": 1.4062156819679332, + "grad_norm": 1.9774562120437622, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7240501642227173, + "num_tokens": 328659615.0, + "step": 12805 + }, + { + "epoch": 1.406325499670547, + "grad_norm": 2.451887607574463, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7318230271339417, + "num_tokens": 328681232.0, + "step": 12806 + }, + { + "epoch": 1.4064353173731605, + "grad_norm": 2.570364475250244, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7400244474411011, + "num_tokens": 328701389.0, + "step": 12807 + }, + { + "epoch": 1.4065451350757743, + "grad_norm": 2.0890870094299316, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7194865942001343, + "num_tokens": 328728271.0, + "step": 12808 + }, + { + "epoch": 1.4066549527783878, + "grad_norm": 2.261373281478882, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7218001484870911, + "num_tokens": 328752680.0, + "step": 12809 + }, + { + "epoch": 1.4067647704810016, + "grad_norm": 2.5668106079101562, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7421221137046814, + "num_tokens": 328773244.0, + "step": 12810 + }, + { + "epoch": 1.4068745881836153, + "grad_norm": 2.235464572906494, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.693461537361145, + "num_tokens": 328800694.0, + "step": 12811 + }, + { + "epoch": 1.4069844058862289, + "grad_norm": 2.5270814895629883, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7234396934509277, + "num_tokens": 328823430.0, + "step": 12812 + }, + { + "epoch": 1.4070942235888424, + "grad_norm": 2.3045530319213867, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7231391668319702, + "num_tokens": 328847236.0, + "step": 12813 + }, + { + "epoch": 1.4072040412914562, + "grad_norm": 2.28861927986145, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7102744579315186, + "num_tokens": 328874339.0, + "step": 12814 + }, + { + "epoch": 1.40731385899407, + "grad_norm": 2.4352738857269287, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7235490083694458, + "num_tokens": 328898062.0, + "step": 12815 + }, + { + "epoch": 1.4074236766966834, + "grad_norm": 2.1662821769714355, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7154327034950256, + "num_tokens": 328926172.0, + "step": 12816 + }, + { + "epoch": 1.4075334943992972, + "grad_norm": 2.402099609375, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7181881070137024, + "num_tokens": 328949929.0, + "step": 12817 + }, + { + "epoch": 1.4076433121019107, + "grad_norm": 2.1004045009613037, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7218229174613953, + "num_tokens": 328977567.0, + "step": 12818 + }, + { + "epoch": 1.4077531298045245, + "grad_norm": 2.0595803260803223, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7043474316596985, + "num_tokens": 329008615.0, + "step": 12819 + }, + { + "epoch": 1.4078629475071383, + "grad_norm": 2.3999972343444824, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.722468376159668, + "num_tokens": 329030355.0, + "step": 12820 + }, + { + "epoch": 1.4079727652097518, + "grad_norm": 2.3213257789611816, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7386451959609985, + "num_tokens": 329052069.0, + "step": 12821 + }, + { + "epoch": 1.4080825829123655, + "grad_norm": 2.156456708908081, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7319981455802917, + "num_tokens": 329077789.0, + "step": 12822 + }, + { + "epoch": 1.408192400614979, + "grad_norm": 2.228440046310425, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7452147006988525, + "num_tokens": 329100853.0, + "step": 12823 + }, + { + "epoch": 1.4083022183175928, + "grad_norm": 2.5378098487854004, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7428581714630127, + "num_tokens": 329122317.0, + "step": 12824 + }, + { + "epoch": 1.4084120360202066, + "grad_norm": 2.758373737335205, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7321692109107971, + "num_tokens": 329140239.0, + "step": 12825 + }, + { + "epoch": 1.4085218537228201, + "grad_norm": 2.2021191120147705, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7189499139785767, + "num_tokens": 329166707.0, + "step": 12826 + }, + { + "epoch": 1.4086316714254337, + "grad_norm": 2.441291332244873, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.711355447769165, + "num_tokens": 329189357.0, + "step": 12827 + }, + { + "epoch": 1.4087414891280474, + "grad_norm": 2.1460518836975098, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7169239521026611, + "num_tokens": 329217611.0, + "step": 12828 + }, + { + "epoch": 1.4088513068306612, + "grad_norm": 1.957504391670227, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.708321213722229, + "num_tokens": 329251617.0, + "step": 12829 + }, + { + "epoch": 1.4089611245332747, + "grad_norm": 2.363081693649292, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.738461971282959, + "num_tokens": 329274922.0, + "step": 12830 + }, + { + "epoch": 1.4090709422358885, + "grad_norm": 2.2255778312683105, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7063800096511841, + "num_tokens": 329301021.0, + "step": 12831 + }, + { + "epoch": 1.409180759938502, + "grad_norm": 2.3142549991607666, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7191159725189209, + "num_tokens": 329324990.0, + "step": 12832 + }, + { + "epoch": 1.4092905776411158, + "grad_norm": 2.0491220951080322, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7134000658988953, + "num_tokens": 329354387.0, + "step": 12833 + }, + { + "epoch": 1.4094003953437295, + "grad_norm": 2.545930862426758, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7299580574035645, + "num_tokens": 329374455.0, + "step": 12834 + }, + { + "epoch": 1.409510213046343, + "grad_norm": 2.312135934829712, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7338773012161255, + "num_tokens": 329398077.0, + "step": 12835 + }, + { + "epoch": 1.4096200307489568, + "grad_norm": 2.154655694961548, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7304860353469849, + "num_tokens": 329424030.0, + "step": 12836 + }, + { + "epoch": 1.4097298484515703, + "grad_norm": 2.4555583000183105, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7197830677032471, + "num_tokens": 329445938.0, + "step": 12837 + }, + { + "epoch": 1.409839666154184, + "grad_norm": 2.028810739517212, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7188135385513306, + "num_tokens": 329475940.0, + "step": 12838 + }, + { + "epoch": 1.4099494838567976, + "grad_norm": 2.201557159423828, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7481586933135986, + "num_tokens": 329500716.0, + "step": 12839 + }, + { + "epoch": 1.4100593015594114, + "grad_norm": 2.0648193359375, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.7004346251487732, + "num_tokens": 329531157.0, + "step": 12840 + }, + { + "epoch": 1.410169119262025, + "grad_norm": 2.2813644409179688, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7292673587799072, + "num_tokens": 329555486.0, + "step": 12841 + }, + { + "epoch": 1.4102789369646387, + "grad_norm": 2.053368091583252, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7034841775894165, + "num_tokens": 329587381.0, + "step": 12842 + }, + { + "epoch": 1.4103887546672524, + "grad_norm": 2.0703954696655273, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6948933601379395, + "num_tokens": 329615677.0, + "step": 12843 + }, + { + "epoch": 1.410498572369866, + "grad_norm": 2.227134943008423, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.704779863357544, + "num_tokens": 329642282.0, + "step": 12844 + }, + { + "epoch": 1.4106083900724797, + "grad_norm": 2.033280849456787, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7033985257148743, + "num_tokens": 329671747.0, + "step": 12845 + }, + { + "epoch": 1.4107182077750933, + "grad_norm": 2.258408308029175, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7093379497528076, + "num_tokens": 329698923.0, + "step": 12846 + }, + { + "epoch": 1.410828025477707, + "grad_norm": 2.0237374305725098, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7111202478408813, + "num_tokens": 329728909.0, + "step": 12847 + }, + { + "epoch": 1.4109378431803208, + "grad_norm": 2.3029816150665283, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7027256488800049, + "num_tokens": 329754568.0, + "step": 12848 + }, + { + "epoch": 1.4110476608829343, + "grad_norm": 2.1554744243621826, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7130953073501587, + "num_tokens": 329783791.0, + "step": 12849 + }, + { + "epoch": 1.4111574785855479, + "grad_norm": 2.1273648738861084, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6974329352378845, + "num_tokens": 329813275.0, + "step": 12850 + }, + { + "epoch": 1.4112672962881616, + "grad_norm": 2.325227737426758, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.721561074256897, + "num_tokens": 329837735.0, + "step": 12851 + }, + { + "epoch": 1.4113771139907754, + "grad_norm": 2.5087223052978516, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7318791747093201, + "num_tokens": 329862024.0, + "step": 12852 + }, + { + "epoch": 1.411486931693389, + "grad_norm": 2.193540573120117, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7104614973068237, + "num_tokens": 329890779.0, + "step": 12853 + }, + { + "epoch": 1.4115967493960027, + "grad_norm": 2.5134198665618896, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7397218346595764, + "num_tokens": 329911546.0, + "step": 12854 + }, + { + "epoch": 1.4117065670986162, + "grad_norm": 2.109184503555298, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.731622576713562, + "num_tokens": 329939665.0, + "step": 12855 + }, + { + "epoch": 1.41181638480123, + "grad_norm": 1.991441249847412, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7157896161079407, + "num_tokens": 329972692.0, + "step": 12856 + }, + { + "epoch": 1.4119262025038437, + "grad_norm": 1.9076751470565796, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.730968713760376, + "num_tokens": 330006326.0, + "step": 12857 + }, + { + "epoch": 1.4120360202064572, + "grad_norm": 2.433852434158325, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7340129613876343, + "num_tokens": 330027675.0, + "step": 12858 + }, + { + "epoch": 1.412145837909071, + "grad_norm": 2.345946788787842, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7209566831588745, + "num_tokens": 330052656.0, + "step": 12859 + }, + { + "epoch": 1.4122556556116845, + "grad_norm": 2.082001209259033, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6880304217338562, + "num_tokens": 330084567.0, + "step": 12860 + }, + { + "epoch": 1.4123654733142983, + "grad_norm": 2.0994904041290283, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7143777012825012, + "num_tokens": 330114320.0, + "step": 12861 + }, + { + "epoch": 1.412475291016912, + "grad_norm": 2.28452730178833, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7090867161750793, + "num_tokens": 330139129.0, + "step": 12862 + }, + { + "epoch": 1.4125851087195256, + "grad_norm": 2.328397750854492, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7158083319664001, + "num_tokens": 330162574.0, + "step": 12863 + }, + { + "epoch": 1.4126949264221391, + "grad_norm": 2.1475343704223633, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7181304693222046, + "num_tokens": 330190876.0, + "step": 12864 + }, + { + "epoch": 1.4128047441247529, + "grad_norm": 2.3989038467407227, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7227213382720947, + "num_tokens": 330213507.0, + "step": 12865 + }, + { + "epoch": 1.4129145618273666, + "grad_norm": 2.228024482727051, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7232729196548462, + "num_tokens": 330238428.0, + "step": 12866 + }, + { + "epoch": 1.4130243795299802, + "grad_norm": 2.360560655593872, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7325100302696228, + "num_tokens": 330263147.0, + "step": 12867 + }, + { + "epoch": 1.413134197232594, + "grad_norm": 2.2047007083892822, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7308258414268494, + "num_tokens": 330291028.0, + "step": 12868 + }, + { + "epoch": 1.4132440149352075, + "grad_norm": 1.8353878259658813, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7166790962219238, + "num_tokens": 330329961.0, + "step": 12869 + }, + { + "epoch": 1.4133538326378212, + "grad_norm": 1.971112608909607, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6884702444076538, + "num_tokens": 330364566.0, + "step": 12870 + }, + { + "epoch": 1.413463650340435, + "grad_norm": 2.290252685546875, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7083379030227661, + "num_tokens": 330390165.0, + "step": 12871 + }, + { + "epoch": 1.4135734680430485, + "grad_norm": 2.552025556564331, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7162426710128784, + "num_tokens": 330410417.0, + "step": 12872 + }, + { + "epoch": 1.4136832857456623, + "grad_norm": 2.2981324195861816, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7255387902259827, + "num_tokens": 330434556.0, + "step": 12873 + }, + { + "epoch": 1.4137931034482758, + "grad_norm": 2.0976743698120117, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7457680702209473, + "num_tokens": 330462050.0, + "step": 12874 + }, + { + "epoch": 1.4139029211508896, + "grad_norm": 2.349445104598999, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7318068742752075, + "num_tokens": 330484667.0, + "step": 12875 + }, + { + "epoch": 1.4140127388535033, + "grad_norm": 2.353872299194336, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7323752641677856, + "num_tokens": 330507087.0, + "step": 12876 + }, + { + "epoch": 1.4141225565561168, + "grad_norm": 1.9650641679763794, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7298129200935364, + "num_tokens": 330536620.0, + "step": 12877 + }, + { + "epoch": 1.4142323742587304, + "grad_norm": 2.391963481903076, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7278872728347778, + "num_tokens": 330560706.0, + "step": 12878 + }, + { + "epoch": 1.4143421919613441, + "grad_norm": 2.314136028289795, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7107798457145691, + "num_tokens": 330586773.0, + "step": 12879 + }, + { + "epoch": 1.414452009663958, + "grad_norm": 2.1019504070281982, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.728623628616333, + "num_tokens": 330614482.0, + "step": 12880 + }, + { + "epoch": 1.4145618273665714, + "grad_norm": 1.8745460510253906, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7017365097999573, + "num_tokens": 330648279.0, + "step": 12881 + }, + { + "epoch": 1.4146716450691852, + "grad_norm": 2.1544008255004883, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7129741907119751, + "num_tokens": 330675021.0, + "step": 12882 + }, + { + "epoch": 1.4147814627717987, + "grad_norm": 2.4461097717285156, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7206487655639648, + "num_tokens": 330698464.0, + "step": 12883 + }, + { + "epoch": 1.4148912804744125, + "grad_norm": 1.9599517583847046, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6991616487503052, + "num_tokens": 330731761.0, + "step": 12884 + }, + { + "epoch": 1.4150010981770262, + "grad_norm": 2.3722596168518066, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7344794869422913, + "num_tokens": 330755366.0, + "step": 12885 + }, + { + "epoch": 1.4151109158796398, + "grad_norm": 2.0854954719543457, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7047742605209351, + "num_tokens": 330786353.0, + "step": 12886 + }, + { + "epoch": 1.4152207335822535, + "grad_norm": 2.2984375953674316, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7213375568389893, + "num_tokens": 330811963.0, + "step": 12887 + }, + { + "epoch": 1.415330551284867, + "grad_norm": 2.0825018882751465, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.714468240737915, + "num_tokens": 330840125.0, + "step": 12888 + }, + { + "epoch": 1.4154403689874808, + "grad_norm": 2.084028720855713, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7571828365325928, + "num_tokens": 330867999.0, + "step": 12889 + }, + { + "epoch": 1.4155501866900946, + "grad_norm": 2.2502176761627197, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7331609725952148, + "num_tokens": 330892911.0, + "step": 12890 + }, + { + "epoch": 1.4156600043927081, + "grad_norm": 2.205096960067749, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7244151830673218, + "num_tokens": 330919409.0, + "step": 12891 + }, + { + "epoch": 1.4157698220953217, + "grad_norm": 1.8584105968475342, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7260458469390869, + "num_tokens": 330953595.0, + "step": 12892 + }, + { + "epoch": 1.4158796397979354, + "grad_norm": 2.2805211544036865, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7156535387039185, + "num_tokens": 330977170.0, + "step": 12893 + }, + { + "epoch": 1.4159894575005492, + "grad_norm": 2.1250159740448, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7157939672470093, + "num_tokens": 331005173.0, + "step": 12894 + }, + { + "epoch": 1.4160992752031627, + "grad_norm": 2.4192798137664795, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.710896372795105, + "num_tokens": 331028863.0, + "step": 12895 + }, + { + "epoch": 1.4162090929057765, + "grad_norm": 2.429121732711792, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7149333357810974, + "num_tokens": 331051221.0, + "step": 12896 + }, + { + "epoch": 1.41631891060839, + "grad_norm": 2.0868418216705322, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7036499977111816, + "num_tokens": 331080883.0, + "step": 12897 + }, + { + "epoch": 1.4164287283110037, + "grad_norm": 2.3301258087158203, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7327914237976074, + "num_tokens": 331103423.0, + "step": 12898 + }, + { + "epoch": 1.4165385460136175, + "grad_norm": 2.0170578956604004, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7071629166603088, + "num_tokens": 331136771.0, + "step": 12899 + }, + { + "epoch": 1.416648363716231, + "grad_norm": 2.2894864082336426, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7021750211715698, + "num_tokens": 331161641.0, + "step": 12900 + }, + { + "epoch": 1.4167581814188448, + "grad_norm": 2.1487529277801514, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.720963180065155, + "num_tokens": 331188430.0, + "step": 12901 + }, + { + "epoch": 1.4168679991214583, + "grad_norm": 1.9891332387924194, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7075679302215576, + "num_tokens": 331221374.0, + "step": 12902 + }, + { + "epoch": 1.416977816824072, + "grad_norm": 2.147005796432495, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7072423696517944, + "num_tokens": 331249621.0, + "step": 12903 + }, + { + "epoch": 1.4170876345266856, + "grad_norm": 2.407102108001709, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7322404384613037, + "num_tokens": 331270617.0, + "step": 12904 + }, + { + "epoch": 1.4171974522292994, + "grad_norm": 2.093388557434082, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7221781611442566, + "num_tokens": 331297458.0, + "step": 12905 + }, + { + "epoch": 1.417307269931913, + "grad_norm": 2.210653305053711, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7247488498687744, + "num_tokens": 331322283.0, + "step": 12906 + }, + { + "epoch": 1.4174170876345267, + "grad_norm": 2.1080055236816406, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7457863092422485, + "num_tokens": 331350480.0, + "step": 12907 + }, + { + "epoch": 1.4175269053371404, + "grad_norm": 2.3314859867095947, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7416789531707764, + "num_tokens": 331371411.0, + "step": 12908 + }, + { + "epoch": 1.417636723039754, + "grad_norm": 2.34913969039917, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7129684686660767, + "num_tokens": 331395284.0, + "step": 12909 + }, + { + "epoch": 1.4177465407423677, + "grad_norm": 1.957118272781372, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7169665694236755, + "num_tokens": 331429471.0, + "step": 12910 + }, + { + "epoch": 1.4178563584449813, + "grad_norm": 2.12418532371521, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7191128134727478, + "num_tokens": 331460436.0, + "step": 12911 + }, + { + "epoch": 1.417966176147595, + "grad_norm": 2.279714584350586, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7104581594467163, + "num_tokens": 331485076.0, + "step": 12912 + }, + { + "epoch": 1.4180759938502088, + "grad_norm": 2.491021156311035, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7169525623321533, + "num_tokens": 331505261.0, + "step": 12913 + }, + { + "epoch": 1.4181858115528223, + "grad_norm": 2.2281134128570557, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7109200954437256, + "num_tokens": 331534677.0, + "step": 12914 + }, + { + "epoch": 1.4182956292554358, + "grad_norm": 2.1308815479278564, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.708419680595398, + "num_tokens": 331562272.0, + "step": 12915 + }, + { + "epoch": 1.4184054469580496, + "grad_norm": 2.2673988342285156, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7236685156822205, + "num_tokens": 331586422.0, + "step": 12916 + }, + { + "epoch": 1.4185152646606634, + "grad_norm": 2.298362970352173, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7313359975814819, + "num_tokens": 331610994.0, + "step": 12917 + }, + { + "epoch": 1.418625082363277, + "grad_norm": 2.277042865753174, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.717165470123291, + "num_tokens": 331637226.0, + "step": 12918 + }, + { + "epoch": 1.4187349000658906, + "grad_norm": 2.138627767562866, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7223196029663086, + "num_tokens": 331664675.0, + "step": 12919 + }, + { + "epoch": 1.4188447177685042, + "grad_norm": 2.273282289505005, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7385832667350769, + "num_tokens": 331689606.0, + "step": 12920 + }, + { + "epoch": 1.418954535471118, + "grad_norm": 2.0915069580078125, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.71031653881073, + "num_tokens": 331717613.0, + "step": 12921 + }, + { + "epoch": 1.4190643531737317, + "grad_norm": 2.0657098293304443, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7335871458053589, + "num_tokens": 331744614.0, + "step": 12922 + }, + { + "epoch": 1.4191741708763452, + "grad_norm": 2.2295892238616943, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7236173748970032, + "num_tokens": 331769117.0, + "step": 12923 + }, + { + "epoch": 1.419283988578959, + "grad_norm": 2.053849935531616, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6966308951377869, + "num_tokens": 331800718.0, + "step": 12924 + }, + { + "epoch": 1.4193938062815725, + "grad_norm": 2.104905843734741, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7375577688217163, + "num_tokens": 331826806.0, + "step": 12925 + }, + { + "epoch": 1.4195036239841863, + "grad_norm": 2.3747661113739014, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.721778929233551, + "num_tokens": 331849743.0, + "step": 12926 + }, + { + "epoch": 1.4196134416868, + "grad_norm": 2.0710151195526123, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7183168530464172, + "num_tokens": 331878800.0, + "step": 12927 + }, + { + "epoch": 1.4197232593894136, + "grad_norm": 2.1675326824188232, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7222859859466553, + "num_tokens": 331905490.0, + "step": 12928 + }, + { + "epoch": 1.419833077092027, + "grad_norm": 2.171783685684204, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7162023782730103, + "num_tokens": 331931366.0, + "step": 12929 + }, + { + "epoch": 1.4199428947946409, + "grad_norm": 1.954646348953247, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7205784320831299, + "num_tokens": 331963104.0, + "step": 12930 + }, + { + "epoch": 1.4200527124972546, + "grad_norm": 2.178274393081665, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7295365333557129, + "num_tokens": 331988019.0, + "step": 12931 + }, + { + "epoch": 1.4201625301998682, + "grad_norm": 2.466052532196045, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7261843681335449, + "num_tokens": 332009147.0, + "step": 12932 + }, + { + "epoch": 1.420272347902482, + "grad_norm": 2.0355851650238037, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.686613917350769, + "num_tokens": 332040982.0, + "step": 12933 + }, + { + "epoch": 1.4203821656050954, + "grad_norm": 2.225714921951294, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7147902250289917, + "num_tokens": 332068148.0, + "step": 12934 + }, + { + "epoch": 1.4204919833077092, + "grad_norm": 2.4284794330596924, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7169467210769653, + "num_tokens": 332090210.0, + "step": 12935 + }, + { + "epoch": 1.420601801010323, + "grad_norm": 2.429560422897339, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7335197925567627, + "num_tokens": 332111880.0, + "step": 12936 + }, + { + "epoch": 1.4207116187129365, + "grad_norm": 2.211869955062866, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7151227593421936, + "num_tokens": 332141743.0, + "step": 12937 + }, + { + "epoch": 1.4208214364155503, + "grad_norm": 2.9032790660858154, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7169026136398315, + "num_tokens": 332159880.0, + "step": 12938 + }, + { + "epoch": 1.4209312541181638, + "grad_norm": 2.5976998805999756, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7130513191223145, + "num_tokens": 332181511.0, + "step": 12939 + }, + { + "epoch": 1.4210410718207775, + "grad_norm": 2.148207426071167, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7048628926277161, + "num_tokens": 332207954.0, + "step": 12940 + }, + { + "epoch": 1.4211508895233913, + "grad_norm": 2.062373161315918, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.697179913520813, + "num_tokens": 332239806.0, + "step": 12941 + }, + { + "epoch": 1.4212607072260048, + "grad_norm": 2.387556552886963, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7168399691581726, + "num_tokens": 332264246.0, + "step": 12942 + }, + { + "epoch": 1.4213705249286184, + "grad_norm": 2.2648589611053467, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7264896631240845, + "num_tokens": 332291470.0, + "step": 12943 + }, + { + "epoch": 1.4214803426312321, + "grad_norm": 2.1085355281829834, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.739368200302124, + "num_tokens": 332318868.0, + "step": 12944 + }, + { + "epoch": 1.4215901603338459, + "grad_norm": 2.3113794326782227, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.757880687713623, + "num_tokens": 332340978.0, + "step": 12945 + }, + { + "epoch": 1.4216999780364594, + "grad_norm": 2.3055014610290527, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.689089298248291, + "num_tokens": 332367683.0, + "step": 12946 + }, + { + "epoch": 1.4218097957390732, + "grad_norm": 2.3815107345581055, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7141114473342896, + "num_tokens": 332391702.0, + "step": 12947 + }, + { + "epoch": 1.4219196134416867, + "grad_norm": 2.431295871734619, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7359585762023926, + "num_tokens": 332413836.0, + "step": 12948 + }, + { + "epoch": 1.4220294311443005, + "grad_norm": 2.2948577404022217, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7054991126060486, + "num_tokens": 332441038.0, + "step": 12949 + }, + { + "epoch": 1.4221392488469142, + "grad_norm": 2.5356051921844482, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7035195827484131, + "num_tokens": 332463753.0, + "step": 12950 + }, + { + "epoch": 1.4222490665495278, + "grad_norm": 2.1847434043884277, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7240968942642212, + "num_tokens": 332488403.0, + "step": 12951 + }, + { + "epoch": 1.4223588842521415, + "grad_norm": 2.483128309249878, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7276074290275574, + "num_tokens": 332508994.0, + "step": 12952 + }, + { + "epoch": 1.422468701954755, + "grad_norm": 2.360966205596924, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7150731086730957, + "num_tokens": 332533132.0, + "step": 12953 + }, + { + "epoch": 1.4225785196573688, + "grad_norm": 2.650686502456665, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7418190240859985, + "num_tokens": 332551012.0, + "step": 12954 + }, + { + "epoch": 1.4226883373599823, + "grad_norm": 2.462238073348999, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7070968151092529, + "num_tokens": 332573486.0, + "step": 12955 + }, + { + "epoch": 1.422798155062596, + "grad_norm": 2.25588321685791, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7126790285110474, + "num_tokens": 332601375.0, + "step": 12956 + }, + { + "epoch": 1.4229079727652096, + "grad_norm": 2.64925217628479, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7279283404350281, + "num_tokens": 332625851.0, + "step": 12957 + }, + { + "epoch": 1.4230177904678234, + "grad_norm": 2.282994270324707, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7070491313934326, + "num_tokens": 332650870.0, + "step": 12958 + }, + { + "epoch": 1.4231276081704372, + "grad_norm": 2.3984713554382324, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7229214906692505, + "num_tokens": 332674132.0, + "step": 12959 + }, + { + "epoch": 1.4232374258730507, + "grad_norm": 2.4664852619171143, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7140975594520569, + "num_tokens": 332696132.0, + "step": 12960 + }, + { + "epoch": 1.4233472435756644, + "grad_norm": 2.1892547607421875, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7178863286972046, + "num_tokens": 332722978.0, + "step": 12961 + }, + { + "epoch": 1.423457061278278, + "grad_norm": 2.306596040725708, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7280606031417847, + "num_tokens": 332747541.0, + "step": 12962 + }, + { + "epoch": 1.4235668789808917, + "grad_norm": 2.392836093902588, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.732380211353302, + "num_tokens": 332769913.0, + "step": 12963 + }, + { + "epoch": 1.4236766966835055, + "grad_norm": 2.185260057449341, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7327117919921875, + "num_tokens": 332795302.0, + "step": 12964 + }, + { + "epoch": 1.423786514386119, + "grad_norm": 2.428001880645752, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7119594812393188, + "num_tokens": 332819305.0, + "step": 12965 + }, + { + "epoch": 1.4238963320887328, + "grad_norm": 2.5002591609954834, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7425476908683777, + "num_tokens": 332840834.0, + "step": 12966 + }, + { + "epoch": 1.4240061497913463, + "grad_norm": 2.307680368423462, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7226837277412415, + "num_tokens": 332865821.0, + "step": 12967 + }, + { + "epoch": 1.42411596749396, + "grad_norm": 2.174976348876953, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7127458453178406, + "num_tokens": 332893423.0, + "step": 12968 + }, + { + "epoch": 1.4242257851965736, + "grad_norm": 2.248518943786621, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7159628868103027, + "num_tokens": 332918583.0, + "step": 12969 + }, + { + "epoch": 1.4243356028991874, + "grad_norm": 2.153765916824341, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.701356053352356, + "num_tokens": 332944872.0, + "step": 12970 + }, + { + "epoch": 1.424445420601801, + "grad_norm": 2.360888957977295, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.70240318775177, + "num_tokens": 332969198.0, + "step": 12971 + }, + { + "epoch": 1.4245552383044147, + "grad_norm": 2.309607982635498, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7189607620239258, + "num_tokens": 332992439.0, + "step": 12972 + }, + { + "epoch": 1.4246650560070284, + "grad_norm": 2.4394872188568115, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.731621503829956, + "num_tokens": 333014165.0, + "step": 12973 + }, + { + "epoch": 1.424774873709642, + "grad_norm": 1.8567622900009155, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7178022861480713, + "num_tokens": 333047163.0, + "step": 12974 + }, + { + "epoch": 1.4248846914122557, + "grad_norm": 2.488287925720215, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.720949113368988, + "num_tokens": 333071635.0, + "step": 12975 + }, + { + "epoch": 1.4249945091148692, + "grad_norm": 2.5609540939331055, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7036753296852112, + "num_tokens": 333095284.0, + "step": 12976 + }, + { + "epoch": 1.425104326817483, + "grad_norm": 2.3376147747039795, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7000902891159058, + "num_tokens": 333119766.0, + "step": 12977 + }, + { + "epoch": 1.4252141445200968, + "grad_norm": 2.3385026454925537, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7113533020019531, + "num_tokens": 333143800.0, + "step": 12978 + }, + { + "epoch": 1.4253239622227103, + "grad_norm": 2.5029850006103516, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7250916361808777, + "num_tokens": 333165390.0, + "step": 12979 + }, + { + "epoch": 1.4254337799253238, + "grad_norm": 2.482454538345337, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.728935718536377, + "num_tokens": 333187265.0, + "step": 12980 + }, + { + "epoch": 1.4255435976279376, + "grad_norm": 2.4773056507110596, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7143869400024414, + "num_tokens": 333208630.0, + "step": 12981 + }, + { + "epoch": 1.4256534153305513, + "grad_norm": 2.622518301010132, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7332117557525635, + "num_tokens": 333229299.0, + "step": 12982 + }, + { + "epoch": 1.4257632330331649, + "grad_norm": 2.1948769092559814, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7260950803756714, + "num_tokens": 333256200.0, + "step": 12983 + }, + { + "epoch": 1.4258730507357786, + "grad_norm": 2.2277183532714844, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7237893342971802, + "num_tokens": 333283203.0, + "step": 12984 + }, + { + "epoch": 1.4259828684383922, + "grad_norm": 2.202699661254883, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6945270299911499, + "num_tokens": 333311550.0, + "step": 12985 + }, + { + "epoch": 1.426092686141006, + "grad_norm": 2.0876753330230713, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7104212045669556, + "num_tokens": 333342568.0, + "step": 12986 + }, + { + "epoch": 1.4262025038436197, + "grad_norm": 2.3083457946777344, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.736233115196228, + "num_tokens": 333367202.0, + "step": 12987 + }, + { + "epoch": 1.4263123215462332, + "grad_norm": 2.2543742656707764, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7234026193618774, + "num_tokens": 333393002.0, + "step": 12988 + }, + { + "epoch": 1.426422139248847, + "grad_norm": 2.5475499629974365, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7293365001678467, + "num_tokens": 333413625.0, + "step": 12989 + }, + { + "epoch": 1.4265319569514605, + "grad_norm": 2.092463731765747, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7021987438201904, + "num_tokens": 333443085.0, + "step": 12990 + }, + { + "epoch": 1.4266417746540743, + "grad_norm": 2.3944249153137207, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6932903528213501, + "num_tokens": 333468166.0, + "step": 12991 + }, + { + "epoch": 1.426751592356688, + "grad_norm": 2.3663742542266846, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7297787070274353, + "num_tokens": 333492254.0, + "step": 12992 + }, + { + "epoch": 1.4268614100593016, + "grad_norm": 2.258204698562622, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7241562008857727, + "num_tokens": 333516916.0, + "step": 12993 + }, + { + "epoch": 1.426971227761915, + "grad_norm": 2.231738567352295, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7129177451133728, + "num_tokens": 333542332.0, + "step": 12994 + }, + { + "epoch": 1.4270810454645289, + "grad_norm": 2.6074411869049072, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7293788194656372, + "num_tokens": 333560608.0, + "step": 12995 + }, + { + "epoch": 1.4271908631671426, + "grad_norm": 2.3662991523742676, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7309308052062988, + "num_tokens": 333584579.0, + "step": 12996 + }, + { + "epoch": 1.4273006808697561, + "grad_norm": 2.289999485015869, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6765252351760864, + "num_tokens": 333612162.0, + "step": 12997 + }, + { + "epoch": 1.42741049857237, + "grad_norm": 2.1872806549072266, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7189623713493347, + "num_tokens": 333641804.0, + "step": 12998 + }, + { + "epoch": 1.4275203162749834, + "grad_norm": 2.410017967224121, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7089509963989258, + "num_tokens": 333665648.0, + "step": 12999 + }, + { + "epoch": 1.4276301339775972, + "grad_norm": 2.2532382011413574, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7295025587081909, + "num_tokens": 333689645.0, + "step": 13000 + }, + { + "epoch": 1.427739951680211, + "grad_norm": 2.0904126167297363, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7162368893623352, + "num_tokens": 333717168.0, + "step": 13001 + }, + { + "epoch": 1.4278497693828245, + "grad_norm": 2.2412021160125732, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7314797639846802, + "num_tokens": 333741166.0, + "step": 13002 + }, + { + "epoch": 1.4279595870854382, + "grad_norm": 2.2755792140960693, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7282894849777222, + "num_tokens": 333764607.0, + "step": 13003 + }, + { + "epoch": 1.4280694047880518, + "grad_norm": 2.372734308242798, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7052649259567261, + "num_tokens": 333789009.0, + "step": 13004 + }, + { + "epoch": 1.4281792224906655, + "grad_norm": 2.1011698246002197, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7062922716140747, + "num_tokens": 333817517.0, + "step": 13005 + }, + { + "epoch": 1.4282890401932793, + "grad_norm": 2.167783498764038, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7050677537918091, + "num_tokens": 333845718.0, + "step": 13006 + }, + { + "epoch": 1.4283988578958928, + "grad_norm": 2.019484758377075, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7130428552627563, + "num_tokens": 333876021.0, + "step": 13007 + }, + { + "epoch": 1.4285086755985064, + "grad_norm": 2.3816709518432617, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7213760018348694, + "num_tokens": 333898473.0, + "step": 13008 + }, + { + "epoch": 1.4286184933011201, + "grad_norm": 2.1506381034851074, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7141151428222656, + "num_tokens": 333926506.0, + "step": 13009 + }, + { + "epoch": 1.4287283110037339, + "grad_norm": 2.252683639526367, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6999778747558594, + "num_tokens": 333952624.0, + "step": 13010 + }, + { + "epoch": 1.4288381287063474, + "grad_norm": 2.1942479610443115, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7251316905021667, + "num_tokens": 333978960.0, + "step": 13011 + }, + { + "epoch": 1.4289479464089612, + "grad_norm": 2.209117889404297, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7137846946716309, + "num_tokens": 334004340.0, + "step": 13012 + }, + { + "epoch": 1.4290577641115747, + "grad_norm": 2.1269917488098145, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.69725102186203, + "num_tokens": 334034301.0, + "step": 13013 + }, + { + "epoch": 1.4291675818141885, + "grad_norm": 2.309675931930542, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.709684431552887, + "num_tokens": 334059842.0, + "step": 13014 + }, + { + "epoch": 1.4292773995168022, + "grad_norm": 2.351039171218872, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.714418351650238, + "num_tokens": 334082627.0, + "step": 13015 + }, + { + "epoch": 1.4293872172194158, + "grad_norm": 2.4907517433166504, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7176567316055298, + "num_tokens": 334107934.0, + "step": 13016 + }, + { + "epoch": 1.4294970349220295, + "grad_norm": 2.155832290649414, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7097857594490051, + "num_tokens": 334137037.0, + "step": 13017 + }, + { + "epoch": 1.429606852624643, + "grad_norm": 2.547947645187378, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7350168228149414, + "num_tokens": 334157630.0, + "step": 13018 + }, + { + "epoch": 1.4297166703272568, + "grad_norm": 2.365462064743042, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7438787817955017, + "num_tokens": 334179400.0, + "step": 13019 + }, + { + "epoch": 1.4298264880298703, + "grad_norm": 2.058936357498169, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7189176678657532, + "num_tokens": 334208391.0, + "step": 13020 + }, + { + "epoch": 1.429936305732484, + "grad_norm": 2.479259729385376, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7158298492431641, + "num_tokens": 334231207.0, + "step": 13021 + }, + { + "epoch": 1.4300461234350976, + "grad_norm": 2.0612952709198, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7209651470184326, + "num_tokens": 334261449.0, + "step": 13022 + }, + { + "epoch": 1.4301559411377114, + "grad_norm": 2.3321001529693604, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7025570869445801, + "num_tokens": 334287360.0, + "step": 13023 + }, + { + "epoch": 1.4302657588403251, + "grad_norm": 2.428896188735962, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.754427433013916, + "num_tokens": 334307575.0, + "step": 13024 + }, + { + "epoch": 1.4303755765429387, + "grad_norm": 2.412548303604126, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.72206711769104, + "num_tokens": 334331068.0, + "step": 13025 + }, + { + "epoch": 1.4304853942455524, + "grad_norm": 2.0556986331939697, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7123077511787415, + "num_tokens": 334360914.0, + "step": 13026 + }, + { + "epoch": 1.430595211948166, + "grad_norm": 2.276554584503174, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7357932925224304, + "num_tokens": 334384835.0, + "step": 13027 + }, + { + "epoch": 1.4307050296507797, + "grad_norm": 2.1660919189453125, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7263402938842773, + "num_tokens": 334412076.0, + "step": 13028 + }, + { + "epoch": 1.4308148473533935, + "grad_norm": 2.0295395851135254, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6958028078079224, + "num_tokens": 334444674.0, + "step": 13029 + }, + { + "epoch": 1.430924665056007, + "grad_norm": 2.050638198852539, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.705142617225647, + "num_tokens": 334474359.0, + "step": 13030 + }, + { + "epoch": 1.4310344827586206, + "grad_norm": 2.3724966049194336, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7193339467048645, + "num_tokens": 334497624.0, + "step": 13031 + }, + { + "epoch": 1.4311443004612343, + "grad_norm": 2.0828664302825928, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7344459891319275, + "num_tokens": 334525017.0, + "step": 13032 + }, + { + "epoch": 1.431254118163848, + "grad_norm": 2.154207944869995, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7058813571929932, + "num_tokens": 334552104.0, + "step": 13033 + }, + { + "epoch": 1.4313639358664616, + "grad_norm": 2.2459185123443604, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7087308168411255, + "num_tokens": 334575968.0, + "step": 13034 + }, + { + "epoch": 1.4314737535690754, + "grad_norm": 2.4903674125671387, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7223086953163147, + "num_tokens": 334599128.0, + "step": 13035 + }, + { + "epoch": 1.431583571271689, + "grad_norm": 2.405231475830078, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7279092669487, + "num_tokens": 334622521.0, + "step": 13036 + }, + { + "epoch": 1.4316933889743026, + "grad_norm": 2.079603910446167, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7174690961837769, + "num_tokens": 334651278.0, + "step": 13037 + }, + { + "epoch": 1.4318032066769164, + "grad_norm": 2.1008450984954834, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.737876832485199, + "num_tokens": 334679624.0, + "step": 13038 + }, + { + "epoch": 1.43191302437953, + "grad_norm": 2.211580753326416, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6981039047241211, + "num_tokens": 334705921.0, + "step": 13039 + }, + { + "epoch": 1.4320228420821437, + "grad_norm": 2.097888708114624, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7298473119735718, + "num_tokens": 334733777.0, + "step": 13040 + }, + { + "epoch": 1.4321326597847572, + "grad_norm": 2.2204458713531494, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6959431767463684, + "num_tokens": 334762447.0, + "step": 13041 + }, + { + "epoch": 1.432242477487371, + "grad_norm": 2.180373191833496, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7182121276855469, + "num_tokens": 334790070.0, + "step": 13042 + }, + { + "epoch": 1.4323522951899847, + "grad_norm": 2.1218159198760986, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7363651990890503, + "num_tokens": 334815834.0, + "step": 13043 + }, + { + "epoch": 1.4324621128925983, + "grad_norm": 2.3450844287872314, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7145313024520874, + "num_tokens": 334839576.0, + "step": 13044 + }, + { + "epoch": 1.4325719305952118, + "grad_norm": 2.3147168159484863, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7234234809875488, + "num_tokens": 334861930.0, + "step": 13045 + }, + { + "epoch": 1.4326817482978256, + "grad_norm": 1.7893602848052979, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7196701765060425, + "num_tokens": 334896959.0, + "step": 13046 + }, + { + "epoch": 1.4327915660004393, + "grad_norm": 2.113121271133423, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7042407989501953, + "num_tokens": 334924085.0, + "step": 13047 + }, + { + "epoch": 1.4329013837030529, + "grad_norm": 2.327387571334839, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7339988946914673, + "num_tokens": 334949003.0, + "step": 13048 + }, + { + "epoch": 1.4330112014056666, + "grad_norm": 2.5874249935150146, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7153782248497009, + "num_tokens": 334970215.0, + "step": 13049 + }, + { + "epoch": 1.4331210191082802, + "grad_norm": 1.901108980178833, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7236225008964539, + "num_tokens": 335004681.0, + "step": 13050 + }, + { + "epoch": 1.433230836810894, + "grad_norm": 2.285722017288208, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7278260588645935, + "num_tokens": 335030212.0, + "step": 13051 + }, + { + "epoch": 1.4333406545135077, + "grad_norm": 2.244507074356079, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7129017114639282, + "num_tokens": 335054933.0, + "step": 13052 + }, + { + "epoch": 1.4334504722161212, + "grad_norm": 2.3831920623779297, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7223632335662842, + "num_tokens": 335078532.0, + "step": 13053 + }, + { + "epoch": 1.433560289918735, + "grad_norm": 2.625715494155884, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7382054328918457, + "num_tokens": 335096664.0, + "step": 13054 + }, + { + "epoch": 1.4336701076213485, + "grad_norm": 2.286983013153076, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.731892466545105, + "num_tokens": 335120768.0, + "step": 13055 + }, + { + "epoch": 1.4337799253239623, + "grad_norm": 2.3422911167144775, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7090607285499573, + "num_tokens": 335145402.0, + "step": 13056 + }, + { + "epoch": 1.433889743026576, + "grad_norm": 2.2717888355255127, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7262462377548218, + "num_tokens": 335169665.0, + "step": 13057 + }, + { + "epoch": 1.4339995607291895, + "grad_norm": 2.3883838653564453, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7145273685455322, + "num_tokens": 335193041.0, + "step": 13058 + }, + { + "epoch": 1.434109378431803, + "grad_norm": 2.2550947666168213, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7196404933929443, + "num_tokens": 335218514.0, + "step": 13059 + }, + { + "epoch": 1.4342191961344168, + "grad_norm": 2.236686944961548, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7111179828643799, + "num_tokens": 335247570.0, + "step": 13060 + }, + { + "epoch": 1.4343290138370306, + "grad_norm": 2.4514143466949463, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7180472612380981, + "num_tokens": 335271968.0, + "step": 13061 + }, + { + "epoch": 1.4344388315396441, + "grad_norm": 2.4841582775115967, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7249654531478882, + "num_tokens": 335296092.0, + "step": 13062 + }, + { + "epoch": 1.434548649242258, + "grad_norm": 2.37115740776062, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7024945020675659, + "num_tokens": 335321978.0, + "step": 13063 + }, + { + "epoch": 1.4346584669448714, + "grad_norm": 2.497541666030884, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.733693540096283, + "num_tokens": 335340375.0, + "step": 13064 + }, + { + "epoch": 1.4347682846474852, + "grad_norm": 2.337958812713623, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.717043399810791, + "num_tokens": 335364733.0, + "step": 13065 + }, + { + "epoch": 1.434878102350099, + "grad_norm": 2.3119053840637207, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7227794528007507, + "num_tokens": 335389583.0, + "step": 13066 + }, + { + "epoch": 1.4349879200527125, + "grad_norm": 2.317978620529175, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7084313631057739, + "num_tokens": 335413437.0, + "step": 13067 + }, + { + "epoch": 1.4350977377553262, + "grad_norm": 2.2739269733428955, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7192678451538086, + "num_tokens": 335437975.0, + "step": 13068 + }, + { + "epoch": 1.4352075554579398, + "grad_norm": 2.851547956466675, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7453306913375854, + "num_tokens": 335456885.0, + "step": 13069 + }, + { + "epoch": 1.4353173731605535, + "grad_norm": 2.047379493713379, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7314869165420532, + "num_tokens": 335486124.0, + "step": 13070 + }, + { + "epoch": 1.4354271908631673, + "grad_norm": 2.116240978240967, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7128661870956421, + "num_tokens": 335516451.0, + "step": 13071 + }, + { + "epoch": 1.4355370085657808, + "grad_norm": 2.4336318969726562, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7195308208465576, + "num_tokens": 335539743.0, + "step": 13072 + }, + { + "epoch": 1.4356468262683943, + "grad_norm": 2.283174991607666, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7135313749313354, + "num_tokens": 335565743.0, + "step": 13073 + }, + { + "epoch": 1.435756643971008, + "grad_norm": 2.308588981628418, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7336759567260742, + "num_tokens": 335586912.0, + "step": 13074 + }, + { + "epoch": 1.4358664616736219, + "grad_norm": 2.221360206604004, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.709617018699646, + "num_tokens": 335613749.0, + "step": 13075 + }, + { + "epoch": 1.4359762793762354, + "grad_norm": 1.9839808940887451, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7328612804412842, + "num_tokens": 335645310.0, + "step": 13076 + }, + { + "epoch": 1.4360860970788492, + "grad_norm": 2.0529637336730957, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6894816160202026, + "num_tokens": 335675910.0, + "step": 13077 + }, + { + "epoch": 1.4361959147814627, + "grad_norm": 2.601048707962036, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7302001714706421, + "num_tokens": 335697040.0, + "step": 13078 + }, + { + "epoch": 1.4363057324840764, + "grad_norm": 2.1931755542755127, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7066840529441833, + "num_tokens": 335723377.0, + "step": 13079 + }, + { + "epoch": 1.4364155501866902, + "grad_norm": 2.5345284938812256, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7210249304771423, + "num_tokens": 335745957.0, + "step": 13080 + }, + { + "epoch": 1.4365253678893037, + "grad_norm": 2.5537285804748535, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7293401956558228, + "num_tokens": 335767355.0, + "step": 13081 + }, + { + "epoch": 1.4366351855919175, + "grad_norm": 2.581881046295166, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7251863479614258, + "num_tokens": 335786381.0, + "step": 13082 + }, + { + "epoch": 1.436745003294531, + "grad_norm": 2.3788094520568848, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7140970230102539, + "num_tokens": 335811523.0, + "step": 13083 + }, + { + "epoch": 1.4368548209971448, + "grad_norm": 2.155384063720703, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7072142958641052, + "num_tokens": 335839296.0, + "step": 13084 + }, + { + "epoch": 1.4369646386997583, + "grad_norm": 2.1806416511535645, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7263170480728149, + "num_tokens": 335865669.0, + "step": 13085 + }, + { + "epoch": 1.437074456402372, + "grad_norm": 2.2944557666778564, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7222083210945129, + "num_tokens": 335891870.0, + "step": 13086 + }, + { + "epoch": 1.4371842741049856, + "grad_norm": 2.1784565448760986, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7104791402816772, + "num_tokens": 335917119.0, + "step": 13087 + }, + { + "epoch": 1.4372940918075994, + "grad_norm": 2.5102522373199463, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.718246579170227, + "num_tokens": 335937877.0, + "step": 13088 + }, + { + "epoch": 1.4374039095102131, + "grad_norm": 2.119983434677124, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7259513735771179, + "num_tokens": 335965413.0, + "step": 13089 + }, + { + "epoch": 1.4375137272128267, + "grad_norm": 2.2926254272460938, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7139949798583984, + "num_tokens": 335994768.0, + "step": 13090 + }, + { + "epoch": 1.4376235449154404, + "grad_norm": 2.429830551147461, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7391443252563477, + "num_tokens": 336017316.0, + "step": 13091 + }, + { + "epoch": 1.437733362618054, + "grad_norm": 2.4841091632843018, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7159632444381714, + "num_tokens": 336039935.0, + "step": 13092 + }, + { + "epoch": 1.4378431803206677, + "grad_norm": 2.3384644985198975, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7066119909286499, + "num_tokens": 336062868.0, + "step": 13093 + }, + { + "epoch": 1.4379529980232815, + "grad_norm": 2.123383045196533, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7185039520263672, + "num_tokens": 336091360.0, + "step": 13094 + }, + { + "epoch": 1.438062815725895, + "grad_norm": 2.153369665145874, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7324062585830688, + "num_tokens": 336118501.0, + "step": 13095 + }, + { + "epoch": 1.4381726334285085, + "grad_norm": 2.3260974884033203, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7238432168960571, + "num_tokens": 336142183.0, + "step": 13096 + }, + { + "epoch": 1.4382824511311223, + "grad_norm": 2.1227846145629883, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.704450249671936, + "num_tokens": 336171436.0, + "step": 13097 + }, + { + "epoch": 1.438392268833736, + "grad_norm": 2.3470704555511475, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7138961553573608, + "num_tokens": 336195049.0, + "step": 13098 + }, + { + "epoch": 1.4385020865363496, + "grad_norm": 2.777029514312744, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7273608446121216, + "num_tokens": 336214540.0, + "step": 13099 + }, + { + "epoch": 1.4386119042389633, + "grad_norm": 2.3809802532196045, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.729170560836792, + "num_tokens": 336237427.0, + "step": 13100 + }, + { + "epoch": 1.4387217219415769, + "grad_norm": 2.245626926422119, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7223007678985596, + "num_tokens": 336262368.0, + "step": 13101 + }, + { + "epoch": 1.4388315396441906, + "grad_norm": 2.124610662460327, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7196540832519531, + "num_tokens": 336291035.0, + "step": 13102 + }, + { + "epoch": 1.4389413573468044, + "grad_norm": 2.2237398624420166, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7174999713897705, + "num_tokens": 336316694.0, + "step": 13103 + }, + { + "epoch": 1.439051175049418, + "grad_norm": 1.949479341506958, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7360478043556213, + "num_tokens": 336347431.0, + "step": 13104 + }, + { + "epoch": 1.4391609927520317, + "grad_norm": 1.9166018962860107, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7401368618011475, + "num_tokens": 336379167.0, + "step": 13105 + }, + { + "epoch": 1.4392708104546452, + "grad_norm": 1.9415128231048584, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7127412557601929, + "num_tokens": 336412257.0, + "step": 13106 + }, + { + "epoch": 1.439380628157259, + "grad_norm": 2.1616547107696533, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7232047915458679, + "num_tokens": 336441305.0, + "step": 13107 + }, + { + "epoch": 1.4394904458598727, + "grad_norm": 2.0200552940368652, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7013295292854309, + "num_tokens": 336470558.0, + "step": 13108 + }, + { + "epoch": 1.4396002635624863, + "grad_norm": 2.129720449447632, + "learning_rate": 1e-06, + "loss": 1.0801, + "mean_token_accuracy": 0.6805224418640137, + "num_tokens": 336498749.0, + "step": 13109 + }, + { + "epoch": 1.4397100812650998, + "grad_norm": 2.158447027206421, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7064462304115295, + "num_tokens": 336526606.0, + "step": 13110 + }, + { + "epoch": 1.4398198989677136, + "grad_norm": 2.155413866043091, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7368566393852234, + "num_tokens": 336550870.0, + "step": 13111 + }, + { + "epoch": 1.4399297166703273, + "grad_norm": 2.143381357192993, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7159678339958191, + "num_tokens": 336579032.0, + "step": 13112 + }, + { + "epoch": 1.4400395343729409, + "grad_norm": 2.2010421752929688, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7123445272445679, + "num_tokens": 336604292.0, + "step": 13113 + }, + { + "epoch": 1.4401493520755546, + "grad_norm": 2.3035476207733154, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7116317749023438, + "num_tokens": 336630463.0, + "step": 13114 + }, + { + "epoch": 1.4402591697781681, + "grad_norm": 2.1018264293670654, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7114312648773193, + "num_tokens": 336657787.0, + "step": 13115 + }, + { + "epoch": 1.440368987480782, + "grad_norm": 2.3801255226135254, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7246041297912598, + "num_tokens": 336681692.0, + "step": 13116 + }, + { + "epoch": 1.4404788051833957, + "grad_norm": 2.40872859954834, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.737450361251831, + "num_tokens": 336703501.0, + "step": 13117 + }, + { + "epoch": 1.4405886228860092, + "grad_norm": 2.2241134643554688, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7199327349662781, + "num_tokens": 336728297.0, + "step": 13118 + }, + { + "epoch": 1.440698440588623, + "grad_norm": 2.201911449432373, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.697114109992981, + "num_tokens": 336757073.0, + "step": 13119 + }, + { + "epoch": 1.4408082582912365, + "grad_norm": 2.576857089996338, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.720257043838501, + "num_tokens": 336778150.0, + "step": 13120 + }, + { + "epoch": 1.4409180759938502, + "grad_norm": 2.3656094074249268, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7129730582237244, + "num_tokens": 336802514.0, + "step": 13121 + }, + { + "epoch": 1.441027893696464, + "grad_norm": 2.2098734378814697, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7094616889953613, + "num_tokens": 336828557.0, + "step": 13122 + }, + { + "epoch": 1.4411377113990775, + "grad_norm": 2.4039509296417236, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7254500389099121, + "num_tokens": 336850552.0, + "step": 13123 + }, + { + "epoch": 1.441247529101691, + "grad_norm": 2.1183536052703857, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7136901021003723, + "num_tokens": 336878152.0, + "step": 13124 + }, + { + "epoch": 1.4413573468043048, + "grad_norm": 2.3587474822998047, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7017546892166138, + "num_tokens": 336902539.0, + "step": 13125 + }, + { + "epoch": 1.4414671645069186, + "grad_norm": 2.0685205459594727, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7166708707809448, + "num_tokens": 336930911.0, + "step": 13126 + }, + { + "epoch": 1.4415769822095321, + "grad_norm": 2.5197579860687256, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7197751402854919, + "num_tokens": 336954982.0, + "step": 13127 + }, + { + "epoch": 1.4416867999121459, + "grad_norm": 2.2167153358459473, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7064180374145508, + "num_tokens": 336980892.0, + "step": 13128 + }, + { + "epoch": 1.4417966176147594, + "grad_norm": 2.1526033878326416, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7020201086997986, + "num_tokens": 337011446.0, + "step": 13129 + }, + { + "epoch": 1.4419064353173732, + "grad_norm": 2.3651938438415527, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7333720922470093, + "num_tokens": 337034912.0, + "step": 13130 + }, + { + "epoch": 1.442016253019987, + "grad_norm": 2.1486470699310303, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7142758965492249, + "num_tokens": 337061949.0, + "step": 13131 + }, + { + "epoch": 1.4421260707226005, + "grad_norm": 2.2824666500091553, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7242379188537598, + "num_tokens": 337084884.0, + "step": 13132 + }, + { + "epoch": 1.4422358884252142, + "grad_norm": 2.361621618270874, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7260636687278748, + "num_tokens": 337108243.0, + "step": 13133 + }, + { + "epoch": 1.4423457061278278, + "grad_norm": 2.2229933738708496, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.707166314125061, + "num_tokens": 337135241.0, + "step": 13134 + }, + { + "epoch": 1.4424555238304415, + "grad_norm": 2.215951681137085, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.728583574295044, + "num_tokens": 337160281.0, + "step": 13135 + }, + { + "epoch": 1.442565341533055, + "grad_norm": 2.133162021636963, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7204432487487793, + "num_tokens": 337188226.0, + "step": 13136 + }, + { + "epoch": 1.4426751592356688, + "grad_norm": 2.1425435543060303, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.708806574344635, + "num_tokens": 337215031.0, + "step": 13137 + }, + { + "epoch": 1.4427849769382823, + "grad_norm": 2.4075324535369873, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7254579663276672, + "num_tokens": 337237106.0, + "step": 13138 + }, + { + "epoch": 1.442894794640896, + "grad_norm": 2.1889147758483887, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.735629677772522, + "num_tokens": 337263545.0, + "step": 13139 + }, + { + "epoch": 1.4430046123435099, + "grad_norm": 2.4150350093841553, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7209442257881165, + "num_tokens": 337286989.0, + "step": 13140 + }, + { + "epoch": 1.4431144300461234, + "grad_norm": 2.6303598880767822, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7327688932418823, + "num_tokens": 337306021.0, + "step": 13141 + }, + { + "epoch": 1.4432242477487371, + "grad_norm": 2.0080387592315674, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6986327171325684, + "num_tokens": 337339850.0, + "step": 13142 + }, + { + "epoch": 1.4433340654513507, + "grad_norm": 2.5179481506347656, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7252673506736755, + "num_tokens": 337361707.0, + "step": 13143 + }, + { + "epoch": 1.4434438831539644, + "grad_norm": 2.332242250442505, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7187440991401672, + "num_tokens": 337387733.0, + "step": 13144 + }, + { + "epoch": 1.4435537008565782, + "grad_norm": 1.9517868757247925, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.706383466720581, + "num_tokens": 337424224.0, + "step": 13145 + }, + { + "epoch": 1.4436635185591917, + "grad_norm": 2.331594467163086, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7084182500839233, + "num_tokens": 337449790.0, + "step": 13146 + }, + { + "epoch": 1.4437733362618055, + "grad_norm": 2.5322415828704834, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7541452646255493, + "num_tokens": 337468880.0, + "step": 13147 + }, + { + "epoch": 1.443883153964419, + "grad_norm": 2.078664541244507, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6919378042221069, + "num_tokens": 337496913.0, + "step": 13148 + }, + { + "epoch": 1.4439929716670328, + "grad_norm": 2.086380958557129, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7333100438117981, + "num_tokens": 337525221.0, + "step": 13149 + }, + { + "epoch": 1.4441027893696463, + "grad_norm": 2.375551223754883, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7064833045005798, + "num_tokens": 337548765.0, + "step": 13150 + }, + { + "epoch": 1.44421260707226, + "grad_norm": 2.1138577461242676, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7266606092453003, + "num_tokens": 337576488.0, + "step": 13151 + }, + { + "epoch": 1.4443224247748736, + "grad_norm": 2.0876259803771973, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7227394580841064, + "num_tokens": 337605342.0, + "step": 13152 + }, + { + "epoch": 1.4444322424774874, + "grad_norm": 2.2942309379577637, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7307088375091553, + "num_tokens": 337630088.0, + "step": 13153 + }, + { + "epoch": 1.4445420601801011, + "grad_norm": 2.7547478675842285, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.746801495552063, + "num_tokens": 337648221.0, + "step": 13154 + }, + { + "epoch": 1.4446518778827147, + "grad_norm": 2.4989607334136963, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7227132320404053, + "num_tokens": 337670867.0, + "step": 13155 + }, + { + "epoch": 1.4447616955853284, + "grad_norm": 2.0312089920043945, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6983071565628052, + "num_tokens": 337701931.0, + "step": 13156 + }, + { + "epoch": 1.444871513287942, + "grad_norm": 2.4197394847869873, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7355841398239136, + "num_tokens": 337724325.0, + "step": 13157 + }, + { + "epoch": 1.4449813309905557, + "grad_norm": 2.1230905055999756, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7330408692359924, + "num_tokens": 337752849.0, + "step": 13158 + }, + { + "epoch": 1.4450911486931695, + "grad_norm": 2.4811103343963623, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7418557405471802, + "num_tokens": 337776019.0, + "step": 13159 + }, + { + "epoch": 1.445200966395783, + "grad_norm": 2.109139919281006, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7314872741699219, + "num_tokens": 337802036.0, + "step": 13160 + }, + { + "epoch": 1.4453107840983965, + "grad_norm": 2.194932699203491, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7428262233734131, + "num_tokens": 337826593.0, + "step": 13161 + }, + { + "epoch": 1.4454206018010103, + "grad_norm": 2.4450178146362305, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7329251170158386, + "num_tokens": 337849528.0, + "step": 13162 + }, + { + "epoch": 1.445530419503624, + "grad_norm": 2.1418087482452393, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7220306396484375, + "num_tokens": 337878984.0, + "step": 13163 + }, + { + "epoch": 1.4456402372062376, + "grad_norm": 2.107837200164795, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7271180152893066, + "num_tokens": 337909063.0, + "step": 13164 + }, + { + "epoch": 1.4457500549088513, + "grad_norm": 2.1713809967041016, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7279800772666931, + "num_tokens": 337936527.0, + "step": 13165 + }, + { + "epoch": 1.4458598726114649, + "grad_norm": 2.5732648372650146, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7465025186538696, + "num_tokens": 337956434.0, + "step": 13166 + }, + { + "epoch": 1.4459696903140786, + "grad_norm": 2.314918279647827, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7263634204864502, + "num_tokens": 337979922.0, + "step": 13167 + }, + { + "epoch": 1.4460795080166924, + "grad_norm": 2.269028902053833, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7139780521392822, + "num_tokens": 338005553.0, + "step": 13168 + }, + { + "epoch": 1.446189325719306, + "grad_norm": 2.1801626682281494, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7263568639755249, + "num_tokens": 338030746.0, + "step": 13169 + }, + { + "epoch": 1.4462991434219197, + "grad_norm": 2.1326868534088135, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7051702737808228, + "num_tokens": 338060704.0, + "step": 13170 + }, + { + "epoch": 1.4464089611245332, + "grad_norm": 2.5137805938720703, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7078264355659485, + "num_tokens": 338082111.0, + "step": 13171 + }, + { + "epoch": 1.446518778827147, + "grad_norm": 2.395780324935913, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7280540466308594, + "num_tokens": 338105427.0, + "step": 13172 + }, + { + "epoch": 1.4466285965297607, + "grad_norm": 2.224916696548462, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7322871685028076, + "num_tokens": 338131540.0, + "step": 13173 + }, + { + "epoch": 1.4467384142323743, + "grad_norm": 2.6345438957214355, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7431260347366333, + "num_tokens": 338149573.0, + "step": 13174 + }, + { + "epoch": 1.4468482319349878, + "grad_norm": 2.1694440841674805, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7109516859054565, + "num_tokens": 338179614.0, + "step": 13175 + }, + { + "epoch": 1.4469580496376016, + "grad_norm": 2.1715054512023926, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7339516282081604, + "num_tokens": 338209424.0, + "step": 13176 + }, + { + "epoch": 1.4470678673402153, + "grad_norm": 2.3573203086853027, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7145594358444214, + "num_tokens": 338231960.0, + "step": 13177 + }, + { + "epoch": 1.4471776850428288, + "grad_norm": 2.2362048625946045, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7263902425765991, + "num_tokens": 338257115.0, + "step": 13178 + }, + { + "epoch": 1.4472875027454426, + "grad_norm": 2.639113664627075, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7260444164276123, + "num_tokens": 338277642.0, + "step": 13179 + }, + { + "epoch": 1.4473973204480561, + "grad_norm": 2.17521333694458, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7146514654159546, + "num_tokens": 338306123.0, + "step": 13180 + }, + { + "epoch": 1.44750713815067, + "grad_norm": 2.71010160446167, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7511401176452637, + "num_tokens": 338325688.0, + "step": 13181 + }, + { + "epoch": 1.4476169558532836, + "grad_norm": 2.3492431640625, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7379885911941528, + "num_tokens": 338347608.0, + "step": 13182 + }, + { + "epoch": 1.4477267735558972, + "grad_norm": 2.2624735832214355, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7130621671676636, + "num_tokens": 338373230.0, + "step": 13183 + }, + { + "epoch": 1.447836591258511, + "grad_norm": 2.1607506275177, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7478266954421997, + "num_tokens": 338398707.0, + "step": 13184 + }, + { + "epoch": 1.4479464089611245, + "grad_norm": 2.0871376991271973, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7401109933853149, + "num_tokens": 338426145.0, + "step": 13185 + }, + { + "epoch": 1.4480562266637382, + "grad_norm": 2.283690929412842, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7239492535591125, + "num_tokens": 338452339.0, + "step": 13186 + }, + { + "epoch": 1.448166044366352, + "grad_norm": 2.312525749206543, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7323175668716431, + "num_tokens": 338476149.0, + "step": 13187 + }, + { + "epoch": 1.4482758620689655, + "grad_norm": 2.2640011310577393, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7130308151245117, + "num_tokens": 338503236.0, + "step": 13188 + }, + { + "epoch": 1.448385679771579, + "grad_norm": 2.0056073665618896, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7091954350471497, + "num_tokens": 338531750.0, + "step": 13189 + }, + { + "epoch": 1.4484954974741928, + "grad_norm": 2.159618616104126, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7253809571266174, + "num_tokens": 338559016.0, + "step": 13190 + }, + { + "epoch": 1.4486053151768066, + "grad_norm": 2.1484081745147705, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.705078661441803, + "num_tokens": 338585232.0, + "step": 13191 + }, + { + "epoch": 1.44871513287942, + "grad_norm": 2.261324882507324, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7286987900733948, + "num_tokens": 338610044.0, + "step": 13192 + }, + { + "epoch": 1.4488249505820339, + "grad_norm": 2.243628740310669, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7273805141448975, + "num_tokens": 338635096.0, + "step": 13193 + }, + { + "epoch": 1.4489347682846474, + "grad_norm": 2.1761724948883057, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6897484660148621, + "num_tokens": 338664089.0, + "step": 13194 + }, + { + "epoch": 1.4490445859872612, + "grad_norm": 2.046665906906128, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7036575078964233, + "num_tokens": 338695270.0, + "step": 13195 + }, + { + "epoch": 1.449154403689875, + "grad_norm": 2.4047610759735107, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.722294807434082, + "num_tokens": 338718189.0, + "step": 13196 + }, + { + "epoch": 1.4492642213924884, + "grad_norm": 2.1622345447540283, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7088909149169922, + "num_tokens": 338745505.0, + "step": 13197 + }, + { + "epoch": 1.4493740390951022, + "grad_norm": 2.1698763370513916, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7421293258666992, + "num_tokens": 338773274.0, + "step": 13198 + }, + { + "epoch": 1.4494838567977157, + "grad_norm": 2.194383382797241, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7138313055038452, + "num_tokens": 338799412.0, + "step": 13199 + }, + { + "epoch": 1.4495936745003295, + "grad_norm": 1.957794189453125, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7601351141929626, + "num_tokens": 338827350.0, + "step": 13200 + }, + { + "epoch": 1.449703492202943, + "grad_norm": 2.333890914916992, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7285571098327637, + "num_tokens": 338852119.0, + "step": 13201 + }, + { + "epoch": 1.4498133099055568, + "grad_norm": 2.2527313232421875, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6983864307403564, + "num_tokens": 338877890.0, + "step": 13202 + }, + { + "epoch": 1.4499231276081703, + "grad_norm": 2.104851722717285, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7016609907150269, + "num_tokens": 338906689.0, + "step": 13203 + }, + { + "epoch": 1.450032945310784, + "grad_norm": 1.9994182586669922, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7140031456947327, + "num_tokens": 338937863.0, + "step": 13204 + }, + { + "epoch": 1.4501427630133978, + "grad_norm": 2.3351545333862305, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7036228775978088, + "num_tokens": 338962175.0, + "step": 13205 + }, + { + "epoch": 1.4502525807160114, + "grad_norm": 2.264021873474121, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7512458562850952, + "num_tokens": 338986515.0, + "step": 13206 + }, + { + "epoch": 1.4503623984186251, + "grad_norm": 2.2242891788482666, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6940263509750366, + "num_tokens": 339012831.0, + "step": 13207 + }, + { + "epoch": 1.4504722161212387, + "grad_norm": 2.0449116230010986, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7293757200241089, + "num_tokens": 339042268.0, + "step": 13208 + }, + { + "epoch": 1.4505820338238524, + "grad_norm": 2.0278635025024414, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.730576753616333, + "num_tokens": 339071253.0, + "step": 13209 + }, + { + "epoch": 1.4506918515264662, + "grad_norm": 2.3438189029693604, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7291101813316345, + "num_tokens": 339094022.0, + "step": 13210 + }, + { + "epoch": 1.4508016692290797, + "grad_norm": 2.2435741424560547, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7155486941337585, + "num_tokens": 339118744.0, + "step": 13211 + }, + { + "epoch": 1.4509114869316933, + "grad_norm": 2.2642440795898438, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7106581926345825, + "num_tokens": 339146063.0, + "step": 13212 + }, + { + "epoch": 1.451021304634307, + "grad_norm": 2.3283259868621826, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7162836790084839, + "num_tokens": 339170676.0, + "step": 13213 + }, + { + "epoch": 1.4511311223369208, + "grad_norm": 2.196371078491211, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7226595282554626, + "num_tokens": 339198035.0, + "step": 13214 + }, + { + "epoch": 1.4512409400395343, + "grad_norm": 2.0881972312927246, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.679059624671936, + "num_tokens": 339227954.0, + "step": 13215 + }, + { + "epoch": 1.451350757742148, + "grad_norm": 2.4354007244110107, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7260844707489014, + "num_tokens": 339249659.0, + "step": 13216 + }, + { + "epoch": 1.4514605754447616, + "grad_norm": 2.0574145317077637, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7196069955825806, + "num_tokens": 339278479.0, + "step": 13217 + }, + { + "epoch": 1.4515703931473753, + "grad_norm": 2.338139057159424, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7227925062179565, + "num_tokens": 339301949.0, + "step": 13218 + }, + { + "epoch": 1.451680210849989, + "grad_norm": 2.3343050479888916, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7587300539016724, + "num_tokens": 339324621.0, + "step": 13219 + }, + { + "epoch": 1.4517900285526026, + "grad_norm": 2.6004014015197754, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7207939624786377, + "num_tokens": 339345311.0, + "step": 13220 + }, + { + "epoch": 1.4518998462552164, + "grad_norm": 2.1740500926971436, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.717185378074646, + "num_tokens": 339372179.0, + "step": 13221 + }, + { + "epoch": 1.45200966395783, + "grad_norm": 2.46331524848938, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.689473032951355, + "num_tokens": 339396059.0, + "step": 13222 + }, + { + "epoch": 1.4521194816604437, + "grad_norm": 2.29984974861145, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7122912406921387, + "num_tokens": 339420195.0, + "step": 13223 + }, + { + "epoch": 1.4522292993630574, + "grad_norm": 2.113562822341919, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6777126789093018, + "num_tokens": 339449530.0, + "step": 13224 + }, + { + "epoch": 1.452339117065671, + "grad_norm": 2.147895336151123, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6967747211456299, + "num_tokens": 339478318.0, + "step": 13225 + }, + { + "epoch": 1.4524489347682845, + "grad_norm": 2.5827088356018066, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7263708114624023, + "num_tokens": 339499963.0, + "step": 13226 + }, + { + "epoch": 1.4525587524708983, + "grad_norm": 2.3015408515930176, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7329713106155396, + "num_tokens": 339522318.0, + "step": 13227 + }, + { + "epoch": 1.452668570173512, + "grad_norm": 2.0519845485687256, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7256765365600586, + "num_tokens": 339552112.0, + "step": 13228 + }, + { + "epoch": 1.4527783878761256, + "grad_norm": 2.1895978450775146, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7195127010345459, + "num_tokens": 339578964.0, + "step": 13229 + }, + { + "epoch": 1.4528882055787393, + "grad_norm": 2.34273362159729, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7160193920135498, + "num_tokens": 339604052.0, + "step": 13230 + }, + { + "epoch": 1.4529980232813529, + "grad_norm": 2.4272522926330566, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7129886746406555, + "num_tokens": 339626589.0, + "step": 13231 + }, + { + "epoch": 1.4531078409839666, + "grad_norm": 2.1552317142486572, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.70189368724823, + "num_tokens": 339653287.0, + "step": 13232 + }, + { + "epoch": 1.4532176586865804, + "grad_norm": 2.3070104122161865, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7359593510627747, + "num_tokens": 339676163.0, + "step": 13233 + }, + { + "epoch": 1.453327476389194, + "grad_norm": 1.89431631565094, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7272473573684692, + "num_tokens": 339707805.0, + "step": 13234 + }, + { + "epoch": 1.4534372940918077, + "grad_norm": 2.3045144081115723, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7116122245788574, + "num_tokens": 339733393.0, + "step": 13235 + }, + { + "epoch": 1.4535471117944212, + "grad_norm": 2.042836904525757, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7311102151870728, + "num_tokens": 339762093.0, + "step": 13236 + }, + { + "epoch": 1.453656929497035, + "grad_norm": 2.0529794692993164, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7024363279342651, + "num_tokens": 339793466.0, + "step": 13237 + }, + { + "epoch": 1.4537667471996487, + "grad_norm": 1.9245127439498901, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7300652265548706, + "num_tokens": 339826903.0, + "step": 13238 + }, + { + "epoch": 1.4538765649022622, + "grad_norm": 2.012334108352661, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6897750496864319, + "num_tokens": 339859270.0, + "step": 13239 + }, + { + "epoch": 1.4539863826048758, + "grad_norm": 2.1790690422058105, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7313656806945801, + "num_tokens": 339886475.0, + "step": 13240 + }, + { + "epoch": 1.4540962003074895, + "grad_norm": 2.216573476791382, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7234013080596924, + "num_tokens": 339913087.0, + "step": 13241 + }, + { + "epoch": 1.4542060180101033, + "grad_norm": 2.1596813201904297, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6960984468460083, + "num_tokens": 339941206.0, + "step": 13242 + }, + { + "epoch": 1.4543158357127168, + "grad_norm": 2.3115158081054688, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7405065298080444, + "num_tokens": 339965383.0, + "step": 13243 + }, + { + "epoch": 1.4544256534153306, + "grad_norm": 2.177029609680176, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7106097936630249, + "num_tokens": 339994105.0, + "step": 13244 + }, + { + "epoch": 1.4545354711179441, + "grad_norm": 2.34010910987854, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.733920156955719, + "num_tokens": 340018100.0, + "step": 13245 + }, + { + "epoch": 1.4546452888205579, + "grad_norm": 2.104992151260376, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.717681884765625, + "num_tokens": 340047241.0, + "step": 13246 + }, + { + "epoch": 1.4547551065231716, + "grad_norm": 2.295042037963867, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7304225564002991, + "num_tokens": 340071999.0, + "step": 13247 + }, + { + "epoch": 1.4548649242257852, + "grad_norm": 2.3086235523223877, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7111656069755554, + "num_tokens": 340096272.0, + "step": 13248 + }, + { + "epoch": 1.454974741928399, + "grad_norm": 2.0377085208892822, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7111287117004395, + "num_tokens": 340127629.0, + "step": 13249 + }, + { + "epoch": 1.4550845596310125, + "grad_norm": 2.382211685180664, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7158575057983398, + "num_tokens": 340152691.0, + "step": 13250 + }, + { + "epoch": 1.4551943773336262, + "grad_norm": 2.0513696670532227, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7225930094718933, + "num_tokens": 340181797.0, + "step": 13251 + }, + { + "epoch": 1.45530419503624, + "grad_norm": 2.1331167221069336, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.70680832862854, + "num_tokens": 340209224.0, + "step": 13252 + }, + { + "epoch": 1.4554140127388535, + "grad_norm": 2.2808635234832764, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7515468001365662, + "num_tokens": 340231539.0, + "step": 13253 + }, + { + "epoch": 1.455523830441467, + "grad_norm": 1.9517589807510376, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6961817741394043, + "num_tokens": 340265117.0, + "step": 13254 + }, + { + "epoch": 1.4556336481440808, + "grad_norm": 2.0485379695892334, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7039214372634888, + "num_tokens": 340295898.0, + "step": 13255 + }, + { + "epoch": 1.4557434658466946, + "grad_norm": 2.3159689903259277, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7246994972229004, + "num_tokens": 340318562.0, + "step": 13256 + }, + { + "epoch": 1.455853283549308, + "grad_norm": 2.1136491298675537, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7404354810714722, + "num_tokens": 340344799.0, + "step": 13257 + }, + { + "epoch": 1.4559631012519219, + "grad_norm": 2.1939096450805664, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.716224193572998, + "num_tokens": 340371779.0, + "step": 13258 + }, + { + "epoch": 1.4560729189545354, + "grad_norm": 2.365476369857788, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7328431606292725, + "num_tokens": 340395271.0, + "step": 13259 + }, + { + "epoch": 1.4561827366571491, + "grad_norm": 2.429837226867676, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7519797086715698, + "num_tokens": 340417496.0, + "step": 13260 + }, + { + "epoch": 1.456292554359763, + "grad_norm": 2.4076478481292725, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.723152756690979, + "num_tokens": 340442033.0, + "step": 13261 + }, + { + "epoch": 1.4564023720623764, + "grad_norm": 2.26664400100708, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.722882866859436, + "num_tokens": 340468721.0, + "step": 13262 + }, + { + "epoch": 1.4565121897649902, + "grad_norm": 2.632544755935669, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7282706499099731, + "num_tokens": 340489264.0, + "step": 13263 + }, + { + "epoch": 1.4566220074676037, + "grad_norm": 2.244032382965088, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7370505928993225, + "num_tokens": 340513364.0, + "step": 13264 + }, + { + "epoch": 1.4567318251702175, + "grad_norm": 2.2951488494873047, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6948526501655579, + "num_tokens": 340538694.0, + "step": 13265 + }, + { + "epoch": 1.456841642872831, + "grad_norm": 2.44547963142395, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7169628739356995, + "num_tokens": 340561400.0, + "step": 13266 + }, + { + "epoch": 1.4569514605754448, + "grad_norm": 2.228041887283325, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7354942560195923, + "num_tokens": 340585689.0, + "step": 13267 + }, + { + "epoch": 1.4570612782780583, + "grad_norm": 1.8551883697509766, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6941927671432495, + "num_tokens": 340624137.0, + "step": 13268 + }, + { + "epoch": 1.457171095980672, + "grad_norm": 2.247077226638794, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.72492915391922, + "num_tokens": 340649288.0, + "step": 13269 + }, + { + "epoch": 1.4572809136832858, + "grad_norm": 2.141599655151367, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7185621857643127, + "num_tokens": 340677472.0, + "step": 13270 + }, + { + "epoch": 1.4573907313858994, + "grad_norm": 2.202702045440674, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7239902019500732, + "num_tokens": 340702851.0, + "step": 13271 + }, + { + "epoch": 1.4575005490885131, + "grad_norm": 2.1258347034454346, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7123484015464783, + "num_tokens": 340730294.0, + "step": 13272 + }, + { + "epoch": 1.4576103667911267, + "grad_norm": 2.3617422580718994, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7054617404937744, + "num_tokens": 340757012.0, + "step": 13273 + }, + { + "epoch": 1.4577201844937404, + "grad_norm": 2.315915822982788, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7059884071350098, + "num_tokens": 340783036.0, + "step": 13274 + }, + { + "epoch": 1.4578300021963542, + "grad_norm": 2.138108015060425, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7275629043579102, + "num_tokens": 340812294.0, + "step": 13275 + }, + { + "epoch": 1.4579398198989677, + "grad_norm": 2.108973741531372, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7249259948730469, + "num_tokens": 340838783.0, + "step": 13276 + }, + { + "epoch": 1.4580496376015812, + "grad_norm": 1.9738609790802002, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7147876024246216, + "num_tokens": 340871858.0, + "step": 13277 + }, + { + "epoch": 1.458159455304195, + "grad_norm": 2.102725028991699, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7396024465560913, + "num_tokens": 340900197.0, + "step": 13278 + }, + { + "epoch": 1.4582692730068088, + "grad_norm": 2.17305064201355, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6973942518234253, + "num_tokens": 340928280.0, + "step": 13279 + }, + { + "epoch": 1.4583790907094223, + "grad_norm": 2.4684255123138428, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7213977575302124, + "num_tokens": 340950382.0, + "step": 13280 + }, + { + "epoch": 1.458488908412036, + "grad_norm": 2.118776559829712, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.705374002456665, + "num_tokens": 340979457.0, + "step": 13281 + }, + { + "epoch": 1.4585987261146496, + "grad_norm": 2.3389382362365723, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7059762477874756, + "num_tokens": 341003678.0, + "step": 13282 + }, + { + "epoch": 1.4587085438172633, + "grad_norm": 2.3485188484191895, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7159335613250732, + "num_tokens": 341028201.0, + "step": 13283 + }, + { + "epoch": 1.458818361519877, + "grad_norm": 2.398792266845703, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7281243801116943, + "num_tokens": 341052177.0, + "step": 13284 + }, + { + "epoch": 1.4589281792224906, + "grad_norm": 2.622600555419922, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7281209826469421, + "num_tokens": 341072580.0, + "step": 13285 + }, + { + "epoch": 1.4590379969251044, + "grad_norm": 2.562587022781372, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7205150127410889, + "num_tokens": 341094112.0, + "step": 13286 + }, + { + "epoch": 1.459147814627718, + "grad_norm": 2.3650965690612793, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7010089755058289, + "num_tokens": 341119234.0, + "step": 13287 + }, + { + "epoch": 1.4592576323303317, + "grad_norm": 2.2459545135498047, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7197222709655762, + "num_tokens": 341144308.0, + "step": 13288 + }, + { + "epoch": 1.4593674500329454, + "grad_norm": 2.203301191329956, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7223867177963257, + "num_tokens": 341171990.0, + "step": 13289 + }, + { + "epoch": 1.459477267735559, + "grad_norm": 2.263362407684326, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7050893306732178, + "num_tokens": 341198665.0, + "step": 13290 + }, + { + "epoch": 1.4595870854381725, + "grad_norm": 2.0159146785736084, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.695701003074646, + "num_tokens": 341228513.0, + "step": 13291 + }, + { + "epoch": 1.4596969031407863, + "grad_norm": 2.2095279693603516, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.731194257736206, + "num_tokens": 341254976.0, + "step": 13292 + }, + { + "epoch": 1.4598067208434, + "grad_norm": 2.365935802459717, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7208876609802246, + "num_tokens": 341277905.0, + "step": 13293 + }, + { + "epoch": 1.4599165385460136, + "grad_norm": 2.0180740356445312, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.736316442489624, + "num_tokens": 341305323.0, + "step": 13294 + }, + { + "epoch": 1.4600263562486273, + "grad_norm": 2.4978184700012207, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7508621215820312, + "num_tokens": 341324097.0, + "step": 13295 + }, + { + "epoch": 1.4601361739512408, + "grad_norm": 2.040987968444824, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6849092245101929, + "num_tokens": 341356430.0, + "step": 13296 + }, + { + "epoch": 1.4602459916538546, + "grad_norm": 2.084792375564575, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7070029973983765, + "num_tokens": 341385648.0, + "step": 13297 + }, + { + "epoch": 1.4603558093564684, + "grad_norm": 2.1818180084228516, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7302958965301514, + "num_tokens": 341415067.0, + "step": 13298 + }, + { + "epoch": 1.460465627059082, + "grad_norm": 1.8599246740341187, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7288058996200562, + "num_tokens": 341451280.0, + "step": 13299 + }, + { + "epoch": 1.4605754447616957, + "grad_norm": 2.4607086181640625, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7312151193618774, + "num_tokens": 341472606.0, + "step": 13300 + }, + { + "epoch": 1.4606852624643092, + "grad_norm": 1.999534010887146, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7170162796974182, + "num_tokens": 341501624.0, + "step": 13301 + }, + { + "epoch": 1.460795080166923, + "grad_norm": 2.4083869457244873, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7223572731018066, + "num_tokens": 341525986.0, + "step": 13302 + }, + { + "epoch": 1.4609048978695367, + "grad_norm": 2.167633533477783, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7259423732757568, + "num_tokens": 341553688.0, + "step": 13303 + }, + { + "epoch": 1.4610147155721502, + "grad_norm": 2.4748289585113525, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7262980341911316, + "num_tokens": 341575411.0, + "step": 13304 + }, + { + "epoch": 1.4611245332747638, + "grad_norm": 2.3743896484375, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7192237377166748, + "num_tokens": 341600430.0, + "step": 13305 + }, + { + "epoch": 1.4612343509773775, + "grad_norm": 2.243452548980713, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6935147047042847, + "num_tokens": 341630453.0, + "step": 13306 + }, + { + "epoch": 1.4613441686799913, + "grad_norm": 2.1684160232543945, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7208437323570251, + "num_tokens": 341657479.0, + "step": 13307 + }, + { + "epoch": 1.4614539863826048, + "grad_norm": 2.088472604751587, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7396401166915894, + "num_tokens": 341685121.0, + "step": 13308 + }, + { + "epoch": 1.4615638040852186, + "grad_norm": 2.552422523498535, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7462255358695984, + "num_tokens": 341704156.0, + "step": 13309 + }, + { + "epoch": 1.4616736217878321, + "grad_norm": 2.1900601387023926, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7148622274398804, + "num_tokens": 341728927.0, + "step": 13310 + }, + { + "epoch": 1.4617834394904459, + "grad_norm": 2.0704383850097656, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7024092674255371, + "num_tokens": 341759300.0, + "step": 13311 + }, + { + "epoch": 1.4618932571930596, + "grad_norm": 2.158724069595337, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7158680558204651, + "num_tokens": 341787124.0, + "step": 13312 + }, + { + "epoch": 1.4620030748956732, + "grad_norm": 2.2231557369232178, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7202759981155396, + "num_tokens": 341814271.0, + "step": 13313 + }, + { + "epoch": 1.462112892598287, + "grad_norm": 2.4032979011535645, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6924058198928833, + "num_tokens": 341837971.0, + "step": 13314 + }, + { + "epoch": 1.4622227103009005, + "grad_norm": 2.1480116844177246, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7140500545501709, + "num_tokens": 341865197.0, + "step": 13315 + }, + { + "epoch": 1.4623325280035142, + "grad_norm": 2.3631765842437744, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7184516191482544, + "num_tokens": 341888819.0, + "step": 13316 + }, + { + "epoch": 1.462442345706128, + "grad_norm": 2.4973819255828857, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7202202081680298, + "num_tokens": 341913099.0, + "step": 13317 + }, + { + "epoch": 1.4625521634087415, + "grad_norm": 2.3659422397613525, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7107011675834656, + "num_tokens": 341937528.0, + "step": 13318 + }, + { + "epoch": 1.462661981111355, + "grad_norm": 2.431396961212158, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7144922614097595, + "num_tokens": 341962623.0, + "step": 13319 + }, + { + "epoch": 1.4627717988139688, + "grad_norm": 2.2571263313293457, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7232205271720886, + "num_tokens": 341990467.0, + "step": 13320 + }, + { + "epoch": 1.4628816165165826, + "grad_norm": 2.3352649211883545, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7316215634346008, + "num_tokens": 342014171.0, + "step": 13321 + }, + { + "epoch": 1.462991434219196, + "grad_norm": 2.0588274002075195, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.709625780582428, + "num_tokens": 342046925.0, + "step": 13322 + }, + { + "epoch": 1.4631012519218098, + "grad_norm": 2.340043544769287, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7181493639945984, + "num_tokens": 342071546.0, + "step": 13323 + }, + { + "epoch": 1.4632110696244234, + "grad_norm": 2.4344260692596436, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7448459267616272, + "num_tokens": 342094702.0, + "step": 13324 + }, + { + "epoch": 1.4633208873270371, + "grad_norm": 2.415142059326172, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.73014235496521, + "num_tokens": 342118190.0, + "step": 13325 + }, + { + "epoch": 1.463430705029651, + "grad_norm": 2.5149917602539062, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.732563316822052, + "num_tokens": 342139267.0, + "step": 13326 + }, + { + "epoch": 1.4635405227322644, + "grad_norm": 2.1947004795074463, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7160574197769165, + "num_tokens": 342166758.0, + "step": 13327 + }, + { + "epoch": 1.4636503404348782, + "grad_norm": 2.145906686782837, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7097371816635132, + "num_tokens": 342195686.0, + "step": 13328 + }, + { + "epoch": 1.4637601581374917, + "grad_norm": 2.257194757461548, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7146928906440735, + "num_tokens": 342220269.0, + "step": 13329 + }, + { + "epoch": 1.4638699758401055, + "grad_norm": 2.3853588104248047, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7376659512519836, + "num_tokens": 342242341.0, + "step": 13330 + }, + { + "epoch": 1.463979793542719, + "grad_norm": 2.314222812652588, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7167971730232239, + "num_tokens": 342269282.0, + "step": 13331 + }, + { + "epoch": 1.4640896112453328, + "grad_norm": 2.184375762939453, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7142179012298584, + "num_tokens": 342296196.0, + "step": 13332 + }, + { + "epoch": 1.4641994289479463, + "grad_norm": 2.5980305671691895, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7194653153419495, + "num_tokens": 342317495.0, + "step": 13333 + }, + { + "epoch": 1.46430924665056, + "grad_norm": 2.352382183074951, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7357858419418335, + "num_tokens": 342342977.0, + "step": 13334 + }, + { + "epoch": 1.4644190643531738, + "grad_norm": 2.342061758041382, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7018020749092102, + "num_tokens": 342367364.0, + "step": 13335 + }, + { + "epoch": 1.4645288820557874, + "grad_norm": 2.510368585586548, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7365439534187317, + "num_tokens": 342388457.0, + "step": 13336 + }, + { + "epoch": 1.464638699758401, + "grad_norm": 2.228682518005371, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7279343605041504, + "num_tokens": 342414109.0, + "step": 13337 + }, + { + "epoch": 1.4647485174610146, + "grad_norm": 2.3141801357269287, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7477779388427734, + "num_tokens": 342437580.0, + "step": 13338 + }, + { + "epoch": 1.4648583351636284, + "grad_norm": 2.1063106060028076, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7183359861373901, + "num_tokens": 342467456.0, + "step": 13339 + }, + { + "epoch": 1.4649681528662422, + "grad_norm": 2.5991177558898926, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7154201865196228, + "num_tokens": 342488270.0, + "step": 13340 + }, + { + "epoch": 1.4650779705688557, + "grad_norm": 2.5686047077178955, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7228380441665649, + "num_tokens": 342509974.0, + "step": 13341 + }, + { + "epoch": 1.4651877882714692, + "grad_norm": 2.519850969314575, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.702070951461792, + "num_tokens": 342532450.0, + "step": 13342 + }, + { + "epoch": 1.465297605974083, + "grad_norm": 2.2486114501953125, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7114505767822266, + "num_tokens": 342559122.0, + "step": 13343 + }, + { + "epoch": 1.4654074236766967, + "grad_norm": 2.1681814193725586, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7235543131828308, + "num_tokens": 342587202.0, + "step": 13344 + }, + { + "epoch": 1.4655172413793103, + "grad_norm": 2.0541579723358154, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6949050426483154, + "num_tokens": 342617998.0, + "step": 13345 + }, + { + "epoch": 1.465627059081924, + "grad_norm": 2.1249358654022217, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7222722768783569, + "num_tokens": 342645902.0, + "step": 13346 + }, + { + "epoch": 1.4657368767845376, + "grad_norm": 2.5072999000549316, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7455533742904663, + "num_tokens": 342665531.0, + "step": 13347 + }, + { + "epoch": 1.4658466944871513, + "grad_norm": 2.083799362182617, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7154421210289001, + "num_tokens": 342694121.0, + "step": 13348 + }, + { + "epoch": 1.465956512189765, + "grad_norm": 2.160278081893921, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7209856510162354, + "num_tokens": 342721214.0, + "step": 13349 + }, + { + "epoch": 1.4660663298923786, + "grad_norm": 2.1792595386505127, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7437022924423218, + "num_tokens": 342746479.0, + "step": 13350 + }, + { + "epoch": 1.4661761475949924, + "grad_norm": 1.998337745666504, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7103927135467529, + "num_tokens": 342777033.0, + "step": 13351 + }, + { + "epoch": 1.466285965297606, + "grad_norm": 2.296980142593384, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7132112979888916, + "num_tokens": 342802415.0, + "step": 13352 + }, + { + "epoch": 1.4663957830002197, + "grad_norm": 2.5015013217926025, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7267355918884277, + "num_tokens": 342823544.0, + "step": 13353 + }, + { + "epoch": 1.4665056007028334, + "grad_norm": 2.2302277088165283, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7190209627151489, + "num_tokens": 342851433.0, + "step": 13354 + }, + { + "epoch": 1.466615418405447, + "grad_norm": 2.1780331134796143, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7331690788269043, + "num_tokens": 342880196.0, + "step": 13355 + }, + { + "epoch": 1.4667252361080605, + "grad_norm": 2.2450695037841797, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.720855176448822, + "num_tokens": 342906672.0, + "step": 13356 + }, + { + "epoch": 1.4668350538106742, + "grad_norm": 2.161982536315918, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6968549489974976, + "num_tokens": 342937046.0, + "step": 13357 + }, + { + "epoch": 1.466944871513288, + "grad_norm": 2.186279296875, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7128412127494812, + "num_tokens": 342963616.0, + "step": 13358 + }, + { + "epoch": 1.4670546892159015, + "grad_norm": 2.4645261764526367, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7381759881973267, + "num_tokens": 342984769.0, + "step": 13359 + }, + { + "epoch": 1.4671645069185153, + "grad_norm": 2.3239481449127197, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7088697552680969, + "num_tokens": 343009036.0, + "step": 13360 + }, + { + "epoch": 1.4672743246211288, + "grad_norm": 2.4660770893096924, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7035956382751465, + "num_tokens": 343030727.0, + "step": 13361 + }, + { + "epoch": 1.4673841423237426, + "grad_norm": 2.316260814666748, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7383514642715454, + "num_tokens": 343052983.0, + "step": 13362 + }, + { + "epoch": 1.4674939600263563, + "grad_norm": 2.0502805709838867, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6937583088874817, + "num_tokens": 343081356.0, + "step": 13363 + }, + { + "epoch": 1.4676037777289699, + "grad_norm": 2.264657497406006, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7266812324523926, + "num_tokens": 343106640.0, + "step": 13364 + }, + { + "epoch": 1.4677135954315836, + "grad_norm": 1.9756799936294556, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7286607623100281, + "num_tokens": 343137613.0, + "step": 13365 + }, + { + "epoch": 1.4678234131341972, + "grad_norm": 2.135416030883789, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7025377154350281, + "num_tokens": 343165921.0, + "step": 13366 + }, + { + "epoch": 1.467933230836811, + "grad_norm": 2.2778642177581787, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7190051078796387, + "num_tokens": 343190915.0, + "step": 13367 + }, + { + "epoch": 1.4680430485394247, + "grad_norm": 2.177511215209961, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.710453987121582, + "num_tokens": 343216814.0, + "step": 13368 + }, + { + "epoch": 1.4681528662420382, + "grad_norm": 2.5757205486297607, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7033761739730835, + "num_tokens": 343237774.0, + "step": 13369 + }, + { + "epoch": 1.4682626839446518, + "grad_norm": 2.2325141429901123, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7224693298339844, + "num_tokens": 343264167.0, + "step": 13370 + }, + { + "epoch": 1.4683725016472655, + "grad_norm": 2.2262892723083496, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7287605404853821, + "num_tokens": 343290639.0, + "step": 13371 + }, + { + "epoch": 1.4684823193498793, + "grad_norm": 2.392998218536377, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7244008779525757, + "num_tokens": 343312698.0, + "step": 13372 + }, + { + "epoch": 1.4685921370524928, + "grad_norm": 2.1210031509399414, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.70540851354599, + "num_tokens": 343341525.0, + "step": 13373 + }, + { + "epoch": 1.4687019547551066, + "grad_norm": 2.466535806655884, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7175313234329224, + "num_tokens": 343363126.0, + "step": 13374 + }, + { + "epoch": 1.46881177245772, + "grad_norm": 2.314863443374634, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7150369882583618, + "num_tokens": 343388545.0, + "step": 13375 + }, + { + "epoch": 1.4689215901603339, + "grad_norm": 2.3091139793395996, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7062492966651917, + "num_tokens": 343414942.0, + "step": 13376 + }, + { + "epoch": 1.4690314078629476, + "grad_norm": 2.021669387817383, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7154271602630615, + "num_tokens": 343445809.0, + "step": 13377 + }, + { + "epoch": 1.4691412255655611, + "grad_norm": 2.3153927326202393, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7493510842323303, + "num_tokens": 343467771.0, + "step": 13378 + }, + { + "epoch": 1.469251043268175, + "grad_norm": 2.173915386199951, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7187168598175049, + "num_tokens": 343493555.0, + "step": 13379 + }, + { + "epoch": 1.4693608609707884, + "grad_norm": 2.234282970428467, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7377744913101196, + "num_tokens": 343519367.0, + "step": 13380 + }, + { + "epoch": 1.4694706786734022, + "grad_norm": 2.2410340309143066, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.722156286239624, + "num_tokens": 343545129.0, + "step": 13381 + }, + { + "epoch": 1.4695804963760157, + "grad_norm": 2.3442893028259277, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7285250425338745, + "num_tokens": 343569160.0, + "step": 13382 + }, + { + "epoch": 1.4696903140786295, + "grad_norm": 1.9921339750289917, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.715554416179657, + "num_tokens": 343601921.0, + "step": 13383 + }, + { + "epoch": 1.469800131781243, + "grad_norm": 1.885447382926941, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7243465185165405, + "num_tokens": 343636110.0, + "step": 13384 + }, + { + "epoch": 1.4699099494838568, + "grad_norm": 2.0917370319366455, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7184191942214966, + "num_tokens": 343662576.0, + "step": 13385 + }, + { + "epoch": 1.4700197671864705, + "grad_norm": 2.0298519134521484, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6969438791275024, + "num_tokens": 343694322.0, + "step": 13386 + }, + { + "epoch": 1.470129584889084, + "grad_norm": 2.292062759399414, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7171990871429443, + "num_tokens": 343720745.0, + "step": 13387 + }, + { + "epoch": 1.4702394025916978, + "grad_norm": 2.2201197147369385, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.731721043586731, + "num_tokens": 343745112.0, + "step": 13388 + }, + { + "epoch": 1.4703492202943114, + "grad_norm": 2.0323402881622314, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7056735157966614, + "num_tokens": 343778173.0, + "step": 13389 + }, + { + "epoch": 1.4704590379969251, + "grad_norm": 2.3830950260162354, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7247976064682007, + "num_tokens": 343802962.0, + "step": 13390 + }, + { + "epoch": 1.4705688556995389, + "grad_norm": 2.413062810897827, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7107055187225342, + "num_tokens": 343828258.0, + "step": 13391 + }, + { + "epoch": 1.4706786734021524, + "grad_norm": 2.154014825820923, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7201143503189087, + "num_tokens": 343856706.0, + "step": 13392 + }, + { + "epoch": 1.4707884911047662, + "grad_norm": 2.316986322402954, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7124186754226685, + "num_tokens": 343880409.0, + "step": 13393 + }, + { + "epoch": 1.4708983088073797, + "grad_norm": 2.040681838989258, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6956369876861572, + "num_tokens": 343911446.0, + "step": 13394 + }, + { + "epoch": 1.4710081265099935, + "grad_norm": 2.708909749984741, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7387468814849854, + "num_tokens": 343930541.0, + "step": 13395 + }, + { + "epoch": 1.471117944212607, + "grad_norm": 2.169243335723877, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7122708559036255, + "num_tokens": 343958967.0, + "step": 13396 + }, + { + "epoch": 1.4712277619152208, + "grad_norm": 2.5784716606140137, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7204694747924805, + "num_tokens": 343980147.0, + "step": 13397 + }, + { + "epoch": 1.4713375796178343, + "grad_norm": 2.170494318008423, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.717294454574585, + "num_tokens": 344006669.0, + "step": 13398 + }, + { + "epoch": 1.471447397320448, + "grad_norm": 2.656189203262329, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7225775718688965, + "num_tokens": 344024845.0, + "step": 13399 + }, + { + "epoch": 1.4715572150230618, + "grad_norm": 1.9865111112594604, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7135529518127441, + "num_tokens": 344056347.0, + "step": 13400 + }, + { + "epoch": 1.4716670327256753, + "grad_norm": 2.400367021560669, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.733980655670166, + "num_tokens": 344078612.0, + "step": 13401 + }, + { + "epoch": 1.471776850428289, + "grad_norm": 2.1774301528930664, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7246768474578857, + "num_tokens": 344105416.0, + "step": 13402 + }, + { + "epoch": 1.4718866681309026, + "grad_norm": 2.4218337535858154, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7669760584831238, + "num_tokens": 344126569.0, + "step": 13403 + }, + { + "epoch": 1.4719964858335164, + "grad_norm": 2.3946471214294434, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7075356245040894, + "num_tokens": 344151374.0, + "step": 13404 + }, + { + "epoch": 1.4721063035361301, + "grad_norm": 2.326472520828247, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7123957872390747, + "num_tokens": 344176055.0, + "step": 13405 + }, + { + "epoch": 1.4722161212387437, + "grad_norm": 2.5804591178894043, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7197598218917847, + "num_tokens": 344197385.0, + "step": 13406 + }, + { + "epoch": 1.4723259389413572, + "grad_norm": 2.380946397781372, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7261511087417603, + "num_tokens": 344221133.0, + "step": 13407 + }, + { + "epoch": 1.472435756643971, + "grad_norm": 2.1812736988067627, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6956689357757568, + "num_tokens": 344251953.0, + "step": 13408 + }, + { + "epoch": 1.4725455743465847, + "grad_norm": 2.0366463661193848, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7305787801742554, + "num_tokens": 344280009.0, + "step": 13409 + }, + { + "epoch": 1.4726553920491983, + "grad_norm": 2.0503652095794678, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7134842276573181, + "num_tokens": 344310659.0, + "step": 13410 + }, + { + "epoch": 1.472765209751812, + "grad_norm": 2.335223436355591, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7208112478256226, + "num_tokens": 344333611.0, + "step": 13411 + }, + { + "epoch": 1.4728750274544256, + "grad_norm": 1.8790782690048218, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7116470336914062, + "num_tokens": 344366577.0, + "step": 13412 + }, + { + "epoch": 1.4729848451570393, + "grad_norm": 2.1161515712738037, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7272816896438599, + "num_tokens": 344394520.0, + "step": 13413 + }, + { + "epoch": 1.473094662859653, + "grad_norm": 2.038675546646118, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7195090055465698, + "num_tokens": 344422338.0, + "step": 13414 + }, + { + "epoch": 1.4732044805622666, + "grad_norm": 2.0844533443450928, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.712776780128479, + "num_tokens": 344452872.0, + "step": 13415 + }, + { + "epoch": 1.4733142982648804, + "grad_norm": 2.2502152919769287, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7455966472625732, + "num_tokens": 344477899.0, + "step": 13416 + }, + { + "epoch": 1.473424115967494, + "grad_norm": 2.354313611984253, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7411638498306274, + "num_tokens": 344500674.0, + "step": 13417 + }, + { + "epoch": 1.4735339336701077, + "grad_norm": 2.103442668914795, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7186712026596069, + "num_tokens": 344527559.0, + "step": 13418 + }, + { + "epoch": 1.4736437513727214, + "grad_norm": 1.9680252075195312, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7131766080856323, + "num_tokens": 344559151.0, + "step": 13419 + }, + { + "epoch": 1.473753569075335, + "grad_norm": 2.1140034198760986, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7166458368301392, + "num_tokens": 344585469.0, + "step": 13420 + }, + { + "epoch": 1.4738633867779485, + "grad_norm": 2.3660619258880615, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7167415022850037, + "num_tokens": 344609337.0, + "step": 13421 + }, + { + "epoch": 1.4739732044805622, + "grad_norm": 2.07742977142334, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7067627906799316, + "num_tokens": 344639494.0, + "step": 13422 + }, + { + "epoch": 1.474083022183176, + "grad_norm": 2.6887896060943604, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7424418926239014, + "num_tokens": 344656246.0, + "step": 13423 + }, + { + "epoch": 1.4741928398857895, + "grad_norm": 2.440600872039795, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7208379507064819, + "num_tokens": 344679248.0, + "step": 13424 + }, + { + "epoch": 1.4743026575884033, + "grad_norm": 1.9920525550842285, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6829550266265869, + "num_tokens": 344712141.0, + "step": 13425 + }, + { + "epoch": 1.4744124752910168, + "grad_norm": 2.2730071544647217, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7032060027122498, + "num_tokens": 344738625.0, + "step": 13426 + }, + { + "epoch": 1.4745222929936306, + "grad_norm": 2.511075496673584, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7130991220474243, + "num_tokens": 344761132.0, + "step": 13427 + }, + { + "epoch": 1.4746321106962443, + "grad_norm": 2.1273305416107178, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7170360088348389, + "num_tokens": 344789616.0, + "step": 13428 + }, + { + "epoch": 1.4747419283988579, + "grad_norm": 2.0734364986419678, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7222859859466553, + "num_tokens": 344817836.0, + "step": 13429 + }, + { + "epoch": 1.4748517461014716, + "grad_norm": 2.2565622329711914, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7046290636062622, + "num_tokens": 344844425.0, + "step": 13430 + }, + { + "epoch": 1.4749615638040852, + "grad_norm": 2.272772789001465, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7220703959465027, + "num_tokens": 344869117.0, + "step": 13431 + }, + { + "epoch": 1.475071381506699, + "grad_norm": 2.3386311531066895, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7220892310142517, + "num_tokens": 344893033.0, + "step": 13432 + }, + { + "epoch": 1.4751811992093127, + "grad_norm": 2.28829026222229, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.712192177772522, + "num_tokens": 344917795.0, + "step": 13433 + }, + { + "epoch": 1.4752910169119262, + "grad_norm": 1.9896646738052368, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6989973783493042, + "num_tokens": 344948160.0, + "step": 13434 + }, + { + "epoch": 1.4754008346145397, + "grad_norm": 2.4737071990966797, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7465609312057495, + "num_tokens": 344969414.0, + "step": 13435 + }, + { + "epoch": 1.4755106523171535, + "grad_norm": 2.509016752243042, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7341869473457336, + "num_tokens": 344990971.0, + "step": 13436 + }, + { + "epoch": 1.4756204700197673, + "grad_norm": 2.2105255126953125, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7165360450744629, + "num_tokens": 345017408.0, + "step": 13437 + }, + { + "epoch": 1.4757302877223808, + "grad_norm": 2.4081852436065674, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7433359622955322, + "num_tokens": 345038772.0, + "step": 13438 + }, + { + "epoch": 1.4758401054249946, + "grad_norm": 2.101996421813965, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7031049132347107, + "num_tokens": 345067236.0, + "step": 13439 + }, + { + "epoch": 1.475949923127608, + "grad_norm": 2.0141148567199707, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7265354990959167, + "num_tokens": 345099262.0, + "step": 13440 + }, + { + "epoch": 1.4760597408302218, + "grad_norm": 2.48168683052063, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7246742844581604, + "num_tokens": 345121076.0, + "step": 13441 + }, + { + "epoch": 1.4761695585328356, + "grad_norm": 2.0945401191711426, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6910682916641235, + "num_tokens": 345152176.0, + "step": 13442 + }, + { + "epoch": 1.4762793762354491, + "grad_norm": 2.276510715484619, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7187331914901733, + "num_tokens": 345177811.0, + "step": 13443 + }, + { + "epoch": 1.476389193938063, + "grad_norm": 2.06524658203125, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7304785251617432, + "num_tokens": 345207046.0, + "step": 13444 + }, + { + "epoch": 1.4764990116406764, + "grad_norm": 2.4531543254852295, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.706934928894043, + "num_tokens": 345231299.0, + "step": 13445 + }, + { + "epoch": 1.4766088293432902, + "grad_norm": 2.314546823501587, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7180179357528687, + "num_tokens": 345257844.0, + "step": 13446 + }, + { + "epoch": 1.4767186470459037, + "grad_norm": 2.5209600925445557, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.732017457485199, + "num_tokens": 345277956.0, + "step": 13447 + }, + { + "epoch": 1.4768284647485175, + "grad_norm": 2.270637273788452, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7213821411132812, + "num_tokens": 345303668.0, + "step": 13448 + }, + { + "epoch": 1.476938282451131, + "grad_norm": 2.209855556488037, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7265378832817078, + "num_tokens": 345330288.0, + "step": 13449 + }, + { + "epoch": 1.4770481001537448, + "grad_norm": 2.7664153575897217, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.746336817741394, + "num_tokens": 345347975.0, + "step": 13450 + }, + { + "epoch": 1.4771579178563585, + "grad_norm": 2.396629571914673, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7179538011550903, + "num_tokens": 345372039.0, + "step": 13451 + }, + { + "epoch": 1.477267735558972, + "grad_norm": 2.476435422897339, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.720276951789856, + "num_tokens": 345393607.0, + "step": 13452 + }, + { + "epoch": 1.4773775532615858, + "grad_norm": 2.515176296234131, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7291479706764221, + "num_tokens": 345414221.0, + "step": 13453 + }, + { + "epoch": 1.4774873709641994, + "grad_norm": 2.0989139080047607, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7021125555038452, + "num_tokens": 345442787.0, + "step": 13454 + }, + { + "epoch": 1.477597188666813, + "grad_norm": 2.678157091140747, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7273625135421753, + "num_tokens": 345462339.0, + "step": 13455 + }, + { + "epoch": 1.4777070063694269, + "grad_norm": 2.614179849624634, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7110913991928101, + "num_tokens": 345482862.0, + "step": 13456 + }, + { + "epoch": 1.4778168240720404, + "grad_norm": 1.9583510160446167, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7115779519081116, + "num_tokens": 345513505.0, + "step": 13457 + }, + { + "epoch": 1.477926641774654, + "grad_norm": 2.1549301147460938, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7292544841766357, + "num_tokens": 345539633.0, + "step": 13458 + }, + { + "epoch": 1.4780364594772677, + "grad_norm": 2.2535765171051025, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7226454615592957, + "num_tokens": 345563519.0, + "step": 13459 + }, + { + "epoch": 1.4781462771798815, + "grad_norm": 1.8270643949508667, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7053967714309692, + "num_tokens": 345599628.0, + "step": 13460 + }, + { + "epoch": 1.478256094882495, + "grad_norm": 2.1658332347869873, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.701653778553009, + "num_tokens": 345626394.0, + "step": 13461 + }, + { + "epoch": 1.4783659125851087, + "grad_norm": 2.2364935874938965, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7041797637939453, + "num_tokens": 345652751.0, + "step": 13462 + }, + { + "epoch": 1.4784757302877223, + "grad_norm": 2.0674612522125244, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7187657356262207, + "num_tokens": 345680921.0, + "step": 13463 + }, + { + "epoch": 1.478585547990336, + "grad_norm": 2.284231662750244, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7087551355361938, + "num_tokens": 345706665.0, + "step": 13464 + }, + { + "epoch": 1.4786953656929498, + "grad_norm": 2.2406275272369385, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7296724915504456, + "num_tokens": 345732760.0, + "step": 13465 + }, + { + "epoch": 1.4788051833955633, + "grad_norm": 2.162123680114746, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6932245492935181, + "num_tokens": 345760777.0, + "step": 13466 + }, + { + "epoch": 1.478915001098177, + "grad_norm": 2.1723315715789795, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7384167909622192, + "num_tokens": 345787169.0, + "step": 13467 + }, + { + "epoch": 1.4790248188007906, + "grad_norm": 2.2856030464172363, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7219369411468506, + "num_tokens": 345812158.0, + "step": 13468 + }, + { + "epoch": 1.4791346365034044, + "grad_norm": 2.2019331455230713, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7060049772262573, + "num_tokens": 345839521.0, + "step": 13469 + }, + { + "epoch": 1.4792444542060181, + "grad_norm": 2.072854995727539, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7073825001716614, + "num_tokens": 345871745.0, + "step": 13470 + }, + { + "epoch": 1.4793542719086317, + "grad_norm": 2.1072330474853516, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7292499542236328, + "num_tokens": 345898543.0, + "step": 13471 + }, + { + "epoch": 1.4794640896112452, + "grad_norm": 2.4399101734161377, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7053745985031128, + "num_tokens": 345924570.0, + "step": 13472 + }, + { + "epoch": 1.479573907313859, + "grad_norm": 2.444021701812744, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7362967729568481, + "num_tokens": 345947973.0, + "step": 13473 + }, + { + "epoch": 1.4796837250164727, + "grad_norm": 2.415503740310669, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7421202659606934, + "num_tokens": 345968610.0, + "step": 13474 + }, + { + "epoch": 1.4797935427190863, + "grad_norm": 2.069852113723755, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7226592898368835, + "num_tokens": 346000519.0, + "step": 13475 + }, + { + "epoch": 1.4799033604217, + "grad_norm": 2.1610612869262695, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7341076135635376, + "num_tokens": 346026126.0, + "step": 13476 + }, + { + "epoch": 1.4800131781243135, + "grad_norm": 2.751328706741333, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7089636921882629, + "num_tokens": 346046092.0, + "step": 13477 + }, + { + "epoch": 1.4801229958269273, + "grad_norm": 2.218690872192383, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7202691435813904, + "num_tokens": 346074942.0, + "step": 13478 + }, + { + "epoch": 1.480232813529541, + "grad_norm": 2.127696990966797, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7241429090499878, + "num_tokens": 346100282.0, + "step": 13479 + }, + { + "epoch": 1.4803426312321546, + "grad_norm": 2.5757176876068115, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7375196814537048, + "num_tokens": 346120056.0, + "step": 13480 + }, + { + "epoch": 1.4804524489347684, + "grad_norm": 2.5996270179748535, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7284720540046692, + "num_tokens": 346140704.0, + "step": 13481 + }, + { + "epoch": 1.4805622666373819, + "grad_norm": 2.1165359020233154, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6929970979690552, + "num_tokens": 346171017.0, + "step": 13482 + }, + { + "epoch": 1.4806720843399956, + "grad_norm": 2.325225591659546, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7226086854934692, + "num_tokens": 346195031.0, + "step": 13483 + }, + { + "epoch": 1.4807819020426094, + "grad_norm": 2.1177592277526855, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.722000241279602, + "num_tokens": 346224882.0, + "step": 13484 + }, + { + "epoch": 1.480891719745223, + "grad_norm": 2.280407190322876, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7438488006591797, + "num_tokens": 346250088.0, + "step": 13485 + }, + { + "epoch": 1.4810015374478365, + "grad_norm": 2.411813974380493, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7299003005027771, + "num_tokens": 346272341.0, + "step": 13486 + }, + { + "epoch": 1.4811113551504502, + "grad_norm": 2.168630361557007, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7064510583877563, + "num_tokens": 346300716.0, + "step": 13487 + }, + { + "epoch": 1.481221172853064, + "grad_norm": 2.2435290813446045, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7358169555664062, + "num_tokens": 346325286.0, + "step": 13488 + }, + { + "epoch": 1.4813309905556775, + "grad_norm": 2.2063281536102295, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7271544933319092, + "num_tokens": 346349241.0, + "step": 13489 + }, + { + "epoch": 1.4814408082582913, + "grad_norm": 2.203740119934082, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7187983989715576, + "num_tokens": 346372990.0, + "step": 13490 + }, + { + "epoch": 1.4815506259609048, + "grad_norm": 2.307077646255493, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7261638641357422, + "num_tokens": 346398734.0, + "step": 13491 + }, + { + "epoch": 1.4816604436635186, + "grad_norm": 2.358517646789551, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7234094142913818, + "num_tokens": 346422260.0, + "step": 13492 + }, + { + "epoch": 1.4817702613661323, + "grad_norm": 2.1941347122192383, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7139081954956055, + "num_tokens": 346449169.0, + "step": 13493 + }, + { + "epoch": 1.4818800790687459, + "grad_norm": 2.548995018005371, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.725333571434021, + "num_tokens": 346470590.0, + "step": 13494 + }, + { + "epoch": 1.4819898967713596, + "grad_norm": 2.390979290008545, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7572535276412964, + "num_tokens": 346489997.0, + "step": 13495 + }, + { + "epoch": 1.4820997144739732, + "grad_norm": 2.141045093536377, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7160454988479614, + "num_tokens": 346516948.0, + "step": 13496 + }, + { + "epoch": 1.482209532176587, + "grad_norm": 2.1638832092285156, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.727782130241394, + "num_tokens": 346543868.0, + "step": 13497 + }, + { + "epoch": 1.4823193498792007, + "grad_norm": 2.3574280738830566, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7395104169845581, + "num_tokens": 346566681.0, + "step": 13498 + }, + { + "epoch": 1.4824291675818142, + "grad_norm": 2.2909719944000244, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7173405885696411, + "num_tokens": 346592961.0, + "step": 13499 + }, + { + "epoch": 1.4825389852844277, + "grad_norm": 2.1814990043640137, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7081270217895508, + "num_tokens": 346620595.0, + "step": 13500 + }, + { + "epoch": 1.4826488029870415, + "grad_norm": 2.3443267345428467, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7202571034431458, + "num_tokens": 346644566.0, + "step": 13501 + }, + { + "epoch": 1.4827586206896552, + "grad_norm": 2.175659418106079, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7235316634178162, + "num_tokens": 346672291.0, + "step": 13502 + }, + { + "epoch": 1.4828684383922688, + "grad_norm": 2.387752056121826, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7224080562591553, + "num_tokens": 346696012.0, + "step": 13503 + }, + { + "epoch": 1.4829782560948825, + "grad_norm": 2.1478629112243652, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7084701061248779, + "num_tokens": 346725159.0, + "step": 13504 + }, + { + "epoch": 1.483088073797496, + "grad_norm": 2.0160953998565674, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7162674069404602, + "num_tokens": 346755961.0, + "step": 13505 + }, + { + "epoch": 1.4831978915001098, + "grad_norm": 2.0457863807678223, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.709641695022583, + "num_tokens": 346785104.0, + "step": 13506 + }, + { + "epoch": 1.4833077092027236, + "grad_norm": 2.0834457874298096, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6950356960296631, + "num_tokens": 346816677.0, + "step": 13507 + }, + { + "epoch": 1.4834175269053371, + "grad_norm": 2.3185486793518066, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7091428637504578, + "num_tokens": 346840489.0, + "step": 13508 + }, + { + "epoch": 1.4835273446079509, + "grad_norm": 2.072345018386841, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.710429847240448, + "num_tokens": 346871004.0, + "step": 13509 + }, + { + "epoch": 1.4836371623105644, + "grad_norm": 2.226609468460083, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7368481159210205, + "num_tokens": 346897153.0, + "step": 13510 + }, + { + "epoch": 1.4837469800131782, + "grad_norm": 1.9954029321670532, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.719176709651947, + "num_tokens": 346927192.0, + "step": 13511 + }, + { + "epoch": 1.4838567977157917, + "grad_norm": 2.8915913105010986, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7309573888778687, + "num_tokens": 346945717.0, + "step": 13512 + }, + { + "epoch": 1.4839666154184055, + "grad_norm": 2.44270920753479, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7328940629959106, + "num_tokens": 346968789.0, + "step": 13513 + }, + { + "epoch": 1.484076433121019, + "grad_norm": 2.2314975261688232, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7327918410301208, + "num_tokens": 346995203.0, + "step": 13514 + }, + { + "epoch": 1.4841862508236328, + "grad_norm": 2.2627339363098145, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7184435129165649, + "num_tokens": 347020652.0, + "step": 13515 + }, + { + "epoch": 1.4842960685262465, + "grad_norm": 2.497204542160034, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7167285680770874, + "num_tokens": 347043954.0, + "step": 13516 + }, + { + "epoch": 1.48440588622886, + "grad_norm": 2.541067600250244, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7390207648277283, + "num_tokens": 347065307.0, + "step": 13517 + }, + { + "epoch": 1.4845157039314738, + "grad_norm": 2.211897134780884, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7169957160949707, + "num_tokens": 347090658.0, + "step": 13518 + }, + { + "epoch": 1.4846255216340873, + "grad_norm": 2.2906875610351562, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7221961617469788, + "num_tokens": 347114886.0, + "step": 13519 + }, + { + "epoch": 1.484735339336701, + "grad_norm": 2.3850924968719482, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7359327077865601, + "num_tokens": 347137633.0, + "step": 13520 + }, + { + "epoch": 1.4848451570393149, + "grad_norm": 2.445824384689331, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7304743528366089, + "num_tokens": 347159496.0, + "step": 13521 + }, + { + "epoch": 1.4849549747419284, + "grad_norm": 2.120600700378418, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.720566987991333, + "num_tokens": 347187348.0, + "step": 13522 + }, + { + "epoch": 1.485064792444542, + "grad_norm": 2.104586601257324, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7136519551277161, + "num_tokens": 347215259.0, + "step": 13523 + }, + { + "epoch": 1.4851746101471557, + "grad_norm": 2.109442949295044, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7134804725646973, + "num_tokens": 347241940.0, + "step": 13524 + }, + { + "epoch": 1.4852844278497694, + "grad_norm": 2.2900030612945557, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.700372040271759, + "num_tokens": 347269607.0, + "step": 13525 + }, + { + "epoch": 1.485394245552383, + "grad_norm": 2.319321870803833, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7104034423828125, + "num_tokens": 347293055.0, + "step": 13526 + }, + { + "epoch": 1.4855040632549967, + "grad_norm": 2.2921955585479736, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7343862056732178, + "num_tokens": 347316416.0, + "step": 13527 + }, + { + "epoch": 1.4856138809576103, + "grad_norm": 2.2834787368774414, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7046875953674316, + "num_tokens": 347342354.0, + "step": 13528 + }, + { + "epoch": 1.485723698660224, + "grad_norm": 2.3533294200897217, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7063500285148621, + "num_tokens": 347367513.0, + "step": 13529 + }, + { + "epoch": 1.4858335163628378, + "grad_norm": 2.2862119674682617, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7070826292037964, + "num_tokens": 347392542.0, + "step": 13530 + }, + { + "epoch": 1.4859433340654513, + "grad_norm": 2.4025309085845947, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7184129953384399, + "num_tokens": 347415966.0, + "step": 13531 + }, + { + "epoch": 1.486053151768065, + "grad_norm": 2.333230495452881, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7223142981529236, + "num_tokens": 347439062.0, + "step": 13532 + }, + { + "epoch": 1.4861629694706786, + "grad_norm": 2.4332058429718018, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7109511494636536, + "num_tokens": 347461495.0, + "step": 13533 + }, + { + "epoch": 1.4862727871732924, + "grad_norm": 2.2049155235290527, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7187591195106506, + "num_tokens": 347491206.0, + "step": 13534 + }, + { + "epoch": 1.4863826048759061, + "grad_norm": 2.347435712814331, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.737949788570404, + "num_tokens": 347515685.0, + "step": 13535 + }, + { + "epoch": 1.4864924225785197, + "grad_norm": 2.2148027420043945, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7038480043411255, + "num_tokens": 347544448.0, + "step": 13536 + }, + { + "epoch": 1.4866022402811332, + "grad_norm": 2.231505870819092, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7337175607681274, + "num_tokens": 347567270.0, + "step": 13537 + }, + { + "epoch": 1.486712057983747, + "grad_norm": 2.205090284347534, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7635178565979004, + "num_tokens": 347593106.0, + "step": 13538 + }, + { + "epoch": 1.4868218756863607, + "grad_norm": 2.1098318099975586, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7228941321372986, + "num_tokens": 347622398.0, + "step": 13539 + }, + { + "epoch": 1.4869316933889742, + "grad_norm": 2.0681653022766113, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7039426565170288, + "num_tokens": 347653823.0, + "step": 13540 + }, + { + "epoch": 1.487041511091588, + "grad_norm": 2.0940849781036377, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6943424940109253, + "num_tokens": 347683169.0, + "step": 13541 + }, + { + "epoch": 1.4871513287942015, + "grad_norm": 2.24629807472229, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7165618538856506, + "num_tokens": 347707165.0, + "step": 13542 + }, + { + "epoch": 1.4872611464968153, + "grad_norm": 1.960677981376648, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.715293288230896, + "num_tokens": 347739911.0, + "step": 13543 + }, + { + "epoch": 1.487370964199429, + "grad_norm": 2.4973413944244385, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7503392696380615, + "num_tokens": 347760393.0, + "step": 13544 + }, + { + "epoch": 1.4874807819020426, + "grad_norm": 2.3503992557525635, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.752982497215271, + "num_tokens": 347783027.0, + "step": 13545 + }, + { + "epoch": 1.4875905996046563, + "grad_norm": 2.0342509746551514, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7126597762107849, + "num_tokens": 347812684.0, + "step": 13546 + }, + { + "epoch": 1.4877004173072699, + "grad_norm": 2.231945753097534, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7095422744750977, + "num_tokens": 347840413.0, + "step": 13547 + }, + { + "epoch": 1.4878102350098836, + "grad_norm": 2.4554224014282227, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7201681733131409, + "num_tokens": 347861887.0, + "step": 13548 + }, + { + "epoch": 1.4879200527124974, + "grad_norm": 2.3102474212646484, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7098186016082764, + "num_tokens": 347887298.0, + "step": 13549 + }, + { + "epoch": 1.488029870415111, + "grad_norm": 2.460820198059082, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7252469062805176, + "num_tokens": 347909624.0, + "step": 13550 + }, + { + "epoch": 1.4881396881177245, + "grad_norm": 2.1581335067749023, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7224655747413635, + "num_tokens": 347935806.0, + "step": 13551 + }, + { + "epoch": 1.4882495058203382, + "grad_norm": 2.233759880065918, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7176952362060547, + "num_tokens": 347959888.0, + "step": 13552 + }, + { + "epoch": 1.488359323522952, + "grad_norm": 2.2452118396759033, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7361800074577332, + "num_tokens": 347985155.0, + "step": 13553 + }, + { + "epoch": 1.4884691412255655, + "grad_norm": 2.4644625186920166, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.725265383720398, + "num_tokens": 348006752.0, + "step": 13554 + }, + { + "epoch": 1.4885789589281793, + "grad_norm": 2.377739906311035, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7214177846908569, + "num_tokens": 348031950.0, + "step": 13555 + }, + { + "epoch": 1.4886887766307928, + "grad_norm": 2.3127176761627197, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7152753472328186, + "num_tokens": 348056703.0, + "step": 13556 + }, + { + "epoch": 1.4887985943334066, + "grad_norm": 1.9409781694412231, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.70360267162323, + "num_tokens": 348091499.0, + "step": 13557 + }, + { + "epoch": 1.4889084120360203, + "grad_norm": 2.250742197036743, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7171332240104675, + "num_tokens": 348115932.0, + "step": 13558 + }, + { + "epoch": 1.4890182297386338, + "grad_norm": 2.3633899688720703, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7179258465766907, + "num_tokens": 348139085.0, + "step": 13559 + }, + { + "epoch": 1.4891280474412476, + "grad_norm": 2.355708122253418, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7191481590270996, + "num_tokens": 348162461.0, + "step": 13560 + }, + { + "epoch": 1.4892378651438611, + "grad_norm": 2.2814102172851562, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6897190809249878, + "num_tokens": 348189402.0, + "step": 13561 + }, + { + "epoch": 1.489347682846475, + "grad_norm": 2.1782515048980713, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7074475288391113, + "num_tokens": 348216063.0, + "step": 13562 + }, + { + "epoch": 1.4894575005490884, + "grad_norm": 2.466284990310669, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7033891081809998, + "num_tokens": 348240288.0, + "step": 13563 + }, + { + "epoch": 1.4895673182517022, + "grad_norm": 2.499584197998047, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7203304767608643, + "num_tokens": 348263374.0, + "step": 13564 + }, + { + "epoch": 1.4896771359543157, + "grad_norm": 2.0930404663085938, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6971487402915955, + "num_tokens": 348292768.0, + "step": 13565 + }, + { + "epoch": 1.4897869536569295, + "grad_norm": 2.2976386547088623, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7347357273101807, + "num_tokens": 348317415.0, + "step": 13566 + }, + { + "epoch": 1.4898967713595432, + "grad_norm": 2.227511405944824, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.727226734161377, + "num_tokens": 348342733.0, + "step": 13567 + }, + { + "epoch": 1.4900065890621568, + "grad_norm": 2.019881248474121, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7310437560081482, + "num_tokens": 348374147.0, + "step": 13568 + }, + { + "epoch": 1.4901164067647705, + "grad_norm": 2.1068058013916016, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7203123569488525, + "num_tokens": 348404473.0, + "step": 13569 + }, + { + "epoch": 1.490226224467384, + "grad_norm": 2.27193546295166, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.72132408618927, + "num_tokens": 348430261.0, + "step": 13570 + }, + { + "epoch": 1.4903360421699978, + "grad_norm": 2.2082631587982178, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7194486856460571, + "num_tokens": 348456982.0, + "step": 13571 + }, + { + "epoch": 1.4904458598726116, + "grad_norm": 2.647322654724121, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7544776201248169, + "num_tokens": 348475874.0, + "step": 13572 + }, + { + "epoch": 1.4905556775752251, + "grad_norm": 1.9350621700286865, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.722380518913269, + "num_tokens": 348507186.0, + "step": 13573 + }, + { + "epoch": 1.4906654952778389, + "grad_norm": 2.2854678630828857, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7123830318450928, + "num_tokens": 348533100.0, + "step": 13574 + }, + { + "epoch": 1.4907753129804524, + "grad_norm": 2.6553361415863037, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7093202471733093, + "num_tokens": 348552508.0, + "step": 13575 + }, + { + "epoch": 1.4908851306830662, + "grad_norm": 2.22397780418396, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7143668532371521, + "num_tokens": 348577208.0, + "step": 13576 + }, + { + "epoch": 1.4909949483856797, + "grad_norm": 2.165715217590332, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7056989669799805, + "num_tokens": 348605324.0, + "step": 13577 + }, + { + "epoch": 1.4911047660882935, + "grad_norm": 2.0525732040405273, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7091268301010132, + "num_tokens": 348637829.0, + "step": 13578 + }, + { + "epoch": 1.491214583790907, + "grad_norm": 1.9581801891326904, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7125429511070251, + "num_tokens": 348670174.0, + "step": 13579 + }, + { + "epoch": 1.4913244014935207, + "grad_norm": 2.4559402465820312, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7032545804977417, + "num_tokens": 348694218.0, + "step": 13580 + }, + { + "epoch": 1.4914342191961345, + "grad_norm": 2.57436203956604, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7516036033630371, + "num_tokens": 348713179.0, + "step": 13581 + }, + { + "epoch": 1.491544036898748, + "grad_norm": 2.4232594966888428, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7362299561500549, + "num_tokens": 348735207.0, + "step": 13582 + }, + { + "epoch": 1.4916538546013618, + "grad_norm": 2.2936172485351562, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7210618257522583, + "num_tokens": 348759562.0, + "step": 13583 + }, + { + "epoch": 1.4917636723039753, + "grad_norm": 2.4088473320007324, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7379122972488403, + "num_tokens": 348782460.0, + "step": 13584 + }, + { + "epoch": 1.491873490006589, + "grad_norm": 2.255537509918213, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6936074495315552, + "num_tokens": 348808878.0, + "step": 13585 + }, + { + "epoch": 1.4919833077092028, + "grad_norm": 2.1879420280456543, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7396312952041626, + "num_tokens": 348835258.0, + "step": 13586 + }, + { + "epoch": 1.4920931254118164, + "grad_norm": 2.4021503925323486, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7239018082618713, + "num_tokens": 348858970.0, + "step": 13587 + }, + { + "epoch": 1.49220294311443, + "grad_norm": 2.3956987857818604, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7216792702674866, + "num_tokens": 348883050.0, + "step": 13588 + }, + { + "epoch": 1.4923127608170437, + "grad_norm": 2.2397983074188232, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7096840143203735, + "num_tokens": 348909837.0, + "step": 13589 + }, + { + "epoch": 1.4924225785196574, + "grad_norm": 2.347806453704834, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7263513803482056, + "num_tokens": 348935216.0, + "step": 13590 + }, + { + "epoch": 1.492532396222271, + "grad_norm": 2.0982789993286133, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7239100933074951, + "num_tokens": 348963853.0, + "step": 13591 + }, + { + "epoch": 1.4926422139248847, + "grad_norm": 2.3926286697387695, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7420775890350342, + "num_tokens": 348986397.0, + "step": 13592 + }, + { + "epoch": 1.4927520316274983, + "grad_norm": 2.4671547412872314, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7224640250205994, + "num_tokens": 349008172.0, + "step": 13593 + }, + { + "epoch": 1.492861849330112, + "grad_norm": 2.077036142349243, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7533021569252014, + "num_tokens": 349034244.0, + "step": 13594 + }, + { + "epoch": 1.4929716670327258, + "grad_norm": 2.3189754486083984, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7255513668060303, + "num_tokens": 349059679.0, + "step": 13595 + }, + { + "epoch": 1.4930814847353393, + "grad_norm": 2.2656772136688232, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7154678106307983, + "num_tokens": 349084897.0, + "step": 13596 + }, + { + "epoch": 1.493191302437953, + "grad_norm": 2.318751335144043, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7382886409759521, + "num_tokens": 349108537.0, + "step": 13597 + }, + { + "epoch": 1.4933011201405666, + "grad_norm": 2.265798807144165, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7422919273376465, + "num_tokens": 349132220.0, + "step": 13598 + }, + { + "epoch": 1.4934109378431804, + "grad_norm": 2.2135794162750244, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7050321102142334, + "num_tokens": 349160905.0, + "step": 13599 + }, + { + "epoch": 1.493520755545794, + "grad_norm": 2.1131508350372314, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.697504997253418, + "num_tokens": 349189944.0, + "step": 13600 + }, + { + "epoch": 1.4936305732484076, + "grad_norm": 2.583550214767456, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7256692051887512, + "num_tokens": 349209953.0, + "step": 13601 + }, + { + "epoch": 1.4937403909510212, + "grad_norm": 2.0435774326324463, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7328746318817139, + "num_tokens": 349239797.0, + "step": 13602 + }, + { + "epoch": 1.493850208653635, + "grad_norm": 2.063086748123169, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7094619274139404, + "num_tokens": 349270554.0, + "step": 13603 + }, + { + "epoch": 1.4939600263562487, + "grad_norm": 2.4210410118103027, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7403060793876648, + "num_tokens": 349291341.0, + "step": 13604 + }, + { + "epoch": 1.4940698440588622, + "grad_norm": 2.1155221462249756, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7205075025558472, + "num_tokens": 349319254.0, + "step": 13605 + }, + { + "epoch": 1.494179661761476, + "grad_norm": 2.2240703105926514, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7561287879943848, + "num_tokens": 349343739.0, + "step": 13606 + }, + { + "epoch": 1.4942894794640895, + "grad_norm": 2.251675844192505, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.731632649898529, + "num_tokens": 349368013.0, + "step": 13607 + }, + { + "epoch": 1.4943992971667033, + "grad_norm": 2.2541685104370117, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7200067639350891, + "num_tokens": 349395288.0, + "step": 13608 + }, + { + "epoch": 1.494509114869317, + "grad_norm": 2.254960775375366, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7260252237319946, + "num_tokens": 349422761.0, + "step": 13609 + }, + { + "epoch": 1.4946189325719306, + "grad_norm": 2.2778728008270264, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7164419889450073, + "num_tokens": 349448237.0, + "step": 13610 + }, + { + "epoch": 1.4947287502745443, + "grad_norm": 2.479829788208008, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7137542366981506, + "num_tokens": 349468733.0, + "step": 13611 + }, + { + "epoch": 1.4948385679771579, + "grad_norm": 2.072601795196533, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6924980878829956, + "num_tokens": 349498259.0, + "step": 13612 + }, + { + "epoch": 1.4949483856797716, + "grad_norm": 2.117344856262207, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7058314085006714, + "num_tokens": 349528325.0, + "step": 13613 + }, + { + "epoch": 1.4950582033823854, + "grad_norm": 2.275378942489624, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7182155847549438, + "num_tokens": 349553215.0, + "step": 13614 + }, + { + "epoch": 1.495168021084999, + "grad_norm": 2.152534246444702, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6981663107872009, + "num_tokens": 349581334.0, + "step": 13615 + }, + { + "epoch": 1.4952778387876124, + "grad_norm": 2.344817638397217, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7259414792060852, + "num_tokens": 349605857.0, + "step": 13616 + }, + { + "epoch": 1.4953876564902262, + "grad_norm": 2.1933934688568115, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6983983516693115, + "num_tokens": 349633502.0, + "step": 13617 + }, + { + "epoch": 1.49549747419284, + "grad_norm": 2.215147018432617, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7316169142723083, + "num_tokens": 349659549.0, + "step": 13618 + }, + { + "epoch": 1.4956072918954535, + "grad_norm": 2.066215753555298, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7038174867630005, + "num_tokens": 349688090.0, + "step": 13619 + }, + { + "epoch": 1.4957171095980673, + "grad_norm": 2.204420328140259, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7023113369941711, + "num_tokens": 349716306.0, + "step": 13620 + }, + { + "epoch": 1.4958269273006808, + "grad_norm": 2.35371732711792, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7201199531555176, + "num_tokens": 349741808.0, + "step": 13621 + }, + { + "epoch": 1.4959367450032945, + "grad_norm": 2.314805030822754, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7345870733261108, + "num_tokens": 349766127.0, + "step": 13622 + }, + { + "epoch": 1.4960465627059083, + "grad_norm": 2.174065589904785, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7248876094818115, + "num_tokens": 349793900.0, + "step": 13623 + }, + { + "epoch": 1.4961563804085218, + "grad_norm": 2.338526487350464, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.741471529006958, + "num_tokens": 349818079.0, + "step": 13624 + }, + { + "epoch": 1.4962661981111356, + "grad_norm": 2.0423645973205566, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6851375699043274, + "num_tokens": 349850582.0, + "step": 13625 + }, + { + "epoch": 1.4963760158137491, + "grad_norm": 2.0644803047180176, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7135784029960632, + "num_tokens": 349880216.0, + "step": 13626 + }, + { + "epoch": 1.4964858335163629, + "grad_norm": 2.0973894596099854, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7111515998840332, + "num_tokens": 349910875.0, + "step": 13627 + }, + { + "epoch": 1.4965956512189764, + "grad_norm": 2.4081594944000244, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7192096710205078, + "num_tokens": 349934139.0, + "step": 13628 + }, + { + "epoch": 1.4967054689215902, + "grad_norm": 2.341021776199341, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7046738862991333, + "num_tokens": 349958819.0, + "step": 13629 + }, + { + "epoch": 1.4968152866242037, + "grad_norm": 2.7303171157836914, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7284554243087769, + "num_tokens": 349976933.0, + "step": 13630 + }, + { + "epoch": 1.4969251043268175, + "grad_norm": 2.0878281593322754, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6844593286514282, + "num_tokens": 350005843.0, + "step": 13631 + }, + { + "epoch": 1.4970349220294312, + "grad_norm": 2.3306219577789307, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7539491653442383, + "num_tokens": 350027407.0, + "step": 13632 + }, + { + "epoch": 1.4971447397320448, + "grad_norm": 2.566690683364868, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7206164598464966, + "num_tokens": 350049304.0, + "step": 13633 + }, + { + "epoch": 1.4972545574346585, + "grad_norm": 2.648108720779419, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.74235999584198, + "num_tokens": 350068494.0, + "step": 13634 + }, + { + "epoch": 1.497364375137272, + "grad_norm": 2.06020450592041, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7212471961975098, + "num_tokens": 350099407.0, + "step": 13635 + }, + { + "epoch": 1.4974741928398858, + "grad_norm": 2.3556158542633057, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7346138954162598, + "num_tokens": 350122218.0, + "step": 13636 + }, + { + "epoch": 1.4975840105424996, + "grad_norm": 2.4466681480407715, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7266885042190552, + "num_tokens": 350144883.0, + "step": 13637 + }, + { + "epoch": 1.497693828245113, + "grad_norm": 2.1166253089904785, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7181836366653442, + "num_tokens": 350174800.0, + "step": 13638 + }, + { + "epoch": 1.4978036459477266, + "grad_norm": 2.1159825325012207, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6988276839256287, + "num_tokens": 350203854.0, + "step": 13639 + }, + { + "epoch": 1.4979134636503404, + "grad_norm": 2.2078444957733154, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7358630895614624, + "num_tokens": 350231548.0, + "step": 13640 + }, + { + "epoch": 1.4980232813529542, + "grad_norm": 1.9952234029769897, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7116729617118835, + "num_tokens": 350265346.0, + "step": 13641 + }, + { + "epoch": 1.4981330990555677, + "grad_norm": 2.712728261947632, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7083487510681152, + "num_tokens": 350284870.0, + "step": 13642 + }, + { + "epoch": 1.4982429167581814, + "grad_norm": 2.69490909576416, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7370673418045044, + "num_tokens": 350304201.0, + "step": 13643 + }, + { + "epoch": 1.498352734460795, + "grad_norm": 2.307410478591919, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7193376421928406, + "num_tokens": 350326701.0, + "step": 13644 + }, + { + "epoch": 1.4984625521634087, + "grad_norm": 2.1168344020843506, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7296065092086792, + "num_tokens": 350353687.0, + "step": 13645 + }, + { + "epoch": 1.4985723698660225, + "grad_norm": 2.4893407821655273, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7368435859680176, + "num_tokens": 350373028.0, + "step": 13646 + }, + { + "epoch": 1.498682187568636, + "grad_norm": 2.2049448490142822, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.734261155128479, + "num_tokens": 350397717.0, + "step": 13647 + }, + { + "epoch": 1.4987920052712498, + "grad_norm": 2.1685826778411865, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7022011876106262, + "num_tokens": 350427189.0, + "step": 13648 + }, + { + "epoch": 1.4989018229738633, + "grad_norm": 2.7497682571411133, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7187372446060181, + "num_tokens": 350445012.0, + "step": 13649 + }, + { + "epoch": 1.499011640676477, + "grad_norm": 2.4501850605010986, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7256288528442383, + "num_tokens": 350466043.0, + "step": 13650 + }, + { + "epoch": 1.4991214583790908, + "grad_norm": 2.1447036266326904, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.723315954208374, + "num_tokens": 350493062.0, + "step": 13651 + }, + { + "epoch": 1.4992312760817044, + "grad_norm": 2.283808708190918, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7067517638206482, + "num_tokens": 350520773.0, + "step": 13652 + }, + { + "epoch": 1.499341093784318, + "grad_norm": 2.5659706592559814, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7189777493476868, + "num_tokens": 350544184.0, + "step": 13653 + }, + { + "epoch": 1.4994509114869317, + "grad_norm": 2.020826816558838, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7138595581054688, + "num_tokens": 350571982.0, + "step": 13654 + }, + { + "epoch": 1.4995607291895454, + "grad_norm": 3.216193437576294, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7294110059738159, + "num_tokens": 350593704.0, + "step": 13655 + }, + { + "epoch": 1.499670546892159, + "grad_norm": 2.2866528034210205, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7311728000640869, + "num_tokens": 350617755.0, + "step": 13656 + }, + { + "epoch": 1.4997803645947727, + "grad_norm": 2.2219762802124023, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7185872197151184, + "num_tokens": 350643121.0, + "step": 13657 + }, + { + "epoch": 1.4998901822973862, + "grad_norm": 2.2766025066375732, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7213799357414246, + "num_tokens": 350667868.0, + "step": 13658 + }, + { + "epoch": 1.5, + "grad_norm": 2.4631381034851074, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7171552777290344, + "num_tokens": 350692931.0, + "step": 13659 + }, + { + "epoch": 1.5001098177026138, + "grad_norm": 2.413595199584961, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7079167366027832, + "num_tokens": 350716506.0, + "step": 13660 + }, + { + "epoch": 1.5002196354052273, + "grad_norm": 2.52653169631958, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7290574312210083, + "num_tokens": 350737868.0, + "step": 13661 + }, + { + "epoch": 1.5003294531078408, + "grad_norm": 1.8067089319229126, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7258228659629822, + "num_tokens": 350774120.0, + "step": 13662 + }, + { + "epoch": 1.5004392708104546, + "grad_norm": 2.2336015701293945, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7261373400688171, + "num_tokens": 350798619.0, + "step": 13663 + }, + { + "epoch": 1.5005490885130683, + "grad_norm": 2.1291210651397705, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7109968662261963, + "num_tokens": 350826485.0, + "step": 13664 + }, + { + "epoch": 1.500658906215682, + "grad_norm": 2.063446283340454, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7146495580673218, + "num_tokens": 350856275.0, + "step": 13665 + }, + { + "epoch": 1.5007687239182956, + "grad_norm": 2.00701642036438, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7111330032348633, + "num_tokens": 350887983.0, + "step": 13666 + }, + { + "epoch": 1.5008785416209092, + "grad_norm": 2.2621006965637207, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6970498561859131, + "num_tokens": 350915368.0, + "step": 13667 + }, + { + "epoch": 1.500988359323523, + "grad_norm": 1.9648643732070923, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7109416127204895, + "num_tokens": 350948122.0, + "step": 13668 + }, + { + "epoch": 1.5010981770261367, + "grad_norm": 2.5249857902526855, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7262792587280273, + "num_tokens": 350968454.0, + "step": 13669 + }, + { + "epoch": 1.5012079947287504, + "grad_norm": 2.2869389057159424, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7483358383178711, + "num_tokens": 350992978.0, + "step": 13670 + }, + { + "epoch": 1.501317812431364, + "grad_norm": 2.402514696121216, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7269207239151001, + "num_tokens": 351014920.0, + "step": 13671 + }, + { + "epoch": 1.5014276301339775, + "grad_norm": 2.2914061546325684, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7270476818084717, + "num_tokens": 351041235.0, + "step": 13672 + }, + { + "epoch": 1.5015374478365913, + "grad_norm": 2.439377546310425, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7071993350982666, + "num_tokens": 351064925.0, + "step": 13673 + }, + { + "epoch": 1.501647265539205, + "grad_norm": 2.1337077617645264, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7219741344451904, + "num_tokens": 351092818.0, + "step": 13674 + }, + { + "epoch": 1.5017570832418186, + "grad_norm": 2.826411485671997, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7366406917572021, + "num_tokens": 351109741.0, + "step": 13675 + }, + { + "epoch": 1.501866900944432, + "grad_norm": 2.199554443359375, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7228773236274719, + "num_tokens": 351135201.0, + "step": 13676 + }, + { + "epoch": 1.5019767186470458, + "grad_norm": 2.332993507385254, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7116144895553589, + "num_tokens": 351159395.0, + "step": 13677 + }, + { + "epoch": 1.5020865363496596, + "grad_norm": 2.50597882270813, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.73401939868927, + "num_tokens": 351179889.0, + "step": 13678 + }, + { + "epoch": 1.5021963540522734, + "grad_norm": 2.4066269397735596, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7195080518722534, + "num_tokens": 351204405.0, + "step": 13679 + }, + { + "epoch": 1.502306171754887, + "grad_norm": 2.064622640609741, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6891070008277893, + "num_tokens": 351235788.0, + "step": 13680 + }, + { + "epoch": 1.5024159894575004, + "grad_norm": 2.7615270614624023, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.723336935043335, + "num_tokens": 351253689.0, + "step": 13681 + }, + { + "epoch": 1.5025258071601142, + "grad_norm": 2.0121312141418457, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7358212471008301, + "num_tokens": 351284128.0, + "step": 13682 + }, + { + "epoch": 1.502635624862728, + "grad_norm": 2.6829981803894043, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7564007043838501, + "num_tokens": 351301716.0, + "step": 13683 + }, + { + "epoch": 1.5027454425653415, + "grad_norm": 2.2364258766174316, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7206529378890991, + "num_tokens": 351326147.0, + "step": 13684 + }, + { + "epoch": 1.5028552602679552, + "grad_norm": 2.6885337829589844, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7141369581222534, + "num_tokens": 351346754.0, + "step": 13685 + }, + { + "epoch": 1.5029650779705688, + "grad_norm": 2.112761974334717, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7434145212173462, + "num_tokens": 351374122.0, + "step": 13686 + }, + { + "epoch": 1.5030748956731825, + "grad_norm": 2.354710578918457, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7189544439315796, + "num_tokens": 351397899.0, + "step": 13687 + }, + { + "epoch": 1.5031847133757963, + "grad_norm": 2.647118091583252, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7181843519210815, + "num_tokens": 351418480.0, + "step": 13688 + }, + { + "epoch": 1.5032945310784098, + "grad_norm": 2.5041844844818115, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7334558963775635, + "num_tokens": 351440511.0, + "step": 13689 + }, + { + "epoch": 1.5034043487810234, + "grad_norm": 2.435852289199829, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7317674160003662, + "num_tokens": 351461592.0, + "step": 13690 + }, + { + "epoch": 1.5035141664836371, + "grad_norm": 2.2782185077667236, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7287797927856445, + "num_tokens": 351485278.0, + "step": 13691 + }, + { + "epoch": 1.5036239841862509, + "grad_norm": 1.9938790798187256, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7210502624511719, + "num_tokens": 351516165.0, + "step": 13692 + }, + { + "epoch": 1.5037338018888646, + "grad_norm": 2.049797296524048, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7098734378814697, + "num_tokens": 351547158.0, + "step": 13693 + }, + { + "epoch": 1.5038436195914782, + "grad_norm": 2.3123836517333984, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7168104648590088, + "num_tokens": 351571310.0, + "step": 13694 + }, + { + "epoch": 1.5039534372940917, + "grad_norm": 2.3017799854278564, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7307100296020508, + "num_tokens": 351595728.0, + "step": 13695 + }, + { + "epoch": 1.5040632549967055, + "grad_norm": 2.1385233402252197, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7091992497444153, + "num_tokens": 351624533.0, + "step": 13696 + }, + { + "epoch": 1.5041730726993192, + "grad_norm": 2.2722108364105225, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7174160480499268, + "num_tokens": 351651357.0, + "step": 13697 + }, + { + "epoch": 1.5042828904019327, + "grad_norm": 2.0950939655303955, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7084718346595764, + "num_tokens": 351681444.0, + "step": 13698 + }, + { + "epoch": 1.5043927081045465, + "grad_norm": 2.177302360534668, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7300865650177002, + "num_tokens": 351708225.0, + "step": 13699 + }, + { + "epoch": 1.50450252580716, + "grad_norm": 2.1769120693206787, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.712369441986084, + "num_tokens": 351734838.0, + "step": 13700 + }, + { + "epoch": 1.5046123435097738, + "grad_norm": 1.9572352170944214, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7031128406524658, + "num_tokens": 351766806.0, + "step": 13701 + }, + { + "epoch": 1.5047221612123876, + "grad_norm": 2.3828701972961426, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7204788327217102, + "num_tokens": 351791246.0, + "step": 13702 + }, + { + "epoch": 1.504831978915001, + "grad_norm": 2.10357928276062, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7280958890914917, + "num_tokens": 351818879.0, + "step": 13703 + }, + { + "epoch": 1.5049417966176146, + "grad_norm": 2.435914993286133, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7448724508285522, + "num_tokens": 351839485.0, + "step": 13704 + }, + { + "epoch": 1.5050516143202284, + "grad_norm": 2.5782992839813232, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.728830099105835, + "num_tokens": 351859001.0, + "step": 13705 + }, + { + "epoch": 1.5051614320228421, + "grad_norm": 2.371915817260742, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7396378517150879, + "num_tokens": 351880986.0, + "step": 13706 + }, + { + "epoch": 1.505271249725456, + "grad_norm": 2.467276096343994, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7222921252250671, + "num_tokens": 351903349.0, + "step": 13707 + }, + { + "epoch": 1.5053810674280694, + "grad_norm": 2.420638084411621, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7453053593635559, + "num_tokens": 351924946.0, + "step": 13708 + }, + { + "epoch": 1.505490885130683, + "grad_norm": 2.4583241939544678, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7198540568351746, + "num_tokens": 351947812.0, + "step": 13709 + }, + { + "epoch": 1.5056007028332967, + "grad_norm": 2.5840260982513428, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7223127484321594, + "num_tokens": 351968082.0, + "step": 13710 + }, + { + "epoch": 1.5057105205359105, + "grad_norm": 2.519355297088623, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7342169880867004, + "num_tokens": 351988712.0, + "step": 13711 + }, + { + "epoch": 1.505820338238524, + "grad_norm": 2.073396921157837, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7335925102233887, + "num_tokens": 352015832.0, + "step": 13712 + }, + { + "epoch": 1.5059301559411375, + "grad_norm": 2.3966331481933594, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7314496040344238, + "num_tokens": 352039794.0, + "step": 13713 + }, + { + "epoch": 1.5060399736437513, + "grad_norm": 2.2386550903320312, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7072901725769043, + "num_tokens": 352066191.0, + "step": 13714 + }, + { + "epoch": 1.506149791346365, + "grad_norm": 2.130542278289795, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7283924221992493, + "num_tokens": 352092119.0, + "step": 13715 + }, + { + "epoch": 1.5062596090489788, + "grad_norm": 2.221177816390991, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7453873753547668, + "num_tokens": 352117679.0, + "step": 13716 + }, + { + "epoch": 1.5063694267515924, + "grad_norm": 2.5517547130584717, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7185661792755127, + "num_tokens": 352139101.0, + "step": 13717 + }, + { + "epoch": 1.506479244454206, + "grad_norm": 1.9860635995864868, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7388123869895935, + "num_tokens": 352168014.0, + "step": 13718 + }, + { + "epoch": 1.5065890621568196, + "grad_norm": 2.1422855854034424, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.713955819606781, + "num_tokens": 352194216.0, + "step": 13719 + }, + { + "epoch": 1.5066988798594334, + "grad_norm": 2.270630121231079, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7152932286262512, + "num_tokens": 352219361.0, + "step": 13720 + }, + { + "epoch": 1.5068086975620472, + "grad_norm": 2.2422637939453125, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7183873653411865, + "num_tokens": 352244863.0, + "step": 13721 + }, + { + "epoch": 1.5069185152646607, + "grad_norm": 1.9505369663238525, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7074293494224548, + "num_tokens": 352276831.0, + "step": 13722 + }, + { + "epoch": 1.5070283329672742, + "grad_norm": 2.715604066848755, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7224783897399902, + "num_tokens": 352296680.0, + "step": 13723 + }, + { + "epoch": 1.507138150669888, + "grad_norm": 2.2700421810150146, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7142527103424072, + "num_tokens": 352323042.0, + "step": 13724 + }, + { + "epoch": 1.5072479683725017, + "grad_norm": 2.103510618209839, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7138043642044067, + "num_tokens": 352350398.0, + "step": 13725 + }, + { + "epoch": 1.5073577860751153, + "grad_norm": 2.2126219272613525, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.724911093711853, + "num_tokens": 352377883.0, + "step": 13726 + }, + { + "epoch": 1.5074676037777288, + "grad_norm": 2.2823359966278076, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7137807011604309, + "num_tokens": 352404592.0, + "step": 13727 + }, + { + "epoch": 1.5075774214803426, + "grad_norm": 2.0949547290802, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7141174077987671, + "num_tokens": 352433643.0, + "step": 13728 + }, + { + "epoch": 1.5076872391829563, + "grad_norm": 2.102102518081665, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7123895883560181, + "num_tokens": 352460996.0, + "step": 13729 + }, + { + "epoch": 1.50779705688557, + "grad_norm": 2.1739320755004883, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7189652919769287, + "num_tokens": 352489795.0, + "step": 13730 + }, + { + "epoch": 1.5079068745881836, + "grad_norm": 2.0022952556610107, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7486746907234192, + "num_tokens": 352520861.0, + "step": 13731 + }, + { + "epoch": 1.5080166922907972, + "grad_norm": 2.0490565299987793, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7121257781982422, + "num_tokens": 352550612.0, + "step": 13732 + }, + { + "epoch": 1.508126509993411, + "grad_norm": 2.5824086666107178, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7295281291007996, + "num_tokens": 352571304.0, + "step": 13733 + }, + { + "epoch": 1.5082363276960247, + "grad_norm": 2.1332874298095703, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7097442150115967, + "num_tokens": 352598956.0, + "step": 13734 + }, + { + "epoch": 1.5083461453986382, + "grad_norm": 2.281843900680542, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7154910564422607, + "num_tokens": 352623163.0, + "step": 13735 + }, + { + "epoch": 1.508455963101252, + "grad_norm": 2.2651495933532715, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7116191387176514, + "num_tokens": 352650777.0, + "step": 13736 + }, + { + "epoch": 1.5085657808038655, + "grad_norm": 2.293773651123047, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6974729299545288, + "num_tokens": 352677997.0, + "step": 13737 + }, + { + "epoch": 1.5086755985064793, + "grad_norm": 2.1972265243530273, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7271840572357178, + "num_tokens": 352704588.0, + "step": 13738 + }, + { + "epoch": 1.508785416209093, + "grad_norm": 2.121772527694702, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7156933546066284, + "num_tokens": 352734508.0, + "step": 13739 + }, + { + "epoch": 1.5088952339117065, + "grad_norm": 1.8773624897003174, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6957768797874451, + "num_tokens": 352772490.0, + "step": 13740 + }, + { + "epoch": 1.50900505161432, + "grad_norm": 2.468397378921509, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7281441688537598, + "num_tokens": 352794723.0, + "step": 13741 + }, + { + "epoch": 1.5091148693169338, + "grad_norm": 2.33886456489563, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7098994255065918, + "num_tokens": 352819339.0, + "step": 13742 + }, + { + "epoch": 1.5092246870195476, + "grad_norm": 2.107707977294922, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7008683085441589, + "num_tokens": 352848861.0, + "step": 13743 + }, + { + "epoch": 1.5093345047221614, + "grad_norm": 2.383753538131714, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.69740891456604, + "num_tokens": 352874006.0, + "step": 13744 + }, + { + "epoch": 1.5094443224247749, + "grad_norm": 2.2607836723327637, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.716598391532898, + "num_tokens": 352899812.0, + "step": 13745 + }, + { + "epoch": 1.5095541401273884, + "grad_norm": 2.2366652488708496, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7241480350494385, + "num_tokens": 352925004.0, + "step": 13746 + }, + { + "epoch": 1.5096639578300022, + "grad_norm": 2.111448049545288, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7286968231201172, + "num_tokens": 352951551.0, + "step": 13747 + }, + { + "epoch": 1.509773775532616, + "grad_norm": 2.10972261428833, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.720369815826416, + "num_tokens": 352981457.0, + "step": 13748 + }, + { + "epoch": 1.5098835932352295, + "grad_norm": 2.231289863586426, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7202029228210449, + "num_tokens": 353007473.0, + "step": 13749 + }, + { + "epoch": 1.5099934109378432, + "grad_norm": 2.3074893951416016, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7371645569801331, + "num_tokens": 353031574.0, + "step": 13750 + }, + { + "epoch": 1.5101032286404568, + "grad_norm": 2.2171995639801025, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7181527614593506, + "num_tokens": 353059373.0, + "step": 13751 + }, + { + "epoch": 1.5102130463430705, + "grad_norm": 2.021186590194702, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7232398986816406, + "num_tokens": 353090703.0, + "step": 13752 + }, + { + "epoch": 1.5103228640456843, + "grad_norm": 2.2893483638763428, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.746288537979126, + "num_tokens": 353114032.0, + "step": 13753 + }, + { + "epoch": 1.5104326817482978, + "grad_norm": 2.062985420227051, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.70536208152771, + "num_tokens": 353144704.0, + "step": 13754 + }, + { + "epoch": 1.5105424994509113, + "grad_norm": 2.060260057449341, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7217966914176941, + "num_tokens": 353172119.0, + "step": 13755 + }, + { + "epoch": 1.510652317153525, + "grad_norm": 2.5054256916046143, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.737234354019165, + "num_tokens": 353193515.0, + "step": 13756 + }, + { + "epoch": 1.5107621348561389, + "grad_norm": 2.259533643722534, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7179688215255737, + "num_tokens": 353216988.0, + "step": 13757 + }, + { + "epoch": 1.5108719525587526, + "grad_norm": 2.0345966815948486, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7032017707824707, + "num_tokens": 353245188.0, + "step": 13758 + }, + { + "epoch": 1.5109817702613662, + "grad_norm": 2.5145559310913086, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7583695650100708, + "num_tokens": 353266468.0, + "step": 13759 + }, + { + "epoch": 1.5110915879639797, + "grad_norm": 2.2470335960388184, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6789606213569641, + "num_tokens": 353294144.0, + "step": 13760 + }, + { + "epoch": 1.5112014056665934, + "grad_norm": 2.362919569015503, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7406508922576904, + "num_tokens": 353316627.0, + "step": 13761 + }, + { + "epoch": 1.5113112233692072, + "grad_norm": 2.4334306716918945, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7352200746536255, + "num_tokens": 353340109.0, + "step": 13762 + }, + { + "epoch": 1.5114210410718207, + "grad_norm": 2.094569683074951, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6918519139289856, + "num_tokens": 353370197.0, + "step": 13763 + }, + { + "epoch": 1.5115308587744345, + "grad_norm": 2.220418691635132, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7335904240608215, + "num_tokens": 353396979.0, + "step": 13764 + }, + { + "epoch": 1.511640676477048, + "grad_norm": 2.169187307357788, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7227925062179565, + "num_tokens": 353424215.0, + "step": 13765 + }, + { + "epoch": 1.5117504941796618, + "grad_norm": 2.2050158977508545, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7375777363777161, + "num_tokens": 353449134.0, + "step": 13766 + }, + { + "epoch": 1.5118603118822755, + "grad_norm": 2.157461643218994, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7200168967247009, + "num_tokens": 353476610.0, + "step": 13767 + }, + { + "epoch": 1.511970129584889, + "grad_norm": 1.980293869972229, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7255548238754272, + "num_tokens": 353506847.0, + "step": 13768 + }, + { + "epoch": 1.5120799472875026, + "grad_norm": 2.4076735973358154, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7100081443786621, + "num_tokens": 353530425.0, + "step": 13769 + }, + { + "epoch": 1.5121897649901164, + "grad_norm": 2.346547842025757, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7160966992378235, + "num_tokens": 353554601.0, + "step": 13770 + }, + { + "epoch": 1.5122995826927301, + "grad_norm": 1.966995120048523, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7163041234016418, + "num_tokens": 353587802.0, + "step": 13771 + }, + { + "epoch": 1.5124094003953439, + "grad_norm": 2.4078996181488037, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7215684652328491, + "num_tokens": 353610076.0, + "step": 13772 + }, + { + "epoch": 1.5125192180979574, + "grad_norm": 2.11057448387146, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7040895223617554, + "num_tokens": 353637827.0, + "step": 13773 + }, + { + "epoch": 1.512629035800571, + "grad_norm": 2.1819653511047363, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7119901776313782, + "num_tokens": 353666396.0, + "step": 13774 + }, + { + "epoch": 1.5127388535031847, + "grad_norm": 2.5325074195861816, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7342748641967773, + "num_tokens": 353687208.0, + "step": 13775 + }, + { + "epoch": 1.5128486712057985, + "grad_norm": 1.867386817932129, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7198036313056946, + "num_tokens": 353719494.0, + "step": 13776 + }, + { + "epoch": 1.512958488908412, + "grad_norm": 2.028197765350342, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7270816564559937, + "num_tokens": 353749116.0, + "step": 13777 + }, + { + "epoch": 1.5130683066110255, + "grad_norm": 2.395272970199585, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7117874622344971, + "num_tokens": 353773161.0, + "step": 13778 + }, + { + "epoch": 1.5131781243136393, + "grad_norm": 2.480452299118042, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7098565101623535, + "num_tokens": 353795806.0, + "step": 13779 + }, + { + "epoch": 1.513287942016253, + "grad_norm": 2.436730146408081, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7324836850166321, + "num_tokens": 353818347.0, + "step": 13780 + }, + { + "epoch": 1.5133977597188668, + "grad_norm": 2.354626417160034, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.743485689163208, + "num_tokens": 353842155.0, + "step": 13781 + }, + { + "epoch": 1.5135075774214803, + "grad_norm": 2.3452718257904053, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.723447322845459, + "num_tokens": 353865975.0, + "step": 13782 + }, + { + "epoch": 1.5136173951240939, + "grad_norm": 2.1067214012145996, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.731191873550415, + "num_tokens": 353895232.0, + "step": 13783 + }, + { + "epoch": 1.5137272128267076, + "grad_norm": 2.582258939743042, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.727747917175293, + "num_tokens": 353915837.0, + "step": 13784 + }, + { + "epoch": 1.5138370305293214, + "grad_norm": 1.9524635076522827, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.699160099029541, + "num_tokens": 353947067.0, + "step": 13785 + }, + { + "epoch": 1.5139468482319351, + "grad_norm": 2.2959482669830322, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7406131029129028, + "num_tokens": 353972204.0, + "step": 13786 + }, + { + "epoch": 1.5140566659345487, + "grad_norm": 2.4738094806671143, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7308863997459412, + "num_tokens": 353993078.0, + "step": 13787 + }, + { + "epoch": 1.5141664836371622, + "grad_norm": 2.302126884460449, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7111918926239014, + "num_tokens": 354018853.0, + "step": 13788 + }, + { + "epoch": 1.514276301339776, + "grad_norm": 2.244863510131836, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7226302623748779, + "num_tokens": 354045441.0, + "step": 13789 + }, + { + "epoch": 1.5143861190423897, + "grad_norm": 1.9940619468688965, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7280890941619873, + "num_tokens": 354075251.0, + "step": 13790 + }, + { + "epoch": 1.5144959367450033, + "grad_norm": 2.2884297370910645, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7029348611831665, + "num_tokens": 354100019.0, + "step": 13791 + }, + { + "epoch": 1.5146057544476168, + "grad_norm": 2.0880045890808105, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7010740041732788, + "num_tokens": 354127867.0, + "step": 13792 + }, + { + "epoch": 1.5147155721502306, + "grad_norm": 2.2680704593658447, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7111880779266357, + "num_tokens": 354153410.0, + "step": 13793 + }, + { + "epoch": 1.5148253898528443, + "grad_norm": 1.9829601049423218, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7063322067260742, + "num_tokens": 354187531.0, + "step": 13794 + }, + { + "epoch": 1.514935207555458, + "grad_norm": 2.3206379413604736, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7278225421905518, + "num_tokens": 354210903.0, + "step": 13795 + }, + { + "epoch": 1.5150450252580716, + "grad_norm": 2.353890895843506, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7423710823059082, + "num_tokens": 354234692.0, + "step": 13796 + }, + { + "epoch": 1.5151548429606851, + "grad_norm": 2.2012102603912354, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7034132480621338, + "num_tokens": 354264274.0, + "step": 13797 + }, + { + "epoch": 1.515264660663299, + "grad_norm": 2.220069169998169, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7205829620361328, + "num_tokens": 354288729.0, + "step": 13798 + }, + { + "epoch": 1.5153744783659127, + "grad_norm": 2.5742831230163574, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7396361827850342, + "num_tokens": 354307969.0, + "step": 13799 + }, + { + "epoch": 1.5154842960685262, + "grad_norm": 2.2120985984802246, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7221212387084961, + "num_tokens": 354333031.0, + "step": 13800 + }, + { + "epoch": 1.51559411377114, + "grad_norm": 2.2779300212860107, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7185869216918945, + "num_tokens": 354357755.0, + "step": 13801 + }, + { + "epoch": 1.5157039314737535, + "grad_norm": 2.2742505073547363, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7135564088821411, + "num_tokens": 354384890.0, + "step": 13802 + }, + { + "epoch": 1.5158137491763672, + "grad_norm": 2.12302565574646, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7249934077262878, + "num_tokens": 354412226.0, + "step": 13803 + }, + { + "epoch": 1.515923566878981, + "grad_norm": 2.0799379348754883, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7114856243133545, + "num_tokens": 354441152.0, + "step": 13804 + }, + { + "epoch": 1.5160333845815945, + "grad_norm": 1.9567824602127075, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7115269899368286, + "num_tokens": 354472895.0, + "step": 13805 + }, + { + "epoch": 1.516143202284208, + "grad_norm": 2.4415249824523926, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7362674474716187, + "num_tokens": 354495385.0, + "step": 13806 + }, + { + "epoch": 1.5162530199868218, + "grad_norm": 2.277071237564087, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7159374356269836, + "num_tokens": 354521647.0, + "step": 13807 + }, + { + "epoch": 1.5163628376894356, + "grad_norm": 2.327476739883423, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7248991131782532, + "num_tokens": 354544592.0, + "step": 13808 + }, + { + "epoch": 1.5164726553920493, + "grad_norm": 2.362060546875, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7194095849990845, + "num_tokens": 354568585.0, + "step": 13809 + }, + { + "epoch": 1.5165824730946629, + "grad_norm": 2.909250497817993, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7445374131202698, + "num_tokens": 354586668.0, + "step": 13810 + }, + { + "epoch": 1.5166922907972764, + "grad_norm": 2.332387924194336, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7205312252044678, + "num_tokens": 354610040.0, + "step": 13811 + }, + { + "epoch": 1.5168021084998902, + "grad_norm": 2.274491548538208, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7353997230529785, + "num_tokens": 354633961.0, + "step": 13812 + }, + { + "epoch": 1.516911926202504, + "grad_norm": 2.0791707038879395, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7061858177185059, + "num_tokens": 354663020.0, + "step": 13813 + }, + { + "epoch": 1.5170217439051175, + "grad_norm": 2.166886806488037, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7330652475357056, + "num_tokens": 354690340.0, + "step": 13814 + }, + { + "epoch": 1.5171315616077312, + "grad_norm": 2.271324634552002, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7082371711730957, + "num_tokens": 354715519.0, + "step": 13815 + }, + { + "epoch": 1.5172413793103448, + "grad_norm": 2.270603895187378, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7261658906936646, + "num_tokens": 354740507.0, + "step": 13816 + }, + { + "epoch": 1.5173511970129585, + "grad_norm": 2.0987863540649414, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7136585712432861, + "num_tokens": 354772071.0, + "step": 13817 + }, + { + "epoch": 1.5174610147155723, + "grad_norm": 2.2362916469573975, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7137879133224487, + "num_tokens": 354799060.0, + "step": 13818 + }, + { + "epoch": 1.5175708324181858, + "grad_norm": 2.2924346923828125, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7213544249534607, + "num_tokens": 354823265.0, + "step": 13819 + }, + { + "epoch": 1.5176806501207993, + "grad_norm": 2.2937545776367188, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7233052253723145, + "num_tokens": 354847357.0, + "step": 13820 + }, + { + "epoch": 1.517790467823413, + "grad_norm": 2.185875177383423, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7366738319396973, + "num_tokens": 354873700.0, + "step": 13821 + }, + { + "epoch": 1.5179002855260268, + "grad_norm": 1.9847129583358765, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7263392210006714, + "num_tokens": 354904323.0, + "step": 13822 + }, + { + "epoch": 1.5180101032286406, + "grad_norm": 2.09804630279541, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7240073084831238, + "num_tokens": 354933779.0, + "step": 13823 + }, + { + "epoch": 1.5181199209312541, + "grad_norm": 2.57492733001709, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7325962781906128, + "num_tokens": 354954386.0, + "step": 13824 + }, + { + "epoch": 1.5182297386338677, + "grad_norm": 2.702904462814331, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7153252363204956, + "num_tokens": 354973849.0, + "step": 13825 + }, + { + "epoch": 1.5183395563364814, + "grad_norm": 2.2409329414367676, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7225397229194641, + "num_tokens": 355000407.0, + "step": 13826 + }, + { + "epoch": 1.5184493740390952, + "grad_norm": 2.409315347671509, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7477996945381165, + "num_tokens": 355022722.0, + "step": 13827 + }, + { + "epoch": 1.5185591917417087, + "grad_norm": 2.1443018913269043, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.722394585609436, + "num_tokens": 355049330.0, + "step": 13828 + }, + { + "epoch": 1.5186690094443223, + "grad_norm": 2.4800806045532227, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7024831771850586, + "num_tokens": 355071477.0, + "step": 13829 + }, + { + "epoch": 1.518778827146936, + "grad_norm": 2.5181641578674316, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7471789121627808, + "num_tokens": 355091622.0, + "step": 13830 + }, + { + "epoch": 1.5188886448495498, + "grad_norm": 2.2723422050476074, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.736082911491394, + "num_tokens": 355114632.0, + "step": 13831 + }, + { + "epoch": 1.5189984625521635, + "grad_norm": 2.155144691467285, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7409014105796814, + "num_tokens": 355140488.0, + "step": 13832 + }, + { + "epoch": 1.519108280254777, + "grad_norm": 2.6262290477752686, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7220120429992676, + "num_tokens": 355161829.0, + "step": 13833 + }, + { + "epoch": 1.5192180979573906, + "grad_norm": 2.355199098587036, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7142144441604614, + "num_tokens": 355187061.0, + "step": 13834 + }, + { + "epoch": 1.5193279156600044, + "grad_norm": 2.1951284408569336, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7127006649971008, + "num_tokens": 355214243.0, + "step": 13835 + }, + { + "epoch": 1.5194377333626181, + "grad_norm": 2.233091115951538, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7383952140808105, + "num_tokens": 355237757.0, + "step": 13836 + }, + { + "epoch": 1.5195475510652319, + "grad_norm": 2.3023006916046143, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7196044325828552, + "num_tokens": 355260914.0, + "step": 13837 + }, + { + "epoch": 1.5196573687678454, + "grad_norm": 2.353761672973633, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.748794436454773, + "num_tokens": 355282639.0, + "step": 13838 + }, + { + "epoch": 1.519767186470459, + "grad_norm": 2.2909679412841797, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7160400152206421, + "num_tokens": 355308983.0, + "step": 13839 + }, + { + "epoch": 1.5198770041730727, + "grad_norm": 2.310025930404663, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7169503569602966, + "num_tokens": 355331640.0, + "step": 13840 + }, + { + "epoch": 1.5199868218756865, + "grad_norm": 2.0453367233276367, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.722438395023346, + "num_tokens": 355361401.0, + "step": 13841 + }, + { + "epoch": 1.5200966395783, + "grad_norm": 2.159437894821167, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6905187368392944, + "num_tokens": 355388955.0, + "step": 13842 + }, + { + "epoch": 1.5202064572809135, + "grad_norm": 2.1606740951538086, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7049334645271301, + "num_tokens": 355416776.0, + "step": 13843 + }, + { + "epoch": 1.5203162749835273, + "grad_norm": 2.143521308898926, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7110142707824707, + "num_tokens": 355444822.0, + "step": 13844 + }, + { + "epoch": 1.520426092686141, + "grad_norm": 2.3740339279174805, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7437294125556946, + "num_tokens": 355466986.0, + "step": 13845 + }, + { + "epoch": 1.5205359103887548, + "grad_norm": 2.295804977416992, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7074609994888306, + "num_tokens": 355494633.0, + "step": 13846 + }, + { + "epoch": 1.5206457280913683, + "grad_norm": 2.2124881744384766, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7165292501449585, + "num_tokens": 355520515.0, + "step": 13847 + }, + { + "epoch": 1.5207555457939819, + "grad_norm": 2.079153537750244, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.714191198348999, + "num_tokens": 355550134.0, + "step": 13848 + }, + { + "epoch": 1.5208653634965956, + "grad_norm": 2.3018500804901123, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7021975517272949, + "num_tokens": 355576725.0, + "step": 13849 + }, + { + "epoch": 1.5209751811992094, + "grad_norm": 2.352935791015625, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7258283495903015, + "num_tokens": 355599196.0, + "step": 13850 + }, + { + "epoch": 1.5210849989018231, + "grad_norm": 2.0044872760772705, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.694566011428833, + "num_tokens": 355632250.0, + "step": 13851 + }, + { + "epoch": 1.5211948166044367, + "grad_norm": 2.3896989822387695, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.71775221824646, + "num_tokens": 355656383.0, + "step": 13852 + }, + { + "epoch": 1.5213046343070502, + "grad_norm": 2.3150618076324463, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.727692723274231, + "num_tokens": 355679622.0, + "step": 13853 + }, + { + "epoch": 1.521414452009664, + "grad_norm": 2.456923723220825, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6949315071105957, + "num_tokens": 355702587.0, + "step": 13854 + }, + { + "epoch": 1.5215242697122777, + "grad_norm": 2.0476973056793213, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7244553565979004, + "num_tokens": 355733397.0, + "step": 13855 + }, + { + "epoch": 1.5216340874148913, + "grad_norm": 2.7542524337768555, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7235937714576721, + "num_tokens": 355751576.0, + "step": 13856 + }, + { + "epoch": 1.5217439051175048, + "grad_norm": 2.6272621154785156, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7323569059371948, + "num_tokens": 355771416.0, + "step": 13857 + }, + { + "epoch": 1.5218537228201185, + "grad_norm": 2.385002374649048, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7231137752532959, + "num_tokens": 355794415.0, + "step": 13858 + }, + { + "epoch": 1.5219635405227323, + "grad_norm": 2.254220724105835, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7149162888526917, + "num_tokens": 355819927.0, + "step": 13859 + }, + { + "epoch": 1.522073358225346, + "grad_norm": 2.2018496990203857, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7063840627670288, + "num_tokens": 355846373.0, + "step": 13860 + }, + { + "epoch": 1.5221831759279596, + "grad_norm": 2.353976011276245, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7047966718673706, + "num_tokens": 355871342.0, + "step": 13861 + }, + { + "epoch": 1.5222929936305731, + "grad_norm": 2.3179619312286377, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7499304413795471, + "num_tokens": 355892853.0, + "step": 13862 + }, + { + "epoch": 1.522402811333187, + "grad_norm": 2.457845687866211, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7116748094558716, + "num_tokens": 355914458.0, + "step": 13863 + }, + { + "epoch": 1.5225126290358006, + "grad_norm": 1.9693272113800049, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7126127481460571, + "num_tokens": 355945304.0, + "step": 13864 + }, + { + "epoch": 1.5226224467384142, + "grad_norm": 1.9894133806228638, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.701417088508606, + "num_tokens": 355975804.0, + "step": 13865 + }, + { + "epoch": 1.522732264441028, + "grad_norm": 2.2787516117095947, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7112253904342651, + "num_tokens": 356000132.0, + "step": 13866 + }, + { + "epoch": 1.5228420821436415, + "grad_norm": 2.148029327392578, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6981021761894226, + "num_tokens": 356027803.0, + "step": 13867 + }, + { + "epoch": 1.5229518998462552, + "grad_norm": 2.143009662628174, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.711330771446228, + "num_tokens": 356054020.0, + "step": 13868 + }, + { + "epoch": 1.523061717548869, + "grad_norm": 2.325667381286621, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7366798520088196, + "num_tokens": 356076226.0, + "step": 13869 + }, + { + "epoch": 1.5231715352514825, + "grad_norm": 2.5767476558685303, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7353289127349854, + "num_tokens": 356097332.0, + "step": 13870 + }, + { + "epoch": 1.523281352954096, + "grad_norm": 2.5508177280426025, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7305055856704712, + "num_tokens": 356118824.0, + "step": 13871 + }, + { + "epoch": 1.5233911706567098, + "grad_norm": 2.4118504524230957, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7309994101524353, + "num_tokens": 356140106.0, + "step": 13872 + }, + { + "epoch": 1.5235009883593236, + "grad_norm": 2.283722162246704, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7052830457687378, + "num_tokens": 356167008.0, + "step": 13873 + }, + { + "epoch": 1.5236108060619373, + "grad_norm": 2.3730132579803467, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7362455725669861, + "num_tokens": 356189197.0, + "step": 13874 + }, + { + "epoch": 1.5237206237645509, + "grad_norm": 2.3154282569885254, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7112110257148743, + "num_tokens": 356214917.0, + "step": 13875 + }, + { + "epoch": 1.5238304414671644, + "grad_norm": 2.3177125453948975, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7027137875556946, + "num_tokens": 356244772.0, + "step": 13876 + }, + { + "epoch": 1.5239402591697782, + "grad_norm": 2.5964245796203613, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7252388000488281, + "num_tokens": 356264443.0, + "step": 13877 + }, + { + "epoch": 1.524050076872392, + "grad_norm": 2.2662720680236816, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7102152705192566, + "num_tokens": 356289235.0, + "step": 13878 + }, + { + "epoch": 1.5241598945750054, + "grad_norm": 2.084247350692749, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.698681116104126, + "num_tokens": 356320086.0, + "step": 13879 + }, + { + "epoch": 1.5242697122776192, + "grad_norm": 2.297118663787842, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7033594250679016, + "num_tokens": 356343153.0, + "step": 13880 + }, + { + "epoch": 1.5243795299802327, + "grad_norm": 2.276332139968872, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7575002908706665, + "num_tokens": 356364539.0, + "step": 13881 + }, + { + "epoch": 1.5244893476828465, + "grad_norm": 2.343015432357788, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7160547971725464, + "num_tokens": 356388337.0, + "step": 13882 + }, + { + "epoch": 1.5245991653854603, + "grad_norm": 2.4401636123657227, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7061029076576233, + "num_tokens": 356410427.0, + "step": 13883 + }, + { + "epoch": 1.5247089830880738, + "grad_norm": 2.374668836593628, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.727493166923523, + "num_tokens": 356432378.0, + "step": 13884 + }, + { + "epoch": 1.5248188007906873, + "grad_norm": 2.2927122116088867, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.727483868598938, + "num_tokens": 356457031.0, + "step": 13885 + }, + { + "epoch": 1.524928618493301, + "grad_norm": 2.2849724292755127, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7495237588882446, + "num_tokens": 356480190.0, + "step": 13886 + }, + { + "epoch": 1.5250384361959148, + "grad_norm": 2.419830083847046, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7232603430747986, + "num_tokens": 356501223.0, + "step": 13887 + }, + { + "epoch": 1.5251482538985286, + "grad_norm": 2.3063838481903076, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.712497889995575, + "num_tokens": 356525284.0, + "step": 13888 + }, + { + "epoch": 1.5252580716011421, + "grad_norm": 2.2516348361968994, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7231742143630981, + "num_tokens": 356550196.0, + "step": 13889 + }, + { + "epoch": 1.5253678893037557, + "grad_norm": 2.24393892288208, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7142150402069092, + "num_tokens": 356576558.0, + "step": 13890 + }, + { + "epoch": 1.5254777070063694, + "grad_norm": 2.315711736679077, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.700635552406311, + "num_tokens": 356601227.0, + "step": 13891 + }, + { + "epoch": 1.5255875247089832, + "grad_norm": 2.2656989097595215, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7132551074028015, + "num_tokens": 356627020.0, + "step": 13892 + }, + { + "epoch": 1.5256973424115967, + "grad_norm": 2.3720479011535645, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7146601676940918, + "num_tokens": 356650881.0, + "step": 13893 + }, + { + "epoch": 1.5258071601142102, + "grad_norm": 2.227385997772217, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7000316381454468, + "num_tokens": 356676734.0, + "step": 13894 + }, + { + "epoch": 1.525916977816824, + "grad_norm": 2.159183979034424, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7120928764343262, + "num_tokens": 356704017.0, + "step": 13895 + }, + { + "epoch": 1.5260267955194378, + "grad_norm": 2.0766658782958984, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7075295448303223, + "num_tokens": 356732401.0, + "step": 13896 + }, + { + "epoch": 1.5261366132220515, + "grad_norm": 2.450244665145874, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7320828437805176, + "num_tokens": 356753748.0, + "step": 13897 + }, + { + "epoch": 1.526246430924665, + "grad_norm": 2.416290521621704, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7294036149978638, + "num_tokens": 356775941.0, + "step": 13898 + }, + { + "epoch": 1.5263562486272786, + "grad_norm": 2.2683916091918945, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7193444967269897, + "num_tokens": 356800107.0, + "step": 13899 + }, + { + "epoch": 1.5264660663298923, + "grad_norm": 2.514472723007202, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7332074046134949, + "num_tokens": 356820709.0, + "step": 13900 + }, + { + "epoch": 1.526575884032506, + "grad_norm": 2.3746824264526367, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7121899127960205, + "num_tokens": 356845543.0, + "step": 13901 + }, + { + "epoch": 1.5266857017351199, + "grad_norm": 2.3483128547668457, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7456957101821899, + "num_tokens": 356868041.0, + "step": 13902 + }, + { + "epoch": 1.5267955194377334, + "grad_norm": 2.297457456588745, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7133603692054749, + "num_tokens": 356892714.0, + "step": 13903 + }, + { + "epoch": 1.526905337140347, + "grad_norm": 2.1829423904418945, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7144386768341064, + "num_tokens": 356918791.0, + "step": 13904 + }, + { + "epoch": 1.5270151548429607, + "grad_norm": 2.557162046432495, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7163800597190857, + "num_tokens": 356939634.0, + "step": 13905 + }, + { + "epoch": 1.5271249725455744, + "grad_norm": 2.398531675338745, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7173540592193604, + "num_tokens": 356962241.0, + "step": 13906 + }, + { + "epoch": 1.527234790248188, + "grad_norm": 2.0447945594787598, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7301733493804932, + "num_tokens": 356992254.0, + "step": 13907 + }, + { + "epoch": 1.5273446079508015, + "grad_norm": 2.4902851581573486, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7286797761917114, + "num_tokens": 357014561.0, + "step": 13908 + }, + { + "epoch": 1.5274544256534153, + "grad_norm": 2.3359365463256836, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7428129315376282, + "num_tokens": 357038219.0, + "step": 13909 + }, + { + "epoch": 1.527564243356029, + "grad_norm": 2.7448644638061523, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7369865775108337, + "num_tokens": 357057113.0, + "step": 13910 + }, + { + "epoch": 1.5276740610586428, + "grad_norm": 2.4032201766967773, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7357398867607117, + "num_tokens": 357079291.0, + "step": 13911 + }, + { + "epoch": 1.5277838787612563, + "grad_norm": 2.1281352043151855, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.724687933921814, + "num_tokens": 357106256.0, + "step": 13912 + }, + { + "epoch": 1.5278936964638699, + "grad_norm": 2.2080583572387695, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7131670713424683, + "num_tokens": 357132810.0, + "step": 13913 + }, + { + "epoch": 1.5280035141664836, + "grad_norm": 2.455061912536621, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7220269441604614, + "num_tokens": 357156549.0, + "step": 13914 + }, + { + "epoch": 1.5281133318690974, + "grad_norm": 2.224196434020996, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.724231481552124, + "num_tokens": 357182789.0, + "step": 13915 + }, + { + "epoch": 1.5282231495717111, + "grad_norm": 2.3177618980407715, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7088754773139954, + "num_tokens": 357208663.0, + "step": 13916 + }, + { + "epoch": 1.5283329672743247, + "grad_norm": 2.062662363052368, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7124906778335571, + "num_tokens": 357238598.0, + "step": 13917 + }, + { + "epoch": 1.5284427849769382, + "grad_norm": 2.3613057136535645, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7245052456855774, + "num_tokens": 357262704.0, + "step": 13918 + }, + { + "epoch": 1.528552602679552, + "grad_norm": 2.4697751998901367, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7399941682815552, + "num_tokens": 357284885.0, + "step": 13919 + }, + { + "epoch": 1.5286624203821657, + "grad_norm": 2.410921812057495, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7260282039642334, + "num_tokens": 357307591.0, + "step": 13920 + }, + { + "epoch": 1.5287722380847792, + "grad_norm": 1.8186827898025513, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7046719789505005, + "num_tokens": 357345389.0, + "step": 13921 + }, + { + "epoch": 1.5288820557873928, + "grad_norm": 2.456036329269409, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7298318147659302, + "num_tokens": 357366744.0, + "step": 13922 + }, + { + "epoch": 1.5289918734900065, + "grad_norm": 2.387007236480713, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7543095350265503, + "num_tokens": 357390115.0, + "step": 13923 + }, + { + "epoch": 1.5291016911926203, + "grad_norm": 2.1960816383361816, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7019332051277161, + "num_tokens": 357418970.0, + "step": 13924 + }, + { + "epoch": 1.529211508895234, + "grad_norm": 2.51798939704895, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7279011011123657, + "num_tokens": 357439363.0, + "step": 13925 + }, + { + "epoch": 1.5293213265978476, + "grad_norm": 2.1558592319488525, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7495154142379761, + "num_tokens": 357463699.0, + "step": 13926 + }, + { + "epoch": 1.5294311443004611, + "grad_norm": 2.2553675174713135, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7197320461273193, + "num_tokens": 357488624.0, + "step": 13927 + }, + { + "epoch": 1.5295409620030749, + "grad_norm": 2.077636957168579, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7149085998535156, + "num_tokens": 357517287.0, + "step": 13928 + }, + { + "epoch": 1.5296507797056886, + "grad_norm": 2.404755115509033, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7341710329055786, + "num_tokens": 357540954.0, + "step": 13929 + }, + { + "epoch": 1.5297605974083022, + "grad_norm": 2.1304702758789062, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7118485569953918, + "num_tokens": 357569422.0, + "step": 13930 + }, + { + "epoch": 1.529870415110916, + "grad_norm": 2.1536176204681396, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7350434064865112, + "num_tokens": 357596391.0, + "step": 13931 + }, + { + "epoch": 1.5299802328135295, + "grad_norm": 2.3335609436035156, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7420661449432373, + "num_tokens": 357618024.0, + "step": 13932 + }, + { + "epoch": 1.5300900505161432, + "grad_norm": 2.454458475112915, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7333493828773499, + "num_tokens": 357638396.0, + "step": 13933 + }, + { + "epoch": 1.530199868218757, + "grad_norm": 2.51515793800354, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7363753318786621, + "num_tokens": 357660943.0, + "step": 13934 + }, + { + "epoch": 1.5303096859213705, + "grad_norm": 2.650089979171753, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.729353666305542, + "num_tokens": 357682667.0, + "step": 13935 + }, + { + "epoch": 1.530419503623984, + "grad_norm": 2.2033464908599854, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7111663818359375, + "num_tokens": 357711511.0, + "step": 13936 + }, + { + "epoch": 1.5305293213265978, + "grad_norm": 2.2205452919006348, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7099010944366455, + "num_tokens": 357736870.0, + "step": 13937 + }, + { + "epoch": 1.5306391390292116, + "grad_norm": 2.198979377746582, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.738676905632019, + "num_tokens": 357762638.0, + "step": 13938 + }, + { + "epoch": 1.5307489567318253, + "grad_norm": 2.2556166648864746, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7203086614608765, + "num_tokens": 357787441.0, + "step": 13939 + }, + { + "epoch": 1.5308587744344389, + "grad_norm": 1.9474573135375977, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7147790193557739, + "num_tokens": 357819334.0, + "step": 13940 + }, + { + "epoch": 1.5309685921370524, + "grad_norm": 2.023855209350586, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6757087707519531, + "num_tokens": 357855211.0, + "step": 13941 + }, + { + "epoch": 1.5310784098396661, + "grad_norm": 2.380195379257202, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7104605436325073, + "num_tokens": 357878163.0, + "step": 13942 + }, + { + "epoch": 1.53118822754228, + "grad_norm": 2.1708624362945557, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.730588972568512, + "num_tokens": 357904485.0, + "step": 13943 + }, + { + "epoch": 1.5312980452448934, + "grad_norm": 2.089273691177368, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7048323154449463, + "num_tokens": 357933989.0, + "step": 13944 + }, + { + "epoch": 1.5314078629475072, + "grad_norm": 2.1596052646636963, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.714478075504303, + "num_tokens": 357962691.0, + "step": 13945 + }, + { + "epoch": 1.5315176806501207, + "grad_norm": 2.410456418991089, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7143731713294983, + "num_tokens": 357984511.0, + "step": 13946 + }, + { + "epoch": 1.5316274983527345, + "grad_norm": 2.053849935531616, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7293466329574585, + "num_tokens": 358014559.0, + "step": 13947 + }, + { + "epoch": 1.5317373160553482, + "grad_norm": 2.0612592697143555, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7239384055137634, + "num_tokens": 358047940.0, + "step": 13948 + }, + { + "epoch": 1.5318471337579618, + "grad_norm": 2.1988446712493896, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7012391686439514, + "num_tokens": 358074851.0, + "step": 13949 + }, + { + "epoch": 1.5319569514605753, + "grad_norm": 2.2965118885040283, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7177980542182922, + "num_tokens": 358099861.0, + "step": 13950 + }, + { + "epoch": 1.532066769163189, + "grad_norm": 2.3488059043884277, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7189517021179199, + "num_tokens": 358123440.0, + "step": 13951 + }, + { + "epoch": 1.5321765868658028, + "grad_norm": 1.8946938514709473, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6772823333740234, + "num_tokens": 358161624.0, + "step": 13952 + }, + { + "epoch": 1.5322864045684166, + "grad_norm": 2.3069005012512207, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7118217945098877, + "num_tokens": 358188641.0, + "step": 13953 + }, + { + "epoch": 1.5323962222710301, + "grad_norm": 2.2525060176849365, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7396340370178223, + "num_tokens": 358213607.0, + "step": 13954 + }, + { + "epoch": 1.5325060399736437, + "grad_norm": 2.516387939453125, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7105667591094971, + "num_tokens": 358237250.0, + "step": 13955 + }, + { + "epoch": 1.5326158576762574, + "grad_norm": 2.229316473007202, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7163174152374268, + "num_tokens": 358262971.0, + "step": 13956 + }, + { + "epoch": 1.5327256753788712, + "grad_norm": 2.160736560821533, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7135210037231445, + "num_tokens": 358290876.0, + "step": 13957 + }, + { + "epoch": 1.5328354930814847, + "grad_norm": 2.0601181983947754, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7107775211334229, + "num_tokens": 358320681.0, + "step": 13958 + }, + { + "epoch": 1.5329453107840982, + "grad_norm": 2.341026544570923, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7247677445411682, + "num_tokens": 358344227.0, + "step": 13959 + }, + { + "epoch": 1.533055128486712, + "grad_norm": 2.117321729660034, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7264673113822937, + "num_tokens": 358372096.0, + "step": 13960 + }, + { + "epoch": 1.5331649461893258, + "grad_norm": 2.163764715194702, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7254196405410767, + "num_tokens": 358399716.0, + "step": 13961 + }, + { + "epoch": 1.5332747638919395, + "grad_norm": 2.1056275367736816, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6977571845054626, + "num_tokens": 358427525.0, + "step": 13962 + }, + { + "epoch": 1.533384581594553, + "grad_norm": 2.4441049098968506, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7323876619338989, + "num_tokens": 358450055.0, + "step": 13963 + }, + { + "epoch": 1.5334943992971666, + "grad_norm": 2.1339972019195557, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6971178650856018, + "num_tokens": 358481620.0, + "step": 13964 + }, + { + "epoch": 1.5336042169997803, + "grad_norm": 2.575279951095581, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7118021249771118, + "num_tokens": 358501383.0, + "step": 13965 + }, + { + "epoch": 1.533714034702394, + "grad_norm": 2.4391863346099854, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7166697978973389, + "num_tokens": 358525079.0, + "step": 13966 + }, + { + "epoch": 1.5338238524050078, + "grad_norm": 2.36269211769104, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6911225318908691, + "num_tokens": 358550888.0, + "step": 13967 + }, + { + "epoch": 1.5339336701076214, + "grad_norm": 2.4119956493377686, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7168178558349609, + "num_tokens": 358574128.0, + "step": 13968 + }, + { + "epoch": 1.534043487810235, + "grad_norm": 2.3614118099212646, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7311108112335205, + "num_tokens": 358595330.0, + "step": 13969 + }, + { + "epoch": 1.5341533055128487, + "grad_norm": 2.0815236568450928, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7021335363388062, + "num_tokens": 358625220.0, + "step": 13970 + }, + { + "epoch": 1.5342631232154624, + "grad_norm": 2.796759605407715, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7256798148155212, + "num_tokens": 358644102.0, + "step": 13971 + }, + { + "epoch": 1.534372940918076, + "grad_norm": 2.0291168689727783, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6966952681541443, + "num_tokens": 358676199.0, + "step": 13972 + }, + { + "epoch": 1.5344827586206895, + "grad_norm": 2.582282304763794, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7519651651382446, + "num_tokens": 358697232.0, + "step": 13973 + }, + { + "epoch": 1.5345925763233033, + "grad_norm": 2.0023937225341797, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7022034525871277, + "num_tokens": 358727744.0, + "step": 13974 + }, + { + "epoch": 1.534702394025917, + "grad_norm": 2.5536205768585205, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7319390177726746, + "num_tokens": 358747858.0, + "step": 13975 + }, + { + "epoch": 1.5348122117285308, + "grad_norm": 2.2431604862213135, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7218132615089417, + "num_tokens": 358772312.0, + "step": 13976 + }, + { + "epoch": 1.5349220294311443, + "grad_norm": 1.9832907915115356, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6920203566551208, + "num_tokens": 358803974.0, + "step": 13977 + }, + { + "epoch": 1.5350318471337578, + "grad_norm": 2.3093111515045166, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7298280596733093, + "num_tokens": 358827598.0, + "step": 13978 + }, + { + "epoch": 1.5351416648363716, + "grad_norm": 2.361156463623047, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7120402455329895, + "num_tokens": 358853448.0, + "step": 13979 + }, + { + "epoch": 1.5352514825389854, + "grad_norm": 2.4375035762786865, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7215203046798706, + "num_tokens": 358876708.0, + "step": 13980 + }, + { + "epoch": 1.535361300241599, + "grad_norm": 2.7832136154174805, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7283885478973389, + "num_tokens": 358896632.0, + "step": 13981 + }, + { + "epoch": 1.5354711179442126, + "grad_norm": 2.2587358951568604, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7208244204521179, + "num_tokens": 358921869.0, + "step": 13982 + }, + { + "epoch": 1.5355809356468262, + "grad_norm": 2.312703847885132, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7094219923019409, + "num_tokens": 358948689.0, + "step": 13983 + }, + { + "epoch": 1.53569075334944, + "grad_norm": 2.1208207607269287, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.708373486995697, + "num_tokens": 358977296.0, + "step": 13984 + }, + { + "epoch": 1.5358005710520537, + "grad_norm": 2.333730697631836, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7233262062072754, + "num_tokens": 359000820.0, + "step": 13985 + }, + { + "epoch": 1.5359103887546672, + "grad_norm": 2.3724687099456787, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7036386132240295, + "num_tokens": 359025691.0, + "step": 13986 + }, + { + "epoch": 1.5360202064572808, + "grad_norm": 2.1966848373413086, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7110989689826965, + "num_tokens": 359050643.0, + "step": 13987 + }, + { + "epoch": 1.5361300241598945, + "grad_norm": 2.4188296794891357, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.730074405670166, + "num_tokens": 359072885.0, + "step": 13988 + }, + { + "epoch": 1.5362398418625083, + "grad_norm": 2.0200233459472656, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7298306226730347, + "num_tokens": 359101843.0, + "step": 13989 + }, + { + "epoch": 1.536349659565122, + "grad_norm": 2.2041842937469482, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7059643268585205, + "num_tokens": 359128772.0, + "step": 13990 + }, + { + "epoch": 1.5364594772677356, + "grad_norm": 2.1690104007720947, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7151821851730347, + "num_tokens": 359156613.0, + "step": 13991 + }, + { + "epoch": 1.536569294970349, + "grad_norm": 2.2507572174072266, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6971138715744019, + "num_tokens": 359182272.0, + "step": 13992 + }, + { + "epoch": 1.5366791126729629, + "grad_norm": 2.3887555599212646, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7137389183044434, + "num_tokens": 359206004.0, + "step": 13993 + }, + { + "epoch": 1.5367889303755766, + "grad_norm": 2.4655256271362305, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7301323413848877, + "num_tokens": 359227879.0, + "step": 13994 + }, + { + "epoch": 1.5368987480781902, + "grad_norm": 2.254312515258789, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6990692615509033, + "num_tokens": 359253938.0, + "step": 13995 + }, + { + "epoch": 1.537008565780804, + "grad_norm": 2.2347476482391357, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7235189080238342, + "num_tokens": 359279378.0, + "step": 13996 + }, + { + "epoch": 1.5371183834834174, + "grad_norm": 2.3416078090667725, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7532142400741577, + "num_tokens": 359301183.0, + "step": 13997 + }, + { + "epoch": 1.5372282011860312, + "grad_norm": 1.920760989189148, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.742996096611023, + "num_tokens": 359333507.0, + "step": 13998 + }, + { + "epoch": 1.537338018888645, + "grad_norm": 2.251011848449707, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7254006862640381, + "num_tokens": 359358638.0, + "step": 13999 + }, + { + "epoch": 1.5374478365912585, + "grad_norm": 2.242366075515747, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7013239860534668, + "num_tokens": 359382697.0, + "step": 14000 + }, + { + "epoch": 1.537557654293872, + "grad_norm": 2.6275880336761475, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7194136381149292, + "num_tokens": 359402367.0, + "step": 14001 + }, + { + "epoch": 1.5376674719964858, + "grad_norm": 1.97516667842865, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7162185907363892, + "num_tokens": 359432873.0, + "step": 14002 + }, + { + "epoch": 1.5377772896990995, + "grad_norm": 2.241635322570801, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7199554443359375, + "num_tokens": 359458806.0, + "step": 14003 + }, + { + "epoch": 1.5378871074017133, + "grad_norm": 2.434915542602539, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7136138677597046, + "num_tokens": 359481571.0, + "step": 14004 + }, + { + "epoch": 1.5379969251043268, + "grad_norm": 2.2713208198547363, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.725283145904541, + "num_tokens": 359507017.0, + "step": 14005 + }, + { + "epoch": 1.5381067428069404, + "grad_norm": 2.211935043334961, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.70246422290802, + "num_tokens": 359535002.0, + "step": 14006 + }, + { + "epoch": 1.5382165605095541, + "grad_norm": 2.1161818504333496, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7473090291023254, + "num_tokens": 359560516.0, + "step": 14007 + }, + { + "epoch": 1.5383263782121679, + "grad_norm": 2.2760961055755615, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7231534123420715, + "num_tokens": 359585237.0, + "step": 14008 + }, + { + "epoch": 1.5384361959147814, + "grad_norm": 2.119215726852417, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7386395335197449, + "num_tokens": 359613495.0, + "step": 14009 + }, + { + "epoch": 1.538546013617395, + "grad_norm": 1.9494476318359375, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6882944107055664, + "num_tokens": 359646447.0, + "step": 14010 + }, + { + "epoch": 1.5386558313200087, + "grad_norm": 2.2159476280212402, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7308453917503357, + "num_tokens": 359671034.0, + "step": 14011 + }, + { + "epoch": 1.5387656490226225, + "grad_norm": 2.414048194885254, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7365761995315552, + "num_tokens": 359692985.0, + "step": 14012 + }, + { + "epoch": 1.5388754667252362, + "grad_norm": 2.3475232124328613, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7053780555725098, + "num_tokens": 359717155.0, + "step": 14013 + }, + { + "epoch": 1.5389852844278498, + "grad_norm": 2.0703177452087402, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7125988006591797, + "num_tokens": 359746834.0, + "step": 14014 + }, + { + "epoch": 1.5390951021304633, + "grad_norm": 2.4963126182556152, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7532020807266235, + "num_tokens": 359769370.0, + "step": 14015 + }, + { + "epoch": 1.539204919833077, + "grad_norm": 2.5300471782684326, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7464977502822876, + "num_tokens": 359791244.0, + "step": 14016 + }, + { + "epoch": 1.5393147375356908, + "grad_norm": 2.3697128295898438, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7247142195701599, + "num_tokens": 359814542.0, + "step": 14017 + }, + { + "epoch": 1.5394245552383046, + "grad_norm": 2.1649036407470703, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7202462553977966, + "num_tokens": 359841311.0, + "step": 14018 + }, + { + "epoch": 1.539534372940918, + "grad_norm": 2.594527244567871, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7417599558830261, + "num_tokens": 359862273.0, + "step": 14019 + }, + { + "epoch": 1.5396441906435316, + "grad_norm": 2.2004129886627197, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7191051244735718, + "num_tokens": 359889328.0, + "step": 14020 + }, + { + "epoch": 1.5397540083461454, + "grad_norm": 2.2742247581481934, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7010303735733032, + "num_tokens": 359915204.0, + "step": 14021 + }, + { + "epoch": 1.5398638260487592, + "grad_norm": 2.1346447467803955, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7135690450668335, + "num_tokens": 359941427.0, + "step": 14022 + }, + { + "epoch": 1.5399736437513727, + "grad_norm": 2.123114824295044, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7217748761177063, + "num_tokens": 359970055.0, + "step": 14023 + }, + { + "epoch": 1.5400834614539862, + "grad_norm": 2.7112209796905518, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7434794902801514, + "num_tokens": 359988582.0, + "step": 14024 + }, + { + "epoch": 1.5401932791566, + "grad_norm": 2.368412733078003, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7307144403457642, + "num_tokens": 360010651.0, + "step": 14025 + }, + { + "epoch": 1.5403030968592137, + "grad_norm": 1.8228996992111206, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6847168207168579, + "num_tokens": 360046192.0, + "step": 14026 + }, + { + "epoch": 1.5404129145618275, + "grad_norm": 2.051691770553589, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.711983859539032, + "num_tokens": 360076597.0, + "step": 14027 + }, + { + "epoch": 1.540522732264441, + "grad_norm": 2.424978733062744, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.723127543926239, + "num_tokens": 360097999.0, + "step": 14028 + }, + { + "epoch": 1.5406325499670546, + "grad_norm": 1.8905963897705078, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7170533537864685, + "num_tokens": 360131985.0, + "step": 14029 + }, + { + "epoch": 1.5407423676696683, + "grad_norm": 2.3063886165618896, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.6992661952972412, + "num_tokens": 360155721.0, + "step": 14030 + }, + { + "epoch": 1.540852185372282, + "grad_norm": 2.027647018432617, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7081223726272583, + "num_tokens": 360185781.0, + "step": 14031 + }, + { + "epoch": 1.5409620030748958, + "grad_norm": 2.2334063053131104, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.733894407749176, + "num_tokens": 360210390.0, + "step": 14032 + }, + { + "epoch": 1.5410718207775094, + "grad_norm": 2.3884315490722656, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7303158044815063, + "num_tokens": 360232939.0, + "step": 14033 + }, + { + "epoch": 1.541181638480123, + "grad_norm": 2.404539108276367, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7150824069976807, + "num_tokens": 360256424.0, + "step": 14034 + }, + { + "epoch": 1.5412914561827367, + "grad_norm": 2.3707473278045654, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7064526081085205, + "num_tokens": 360281678.0, + "step": 14035 + }, + { + "epoch": 1.5414012738853504, + "grad_norm": 2.3141446113586426, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7263380289077759, + "num_tokens": 360305573.0, + "step": 14036 + }, + { + "epoch": 1.541511091587964, + "grad_norm": 1.9746373891830444, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.703854501247406, + "num_tokens": 360338249.0, + "step": 14037 + }, + { + "epoch": 1.5416209092905775, + "grad_norm": 2.549525022506714, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7421300411224365, + "num_tokens": 360358206.0, + "step": 14038 + }, + { + "epoch": 1.5417307269931912, + "grad_norm": 2.4945385456085205, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.707047164440155, + "num_tokens": 360379701.0, + "step": 14039 + }, + { + "epoch": 1.541840544695805, + "grad_norm": 2.714219093322754, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7207008600234985, + "num_tokens": 360399086.0, + "step": 14040 + }, + { + "epoch": 1.5419503623984188, + "grad_norm": 2.0694668292999268, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7273823618888855, + "num_tokens": 360428678.0, + "step": 14041 + }, + { + "epoch": 1.5420601801010323, + "grad_norm": 2.6044654846191406, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7141937017440796, + "num_tokens": 360450798.0, + "step": 14042 + }, + { + "epoch": 1.5421699978036458, + "grad_norm": 2.327014684677124, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7321184873580933, + "num_tokens": 360473803.0, + "step": 14043 + }, + { + "epoch": 1.5422798155062596, + "grad_norm": 2.0595715045928955, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.704535722732544, + "num_tokens": 360503477.0, + "step": 14044 + }, + { + "epoch": 1.5423896332088733, + "grad_norm": 2.5102288722991943, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7066649794578552, + "num_tokens": 360525092.0, + "step": 14045 + }, + { + "epoch": 1.5424994509114869, + "grad_norm": 2.3302738666534424, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7004222273826599, + "num_tokens": 360550533.0, + "step": 14046 + }, + { + "epoch": 1.5426092686141006, + "grad_norm": 2.107219934463501, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7141411304473877, + "num_tokens": 360578240.0, + "step": 14047 + }, + { + "epoch": 1.5427190863167142, + "grad_norm": 2.2489609718322754, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7164832353591919, + "num_tokens": 360604193.0, + "step": 14048 + }, + { + "epoch": 1.542828904019328, + "grad_norm": 2.27628493309021, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7059341669082642, + "num_tokens": 360630777.0, + "step": 14049 + }, + { + "epoch": 1.5429387217219417, + "grad_norm": 2.25286865234375, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7073808908462524, + "num_tokens": 360658016.0, + "step": 14050 + }, + { + "epoch": 1.5430485394245552, + "grad_norm": 2.036611557006836, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.684928297996521, + "num_tokens": 360691237.0, + "step": 14051 + }, + { + "epoch": 1.5431583571271688, + "grad_norm": 2.2820067405700684, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7019752860069275, + "num_tokens": 360717435.0, + "step": 14052 + }, + { + "epoch": 1.5432681748297825, + "grad_norm": 2.243786096572876, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6948885917663574, + "num_tokens": 360745716.0, + "step": 14053 + }, + { + "epoch": 1.5433779925323963, + "grad_norm": 2.233762264251709, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7046791911125183, + "num_tokens": 360771320.0, + "step": 14054 + }, + { + "epoch": 1.54348781023501, + "grad_norm": 2.4433701038360596, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7421799898147583, + "num_tokens": 360791982.0, + "step": 14055 + }, + { + "epoch": 1.5435976279376236, + "grad_norm": 2.1756415367126465, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7047085762023926, + "num_tokens": 360818735.0, + "step": 14056 + }, + { + "epoch": 1.543707445640237, + "grad_norm": 2.2258975505828857, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7151827812194824, + "num_tokens": 360847367.0, + "step": 14057 + }, + { + "epoch": 1.5438172633428509, + "grad_norm": 2.1204373836517334, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7023128271102905, + "num_tokens": 360881281.0, + "step": 14058 + }, + { + "epoch": 1.5439270810454646, + "grad_norm": 2.4095396995544434, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6946665048599243, + "num_tokens": 360906074.0, + "step": 14059 + }, + { + "epoch": 1.5440368987480781, + "grad_norm": 2.212247371673584, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7088875770568848, + "num_tokens": 360932058.0, + "step": 14060 + }, + { + "epoch": 1.544146716450692, + "grad_norm": 2.1531660556793213, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7069422602653503, + "num_tokens": 360959057.0, + "step": 14061 + }, + { + "epoch": 1.5442565341533054, + "grad_norm": 2.2582011222839355, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7721140384674072, + "num_tokens": 360981482.0, + "step": 14062 + }, + { + "epoch": 1.5443663518559192, + "grad_norm": 2.3297464847564697, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7425479888916016, + "num_tokens": 361005165.0, + "step": 14063 + }, + { + "epoch": 1.544476169558533, + "grad_norm": 2.2222342491149902, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7190062999725342, + "num_tokens": 361032057.0, + "step": 14064 + }, + { + "epoch": 1.5445859872611465, + "grad_norm": 2.292548179626465, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7415820956230164, + "num_tokens": 361055122.0, + "step": 14065 + }, + { + "epoch": 1.54469580496376, + "grad_norm": 2.4924135208129883, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7258496880531311, + "num_tokens": 361075384.0, + "step": 14066 + }, + { + "epoch": 1.5448056226663738, + "grad_norm": 2.159468650817871, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7517862319946289, + "num_tokens": 361099647.0, + "step": 14067 + }, + { + "epoch": 1.5449154403689875, + "grad_norm": 2.329317092895508, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.727676272392273, + "num_tokens": 361124036.0, + "step": 14068 + }, + { + "epoch": 1.5450252580716013, + "grad_norm": 2.372580051422119, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7022181749343872, + "num_tokens": 361147653.0, + "step": 14069 + }, + { + "epoch": 1.5451350757742148, + "grad_norm": 2.0157148838043213, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6929482221603394, + "num_tokens": 361178390.0, + "step": 14070 + }, + { + "epoch": 1.5452448934768284, + "grad_norm": 2.424067497253418, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7270686626434326, + "num_tokens": 361200168.0, + "step": 14071 + }, + { + "epoch": 1.5453547111794421, + "grad_norm": 1.984532356262207, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7111022472381592, + "num_tokens": 361233361.0, + "step": 14072 + }, + { + "epoch": 1.5454645288820559, + "grad_norm": 2.2331316471099854, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6990050077438354, + "num_tokens": 361261946.0, + "step": 14073 + }, + { + "epoch": 1.5455743465846694, + "grad_norm": 2.399439573287964, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7315572500228882, + "num_tokens": 361283457.0, + "step": 14074 + }, + { + "epoch": 1.545684164287283, + "grad_norm": 2.464818000793457, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7332594394683838, + "num_tokens": 361304452.0, + "step": 14075 + }, + { + "epoch": 1.5457939819898967, + "grad_norm": 2.1508750915527344, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7101893424987793, + "num_tokens": 361331247.0, + "step": 14076 + }, + { + "epoch": 1.5459037996925105, + "grad_norm": 2.4174110889434814, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7064312696456909, + "num_tokens": 361355459.0, + "step": 14077 + }, + { + "epoch": 1.5460136173951242, + "grad_norm": 2.3969156742095947, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7102665901184082, + "num_tokens": 361378484.0, + "step": 14078 + }, + { + "epoch": 1.5461234350977378, + "grad_norm": 2.1763172149658203, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6956387758255005, + "num_tokens": 361409436.0, + "step": 14079 + }, + { + "epoch": 1.5462332528003513, + "grad_norm": 2.445523977279663, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7328246831893921, + "num_tokens": 361431990.0, + "step": 14080 + }, + { + "epoch": 1.546343070502965, + "grad_norm": 2.1590702533721924, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7217944860458374, + "num_tokens": 361461108.0, + "step": 14081 + }, + { + "epoch": 1.5464528882055788, + "grad_norm": 2.50467586517334, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7148091793060303, + "num_tokens": 361483143.0, + "step": 14082 + }, + { + "epoch": 1.5465627059081926, + "grad_norm": 2.193664312362671, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7050656676292419, + "num_tokens": 361511863.0, + "step": 14083 + }, + { + "epoch": 1.546672523610806, + "grad_norm": 2.485044002532959, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7182375192642212, + "num_tokens": 361535169.0, + "step": 14084 + }, + { + "epoch": 1.5467823413134196, + "grad_norm": 2.0809648036956787, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7194214463233948, + "num_tokens": 361562463.0, + "step": 14085 + }, + { + "epoch": 1.5468921590160334, + "grad_norm": 2.236624002456665, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7138444185256958, + "num_tokens": 361587046.0, + "step": 14086 + }, + { + "epoch": 1.5470019767186471, + "grad_norm": 2.210289239883423, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7260611057281494, + "num_tokens": 361613300.0, + "step": 14087 + }, + { + "epoch": 1.5471117944212607, + "grad_norm": 2.3981082439422607, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7167714834213257, + "num_tokens": 361637765.0, + "step": 14088 + }, + { + "epoch": 1.5472216121238742, + "grad_norm": 2.7788851261138916, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7291219234466553, + "num_tokens": 361657062.0, + "step": 14089 + }, + { + "epoch": 1.547331429826488, + "grad_norm": 2.4698729515075684, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7343140840530396, + "num_tokens": 361678105.0, + "step": 14090 + }, + { + "epoch": 1.5474412475291017, + "grad_norm": 2.093731641769409, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6987887620925903, + "num_tokens": 361707918.0, + "step": 14091 + }, + { + "epoch": 1.5475510652317155, + "grad_norm": 2.5118751525878906, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7283346652984619, + "num_tokens": 361728791.0, + "step": 14092 + }, + { + "epoch": 1.547660882934329, + "grad_norm": 2.4351158142089844, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.73277747631073, + "num_tokens": 361751107.0, + "step": 14093 + }, + { + "epoch": 1.5477707006369426, + "grad_norm": 2.402836561203003, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7253524661064148, + "num_tokens": 361773378.0, + "step": 14094 + }, + { + "epoch": 1.5478805183395563, + "grad_norm": 2.298246383666992, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7249473333358765, + "num_tokens": 361796617.0, + "step": 14095 + }, + { + "epoch": 1.54799033604217, + "grad_norm": 2.3251595497131348, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7251477837562561, + "num_tokens": 361822411.0, + "step": 14096 + }, + { + "epoch": 1.5481001537447838, + "grad_norm": 2.0849263668060303, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7323576807975769, + "num_tokens": 361850148.0, + "step": 14097 + }, + { + "epoch": 1.5482099714473974, + "grad_norm": 2.432269334793091, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7145567536354065, + "num_tokens": 361872608.0, + "step": 14098 + }, + { + "epoch": 1.548319789150011, + "grad_norm": 2.1059908866882324, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.724388599395752, + "num_tokens": 361901042.0, + "step": 14099 + }, + { + "epoch": 1.5484296068526247, + "grad_norm": 1.9785975217819214, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7048697471618652, + "num_tokens": 361933314.0, + "step": 14100 + }, + { + "epoch": 1.5485394245552384, + "grad_norm": 2.3905248641967773, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7254087924957275, + "num_tokens": 361956105.0, + "step": 14101 + }, + { + "epoch": 1.548649242257852, + "grad_norm": 2.165985345840454, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6976689696311951, + "num_tokens": 361984794.0, + "step": 14102 + }, + { + "epoch": 1.5487590599604655, + "grad_norm": 2.115039110183716, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7320201992988586, + "num_tokens": 362011620.0, + "step": 14103 + }, + { + "epoch": 1.5488688776630792, + "grad_norm": 2.153398275375366, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7183786034584045, + "num_tokens": 362038804.0, + "step": 14104 + }, + { + "epoch": 1.548978695365693, + "grad_norm": 2.0365920066833496, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7000914216041565, + "num_tokens": 362069476.0, + "step": 14105 + }, + { + "epoch": 1.5490885130683067, + "grad_norm": 2.181833028793335, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6979297399520874, + "num_tokens": 362097245.0, + "step": 14106 + }, + { + "epoch": 1.5491983307709203, + "grad_norm": 2.2744998931884766, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7216530442237854, + "num_tokens": 362121348.0, + "step": 14107 + }, + { + "epoch": 1.5493081484735338, + "grad_norm": 2.432546615600586, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7203948497772217, + "num_tokens": 362144635.0, + "step": 14108 + }, + { + "epoch": 1.5494179661761476, + "grad_norm": 2.371216297149658, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.719852089881897, + "num_tokens": 362165745.0, + "step": 14109 + }, + { + "epoch": 1.5495277838787613, + "grad_norm": 2.0221121311187744, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7096112370491028, + "num_tokens": 362198120.0, + "step": 14110 + }, + { + "epoch": 1.5496376015813749, + "grad_norm": 2.6473047733306885, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7273790240287781, + "num_tokens": 362218548.0, + "step": 14111 + }, + { + "epoch": 1.5497474192839886, + "grad_norm": 2.2053000926971436, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6946940422058105, + "num_tokens": 362246825.0, + "step": 14112 + }, + { + "epoch": 1.5498572369866022, + "grad_norm": 1.9842560291290283, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7253538966178894, + "num_tokens": 362279565.0, + "step": 14113 + }, + { + "epoch": 1.549967054689216, + "grad_norm": 2.5416207313537598, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7001169323921204, + "num_tokens": 362301741.0, + "step": 14114 + }, + { + "epoch": 1.5500768723918297, + "grad_norm": 2.3987348079681396, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7302208542823792, + "num_tokens": 362323466.0, + "step": 14115 + }, + { + "epoch": 1.5501866900944432, + "grad_norm": 2.270214557647705, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7219661474227905, + "num_tokens": 362349164.0, + "step": 14116 + }, + { + "epoch": 1.5502965077970567, + "grad_norm": 2.679626703262329, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7227932214736938, + "num_tokens": 362369712.0, + "step": 14117 + }, + { + "epoch": 1.5504063254996705, + "grad_norm": 2.2440154552459717, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7448288798332214, + "num_tokens": 362393457.0, + "step": 14118 + }, + { + "epoch": 1.5505161432022843, + "grad_norm": 2.464226722717285, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7348348498344421, + "num_tokens": 362413437.0, + "step": 14119 + }, + { + "epoch": 1.550625960904898, + "grad_norm": 2.2824654579162598, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7368677258491516, + "num_tokens": 362437932.0, + "step": 14120 + }, + { + "epoch": 1.5507357786075116, + "grad_norm": 2.42716646194458, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7220854759216309, + "num_tokens": 362460671.0, + "step": 14121 + }, + { + "epoch": 1.550845596310125, + "grad_norm": 2.013658285140991, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7239242792129517, + "num_tokens": 362491942.0, + "step": 14122 + }, + { + "epoch": 1.5509554140127388, + "grad_norm": 2.186582326889038, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7064337134361267, + "num_tokens": 362519339.0, + "step": 14123 + }, + { + "epoch": 1.5510652317153526, + "grad_norm": 2.3727502822875977, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7189039587974548, + "num_tokens": 362544026.0, + "step": 14124 + }, + { + "epoch": 1.5511750494179661, + "grad_norm": 2.110029697418213, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7254939079284668, + "num_tokens": 362572617.0, + "step": 14125 + }, + { + "epoch": 1.55128486712058, + "grad_norm": 2.2008068561553955, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7075241804122925, + "num_tokens": 362599810.0, + "step": 14126 + }, + { + "epoch": 1.5513946848231934, + "grad_norm": 2.411534309387207, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7478834986686707, + "num_tokens": 362621329.0, + "step": 14127 + }, + { + "epoch": 1.5515045025258072, + "grad_norm": 2.204559087753296, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7395689487457275, + "num_tokens": 362646002.0, + "step": 14128 + }, + { + "epoch": 1.551614320228421, + "grad_norm": 2.2908270359039307, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7146493196487427, + "num_tokens": 362670906.0, + "step": 14129 + }, + { + "epoch": 1.5517241379310345, + "grad_norm": 2.386542320251465, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7421863675117493, + "num_tokens": 362694577.0, + "step": 14130 + }, + { + "epoch": 1.551833955633648, + "grad_norm": 2.0953640937805176, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.724962592124939, + "num_tokens": 362720884.0, + "step": 14131 + }, + { + "epoch": 1.5519437733362618, + "grad_norm": 2.3182976245880127, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7218811511993408, + "num_tokens": 362748090.0, + "step": 14132 + }, + { + "epoch": 1.5520535910388755, + "grad_norm": 2.2949109077453613, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7304865717887878, + "num_tokens": 362772692.0, + "step": 14133 + }, + { + "epoch": 1.5521634087414893, + "grad_norm": 2.707794189453125, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7446320056915283, + "num_tokens": 362792034.0, + "step": 14134 + }, + { + "epoch": 1.5522732264441028, + "grad_norm": 2.0112040042877197, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7307918071746826, + "num_tokens": 362823169.0, + "step": 14135 + }, + { + "epoch": 1.5523830441467164, + "grad_norm": 2.199857234954834, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7195529341697693, + "num_tokens": 362849495.0, + "step": 14136 + }, + { + "epoch": 1.55249286184933, + "grad_norm": 1.9642714262008667, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7373381853103638, + "num_tokens": 362879168.0, + "step": 14137 + }, + { + "epoch": 1.5526026795519439, + "grad_norm": 2.38460636138916, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6979027390480042, + "num_tokens": 362903319.0, + "step": 14138 + }, + { + "epoch": 1.5527124972545574, + "grad_norm": 2.3150265216827393, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7098383903503418, + "num_tokens": 362927840.0, + "step": 14139 + }, + { + "epoch": 1.552822314957171, + "grad_norm": 2.4668266773223877, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7195234298706055, + "num_tokens": 362951568.0, + "step": 14140 + }, + { + "epoch": 1.5529321326597847, + "grad_norm": 2.310711145401001, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7160369157791138, + "num_tokens": 362974848.0, + "step": 14141 + }, + { + "epoch": 1.5530419503623984, + "grad_norm": 1.9671822786331177, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7079439759254456, + "num_tokens": 363008911.0, + "step": 14142 + }, + { + "epoch": 1.5531517680650122, + "grad_norm": 2.3897876739501953, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7212411165237427, + "num_tokens": 363031291.0, + "step": 14143 + }, + { + "epoch": 1.5532615857676257, + "grad_norm": 2.547121047973633, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7202863097190857, + "num_tokens": 363051539.0, + "step": 14144 + }, + { + "epoch": 1.5533714034702393, + "grad_norm": 2.1077940464019775, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7108898162841797, + "num_tokens": 363078094.0, + "step": 14145 + }, + { + "epoch": 1.553481221172853, + "grad_norm": 2.1998424530029297, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.730499267578125, + "num_tokens": 363104165.0, + "step": 14146 + }, + { + "epoch": 1.5535910388754668, + "grad_norm": 2.3023626804351807, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7122466564178467, + "num_tokens": 363129261.0, + "step": 14147 + }, + { + "epoch": 1.5537008565780805, + "grad_norm": 2.3778789043426514, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7405025959014893, + "num_tokens": 363152707.0, + "step": 14148 + }, + { + "epoch": 1.553810674280694, + "grad_norm": 2.3209829330444336, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7265349626541138, + "num_tokens": 363175679.0, + "step": 14149 + }, + { + "epoch": 1.5539204919833076, + "grad_norm": 2.2900776863098145, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7042003273963928, + "num_tokens": 363198802.0, + "step": 14150 + }, + { + "epoch": 1.5540303096859214, + "grad_norm": 2.417936086654663, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7484918832778931, + "num_tokens": 363220203.0, + "step": 14151 + }, + { + "epoch": 1.5541401273885351, + "grad_norm": 2.2500810623168945, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7227243185043335, + "num_tokens": 363245862.0, + "step": 14152 + }, + { + "epoch": 1.5542499450911487, + "grad_norm": 2.082160234451294, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7204064726829529, + "num_tokens": 363274786.0, + "step": 14153 + }, + { + "epoch": 1.5543597627937622, + "grad_norm": 2.422244071960449, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7206389904022217, + "num_tokens": 363297656.0, + "step": 14154 + }, + { + "epoch": 1.554469580496376, + "grad_norm": 2.1895060539245605, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7328954935073853, + "num_tokens": 363324941.0, + "step": 14155 + }, + { + "epoch": 1.5545793981989897, + "grad_norm": 2.0860373973846436, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6946590542793274, + "num_tokens": 363357149.0, + "step": 14156 + }, + { + "epoch": 1.5546892159016035, + "grad_norm": 2.2368078231811523, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7295946478843689, + "num_tokens": 363385463.0, + "step": 14157 + }, + { + "epoch": 1.554799033604217, + "grad_norm": 2.074058771133423, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7097287178039551, + "num_tokens": 363415997.0, + "step": 14158 + }, + { + "epoch": 1.5549088513068305, + "grad_norm": 2.020630121231079, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.705355167388916, + "num_tokens": 363445899.0, + "step": 14159 + }, + { + "epoch": 1.5550186690094443, + "grad_norm": 2.1923155784606934, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7000917196273804, + "num_tokens": 363475931.0, + "step": 14160 + }, + { + "epoch": 1.555128486712058, + "grad_norm": 2.1608662605285645, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7192075848579407, + "num_tokens": 363504231.0, + "step": 14161 + }, + { + "epoch": 1.5552383044146716, + "grad_norm": 2.478086233139038, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6973673105239868, + "num_tokens": 363527777.0, + "step": 14162 + }, + { + "epoch": 1.5553481221172853, + "grad_norm": 2.2793142795562744, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.700339138507843, + "num_tokens": 363550922.0, + "step": 14163 + }, + { + "epoch": 1.5554579398198989, + "grad_norm": 2.531667470932007, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7331957817077637, + "num_tokens": 363571287.0, + "step": 14164 + }, + { + "epoch": 1.5555677575225126, + "grad_norm": 2.257460832595825, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7173924446105957, + "num_tokens": 363596451.0, + "step": 14165 + }, + { + "epoch": 1.5556775752251264, + "grad_norm": 2.3137519359588623, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7067227363586426, + "num_tokens": 363621788.0, + "step": 14166 + }, + { + "epoch": 1.55578739292774, + "grad_norm": 2.0932185649871826, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7185174822807312, + "num_tokens": 363651995.0, + "step": 14167 + }, + { + "epoch": 1.5558972106303535, + "grad_norm": 2.2135863304138184, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7051778435707092, + "num_tokens": 363678512.0, + "step": 14168 + }, + { + "epoch": 1.5560070283329672, + "grad_norm": 2.2658438682556152, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7218104600906372, + "num_tokens": 363703009.0, + "step": 14169 + }, + { + "epoch": 1.556116846035581, + "grad_norm": 2.1953978538513184, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7174324989318848, + "num_tokens": 363731036.0, + "step": 14170 + }, + { + "epoch": 1.5562266637381947, + "grad_norm": 2.257387161254883, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7176529765129089, + "num_tokens": 363756525.0, + "step": 14171 + }, + { + "epoch": 1.5563364814408083, + "grad_norm": 2.4076621532440186, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7287188768386841, + "num_tokens": 363779276.0, + "step": 14172 + }, + { + "epoch": 1.5564462991434218, + "grad_norm": 2.375892400741577, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7134681940078735, + "num_tokens": 363803675.0, + "step": 14173 + }, + { + "epoch": 1.5565561168460356, + "grad_norm": 2.4831409454345703, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7260630130767822, + "num_tokens": 363825939.0, + "step": 14174 + }, + { + "epoch": 1.5566659345486493, + "grad_norm": 2.1847569942474365, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7058011889457703, + "num_tokens": 363854130.0, + "step": 14175 + }, + { + "epoch": 1.5567757522512629, + "grad_norm": 2.529975652694702, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6990736722946167, + "num_tokens": 363875939.0, + "step": 14176 + }, + { + "epoch": 1.5568855699538766, + "grad_norm": 2.567174196243286, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7349551916122437, + "num_tokens": 363897243.0, + "step": 14177 + }, + { + "epoch": 1.5569953876564901, + "grad_norm": 2.5140044689178467, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7220988273620605, + "num_tokens": 363917150.0, + "step": 14178 + }, + { + "epoch": 1.557105205359104, + "grad_norm": 2.261739730834961, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7203817963600159, + "num_tokens": 363942312.0, + "step": 14179 + }, + { + "epoch": 1.5572150230617177, + "grad_norm": 2.1043286323547363, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.706362783908844, + "num_tokens": 363968476.0, + "step": 14180 + }, + { + "epoch": 1.5573248407643312, + "grad_norm": 2.265941858291626, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7349101305007935, + "num_tokens": 363991866.0, + "step": 14181 + }, + { + "epoch": 1.5574346584669447, + "grad_norm": 2.1886610984802246, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7246584892272949, + "num_tokens": 364017218.0, + "step": 14182 + }, + { + "epoch": 1.5575444761695585, + "grad_norm": 2.354250907897949, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7526282668113708, + "num_tokens": 364038005.0, + "step": 14183 + }, + { + "epoch": 1.5576542938721722, + "grad_norm": 2.475027561187744, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7219852209091187, + "num_tokens": 364060259.0, + "step": 14184 + }, + { + "epoch": 1.557764111574786, + "grad_norm": 2.2001750469207764, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7267405390739441, + "num_tokens": 364087592.0, + "step": 14185 + }, + { + "epoch": 1.5578739292773995, + "grad_norm": 2.288325548171997, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7232112288475037, + "num_tokens": 364111429.0, + "step": 14186 + }, + { + "epoch": 1.557983746980013, + "grad_norm": 2.3846213817596436, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7051315307617188, + "num_tokens": 364136046.0, + "step": 14187 + }, + { + "epoch": 1.5580935646826268, + "grad_norm": 2.1073925495147705, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7280837297439575, + "num_tokens": 364168371.0, + "step": 14188 + }, + { + "epoch": 1.5582033823852406, + "grad_norm": 1.9653161764144897, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7043026685714722, + "num_tokens": 364200467.0, + "step": 14189 + }, + { + "epoch": 1.5583132000878541, + "grad_norm": 2.3234598636627197, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6993185877799988, + "num_tokens": 364226858.0, + "step": 14190 + }, + { + "epoch": 1.5584230177904677, + "grad_norm": 2.25285005569458, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.726655900478363, + "num_tokens": 364254105.0, + "step": 14191 + }, + { + "epoch": 1.5585328354930814, + "grad_norm": 2.1558289527893066, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7145421504974365, + "num_tokens": 364283220.0, + "step": 14192 + }, + { + "epoch": 1.5586426531956952, + "grad_norm": 2.341679573059082, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7374904751777649, + "num_tokens": 364305568.0, + "step": 14193 + }, + { + "epoch": 1.558752470898309, + "grad_norm": 2.1216161251068115, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7329579591751099, + "num_tokens": 364330561.0, + "step": 14194 + }, + { + "epoch": 1.5588622886009225, + "grad_norm": 2.4000794887542725, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7511234879493713, + "num_tokens": 364353848.0, + "step": 14195 + }, + { + "epoch": 1.558972106303536, + "grad_norm": 2.784299850463867, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7307893633842468, + "num_tokens": 364373126.0, + "step": 14196 + }, + { + "epoch": 1.5590819240061498, + "grad_norm": 2.400588035583496, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7113379836082458, + "num_tokens": 364398659.0, + "step": 14197 + }, + { + "epoch": 1.5591917417087635, + "grad_norm": 2.196823835372925, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7352725267410278, + "num_tokens": 364425426.0, + "step": 14198 + }, + { + "epoch": 1.5593015594113773, + "grad_norm": 2.1846463680267334, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.73260098695755, + "num_tokens": 364450536.0, + "step": 14199 + }, + { + "epoch": 1.5594113771139908, + "grad_norm": 2.0325064659118652, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.719557523727417, + "num_tokens": 364479343.0, + "step": 14200 + }, + { + "epoch": 1.5595211948166043, + "grad_norm": 2.222083568572998, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6890764832496643, + "num_tokens": 364507826.0, + "step": 14201 + }, + { + "epoch": 1.559631012519218, + "grad_norm": 2.1153533458709717, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7221239805221558, + "num_tokens": 364534711.0, + "step": 14202 + }, + { + "epoch": 1.5597408302218319, + "grad_norm": 2.512249708175659, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.750438928604126, + "num_tokens": 364556217.0, + "step": 14203 + }, + { + "epoch": 1.5598506479244454, + "grad_norm": 2.416545867919922, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7113296985626221, + "num_tokens": 364581181.0, + "step": 14204 + }, + { + "epoch": 1.559960465627059, + "grad_norm": 2.1424989700317383, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7255959510803223, + "num_tokens": 364609243.0, + "step": 14205 + }, + { + "epoch": 1.5600702833296727, + "grad_norm": 2.326209545135498, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6982896327972412, + "num_tokens": 364634313.0, + "step": 14206 + }, + { + "epoch": 1.5601801010322864, + "grad_norm": 2.3162193298339844, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7096481919288635, + "num_tokens": 364658098.0, + "step": 14207 + }, + { + "epoch": 1.5602899187349002, + "grad_norm": 2.2759532928466797, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7367368936538696, + "num_tokens": 364682540.0, + "step": 14208 + }, + { + "epoch": 1.5603997364375137, + "grad_norm": 2.0347208976745605, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7048835158348083, + "num_tokens": 364713751.0, + "step": 14209 + }, + { + "epoch": 1.5605095541401273, + "grad_norm": 2.2400519847869873, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7211272716522217, + "num_tokens": 364737870.0, + "step": 14210 + }, + { + "epoch": 1.560619371842741, + "grad_norm": 2.489151954650879, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7317028045654297, + "num_tokens": 364757430.0, + "step": 14211 + }, + { + "epoch": 1.5607291895453548, + "grad_norm": 2.1443116664886475, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.725761890411377, + "num_tokens": 364785292.0, + "step": 14212 + }, + { + "epoch": 1.5608390072479685, + "grad_norm": 2.206547975540161, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7266680002212524, + "num_tokens": 364812151.0, + "step": 14213 + }, + { + "epoch": 1.560948824950582, + "grad_norm": 2.130068063735962, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7206175327301025, + "num_tokens": 364838828.0, + "step": 14214 + }, + { + "epoch": 1.5610586426531956, + "grad_norm": 2.067159414291382, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7152978181838989, + "num_tokens": 364869217.0, + "step": 14215 + }, + { + "epoch": 1.5611684603558094, + "grad_norm": 2.1418297290802, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7035338878631592, + "num_tokens": 364899185.0, + "step": 14216 + }, + { + "epoch": 1.5612782780584231, + "grad_norm": 2.404472589492798, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7178166508674622, + "num_tokens": 364921714.0, + "step": 14217 + }, + { + "epoch": 1.5613880957610367, + "grad_norm": 1.9846487045288086, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7265421748161316, + "num_tokens": 364953680.0, + "step": 14218 + }, + { + "epoch": 1.5614979134636502, + "grad_norm": 2.575582265853882, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7363764643669128, + "num_tokens": 364975582.0, + "step": 14219 + }, + { + "epoch": 1.561607731166264, + "grad_norm": 2.646195888519287, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7256898880004883, + "num_tokens": 364995349.0, + "step": 14220 + }, + { + "epoch": 1.5617175488688777, + "grad_norm": 2.577785015106201, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7317359447479248, + "num_tokens": 365016983.0, + "step": 14221 + }, + { + "epoch": 1.5618273665714915, + "grad_norm": 2.3987419605255127, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7218482494354248, + "num_tokens": 365042193.0, + "step": 14222 + }, + { + "epoch": 1.561937184274105, + "grad_norm": 2.401182174682617, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7107155323028564, + "num_tokens": 365066977.0, + "step": 14223 + }, + { + "epoch": 1.5620470019767185, + "grad_norm": 2.1502017974853516, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7304126024246216, + "num_tokens": 365094654.0, + "step": 14224 + }, + { + "epoch": 1.5621568196793323, + "grad_norm": 2.511321783065796, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7116013765335083, + "num_tokens": 365114853.0, + "step": 14225 + }, + { + "epoch": 1.562266637381946, + "grad_norm": 2.2022078037261963, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.6948570013046265, + "num_tokens": 365142245.0, + "step": 14226 + }, + { + "epoch": 1.5623764550845596, + "grad_norm": 2.223634958267212, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7253902554512024, + "num_tokens": 365170246.0, + "step": 14227 + }, + { + "epoch": 1.5624862727871733, + "grad_norm": 2.571183204650879, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7237626314163208, + "num_tokens": 365190069.0, + "step": 14228 + }, + { + "epoch": 1.5625960904897869, + "grad_norm": 2.2850427627563477, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7225148677825928, + "num_tokens": 365216555.0, + "step": 14229 + }, + { + "epoch": 1.5627059081924006, + "grad_norm": 2.218578815460205, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6951802968978882, + "num_tokens": 365244245.0, + "step": 14230 + }, + { + "epoch": 1.5628157258950144, + "grad_norm": 2.3353428840637207, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7209213972091675, + "num_tokens": 365269668.0, + "step": 14231 + }, + { + "epoch": 1.562925543597628, + "grad_norm": 2.34291410446167, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.729392409324646, + "num_tokens": 365292700.0, + "step": 14232 + }, + { + "epoch": 1.5630353613002415, + "grad_norm": 2.316652774810791, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7232880592346191, + "num_tokens": 365316771.0, + "step": 14233 + }, + { + "epoch": 1.5631451790028552, + "grad_norm": 2.2232487201690674, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7035740613937378, + "num_tokens": 365342847.0, + "step": 14234 + }, + { + "epoch": 1.563254996705469, + "grad_norm": 2.661696434020996, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7471243143081665, + "num_tokens": 365361372.0, + "step": 14235 + }, + { + "epoch": 1.5633648144080827, + "grad_norm": 2.0699825286865234, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7145700454711914, + "num_tokens": 365391282.0, + "step": 14236 + }, + { + "epoch": 1.5634746321106963, + "grad_norm": 2.067537307739258, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7126210927963257, + "num_tokens": 365418927.0, + "step": 14237 + }, + { + "epoch": 1.5635844498133098, + "grad_norm": 2.165367841720581, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7243287563323975, + "num_tokens": 365447374.0, + "step": 14238 + }, + { + "epoch": 1.5636942675159236, + "grad_norm": 2.3270184993743896, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.701914370059967, + "num_tokens": 365474464.0, + "step": 14239 + }, + { + "epoch": 1.5638040852185373, + "grad_norm": 2.4361307621002197, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7353883981704712, + "num_tokens": 365495194.0, + "step": 14240 + }, + { + "epoch": 1.5639139029211508, + "grad_norm": 2.1285901069641113, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7256159782409668, + "num_tokens": 365522927.0, + "step": 14241 + }, + { + "epoch": 1.5640237206237646, + "grad_norm": 2.0536463260650635, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7014812231063843, + "num_tokens": 365553899.0, + "step": 14242 + }, + { + "epoch": 1.5641335383263781, + "grad_norm": 2.4854354858398438, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7222208380699158, + "num_tokens": 365577046.0, + "step": 14243 + }, + { + "epoch": 1.564243356028992, + "grad_norm": 2.052912473678589, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6999848484992981, + "num_tokens": 365607360.0, + "step": 14244 + }, + { + "epoch": 1.5643531737316057, + "grad_norm": 2.044288396835327, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7105326056480408, + "num_tokens": 365637718.0, + "step": 14245 + }, + { + "epoch": 1.5644629914342192, + "grad_norm": 2.0514512062072754, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.718643307685852, + "num_tokens": 365666621.0, + "step": 14246 + }, + { + "epoch": 1.5645728091368327, + "grad_norm": 2.1977272033691406, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7435011863708496, + "num_tokens": 365691788.0, + "step": 14247 + }, + { + "epoch": 1.5646826268394465, + "grad_norm": 2.3599531650543213, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7175416946411133, + "num_tokens": 365717545.0, + "step": 14248 + }, + { + "epoch": 1.5647924445420602, + "grad_norm": 2.2587392330169678, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.728486955165863, + "num_tokens": 365743048.0, + "step": 14249 + }, + { + "epoch": 1.564902262244674, + "grad_norm": 2.1283180713653564, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.717767596244812, + "num_tokens": 365769299.0, + "step": 14250 + }, + { + "epoch": 1.5650120799472875, + "grad_norm": 2.5150506496429443, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7198630571365356, + "num_tokens": 365789752.0, + "step": 14251 + }, + { + "epoch": 1.565121897649901, + "grad_norm": 2.378326177597046, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.709032416343689, + "num_tokens": 365813074.0, + "step": 14252 + }, + { + "epoch": 1.5652317153525148, + "grad_norm": 2.129042387008667, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7219734191894531, + "num_tokens": 365840889.0, + "step": 14253 + }, + { + "epoch": 1.5653415330551286, + "grad_norm": 2.149732828140259, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7115907073020935, + "num_tokens": 365868733.0, + "step": 14254 + }, + { + "epoch": 1.565451350757742, + "grad_norm": 2.7959306240081787, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.745997428894043, + "num_tokens": 365886245.0, + "step": 14255 + }, + { + "epoch": 1.5655611684603556, + "grad_norm": 2.620124578475952, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7008244395256042, + "num_tokens": 365907834.0, + "step": 14256 + }, + { + "epoch": 1.5656709861629694, + "grad_norm": 2.0201594829559326, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7166998386383057, + "num_tokens": 365942223.0, + "step": 14257 + }, + { + "epoch": 1.5657808038655832, + "grad_norm": 2.5186662673950195, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7433537840843201, + "num_tokens": 365964468.0, + "step": 14258 + }, + { + "epoch": 1.565890621568197, + "grad_norm": 2.1105029582977295, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7145091891288757, + "num_tokens": 365996025.0, + "step": 14259 + }, + { + "epoch": 1.5660004392708105, + "grad_norm": 2.227550506591797, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7358986735343933, + "num_tokens": 366018342.0, + "step": 14260 + }, + { + "epoch": 1.566110256973424, + "grad_norm": 2.124974012374878, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7106636762619019, + "num_tokens": 366048075.0, + "step": 14261 + }, + { + "epoch": 1.5662200746760377, + "grad_norm": 2.100271224975586, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.70426344871521, + "num_tokens": 366077565.0, + "step": 14262 + }, + { + "epoch": 1.5663298923786515, + "grad_norm": 2.3990559577941895, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7106320261955261, + "num_tokens": 366101839.0, + "step": 14263 + }, + { + "epoch": 1.5664397100812653, + "grad_norm": 2.2662651538848877, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7101845741271973, + "num_tokens": 366127253.0, + "step": 14264 + }, + { + "epoch": 1.5665495277838788, + "grad_norm": 2.1717052459716797, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7127233743667603, + "num_tokens": 366155434.0, + "step": 14265 + }, + { + "epoch": 1.5666593454864923, + "grad_norm": 2.164496421813965, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7097129821777344, + "num_tokens": 366183637.0, + "step": 14266 + }, + { + "epoch": 1.566769163189106, + "grad_norm": 1.923083782196045, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6988643407821655, + "num_tokens": 366217836.0, + "step": 14267 + }, + { + "epoch": 1.5668789808917198, + "grad_norm": 2.2062013149261475, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7209603190422058, + "num_tokens": 366243906.0, + "step": 14268 + }, + { + "epoch": 1.5669887985943334, + "grad_norm": 2.2381277084350586, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7084835171699524, + "num_tokens": 366268640.0, + "step": 14269 + }, + { + "epoch": 1.567098616296947, + "grad_norm": 2.292053699493408, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7279127836227417, + "num_tokens": 366293751.0, + "step": 14270 + }, + { + "epoch": 1.5672084339995607, + "grad_norm": 2.05088210105896, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7249271273612976, + "num_tokens": 366325206.0, + "step": 14271 + }, + { + "epoch": 1.5673182517021744, + "grad_norm": 2.1963305473327637, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7161574959754944, + "num_tokens": 366349744.0, + "step": 14272 + }, + { + "epoch": 1.5674280694047882, + "grad_norm": 2.303231954574585, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7284106016159058, + "num_tokens": 366375042.0, + "step": 14273 + }, + { + "epoch": 1.5675378871074017, + "grad_norm": 2.290055513381958, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7180147171020508, + "num_tokens": 366400273.0, + "step": 14274 + }, + { + "epoch": 1.5676477048100153, + "grad_norm": 2.383139133453369, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7166092395782471, + "num_tokens": 366425586.0, + "step": 14275 + }, + { + "epoch": 1.567757522512629, + "grad_norm": 2.386756658554077, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7400122880935669, + "num_tokens": 366448006.0, + "step": 14276 + }, + { + "epoch": 1.5678673402152428, + "grad_norm": 2.265519142150879, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7348706722259521, + "num_tokens": 366473279.0, + "step": 14277 + }, + { + "epoch": 1.5679771579178565, + "grad_norm": 2.0234572887420654, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7121760845184326, + "num_tokens": 366503809.0, + "step": 14278 + }, + { + "epoch": 1.56808697562047, + "grad_norm": 2.350557804107666, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7267839312553406, + "num_tokens": 366528891.0, + "step": 14279 + }, + { + "epoch": 1.5681967933230836, + "grad_norm": 2.16245174407959, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6867203712463379, + "num_tokens": 366559444.0, + "step": 14280 + }, + { + "epoch": 1.5683066110256974, + "grad_norm": 2.1432883739471436, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7250630855560303, + "num_tokens": 366586862.0, + "step": 14281 + }, + { + "epoch": 1.568416428728311, + "grad_norm": 2.3359601497650146, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7139497995376587, + "num_tokens": 366612572.0, + "step": 14282 + }, + { + "epoch": 1.5685262464309246, + "grad_norm": 1.933565616607666, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7183887362480164, + "num_tokens": 366644665.0, + "step": 14283 + }, + { + "epoch": 1.5686360641335382, + "grad_norm": 2.055464506149292, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7093839645385742, + "num_tokens": 366672990.0, + "step": 14284 + }, + { + "epoch": 1.568745881836152, + "grad_norm": 2.3573811054229736, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7375614643096924, + "num_tokens": 366696187.0, + "step": 14285 + }, + { + "epoch": 1.5688556995387657, + "grad_norm": 2.2847957611083984, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7247857451438904, + "num_tokens": 366719078.0, + "step": 14286 + }, + { + "epoch": 1.5689655172413794, + "grad_norm": 2.2772786617279053, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7541356086730957, + "num_tokens": 366741709.0, + "step": 14287 + }, + { + "epoch": 1.569075334943993, + "grad_norm": 2.21527099609375, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7280750274658203, + "num_tokens": 366766106.0, + "step": 14288 + }, + { + "epoch": 1.5691851526466065, + "grad_norm": 2.11251163482666, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7111503481864929, + "num_tokens": 366793920.0, + "step": 14289 + }, + { + "epoch": 1.5692949703492203, + "grad_norm": 2.171907663345337, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7320245504379272, + "num_tokens": 366818471.0, + "step": 14290 + }, + { + "epoch": 1.569404788051834, + "grad_norm": 1.879854440689087, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.714029848575592, + "num_tokens": 366853055.0, + "step": 14291 + }, + { + "epoch": 1.5695146057544476, + "grad_norm": 2.1531667709350586, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.731340229511261, + "num_tokens": 366879792.0, + "step": 14292 + }, + { + "epoch": 1.5696244234570613, + "grad_norm": 2.3631834983825684, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7417187690734863, + "num_tokens": 366903363.0, + "step": 14293 + }, + { + "epoch": 1.5697342411596749, + "grad_norm": 2.0521533489227295, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7192621231079102, + "num_tokens": 366932851.0, + "step": 14294 + }, + { + "epoch": 1.5698440588622886, + "grad_norm": 2.2830464839935303, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.724082887172699, + "num_tokens": 366957699.0, + "step": 14295 + }, + { + "epoch": 1.5699538765649024, + "grad_norm": 2.330949544906616, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7439106702804565, + "num_tokens": 366981270.0, + "step": 14296 + }, + { + "epoch": 1.570063694267516, + "grad_norm": 2.0647244453430176, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.6997156143188477, + "num_tokens": 367012450.0, + "step": 14297 + }, + { + "epoch": 1.5701735119701294, + "grad_norm": 2.331315755844116, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7338734865188599, + "num_tokens": 367035384.0, + "step": 14298 + }, + { + "epoch": 1.5702833296727432, + "grad_norm": 2.123581886291504, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7080926895141602, + "num_tokens": 367064336.0, + "step": 14299 + }, + { + "epoch": 1.570393147375357, + "grad_norm": 2.270800828933716, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7056186199188232, + "num_tokens": 367093392.0, + "step": 14300 + }, + { + "epoch": 1.5705029650779707, + "grad_norm": 2.1840097904205322, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6952921748161316, + "num_tokens": 367121099.0, + "step": 14301 + }, + { + "epoch": 1.5706127827805842, + "grad_norm": 2.433689594268799, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.70985347032547, + "num_tokens": 367144410.0, + "step": 14302 + }, + { + "epoch": 1.5707226004831978, + "grad_norm": 2.373363733291626, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7433887720108032, + "num_tokens": 367166559.0, + "step": 14303 + }, + { + "epoch": 1.5708324181858115, + "grad_norm": 2.0352134704589844, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7104021906852722, + "num_tokens": 367196583.0, + "step": 14304 + }, + { + "epoch": 1.5709422358884253, + "grad_norm": 2.248722791671753, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7420439124107361, + "num_tokens": 367219486.0, + "step": 14305 + }, + { + "epoch": 1.5710520535910388, + "grad_norm": 2.1265692710876465, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7058707475662231, + "num_tokens": 367249108.0, + "step": 14306 + }, + { + "epoch": 1.5711618712936526, + "grad_norm": 2.2238855361938477, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7015862464904785, + "num_tokens": 367274958.0, + "step": 14307 + }, + { + "epoch": 1.5712716889962661, + "grad_norm": 2.1170530319213867, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7443538904190063, + "num_tokens": 367299400.0, + "step": 14308 + }, + { + "epoch": 1.5713815066988799, + "grad_norm": 2.3124380111694336, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7413545250892639, + "num_tokens": 367322209.0, + "step": 14309 + }, + { + "epoch": 1.5714913244014936, + "grad_norm": 2.105851650238037, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7287887930870056, + "num_tokens": 367350337.0, + "step": 14310 + }, + { + "epoch": 1.5716011421041072, + "grad_norm": 1.984925627708435, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7093595266342163, + "num_tokens": 367381390.0, + "step": 14311 + }, + { + "epoch": 1.5717109598067207, + "grad_norm": 2.140599489212036, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7159058451652527, + "num_tokens": 367408305.0, + "step": 14312 + }, + { + "epoch": 1.5718207775093345, + "grad_norm": 2.3529510498046875, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7200919389724731, + "num_tokens": 367430719.0, + "step": 14313 + }, + { + "epoch": 1.5719305952119482, + "grad_norm": 2.4685235023498535, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7438359260559082, + "num_tokens": 367452182.0, + "step": 14314 + }, + { + "epoch": 1.572040412914562, + "grad_norm": 2.2785985469818115, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7471548914909363, + "num_tokens": 367475992.0, + "step": 14315 + }, + { + "epoch": 1.5721502306171755, + "grad_norm": 2.0637764930725098, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.716684103012085, + "num_tokens": 367506694.0, + "step": 14316 + }, + { + "epoch": 1.572260048319789, + "grad_norm": 2.289668321609497, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.69432133436203, + "num_tokens": 367532555.0, + "step": 14317 + }, + { + "epoch": 1.5723698660224028, + "grad_norm": 2.1981351375579834, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7012836337089539, + "num_tokens": 367558939.0, + "step": 14318 + }, + { + "epoch": 1.5724796837250166, + "grad_norm": 2.1727566719055176, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7084754705429077, + "num_tokens": 367587777.0, + "step": 14319 + }, + { + "epoch": 1.57258950142763, + "grad_norm": 2.2485787868499756, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7212435603141785, + "num_tokens": 367613892.0, + "step": 14320 + }, + { + "epoch": 1.5726993191302436, + "grad_norm": 2.054108142852783, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.701683759689331, + "num_tokens": 367644896.0, + "step": 14321 + }, + { + "epoch": 1.5728091368328574, + "grad_norm": 2.4463815689086914, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7217302322387695, + "num_tokens": 367666921.0, + "step": 14322 + }, + { + "epoch": 1.5729189545354711, + "grad_norm": 2.083780288696289, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7152245044708252, + "num_tokens": 367695663.0, + "step": 14323 + }, + { + "epoch": 1.573028772238085, + "grad_norm": 2.189976453781128, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.716201901435852, + "num_tokens": 367722398.0, + "step": 14324 + }, + { + "epoch": 1.5731385899406984, + "grad_norm": 2.4971282482147217, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.719041645526886, + "num_tokens": 367743924.0, + "step": 14325 + }, + { + "epoch": 1.573248407643312, + "grad_norm": 2.1685972213745117, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7218116521835327, + "num_tokens": 367769508.0, + "step": 14326 + }, + { + "epoch": 1.5733582253459257, + "grad_norm": 1.9466290473937988, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7032774686813354, + "num_tokens": 367801855.0, + "step": 14327 + }, + { + "epoch": 1.5734680430485395, + "grad_norm": 2.400808572769165, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7378929257392883, + "num_tokens": 367824130.0, + "step": 14328 + }, + { + "epoch": 1.5735778607511532, + "grad_norm": 2.3431410789489746, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.697825014591217, + "num_tokens": 367848538.0, + "step": 14329 + }, + { + "epoch": 1.5736876784537668, + "grad_norm": 2.1129748821258545, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6973429918289185, + "num_tokens": 367878362.0, + "step": 14330 + }, + { + "epoch": 1.5737974961563803, + "grad_norm": 2.5152621269226074, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7426360249519348, + "num_tokens": 367897746.0, + "step": 14331 + }, + { + "epoch": 1.573907313858994, + "grad_norm": 2.157928705215454, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7200006246566772, + "num_tokens": 367922774.0, + "step": 14332 + }, + { + "epoch": 1.5740171315616078, + "grad_norm": 2.281686305999756, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7093859910964966, + "num_tokens": 367948105.0, + "step": 14333 + }, + { + "epoch": 1.5741269492642214, + "grad_norm": 2.231313467025757, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6860827207565308, + "num_tokens": 367975537.0, + "step": 14334 + }, + { + "epoch": 1.574236766966835, + "grad_norm": 2.1294138431549072, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7069970369338989, + "num_tokens": 368002456.0, + "step": 14335 + }, + { + "epoch": 1.5743465846694487, + "grad_norm": 2.188981771469116, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7208123207092285, + "num_tokens": 368028328.0, + "step": 14336 + }, + { + "epoch": 1.5744564023720624, + "grad_norm": 2.2677714824676514, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6944669485092163, + "num_tokens": 368054521.0, + "step": 14337 + }, + { + "epoch": 1.5745662200746762, + "grad_norm": 2.5344998836517334, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7262635231018066, + "num_tokens": 368075378.0, + "step": 14338 + }, + { + "epoch": 1.5746760377772897, + "grad_norm": 2.237438917160034, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7238888740539551, + "num_tokens": 368099962.0, + "step": 14339 + }, + { + "epoch": 1.5747858554799032, + "grad_norm": 2.1354310512542725, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7060390710830688, + "num_tokens": 368127255.0, + "step": 14340 + }, + { + "epoch": 1.574895673182517, + "grad_norm": 2.172673225402832, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7315391302108765, + "num_tokens": 368154602.0, + "step": 14341 + }, + { + "epoch": 1.5750054908851308, + "grad_norm": 2.4865305423736572, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7373645305633545, + "num_tokens": 368177411.0, + "step": 14342 + }, + { + "epoch": 1.5751153085877443, + "grad_norm": 2.534536838531494, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7207633852958679, + "num_tokens": 368198269.0, + "step": 14343 + }, + { + "epoch": 1.575225126290358, + "grad_norm": 1.9723299741744995, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6952603459358215, + "num_tokens": 368232389.0, + "step": 14344 + }, + { + "epoch": 1.5753349439929716, + "grad_norm": 2.128192901611328, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7336217164993286, + "num_tokens": 368259709.0, + "step": 14345 + }, + { + "epoch": 1.5754447616955853, + "grad_norm": 2.2824621200561523, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7133030891418457, + "num_tokens": 368284242.0, + "step": 14346 + }, + { + "epoch": 1.575554579398199, + "grad_norm": 2.1550145149230957, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6990181803703308, + "num_tokens": 368312048.0, + "step": 14347 + }, + { + "epoch": 1.5756643971008126, + "grad_norm": 2.394341468811035, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7337069511413574, + "num_tokens": 368334659.0, + "step": 14348 + }, + { + "epoch": 1.5757742148034262, + "grad_norm": 2.085869550704956, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6999013423919678, + "num_tokens": 368369170.0, + "step": 14349 + }, + { + "epoch": 1.57588403250604, + "grad_norm": 2.4103665351867676, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7265897989273071, + "num_tokens": 368391233.0, + "step": 14350 + }, + { + "epoch": 1.5759938502086537, + "grad_norm": 2.2177176475524902, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.733608603477478, + "num_tokens": 368415659.0, + "step": 14351 + }, + { + "epoch": 1.5761036679112674, + "grad_norm": 2.210538625717163, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7039929628372192, + "num_tokens": 368443312.0, + "step": 14352 + }, + { + "epoch": 1.576213485613881, + "grad_norm": 2.3939082622528076, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7214970588684082, + "num_tokens": 368465845.0, + "step": 14353 + }, + { + "epoch": 1.5763233033164945, + "grad_norm": 2.274709463119507, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7178143262863159, + "num_tokens": 368490310.0, + "step": 14354 + }, + { + "epoch": 1.5764331210191083, + "grad_norm": 2.281845808029175, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7291378378868103, + "num_tokens": 368514800.0, + "step": 14355 + }, + { + "epoch": 1.576542938721722, + "grad_norm": 2.165436029434204, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6904247999191284, + "num_tokens": 368544446.0, + "step": 14356 + }, + { + "epoch": 1.5766527564243356, + "grad_norm": 2.2913615703582764, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7444158792495728, + "num_tokens": 368567332.0, + "step": 14357 + }, + { + "epoch": 1.5767625741269493, + "grad_norm": 2.1147053241729736, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7477951049804688, + "num_tokens": 368596150.0, + "step": 14358 + }, + { + "epoch": 1.5768723918295628, + "grad_norm": 1.9821988344192505, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7249414324760437, + "num_tokens": 368627242.0, + "step": 14359 + }, + { + "epoch": 1.5769822095321766, + "grad_norm": 2.2661993503570557, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7025178670883179, + "num_tokens": 368652758.0, + "step": 14360 + }, + { + "epoch": 1.5770920272347904, + "grad_norm": 2.2387101650238037, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7102376222610474, + "num_tokens": 368679645.0, + "step": 14361 + }, + { + "epoch": 1.577201844937404, + "grad_norm": 2.2952425479888916, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7201273441314697, + "num_tokens": 368704983.0, + "step": 14362 + }, + { + "epoch": 1.5773116626400174, + "grad_norm": 1.9973188638687134, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7027038335800171, + "num_tokens": 368736885.0, + "step": 14363 + }, + { + "epoch": 1.5774214803426312, + "grad_norm": 2.3390655517578125, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.737534761428833, + "num_tokens": 368762432.0, + "step": 14364 + }, + { + "epoch": 1.577531298045245, + "grad_norm": 2.093630075454712, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6988682746887207, + "num_tokens": 368791070.0, + "step": 14365 + }, + { + "epoch": 1.5776411157478587, + "grad_norm": 1.942320704460144, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7007895708084106, + "num_tokens": 368824069.0, + "step": 14366 + }, + { + "epoch": 1.5777509334504722, + "grad_norm": 2.2139997482299805, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7130041122436523, + "num_tokens": 368851098.0, + "step": 14367 + }, + { + "epoch": 1.5778607511530858, + "grad_norm": 2.4013938903808594, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7321994304656982, + "num_tokens": 368872221.0, + "step": 14368 + }, + { + "epoch": 1.5779705688556995, + "grad_norm": 1.9846478700637817, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7161285877227783, + "num_tokens": 368903317.0, + "step": 14369 + }, + { + "epoch": 1.5780803865583133, + "grad_norm": 2.304442882537842, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7086533308029175, + "num_tokens": 368927861.0, + "step": 14370 + }, + { + "epoch": 1.5781902042609268, + "grad_norm": 2.20413875579834, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.710553765296936, + "num_tokens": 368955070.0, + "step": 14371 + }, + { + "epoch": 1.5783000219635406, + "grad_norm": 2.250194787979126, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7348283529281616, + "num_tokens": 368979308.0, + "step": 14372 + }, + { + "epoch": 1.5784098396661541, + "grad_norm": 2.3753230571746826, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7190661430358887, + "num_tokens": 369002180.0, + "step": 14373 + }, + { + "epoch": 1.5785196573687679, + "grad_norm": 2.1964457035064697, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7081459760665894, + "num_tokens": 369028448.0, + "step": 14374 + }, + { + "epoch": 1.5786294750713816, + "grad_norm": 2.7929813861846924, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7456249594688416, + "num_tokens": 369047985.0, + "step": 14375 + }, + { + "epoch": 1.5787392927739952, + "grad_norm": 2.187044143676758, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7443182468414307, + "num_tokens": 369074366.0, + "step": 14376 + }, + { + "epoch": 1.5788491104766087, + "grad_norm": 2.3389055728912354, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.709439754486084, + "num_tokens": 369097236.0, + "step": 14377 + }, + { + "epoch": 1.5789589281792225, + "grad_norm": 2.1874499320983887, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7034615278244019, + "num_tokens": 369123485.0, + "step": 14378 + }, + { + "epoch": 1.5790687458818362, + "grad_norm": 2.131772518157959, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7127505540847778, + "num_tokens": 369151635.0, + "step": 14379 + }, + { + "epoch": 1.57917856358445, + "grad_norm": 2.2729337215423584, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7486552596092224, + "num_tokens": 369173807.0, + "step": 14380 + }, + { + "epoch": 1.5792883812870635, + "grad_norm": 2.128041982650757, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7142629623413086, + "num_tokens": 369200995.0, + "step": 14381 + }, + { + "epoch": 1.579398198989677, + "grad_norm": 2.3559837341308594, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7257243394851685, + "num_tokens": 369222663.0, + "step": 14382 + }, + { + "epoch": 1.5795080166922908, + "grad_norm": 2.262497663497925, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7115331888198853, + "num_tokens": 369247722.0, + "step": 14383 + }, + { + "epoch": 1.5796178343949046, + "grad_norm": 2.0589537620544434, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7247110605239868, + "num_tokens": 369274845.0, + "step": 14384 + }, + { + "epoch": 1.579727652097518, + "grad_norm": 2.2045400142669678, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7201272249221802, + "num_tokens": 369300524.0, + "step": 14385 + }, + { + "epoch": 1.5798374698001316, + "grad_norm": 2.403033971786499, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6867260932922363, + "num_tokens": 369326976.0, + "step": 14386 + }, + { + "epoch": 1.5799472875027454, + "grad_norm": 2.2481415271759033, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7144697904586792, + "num_tokens": 369352076.0, + "step": 14387 + }, + { + "epoch": 1.5800571052053591, + "grad_norm": 2.5272305011749268, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7302377223968506, + "num_tokens": 369372511.0, + "step": 14388 + }, + { + "epoch": 1.580166922907973, + "grad_norm": 2.1326324939727783, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7304818034172058, + "num_tokens": 369397319.0, + "step": 14389 + }, + { + "epoch": 1.5802767406105864, + "grad_norm": 2.1066372394561768, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.716230034828186, + "num_tokens": 369426532.0, + "step": 14390 + }, + { + "epoch": 1.5803865583132, + "grad_norm": 2.0903091430664062, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7073577046394348, + "num_tokens": 369456101.0, + "step": 14391 + }, + { + "epoch": 1.5804963760158137, + "grad_norm": 2.061417579650879, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6817071437835693, + "num_tokens": 369488259.0, + "step": 14392 + }, + { + "epoch": 1.5806061937184275, + "grad_norm": 2.1314380168914795, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7403740882873535, + "num_tokens": 369514923.0, + "step": 14393 + }, + { + "epoch": 1.5807160114210412, + "grad_norm": 2.7160987854003906, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7201076745986938, + "num_tokens": 369535368.0, + "step": 14394 + }, + { + "epoch": 1.5808258291236548, + "grad_norm": 2.803334951400757, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7386433482170105, + "num_tokens": 369554025.0, + "step": 14395 + }, + { + "epoch": 1.5809356468262683, + "grad_norm": 2.3328394889831543, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7130683660507202, + "num_tokens": 369578043.0, + "step": 14396 + }, + { + "epoch": 1.581045464528882, + "grad_norm": 2.351008415222168, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7168852090835571, + "num_tokens": 369603656.0, + "step": 14397 + }, + { + "epoch": 1.5811552822314958, + "grad_norm": 2.3958911895751953, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7214252352714539, + "num_tokens": 369629137.0, + "step": 14398 + }, + { + "epoch": 1.5812650999341094, + "grad_norm": 2.1700186729431152, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7111953496932983, + "num_tokens": 369654252.0, + "step": 14399 + }, + { + "epoch": 1.5813749176367229, + "grad_norm": 2.4115195274353027, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7254005670547485, + "num_tokens": 369675064.0, + "step": 14400 + }, + { + "epoch": 1.5814847353393366, + "grad_norm": 2.176964044570923, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7325325012207031, + "num_tokens": 369699681.0, + "step": 14401 + }, + { + "epoch": 1.5815945530419504, + "grad_norm": 2.3180713653564453, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7246266007423401, + "num_tokens": 369725075.0, + "step": 14402 + }, + { + "epoch": 1.5817043707445642, + "grad_norm": 2.5808231830596924, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.720064640045166, + "num_tokens": 369746237.0, + "step": 14403 + }, + { + "epoch": 1.5818141884471777, + "grad_norm": 2.0485477447509766, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7233072519302368, + "num_tokens": 369775159.0, + "step": 14404 + }, + { + "epoch": 1.5819240061497912, + "grad_norm": 2.08189058303833, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6966750025749207, + "num_tokens": 369804594.0, + "step": 14405 + }, + { + "epoch": 1.582033823852405, + "grad_norm": 2.1836366653442383, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.726030170917511, + "num_tokens": 369829580.0, + "step": 14406 + }, + { + "epoch": 1.5821436415550187, + "grad_norm": 2.057129144668579, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7241113185882568, + "num_tokens": 369858353.0, + "step": 14407 + }, + { + "epoch": 1.5822534592576323, + "grad_norm": 2.059385061264038, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7481851577758789, + "num_tokens": 369886068.0, + "step": 14408 + }, + { + "epoch": 1.582363276960246, + "grad_norm": 2.1023740768432617, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7400424480438232, + "num_tokens": 369915200.0, + "step": 14409 + }, + { + "epoch": 1.5824730946628596, + "grad_norm": 2.078098773956299, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7036395072937012, + "num_tokens": 369946596.0, + "step": 14410 + }, + { + "epoch": 1.5825829123654733, + "grad_norm": 2.259445905685425, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7171576023101807, + "num_tokens": 369971659.0, + "step": 14411 + }, + { + "epoch": 1.582692730068087, + "grad_norm": 2.1644036769866943, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7523718476295471, + "num_tokens": 369997751.0, + "step": 14412 + }, + { + "epoch": 1.5828025477707006, + "grad_norm": 2.3278422355651855, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7224478721618652, + "num_tokens": 370020913.0, + "step": 14413 + }, + { + "epoch": 1.5829123654733142, + "grad_norm": 2.073683261871338, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7200731039047241, + "num_tokens": 370049492.0, + "step": 14414 + }, + { + "epoch": 1.583022183175928, + "grad_norm": 2.175779342651367, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7117822766304016, + "num_tokens": 370077436.0, + "step": 14415 + }, + { + "epoch": 1.5831320008785417, + "grad_norm": 2.5437049865722656, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7215048670768738, + "num_tokens": 370099121.0, + "step": 14416 + }, + { + "epoch": 1.5832418185811554, + "grad_norm": 2.307109832763672, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7167108058929443, + "num_tokens": 370123881.0, + "step": 14417 + }, + { + "epoch": 1.583351636283769, + "grad_norm": 2.1621153354644775, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7142812013626099, + "num_tokens": 370150963.0, + "step": 14418 + }, + { + "epoch": 1.5834614539863825, + "grad_norm": 2.59411358833313, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7131118178367615, + "num_tokens": 370172103.0, + "step": 14419 + }, + { + "epoch": 1.5835712716889963, + "grad_norm": 2.195880889892578, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7018663883209229, + "num_tokens": 370201473.0, + "step": 14420 + }, + { + "epoch": 1.58368108939161, + "grad_norm": 2.4547464847564697, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7378807663917542, + "num_tokens": 370222624.0, + "step": 14421 + }, + { + "epoch": 1.5837909070942235, + "grad_norm": 2.345515012741089, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7395718097686768, + "num_tokens": 370245061.0, + "step": 14422 + }, + { + "epoch": 1.5839007247968373, + "grad_norm": 2.5700552463531494, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.720086932182312, + "num_tokens": 370266535.0, + "step": 14423 + }, + { + "epoch": 1.5840105424994508, + "grad_norm": 2.215230703353882, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7211601734161377, + "num_tokens": 370290945.0, + "step": 14424 + }, + { + "epoch": 1.5841203602020646, + "grad_norm": 2.3357927799224854, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7205615043640137, + "num_tokens": 370313105.0, + "step": 14425 + }, + { + "epoch": 1.5842301779046783, + "grad_norm": 2.207749843597412, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7231485843658447, + "num_tokens": 370339012.0, + "step": 14426 + }, + { + "epoch": 1.5843399956072919, + "grad_norm": 2.028867721557617, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7116246819496155, + "num_tokens": 370370627.0, + "step": 14427 + }, + { + "epoch": 1.5844498133099054, + "grad_norm": 2.5822033882141113, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7181456089019775, + "num_tokens": 370391226.0, + "step": 14428 + }, + { + "epoch": 1.5845596310125192, + "grad_norm": 1.8670480251312256, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7043734192848206, + "num_tokens": 370424981.0, + "step": 14429 + }, + { + "epoch": 1.584669448715133, + "grad_norm": 2.33612322807312, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7236965894699097, + "num_tokens": 370448530.0, + "step": 14430 + }, + { + "epoch": 1.5847792664177467, + "grad_norm": 2.487713575363159, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6929844617843628, + "num_tokens": 370471590.0, + "step": 14431 + }, + { + "epoch": 1.5848890841203602, + "grad_norm": 2.2298431396484375, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.6968179941177368, + "num_tokens": 370501105.0, + "step": 14432 + }, + { + "epoch": 1.5849989018229738, + "grad_norm": 2.148564338684082, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.711472749710083, + "num_tokens": 370528936.0, + "step": 14433 + }, + { + "epoch": 1.5851087195255875, + "grad_norm": 2.64872145652771, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7492325305938721, + "num_tokens": 370548288.0, + "step": 14434 + }, + { + "epoch": 1.5852185372282013, + "grad_norm": 2.0840041637420654, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7271794080734253, + "num_tokens": 370575081.0, + "step": 14435 + }, + { + "epoch": 1.5853283549308148, + "grad_norm": 2.4138379096984863, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7242599725723267, + "num_tokens": 370597797.0, + "step": 14436 + }, + { + "epoch": 1.5854381726334283, + "grad_norm": 2.530860424041748, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7379170656204224, + "num_tokens": 370621231.0, + "step": 14437 + }, + { + "epoch": 1.585547990336042, + "grad_norm": 2.2113254070281982, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7272850275039673, + "num_tokens": 370649441.0, + "step": 14438 + }, + { + "epoch": 1.5856578080386559, + "grad_norm": 2.115990161895752, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7377099990844727, + "num_tokens": 370676271.0, + "step": 14439 + }, + { + "epoch": 1.5857676257412696, + "grad_norm": 2.2562084197998047, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.710923433303833, + "num_tokens": 370703210.0, + "step": 14440 + }, + { + "epoch": 1.5858774434438832, + "grad_norm": 2.279818534851074, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7226302027702332, + "num_tokens": 370729327.0, + "step": 14441 + }, + { + "epoch": 1.5859872611464967, + "grad_norm": 2.1339173316955566, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7153340578079224, + "num_tokens": 370756825.0, + "step": 14442 + }, + { + "epoch": 1.5860970788491104, + "grad_norm": 2.200026035308838, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7178164720535278, + "num_tokens": 370782992.0, + "step": 14443 + }, + { + "epoch": 1.5862068965517242, + "grad_norm": 2.3810369968414307, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7097419500350952, + "num_tokens": 370805615.0, + "step": 14444 + }, + { + "epoch": 1.586316714254338, + "grad_norm": 2.1208503246307373, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7222668528556824, + "num_tokens": 370832864.0, + "step": 14445 + }, + { + "epoch": 1.5864265319569515, + "grad_norm": 2.033048629760742, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7222424745559692, + "num_tokens": 370862777.0, + "step": 14446 + }, + { + "epoch": 1.586536349659565, + "grad_norm": 2.3465940952301025, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7101448774337769, + "num_tokens": 370887022.0, + "step": 14447 + }, + { + "epoch": 1.5866461673621788, + "grad_norm": 2.4891245365142822, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7184761762619019, + "num_tokens": 370908778.0, + "step": 14448 + }, + { + "epoch": 1.5867559850647925, + "grad_norm": 2.2756190299987793, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7239717245101929, + "num_tokens": 370932828.0, + "step": 14449 + }, + { + "epoch": 1.586865802767406, + "grad_norm": 2.128725290298462, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7228652834892273, + "num_tokens": 370959917.0, + "step": 14450 + }, + { + "epoch": 1.5869756204700196, + "grad_norm": 2.3640003204345703, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7429758310317993, + "num_tokens": 370982752.0, + "step": 14451 + }, + { + "epoch": 1.5870854381726334, + "grad_norm": 2.493595600128174, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7276977896690369, + "num_tokens": 371003250.0, + "step": 14452 + }, + { + "epoch": 1.5871952558752471, + "grad_norm": 2.3533506393432617, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7148672342300415, + "num_tokens": 371027648.0, + "step": 14453 + }, + { + "epoch": 1.5873050735778609, + "grad_norm": 2.2937350273132324, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7145116329193115, + "num_tokens": 371055234.0, + "step": 14454 + }, + { + "epoch": 1.5874148912804744, + "grad_norm": 2.2857182025909424, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7340914011001587, + "num_tokens": 371078601.0, + "step": 14455 + }, + { + "epoch": 1.587524708983088, + "grad_norm": 2.3979334831237793, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.70118647813797, + "num_tokens": 371104371.0, + "step": 14456 + }, + { + "epoch": 1.5876345266857017, + "grad_norm": 1.9963740110397339, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7263670563697815, + "num_tokens": 371134820.0, + "step": 14457 + }, + { + "epoch": 1.5877443443883155, + "grad_norm": 2.2484514713287354, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7046384811401367, + "num_tokens": 371161038.0, + "step": 14458 + }, + { + "epoch": 1.5878541620909292, + "grad_norm": 2.1239566802978516, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7002177834510803, + "num_tokens": 371189938.0, + "step": 14459 + }, + { + "epoch": 1.5879639797935428, + "grad_norm": 2.2359514236450195, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7144080400466919, + "num_tokens": 371217010.0, + "step": 14460 + }, + { + "epoch": 1.5880737974961563, + "grad_norm": 2.160874366760254, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7109825611114502, + "num_tokens": 371243016.0, + "step": 14461 + }, + { + "epoch": 1.58818361519877, + "grad_norm": 2.2493999004364014, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6945918798446655, + "num_tokens": 371269771.0, + "step": 14462 + }, + { + "epoch": 1.5882934329013838, + "grad_norm": 2.1747524738311768, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7195678949356079, + "num_tokens": 371297439.0, + "step": 14463 + }, + { + "epoch": 1.5884032506039973, + "grad_norm": 2.6157236099243164, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7371408939361572, + "num_tokens": 371317870.0, + "step": 14464 + }, + { + "epoch": 1.5885130683066109, + "grad_norm": 2.223458766937256, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7498526573181152, + "num_tokens": 371342658.0, + "step": 14465 + }, + { + "epoch": 1.5886228860092246, + "grad_norm": 2.236051082611084, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7111402750015259, + "num_tokens": 371368878.0, + "step": 14466 + }, + { + "epoch": 1.5887327037118384, + "grad_norm": 2.113895893096924, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7123100757598877, + "num_tokens": 371396637.0, + "step": 14467 + }, + { + "epoch": 1.5888425214144521, + "grad_norm": 2.1842453479766846, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7354438900947571, + "num_tokens": 371421671.0, + "step": 14468 + }, + { + "epoch": 1.5889523391170657, + "grad_norm": 2.2804737091064453, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.721712589263916, + "num_tokens": 371444205.0, + "step": 14469 + }, + { + "epoch": 1.5890621568196792, + "grad_norm": 2.16133189201355, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7248044610023499, + "num_tokens": 371471373.0, + "step": 14470 + }, + { + "epoch": 1.589171974522293, + "grad_norm": 1.9651557207107544, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7178643941879272, + "num_tokens": 371503685.0, + "step": 14471 + }, + { + "epoch": 1.5892817922249067, + "grad_norm": 2.2946364879608154, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7360100150108337, + "num_tokens": 371526932.0, + "step": 14472 + }, + { + "epoch": 1.5893916099275203, + "grad_norm": 2.523413896560669, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7314929962158203, + "num_tokens": 371545766.0, + "step": 14473 + }, + { + "epoch": 1.589501427630134, + "grad_norm": 2.0086259841918945, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7151777744293213, + "num_tokens": 371575614.0, + "step": 14474 + }, + { + "epoch": 1.5896112453327476, + "grad_norm": 2.169224262237549, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7228535413742065, + "num_tokens": 371600854.0, + "step": 14475 + }, + { + "epoch": 1.5897210630353613, + "grad_norm": 2.2185685634613037, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7125194072723389, + "num_tokens": 371628744.0, + "step": 14476 + }, + { + "epoch": 1.589830880737975, + "grad_norm": 1.95732581615448, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7261612415313721, + "num_tokens": 371662884.0, + "step": 14477 + }, + { + "epoch": 1.5899406984405886, + "grad_norm": 2.0588483810424805, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7223213911056519, + "num_tokens": 371693571.0, + "step": 14478 + }, + { + "epoch": 1.5900505161432021, + "grad_norm": 2.058734178543091, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7238659858703613, + "num_tokens": 371721121.0, + "step": 14479 + }, + { + "epoch": 1.590160333845816, + "grad_norm": 2.3589253425598145, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7079371213912964, + "num_tokens": 371746613.0, + "step": 14480 + }, + { + "epoch": 1.5902701515484297, + "grad_norm": 2.2795181274414062, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7215110063552856, + "num_tokens": 371770984.0, + "step": 14481 + }, + { + "epoch": 1.5903799692510434, + "grad_norm": 2.356431484222412, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7539443969726562, + "num_tokens": 371792724.0, + "step": 14482 + }, + { + "epoch": 1.590489786953657, + "grad_norm": 2.3176653385162354, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7434664964675903, + "num_tokens": 371814600.0, + "step": 14483 + }, + { + "epoch": 1.5905996046562705, + "grad_norm": 2.0518088340759277, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7379658818244934, + "num_tokens": 371843474.0, + "step": 14484 + }, + { + "epoch": 1.5907094223588842, + "grad_norm": 2.3252124786376953, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6997401118278503, + "num_tokens": 371869309.0, + "step": 14485 + }, + { + "epoch": 1.590819240061498, + "grad_norm": 2.0995426177978516, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7205918431282043, + "num_tokens": 371897508.0, + "step": 14486 + }, + { + "epoch": 1.5909290577641115, + "grad_norm": 2.343350648880005, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7346874475479126, + "num_tokens": 371921480.0, + "step": 14487 + }, + { + "epoch": 1.5910388754667253, + "grad_norm": 2.208549737930298, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7206580638885498, + "num_tokens": 371947077.0, + "step": 14488 + }, + { + "epoch": 1.5911486931693388, + "grad_norm": 2.5061709880828857, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.71124267578125, + "num_tokens": 371969065.0, + "step": 14489 + }, + { + "epoch": 1.5912585108719526, + "grad_norm": 1.9985898733139038, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7320142984390259, + "num_tokens": 371998126.0, + "step": 14490 + }, + { + "epoch": 1.5913683285745663, + "grad_norm": 1.943787693977356, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6865038871765137, + "num_tokens": 372032424.0, + "step": 14491 + }, + { + "epoch": 1.5914781462771799, + "grad_norm": 1.9616628885269165, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.6999804973602295, + "num_tokens": 372065643.0, + "step": 14492 + }, + { + "epoch": 1.5915879639797934, + "grad_norm": 2.0291881561279297, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7025079727172852, + "num_tokens": 372097455.0, + "step": 14493 + }, + { + "epoch": 1.5916977816824072, + "grad_norm": 2.4581305980682373, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7223501205444336, + "num_tokens": 372118869.0, + "step": 14494 + }, + { + "epoch": 1.591807599385021, + "grad_norm": 2.306680679321289, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7230985164642334, + "num_tokens": 372143848.0, + "step": 14495 + }, + { + "epoch": 1.5919174170876347, + "grad_norm": 2.1037824153900146, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.752199649810791, + "num_tokens": 372172004.0, + "step": 14496 + }, + { + "epoch": 1.5920272347902482, + "grad_norm": 1.8338747024536133, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7409005761146545, + "num_tokens": 372204489.0, + "step": 14497 + }, + { + "epoch": 1.5921370524928617, + "grad_norm": 1.891337513923645, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6963093280792236, + "num_tokens": 372240055.0, + "step": 14498 + }, + { + "epoch": 1.5922468701954755, + "grad_norm": 2.0582783222198486, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7049976587295532, + "num_tokens": 372271219.0, + "step": 14499 + }, + { + "epoch": 1.5923566878980893, + "grad_norm": 2.283259630203247, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7434738874435425, + "num_tokens": 372296467.0, + "step": 14500 + }, + { + "epoch": 1.5924665056007028, + "grad_norm": 2.7239696979522705, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7552748322486877, + "num_tokens": 372314479.0, + "step": 14501 + }, + { + "epoch": 1.5925763233033163, + "grad_norm": 2.2184383869171143, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7183889150619507, + "num_tokens": 372344350.0, + "step": 14502 + }, + { + "epoch": 1.59268614100593, + "grad_norm": 2.369210958480835, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7103945016860962, + "num_tokens": 372370994.0, + "step": 14503 + }, + { + "epoch": 1.5927959587085438, + "grad_norm": 2.4285998344421387, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6963088512420654, + "num_tokens": 372395045.0, + "step": 14504 + }, + { + "epoch": 1.5929057764111576, + "grad_norm": 2.5290467739105225, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7167907953262329, + "num_tokens": 372414926.0, + "step": 14505 + }, + { + "epoch": 1.5930155941137711, + "grad_norm": 2.388141393661499, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7213378548622131, + "num_tokens": 372438729.0, + "step": 14506 + }, + { + "epoch": 1.5931254118163847, + "grad_norm": 2.1021225452423096, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.724602222442627, + "num_tokens": 372466433.0, + "step": 14507 + }, + { + "epoch": 1.5932352295189984, + "grad_norm": 2.0335323810577393, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7147908806800842, + "num_tokens": 372495760.0, + "step": 14508 + }, + { + "epoch": 1.5933450472216122, + "grad_norm": 2.446951389312744, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7212798595428467, + "num_tokens": 372516686.0, + "step": 14509 + }, + { + "epoch": 1.593454864924226, + "grad_norm": 2.2791850566864014, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.717354953289032, + "num_tokens": 372541324.0, + "step": 14510 + }, + { + "epoch": 1.5935646826268395, + "grad_norm": 2.1387431621551514, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7128657698631287, + "num_tokens": 372568458.0, + "step": 14511 + }, + { + "epoch": 1.593674500329453, + "grad_norm": 2.2066516876220703, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7265956401824951, + "num_tokens": 372594502.0, + "step": 14512 + }, + { + "epoch": 1.5937843180320668, + "grad_norm": 2.4015119075775146, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7099260091781616, + "num_tokens": 372616916.0, + "step": 14513 + }, + { + "epoch": 1.5938941357346805, + "grad_norm": 2.3455379009246826, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7065645456314087, + "num_tokens": 372642102.0, + "step": 14514 + }, + { + "epoch": 1.594003953437294, + "grad_norm": 2.2311952114105225, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7137671709060669, + "num_tokens": 372668124.0, + "step": 14515 + }, + { + "epoch": 1.5941137711399076, + "grad_norm": 2.6069254875183105, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7270834445953369, + "num_tokens": 372687509.0, + "step": 14516 + }, + { + "epoch": 1.5942235888425214, + "grad_norm": 2.3618013858795166, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7095680832862854, + "num_tokens": 372711679.0, + "step": 14517 + }, + { + "epoch": 1.5943334065451351, + "grad_norm": 2.077521324157715, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6946678757667542, + "num_tokens": 372742907.0, + "step": 14518 + }, + { + "epoch": 1.5944432242477489, + "grad_norm": 2.2377212047576904, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7382150292396545, + "num_tokens": 372767293.0, + "step": 14519 + }, + { + "epoch": 1.5945530419503624, + "grad_norm": 2.2891905307769775, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7240996360778809, + "num_tokens": 372790802.0, + "step": 14520 + }, + { + "epoch": 1.594662859652976, + "grad_norm": 2.2909860610961914, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7255911231040955, + "num_tokens": 372816437.0, + "step": 14521 + }, + { + "epoch": 1.5947726773555897, + "grad_norm": 2.3701436519622803, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7106732130050659, + "num_tokens": 372840096.0, + "step": 14522 + }, + { + "epoch": 1.5948824950582035, + "grad_norm": 2.1133546829223633, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.689386248588562, + "num_tokens": 372869614.0, + "step": 14523 + }, + { + "epoch": 1.5949923127608172, + "grad_norm": 2.0640785694122314, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.733668327331543, + "num_tokens": 372899050.0, + "step": 14524 + }, + { + "epoch": 1.5951021304634307, + "grad_norm": 2.324950933456421, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7121204733848572, + "num_tokens": 372922999.0, + "step": 14525 + }, + { + "epoch": 1.5952119481660443, + "grad_norm": 2.261011838912964, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7021375894546509, + "num_tokens": 372948020.0, + "step": 14526 + }, + { + "epoch": 1.595321765868658, + "grad_norm": 2.1455495357513428, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7095152139663696, + "num_tokens": 372975759.0, + "step": 14527 + }, + { + "epoch": 1.5954315835712718, + "grad_norm": 2.2035183906555176, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7255042791366577, + "num_tokens": 373002640.0, + "step": 14528 + }, + { + "epoch": 1.5955414012738853, + "grad_norm": 2.6165566444396973, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7224621176719666, + "num_tokens": 373024250.0, + "step": 14529 + }, + { + "epoch": 1.5956512189764989, + "grad_norm": 2.105394124984741, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.72062748670578, + "num_tokens": 373052926.0, + "step": 14530 + }, + { + "epoch": 1.5957610366791126, + "grad_norm": 2.4708211421966553, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7278069853782654, + "num_tokens": 373075880.0, + "step": 14531 + }, + { + "epoch": 1.5958708543817264, + "grad_norm": 2.3927600383758545, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7617470026016235, + "num_tokens": 373096665.0, + "step": 14532 + }, + { + "epoch": 1.5959806720843401, + "grad_norm": 2.0979549884796143, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7075504064559937, + "num_tokens": 373127861.0, + "step": 14533 + }, + { + "epoch": 1.5960904897869537, + "grad_norm": 2.3778560161590576, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7328299283981323, + "num_tokens": 373149959.0, + "step": 14534 + }, + { + "epoch": 1.5962003074895672, + "grad_norm": 2.39858078956604, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.701951801776886, + "num_tokens": 373174065.0, + "step": 14535 + }, + { + "epoch": 1.596310125192181, + "grad_norm": 2.2601683139801025, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7172674536705017, + "num_tokens": 373198774.0, + "step": 14536 + }, + { + "epoch": 1.5964199428947947, + "grad_norm": 2.2810049057006836, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7006583213806152, + "num_tokens": 373225384.0, + "step": 14537 + }, + { + "epoch": 1.5965297605974083, + "grad_norm": 2.2062911987304688, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7374658584594727, + "num_tokens": 373251122.0, + "step": 14538 + }, + { + "epoch": 1.596639578300022, + "grad_norm": 2.1841604709625244, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7241486310958862, + "num_tokens": 373278345.0, + "step": 14539 + }, + { + "epoch": 1.5967493960026355, + "grad_norm": 2.1755177974700928, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7138466238975525, + "num_tokens": 373305816.0, + "step": 14540 + }, + { + "epoch": 1.5968592137052493, + "grad_norm": 2.1536014080047607, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7109567523002625, + "num_tokens": 373332862.0, + "step": 14541 + }, + { + "epoch": 1.596969031407863, + "grad_norm": 2.51872181892395, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7224130630493164, + "num_tokens": 373353613.0, + "step": 14542 + }, + { + "epoch": 1.5970788491104766, + "grad_norm": 2.358319044113159, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7323864698410034, + "num_tokens": 373374866.0, + "step": 14543 + }, + { + "epoch": 1.5971886668130901, + "grad_norm": 2.341618061065674, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7253250479698181, + "num_tokens": 373397120.0, + "step": 14544 + }, + { + "epoch": 1.5972984845157039, + "grad_norm": 2.493492841720581, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7241449356079102, + "num_tokens": 373417712.0, + "step": 14545 + }, + { + "epoch": 1.5974083022183176, + "grad_norm": 2.1329963207244873, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7040773630142212, + "num_tokens": 373445629.0, + "step": 14546 + }, + { + "epoch": 1.5975181199209314, + "grad_norm": 2.3250479698181152, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7135949730873108, + "num_tokens": 373470057.0, + "step": 14547 + }, + { + "epoch": 1.597627937623545, + "grad_norm": 2.096057176589966, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7388448119163513, + "num_tokens": 373498535.0, + "step": 14548 + }, + { + "epoch": 1.5977377553261585, + "grad_norm": 2.203641891479492, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7320877313613892, + "num_tokens": 373524427.0, + "step": 14549 + }, + { + "epoch": 1.5978475730287722, + "grad_norm": 2.069633722305298, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.700052797794342, + "num_tokens": 373553940.0, + "step": 14550 + }, + { + "epoch": 1.597957390731386, + "grad_norm": 2.3091604709625244, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7406011819839478, + "num_tokens": 373576471.0, + "step": 14551 + }, + { + "epoch": 1.5980672084339995, + "grad_norm": 2.3315041065216064, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197275161743164, + "num_tokens": 373601415.0, + "step": 14552 + }, + { + "epoch": 1.5981770261366133, + "grad_norm": 2.273005962371826, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.717850923538208, + "num_tokens": 373626247.0, + "step": 14553 + }, + { + "epoch": 1.5982868438392268, + "grad_norm": 2.199723958969116, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7300248146057129, + "num_tokens": 373652962.0, + "step": 14554 + }, + { + "epoch": 1.5983966615418406, + "grad_norm": 2.7957570552825928, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7463521957397461, + "num_tokens": 373671161.0, + "step": 14555 + }, + { + "epoch": 1.5985064792444543, + "grad_norm": 2.70025897026062, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7200581431388855, + "num_tokens": 373690860.0, + "step": 14556 + }, + { + "epoch": 1.5986162969470679, + "grad_norm": 2.1719415187835693, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.709691047668457, + "num_tokens": 373720500.0, + "step": 14557 + }, + { + "epoch": 1.5987261146496814, + "grad_norm": 2.7284657955169678, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.722187876701355, + "num_tokens": 373737788.0, + "step": 14558 + }, + { + "epoch": 1.5988359323522952, + "grad_norm": 2.2229909896850586, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7261168360710144, + "num_tokens": 373764077.0, + "step": 14559 + }, + { + "epoch": 1.598945750054909, + "grad_norm": 2.2889530658721924, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7254682779312134, + "num_tokens": 373788824.0, + "step": 14560 + }, + { + "epoch": 1.5990555677575227, + "grad_norm": 2.244208574295044, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7171245813369751, + "num_tokens": 373815194.0, + "step": 14561 + }, + { + "epoch": 1.5991653854601362, + "grad_norm": 2.3469481468200684, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.717208743095398, + "num_tokens": 373840980.0, + "step": 14562 + }, + { + "epoch": 1.5992752031627497, + "grad_norm": 2.4741146564483643, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7243868708610535, + "num_tokens": 373862591.0, + "step": 14563 + }, + { + "epoch": 1.5993850208653635, + "grad_norm": 2.1619675159454346, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7123009562492371, + "num_tokens": 373890097.0, + "step": 14564 + }, + { + "epoch": 1.5994948385679773, + "grad_norm": 2.252779960632324, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7236894369125366, + "num_tokens": 373917042.0, + "step": 14565 + }, + { + "epoch": 1.5996046562705908, + "grad_norm": 2.1538515090942383, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7296603918075562, + "num_tokens": 373942184.0, + "step": 14566 + }, + { + "epoch": 1.5997144739732043, + "grad_norm": 2.223738670349121, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7115840911865234, + "num_tokens": 373969195.0, + "step": 14567 + }, + { + "epoch": 1.599824291675818, + "grad_norm": 2.197272777557373, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7131115198135376, + "num_tokens": 373996248.0, + "step": 14568 + }, + { + "epoch": 1.5999341093784318, + "grad_norm": 2.1579127311706543, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7341206669807434, + "num_tokens": 374021835.0, + "step": 14569 + }, + { + "epoch": 1.6000439270810456, + "grad_norm": 2.2195918560028076, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7109889984130859, + "num_tokens": 374048390.0, + "step": 14570 + }, + { + "epoch": 1.6001537447836591, + "grad_norm": 2.2210264205932617, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7190731167793274, + "num_tokens": 374073369.0, + "step": 14571 + }, + { + "epoch": 1.6002635624862727, + "grad_norm": 1.9950608015060425, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.72348552942276, + "num_tokens": 374103470.0, + "step": 14572 + }, + { + "epoch": 1.6003733801888864, + "grad_norm": 2.3795576095581055, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6999127864837646, + "num_tokens": 374128807.0, + "step": 14573 + }, + { + "epoch": 1.6004831978915002, + "grad_norm": 2.1768076419830322, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7397640943527222, + "num_tokens": 374155662.0, + "step": 14574 + }, + { + "epoch": 1.600593015594114, + "grad_norm": 2.6000235080718994, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7313162684440613, + "num_tokens": 374176009.0, + "step": 14575 + }, + { + "epoch": 1.6007028332967275, + "grad_norm": 2.150670289993286, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.707318902015686, + "num_tokens": 374202362.0, + "step": 14576 + }, + { + "epoch": 1.600812650999341, + "grad_norm": 2.334428548812866, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7042577862739563, + "num_tokens": 374225606.0, + "step": 14577 + }, + { + "epoch": 1.6009224687019548, + "grad_norm": 2.6868855953216553, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.751696765422821, + "num_tokens": 374242581.0, + "step": 14578 + }, + { + "epoch": 1.6010322864045685, + "grad_norm": 2.6141719818115234, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.736130952835083, + "num_tokens": 374262663.0, + "step": 14579 + }, + { + "epoch": 1.601142104107182, + "grad_norm": 2.0742475986480713, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7214196920394897, + "num_tokens": 374292127.0, + "step": 14580 + }, + { + "epoch": 1.6012519218097956, + "grad_norm": 1.863296389579773, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7326973676681519, + "num_tokens": 374327356.0, + "step": 14581 + }, + { + "epoch": 1.6013617395124093, + "grad_norm": 2.235900402069092, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7092069387435913, + "num_tokens": 374356169.0, + "step": 14582 + }, + { + "epoch": 1.601471557215023, + "grad_norm": 2.49613094329834, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7192944288253784, + "num_tokens": 374378052.0, + "step": 14583 + }, + { + "epoch": 1.6015813749176369, + "grad_norm": 2.220921516418457, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6911493539810181, + "num_tokens": 374407980.0, + "step": 14584 + }, + { + "epoch": 1.6016911926202504, + "grad_norm": 2.2979395389556885, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7306517362594604, + "num_tokens": 374431756.0, + "step": 14585 + }, + { + "epoch": 1.601801010322864, + "grad_norm": 2.304206371307373, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.715013861656189, + "num_tokens": 374457817.0, + "step": 14586 + }, + { + "epoch": 1.6019108280254777, + "grad_norm": 2.2250783443450928, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6946463584899902, + "num_tokens": 374483828.0, + "step": 14587 + }, + { + "epoch": 1.6020206457280914, + "grad_norm": 2.4444780349731445, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7125570774078369, + "num_tokens": 374506135.0, + "step": 14588 + }, + { + "epoch": 1.602130463430705, + "grad_norm": 2.240184783935547, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6890676617622375, + "num_tokens": 374533149.0, + "step": 14589 + }, + { + "epoch": 1.6022402811333187, + "grad_norm": 2.01587176322937, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7161154747009277, + "num_tokens": 374562336.0, + "step": 14590 + }, + { + "epoch": 1.6023500988359323, + "grad_norm": 2.032158136367798, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6973934769630432, + "num_tokens": 374593420.0, + "step": 14591 + }, + { + "epoch": 1.602459916538546, + "grad_norm": 2.8067712783813477, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7274762392044067, + "num_tokens": 374612258.0, + "step": 14592 + }, + { + "epoch": 1.6025697342411598, + "grad_norm": 2.20611572265625, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.71003258228302, + "num_tokens": 374641049.0, + "step": 14593 + }, + { + "epoch": 1.6026795519437733, + "grad_norm": 2.1225602626800537, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7227948307991028, + "num_tokens": 374669652.0, + "step": 14594 + }, + { + "epoch": 1.6027893696463869, + "grad_norm": 2.200450897216797, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6938035488128662, + "num_tokens": 374698694.0, + "step": 14595 + }, + { + "epoch": 1.6028991873490006, + "grad_norm": 2.074613332748413, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7307132482528687, + "num_tokens": 374725898.0, + "step": 14596 + }, + { + "epoch": 1.6030090050516144, + "grad_norm": 2.0544042587280273, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7254932522773743, + "num_tokens": 374754544.0, + "step": 14597 + }, + { + "epoch": 1.6031188227542281, + "grad_norm": 2.1222383975982666, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7159387469291687, + "num_tokens": 374780362.0, + "step": 14598 + }, + { + "epoch": 1.6032286404568417, + "grad_norm": 2.268908739089966, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7084215879440308, + "num_tokens": 374806469.0, + "step": 14599 + }, + { + "epoch": 1.6033384581594552, + "grad_norm": 2.0967047214508057, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.703444242477417, + "num_tokens": 374837485.0, + "step": 14600 + }, + { + "epoch": 1.603448275862069, + "grad_norm": 2.071218252182007, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.724807620048523, + "num_tokens": 374864665.0, + "step": 14601 + }, + { + "epoch": 1.6035580935646827, + "grad_norm": 2.3485541343688965, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7557958364486694, + "num_tokens": 374888198.0, + "step": 14602 + }, + { + "epoch": 1.6036679112672962, + "grad_norm": 2.1667850017547607, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7274841070175171, + "num_tokens": 374914749.0, + "step": 14603 + }, + { + "epoch": 1.60377772896991, + "grad_norm": 2.2828001976013184, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.731881856918335, + "num_tokens": 374938172.0, + "step": 14604 + }, + { + "epoch": 1.6038875466725235, + "grad_norm": 2.3422305583953857, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7481477856636047, + "num_tokens": 374963116.0, + "step": 14605 + }, + { + "epoch": 1.6039973643751373, + "grad_norm": 2.109793186187744, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7246625423431396, + "num_tokens": 374990345.0, + "step": 14606 + }, + { + "epoch": 1.604107182077751, + "grad_norm": 2.2032952308654785, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7177731990814209, + "num_tokens": 375017388.0, + "step": 14607 + }, + { + "epoch": 1.6042169997803646, + "grad_norm": 2.318711280822754, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7368184328079224, + "num_tokens": 375040943.0, + "step": 14608 + }, + { + "epoch": 1.6043268174829781, + "grad_norm": 2.5077526569366455, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7590186595916748, + "num_tokens": 375060554.0, + "step": 14609 + }, + { + "epoch": 1.6044366351855919, + "grad_norm": 1.9799193143844604, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7283723950386047, + "num_tokens": 375093109.0, + "step": 14610 + }, + { + "epoch": 1.6045464528882056, + "grad_norm": 2.159264087677002, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7024773359298706, + "num_tokens": 375122639.0, + "step": 14611 + }, + { + "epoch": 1.6046562705908194, + "grad_norm": 2.1452224254608154, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7023810148239136, + "num_tokens": 375153916.0, + "step": 14612 + }, + { + "epoch": 1.604766088293433, + "grad_norm": 2.271050214767456, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.733953058719635, + "num_tokens": 375178281.0, + "step": 14613 + }, + { + "epoch": 1.6048759059960465, + "grad_norm": 2.2246904373168945, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7267879247665405, + "num_tokens": 375203124.0, + "step": 14614 + }, + { + "epoch": 1.6049857236986602, + "grad_norm": 2.0832571983337402, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7153271436691284, + "num_tokens": 375232537.0, + "step": 14615 + }, + { + "epoch": 1.605095541401274, + "grad_norm": 2.2093393802642822, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7129430174827576, + "num_tokens": 375258463.0, + "step": 14616 + }, + { + "epoch": 1.6052053591038875, + "grad_norm": 2.201197862625122, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.708733856678009, + "num_tokens": 375284387.0, + "step": 14617 + }, + { + "epoch": 1.605315176806501, + "grad_norm": 1.981184959411621, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7180039882659912, + "num_tokens": 375315149.0, + "step": 14618 + }, + { + "epoch": 1.6054249945091148, + "grad_norm": 2.3457038402557373, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7277631759643555, + "num_tokens": 375337446.0, + "step": 14619 + }, + { + "epoch": 1.6055348122117286, + "grad_norm": 1.9993622303009033, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7011513710021973, + "num_tokens": 375367248.0, + "step": 14620 + }, + { + "epoch": 1.6056446299143423, + "grad_norm": 2.15453839302063, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7157390117645264, + "num_tokens": 375394529.0, + "step": 14621 + }, + { + "epoch": 1.6057544476169558, + "grad_norm": 2.252021312713623, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7164409756660461, + "num_tokens": 375422239.0, + "step": 14622 + }, + { + "epoch": 1.6058642653195694, + "grad_norm": 2.041043758392334, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7417124509811401, + "num_tokens": 375449873.0, + "step": 14623 + }, + { + "epoch": 1.6059740830221831, + "grad_norm": 2.1727521419525146, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7045380473136902, + "num_tokens": 375476733.0, + "step": 14624 + }, + { + "epoch": 1.606083900724797, + "grad_norm": 2.6696388721466064, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7339931130409241, + "num_tokens": 375496400.0, + "step": 14625 + }, + { + "epoch": 1.6061937184274107, + "grad_norm": 2.595686197280884, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7326939105987549, + "num_tokens": 375515786.0, + "step": 14626 + }, + { + "epoch": 1.6063035361300242, + "grad_norm": 2.5496997833251953, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7458102703094482, + "num_tokens": 375535596.0, + "step": 14627 + }, + { + "epoch": 1.6064133538326377, + "grad_norm": 2.1466012001037598, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.713119626045227, + "num_tokens": 375563186.0, + "step": 14628 + }, + { + "epoch": 1.6065231715352515, + "grad_norm": 2.3998701572418213, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7204664945602417, + "num_tokens": 375586914.0, + "step": 14629 + }, + { + "epoch": 1.6066329892378652, + "grad_norm": 2.277149200439453, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7086347341537476, + "num_tokens": 375614174.0, + "step": 14630 + }, + { + "epoch": 1.6067428069404788, + "grad_norm": 2.1658895015716553, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6987606287002563, + "num_tokens": 375644423.0, + "step": 14631 + }, + { + "epoch": 1.6068526246430923, + "grad_norm": 2.2405242919921875, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7197493314743042, + "num_tokens": 375669417.0, + "step": 14632 + }, + { + "epoch": 1.606962442345706, + "grad_norm": 2.3988568782806396, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7124701738357544, + "num_tokens": 375692349.0, + "step": 14633 + }, + { + "epoch": 1.6070722600483198, + "grad_norm": 2.252643585205078, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7163968682289124, + "num_tokens": 375718292.0, + "step": 14634 + }, + { + "epoch": 1.6071820777509336, + "grad_norm": 2.005274772644043, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7150853872299194, + "num_tokens": 375750615.0, + "step": 14635 + }, + { + "epoch": 1.6072918954535471, + "grad_norm": 2.2873494625091553, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7088078856468201, + "num_tokens": 375776763.0, + "step": 14636 + }, + { + "epoch": 1.6074017131561606, + "grad_norm": 2.177129030227661, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7215288281440735, + "num_tokens": 375802965.0, + "step": 14637 + }, + { + "epoch": 1.6075115308587744, + "grad_norm": 2.1093292236328125, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7313278913497925, + "num_tokens": 375832129.0, + "step": 14638 + }, + { + "epoch": 1.6076213485613882, + "grad_norm": 2.0627243518829346, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7096078395843506, + "num_tokens": 375862136.0, + "step": 14639 + }, + { + "epoch": 1.607731166264002, + "grad_norm": 2.3475966453552246, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7135735750198364, + "num_tokens": 375886374.0, + "step": 14640 + }, + { + "epoch": 1.6078409839666155, + "grad_norm": 2.4049808979034424, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7339537143707275, + "num_tokens": 375907966.0, + "step": 14641 + }, + { + "epoch": 1.607950801669229, + "grad_norm": 1.9177360534667969, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7293368577957153, + "num_tokens": 375939742.0, + "step": 14642 + }, + { + "epoch": 1.6080606193718427, + "grad_norm": 1.9850049018859863, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7176933288574219, + "num_tokens": 375971631.0, + "step": 14643 + }, + { + "epoch": 1.6081704370744565, + "grad_norm": 2.235556125640869, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7222919464111328, + "num_tokens": 375997909.0, + "step": 14644 + }, + { + "epoch": 1.60828025477707, + "grad_norm": 2.367626428604126, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6994026899337769, + "num_tokens": 376024271.0, + "step": 14645 + }, + { + "epoch": 1.6083900724796836, + "grad_norm": 2.306810140609741, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7298576831817627, + "num_tokens": 376047943.0, + "step": 14646 + }, + { + "epoch": 1.6084998901822973, + "grad_norm": 2.271768569946289, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7255657911300659, + "num_tokens": 376072680.0, + "step": 14647 + }, + { + "epoch": 1.608609707884911, + "grad_norm": 2.2464683055877686, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7319991588592529, + "num_tokens": 376096772.0, + "step": 14648 + }, + { + "epoch": 1.6087195255875248, + "grad_norm": 2.1666853427886963, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7206394076347351, + "num_tokens": 376122837.0, + "step": 14649 + }, + { + "epoch": 1.6088293432901384, + "grad_norm": 2.2670013904571533, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7146469354629517, + "num_tokens": 376149150.0, + "step": 14650 + }, + { + "epoch": 1.608939160992752, + "grad_norm": 2.3359508514404297, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7173862457275391, + "num_tokens": 376171680.0, + "step": 14651 + }, + { + "epoch": 1.6090489786953657, + "grad_norm": 2.391615629196167, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7226493954658508, + "num_tokens": 376195570.0, + "step": 14652 + }, + { + "epoch": 1.6091587963979794, + "grad_norm": 2.6434924602508545, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7123706340789795, + "num_tokens": 376216604.0, + "step": 14653 + }, + { + "epoch": 1.609268614100593, + "grad_norm": 2.332951068878174, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7144037485122681, + "num_tokens": 376240373.0, + "step": 14654 + }, + { + "epoch": 1.6093784318032067, + "grad_norm": 2.109652042388916, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6947731375694275, + "num_tokens": 376270472.0, + "step": 14655 + }, + { + "epoch": 1.6094882495058203, + "grad_norm": 2.221374273300171, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.713948130607605, + "num_tokens": 376296173.0, + "step": 14656 + }, + { + "epoch": 1.609598067208434, + "grad_norm": 2.2254981994628906, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7078043222427368, + "num_tokens": 376323795.0, + "step": 14657 + }, + { + "epoch": 1.6097078849110478, + "grad_norm": 2.5348620414733887, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7400741577148438, + "num_tokens": 376344024.0, + "step": 14658 + }, + { + "epoch": 1.6098177026136613, + "grad_norm": 2.324465036392212, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7235229015350342, + "num_tokens": 376367090.0, + "step": 14659 + }, + { + "epoch": 1.6099275203162748, + "grad_norm": 2.280775547027588, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7199674844741821, + "num_tokens": 376391822.0, + "step": 14660 + }, + { + "epoch": 1.6100373380188886, + "grad_norm": 1.9625300168991089, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7319834232330322, + "num_tokens": 376423498.0, + "step": 14661 + }, + { + "epoch": 1.6101471557215024, + "grad_norm": 2.2952325344085693, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7422336339950562, + "num_tokens": 376449407.0, + "step": 14662 + }, + { + "epoch": 1.6102569734241161, + "grad_norm": 2.115182399749756, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7064294219017029, + "num_tokens": 376480239.0, + "step": 14663 + }, + { + "epoch": 1.6103667911267296, + "grad_norm": 2.1011600494384766, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7042681574821472, + "num_tokens": 376509504.0, + "step": 14664 + }, + { + "epoch": 1.6104766088293432, + "grad_norm": 2.6758220195770264, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7214657664299011, + "num_tokens": 376529245.0, + "step": 14665 + }, + { + "epoch": 1.610586426531957, + "grad_norm": 2.7580513954162598, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7159378528594971, + "num_tokens": 376547553.0, + "step": 14666 + }, + { + "epoch": 1.6106962442345707, + "grad_norm": 2.155087947845459, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7420498132705688, + "num_tokens": 376575816.0, + "step": 14667 + }, + { + "epoch": 1.6108060619371842, + "grad_norm": 2.2766830921173096, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7208781242370605, + "num_tokens": 376602186.0, + "step": 14668 + }, + { + "epoch": 1.610915879639798, + "grad_norm": 2.093271255493164, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7014913558959961, + "num_tokens": 376631373.0, + "step": 14669 + }, + { + "epoch": 1.6110256973424115, + "grad_norm": 2.1477270126342773, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7132424712181091, + "num_tokens": 376658692.0, + "step": 14670 + }, + { + "epoch": 1.6111355150450253, + "grad_norm": 2.271489381790161, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7082023620605469, + "num_tokens": 376685050.0, + "step": 14671 + }, + { + "epoch": 1.611245332747639, + "grad_norm": 1.969899296760559, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6974790096282959, + "num_tokens": 376716454.0, + "step": 14672 + }, + { + "epoch": 1.6113551504502526, + "grad_norm": 2.2623209953308105, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7002396583557129, + "num_tokens": 376743590.0, + "step": 14673 + }, + { + "epoch": 1.611464968152866, + "grad_norm": 2.1581308841705322, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7117727398872375, + "num_tokens": 376771826.0, + "step": 14674 + }, + { + "epoch": 1.6115747858554799, + "grad_norm": 2.254823684692383, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7304076552391052, + "num_tokens": 376798109.0, + "step": 14675 + }, + { + "epoch": 1.6116846035580936, + "grad_norm": 2.412381172180176, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7234750986099243, + "num_tokens": 376819807.0, + "step": 14676 + }, + { + "epoch": 1.6117944212607074, + "grad_norm": 1.9933098554611206, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7130692601203918, + "num_tokens": 376850242.0, + "step": 14677 + }, + { + "epoch": 1.611904238963321, + "grad_norm": 2.420257806777954, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7078858613967896, + "num_tokens": 376873387.0, + "step": 14678 + }, + { + "epoch": 1.6120140566659344, + "grad_norm": 2.2052133083343506, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7346500158309937, + "num_tokens": 376898883.0, + "step": 14679 + }, + { + "epoch": 1.6121238743685482, + "grad_norm": 2.624551296234131, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7444556355476379, + "num_tokens": 376917361.0, + "step": 14680 + }, + { + "epoch": 1.612233692071162, + "grad_norm": 2.491210460662842, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7035599946975708, + "num_tokens": 376938422.0, + "step": 14681 + }, + { + "epoch": 1.6123435097737755, + "grad_norm": 2.5441272258758545, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7352978587150574, + "num_tokens": 376958276.0, + "step": 14682 + }, + { + "epoch": 1.612453327476389, + "grad_norm": 2.2084901332855225, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7318658828735352, + "num_tokens": 376987942.0, + "step": 14683 + }, + { + "epoch": 1.6125631451790028, + "grad_norm": 2.272991418838501, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7153007984161377, + "num_tokens": 377014375.0, + "step": 14684 + }, + { + "epoch": 1.6126729628816165, + "grad_norm": 2.2615976333618164, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7305002212524414, + "num_tokens": 377040379.0, + "step": 14685 + }, + { + "epoch": 1.6127827805842303, + "grad_norm": 2.3428521156311035, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.729911744594574, + "num_tokens": 377064979.0, + "step": 14686 + }, + { + "epoch": 1.6128925982868438, + "grad_norm": 2.1153547763824463, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7184274196624756, + "num_tokens": 377094786.0, + "step": 14687 + }, + { + "epoch": 1.6130024159894574, + "grad_norm": 2.3411238193511963, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7185887098312378, + "num_tokens": 377118827.0, + "step": 14688 + }, + { + "epoch": 1.6131122336920711, + "grad_norm": 2.4385998249053955, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7251203656196594, + "num_tokens": 377140554.0, + "step": 14689 + }, + { + "epoch": 1.6132220513946849, + "grad_norm": 2.5106098651885986, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7124704122543335, + "num_tokens": 377161532.0, + "step": 14690 + }, + { + "epoch": 1.6133318690972986, + "grad_norm": 2.4507408142089844, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7274826169013977, + "num_tokens": 377183515.0, + "step": 14691 + }, + { + "epoch": 1.6134416867999122, + "grad_norm": 2.2808165550231934, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7130975723266602, + "num_tokens": 377209007.0, + "step": 14692 + }, + { + "epoch": 1.6135515045025257, + "grad_norm": 2.3170013427734375, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7150715589523315, + "num_tokens": 377232859.0, + "step": 14693 + }, + { + "epoch": 1.6136613222051395, + "grad_norm": 2.3055295944213867, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7166286110877991, + "num_tokens": 377258221.0, + "step": 14694 + }, + { + "epoch": 1.6137711399077532, + "grad_norm": 2.6073198318481445, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7232077121734619, + "num_tokens": 377278426.0, + "step": 14695 + }, + { + "epoch": 1.6138809576103668, + "grad_norm": 2.3838703632354736, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7222399711608887, + "num_tokens": 377300300.0, + "step": 14696 + }, + { + "epoch": 1.6139907753129803, + "grad_norm": 2.172001600265503, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7195324897766113, + "num_tokens": 377327331.0, + "step": 14697 + }, + { + "epoch": 1.614100593015594, + "grad_norm": 2.248025417327881, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7079331278800964, + "num_tokens": 377352428.0, + "step": 14698 + }, + { + "epoch": 1.6142104107182078, + "grad_norm": 2.715378522872925, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7290439605712891, + "num_tokens": 377369926.0, + "step": 14699 + }, + { + "epoch": 1.6143202284208216, + "grad_norm": 2.288771390914917, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7209960222244263, + "num_tokens": 377397051.0, + "step": 14700 + }, + { + "epoch": 1.614430046123435, + "grad_norm": 2.2901406288146973, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.709969162940979, + "num_tokens": 377423000.0, + "step": 14701 + }, + { + "epoch": 1.6145398638260486, + "grad_norm": 2.2998087406158447, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.718124508857727, + "num_tokens": 377448721.0, + "step": 14702 + }, + { + "epoch": 1.6146496815286624, + "grad_norm": 2.337637186050415, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7188259363174438, + "num_tokens": 377473035.0, + "step": 14703 + }, + { + "epoch": 1.6147594992312762, + "grad_norm": 2.4268438816070557, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7218531370162964, + "num_tokens": 377493636.0, + "step": 14704 + }, + { + "epoch": 1.61486931693389, + "grad_norm": 2.2971808910369873, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7098267674446106, + "num_tokens": 377520398.0, + "step": 14705 + }, + { + "epoch": 1.6149791346365034, + "grad_norm": 2.14595103263855, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7247872948646545, + "num_tokens": 377547519.0, + "step": 14706 + }, + { + "epoch": 1.615088952339117, + "grad_norm": 2.3727149963378906, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7202159762382507, + "num_tokens": 377569975.0, + "step": 14707 + }, + { + "epoch": 1.6151987700417307, + "grad_norm": 2.3438408374786377, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7100085616111755, + "num_tokens": 377594342.0, + "step": 14708 + }, + { + "epoch": 1.6153085877443445, + "grad_norm": 2.0511093139648438, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7389646768569946, + "num_tokens": 377624097.0, + "step": 14709 + }, + { + "epoch": 1.615418405446958, + "grad_norm": 1.9205721616744995, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7005841732025146, + "num_tokens": 377657660.0, + "step": 14710 + }, + { + "epoch": 1.6155282231495716, + "grad_norm": 2.0322792530059814, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6938763856887817, + "num_tokens": 377689364.0, + "step": 14711 + }, + { + "epoch": 1.6156380408521853, + "grad_norm": 2.215578317642212, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7191281318664551, + "num_tokens": 377714935.0, + "step": 14712 + }, + { + "epoch": 1.615747858554799, + "grad_norm": 2.1906237602233887, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.709949791431427, + "num_tokens": 377740097.0, + "step": 14713 + }, + { + "epoch": 1.6158576762574128, + "grad_norm": 2.2847044467926025, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7178261876106262, + "num_tokens": 377767765.0, + "step": 14714 + }, + { + "epoch": 1.6159674939600264, + "grad_norm": 2.8733692169189453, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7427612543106079, + "num_tokens": 377785068.0, + "step": 14715 + }, + { + "epoch": 1.61607731166264, + "grad_norm": 2.292745590209961, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7117207050323486, + "num_tokens": 377810065.0, + "step": 14716 + }, + { + "epoch": 1.6161871293652537, + "grad_norm": 2.3368711471557617, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7285813689231873, + "num_tokens": 377832510.0, + "step": 14717 + }, + { + "epoch": 1.6162969470678674, + "grad_norm": 2.201179027557373, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7371748685836792, + "num_tokens": 377856462.0, + "step": 14718 + }, + { + "epoch": 1.616406764770481, + "grad_norm": 2.132085084915161, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.714850902557373, + "num_tokens": 377884431.0, + "step": 14719 + }, + { + "epoch": 1.6165165824730947, + "grad_norm": 2.5350422859191895, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7209727764129639, + "num_tokens": 377907031.0, + "step": 14720 + }, + { + "epoch": 1.6166264001757082, + "grad_norm": 2.3573923110961914, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.710109531879425, + "num_tokens": 377931800.0, + "step": 14721 + }, + { + "epoch": 1.616736217878322, + "grad_norm": 2.4155149459838867, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7313763499259949, + "num_tokens": 377953444.0, + "step": 14722 + }, + { + "epoch": 1.6168460355809358, + "grad_norm": 2.111668348312378, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7299489378929138, + "num_tokens": 377981905.0, + "step": 14723 + }, + { + "epoch": 1.6169558532835493, + "grad_norm": 2.0716822147369385, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7093372344970703, + "num_tokens": 378012815.0, + "step": 14724 + }, + { + "epoch": 1.6170656709861628, + "grad_norm": 2.227902412414551, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6879751682281494, + "num_tokens": 378038839.0, + "step": 14725 + }, + { + "epoch": 1.6171754886887766, + "grad_norm": 1.9131652116775513, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7169413566589355, + "num_tokens": 378071363.0, + "step": 14726 + }, + { + "epoch": 1.6172853063913903, + "grad_norm": 2.3084025382995605, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7188428044319153, + "num_tokens": 378096204.0, + "step": 14727 + }, + { + "epoch": 1.617395124094004, + "grad_norm": 2.3652188777923584, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7071754932403564, + "num_tokens": 378120640.0, + "step": 14728 + }, + { + "epoch": 1.6175049417966176, + "grad_norm": 2.0297305583953857, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7137695550918579, + "num_tokens": 378152822.0, + "step": 14729 + }, + { + "epoch": 1.6176147594992312, + "grad_norm": 2.2595369815826416, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.705333948135376, + "num_tokens": 378181694.0, + "step": 14730 + }, + { + "epoch": 1.617724577201845, + "grad_norm": 2.124149799346924, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7000781297683716, + "num_tokens": 378210398.0, + "step": 14731 + }, + { + "epoch": 1.6178343949044587, + "grad_norm": 2.338860273361206, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7330424785614014, + "num_tokens": 378234110.0, + "step": 14732 + }, + { + "epoch": 1.6179442126070722, + "grad_norm": 1.8535312414169312, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6980143785476685, + "num_tokens": 378272384.0, + "step": 14733 + }, + { + "epoch": 1.618054030309686, + "grad_norm": 2.503293037414551, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7107325792312622, + "num_tokens": 378296477.0, + "step": 14734 + }, + { + "epoch": 1.6181638480122995, + "grad_norm": 2.2960000038146973, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7237404584884644, + "num_tokens": 378320082.0, + "step": 14735 + }, + { + "epoch": 1.6182736657149133, + "grad_norm": 2.6126179695129395, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7442334294319153, + "num_tokens": 378339988.0, + "step": 14736 + }, + { + "epoch": 1.618383483417527, + "grad_norm": 2.5148892402648926, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.7177325487136841, + "num_tokens": 378366082.0, + "step": 14737 + }, + { + "epoch": 1.6184933011201406, + "grad_norm": 2.316737413406372, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7192845344543457, + "num_tokens": 378392767.0, + "step": 14738 + }, + { + "epoch": 1.618603118822754, + "grad_norm": 2.208477020263672, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.701309323310852, + "num_tokens": 378419107.0, + "step": 14739 + }, + { + "epoch": 1.6187129365253679, + "grad_norm": 2.4624221324920654, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7550929188728333, + "num_tokens": 378439851.0, + "step": 14740 + }, + { + "epoch": 1.6188227542279816, + "grad_norm": 2.183990001678467, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7080804705619812, + "num_tokens": 378466815.0, + "step": 14741 + }, + { + "epoch": 1.6189325719305954, + "grad_norm": 2.054492950439453, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7094492316246033, + "num_tokens": 378498465.0, + "step": 14742 + }, + { + "epoch": 1.619042389633209, + "grad_norm": 1.954802393913269, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7239168882369995, + "num_tokens": 378530724.0, + "step": 14743 + }, + { + "epoch": 1.6191522073358224, + "grad_norm": 2.5391361713409424, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7451046705245972, + "num_tokens": 378550994.0, + "step": 14744 + }, + { + "epoch": 1.6192620250384362, + "grad_norm": 2.2017629146575928, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7292061448097229, + "num_tokens": 378576608.0, + "step": 14745 + }, + { + "epoch": 1.61937184274105, + "grad_norm": 2.200101137161255, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7164804935455322, + "num_tokens": 378602025.0, + "step": 14746 + }, + { + "epoch": 1.6194816604436635, + "grad_norm": 2.4623024463653564, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7243292927742004, + "num_tokens": 378624715.0, + "step": 14747 + }, + { + "epoch": 1.619591478146277, + "grad_norm": 2.128173351287842, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6953273415565491, + "num_tokens": 378652851.0, + "step": 14748 + }, + { + "epoch": 1.6197012958488908, + "grad_norm": 2.095970392227173, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7342570424079895, + "num_tokens": 378681988.0, + "step": 14749 + }, + { + "epoch": 1.6198111135515045, + "grad_norm": 2.309455156326294, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7362727522850037, + "num_tokens": 378705899.0, + "step": 14750 + }, + { + "epoch": 1.6199209312541183, + "grad_norm": 2.2560787200927734, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7522430419921875, + "num_tokens": 378727926.0, + "step": 14751 + }, + { + "epoch": 1.6200307489567318, + "grad_norm": 2.0963265895843506, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7249568104743958, + "num_tokens": 378755816.0, + "step": 14752 + }, + { + "epoch": 1.6201405666593454, + "grad_norm": 2.325716495513916, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7363177537918091, + "num_tokens": 378779189.0, + "step": 14753 + }, + { + "epoch": 1.6202503843619591, + "grad_norm": 2.328404426574707, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7380917072296143, + "num_tokens": 378800368.0, + "step": 14754 + }, + { + "epoch": 1.6203602020645729, + "grad_norm": 2.1471192836761475, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7240719795227051, + "num_tokens": 378828962.0, + "step": 14755 + }, + { + "epoch": 1.6204700197671866, + "grad_norm": 2.2779600620269775, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7182645797729492, + "num_tokens": 378853659.0, + "step": 14756 + }, + { + "epoch": 1.6205798374698002, + "grad_norm": 2.1171936988830566, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7141436338424683, + "num_tokens": 378882381.0, + "step": 14757 + }, + { + "epoch": 1.6206896551724137, + "grad_norm": 2.130392074584961, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7151218056678772, + "num_tokens": 378912215.0, + "step": 14758 + }, + { + "epoch": 1.6207994728750275, + "grad_norm": 2.5288846492767334, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.72254478931427, + "num_tokens": 378932896.0, + "step": 14759 + }, + { + "epoch": 1.6209092905776412, + "grad_norm": 2.150161027908325, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7365743517875671, + "num_tokens": 378961941.0, + "step": 14760 + }, + { + "epoch": 1.6210191082802548, + "grad_norm": 2.350027084350586, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7168727517127991, + "num_tokens": 378986647.0, + "step": 14761 + }, + { + "epoch": 1.6211289259828683, + "grad_norm": 2.197307825088501, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7213716506958008, + "num_tokens": 379014038.0, + "step": 14762 + }, + { + "epoch": 1.621238743685482, + "grad_norm": 2.067457914352417, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7061989307403564, + "num_tokens": 379046674.0, + "step": 14763 + }, + { + "epoch": 1.6213485613880958, + "grad_norm": 2.230933904647827, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7355586886405945, + "num_tokens": 379071922.0, + "step": 14764 + }, + { + "epoch": 1.6214583790907096, + "grad_norm": 2.123079776763916, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7231988906860352, + "num_tokens": 379099218.0, + "step": 14765 + }, + { + "epoch": 1.621568196793323, + "grad_norm": 2.46604061126709, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.712398886680603, + "num_tokens": 379121842.0, + "step": 14766 + }, + { + "epoch": 1.6216780144959366, + "grad_norm": 2.437377452850342, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7153496742248535, + "num_tokens": 379145148.0, + "step": 14767 + }, + { + "epoch": 1.6217878321985504, + "grad_norm": 2.4334981441497803, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7392950057983398, + "num_tokens": 379168358.0, + "step": 14768 + }, + { + "epoch": 1.6218976499011641, + "grad_norm": 2.229135036468506, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7146197557449341, + "num_tokens": 379194620.0, + "step": 14769 + }, + { + "epoch": 1.6220074676037777, + "grad_norm": 2.38360333442688, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7227111458778381, + "num_tokens": 379218805.0, + "step": 14770 + }, + { + "epoch": 1.6221172853063914, + "grad_norm": 2.0282838344573975, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6963664293289185, + "num_tokens": 379249937.0, + "step": 14771 + }, + { + "epoch": 1.622227103009005, + "grad_norm": 2.2151381969451904, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6921626329421997, + "num_tokens": 379278920.0, + "step": 14772 + }, + { + "epoch": 1.6223369207116187, + "grad_norm": 2.6339218616485596, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7115110754966736, + "num_tokens": 379299583.0, + "step": 14773 + }, + { + "epoch": 1.6224467384142325, + "grad_norm": 2.1560986042022705, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7091090083122253, + "num_tokens": 379326102.0, + "step": 14774 + }, + { + "epoch": 1.622556556116846, + "grad_norm": 2.105114459991455, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.740464448928833, + "num_tokens": 379352774.0, + "step": 14775 + }, + { + "epoch": 1.6226663738194596, + "grad_norm": 2.3539934158325195, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7337386012077332, + "num_tokens": 379375729.0, + "step": 14776 + }, + { + "epoch": 1.6227761915220733, + "grad_norm": 2.3400111198425293, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7242929935455322, + "num_tokens": 379399930.0, + "step": 14777 + }, + { + "epoch": 1.622886009224687, + "grad_norm": 2.292606830596924, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7006081938743591, + "num_tokens": 379426517.0, + "step": 14778 + }, + { + "epoch": 1.6229958269273008, + "grad_norm": 2.042388916015625, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7168934345245361, + "num_tokens": 379456971.0, + "step": 14779 + }, + { + "epoch": 1.6231056446299144, + "grad_norm": 2.2274739742279053, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7154309153556824, + "num_tokens": 379483775.0, + "step": 14780 + }, + { + "epoch": 1.623215462332528, + "grad_norm": 2.392578125, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7140964865684509, + "num_tokens": 379507680.0, + "step": 14781 + }, + { + "epoch": 1.6233252800351416, + "grad_norm": 2.0761327743530273, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.724496066570282, + "num_tokens": 379536390.0, + "step": 14782 + }, + { + "epoch": 1.6234350977377554, + "grad_norm": 2.2418768405914307, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7139559984207153, + "num_tokens": 379560635.0, + "step": 14783 + }, + { + "epoch": 1.623544915440369, + "grad_norm": 2.0934972763061523, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7278368473052979, + "num_tokens": 379589694.0, + "step": 14784 + }, + { + "epoch": 1.6236547331429827, + "grad_norm": 2.3310835361480713, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.71428382396698, + "num_tokens": 379613584.0, + "step": 14785 + }, + { + "epoch": 1.6237645508455962, + "grad_norm": 1.91183340549469, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.71564781665802, + "num_tokens": 379645453.0, + "step": 14786 + }, + { + "epoch": 1.62387436854821, + "grad_norm": 2.084073066711426, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7286858558654785, + "num_tokens": 379673349.0, + "step": 14787 + }, + { + "epoch": 1.6239841862508237, + "grad_norm": 2.306795835494995, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7344028949737549, + "num_tokens": 379697236.0, + "step": 14788 + }, + { + "epoch": 1.6240940039534373, + "grad_norm": 2.1882944107055664, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7109941840171814, + "num_tokens": 379724711.0, + "step": 14789 + }, + { + "epoch": 1.6242038216560508, + "grad_norm": 2.2753844261169434, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7083117365837097, + "num_tokens": 379752146.0, + "step": 14790 + }, + { + "epoch": 1.6243136393586646, + "grad_norm": 1.9361189603805542, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7002785801887512, + "num_tokens": 379786631.0, + "step": 14791 + }, + { + "epoch": 1.6244234570612783, + "grad_norm": 2.061307191848755, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.701198160648346, + "num_tokens": 379817986.0, + "step": 14792 + }, + { + "epoch": 1.624533274763892, + "grad_norm": 1.9831657409667969, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7155174612998962, + "num_tokens": 379848615.0, + "step": 14793 + }, + { + "epoch": 1.6246430924665056, + "grad_norm": 2.2517457008361816, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7346019744873047, + "num_tokens": 379873662.0, + "step": 14794 + }, + { + "epoch": 1.6247529101691192, + "grad_norm": 2.3269903659820557, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7219166159629822, + "num_tokens": 379897565.0, + "step": 14795 + }, + { + "epoch": 1.624862727871733, + "grad_norm": 2.117737054824829, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6972020268440247, + "num_tokens": 379927822.0, + "step": 14796 + }, + { + "epoch": 1.6249725455743467, + "grad_norm": 2.6011738777160645, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7541744709014893, + "num_tokens": 379946064.0, + "step": 14797 + }, + { + "epoch": 1.6250823632769602, + "grad_norm": 2.2629964351654053, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7056475877761841, + "num_tokens": 379971834.0, + "step": 14798 + }, + { + "epoch": 1.6251921809795737, + "grad_norm": 2.3199398517608643, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7353105545043945, + "num_tokens": 379995493.0, + "step": 14799 + }, + { + "epoch": 1.6253019986821875, + "grad_norm": 2.2336270809173584, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7229448556900024, + "num_tokens": 380022235.0, + "step": 14800 + }, + { + "epoch": 1.6254118163848013, + "grad_norm": 2.0368294715881348, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7125792503356934, + "num_tokens": 380054004.0, + "step": 14801 + }, + { + "epoch": 1.625521634087415, + "grad_norm": 2.176213502883911, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7268623113632202, + "num_tokens": 380079225.0, + "step": 14802 + }, + { + "epoch": 1.6256314517900285, + "grad_norm": 2.3183045387268066, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7113455533981323, + "num_tokens": 380103759.0, + "step": 14803 + }, + { + "epoch": 1.625741269492642, + "grad_norm": 2.3519551753997803, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7305812239646912, + "num_tokens": 380126358.0, + "step": 14804 + }, + { + "epoch": 1.6258510871952558, + "grad_norm": 2.1092002391815186, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7431140542030334, + "num_tokens": 380153411.0, + "step": 14805 + }, + { + "epoch": 1.6259609048978696, + "grad_norm": 1.9860910177230835, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7039182186126709, + "num_tokens": 380184820.0, + "step": 14806 + }, + { + "epoch": 1.6260707226004834, + "grad_norm": 2.037102699279785, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7207092046737671, + "num_tokens": 380215235.0, + "step": 14807 + }, + { + "epoch": 1.6261805403030969, + "grad_norm": 2.30710768699646, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7194536924362183, + "num_tokens": 380241907.0, + "step": 14808 + }, + { + "epoch": 1.6262903580057104, + "grad_norm": 2.19062876701355, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7182466983795166, + "num_tokens": 380268191.0, + "step": 14809 + }, + { + "epoch": 1.6264001757083242, + "grad_norm": 2.2200088500976562, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.743323802947998, + "num_tokens": 380292136.0, + "step": 14810 + }, + { + "epoch": 1.626509993410938, + "grad_norm": 2.573439598083496, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7129274606704712, + "num_tokens": 380312721.0, + "step": 14811 + }, + { + "epoch": 1.6266198111135515, + "grad_norm": 2.4566657543182373, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7154746055603027, + "num_tokens": 380334616.0, + "step": 14812 + }, + { + "epoch": 1.626729628816165, + "grad_norm": 2.358548641204834, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.740465521812439, + "num_tokens": 380359120.0, + "step": 14813 + }, + { + "epoch": 1.6268394465187788, + "grad_norm": 2.1105918884277344, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7203278541564941, + "num_tokens": 380389641.0, + "step": 14814 + }, + { + "epoch": 1.6269492642213925, + "grad_norm": 2.314192056655884, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7203139066696167, + "num_tokens": 380413421.0, + "step": 14815 + }, + { + "epoch": 1.6270590819240063, + "grad_norm": 2.1408307552337646, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7249103784561157, + "num_tokens": 380440112.0, + "step": 14816 + }, + { + "epoch": 1.6271688996266198, + "grad_norm": 2.03395414352417, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7181025743484497, + "num_tokens": 380471340.0, + "step": 14817 + }, + { + "epoch": 1.6272787173292333, + "grad_norm": 2.5133042335510254, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7516427040100098, + "num_tokens": 380490297.0, + "step": 14818 + }, + { + "epoch": 1.627388535031847, + "grad_norm": 2.2865169048309326, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7124840021133423, + "num_tokens": 380515849.0, + "step": 14819 + }, + { + "epoch": 1.6274983527344609, + "grad_norm": 1.9872491359710693, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7161314487457275, + "num_tokens": 380545875.0, + "step": 14820 + }, + { + "epoch": 1.6276081704370746, + "grad_norm": 2.0848569869995117, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7216744422912598, + "num_tokens": 380574133.0, + "step": 14821 + }, + { + "epoch": 1.6277179881396882, + "grad_norm": 2.153402805328369, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7405166625976562, + "num_tokens": 380599523.0, + "step": 14822 + }, + { + "epoch": 1.6278278058423017, + "grad_norm": 2.3879668712615967, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7367856502532959, + "num_tokens": 380622317.0, + "step": 14823 + }, + { + "epoch": 1.6279376235449154, + "grad_norm": 2.363755226135254, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7305525541305542, + "num_tokens": 380646263.0, + "step": 14824 + }, + { + "epoch": 1.6280474412475292, + "grad_norm": 2.05928111076355, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7172773480415344, + "num_tokens": 380676139.0, + "step": 14825 + }, + { + "epoch": 1.6281572589501427, + "grad_norm": 2.2988545894622803, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7051222324371338, + "num_tokens": 380701656.0, + "step": 14826 + }, + { + "epoch": 1.6282670766527563, + "grad_norm": 2.0874853134155273, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.715873658657074, + "num_tokens": 380731394.0, + "step": 14827 + }, + { + "epoch": 1.62837689435537, + "grad_norm": 2.2251014709472656, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7183223962783813, + "num_tokens": 380758197.0, + "step": 14828 + }, + { + "epoch": 1.6284867120579838, + "grad_norm": 2.301697015762329, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7006843090057373, + "num_tokens": 380784491.0, + "step": 14829 + }, + { + "epoch": 1.6285965297605975, + "grad_norm": 2.082705020904541, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7039836645126343, + "num_tokens": 380813430.0, + "step": 14830 + }, + { + "epoch": 1.628706347463211, + "grad_norm": 2.1495518684387207, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7227615714073181, + "num_tokens": 380840963.0, + "step": 14831 + }, + { + "epoch": 1.6288161651658246, + "grad_norm": 2.215392589569092, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7056884765625, + "num_tokens": 380867148.0, + "step": 14832 + }, + { + "epoch": 1.6289259828684384, + "grad_norm": 2.3787505626678467, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7268447875976562, + "num_tokens": 380891105.0, + "step": 14833 + }, + { + "epoch": 1.6290358005710521, + "grad_norm": 2.3527653217315674, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7156796455383301, + "num_tokens": 380913572.0, + "step": 14834 + }, + { + "epoch": 1.6291456182736657, + "grad_norm": 2.418843984603882, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7371996641159058, + "num_tokens": 380936747.0, + "step": 14835 + }, + { + "epoch": 1.6292554359762794, + "grad_norm": 2.127502202987671, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7009597420692444, + "num_tokens": 380964788.0, + "step": 14836 + }, + { + "epoch": 1.629365253678893, + "grad_norm": 1.8700525760650635, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.689656138420105, + "num_tokens": 381001103.0, + "step": 14837 + }, + { + "epoch": 1.6294750713815067, + "grad_norm": 2.7190909385681152, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7367429137229919, + "num_tokens": 381018446.0, + "step": 14838 + }, + { + "epoch": 1.6295848890841205, + "grad_norm": 2.398696184158325, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6978248357772827, + "num_tokens": 381043177.0, + "step": 14839 + }, + { + "epoch": 1.629694706786734, + "grad_norm": 1.9972217082977295, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6991570591926575, + "num_tokens": 381074764.0, + "step": 14840 + }, + { + "epoch": 1.6298045244893475, + "grad_norm": 2.4350688457489014, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7382674217224121, + "num_tokens": 381096228.0, + "step": 14841 + }, + { + "epoch": 1.6299143421919613, + "grad_norm": 2.305645227432251, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7239943146705627, + "num_tokens": 381119766.0, + "step": 14842 + }, + { + "epoch": 1.630024159894575, + "grad_norm": 2.2145566940307617, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7007927894592285, + "num_tokens": 381145867.0, + "step": 14843 + }, + { + "epoch": 1.6301339775971888, + "grad_norm": 2.2703490257263184, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7243149876594543, + "num_tokens": 381170918.0, + "step": 14844 + }, + { + "epoch": 1.6302437952998023, + "grad_norm": 2.179785966873169, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.716585636138916, + "num_tokens": 381198739.0, + "step": 14845 + }, + { + "epoch": 1.6303536130024159, + "grad_norm": 2.2572944164276123, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7168639898300171, + "num_tokens": 381227951.0, + "step": 14846 + }, + { + "epoch": 1.6304634307050296, + "grad_norm": 2.215919017791748, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7345999479293823, + "num_tokens": 381255382.0, + "step": 14847 + }, + { + "epoch": 1.6305732484076434, + "grad_norm": 2.2257602214813232, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7094216346740723, + "num_tokens": 381281430.0, + "step": 14848 + }, + { + "epoch": 1.630683066110257, + "grad_norm": 2.29176926612854, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.720624566078186, + "num_tokens": 381305372.0, + "step": 14849 + }, + { + "epoch": 1.6307928838128707, + "grad_norm": 2.417485237121582, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7294179797172546, + "num_tokens": 381330873.0, + "step": 14850 + }, + { + "epoch": 1.6309027015154842, + "grad_norm": 2.3779244422912598, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7340503931045532, + "num_tokens": 381354505.0, + "step": 14851 + }, + { + "epoch": 1.631012519218098, + "grad_norm": 2.1831910610198975, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7187552452087402, + "num_tokens": 381381611.0, + "step": 14852 + }, + { + "epoch": 1.6311223369207117, + "grad_norm": 2.304290294647217, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7524130344390869, + "num_tokens": 381403401.0, + "step": 14853 + }, + { + "epoch": 1.6312321546233253, + "grad_norm": 2.605386972427368, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.725020170211792, + "num_tokens": 381422932.0, + "step": 14854 + }, + { + "epoch": 1.6313419723259388, + "grad_norm": 2.0214388370513916, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7321258187294006, + "num_tokens": 381452920.0, + "step": 14855 + }, + { + "epoch": 1.6314517900285526, + "grad_norm": 1.9938608407974243, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7110123634338379, + "num_tokens": 381484484.0, + "step": 14856 + }, + { + "epoch": 1.6315616077311663, + "grad_norm": 2.085876226425171, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7644555568695068, + "num_tokens": 381510890.0, + "step": 14857 + }, + { + "epoch": 1.63167142543378, + "grad_norm": 2.5018670558929443, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7029231786727905, + "num_tokens": 381533208.0, + "step": 14858 + }, + { + "epoch": 1.6317812431363936, + "grad_norm": 2.2844417095184326, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7271590232849121, + "num_tokens": 381557479.0, + "step": 14859 + }, + { + "epoch": 1.6318910608390071, + "grad_norm": 2.2175817489624023, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.70882248878479, + "num_tokens": 381586078.0, + "step": 14860 + }, + { + "epoch": 1.632000878541621, + "grad_norm": 2.232100248336792, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7199596762657166, + "num_tokens": 381610000.0, + "step": 14861 + }, + { + "epoch": 1.6321106962442347, + "grad_norm": 2.266470432281494, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7456587553024292, + "num_tokens": 381633950.0, + "step": 14862 + }, + { + "epoch": 1.6322205139468482, + "grad_norm": 2.1097474098205566, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7272958755493164, + "num_tokens": 381662948.0, + "step": 14863 + }, + { + "epoch": 1.6323303316494617, + "grad_norm": 2.1496167182922363, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7134233713150024, + "num_tokens": 381690190.0, + "step": 14864 + }, + { + "epoch": 1.6324401493520755, + "grad_norm": 2.4262259006500244, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7327863574028015, + "num_tokens": 381711752.0, + "step": 14865 + }, + { + "epoch": 1.6325499670546892, + "grad_norm": 2.250087022781372, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.6989938020706177, + "num_tokens": 381737566.0, + "step": 14866 + }, + { + "epoch": 1.632659784757303, + "grad_norm": 2.2812790870666504, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7248144745826721, + "num_tokens": 381762890.0, + "step": 14867 + }, + { + "epoch": 1.6327696024599165, + "grad_norm": 2.4901599884033203, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.741428017616272, + "num_tokens": 381784606.0, + "step": 14868 + }, + { + "epoch": 1.63287942016253, + "grad_norm": 2.090550184249878, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.6972494721412659, + "num_tokens": 381815001.0, + "step": 14869 + }, + { + "epoch": 1.6329892378651438, + "grad_norm": 2.145672082901001, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7272778749465942, + "num_tokens": 381841121.0, + "step": 14870 + }, + { + "epoch": 1.6330990555677576, + "grad_norm": 2.3918685913085938, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.698077380657196, + "num_tokens": 381867592.0, + "step": 14871 + }, + { + "epoch": 1.6332088732703713, + "grad_norm": 2.414210557937622, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7254930734634399, + "num_tokens": 381890904.0, + "step": 14872 + }, + { + "epoch": 1.6333186909729849, + "grad_norm": 2.3898227214813232, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7083040475845337, + "num_tokens": 381917210.0, + "step": 14873 + }, + { + "epoch": 1.6334285086755984, + "grad_norm": 2.3331751823425293, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7332163453102112, + "num_tokens": 381939000.0, + "step": 14874 + }, + { + "epoch": 1.6335383263782122, + "grad_norm": 2.3165342807769775, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7312078475952148, + "num_tokens": 381964007.0, + "step": 14875 + }, + { + "epoch": 1.633648144080826, + "grad_norm": 2.2804219722747803, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7559636235237122, + "num_tokens": 381988744.0, + "step": 14876 + }, + { + "epoch": 1.6337579617834395, + "grad_norm": 1.9729728698730469, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.710476279258728, + "num_tokens": 382022267.0, + "step": 14877 + }, + { + "epoch": 1.633867779486053, + "grad_norm": 2.5045318603515625, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.736085057258606, + "num_tokens": 382043349.0, + "step": 14878 + }, + { + "epoch": 1.6339775971886668, + "grad_norm": 2.384874105453491, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7210953235626221, + "num_tokens": 382068074.0, + "step": 14879 + }, + { + "epoch": 1.6340874148912805, + "grad_norm": 2.255678415298462, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7072089910507202, + "num_tokens": 382094596.0, + "step": 14880 + }, + { + "epoch": 1.6341972325938943, + "grad_norm": 2.59525990486145, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7050973176956177, + "num_tokens": 382116116.0, + "step": 14881 + }, + { + "epoch": 1.6343070502965078, + "grad_norm": 2.5903916358947754, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7313615679740906, + "num_tokens": 382137117.0, + "step": 14882 + }, + { + "epoch": 1.6344168679991213, + "grad_norm": 2.2922685146331787, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7240752577781677, + "num_tokens": 382162998.0, + "step": 14883 + }, + { + "epoch": 1.634526685701735, + "grad_norm": 2.058788776397705, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7300657033920288, + "num_tokens": 382192367.0, + "step": 14884 + }, + { + "epoch": 1.6346365034043489, + "grad_norm": 2.3584086894989014, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7294154763221741, + "num_tokens": 382216383.0, + "step": 14885 + }, + { + "epoch": 1.6347463211069626, + "grad_norm": 2.098233699798584, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.731048583984375, + "num_tokens": 382247223.0, + "step": 14886 + }, + { + "epoch": 1.6348561388095761, + "grad_norm": 2.34214186668396, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7287425994873047, + "num_tokens": 382269680.0, + "step": 14887 + }, + { + "epoch": 1.6349659565121897, + "grad_norm": 2.4197311401367188, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7273991703987122, + "num_tokens": 382293972.0, + "step": 14888 + }, + { + "epoch": 1.6350757742148034, + "grad_norm": 2.6385445594787598, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.734740138053894, + "num_tokens": 382315614.0, + "step": 14889 + }, + { + "epoch": 1.6351855919174172, + "grad_norm": 2.867365598678589, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7480356693267822, + "num_tokens": 382332143.0, + "step": 14890 + }, + { + "epoch": 1.6352954096200307, + "grad_norm": 2.38244366645813, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7010783553123474, + "num_tokens": 382357095.0, + "step": 14891 + }, + { + "epoch": 1.6354052273226443, + "grad_norm": 2.434192419052124, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7248683571815491, + "num_tokens": 382378917.0, + "step": 14892 + }, + { + "epoch": 1.635515045025258, + "grad_norm": 2.1221275329589844, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7134934663772583, + "num_tokens": 382409987.0, + "step": 14893 + }, + { + "epoch": 1.6356248627278718, + "grad_norm": 2.1840193271636963, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6888734102249146, + "num_tokens": 382437439.0, + "step": 14894 + }, + { + "epoch": 1.6357346804304855, + "grad_norm": 2.4792394638061523, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7233864665031433, + "num_tokens": 382458218.0, + "step": 14895 + }, + { + "epoch": 1.635844498133099, + "grad_norm": 2.325608968734741, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7350327968597412, + "num_tokens": 382481511.0, + "step": 14896 + }, + { + "epoch": 1.6359543158357126, + "grad_norm": 2.1779322624206543, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7111175060272217, + "num_tokens": 382509638.0, + "step": 14897 + }, + { + "epoch": 1.6360641335383264, + "grad_norm": 2.1995017528533936, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7320331335067749, + "num_tokens": 382537296.0, + "step": 14898 + }, + { + "epoch": 1.6361739512409401, + "grad_norm": 2.243419647216797, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6993000507354736, + "num_tokens": 382563402.0, + "step": 14899 + }, + { + "epoch": 1.6362837689435537, + "grad_norm": 2.4886467456817627, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7216916084289551, + "num_tokens": 382585272.0, + "step": 14900 + }, + { + "epoch": 1.6363935866461674, + "grad_norm": 2.8283936977386475, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7537992000579834, + "num_tokens": 382602242.0, + "step": 14901 + }, + { + "epoch": 1.636503404348781, + "grad_norm": 2.3838095664978027, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.74275803565979, + "num_tokens": 382626164.0, + "step": 14902 + }, + { + "epoch": 1.6366132220513947, + "grad_norm": 2.025111436843872, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7217541933059692, + "num_tokens": 382654821.0, + "step": 14903 + }, + { + "epoch": 1.6367230397540085, + "grad_norm": 2.2358341217041016, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7195641994476318, + "num_tokens": 382680786.0, + "step": 14904 + }, + { + "epoch": 1.636832857456622, + "grad_norm": 2.247095823287964, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7220414876937866, + "num_tokens": 382709772.0, + "step": 14905 + }, + { + "epoch": 1.6369426751592355, + "grad_norm": 2.0711381435394287, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7194761633872986, + "num_tokens": 382737447.0, + "step": 14906 + }, + { + "epoch": 1.6370524928618493, + "grad_norm": 2.7481632232666016, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7302469611167908, + "num_tokens": 382757871.0, + "step": 14907 + }, + { + "epoch": 1.637162310564463, + "grad_norm": 1.9931423664093018, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7288689613342285, + "num_tokens": 382786693.0, + "step": 14908 + }, + { + "epoch": 1.6372721282670768, + "grad_norm": 2.1018240451812744, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7004365921020508, + "num_tokens": 382816599.0, + "step": 14909 + }, + { + "epoch": 1.6373819459696903, + "grad_norm": 2.2111897468566895, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7261825799942017, + "num_tokens": 382841020.0, + "step": 14910 + }, + { + "epoch": 1.6374917636723039, + "grad_norm": 2.222485303878784, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.6994830965995789, + "num_tokens": 382866559.0, + "step": 14911 + }, + { + "epoch": 1.6376015813749176, + "grad_norm": 2.3322157859802246, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7380501627922058, + "num_tokens": 382890198.0, + "step": 14912 + }, + { + "epoch": 1.6377113990775314, + "grad_norm": 2.348358392715454, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.6981114149093628, + "num_tokens": 382914336.0, + "step": 14913 + }, + { + "epoch": 1.637821216780145, + "grad_norm": 2.088005781173706, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7341338396072388, + "num_tokens": 382943288.0, + "step": 14914 + }, + { + "epoch": 1.6379310344827587, + "grad_norm": 2.2158892154693604, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7019690275192261, + "num_tokens": 382969759.0, + "step": 14915 + }, + { + "epoch": 1.6380408521853722, + "grad_norm": 2.5389978885650635, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7123597860336304, + "num_tokens": 382989798.0, + "step": 14916 + }, + { + "epoch": 1.638150669887986, + "grad_norm": 2.2786543369293213, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7169537544250488, + "num_tokens": 383015435.0, + "step": 14917 + }, + { + "epoch": 1.6382604875905997, + "grad_norm": 2.516103744506836, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7238431572914124, + "num_tokens": 383036865.0, + "step": 14918 + }, + { + "epoch": 1.6383703052932133, + "grad_norm": 2.4710216522216797, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7415359020233154, + "num_tokens": 383057619.0, + "step": 14919 + }, + { + "epoch": 1.6384801229958268, + "grad_norm": 2.2684783935546875, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7186120748519897, + "num_tokens": 383083487.0, + "step": 14920 + }, + { + "epoch": 1.6385899406984406, + "grad_norm": 2.2169997692108154, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7054250240325928, + "num_tokens": 383111780.0, + "step": 14921 + }, + { + "epoch": 1.6386997584010543, + "grad_norm": 2.188042163848877, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7175937294960022, + "num_tokens": 383138589.0, + "step": 14922 + }, + { + "epoch": 1.638809576103668, + "grad_norm": 2.0406641960144043, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7388681173324585, + "num_tokens": 383166172.0, + "step": 14923 + }, + { + "epoch": 1.6389193938062816, + "grad_norm": 1.9473909139633179, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6878516674041748, + "num_tokens": 383199180.0, + "step": 14924 + }, + { + "epoch": 1.6390292115088951, + "grad_norm": 2.1778202056884766, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7285970449447632, + "num_tokens": 383226577.0, + "step": 14925 + }, + { + "epoch": 1.639139029211509, + "grad_norm": 2.3368051052093506, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7179754972457886, + "num_tokens": 383250563.0, + "step": 14926 + }, + { + "epoch": 1.6392488469141226, + "grad_norm": 2.235830068588257, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7452700734138489, + "num_tokens": 383273456.0, + "step": 14927 + }, + { + "epoch": 1.6393586646167362, + "grad_norm": 2.156834125518799, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7022572755813599, + "num_tokens": 383304107.0, + "step": 14928 + }, + { + "epoch": 1.6394684823193497, + "grad_norm": 2.082324504852295, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7325232028961182, + "num_tokens": 383332284.0, + "step": 14929 + }, + { + "epoch": 1.6395783000219635, + "grad_norm": 2.05224871635437, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7212651968002319, + "num_tokens": 383361901.0, + "step": 14930 + }, + { + "epoch": 1.6396881177245772, + "grad_norm": 1.9028249979019165, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7150413990020752, + "num_tokens": 383395088.0, + "step": 14931 + }, + { + "epoch": 1.639797935427191, + "grad_norm": 2.241057872772217, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7355319261550903, + "num_tokens": 383418332.0, + "step": 14932 + }, + { + "epoch": 1.6399077531298045, + "grad_norm": 2.0736446380615234, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.705820620059967, + "num_tokens": 383448881.0, + "step": 14933 + }, + { + "epoch": 1.640017570832418, + "grad_norm": 2.3391315937042236, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7087196707725525, + "num_tokens": 383473505.0, + "step": 14934 + }, + { + "epoch": 1.6401273885350318, + "grad_norm": 2.2829625606536865, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.743537425994873, + "num_tokens": 383497879.0, + "step": 14935 + }, + { + "epoch": 1.6402372062376456, + "grad_norm": 2.333174467086792, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.723180890083313, + "num_tokens": 383521104.0, + "step": 14936 + }, + { + "epoch": 1.6403470239402593, + "grad_norm": 2.270197868347168, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7384194731712341, + "num_tokens": 383545452.0, + "step": 14937 + }, + { + "epoch": 1.6404568416428729, + "grad_norm": 2.2090485095977783, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7159658074378967, + "num_tokens": 383571416.0, + "step": 14938 + }, + { + "epoch": 1.6405666593454864, + "grad_norm": 2.0811586380004883, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7127685546875, + "num_tokens": 383602027.0, + "step": 14939 + }, + { + "epoch": 1.6406764770481002, + "grad_norm": 2.1882057189941406, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7142512202262878, + "num_tokens": 383628522.0, + "step": 14940 + }, + { + "epoch": 1.640786294750714, + "grad_norm": 2.3323163986206055, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.714568018913269, + "num_tokens": 383653107.0, + "step": 14941 + }, + { + "epoch": 1.6408961124533274, + "grad_norm": 2.3327345848083496, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7213729619979858, + "num_tokens": 383676384.0, + "step": 14942 + }, + { + "epoch": 1.641005930155941, + "grad_norm": 2.1960647106170654, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7279647588729858, + "num_tokens": 383702722.0, + "step": 14943 + }, + { + "epoch": 1.6411157478585547, + "grad_norm": 2.272412061691284, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.711068868637085, + "num_tokens": 383727836.0, + "step": 14944 + }, + { + "epoch": 1.6412255655611685, + "grad_norm": 2.293992519378662, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7147766947746277, + "num_tokens": 383753691.0, + "step": 14945 + }, + { + "epoch": 1.6413353832637823, + "grad_norm": 2.5501797199249268, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7234483957290649, + "num_tokens": 383775147.0, + "step": 14946 + }, + { + "epoch": 1.6414452009663958, + "grad_norm": 2.228947639465332, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.725527286529541, + "num_tokens": 383799876.0, + "step": 14947 + }, + { + "epoch": 1.6415550186690093, + "grad_norm": 2.0265228748321533, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7217158675193787, + "num_tokens": 383829724.0, + "step": 14948 + }, + { + "epoch": 1.641664836371623, + "grad_norm": 2.2659406661987305, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7160841822624207, + "num_tokens": 383856425.0, + "step": 14949 + }, + { + "epoch": 1.6417746540742368, + "grad_norm": 2.3517346382141113, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7094792127609253, + "num_tokens": 383880473.0, + "step": 14950 + }, + { + "epoch": 1.6418844717768504, + "grad_norm": 2.417393445968628, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7475935220718384, + "num_tokens": 383901545.0, + "step": 14951 + }, + { + "epoch": 1.6419942894794641, + "grad_norm": 1.9916985034942627, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7043180465698242, + "num_tokens": 383934357.0, + "step": 14952 + }, + { + "epoch": 1.6421041071820777, + "grad_norm": 2.2656972408294678, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7136655449867249, + "num_tokens": 383958889.0, + "step": 14953 + }, + { + "epoch": 1.6422139248846914, + "grad_norm": 2.648963451385498, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7261561155319214, + "num_tokens": 383978583.0, + "step": 14954 + }, + { + "epoch": 1.6423237425873052, + "grad_norm": 2.1842575073242188, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6941519379615784, + "num_tokens": 384007304.0, + "step": 14955 + }, + { + "epoch": 1.6424335602899187, + "grad_norm": 2.2516841888427734, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7224383354187012, + "num_tokens": 384033905.0, + "step": 14956 + }, + { + "epoch": 1.6425433779925322, + "grad_norm": 2.486325979232788, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7314296960830688, + "num_tokens": 384055022.0, + "step": 14957 + }, + { + "epoch": 1.642653195695146, + "grad_norm": 2.507357358932495, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7369343042373657, + "num_tokens": 384074455.0, + "step": 14958 + }, + { + "epoch": 1.6427630133977598, + "grad_norm": 2.220183849334717, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7340563535690308, + "num_tokens": 384101651.0, + "step": 14959 + }, + { + "epoch": 1.6428728311003735, + "grad_norm": 2.2427239418029785, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7122223973274231, + "num_tokens": 384130233.0, + "step": 14960 + }, + { + "epoch": 1.642982648802987, + "grad_norm": 2.529582977294922, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7280675172805786, + "num_tokens": 384150280.0, + "step": 14961 + }, + { + "epoch": 1.6430924665056006, + "grad_norm": 2.2241997718811035, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7169911861419678, + "num_tokens": 384174666.0, + "step": 14962 + }, + { + "epoch": 1.6432022842082143, + "grad_norm": 2.2791216373443604, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7169966697692871, + "num_tokens": 384200663.0, + "step": 14963 + }, + { + "epoch": 1.643312101910828, + "grad_norm": 2.0653233528137207, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6856057643890381, + "num_tokens": 384232211.0, + "step": 14964 + }, + { + "epoch": 1.6434219196134416, + "grad_norm": 2.288499593734741, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.704559862613678, + "num_tokens": 384257966.0, + "step": 14965 + }, + { + "epoch": 1.6435317373160554, + "grad_norm": 2.144852876663208, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7090036869049072, + "num_tokens": 384284417.0, + "step": 14966 + }, + { + "epoch": 1.643641555018669, + "grad_norm": 2.278874635696411, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7414145469665527, + "num_tokens": 384308219.0, + "step": 14967 + }, + { + "epoch": 1.6437513727212827, + "grad_norm": 2.523146867752075, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7363684177398682, + "num_tokens": 384328461.0, + "step": 14968 + }, + { + "epoch": 1.6438611904238964, + "grad_norm": 2.1366629600524902, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7201075553894043, + "num_tokens": 384361077.0, + "step": 14969 + }, + { + "epoch": 1.64397100812651, + "grad_norm": 2.302065134048462, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7287907600402832, + "num_tokens": 384384047.0, + "step": 14970 + }, + { + "epoch": 1.6440808258291235, + "grad_norm": 2.2057559490203857, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7039456367492676, + "num_tokens": 384413229.0, + "step": 14971 + }, + { + "epoch": 1.6441906435317373, + "grad_norm": 2.266446590423584, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7261596322059631, + "num_tokens": 384438994.0, + "step": 14972 + }, + { + "epoch": 1.644300461234351, + "grad_norm": 2.4802236557006836, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7420187592506409, + "num_tokens": 384460010.0, + "step": 14973 + }, + { + "epoch": 1.6444102789369648, + "grad_norm": 2.1692259311676025, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7203561663627625, + "num_tokens": 384486755.0, + "step": 14974 + }, + { + "epoch": 1.6445200966395783, + "grad_norm": 2.275906801223755, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.740784764289856, + "num_tokens": 384512492.0, + "step": 14975 + }, + { + "epoch": 1.6446299143421919, + "grad_norm": 2.627288579940796, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7218798398971558, + "num_tokens": 384532261.0, + "step": 14976 + }, + { + "epoch": 1.6447397320448056, + "grad_norm": 2.312343120574951, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7054178714752197, + "num_tokens": 384557758.0, + "step": 14977 + }, + { + "epoch": 1.6448495497474194, + "grad_norm": 2.184070587158203, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7177319526672363, + "num_tokens": 384584602.0, + "step": 14978 + }, + { + "epoch": 1.644959367450033, + "grad_norm": 2.0174756050109863, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.707553505897522, + "num_tokens": 384617061.0, + "step": 14979 + }, + { + "epoch": 1.6450691851526464, + "grad_norm": 2.2167115211486816, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7321211099624634, + "num_tokens": 384643248.0, + "step": 14980 + }, + { + "epoch": 1.6451790028552602, + "grad_norm": 2.429373025894165, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.717502772808075, + "num_tokens": 384666963.0, + "step": 14981 + }, + { + "epoch": 1.645288820557874, + "grad_norm": 2.2474780082702637, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7227427959442139, + "num_tokens": 384693499.0, + "step": 14982 + }, + { + "epoch": 1.6453986382604877, + "grad_norm": 2.2890682220458984, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7209668159484863, + "num_tokens": 384719118.0, + "step": 14983 + }, + { + "epoch": 1.6455084559631012, + "grad_norm": 2.167823553085327, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6929691433906555, + "num_tokens": 384748990.0, + "step": 14984 + }, + { + "epoch": 1.6456182736657148, + "grad_norm": 2.148486375808716, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7145718336105347, + "num_tokens": 384778553.0, + "step": 14985 + }, + { + "epoch": 1.6457280913683285, + "grad_norm": 1.931696891784668, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7010427713394165, + "num_tokens": 384811219.0, + "step": 14986 + }, + { + "epoch": 1.6458379090709423, + "grad_norm": 2.24316143989563, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6884803771972656, + "num_tokens": 384838403.0, + "step": 14987 + }, + { + "epoch": 1.645947726773556, + "grad_norm": 2.2450201511383057, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7213253974914551, + "num_tokens": 384863860.0, + "step": 14988 + }, + { + "epoch": 1.6460575444761696, + "grad_norm": 2.5262503623962402, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7265620827674866, + "num_tokens": 384886246.0, + "step": 14989 + }, + { + "epoch": 1.6461673621787831, + "grad_norm": 2.4168756008148193, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7007911205291748, + "num_tokens": 384910538.0, + "step": 14990 + }, + { + "epoch": 1.6462771798813969, + "grad_norm": 2.2060303688049316, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7499026656150818, + "num_tokens": 384934671.0, + "step": 14991 + }, + { + "epoch": 1.6463869975840106, + "grad_norm": 2.1833019256591797, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7125625014305115, + "num_tokens": 384963722.0, + "step": 14992 + }, + { + "epoch": 1.6464968152866242, + "grad_norm": 2.4262888431549072, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7427022457122803, + "num_tokens": 384984843.0, + "step": 14993 + }, + { + "epoch": 1.6466066329892377, + "grad_norm": 2.350698232650757, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7206292152404785, + "num_tokens": 385010055.0, + "step": 14994 + }, + { + "epoch": 1.6467164506918515, + "grad_norm": 2.0539028644561768, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7023816108703613, + "num_tokens": 385040038.0, + "step": 14995 + }, + { + "epoch": 1.6468262683944652, + "grad_norm": 2.4361064434051514, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7179436683654785, + "num_tokens": 385061820.0, + "step": 14996 + }, + { + "epoch": 1.646936086097079, + "grad_norm": 2.2138984203338623, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7045772671699524, + "num_tokens": 385089475.0, + "step": 14997 + }, + { + "epoch": 1.6470459037996925, + "grad_norm": 2.1629793643951416, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6980721950531006, + "num_tokens": 385116501.0, + "step": 14998 + }, + { + "epoch": 1.647155721502306, + "grad_norm": 2.3109896183013916, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7282067537307739, + "num_tokens": 385142366.0, + "step": 14999 + }, + { + "epoch": 1.6472655392049198, + "grad_norm": 2.4651830196380615, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.707785427570343, + "num_tokens": 385165687.0, + "step": 15000 + }, + { + "epoch": 1.6473753569075336, + "grad_norm": 2.160374879837036, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7637226581573486, + "num_tokens": 385188964.0, + "step": 15001 + }, + { + "epoch": 1.6474851746101473, + "grad_norm": 2.11389422416687, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6932433247566223, + "num_tokens": 385216354.0, + "step": 15002 + }, + { + "epoch": 1.6475949923127609, + "grad_norm": 2.2843735218048096, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7102046608924866, + "num_tokens": 385241356.0, + "step": 15003 + }, + { + "epoch": 1.6477048100153744, + "grad_norm": 2.523224115371704, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.714255154132843, + "num_tokens": 385264073.0, + "step": 15004 + }, + { + "epoch": 1.6478146277179881, + "grad_norm": 2.2295188903808594, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7172456979751587, + "num_tokens": 385290615.0, + "step": 15005 + }, + { + "epoch": 1.647924445420602, + "grad_norm": 2.317621946334839, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7312018871307373, + "num_tokens": 385314703.0, + "step": 15006 + }, + { + "epoch": 1.6480342631232154, + "grad_norm": 2.287648916244507, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7291696071624756, + "num_tokens": 385338812.0, + "step": 15007 + }, + { + "epoch": 1.648144080825829, + "grad_norm": 2.396970510482788, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7222509980201721, + "num_tokens": 385362802.0, + "step": 15008 + }, + { + "epoch": 1.6482538985284427, + "grad_norm": 2.1338136196136475, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7208839058876038, + "num_tokens": 385391766.0, + "step": 15009 + }, + { + "epoch": 1.6483637162310565, + "grad_norm": 2.107165813446045, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6925145983695984, + "num_tokens": 385422144.0, + "step": 15010 + }, + { + "epoch": 1.6484735339336702, + "grad_norm": 2.499748706817627, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7086303234100342, + "num_tokens": 385444781.0, + "step": 15011 + }, + { + "epoch": 1.6485833516362838, + "grad_norm": 2.007683038711548, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6931543350219727, + "num_tokens": 385477844.0, + "step": 15012 + }, + { + "epoch": 1.6486931693388973, + "grad_norm": 2.030338764190674, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.722948431968689, + "num_tokens": 385508387.0, + "step": 15013 + }, + { + "epoch": 1.648802987041511, + "grad_norm": 2.256892442703247, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7096868753433228, + "num_tokens": 385534845.0, + "step": 15014 + }, + { + "epoch": 1.6489128047441248, + "grad_norm": 2.2816784381866455, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7352761626243591, + "num_tokens": 385560994.0, + "step": 15015 + }, + { + "epoch": 1.6490226224467384, + "grad_norm": 2.335859537124634, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7284285426139832, + "num_tokens": 385585724.0, + "step": 15016 + }, + { + "epoch": 1.6491324401493521, + "grad_norm": 2.229379892349243, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7176149487495422, + "num_tokens": 385612087.0, + "step": 15017 + }, + { + "epoch": 1.6492422578519657, + "grad_norm": 2.220499277114868, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7547867298126221, + "num_tokens": 385636595.0, + "step": 15018 + }, + { + "epoch": 1.6493520755545794, + "grad_norm": 2.0008890628814697, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7155240774154663, + "num_tokens": 385666418.0, + "step": 15019 + }, + { + "epoch": 1.6494618932571932, + "grad_norm": 2.169595956802368, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.705829381942749, + "num_tokens": 385695568.0, + "step": 15020 + }, + { + "epoch": 1.6495717109598067, + "grad_norm": 2.124175786972046, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7204568386077881, + "num_tokens": 385726019.0, + "step": 15021 + }, + { + "epoch": 1.6496815286624202, + "grad_norm": 1.8402925729751587, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7017467021942139, + "num_tokens": 385763169.0, + "step": 15022 + }, + { + "epoch": 1.649791346365034, + "grad_norm": 2.50022292137146, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7173176407814026, + "num_tokens": 385786558.0, + "step": 15023 + }, + { + "epoch": 1.6499011640676478, + "grad_norm": 2.1766135692596436, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7332789301872253, + "num_tokens": 385812707.0, + "step": 15024 + }, + { + "epoch": 1.6500109817702615, + "grad_norm": 2.2126262187957764, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7104848623275757, + "num_tokens": 385839328.0, + "step": 15025 + }, + { + "epoch": 1.650120799472875, + "grad_norm": 2.1268162727355957, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7202881574630737, + "num_tokens": 385869267.0, + "step": 15026 + }, + { + "epoch": 1.6502306171754886, + "grad_norm": 2.2265243530273438, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7031905651092529, + "num_tokens": 385895458.0, + "step": 15027 + }, + { + "epoch": 1.6503404348781023, + "grad_norm": 2.0231311321258545, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7079657316207886, + "num_tokens": 385927697.0, + "step": 15028 + }, + { + "epoch": 1.650450252580716, + "grad_norm": 2.3365018367767334, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7006083130836487, + "num_tokens": 385953428.0, + "step": 15029 + }, + { + "epoch": 1.6505600702833296, + "grad_norm": 2.2189908027648926, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7517589330673218, + "num_tokens": 385977999.0, + "step": 15030 + }, + { + "epoch": 1.6506698879859434, + "grad_norm": 2.1989307403564453, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7128182649612427, + "num_tokens": 386004053.0, + "step": 15031 + }, + { + "epoch": 1.650779705688557, + "grad_norm": 2.2197318077087402, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7175258994102478, + "num_tokens": 386031267.0, + "step": 15032 + }, + { + "epoch": 1.6508895233911707, + "grad_norm": 2.1296091079711914, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7216670513153076, + "num_tokens": 386056766.0, + "step": 15033 + }, + { + "epoch": 1.6509993410937844, + "grad_norm": 2.3126959800720215, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.727920413017273, + "num_tokens": 386084441.0, + "step": 15034 + }, + { + "epoch": 1.651109158796398, + "grad_norm": 2.262281894683838, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7321174740791321, + "num_tokens": 386109451.0, + "step": 15035 + }, + { + "epoch": 1.6512189764990115, + "grad_norm": 2.328526735305786, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7226085066795349, + "num_tokens": 386134769.0, + "step": 15036 + }, + { + "epoch": 1.6513287942016253, + "grad_norm": 2.2924108505249023, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7372362613677979, + "num_tokens": 386157540.0, + "step": 15037 + }, + { + "epoch": 1.651438611904239, + "grad_norm": 2.3357112407684326, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7205527424812317, + "num_tokens": 386182279.0, + "step": 15038 + }, + { + "epoch": 1.6515484296068528, + "grad_norm": 2.2445068359375, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6966968774795532, + "num_tokens": 386208694.0, + "step": 15039 + }, + { + "epoch": 1.6516582473094663, + "grad_norm": 2.2538652420043945, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7218793034553528, + "num_tokens": 386234391.0, + "step": 15040 + }, + { + "epoch": 1.6517680650120798, + "grad_norm": 2.0918421745300293, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7162650227546692, + "num_tokens": 386264015.0, + "step": 15041 + }, + { + "epoch": 1.6518778827146936, + "grad_norm": 2.0720889568328857, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7182319760322571, + "num_tokens": 386293064.0, + "step": 15042 + }, + { + "epoch": 1.6519877004173074, + "grad_norm": 1.9209691286087036, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.694144070148468, + "num_tokens": 386324712.0, + "step": 15043 + }, + { + "epoch": 1.652097518119921, + "grad_norm": 2.067706823348999, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.712942898273468, + "num_tokens": 386355035.0, + "step": 15044 + }, + { + "epoch": 1.6522073358225344, + "grad_norm": 2.4359774589538574, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7162798643112183, + "num_tokens": 386377602.0, + "step": 15045 + }, + { + "epoch": 1.6523171535251482, + "grad_norm": 2.500481605529785, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7158627510070801, + "num_tokens": 386399731.0, + "step": 15046 + }, + { + "epoch": 1.652426971227762, + "grad_norm": 2.368417501449585, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7174440622329712, + "num_tokens": 386422986.0, + "step": 15047 + }, + { + "epoch": 1.6525367889303757, + "grad_norm": 2.1104841232299805, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7225126028060913, + "num_tokens": 386451292.0, + "step": 15048 + }, + { + "epoch": 1.6526466066329892, + "grad_norm": 2.112562894821167, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7027455568313599, + "num_tokens": 386480107.0, + "step": 15049 + }, + { + "epoch": 1.6527564243356028, + "grad_norm": 2.2736685276031494, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7269517779350281, + "num_tokens": 386505599.0, + "step": 15050 + }, + { + "epoch": 1.6528662420382165, + "grad_norm": 2.472050905227661, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7171847820281982, + "num_tokens": 386527295.0, + "step": 15051 + }, + { + "epoch": 1.6529760597408303, + "grad_norm": 2.374739408493042, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.730388343334198, + "num_tokens": 386548471.0, + "step": 15052 + }, + { + "epoch": 1.653085877443444, + "grad_norm": 2.234334707260132, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7217450737953186, + "num_tokens": 386574174.0, + "step": 15053 + }, + { + "epoch": 1.6531956951460576, + "grad_norm": 2.389113426208496, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7090688943862915, + "num_tokens": 386596699.0, + "step": 15054 + }, + { + "epoch": 1.653305512848671, + "grad_norm": 2.581378936767578, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7255292534828186, + "num_tokens": 386617969.0, + "step": 15055 + }, + { + "epoch": 1.6534153305512849, + "grad_norm": 2.3006582260131836, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.6998348832130432, + "num_tokens": 386642587.0, + "step": 15056 + }, + { + "epoch": 1.6535251482538986, + "grad_norm": 2.092458963394165, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7536453008651733, + "num_tokens": 386668101.0, + "step": 15057 + }, + { + "epoch": 1.6536349659565122, + "grad_norm": 2.3688302040100098, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7254194021224976, + "num_tokens": 386691621.0, + "step": 15058 + }, + { + "epoch": 1.6537447836591257, + "grad_norm": 2.1298935413360596, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7177348136901855, + "num_tokens": 386717240.0, + "step": 15059 + }, + { + "epoch": 1.6538546013617395, + "grad_norm": 2.3927597999572754, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7372192740440369, + "num_tokens": 386739297.0, + "step": 15060 + }, + { + "epoch": 1.6539644190643532, + "grad_norm": 2.2430713176727295, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.748996376991272, + "num_tokens": 386761256.0, + "step": 15061 + }, + { + "epoch": 1.654074236766967, + "grad_norm": 2.4156510829925537, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7072969675064087, + "num_tokens": 386785044.0, + "step": 15062 + }, + { + "epoch": 1.6541840544695805, + "grad_norm": 2.062490224838257, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7020500898361206, + "num_tokens": 386814471.0, + "step": 15063 + }, + { + "epoch": 1.654293872172194, + "grad_norm": 2.1174237728118896, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7166711091995239, + "num_tokens": 386842872.0, + "step": 15064 + }, + { + "epoch": 1.6544036898748078, + "grad_norm": 2.31435227394104, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7279797196388245, + "num_tokens": 386868680.0, + "step": 15065 + }, + { + "epoch": 1.6545135075774215, + "grad_norm": 2.278698682785034, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7251284122467041, + "num_tokens": 386893714.0, + "step": 15066 + }, + { + "epoch": 1.6546233252800353, + "grad_norm": 2.3457162380218506, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7270725965499878, + "num_tokens": 386919275.0, + "step": 15067 + }, + { + "epoch": 1.6547331429826488, + "grad_norm": 2.4252185821533203, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.738879919052124, + "num_tokens": 386942812.0, + "step": 15068 + }, + { + "epoch": 1.6548429606852624, + "grad_norm": 2.2792060375213623, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7172223925590515, + "num_tokens": 386966686.0, + "step": 15069 + }, + { + "epoch": 1.6549527783878761, + "grad_norm": 2.2040491104125977, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7397009134292603, + "num_tokens": 386993380.0, + "step": 15070 + }, + { + "epoch": 1.65506259609049, + "grad_norm": 1.9159784317016602, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7229434251785278, + "num_tokens": 387025540.0, + "step": 15071 + }, + { + "epoch": 1.6551724137931034, + "grad_norm": 2.1693027019500732, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7210443019866943, + "num_tokens": 387052143.0, + "step": 15072 + }, + { + "epoch": 1.655282231495717, + "grad_norm": 2.552400827407837, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7208955883979797, + "num_tokens": 387072311.0, + "step": 15073 + }, + { + "epoch": 1.6553920491983307, + "grad_norm": 2.7498672008514404, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7233832478523254, + "num_tokens": 387090163.0, + "step": 15074 + }, + { + "epoch": 1.6555018669009445, + "grad_norm": 1.9988276958465576, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7021121978759766, + "num_tokens": 387120426.0, + "step": 15075 + }, + { + "epoch": 1.6556116846035582, + "grad_norm": 2.0015883445739746, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7246309518814087, + "num_tokens": 387151061.0, + "step": 15076 + }, + { + "epoch": 1.6557215023061718, + "grad_norm": 2.394320249557495, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7271884679794312, + "num_tokens": 387174982.0, + "step": 15077 + }, + { + "epoch": 1.6558313200087853, + "grad_norm": 2.486923933029175, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7174270153045654, + "num_tokens": 387197216.0, + "step": 15078 + }, + { + "epoch": 1.655941137711399, + "grad_norm": 2.6251494884490967, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7191148996353149, + "num_tokens": 387219711.0, + "step": 15079 + }, + { + "epoch": 1.6560509554140128, + "grad_norm": 2.1939988136291504, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7266179323196411, + "num_tokens": 387249012.0, + "step": 15080 + }, + { + "epoch": 1.6561607731166264, + "grad_norm": 2.388049364089966, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7066420316696167, + "num_tokens": 387272162.0, + "step": 15081 + }, + { + "epoch": 1.65627059081924, + "grad_norm": 2.054750442504883, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7036484479904175, + "num_tokens": 387302577.0, + "step": 15082 + }, + { + "epoch": 1.6563804085218536, + "grad_norm": 2.1282501220703125, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7164286375045776, + "num_tokens": 387329347.0, + "step": 15083 + }, + { + "epoch": 1.6564902262244674, + "grad_norm": 2.4833924770355225, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7349842190742493, + "num_tokens": 387350021.0, + "step": 15084 + }, + { + "epoch": 1.6566000439270812, + "grad_norm": 2.222041130065918, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7330275177955627, + "num_tokens": 387376935.0, + "step": 15085 + }, + { + "epoch": 1.6567098616296947, + "grad_norm": 2.0059216022491455, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7207711935043335, + "num_tokens": 387405903.0, + "step": 15086 + }, + { + "epoch": 1.6568196793323082, + "grad_norm": 1.9949772357940674, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7231012582778931, + "num_tokens": 387436454.0, + "step": 15087 + }, + { + "epoch": 1.656929497034922, + "grad_norm": 2.425487995147705, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.744154155254364, + "num_tokens": 387457681.0, + "step": 15088 + }, + { + "epoch": 1.6570393147375357, + "grad_norm": 2.2918736934661865, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7463964223861694, + "num_tokens": 387482684.0, + "step": 15089 + }, + { + "epoch": 1.6571491324401495, + "grad_norm": 2.197197198867798, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7209760546684265, + "num_tokens": 387509358.0, + "step": 15090 + }, + { + "epoch": 1.657258950142763, + "grad_norm": 1.9949527978897095, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7307556867599487, + "num_tokens": 387538796.0, + "step": 15091 + }, + { + "epoch": 1.6573687678453766, + "grad_norm": 2.208930492401123, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7053462266921997, + "num_tokens": 387566555.0, + "step": 15092 + }, + { + "epoch": 1.6574785855479903, + "grad_norm": 2.401719331741333, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7408950328826904, + "num_tokens": 387587947.0, + "step": 15093 + }, + { + "epoch": 1.657588403250604, + "grad_norm": 2.21907377243042, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7376394867897034, + "num_tokens": 387612873.0, + "step": 15094 + }, + { + "epoch": 1.6576982209532176, + "grad_norm": 2.536625623703003, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7296850681304932, + "num_tokens": 387634621.0, + "step": 15095 + }, + { + "epoch": 1.6578080386558314, + "grad_norm": 2.2291367053985596, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7118216753005981, + "num_tokens": 387663554.0, + "step": 15096 + }, + { + "epoch": 1.657917856358445, + "grad_norm": 2.2816827297210693, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7284573316574097, + "num_tokens": 387688628.0, + "step": 15097 + }, + { + "epoch": 1.6580276740610587, + "grad_norm": 2.0506558418273926, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7147860527038574, + "num_tokens": 387718369.0, + "step": 15098 + }, + { + "epoch": 1.6581374917636724, + "grad_norm": 2.1253416538238525, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7259243130683899, + "num_tokens": 387746322.0, + "step": 15099 + }, + { + "epoch": 1.658247309466286, + "grad_norm": 2.1501126289367676, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7382797002792358, + "num_tokens": 387771405.0, + "step": 15100 + }, + { + "epoch": 1.6583571271688995, + "grad_norm": 2.7475409507751465, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.740868866443634, + "num_tokens": 387789952.0, + "step": 15101 + }, + { + "epoch": 1.6584669448715132, + "grad_norm": 2.225879430770874, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7304691076278687, + "num_tokens": 387814049.0, + "step": 15102 + }, + { + "epoch": 1.658576762574127, + "grad_norm": 2.197604179382324, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7179099321365356, + "num_tokens": 387839661.0, + "step": 15103 + }, + { + "epoch": 1.6586865802767408, + "grad_norm": 2.5211236476898193, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7534685134887695, + "num_tokens": 387858842.0, + "step": 15104 + }, + { + "epoch": 1.6587963979793543, + "grad_norm": 2.3241584300994873, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7178598642349243, + "num_tokens": 387885021.0, + "step": 15105 + }, + { + "epoch": 1.6589062156819678, + "grad_norm": 2.2811334133148193, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7033436894416809, + "num_tokens": 387909697.0, + "step": 15106 + }, + { + "epoch": 1.6590160333845816, + "grad_norm": 2.084014892578125, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.6971839070320129, + "num_tokens": 387939408.0, + "step": 15107 + }, + { + "epoch": 1.6591258510871953, + "grad_norm": 2.1871113777160645, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.724565863609314, + "num_tokens": 387965213.0, + "step": 15108 + }, + { + "epoch": 1.6592356687898089, + "grad_norm": 2.243157148361206, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.757841944694519, + "num_tokens": 387987594.0, + "step": 15109 + }, + { + "epoch": 1.6593454864924224, + "grad_norm": 2.352062702178955, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7203632593154907, + "num_tokens": 388009394.0, + "step": 15110 + }, + { + "epoch": 1.6594553041950362, + "grad_norm": 2.1187784671783447, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7025719881057739, + "num_tokens": 388036232.0, + "step": 15111 + }, + { + "epoch": 1.65956512189765, + "grad_norm": 1.978292465209961, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6972607970237732, + "num_tokens": 388067566.0, + "step": 15112 + }, + { + "epoch": 1.6596749396002637, + "grad_norm": 2.3323748111724854, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7395319938659668, + "num_tokens": 388088346.0, + "step": 15113 + }, + { + "epoch": 1.6597847573028772, + "grad_norm": 2.3208541870117188, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7218691110610962, + "num_tokens": 388112903.0, + "step": 15114 + }, + { + "epoch": 1.6598945750054908, + "grad_norm": 2.194941997528076, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7007554769515991, + "num_tokens": 388139127.0, + "step": 15115 + }, + { + "epoch": 1.6600043927081045, + "grad_norm": 2.237610101699829, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.715240478515625, + "num_tokens": 388166551.0, + "step": 15116 + }, + { + "epoch": 1.6601142104107183, + "grad_norm": 2.170448064804077, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.725151538848877, + "num_tokens": 388193630.0, + "step": 15117 + }, + { + "epoch": 1.660224028113332, + "grad_norm": 1.9815092086791992, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7108801603317261, + "num_tokens": 388224179.0, + "step": 15118 + }, + { + "epoch": 1.6603338458159456, + "grad_norm": 2.262539863586426, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7154231071472168, + "num_tokens": 388250957.0, + "step": 15119 + }, + { + "epoch": 1.660443663518559, + "grad_norm": 2.5596792697906494, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7312358021736145, + "num_tokens": 388271673.0, + "step": 15120 + }, + { + "epoch": 1.6605534812211729, + "grad_norm": 2.0873961448669434, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7375080585479736, + "num_tokens": 388297123.0, + "step": 15121 + }, + { + "epoch": 1.6606632989237866, + "grad_norm": 2.1408536434173584, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7503829002380371, + "num_tokens": 388321454.0, + "step": 15122 + }, + { + "epoch": 1.6607731166264001, + "grad_norm": 2.017094373703003, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7150446772575378, + "num_tokens": 388352261.0, + "step": 15123 + }, + { + "epoch": 1.6608829343290137, + "grad_norm": 2.5544626712799072, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7229195833206177, + "num_tokens": 388374654.0, + "step": 15124 + }, + { + "epoch": 1.6609927520316274, + "grad_norm": 2.3166775703430176, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7232069969177246, + "num_tokens": 388400075.0, + "step": 15125 + }, + { + "epoch": 1.6611025697342412, + "grad_norm": 2.078287363052368, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7139545679092407, + "num_tokens": 388428776.0, + "step": 15126 + }, + { + "epoch": 1.661212387436855, + "grad_norm": 2.342435121536255, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7166176438331604, + "num_tokens": 388451436.0, + "step": 15127 + }, + { + "epoch": 1.6613222051394685, + "grad_norm": 2.040018320083618, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7127764225006104, + "num_tokens": 388480620.0, + "step": 15128 + }, + { + "epoch": 1.661432022842082, + "grad_norm": 2.3006718158721924, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.729684054851532, + "num_tokens": 388505479.0, + "step": 15129 + }, + { + "epoch": 1.6615418405446958, + "grad_norm": 2.1393327713012695, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.734319806098938, + "num_tokens": 388534446.0, + "step": 15130 + }, + { + "epoch": 1.6616516582473095, + "grad_norm": 2.1206488609313965, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7027438879013062, + "num_tokens": 388564938.0, + "step": 15131 + }, + { + "epoch": 1.661761475949923, + "grad_norm": 2.0928151607513428, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7143669128417969, + "num_tokens": 388592047.0, + "step": 15132 + }, + { + "epoch": 1.6618712936525368, + "grad_norm": 2.213569402694702, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7252970337867737, + "num_tokens": 388618541.0, + "step": 15133 + }, + { + "epoch": 1.6619811113551504, + "grad_norm": 2.068267583847046, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7259308695793152, + "num_tokens": 388647294.0, + "step": 15134 + }, + { + "epoch": 1.6620909290577641, + "grad_norm": 2.2627530097961426, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7007563710212708, + "num_tokens": 388673267.0, + "step": 15135 + }, + { + "epoch": 1.6622007467603779, + "grad_norm": 2.187528371810913, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7081184983253479, + "num_tokens": 388700230.0, + "step": 15136 + }, + { + "epoch": 1.6623105644629914, + "grad_norm": 2.348297357559204, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7242359519004822, + "num_tokens": 388722572.0, + "step": 15137 + }, + { + "epoch": 1.662420382165605, + "grad_norm": 2.081617593765259, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7234921455383301, + "num_tokens": 388754255.0, + "step": 15138 + }, + { + "epoch": 1.6625301998682187, + "grad_norm": 2.1768441200256348, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7208478450775146, + "num_tokens": 388780011.0, + "step": 15139 + }, + { + "epoch": 1.6626400175708325, + "grad_norm": 1.765154242515564, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6917222738265991, + "num_tokens": 388817650.0, + "step": 15140 + }, + { + "epoch": 1.6627498352734462, + "grad_norm": 2.720532178878784, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.74654221534729, + "num_tokens": 388835569.0, + "step": 15141 + }, + { + "epoch": 1.6628596529760598, + "grad_norm": 2.1119935512542725, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.734963059425354, + "num_tokens": 388861341.0, + "step": 15142 + }, + { + "epoch": 1.6629694706786733, + "grad_norm": 2.5171124935150146, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7141979932785034, + "num_tokens": 388883702.0, + "step": 15143 + }, + { + "epoch": 1.663079288381287, + "grad_norm": 2.261282205581665, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7353495359420776, + "num_tokens": 388907185.0, + "step": 15144 + }, + { + "epoch": 1.6631891060839008, + "grad_norm": 2.4842569828033447, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7397647500038147, + "num_tokens": 388931162.0, + "step": 15145 + }, + { + "epoch": 1.6632989237865143, + "grad_norm": 2.3740932941436768, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7445559501647949, + "num_tokens": 388952615.0, + "step": 15146 + }, + { + "epoch": 1.663408741489128, + "grad_norm": 2.166931629180908, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7060028910636902, + "num_tokens": 388979624.0, + "step": 15147 + }, + { + "epoch": 1.6635185591917416, + "grad_norm": 2.231203079223633, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7347127795219421, + "num_tokens": 389003285.0, + "step": 15148 + }, + { + "epoch": 1.6636283768943554, + "grad_norm": 2.1938788890838623, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.6942272186279297, + "num_tokens": 389031131.0, + "step": 15149 + }, + { + "epoch": 1.6637381945969691, + "grad_norm": 2.04923677444458, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7333859205245972, + "num_tokens": 389058855.0, + "step": 15150 + }, + { + "epoch": 1.6638480122995827, + "grad_norm": 2.5591177940368652, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7220081090927124, + "num_tokens": 389078515.0, + "step": 15151 + }, + { + "epoch": 1.6639578300021962, + "grad_norm": 2.5472819805145264, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7029274702072144, + "num_tokens": 389103413.0, + "step": 15152 + }, + { + "epoch": 1.66406764770481, + "grad_norm": 2.637749195098877, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.723319411277771, + "num_tokens": 389123741.0, + "step": 15153 + }, + { + "epoch": 1.6641774654074237, + "grad_norm": 2.2084858417510986, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7063378095626831, + "num_tokens": 389150853.0, + "step": 15154 + }, + { + "epoch": 1.6642872831100375, + "grad_norm": 2.051259994506836, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7163112759590149, + "num_tokens": 389180625.0, + "step": 15155 + }, + { + "epoch": 1.664397100812651, + "grad_norm": 2.1636335849761963, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7161720991134644, + "num_tokens": 389207885.0, + "step": 15156 + }, + { + "epoch": 1.6645069185152646, + "grad_norm": 2.0780842304229736, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7435920238494873, + "num_tokens": 389236904.0, + "step": 15157 + }, + { + "epoch": 1.6646167362178783, + "grad_norm": 2.3139846324920654, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7099438309669495, + "num_tokens": 389261249.0, + "step": 15158 + }, + { + "epoch": 1.664726553920492, + "grad_norm": 2.3297624588012695, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7090705633163452, + "num_tokens": 389287397.0, + "step": 15159 + }, + { + "epoch": 1.6648363716231056, + "grad_norm": 2.068385124206543, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.736445426940918, + "num_tokens": 389315385.0, + "step": 15160 + }, + { + "epoch": 1.6649461893257194, + "grad_norm": 2.2079877853393555, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7353712320327759, + "num_tokens": 389339840.0, + "step": 15161 + }, + { + "epoch": 1.665056007028333, + "grad_norm": 1.9583431482315063, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7072112560272217, + "num_tokens": 389374411.0, + "step": 15162 + }, + { + "epoch": 1.6651658247309467, + "grad_norm": 1.9369052648544312, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7121686935424805, + "num_tokens": 389408038.0, + "step": 15163 + }, + { + "epoch": 1.6652756424335604, + "grad_norm": 2.2156565189361572, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6949689388275146, + "num_tokens": 389434315.0, + "step": 15164 + }, + { + "epoch": 1.665385460136174, + "grad_norm": 2.3506991863250732, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7183570861816406, + "num_tokens": 389456842.0, + "step": 15165 + }, + { + "epoch": 1.6654952778387875, + "grad_norm": 2.227311372756958, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7289873957633972, + "num_tokens": 389483014.0, + "step": 15166 + }, + { + "epoch": 1.6656050955414012, + "grad_norm": 2.3853414058685303, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7284966707229614, + "num_tokens": 389504479.0, + "step": 15167 + }, + { + "epoch": 1.665714913244015, + "grad_norm": 2.0865767002105713, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7082231044769287, + "num_tokens": 389534211.0, + "step": 15168 + }, + { + "epoch": 1.6658247309466288, + "grad_norm": 2.338793992996216, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7418674826622009, + "num_tokens": 389557150.0, + "step": 15169 + }, + { + "epoch": 1.6659345486492423, + "grad_norm": 2.0792720317840576, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7075328230857849, + "num_tokens": 389585189.0, + "step": 15170 + }, + { + "epoch": 1.6660443663518558, + "grad_norm": 2.162646770477295, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7329603433609009, + "num_tokens": 389613516.0, + "step": 15171 + }, + { + "epoch": 1.6661541840544696, + "grad_norm": 2.0994157791137695, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.697401762008667, + "num_tokens": 389643989.0, + "step": 15172 + }, + { + "epoch": 1.6662640017570833, + "grad_norm": 2.128060817718506, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.727880597114563, + "num_tokens": 389671081.0, + "step": 15173 + }, + { + "epoch": 1.6663738194596969, + "grad_norm": 2.0397796630859375, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7160554528236389, + "num_tokens": 389700885.0, + "step": 15174 + }, + { + "epoch": 1.6664836371623104, + "grad_norm": 2.1490864753723145, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7273494601249695, + "num_tokens": 389726549.0, + "step": 15175 + }, + { + "epoch": 1.6665934548649242, + "grad_norm": 2.0168964862823486, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7342908382415771, + "num_tokens": 389757742.0, + "step": 15176 + }, + { + "epoch": 1.666703272567538, + "grad_norm": 2.3010942935943604, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6981832385063171, + "num_tokens": 389784639.0, + "step": 15177 + }, + { + "epoch": 1.6668130902701517, + "grad_norm": 2.035364866256714, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7294246554374695, + "num_tokens": 389812447.0, + "step": 15178 + }, + { + "epoch": 1.6669229079727652, + "grad_norm": 2.2320973873138428, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7222058176994324, + "num_tokens": 389837832.0, + "step": 15179 + }, + { + "epoch": 1.6670327256753787, + "grad_norm": 2.048159122467041, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6942355632781982, + "num_tokens": 389869238.0, + "step": 15180 + }, + { + "epoch": 1.6671425433779925, + "grad_norm": 2.4193363189697266, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7346898317337036, + "num_tokens": 389891530.0, + "step": 15181 + }, + { + "epoch": 1.6672523610806063, + "grad_norm": 2.247056245803833, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.743391215801239, + "num_tokens": 389916640.0, + "step": 15182 + }, + { + "epoch": 1.66736217878322, + "grad_norm": 2.2826128005981445, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7389682531356812, + "num_tokens": 389939522.0, + "step": 15183 + }, + { + "epoch": 1.6674719964858336, + "grad_norm": 2.338663101196289, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7347964644432068, + "num_tokens": 389962436.0, + "step": 15184 + }, + { + "epoch": 1.667581814188447, + "grad_norm": 2.5635297298431396, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7237019538879395, + "num_tokens": 389983477.0, + "step": 15185 + }, + { + "epoch": 1.6676916318910608, + "grad_norm": 2.1188745498657227, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.705539345741272, + "num_tokens": 390012089.0, + "step": 15186 + }, + { + "epoch": 1.6678014495936746, + "grad_norm": 2.4510819911956787, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7182733416557312, + "num_tokens": 390034298.0, + "step": 15187 + }, + { + "epoch": 1.6679112672962881, + "grad_norm": 2.359792947769165, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7086731195449829, + "num_tokens": 390058267.0, + "step": 15188 + }, + { + "epoch": 1.6680210849989017, + "grad_norm": 2.167299747467041, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7082210779190063, + "num_tokens": 390086577.0, + "step": 15189 + }, + { + "epoch": 1.6681309027015154, + "grad_norm": 2.393310546875, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7131071090698242, + "num_tokens": 390109962.0, + "step": 15190 + }, + { + "epoch": 1.6682407204041292, + "grad_norm": 2.033154010772705, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7424179911613464, + "num_tokens": 390138806.0, + "step": 15191 + }, + { + "epoch": 1.668350538106743, + "grad_norm": 2.004746437072754, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7281175255775452, + "num_tokens": 390169934.0, + "step": 15192 + }, + { + "epoch": 1.6684603558093565, + "grad_norm": 2.155738592147827, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6996761560440063, + "num_tokens": 390197605.0, + "step": 15193 + }, + { + "epoch": 1.66857017351197, + "grad_norm": 2.4483134746551514, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7403051853179932, + "num_tokens": 390218294.0, + "step": 15194 + }, + { + "epoch": 1.6686799912145838, + "grad_norm": 2.545469045639038, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7360719442367554, + "num_tokens": 390237176.0, + "step": 15195 + }, + { + "epoch": 1.6687898089171975, + "grad_norm": 2.2134251594543457, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.727249264717102, + "num_tokens": 390263047.0, + "step": 15196 + }, + { + "epoch": 1.668899626619811, + "grad_norm": 2.2251012325286865, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7289844751358032, + "num_tokens": 390287694.0, + "step": 15197 + }, + { + "epoch": 1.6690094443224248, + "grad_norm": 2.283353328704834, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7135095596313477, + "num_tokens": 390313592.0, + "step": 15198 + }, + { + "epoch": 1.6691192620250384, + "grad_norm": 1.875913381576538, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7403704524040222, + "num_tokens": 390346399.0, + "step": 15199 + }, + { + "epoch": 1.669229079727652, + "grad_norm": 2.2414939403533936, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7244279384613037, + "num_tokens": 390372790.0, + "step": 15200 + }, + { + "epoch": 1.6693388974302659, + "grad_norm": 2.2068469524383545, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.719620943069458, + "num_tokens": 390399925.0, + "step": 15201 + }, + { + "epoch": 1.6694487151328794, + "grad_norm": 2.163177967071533, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7137548923492432, + "num_tokens": 390426915.0, + "step": 15202 + }, + { + "epoch": 1.669558532835493, + "grad_norm": 2.170504570007324, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7203919887542725, + "num_tokens": 390453480.0, + "step": 15203 + }, + { + "epoch": 1.6696683505381067, + "grad_norm": 2.2302167415618896, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7446619272232056, + "num_tokens": 390477568.0, + "step": 15204 + }, + { + "epoch": 1.6697781682407205, + "grad_norm": 2.403437376022339, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7088773250579834, + "num_tokens": 390501964.0, + "step": 15205 + }, + { + "epoch": 1.6698879859433342, + "grad_norm": 2.136115550994873, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7210291624069214, + "num_tokens": 390530256.0, + "step": 15206 + }, + { + "epoch": 1.6699978036459477, + "grad_norm": 2.0684847831726074, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7085326910018921, + "num_tokens": 390559244.0, + "step": 15207 + }, + { + "epoch": 1.6701076213485613, + "grad_norm": 2.126758575439453, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7151416540145874, + "num_tokens": 390586965.0, + "step": 15208 + }, + { + "epoch": 1.670217439051175, + "grad_norm": 2.4526207447052, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7121874094009399, + "num_tokens": 390609226.0, + "step": 15209 + }, + { + "epoch": 1.6703272567537888, + "grad_norm": 2.52933931350708, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7129135727882385, + "num_tokens": 390630413.0, + "step": 15210 + }, + { + "epoch": 1.6704370744564023, + "grad_norm": 2.4242889881134033, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7146505117416382, + "num_tokens": 390652909.0, + "step": 15211 + }, + { + "epoch": 1.670546892159016, + "grad_norm": 2.3140079975128174, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7384458780288696, + "num_tokens": 390675529.0, + "step": 15212 + }, + { + "epoch": 1.6706567098616296, + "grad_norm": 2.7541322708129883, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7363447546958923, + "num_tokens": 390693041.0, + "step": 15213 + }, + { + "epoch": 1.6707665275642434, + "grad_norm": 2.3521506786346436, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7285167574882507, + "num_tokens": 390714156.0, + "step": 15214 + }, + { + "epoch": 1.6708763452668571, + "grad_norm": 2.529664993286133, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7157245874404907, + "num_tokens": 390735958.0, + "step": 15215 + }, + { + "epoch": 1.6709861629694707, + "grad_norm": 2.2088558673858643, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7113595008850098, + "num_tokens": 390763756.0, + "step": 15216 + }, + { + "epoch": 1.6710959806720842, + "grad_norm": 2.166682243347168, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7167266607284546, + "num_tokens": 390789578.0, + "step": 15217 + }, + { + "epoch": 1.671205798374698, + "grad_norm": 2.251650810241699, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7204388976097107, + "num_tokens": 390815200.0, + "step": 15218 + }, + { + "epoch": 1.6713156160773117, + "grad_norm": 2.197202682495117, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7361433506011963, + "num_tokens": 390839999.0, + "step": 15219 + }, + { + "epoch": 1.6714254337799255, + "grad_norm": 2.161240816116333, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7118844985961914, + "num_tokens": 390869422.0, + "step": 15220 + }, + { + "epoch": 1.671535251482539, + "grad_norm": 2.0783379077911377, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7285857796669006, + "num_tokens": 390896703.0, + "step": 15221 + }, + { + "epoch": 1.6716450691851525, + "grad_norm": 2.222996950149536, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6955243945121765, + "num_tokens": 390923562.0, + "step": 15222 + }, + { + "epoch": 1.6717548868877663, + "grad_norm": 2.311368703842163, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.7085188031196594, + "num_tokens": 390946936.0, + "step": 15223 + }, + { + "epoch": 1.67186470459038, + "grad_norm": 2.221787929534912, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7216390371322632, + "num_tokens": 390973866.0, + "step": 15224 + }, + { + "epoch": 1.6719745222929936, + "grad_norm": 2.2081799507141113, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7161105871200562, + "num_tokens": 390999712.0, + "step": 15225 + }, + { + "epoch": 1.6720843399956071, + "grad_norm": 2.346010684967041, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7121955752372742, + "num_tokens": 391022266.0, + "step": 15226 + }, + { + "epoch": 1.6721941576982209, + "grad_norm": 1.9488500356674194, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7250044345855713, + "num_tokens": 391053032.0, + "step": 15227 + }, + { + "epoch": 1.6723039754008346, + "grad_norm": 2.642416000366211, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7457902431488037, + "num_tokens": 391070731.0, + "step": 15228 + }, + { + "epoch": 1.6724137931034484, + "grad_norm": 2.076991081237793, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7061743140220642, + "num_tokens": 391100416.0, + "step": 15229 + }, + { + "epoch": 1.672523610806062, + "grad_norm": 2.2034099102020264, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7219377756118774, + "num_tokens": 391128046.0, + "step": 15230 + }, + { + "epoch": 1.6726334285086755, + "grad_norm": 2.4499921798706055, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7312873601913452, + "num_tokens": 391149350.0, + "step": 15231 + }, + { + "epoch": 1.6727432462112892, + "grad_norm": 2.1414482593536377, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7120596170425415, + "num_tokens": 391176840.0, + "step": 15232 + }, + { + "epoch": 1.672853063913903, + "grad_norm": 2.5655605792999268, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7314666509628296, + "num_tokens": 391197796.0, + "step": 15233 + }, + { + "epoch": 1.6729628816165167, + "grad_norm": 2.702465057373047, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7316871285438538, + "num_tokens": 391215813.0, + "step": 15234 + }, + { + "epoch": 1.6730726993191303, + "grad_norm": 2.646052122116089, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7315971851348877, + "num_tokens": 391236241.0, + "step": 15235 + }, + { + "epoch": 1.6731825170217438, + "grad_norm": 2.309204578399658, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7373740673065186, + "num_tokens": 391260126.0, + "step": 15236 + }, + { + "epoch": 1.6732923347243576, + "grad_norm": 2.1871511936187744, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7123560905456543, + "num_tokens": 391287180.0, + "step": 15237 + }, + { + "epoch": 1.6734021524269713, + "grad_norm": 2.5253565311431885, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7374475598335266, + "num_tokens": 391306574.0, + "step": 15238 + }, + { + "epoch": 1.6735119701295849, + "grad_norm": 2.8164045810699463, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7443671822547913, + "num_tokens": 391324438.0, + "step": 15239 + }, + { + "epoch": 1.6736217878321984, + "grad_norm": 2.2750959396362305, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7275177836418152, + "num_tokens": 391348092.0, + "step": 15240 + }, + { + "epoch": 1.6737316055348122, + "grad_norm": 2.37788462638855, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7339329123497009, + "num_tokens": 391371361.0, + "step": 15241 + }, + { + "epoch": 1.673841423237426, + "grad_norm": 2.2356719970703125, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7272224426269531, + "num_tokens": 391395542.0, + "step": 15242 + }, + { + "epoch": 1.6739512409400397, + "grad_norm": 2.2122952938079834, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7229112386703491, + "num_tokens": 391421953.0, + "step": 15243 + }, + { + "epoch": 1.6740610586426532, + "grad_norm": 2.219451427459717, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7486611604690552, + "num_tokens": 391446309.0, + "step": 15244 + }, + { + "epoch": 1.6741708763452667, + "grad_norm": 2.24564790725708, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6977797746658325, + "num_tokens": 391473344.0, + "step": 15245 + }, + { + "epoch": 1.6742806940478805, + "grad_norm": 2.4983367919921875, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7262089252471924, + "num_tokens": 391494270.0, + "step": 15246 + }, + { + "epoch": 1.6743905117504942, + "grad_norm": 2.037661552429199, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7186747193336487, + "num_tokens": 391523075.0, + "step": 15247 + }, + { + "epoch": 1.674500329453108, + "grad_norm": 1.955148458480835, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7272482514381409, + "num_tokens": 391556980.0, + "step": 15248 + }, + { + "epoch": 1.6746101471557215, + "grad_norm": 2.2924349308013916, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7133287787437439, + "num_tokens": 391580310.0, + "step": 15249 + }, + { + "epoch": 1.674719964858335, + "grad_norm": 2.1665709018707275, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7277847528457642, + "num_tokens": 391607929.0, + "step": 15250 + }, + { + "epoch": 1.6748297825609488, + "grad_norm": 1.9555892944335938, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7227914929389954, + "num_tokens": 391639790.0, + "step": 15251 + }, + { + "epoch": 1.6749396002635626, + "grad_norm": 2.179433822631836, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7116823792457581, + "num_tokens": 391665847.0, + "step": 15252 + }, + { + "epoch": 1.6750494179661761, + "grad_norm": 2.208782196044922, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.694216251373291, + "num_tokens": 391692876.0, + "step": 15253 + }, + { + "epoch": 1.6751592356687897, + "grad_norm": 2.214259386062622, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7108072638511658, + "num_tokens": 391719502.0, + "step": 15254 + }, + { + "epoch": 1.6752690533714034, + "grad_norm": 2.7632312774658203, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7357412576675415, + "num_tokens": 391738468.0, + "step": 15255 + }, + { + "epoch": 1.6753788710740172, + "grad_norm": 2.2969655990600586, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.726470947265625, + "num_tokens": 391762427.0, + "step": 15256 + }, + { + "epoch": 1.675488688776631, + "grad_norm": 2.353330373764038, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7108508348464966, + "num_tokens": 391788420.0, + "step": 15257 + }, + { + "epoch": 1.6755985064792445, + "grad_norm": 2.1292471885681152, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7063587307929993, + "num_tokens": 391815976.0, + "step": 15258 + }, + { + "epoch": 1.675708324181858, + "grad_norm": 2.1511502265930176, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7192763090133667, + "num_tokens": 391843114.0, + "step": 15259 + }, + { + "epoch": 1.6758181418844718, + "grad_norm": 2.0181169509887695, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7196083068847656, + "num_tokens": 391872470.0, + "step": 15260 + }, + { + "epoch": 1.6759279595870855, + "grad_norm": 2.335272789001465, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7078561782836914, + "num_tokens": 391897965.0, + "step": 15261 + }, + { + "epoch": 1.676037777289699, + "grad_norm": 2.0005152225494385, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7145527005195618, + "num_tokens": 391929290.0, + "step": 15262 + }, + { + "epoch": 1.6761475949923128, + "grad_norm": 2.0906474590301514, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7126750946044922, + "num_tokens": 391961649.0, + "step": 15263 + }, + { + "epoch": 1.6762574126949263, + "grad_norm": 2.3645589351654053, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7542179822921753, + "num_tokens": 391984340.0, + "step": 15264 + }, + { + "epoch": 1.67636723039754, + "grad_norm": 2.0095701217651367, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7081871628761292, + "num_tokens": 392014437.0, + "step": 15265 + }, + { + "epoch": 1.6764770481001539, + "grad_norm": 2.227520704269409, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7368131279945374, + "num_tokens": 392039182.0, + "step": 15266 + }, + { + "epoch": 1.6765868658027674, + "grad_norm": 2.064277172088623, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7135113477706909, + "num_tokens": 392069136.0, + "step": 15267 + }, + { + "epoch": 1.676696683505381, + "grad_norm": 2.270829677581787, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7319021821022034, + "num_tokens": 392094392.0, + "step": 15268 + }, + { + "epoch": 1.6768065012079947, + "grad_norm": 2.1466851234436035, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7127417325973511, + "num_tokens": 392121855.0, + "step": 15269 + }, + { + "epoch": 1.6769163189106084, + "grad_norm": 2.050497531890869, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7330726385116577, + "num_tokens": 392149731.0, + "step": 15270 + }, + { + "epoch": 1.6770261366132222, + "grad_norm": 2.0908689498901367, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7181699872016907, + "num_tokens": 392179006.0, + "step": 15271 + }, + { + "epoch": 1.6771359543158357, + "grad_norm": 2.0077664852142334, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7095582485198975, + "num_tokens": 392210993.0, + "step": 15272 + }, + { + "epoch": 1.6772457720184493, + "grad_norm": 2.174687385559082, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7088154554367065, + "num_tokens": 392239290.0, + "step": 15273 + }, + { + "epoch": 1.677355589721063, + "grad_norm": 2.1259779930114746, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7064675092697144, + "num_tokens": 392268448.0, + "step": 15274 + }, + { + "epoch": 1.6774654074236768, + "grad_norm": 2.021592140197754, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7072607278823853, + "num_tokens": 392298627.0, + "step": 15275 + }, + { + "epoch": 1.6775752251262903, + "grad_norm": 2.435478448867798, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7163220643997192, + "num_tokens": 392320946.0, + "step": 15276 + }, + { + "epoch": 1.677685042828904, + "grad_norm": 2.1329872608184814, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.719929575920105, + "num_tokens": 392349678.0, + "step": 15277 + }, + { + "epoch": 1.6777948605315176, + "grad_norm": 2.197237730026245, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7487655878067017, + "num_tokens": 392374637.0, + "step": 15278 + }, + { + "epoch": 1.6779046782341314, + "grad_norm": 2.4436187744140625, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7180365920066833, + "num_tokens": 392396207.0, + "step": 15279 + }, + { + "epoch": 1.6780144959367451, + "grad_norm": 2.276151180267334, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7139521241188049, + "num_tokens": 392421448.0, + "step": 15280 + }, + { + "epoch": 1.6781243136393587, + "grad_norm": 2.0535056591033936, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7038686275482178, + "num_tokens": 392452118.0, + "step": 15281 + }, + { + "epoch": 1.6782341313419722, + "grad_norm": 2.283613920211792, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7214218974113464, + "num_tokens": 392476061.0, + "step": 15282 + }, + { + "epoch": 1.678343949044586, + "grad_norm": 2.255391836166382, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7304582595825195, + "num_tokens": 392501002.0, + "step": 15283 + }, + { + "epoch": 1.6784537667471997, + "grad_norm": 2.337912082672119, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7152798771858215, + "num_tokens": 392525564.0, + "step": 15284 + }, + { + "epoch": 1.6785635844498135, + "grad_norm": 2.3999314308166504, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7650314569473267, + "num_tokens": 392546920.0, + "step": 15285 + }, + { + "epoch": 1.678673402152427, + "grad_norm": 2.3417656421661377, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7461255788803101, + "num_tokens": 392569162.0, + "step": 15286 + }, + { + "epoch": 1.6787832198550405, + "grad_norm": 2.1778435707092285, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7084284424781799, + "num_tokens": 392598735.0, + "step": 15287 + }, + { + "epoch": 1.6788930375576543, + "grad_norm": 2.4842395782470703, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7309642434120178, + "num_tokens": 392618909.0, + "step": 15288 + }, + { + "epoch": 1.679002855260268, + "grad_norm": 2.38279128074646, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7055128812789917, + "num_tokens": 392643417.0, + "step": 15289 + }, + { + "epoch": 1.6791126729628816, + "grad_norm": 2.4277758598327637, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7326498627662659, + "num_tokens": 392665550.0, + "step": 15290 + }, + { + "epoch": 1.6792224906654951, + "grad_norm": 2.3664584159851074, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7166756391525269, + "num_tokens": 392689531.0, + "step": 15291 + }, + { + "epoch": 1.6793323083681089, + "grad_norm": 2.024606704711914, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.743247389793396, + "num_tokens": 392717759.0, + "step": 15292 + }, + { + "epoch": 1.6794421260707226, + "grad_norm": 2.364475727081299, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7115928530693054, + "num_tokens": 392743745.0, + "step": 15293 + }, + { + "epoch": 1.6795519437733364, + "grad_norm": 2.218839168548584, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7148197293281555, + "num_tokens": 392770063.0, + "step": 15294 + }, + { + "epoch": 1.67966176147595, + "grad_norm": 2.242108106613159, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7144180536270142, + "num_tokens": 392795424.0, + "step": 15295 + }, + { + "epoch": 1.6797715791785635, + "grad_norm": 2.2929298877716064, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7090927362442017, + "num_tokens": 392820106.0, + "step": 15296 + }, + { + "epoch": 1.6798813968811772, + "grad_norm": 2.331766366958618, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7442212104797363, + "num_tokens": 392841835.0, + "step": 15297 + }, + { + "epoch": 1.679991214583791, + "grad_norm": 2.1551990509033203, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.6999809741973877, + "num_tokens": 392869486.0, + "step": 15298 + }, + { + "epoch": 1.6801010322864047, + "grad_norm": 2.3777332305908203, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7111294865608215, + "num_tokens": 392893264.0, + "step": 15299 + }, + { + "epoch": 1.6802108499890183, + "grad_norm": 2.0325827598571777, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7416677474975586, + "num_tokens": 392922164.0, + "step": 15300 + }, + { + "epoch": 1.6803206676916318, + "grad_norm": 2.32670259475708, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6985507011413574, + "num_tokens": 392950083.0, + "step": 15301 + }, + { + "epoch": 1.6804304853942456, + "grad_norm": 1.9890578985214233, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6890113353729248, + "num_tokens": 392985893.0, + "step": 15302 + }, + { + "epoch": 1.6805403030968593, + "grad_norm": 2.113020420074463, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.727608323097229, + "num_tokens": 393014877.0, + "step": 15303 + }, + { + "epoch": 1.6806501207994728, + "grad_norm": 2.2027745246887207, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7326292991638184, + "num_tokens": 393043166.0, + "step": 15304 + }, + { + "epoch": 1.6807599385020864, + "grad_norm": 2.3332808017730713, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7135138511657715, + "num_tokens": 393065510.0, + "step": 15305 + }, + { + "epoch": 1.6808697562047001, + "grad_norm": 2.0654027462005615, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7001162767410278, + "num_tokens": 393095301.0, + "step": 15306 + }, + { + "epoch": 1.680979573907314, + "grad_norm": 1.9486792087554932, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7119491100311279, + "num_tokens": 393127585.0, + "step": 15307 + }, + { + "epoch": 1.6810893916099277, + "grad_norm": 2.373239755630493, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7296617031097412, + "num_tokens": 393148469.0, + "step": 15308 + }, + { + "epoch": 1.6811992093125412, + "grad_norm": 2.597797155380249, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7389697432518005, + "num_tokens": 393168108.0, + "step": 15309 + }, + { + "epoch": 1.6813090270151547, + "grad_norm": 2.1378941535949707, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6972467303276062, + "num_tokens": 393198530.0, + "step": 15310 + }, + { + "epoch": 1.6814188447177685, + "grad_norm": 2.1555590629577637, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7013310194015503, + "num_tokens": 393224807.0, + "step": 15311 + }, + { + "epoch": 1.6815286624203822, + "grad_norm": 2.0711448192596436, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7102413177490234, + "num_tokens": 393255899.0, + "step": 15312 + }, + { + "epoch": 1.681638480122996, + "grad_norm": 2.5344836711883545, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.751574695110321, + "num_tokens": 393277210.0, + "step": 15313 + }, + { + "epoch": 1.6817482978256095, + "grad_norm": 2.6484596729278564, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7297190427780151, + "num_tokens": 393295897.0, + "step": 15314 + }, + { + "epoch": 1.681858115528223, + "grad_norm": 2.609684467315674, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.732643723487854, + "num_tokens": 393315070.0, + "step": 15315 + }, + { + "epoch": 1.6819679332308368, + "grad_norm": 2.5637917518615723, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7314233183860779, + "num_tokens": 393337401.0, + "step": 15316 + }, + { + "epoch": 1.6820777509334506, + "grad_norm": 2.3900134563446045, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7359288930892944, + "num_tokens": 393360719.0, + "step": 15317 + }, + { + "epoch": 1.6821875686360641, + "grad_norm": 2.0164241790771484, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7133519649505615, + "num_tokens": 393392631.0, + "step": 15318 + }, + { + "epoch": 1.6822973863386776, + "grad_norm": 2.2308175563812256, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6819015741348267, + "num_tokens": 393420790.0, + "step": 15319 + }, + { + "epoch": 1.6824072040412914, + "grad_norm": 2.4643638134002686, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7074471712112427, + "num_tokens": 393444092.0, + "step": 15320 + }, + { + "epoch": 1.6825170217439052, + "grad_norm": 2.097759246826172, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7313867211341858, + "num_tokens": 393472213.0, + "step": 15321 + }, + { + "epoch": 1.682626839446519, + "grad_norm": 2.094618320465088, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7161080837249756, + "num_tokens": 393502375.0, + "step": 15322 + }, + { + "epoch": 1.6827366571491325, + "grad_norm": 2.487494945526123, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7396793365478516, + "num_tokens": 393522886.0, + "step": 15323 + }, + { + "epoch": 1.682846474851746, + "grad_norm": 2.168649911880493, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7147557139396667, + "num_tokens": 393549254.0, + "step": 15324 + }, + { + "epoch": 1.6829562925543597, + "grad_norm": 2.611279249191284, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.766679048538208, + "num_tokens": 393568236.0, + "step": 15325 + }, + { + "epoch": 1.6830661102569735, + "grad_norm": 2.3670856952667236, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7311403155326843, + "num_tokens": 393590694.0, + "step": 15326 + }, + { + "epoch": 1.683175927959587, + "grad_norm": 2.153118371963501, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6945488452911377, + "num_tokens": 393621089.0, + "step": 15327 + }, + { + "epoch": 1.6832857456622008, + "grad_norm": 2.6150896549224854, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7195611596107483, + "num_tokens": 393640769.0, + "step": 15328 + }, + { + "epoch": 1.6833955633648143, + "grad_norm": 2.4655416011810303, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7270227670669556, + "num_tokens": 393661425.0, + "step": 15329 + }, + { + "epoch": 1.683505381067428, + "grad_norm": 2.4131953716278076, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7463370561599731, + "num_tokens": 393682020.0, + "step": 15330 + }, + { + "epoch": 1.6836151987700418, + "grad_norm": 2.2756547927856445, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7311701774597168, + "num_tokens": 393705636.0, + "step": 15331 + }, + { + "epoch": 1.6837250164726554, + "grad_norm": 2.0543618202209473, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.726642370223999, + "num_tokens": 393733784.0, + "step": 15332 + }, + { + "epoch": 1.683834834175269, + "grad_norm": 2.258119821548462, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7084370255470276, + "num_tokens": 393759528.0, + "step": 15333 + }, + { + "epoch": 1.6839446518778827, + "grad_norm": 2.0743887424468994, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.728577733039856, + "num_tokens": 393785849.0, + "step": 15334 + }, + { + "epoch": 1.6840544695804964, + "grad_norm": 1.810694932937622, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7201509475708008, + "num_tokens": 393823124.0, + "step": 15335 + }, + { + "epoch": 1.6841642872831102, + "grad_norm": 2.447551965713501, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7175602316856384, + "num_tokens": 393847224.0, + "step": 15336 + }, + { + "epoch": 1.6842741049857237, + "grad_norm": 2.072112560272217, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7251652479171753, + "num_tokens": 393875985.0, + "step": 15337 + }, + { + "epoch": 1.6843839226883373, + "grad_norm": 2.0618555545806885, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.730927586555481, + "num_tokens": 393903883.0, + "step": 15338 + }, + { + "epoch": 1.684493740390951, + "grad_norm": 2.2014453411102295, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7099265456199646, + "num_tokens": 393931820.0, + "step": 15339 + }, + { + "epoch": 1.6846035580935648, + "grad_norm": 2.220388174057007, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.760717511177063, + "num_tokens": 393956321.0, + "step": 15340 + }, + { + "epoch": 1.6847133757961783, + "grad_norm": 2.3526034355163574, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7495108842849731, + "num_tokens": 393978398.0, + "step": 15341 + }, + { + "epoch": 1.684823193498792, + "grad_norm": 2.3812973499298096, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7238546013832092, + "num_tokens": 394003421.0, + "step": 15342 + }, + { + "epoch": 1.6849330112014056, + "grad_norm": 2.1887195110321045, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7356635332107544, + "num_tokens": 394028819.0, + "step": 15343 + }, + { + "epoch": 1.6850428289040194, + "grad_norm": 2.6137635707855225, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7537748217582703, + "num_tokens": 394047166.0, + "step": 15344 + }, + { + "epoch": 1.685152646606633, + "grad_norm": 2.363734483718872, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7549506425857544, + "num_tokens": 394068017.0, + "step": 15345 + }, + { + "epoch": 1.6852624643092466, + "grad_norm": 2.163227081298828, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7094982862472534, + "num_tokens": 394096799.0, + "step": 15346 + }, + { + "epoch": 1.6853722820118602, + "grad_norm": 2.1868791580200195, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7450586557388306, + "num_tokens": 394122595.0, + "step": 15347 + }, + { + "epoch": 1.685482099714474, + "grad_norm": 2.151130437850952, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7264389395713806, + "num_tokens": 394149961.0, + "step": 15348 + }, + { + "epoch": 1.6855919174170877, + "grad_norm": 2.447404146194458, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7031099796295166, + "num_tokens": 394173061.0, + "step": 15349 + }, + { + "epoch": 1.6857017351197014, + "grad_norm": 2.2185654640197754, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7414462566375732, + "num_tokens": 394199052.0, + "step": 15350 + }, + { + "epoch": 1.685811552822315, + "grad_norm": 2.1080515384674072, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7126715183258057, + "num_tokens": 394227018.0, + "step": 15351 + }, + { + "epoch": 1.6859213705249285, + "grad_norm": 2.2442450523376465, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7153398394584656, + "num_tokens": 394252029.0, + "step": 15352 + }, + { + "epoch": 1.6860311882275423, + "grad_norm": 2.117455244064331, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.708199143409729, + "num_tokens": 394281108.0, + "step": 15353 + }, + { + "epoch": 1.686141005930156, + "grad_norm": 2.0025177001953125, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7011911869049072, + "num_tokens": 394311528.0, + "step": 15354 + }, + { + "epoch": 1.6862508236327696, + "grad_norm": 2.1960527896881104, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7200076580047607, + "num_tokens": 394339039.0, + "step": 15355 + }, + { + "epoch": 1.686360641335383, + "grad_norm": 2.3894639015197754, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7191814184188843, + "num_tokens": 394360819.0, + "step": 15356 + }, + { + "epoch": 1.6864704590379969, + "grad_norm": 2.830204725265503, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7394592761993408, + "num_tokens": 394378031.0, + "step": 15357 + }, + { + "epoch": 1.6865802767406106, + "grad_norm": 2.0638184547424316, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6816755533218384, + "num_tokens": 394410084.0, + "step": 15358 + }, + { + "epoch": 1.6866900944432244, + "grad_norm": 2.4063720703125, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7319344282150269, + "num_tokens": 394431928.0, + "step": 15359 + }, + { + "epoch": 1.686799912145838, + "grad_norm": 2.709601640701294, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7202228903770447, + "num_tokens": 394451686.0, + "step": 15360 + }, + { + "epoch": 1.6869097298484514, + "grad_norm": 2.3438291549682617, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.726058304309845, + "num_tokens": 394475438.0, + "step": 15361 + }, + { + "epoch": 1.6870195475510652, + "grad_norm": 2.162660598754883, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7141597270965576, + "num_tokens": 394502542.0, + "step": 15362 + }, + { + "epoch": 1.687129365253679, + "grad_norm": 2.2895050048828125, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7052320241928101, + "num_tokens": 394528333.0, + "step": 15363 + }, + { + "epoch": 1.6872391829562927, + "grad_norm": 2.119138717651367, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7236036062240601, + "num_tokens": 394556064.0, + "step": 15364 + }, + { + "epoch": 1.6873490006589063, + "grad_norm": 2.407902717590332, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7354514598846436, + "num_tokens": 394580499.0, + "step": 15365 + }, + { + "epoch": 1.6874588183615198, + "grad_norm": 2.0542523860931396, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.7035439014434814, + "num_tokens": 394612378.0, + "step": 15366 + }, + { + "epoch": 1.6875686360641335, + "grad_norm": 2.249547243118286, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7073372602462769, + "num_tokens": 394638489.0, + "step": 15367 + }, + { + "epoch": 1.6876784537667473, + "grad_norm": 2.0368857383728027, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.727133572101593, + "num_tokens": 394667393.0, + "step": 15368 + }, + { + "epoch": 1.6877882714693608, + "grad_norm": 2.3526782989501953, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7646045088768005, + "num_tokens": 394689235.0, + "step": 15369 + }, + { + "epoch": 1.6878980891719744, + "grad_norm": 2.3961684703826904, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7181500196456909, + "num_tokens": 394712465.0, + "step": 15370 + }, + { + "epoch": 1.6880079068745881, + "grad_norm": 2.2730042934417725, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7048407196998596, + "num_tokens": 394736589.0, + "step": 15371 + }, + { + "epoch": 1.6881177245772019, + "grad_norm": 2.646577835083008, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7370584011077881, + "num_tokens": 394756295.0, + "step": 15372 + }, + { + "epoch": 1.6882275422798156, + "grad_norm": 1.9971959590911865, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.713897705078125, + "num_tokens": 394787181.0, + "step": 15373 + }, + { + "epoch": 1.6883373599824292, + "grad_norm": 1.9803063869476318, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.719398558139801, + "num_tokens": 394818961.0, + "step": 15374 + }, + { + "epoch": 1.6884471776850427, + "grad_norm": 2.170686721801758, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.705238938331604, + "num_tokens": 394846220.0, + "step": 15375 + }, + { + "epoch": 1.6885569953876565, + "grad_norm": 2.5387895107269287, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7230180501937866, + "num_tokens": 394867982.0, + "step": 15376 + }, + { + "epoch": 1.6886668130902702, + "grad_norm": 2.244182586669922, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7149966955184937, + "num_tokens": 394894573.0, + "step": 15377 + }, + { + "epoch": 1.6887766307928838, + "grad_norm": 2.099196672439575, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7048819065093994, + "num_tokens": 394926019.0, + "step": 15378 + }, + { + "epoch": 1.6888864484954975, + "grad_norm": 2.378757953643799, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.71783846616745, + "num_tokens": 394951258.0, + "step": 15379 + }, + { + "epoch": 1.688996266198111, + "grad_norm": 2.569713592529297, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7346591949462891, + "num_tokens": 394971622.0, + "step": 15380 + }, + { + "epoch": 1.6891060839007248, + "grad_norm": 2.143364429473877, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7247119545936584, + "num_tokens": 394999797.0, + "step": 15381 + }, + { + "epoch": 1.6892159016033386, + "grad_norm": 2.097221612930298, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.712662398815155, + "num_tokens": 395026935.0, + "step": 15382 + }, + { + "epoch": 1.689325719305952, + "grad_norm": 1.8961764574050903, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7026508450508118, + "num_tokens": 395063684.0, + "step": 15383 + }, + { + "epoch": 1.6894355370085656, + "grad_norm": 2.123591899871826, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7145175337791443, + "num_tokens": 395091278.0, + "step": 15384 + }, + { + "epoch": 1.6895453547111794, + "grad_norm": 2.3241937160491943, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.706013560295105, + "num_tokens": 395118546.0, + "step": 15385 + }, + { + "epoch": 1.6896551724137931, + "grad_norm": 2.359532356262207, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7010400295257568, + "num_tokens": 395142397.0, + "step": 15386 + }, + { + "epoch": 1.689764990116407, + "grad_norm": 2.142235517501831, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7120639085769653, + "num_tokens": 395171237.0, + "step": 15387 + }, + { + "epoch": 1.6898748078190204, + "grad_norm": 2.497891664505005, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7082103490829468, + "num_tokens": 395193737.0, + "step": 15388 + }, + { + "epoch": 1.689984625521634, + "grad_norm": 2.1946024894714355, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7285594940185547, + "num_tokens": 395219425.0, + "step": 15389 + }, + { + "epoch": 1.6900944432242477, + "grad_norm": 2.3558602333068848, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.705859363079071, + "num_tokens": 395242770.0, + "step": 15390 + }, + { + "epoch": 1.6902042609268615, + "grad_norm": 2.2341628074645996, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7279766798019409, + "num_tokens": 395268875.0, + "step": 15391 + }, + { + "epoch": 1.690314078629475, + "grad_norm": 1.8918195962905884, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7249380350112915, + "num_tokens": 395301119.0, + "step": 15392 + }, + { + "epoch": 1.6904238963320888, + "grad_norm": 2.125105857849121, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7165560126304626, + "num_tokens": 395331337.0, + "step": 15393 + }, + { + "epoch": 1.6905337140347023, + "grad_norm": 2.4282586574554443, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7308071851730347, + "num_tokens": 395354211.0, + "step": 15394 + }, + { + "epoch": 1.690643531737316, + "grad_norm": 2.46907377243042, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.712616503238678, + "num_tokens": 395376497.0, + "step": 15395 + }, + { + "epoch": 1.6907533494399298, + "grad_norm": 2.292947292327881, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.727376401424408, + "num_tokens": 395400451.0, + "step": 15396 + }, + { + "epoch": 1.6908631671425434, + "grad_norm": 2.187971591949463, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7039347887039185, + "num_tokens": 395426059.0, + "step": 15397 + }, + { + "epoch": 1.690972984845157, + "grad_norm": 2.056385040283203, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6939879655838013, + "num_tokens": 395460402.0, + "step": 15398 + }, + { + "epoch": 1.6910828025477707, + "grad_norm": 2.3543238639831543, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.72059166431427, + "num_tokens": 395484825.0, + "step": 15399 + }, + { + "epoch": 1.6911926202503844, + "grad_norm": 2.163214683532715, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7096834778785706, + "num_tokens": 395512090.0, + "step": 15400 + }, + { + "epoch": 1.6913024379529982, + "grad_norm": 2.245863914489746, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7115245461463928, + "num_tokens": 395538346.0, + "step": 15401 + }, + { + "epoch": 1.6914122556556117, + "grad_norm": 2.2098405361175537, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7249300479888916, + "num_tokens": 395564555.0, + "step": 15402 + }, + { + "epoch": 1.6915220733582252, + "grad_norm": 2.217787504196167, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7188525795936584, + "num_tokens": 395591803.0, + "step": 15403 + }, + { + "epoch": 1.691631891060839, + "grad_norm": 1.930397629737854, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7022736072540283, + "num_tokens": 395626285.0, + "step": 15404 + }, + { + "epoch": 1.6917417087634528, + "grad_norm": 2.1360411643981934, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7195936441421509, + "num_tokens": 395653862.0, + "step": 15405 + }, + { + "epoch": 1.6918515264660663, + "grad_norm": 2.1713852882385254, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.720348596572876, + "num_tokens": 395681732.0, + "step": 15406 + }, + { + "epoch": 1.6919613441686798, + "grad_norm": 1.903451919555664, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7000797986984253, + "num_tokens": 395715758.0, + "step": 15407 + }, + { + "epoch": 1.6920711618712936, + "grad_norm": 2.425785779953003, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7256621718406677, + "num_tokens": 395737839.0, + "step": 15408 + }, + { + "epoch": 1.6921809795739073, + "grad_norm": 2.191667079925537, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7104914784431458, + "num_tokens": 395768938.0, + "step": 15409 + }, + { + "epoch": 1.692290797276521, + "grad_norm": 1.9851300716400146, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7241706848144531, + "num_tokens": 395799432.0, + "step": 15410 + }, + { + "epoch": 1.6924006149791346, + "grad_norm": 1.9208385944366455, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7010576128959656, + "num_tokens": 395833439.0, + "step": 15411 + }, + { + "epoch": 1.6925104326817482, + "grad_norm": 2.3033740520477295, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7272523641586304, + "num_tokens": 395856497.0, + "step": 15412 + }, + { + "epoch": 1.692620250384362, + "grad_norm": 2.2419145107269287, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.7000043392181396, + "num_tokens": 395883943.0, + "step": 15413 + }, + { + "epoch": 1.6927300680869757, + "grad_norm": 1.9265146255493164, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.744238018989563, + "num_tokens": 395914443.0, + "step": 15414 + }, + { + "epoch": 1.6928398857895894, + "grad_norm": 2.4299259185791016, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7338759899139404, + "num_tokens": 395936241.0, + "step": 15415 + }, + { + "epoch": 1.692949703492203, + "grad_norm": 2.0558884143829346, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7180758118629456, + "num_tokens": 395962844.0, + "step": 15416 + }, + { + "epoch": 1.6930595211948165, + "grad_norm": 2.384270191192627, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7335931062698364, + "num_tokens": 395984062.0, + "step": 15417 + }, + { + "epoch": 1.6931693388974303, + "grad_norm": 2.2273762226104736, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7297032475471497, + "num_tokens": 396009074.0, + "step": 15418 + }, + { + "epoch": 1.693279156600044, + "grad_norm": 2.2114875316619873, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7370429039001465, + "num_tokens": 396034322.0, + "step": 15419 + }, + { + "epoch": 1.6933889743026576, + "grad_norm": 2.399268627166748, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7330816984176636, + "num_tokens": 396055329.0, + "step": 15420 + }, + { + "epoch": 1.693498792005271, + "grad_norm": 2.84743332862854, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7423467040061951, + "num_tokens": 396072680.0, + "step": 15421 + }, + { + "epoch": 1.6936086097078848, + "grad_norm": 1.8459396362304688, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6998758912086487, + "num_tokens": 396109114.0, + "step": 15422 + }, + { + "epoch": 1.6937184274104986, + "grad_norm": 2.3703322410583496, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7137775421142578, + "num_tokens": 396132296.0, + "step": 15423 + }, + { + "epoch": 1.6938282451131124, + "grad_norm": 2.2842254638671875, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6983498334884644, + "num_tokens": 396157043.0, + "step": 15424 + }, + { + "epoch": 1.693938062815726, + "grad_norm": 2.124246835708618, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7104005217552185, + "num_tokens": 396186829.0, + "step": 15425 + }, + { + "epoch": 1.6940478805183394, + "grad_norm": 2.1841375827789307, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7041435241699219, + "num_tokens": 396213112.0, + "step": 15426 + }, + { + "epoch": 1.6941576982209532, + "grad_norm": 2.400007963180542, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7432295083999634, + "num_tokens": 396235764.0, + "step": 15427 + }, + { + "epoch": 1.694267515923567, + "grad_norm": 2.475212812423706, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7473047971725464, + "num_tokens": 396258972.0, + "step": 15428 + }, + { + "epoch": 1.6943773336261807, + "grad_norm": 2.473700761795044, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7296596765518188, + "num_tokens": 396281430.0, + "step": 15429 + }, + { + "epoch": 1.6944871513287942, + "grad_norm": 2.203446865081787, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7178221940994263, + "num_tokens": 396308080.0, + "step": 15430 + }, + { + "epoch": 1.6945969690314078, + "grad_norm": 2.2110562324523926, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7457454204559326, + "num_tokens": 396333833.0, + "step": 15431 + }, + { + "epoch": 1.6947067867340215, + "grad_norm": 2.164149045944214, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7241330146789551, + "num_tokens": 396361989.0, + "step": 15432 + }, + { + "epoch": 1.6948166044366353, + "grad_norm": 2.1113388538360596, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7224004864692688, + "num_tokens": 396388455.0, + "step": 15433 + }, + { + "epoch": 1.6949264221392488, + "grad_norm": 2.0529534816741943, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7250131368637085, + "num_tokens": 396416982.0, + "step": 15434 + }, + { + "epoch": 1.6950362398418624, + "grad_norm": 1.9567996263504028, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6962660551071167, + "num_tokens": 396450007.0, + "step": 15435 + }, + { + "epoch": 1.6951460575444761, + "grad_norm": 2.3155715465545654, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7097596526145935, + "num_tokens": 396474133.0, + "step": 15436 + }, + { + "epoch": 1.6952558752470899, + "grad_norm": 2.4298017024993896, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7151263952255249, + "num_tokens": 396497428.0, + "step": 15437 + }, + { + "epoch": 1.6953656929497036, + "grad_norm": 2.1333179473876953, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7058906555175781, + "num_tokens": 396522414.0, + "step": 15438 + }, + { + "epoch": 1.6954755106523172, + "grad_norm": 2.050600290298462, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7058181762695312, + "num_tokens": 396553679.0, + "step": 15439 + }, + { + "epoch": 1.6955853283549307, + "grad_norm": 2.129185438156128, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7188048362731934, + "num_tokens": 396579247.0, + "step": 15440 + }, + { + "epoch": 1.6956951460575445, + "grad_norm": 2.1883347034454346, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7145538926124573, + "num_tokens": 396605733.0, + "step": 15441 + }, + { + "epoch": 1.6958049637601582, + "grad_norm": 2.657835006713867, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.738024115562439, + "num_tokens": 396625745.0, + "step": 15442 + }, + { + "epoch": 1.6959147814627717, + "grad_norm": 2.2460274696350098, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7253358364105225, + "num_tokens": 396650384.0, + "step": 15443 + }, + { + "epoch": 1.6960245991653855, + "grad_norm": 2.40337872505188, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7188376188278198, + "num_tokens": 396674101.0, + "step": 15444 + }, + { + "epoch": 1.696134416867999, + "grad_norm": 2.5496366024017334, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7229996919631958, + "num_tokens": 396695159.0, + "step": 15445 + }, + { + "epoch": 1.6962442345706128, + "grad_norm": 2.254934310913086, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7594367861747742, + "num_tokens": 396719488.0, + "step": 15446 + }, + { + "epoch": 1.6963540522732266, + "grad_norm": 2.1092405319213867, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7211153507232666, + "num_tokens": 396745481.0, + "step": 15447 + }, + { + "epoch": 1.69646386997584, + "grad_norm": 2.2689313888549805, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7374959588050842, + "num_tokens": 396768864.0, + "step": 15448 + }, + { + "epoch": 1.6965736876784536, + "grad_norm": 2.2771565914154053, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7163568735122681, + "num_tokens": 396793672.0, + "step": 15449 + }, + { + "epoch": 1.6966835053810674, + "grad_norm": 2.0248799324035645, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7355176210403442, + "num_tokens": 396823432.0, + "step": 15450 + }, + { + "epoch": 1.6967933230836811, + "grad_norm": 2.142174482345581, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7300230264663696, + "num_tokens": 396849115.0, + "step": 15451 + }, + { + "epoch": 1.696903140786295, + "grad_norm": 2.4736080169677734, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7534918785095215, + "num_tokens": 396869511.0, + "step": 15452 + }, + { + "epoch": 1.6970129584889084, + "grad_norm": 1.99253511428833, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7053636312484741, + "num_tokens": 396900405.0, + "step": 15453 + }, + { + "epoch": 1.697122776191522, + "grad_norm": 2.34213924407959, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7277824282646179, + "num_tokens": 396924974.0, + "step": 15454 + }, + { + "epoch": 1.6972325938941357, + "grad_norm": 2.292921304702759, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7251914143562317, + "num_tokens": 396951455.0, + "step": 15455 + }, + { + "epoch": 1.6973424115967495, + "grad_norm": 2.444455623626709, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7214785814285278, + "num_tokens": 396973872.0, + "step": 15456 + }, + { + "epoch": 1.697452229299363, + "grad_norm": 2.4586331844329834, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7376660704612732, + "num_tokens": 396995496.0, + "step": 15457 + }, + { + "epoch": 1.6975620470019768, + "grad_norm": 2.0940282344818115, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7086391448974609, + "num_tokens": 397024214.0, + "step": 15458 + }, + { + "epoch": 1.6976718647045903, + "grad_norm": 2.437133312225342, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7027171850204468, + "num_tokens": 397046836.0, + "step": 15459 + }, + { + "epoch": 1.697781682407204, + "grad_norm": 2.439262866973877, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7389624118804932, + "num_tokens": 397068836.0, + "step": 15460 + }, + { + "epoch": 1.6978915001098178, + "grad_norm": 2.4816575050354004, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7301547527313232, + "num_tokens": 397091474.0, + "step": 15461 + }, + { + "epoch": 1.6980013178124314, + "grad_norm": 2.2350358963012695, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7138709425926208, + "num_tokens": 397116773.0, + "step": 15462 + }, + { + "epoch": 1.698111135515045, + "grad_norm": 2.5631191730499268, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7241544723510742, + "num_tokens": 397139673.0, + "step": 15463 + }, + { + "epoch": 1.6982209532176586, + "grad_norm": 1.8640265464782715, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7128287553787231, + "num_tokens": 397171688.0, + "step": 15464 + }, + { + "epoch": 1.6983307709202724, + "grad_norm": 2.3366007804870605, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7208438515663147, + "num_tokens": 397194001.0, + "step": 15465 + }, + { + "epoch": 1.6984405886228862, + "grad_norm": 2.2616937160491943, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7131752967834473, + "num_tokens": 397219020.0, + "step": 15466 + }, + { + "epoch": 1.6985504063254997, + "grad_norm": 2.2938427925109863, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.709028959274292, + "num_tokens": 397244018.0, + "step": 15467 + }, + { + "epoch": 1.6986602240281132, + "grad_norm": 2.2841057777404785, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7258200645446777, + "num_tokens": 397270784.0, + "step": 15468 + }, + { + "epoch": 1.698770041730727, + "grad_norm": 2.026426076889038, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7196086049079895, + "num_tokens": 397304527.0, + "step": 15469 + }, + { + "epoch": 1.6988798594333407, + "grad_norm": 1.974131464958191, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7083182334899902, + "num_tokens": 397334884.0, + "step": 15470 + }, + { + "epoch": 1.6989896771359543, + "grad_norm": 2.3643665313720703, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7126127481460571, + "num_tokens": 397358582.0, + "step": 15471 + }, + { + "epoch": 1.6990994948385678, + "grad_norm": 2.0256829261779785, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7069204449653625, + "num_tokens": 397388133.0, + "step": 15472 + }, + { + "epoch": 1.6992093125411816, + "grad_norm": 2.0687596797943115, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6947947144508362, + "num_tokens": 397416182.0, + "step": 15473 + }, + { + "epoch": 1.6993191302437953, + "grad_norm": 2.7770354747772217, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7140707969665527, + "num_tokens": 397434105.0, + "step": 15474 + }, + { + "epoch": 1.699428947946409, + "grad_norm": 2.256972551345825, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.720978319644928, + "num_tokens": 397458992.0, + "step": 15475 + }, + { + "epoch": 1.6995387656490226, + "grad_norm": 2.16560959815979, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7334436178207397, + "num_tokens": 397484178.0, + "step": 15476 + }, + { + "epoch": 1.6996485833516362, + "grad_norm": 2.37488055229187, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7061989903450012, + "num_tokens": 397508770.0, + "step": 15477 + }, + { + "epoch": 1.69975840105425, + "grad_norm": 2.365415096282959, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7223970890045166, + "num_tokens": 397530163.0, + "step": 15478 + }, + { + "epoch": 1.6998682187568637, + "grad_norm": 2.500242233276367, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7386972904205322, + "num_tokens": 397550398.0, + "step": 15479 + }, + { + "epoch": 1.6999780364594774, + "grad_norm": 2.409252405166626, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7187501788139343, + "num_tokens": 397573538.0, + "step": 15480 + }, + { + "epoch": 1.700087854162091, + "grad_norm": 2.596860647201538, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.747580885887146, + "num_tokens": 397592503.0, + "step": 15481 + }, + { + "epoch": 1.7001976718647045, + "grad_norm": 2.202313184738159, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.714677095413208, + "num_tokens": 397619221.0, + "step": 15482 + }, + { + "epoch": 1.7003074895673183, + "grad_norm": 2.3918144702911377, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7377570867538452, + "num_tokens": 397641524.0, + "step": 15483 + }, + { + "epoch": 1.700417307269932, + "grad_norm": 2.186861753463745, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7091670036315918, + "num_tokens": 397669156.0, + "step": 15484 + }, + { + "epoch": 1.7005271249725455, + "grad_norm": 1.931952714920044, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7191634774208069, + "num_tokens": 397699628.0, + "step": 15485 + }, + { + "epoch": 1.700636942675159, + "grad_norm": 1.92821204662323, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7048673629760742, + "num_tokens": 397733146.0, + "step": 15486 + }, + { + "epoch": 1.7007467603777728, + "grad_norm": 2.2449657917022705, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7232990860939026, + "num_tokens": 397758367.0, + "step": 15487 + }, + { + "epoch": 1.7008565780803866, + "grad_norm": 2.432731866836548, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7255937457084656, + "num_tokens": 397781101.0, + "step": 15488 + }, + { + "epoch": 1.7009663957830004, + "grad_norm": 2.3290297985076904, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7294350862503052, + "num_tokens": 397804411.0, + "step": 15489 + }, + { + "epoch": 1.7010762134856139, + "grad_norm": 2.43208909034729, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7315081357955933, + "num_tokens": 397827532.0, + "step": 15490 + }, + { + "epoch": 1.7011860311882274, + "grad_norm": 2.0335206985473633, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6943310499191284, + "num_tokens": 397861148.0, + "step": 15491 + }, + { + "epoch": 1.7012958488908412, + "grad_norm": 2.153287887573242, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7261474132537842, + "num_tokens": 397885422.0, + "step": 15492 + }, + { + "epoch": 1.701405666593455, + "grad_norm": 2.1517422199249268, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.735114574432373, + "num_tokens": 397909941.0, + "step": 15493 + }, + { + "epoch": 1.7015154842960687, + "grad_norm": 2.1885805130004883, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7322984337806702, + "num_tokens": 397937861.0, + "step": 15494 + }, + { + "epoch": 1.7016253019986822, + "grad_norm": 2.1711628437042236, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7106422185897827, + "num_tokens": 397965375.0, + "step": 15495 + }, + { + "epoch": 1.7017351197012958, + "grad_norm": 2.3049845695495605, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7127689123153687, + "num_tokens": 397990106.0, + "step": 15496 + }, + { + "epoch": 1.7018449374039095, + "grad_norm": 2.220304250717163, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7278162240982056, + "num_tokens": 398015268.0, + "step": 15497 + }, + { + "epoch": 1.7019547551065233, + "grad_norm": 2.0855650901794434, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.7124254703521729, + "num_tokens": 398043910.0, + "step": 15498 + }, + { + "epoch": 1.7020645728091368, + "grad_norm": 2.2350707054138184, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7491707801818848, + "num_tokens": 398068528.0, + "step": 15499 + }, + { + "epoch": 1.7021743905117503, + "grad_norm": 2.0967960357666016, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6981164813041687, + "num_tokens": 398095862.0, + "step": 15500 + }, + { + "epoch": 1.702284208214364, + "grad_norm": 2.2246525287628174, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7184674739837646, + "num_tokens": 398121067.0, + "step": 15501 + }, + { + "epoch": 1.7023940259169779, + "grad_norm": 2.26741886138916, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.744204044342041, + "num_tokens": 398144618.0, + "step": 15502 + }, + { + "epoch": 1.7025038436195916, + "grad_norm": 2.0144128799438477, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6882859468460083, + "num_tokens": 398175992.0, + "step": 15503 + }, + { + "epoch": 1.7026136613222052, + "grad_norm": 2.3676843643188477, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7146138548851013, + "num_tokens": 398201303.0, + "step": 15504 + }, + { + "epoch": 1.7027234790248187, + "grad_norm": 2.6885945796966553, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7288114428520203, + "num_tokens": 398220703.0, + "step": 15505 + }, + { + "epoch": 1.7028332967274324, + "grad_norm": 2.647529363632202, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.728998064994812, + "num_tokens": 398241765.0, + "step": 15506 + }, + { + "epoch": 1.7029431144300462, + "grad_norm": 2.0361592769622803, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7145919799804688, + "num_tokens": 398271229.0, + "step": 15507 + }, + { + "epoch": 1.7030529321326597, + "grad_norm": 2.2925336360931396, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7160870432853699, + "num_tokens": 398297803.0, + "step": 15508 + }, + { + "epoch": 1.7031627498352735, + "grad_norm": 2.4165830612182617, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7421835660934448, + "num_tokens": 398320262.0, + "step": 15509 + }, + { + "epoch": 1.703272567537887, + "grad_norm": 2.2564642429351807, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7421557307243347, + "num_tokens": 398343646.0, + "step": 15510 + }, + { + "epoch": 1.7033823852405008, + "grad_norm": 1.860937237739563, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7162908315658569, + "num_tokens": 398379102.0, + "step": 15511 + }, + { + "epoch": 1.7034922029431145, + "grad_norm": 2.484975576400757, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.730888843536377, + "num_tokens": 398402797.0, + "step": 15512 + }, + { + "epoch": 1.703602020645728, + "grad_norm": 2.6704721450805664, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.736861526966095, + "num_tokens": 398420493.0, + "step": 15513 + }, + { + "epoch": 1.7037118383483416, + "grad_norm": 2.293030261993408, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7232111096382141, + "num_tokens": 398447162.0, + "step": 15514 + }, + { + "epoch": 1.7038216560509554, + "grad_norm": 2.2527706623077393, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7347408533096313, + "num_tokens": 398470865.0, + "step": 15515 + }, + { + "epoch": 1.7039314737535691, + "grad_norm": 2.2152366638183594, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7063255310058594, + "num_tokens": 398496365.0, + "step": 15516 + }, + { + "epoch": 1.7040412914561829, + "grad_norm": 2.3443334102630615, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7212116122245789, + "num_tokens": 398518056.0, + "step": 15517 + }, + { + "epoch": 1.7041511091587964, + "grad_norm": 2.318082809448242, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7151334285736084, + "num_tokens": 398542848.0, + "step": 15518 + }, + { + "epoch": 1.70426092686141, + "grad_norm": 2.5183987617492676, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7278095483779907, + "num_tokens": 398563003.0, + "step": 15519 + }, + { + "epoch": 1.7043707445640237, + "grad_norm": 2.2802326679229736, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.713874340057373, + "num_tokens": 398589251.0, + "step": 15520 + }, + { + "epoch": 1.7044805622666375, + "grad_norm": 2.298933506011963, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7185113430023193, + "num_tokens": 398613040.0, + "step": 15521 + }, + { + "epoch": 1.704590379969251, + "grad_norm": 2.1630167961120605, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7257254719734192, + "num_tokens": 398638069.0, + "step": 15522 + }, + { + "epoch": 1.7047001976718648, + "grad_norm": 2.13725209236145, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7117965817451477, + "num_tokens": 398667419.0, + "step": 15523 + }, + { + "epoch": 1.7048100153744783, + "grad_norm": 2.2882020473480225, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.71108478307724, + "num_tokens": 398693614.0, + "step": 15524 + }, + { + "epoch": 1.704919833077092, + "grad_norm": 2.010089159011841, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7194375991821289, + "num_tokens": 398726043.0, + "step": 15525 + }, + { + "epoch": 1.7050296507797058, + "grad_norm": 2.4965953826904297, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.724551796913147, + "num_tokens": 398746790.0, + "step": 15526 + }, + { + "epoch": 1.7051394684823193, + "grad_norm": 2.3032264709472656, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7152423858642578, + "num_tokens": 398771731.0, + "step": 15527 + }, + { + "epoch": 1.7052492861849329, + "grad_norm": 3.042301654815674, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7336361408233643, + "num_tokens": 398794476.0, + "step": 15528 + }, + { + "epoch": 1.7053591038875466, + "grad_norm": 2.3973259925842285, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7127273678779602, + "num_tokens": 398815864.0, + "step": 15529 + }, + { + "epoch": 1.7054689215901604, + "grad_norm": 2.205291986465454, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7168065309524536, + "num_tokens": 398842255.0, + "step": 15530 + }, + { + "epoch": 1.7055787392927741, + "grad_norm": 2.376892328262329, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.727260410785675, + "num_tokens": 398865426.0, + "step": 15531 + }, + { + "epoch": 1.7056885569953877, + "grad_norm": 1.9500337839126587, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7110926508903503, + "num_tokens": 398899496.0, + "step": 15532 + }, + { + "epoch": 1.7057983746980012, + "grad_norm": 2.283630132675171, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7191988229751587, + "num_tokens": 398924521.0, + "step": 15533 + }, + { + "epoch": 1.705908192400615, + "grad_norm": 2.2220630645751953, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7137292623519897, + "num_tokens": 398951032.0, + "step": 15534 + }, + { + "epoch": 1.7060180101032287, + "grad_norm": 2.3854191303253174, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7329453229904175, + "num_tokens": 398975665.0, + "step": 15535 + }, + { + "epoch": 1.7061278278058423, + "grad_norm": 2.35841703414917, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6919810771942139, + "num_tokens": 399007637.0, + "step": 15536 + }, + { + "epoch": 1.7062376455084558, + "grad_norm": 2.539093017578125, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7314817905426025, + "num_tokens": 399027514.0, + "step": 15537 + }, + { + "epoch": 1.7063474632110696, + "grad_norm": 2.3528687953948975, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7161259651184082, + "num_tokens": 399051673.0, + "step": 15538 + }, + { + "epoch": 1.7064572809136833, + "grad_norm": 2.2869889736175537, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7197961807250977, + "num_tokens": 399075601.0, + "step": 15539 + }, + { + "epoch": 1.706567098616297, + "grad_norm": 2.479926109313965, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.71223384141922, + "num_tokens": 399097744.0, + "step": 15540 + }, + { + "epoch": 1.7066769163189106, + "grad_norm": 2.0336344242095947, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7350714206695557, + "num_tokens": 399129355.0, + "step": 15541 + }, + { + "epoch": 1.7067867340215241, + "grad_norm": 2.1223597526550293, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.722088098526001, + "num_tokens": 399159287.0, + "step": 15542 + }, + { + "epoch": 1.706896551724138, + "grad_norm": 2.392948865890503, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7285915017127991, + "num_tokens": 399181756.0, + "step": 15543 + }, + { + "epoch": 1.7070063694267517, + "grad_norm": 2.327075958251953, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7065566778182983, + "num_tokens": 399207213.0, + "step": 15544 + }, + { + "epoch": 1.7071161871293654, + "grad_norm": 2.443795919418335, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7192639112472534, + "num_tokens": 399229975.0, + "step": 15545 + }, + { + "epoch": 1.707226004831979, + "grad_norm": 2.4126081466674805, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7234488725662231, + "num_tokens": 399251863.0, + "step": 15546 + }, + { + "epoch": 1.7073358225345925, + "grad_norm": 2.03031587600708, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7415029406547546, + "num_tokens": 399277715.0, + "step": 15547 + }, + { + "epoch": 1.7074456402372062, + "grad_norm": 2.283762216567993, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7294101119041443, + "num_tokens": 399301963.0, + "step": 15548 + }, + { + "epoch": 1.70755545793982, + "grad_norm": 2.175158739089966, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7012050151824951, + "num_tokens": 399329763.0, + "step": 15549 + }, + { + "epoch": 1.7076652756424335, + "grad_norm": 2.438857316970825, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.735916793346405, + "num_tokens": 399349482.0, + "step": 15550 + }, + { + "epoch": 1.707775093345047, + "grad_norm": 2.5333778858184814, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7352344989776611, + "num_tokens": 399370216.0, + "step": 15551 + }, + { + "epoch": 1.7078849110476608, + "grad_norm": 2.133061408996582, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7191108465194702, + "num_tokens": 399395937.0, + "step": 15552 + }, + { + "epoch": 1.7079947287502746, + "grad_norm": 2.1759109497070312, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7121679186820984, + "num_tokens": 399422684.0, + "step": 15553 + }, + { + "epoch": 1.7081045464528883, + "grad_norm": 2.4606873989105225, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7350460886955261, + "num_tokens": 399444791.0, + "step": 15554 + }, + { + "epoch": 1.7082143641555019, + "grad_norm": 2.4052040576934814, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7374894618988037, + "num_tokens": 399469131.0, + "step": 15555 + }, + { + "epoch": 1.7083241818581154, + "grad_norm": 2.42433500289917, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7274057269096375, + "num_tokens": 399490561.0, + "step": 15556 + }, + { + "epoch": 1.7084339995607292, + "grad_norm": 2.57766056060791, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7354381680488586, + "num_tokens": 399509387.0, + "step": 15557 + }, + { + "epoch": 1.708543817263343, + "grad_norm": 2.2330849170684814, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7052007913589478, + "num_tokens": 399534311.0, + "step": 15558 + }, + { + "epoch": 1.7086536349659565, + "grad_norm": 2.269416332244873, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7188529968261719, + "num_tokens": 399560870.0, + "step": 15559 + }, + { + "epoch": 1.7087634526685702, + "grad_norm": 2.328051805496216, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7344464063644409, + "num_tokens": 399584791.0, + "step": 15560 + }, + { + "epoch": 1.7088732703711838, + "grad_norm": 2.2915384769439697, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7216439843177795, + "num_tokens": 399609223.0, + "step": 15561 + }, + { + "epoch": 1.7089830880737975, + "grad_norm": 2.389472007751465, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.717442512512207, + "num_tokens": 399632452.0, + "step": 15562 + }, + { + "epoch": 1.7090929057764113, + "grad_norm": 2.152707099914551, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7156289219856262, + "num_tokens": 399657841.0, + "step": 15563 + }, + { + "epoch": 1.7092027234790248, + "grad_norm": 1.9549239873886108, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7116025686264038, + "num_tokens": 399687979.0, + "step": 15564 + }, + { + "epoch": 1.7093125411816383, + "grad_norm": 2.186251640319824, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7408530712127686, + "num_tokens": 399713055.0, + "step": 15565 + }, + { + "epoch": 1.709422358884252, + "grad_norm": 2.3403706550598145, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7304010987281799, + "num_tokens": 399737292.0, + "step": 15566 + }, + { + "epoch": 1.7095321765868658, + "grad_norm": 1.9603525400161743, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7137050032615662, + "num_tokens": 399769659.0, + "step": 15567 + }, + { + "epoch": 1.7096419942894796, + "grad_norm": 2.231085777282715, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7010895013809204, + "num_tokens": 399796984.0, + "step": 15568 + }, + { + "epoch": 1.7097518119920931, + "grad_norm": 2.1830995082855225, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6871156096458435, + "num_tokens": 399823414.0, + "step": 15569 + }, + { + "epoch": 1.7098616296947067, + "grad_norm": 2.4892280101776123, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7303751707077026, + "num_tokens": 399845811.0, + "step": 15570 + }, + { + "epoch": 1.7099714473973204, + "grad_norm": 2.595979928970337, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7282368540763855, + "num_tokens": 399865802.0, + "step": 15571 + }, + { + "epoch": 1.7100812650999342, + "grad_norm": 2.0306882858276367, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7014721632003784, + "num_tokens": 399897091.0, + "step": 15572 + }, + { + "epoch": 1.7101910828025477, + "grad_norm": 2.883465051651001, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7411884069442749, + "num_tokens": 399918043.0, + "step": 15573 + }, + { + "epoch": 1.7103009005051615, + "grad_norm": 2.270904541015625, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7156491279602051, + "num_tokens": 399943500.0, + "step": 15574 + }, + { + "epoch": 1.710410718207775, + "grad_norm": 2.582712411880493, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7270315289497375, + "num_tokens": 399964184.0, + "step": 15575 + }, + { + "epoch": 1.7105205359103888, + "grad_norm": 2.501189947128296, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7394846677780151, + "num_tokens": 399986313.0, + "step": 15576 + }, + { + "epoch": 1.7106303536130025, + "grad_norm": 2.159487247467041, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7264586091041565, + "num_tokens": 400013669.0, + "step": 15577 + }, + { + "epoch": 1.710740171315616, + "grad_norm": 2.1013898849487305, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7254467606544495, + "num_tokens": 400042396.0, + "step": 15578 + }, + { + "epoch": 1.7108499890182296, + "grad_norm": 2.0995657444000244, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7358075976371765, + "num_tokens": 400069118.0, + "step": 15579 + }, + { + "epoch": 1.7109598067208434, + "grad_norm": 2.267415761947632, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7235226035118103, + "num_tokens": 400095051.0, + "step": 15580 + }, + { + "epoch": 1.7110696244234571, + "grad_norm": 2.3454177379608154, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7279704809188843, + "num_tokens": 400120074.0, + "step": 15581 + }, + { + "epoch": 1.7111794421260709, + "grad_norm": 2.5097179412841797, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7308648824691772, + "num_tokens": 400142030.0, + "step": 15582 + }, + { + "epoch": 1.7112892598286844, + "grad_norm": 2.1778242588043213, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.729656457901001, + "num_tokens": 400168062.0, + "step": 15583 + }, + { + "epoch": 1.711399077531298, + "grad_norm": 2.1622345447540283, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7193156480789185, + "num_tokens": 400195164.0, + "step": 15584 + }, + { + "epoch": 1.7115088952339117, + "grad_norm": 2.3333427906036377, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7106016278266907, + "num_tokens": 400221934.0, + "step": 15585 + }, + { + "epoch": 1.7116187129365255, + "grad_norm": 2.0026354789733887, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7163823843002319, + "num_tokens": 400251505.0, + "step": 15586 + }, + { + "epoch": 1.711728530639139, + "grad_norm": 2.1063990592956543, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7366325855255127, + "num_tokens": 400277783.0, + "step": 15587 + }, + { + "epoch": 1.7118383483417525, + "grad_norm": 2.091348171234131, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7515444755554199, + "num_tokens": 400303875.0, + "step": 15588 + }, + { + "epoch": 1.7119481660443663, + "grad_norm": 2.3769116401672363, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6882784366607666, + "num_tokens": 400329812.0, + "step": 15589 + }, + { + "epoch": 1.71205798374698, + "grad_norm": 2.1618118286132812, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7201477289199829, + "num_tokens": 400361104.0, + "step": 15590 + }, + { + "epoch": 1.7121678014495938, + "grad_norm": 2.190342664718628, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7443128824234009, + "num_tokens": 400386654.0, + "step": 15591 + }, + { + "epoch": 1.7122776191522073, + "grad_norm": 2.113766670227051, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.718046247959137, + "num_tokens": 400413319.0, + "step": 15592 + }, + { + "epoch": 1.7123874368548209, + "grad_norm": 2.43583607673645, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7136853337287903, + "num_tokens": 400436423.0, + "step": 15593 + }, + { + "epoch": 1.7124972545574346, + "grad_norm": 2.171541929244995, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7388969659805298, + "num_tokens": 400461020.0, + "step": 15594 + }, + { + "epoch": 1.7126070722600484, + "grad_norm": 2.386941432952881, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7342832088470459, + "num_tokens": 400481978.0, + "step": 15595 + }, + { + "epoch": 1.7127168899626621, + "grad_norm": 2.0425021648406982, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7138882875442505, + "num_tokens": 400510289.0, + "step": 15596 + }, + { + "epoch": 1.7128267076652757, + "grad_norm": 2.030498743057251, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7267752885818481, + "num_tokens": 400540073.0, + "step": 15597 + }, + { + "epoch": 1.7129365253678892, + "grad_norm": 2.109001636505127, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6974039673805237, + "num_tokens": 400570975.0, + "step": 15598 + }, + { + "epoch": 1.713046343070503, + "grad_norm": 2.354703903198242, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.69759202003479, + "num_tokens": 400595201.0, + "step": 15599 + }, + { + "epoch": 1.7131561607731167, + "grad_norm": 2.1618759632110596, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7376832365989685, + "num_tokens": 400623186.0, + "step": 15600 + }, + { + "epoch": 1.7132659784757303, + "grad_norm": 2.3585004806518555, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7082908749580383, + "num_tokens": 400648441.0, + "step": 15601 + }, + { + "epoch": 1.7133757961783438, + "grad_norm": 2.2488458156585693, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7190495729446411, + "num_tokens": 400674685.0, + "step": 15602 + }, + { + "epoch": 1.7134856138809575, + "grad_norm": 2.2159924507141113, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7167967557907104, + "num_tokens": 400700814.0, + "step": 15603 + }, + { + "epoch": 1.7135954315835713, + "grad_norm": 2.1017653942108154, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.726502537727356, + "num_tokens": 400732313.0, + "step": 15604 + }, + { + "epoch": 1.713705249286185, + "grad_norm": 1.7966054677963257, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6993975639343262, + "num_tokens": 400771553.0, + "step": 15605 + }, + { + "epoch": 1.7138150669887986, + "grad_norm": 2.304194688796997, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7096877098083496, + "num_tokens": 400796676.0, + "step": 15606 + }, + { + "epoch": 1.7139248846914121, + "grad_norm": 2.138808488845825, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7291474342346191, + "num_tokens": 400823602.0, + "step": 15607 + }, + { + "epoch": 1.7140347023940259, + "grad_norm": 2.0938847064971924, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7270157337188721, + "num_tokens": 400849735.0, + "step": 15608 + }, + { + "epoch": 1.7141445200966396, + "grad_norm": 2.247473955154419, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6812118291854858, + "num_tokens": 400877245.0, + "step": 15609 + }, + { + "epoch": 1.7142543377992534, + "grad_norm": 2.237003803253174, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7119888663291931, + "num_tokens": 400903647.0, + "step": 15610 + }, + { + "epoch": 1.714364155501867, + "grad_norm": 2.3442742824554443, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.728334903717041, + "num_tokens": 400926848.0, + "step": 15611 + }, + { + "epoch": 1.7144739732044805, + "grad_norm": 2.5297751426696777, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6956729888916016, + "num_tokens": 400948718.0, + "step": 15612 + }, + { + "epoch": 1.7145837909070942, + "grad_norm": 2.2255706787109375, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7122752666473389, + "num_tokens": 400975439.0, + "step": 15613 + }, + { + "epoch": 1.714693608609708, + "grad_norm": 2.326385259628296, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7132799029350281, + "num_tokens": 401001783.0, + "step": 15614 + }, + { + "epoch": 1.7148034263123215, + "grad_norm": 2.2799384593963623, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7156822681427002, + "num_tokens": 401029488.0, + "step": 15615 + }, + { + "epoch": 1.714913244014935, + "grad_norm": 2.3834645748138428, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7389583587646484, + "num_tokens": 401051500.0, + "step": 15616 + }, + { + "epoch": 1.7150230617175488, + "grad_norm": 2.107930898666382, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7167960405349731, + "num_tokens": 401081068.0, + "step": 15617 + }, + { + "epoch": 1.7151328794201626, + "grad_norm": 2.019713878631592, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7038881778717041, + "num_tokens": 401111482.0, + "step": 15618 + }, + { + "epoch": 1.7152426971227763, + "grad_norm": 2.2896361351013184, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7345402240753174, + "num_tokens": 401136178.0, + "step": 15619 + }, + { + "epoch": 1.7153525148253899, + "grad_norm": 2.2744057178497314, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7363288402557373, + "num_tokens": 401160619.0, + "step": 15620 + }, + { + "epoch": 1.7154623325280034, + "grad_norm": 2.487971782684326, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7242690324783325, + "num_tokens": 401181628.0, + "step": 15621 + }, + { + "epoch": 1.7155721502306172, + "grad_norm": 1.9506562948226929, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7032787799835205, + "num_tokens": 401215334.0, + "step": 15622 + }, + { + "epoch": 1.715681967933231, + "grad_norm": 2.1543173789978027, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7283012866973877, + "num_tokens": 401241772.0, + "step": 15623 + }, + { + "epoch": 1.7157917856358444, + "grad_norm": 2.171055316925049, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7225261926651001, + "num_tokens": 401270252.0, + "step": 15624 + }, + { + "epoch": 1.7159016033384582, + "grad_norm": 1.957911491394043, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7326871156692505, + "num_tokens": 401299090.0, + "step": 15625 + }, + { + "epoch": 1.7160114210410717, + "grad_norm": 2.2638707160949707, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7122231721878052, + "num_tokens": 401323587.0, + "step": 15626 + }, + { + "epoch": 1.7161212387436855, + "grad_norm": 2.6311354637145996, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7231279611587524, + "num_tokens": 401344628.0, + "step": 15627 + }, + { + "epoch": 1.7162310564462993, + "grad_norm": 2.513460159301758, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.720826268196106, + "num_tokens": 401367247.0, + "step": 15628 + }, + { + "epoch": 1.7163408741489128, + "grad_norm": 2.514047622680664, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7295764088630676, + "num_tokens": 401389643.0, + "step": 15629 + }, + { + "epoch": 1.7164506918515263, + "grad_norm": 2.254331588745117, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7155098915100098, + "num_tokens": 401416577.0, + "step": 15630 + }, + { + "epoch": 1.71656050955414, + "grad_norm": 2.6545562744140625, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7326139211654663, + "num_tokens": 401437127.0, + "step": 15631 + }, + { + "epoch": 1.7166703272567538, + "grad_norm": 1.9939873218536377, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7175617218017578, + "num_tokens": 401465656.0, + "step": 15632 + }, + { + "epoch": 1.7167801449593676, + "grad_norm": 2.1159470081329346, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7043355107307434, + "num_tokens": 401494627.0, + "step": 15633 + }, + { + "epoch": 1.7168899626619811, + "grad_norm": 2.2153947353363037, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7311985492706299, + "num_tokens": 401519652.0, + "step": 15634 + }, + { + "epoch": 1.7169997803645947, + "grad_norm": 2.182048797607422, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7130972146987915, + "num_tokens": 401546176.0, + "step": 15635 + }, + { + "epoch": 1.7171095980672084, + "grad_norm": 2.1181094646453857, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.697867751121521, + "num_tokens": 401574581.0, + "step": 15636 + }, + { + "epoch": 1.7172194157698222, + "grad_norm": 2.161231756210327, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7211949825286865, + "num_tokens": 401601682.0, + "step": 15637 + }, + { + "epoch": 1.7173292334724357, + "grad_norm": 2.6289587020874023, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7352467179298401, + "num_tokens": 401620354.0, + "step": 15638 + }, + { + "epoch": 1.7174390511750495, + "grad_norm": 2.1725120544433594, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7018967270851135, + "num_tokens": 401647178.0, + "step": 15639 + }, + { + "epoch": 1.717548868877663, + "grad_norm": 2.0068161487579346, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6902037858963013, + "num_tokens": 401677068.0, + "step": 15640 + }, + { + "epoch": 1.7176586865802768, + "grad_norm": 2.367976427078247, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7226871252059937, + "num_tokens": 401701476.0, + "step": 15641 + }, + { + "epoch": 1.7177685042828905, + "grad_norm": 2.0230634212493896, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.711804211139679, + "num_tokens": 401732177.0, + "step": 15642 + }, + { + "epoch": 1.717878321985504, + "grad_norm": 2.154174566268921, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7276031374931335, + "num_tokens": 401759889.0, + "step": 15643 + }, + { + "epoch": 1.7179881396881176, + "grad_norm": 2.4943244457244873, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7223044633865356, + "num_tokens": 401781377.0, + "step": 15644 + }, + { + "epoch": 1.7180979573907313, + "grad_norm": 2.0014591217041016, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7001339197158813, + "num_tokens": 401812067.0, + "step": 15645 + }, + { + "epoch": 1.718207775093345, + "grad_norm": 1.9007110595703125, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7183324098587036, + "num_tokens": 401845679.0, + "step": 15646 + }, + { + "epoch": 1.7183175927959589, + "grad_norm": 2.5547292232513428, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7375401258468628, + "num_tokens": 401866706.0, + "step": 15647 + }, + { + "epoch": 1.7184274104985724, + "grad_norm": 2.336249828338623, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7151230573654175, + "num_tokens": 401889919.0, + "step": 15648 + }, + { + "epoch": 1.718537228201186, + "grad_norm": 2.5068533420562744, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7517508268356323, + "num_tokens": 401909780.0, + "step": 15649 + }, + { + "epoch": 1.7186470459037997, + "grad_norm": 2.1782283782958984, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7362586855888367, + "num_tokens": 401935549.0, + "step": 15650 + }, + { + "epoch": 1.7187568636064134, + "grad_norm": 2.367548942565918, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7107494473457336, + "num_tokens": 401960626.0, + "step": 15651 + }, + { + "epoch": 1.718866681309027, + "grad_norm": 2.330254316329956, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.724073588848114, + "num_tokens": 401983643.0, + "step": 15652 + }, + { + "epoch": 1.7189764990116405, + "grad_norm": 2.273531913757324, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7065914273262024, + "num_tokens": 402008597.0, + "step": 15653 + }, + { + "epoch": 1.7190863167142543, + "grad_norm": 2.394676685333252, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7229511737823486, + "num_tokens": 402030437.0, + "step": 15654 + }, + { + "epoch": 1.719196134416868, + "grad_norm": 2.0969414710998535, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7170760631561279, + "num_tokens": 402057263.0, + "step": 15655 + }, + { + "epoch": 1.7193059521194818, + "grad_norm": 2.1529364585876465, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.701451301574707, + "num_tokens": 402085229.0, + "step": 15656 + }, + { + "epoch": 1.7194157698220953, + "grad_norm": 2.314333200454712, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.721430778503418, + "num_tokens": 402109503.0, + "step": 15657 + }, + { + "epoch": 1.7195255875247089, + "grad_norm": 2.242352247238159, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7194119691848755, + "num_tokens": 402135848.0, + "step": 15658 + }, + { + "epoch": 1.7196354052273226, + "grad_norm": 2.1386220455169678, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7110897302627563, + "num_tokens": 402164939.0, + "step": 15659 + }, + { + "epoch": 1.7197452229299364, + "grad_norm": 2.1059622764587402, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7053285837173462, + "num_tokens": 402193427.0, + "step": 15660 + }, + { + "epoch": 1.7198550406325501, + "grad_norm": 2.395285129547119, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7206165194511414, + "num_tokens": 402216529.0, + "step": 15661 + }, + { + "epoch": 1.7199648583351637, + "grad_norm": 2.5742850303649902, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.736641526222229, + "num_tokens": 402236301.0, + "step": 15662 + }, + { + "epoch": 1.7200746760377772, + "grad_norm": 2.154259204864502, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7182109951972961, + "num_tokens": 402263212.0, + "step": 15663 + }, + { + "epoch": 1.720184493740391, + "grad_norm": 2.0722885131835938, + "learning_rate": 1e-06, + "loss": 1.0753, + "mean_token_accuracy": 0.6903727650642395, + "num_tokens": 402294588.0, + "step": 15664 + }, + { + "epoch": 1.7202943114430047, + "grad_norm": 2.396061420440674, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7165876626968384, + "num_tokens": 402317912.0, + "step": 15665 + }, + { + "epoch": 1.7204041291456182, + "grad_norm": 2.2973785400390625, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7040826082229614, + "num_tokens": 402342595.0, + "step": 15666 + }, + { + "epoch": 1.7205139468482318, + "grad_norm": 1.8807718753814697, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6929645538330078, + "num_tokens": 402376271.0, + "step": 15667 + }, + { + "epoch": 1.7206237645508455, + "grad_norm": 2.5665438175201416, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7270371913909912, + "num_tokens": 402397376.0, + "step": 15668 + }, + { + "epoch": 1.7207335822534593, + "grad_norm": 1.960516333580017, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.712316632270813, + "num_tokens": 402431054.0, + "step": 15669 + }, + { + "epoch": 1.720843399956073, + "grad_norm": 2.2810680866241455, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.725045919418335, + "num_tokens": 402455389.0, + "step": 15670 + }, + { + "epoch": 1.7209532176586866, + "grad_norm": 2.217233180999756, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7183989882469177, + "num_tokens": 402482561.0, + "step": 15671 + }, + { + "epoch": 1.7210630353613001, + "grad_norm": 2.0094211101531982, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7081077694892883, + "num_tokens": 402514428.0, + "step": 15672 + }, + { + "epoch": 1.7211728530639139, + "grad_norm": 2.1624343395233154, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7139174938201904, + "num_tokens": 402544322.0, + "step": 15673 + }, + { + "epoch": 1.7212826707665276, + "grad_norm": 2.238973617553711, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.707733690738678, + "num_tokens": 402570702.0, + "step": 15674 + }, + { + "epoch": 1.7213924884691414, + "grad_norm": 2.1721935272216797, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.6992968916893005, + "num_tokens": 402600165.0, + "step": 15675 + }, + { + "epoch": 1.721502306171755, + "grad_norm": 2.187981605529785, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7309627532958984, + "num_tokens": 402628139.0, + "step": 15676 + }, + { + "epoch": 1.7216121238743685, + "grad_norm": 2.251936197280884, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7428423166275024, + "num_tokens": 402653325.0, + "step": 15677 + }, + { + "epoch": 1.7217219415769822, + "grad_norm": 2.4545605182647705, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7227784395217896, + "num_tokens": 402674018.0, + "step": 15678 + }, + { + "epoch": 1.721831759279596, + "grad_norm": 2.461561679840088, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7375044822692871, + "num_tokens": 402694491.0, + "step": 15679 + }, + { + "epoch": 1.7219415769822095, + "grad_norm": 2.2574660778045654, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7230007648468018, + "num_tokens": 402720591.0, + "step": 15680 + }, + { + "epoch": 1.722051394684823, + "grad_norm": 2.0026862621307373, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7017675638198853, + "num_tokens": 402752990.0, + "step": 15681 + }, + { + "epoch": 1.7221612123874368, + "grad_norm": 2.4133923053741455, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7517492175102234, + "num_tokens": 402773735.0, + "step": 15682 + }, + { + "epoch": 1.7222710300900506, + "grad_norm": 2.088780403137207, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7129495739936829, + "num_tokens": 402802407.0, + "step": 15683 + }, + { + "epoch": 1.7223808477926643, + "grad_norm": 2.175389528274536, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7142159938812256, + "num_tokens": 402829262.0, + "step": 15684 + }, + { + "epoch": 1.7224906654952779, + "grad_norm": 2.3912928104400635, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7141345143318176, + "num_tokens": 402852154.0, + "step": 15685 + }, + { + "epoch": 1.7226004831978914, + "grad_norm": 2.40139102935791, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6969050168991089, + "num_tokens": 402876127.0, + "step": 15686 + }, + { + "epoch": 1.7227103009005051, + "grad_norm": 2.525418758392334, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7335548400878906, + "num_tokens": 402898456.0, + "step": 15687 + }, + { + "epoch": 1.722820118603119, + "grad_norm": 2.1820526123046875, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7156039476394653, + "num_tokens": 402924230.0, + "step": 15688 + }, + { + "epoch": 1.7229299363057324, + "grad_norm": 2.074159622192383, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7098537087440491, + "num_tokens": 402953141.0, + "step": 15689 + }, + { + "epoch": 1.7230397540083462, + "grad_norm": 2.19397234916687, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7186212539672852, + "num_tokens": 402980157.0, + "step": 15690 + }, + { + "epoch": 1.7231495717109597, + "grad_norm": 1.942722201347351, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7193586230278015, + "num_tokens": 403010038.0, + "step": 15691 + }, + { + "epoch": 1.7232593894135735, + "grad_norm": 2.4200735092163086, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7383347153663635, + "num_tokens": 403033853.0, + "step": 15692 + }, + { + "epoch": 1.7233692071161872, + "grad_norm": 2.349740743637085, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7336596250534058, + "num_tokens": 403058025.0, + "step": 15693 + }, + { + "epoch": 1.7234790248188008, + "grad_norm": 2.198760986328125, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7082370519638062, + "num_tokens": 403085332.0, + "step": 15694 + }, + { + "epoch": 1.7235888425214143, + "grad_norm": 2.3307206630706787, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6936804056167603, + "num_tokens": 403118709.0, + "step": 15695 + }, + { + "epoch": 1.723698660224028, + "grad_norm": 2.1875112056732178, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7211259007453918, + "num_tokens": 403144682.0, + "step": 15696 + }, + { + "epoch": 1.7238084779266418, + "grad_norm": 2.174029588699341, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7092790007591248, + "num_tokens": 403173013.0, + "step": 15697 + }, + { + "epoch": 1.7239182956292556, + "grad_norm": 2.29341197013855, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.724632740020752, + "num_tokens": 403198335.0, + "step": 15698 + }, + { + "epoch": 1.7240281133318691, + "grad_norm": 2.095961570739746, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7307658195495605, + "num_tokens": 403225109.0, + "step": 15699 + }, + { + "epoch": 1.7241379310344827, + "grad_norm": 2.1688759326934814, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7168703079223633, + "num_tokens": 403251830.0, + "step": 15700 + }, + { + "epoch": 1.7242477487370964, + "grad_norm": 2.081247091293335, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7105826139450073, + "num_tokens": 403280885.0, + "step": 15701 + }, + { + "epoch": 1.7243575664397102, + "grad_norm": 2.035994052886963, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.702035665512085, + "num_tokens": 403311296.0, + "step": 15702 + }, + { + "epoch": 1.7244673841423237, + "grad_norm": 2.587848663330078, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7428485751152039, + "num_tokens": 403330300.0, + "step": 15703 + }, + { + "epoch": 1.7245772018449375, + "grad_norm": 2.504201650619507, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.740139365196228, + "num_tokens": 403351370.0, + "step": 15704 + }, + { + "epoch": 1.724687019547551, + "grad_norm": 2.497955799102783, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7387092113494873, + "num_tokens": 403371548.0, + "step": 15705 + }, + { + "epoch": 1.7247968372501647, + "grad_norm": 2.3377325534820557, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.7015314102172852, + "num_tokens": 403397452.0, + "step": 15706 + }, + { + "epoch": 1.7249066549527785, + "grad_norm": 2.6948082447052, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7224174737930298, + "num_tokens": 403416960.0, + "step": 15707 + }, + { + "epoch": 1.725016472655392, + "grad_norm": 2.2626354694366455, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7444883584976196, + "num_tokens": 403441613.0, + "step": 15708 + }, + { + "epoch": 1.7251262903580056, + "grad_norm": 2.4016823768615723, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7361233234405518, + "num_tokens": 403464566.0, + "step": 15709 + }, + { + "epoch": 1.7252361080606193, + "grad_norm": 2.3990774154663086, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7104684114456177, + "num_tokens": 403486767.0, + "step": 15710 + }, + { + "epoch": 1.725345925763233, + "grad_norm": 2.213310718536377, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7282963991165161, + "num_tokens": 403512752.0, + "step": 15711 + }, + { + "epoch": 1.7254557434658468, + "grad_norm": 2.464647054672241, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7408573627471924, + "num_tokens": 403533700.0, + "step": 15712 + }, + { + "epoch": 1.7255655611684604, + "grad_norm": 2.5374608039855957, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7282137274742126, + "num_tokens": 403554929.0, + "step": 15713 + }, + { + "epoch": 1.725675378871074, + "grad_norm": 2.1670141220092773, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7102392911911011, + "num_tokens": 403581373.0, + "step": 15714 + }, + { + "epoch": 1.7257851965736877, + "grad_norm": 2.4864261150360107, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.725968599319458, + "num_tokens": 403602386.0, + "step": 15715 + }, + { + "epoch": 1.7258950142763014, + "grad_norm": 2.3182570934295654, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7135862708091736, + "num_tokens": 403626646.0, + "step": 15716 + }, + { + "epoch": 1.726004831978915, + "grad_norm": 2.5982611179351807, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7547115087509155, + "num_tokens": 403645890.0, + "step": 15717 + }, + { + "epoch": 1.7261146496815285, + "grad_norm": 2.3300509452819824, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7491738200187683, + "num_tokens": 403667609.0, + "step": 15718 + }, + { + "epoch": 1.7262244673841423, + "grad_norm": 2.2353086471557617, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7188671231269836, + "num_tokens": 403691318.0, + "step": 15719 + }, + { + "epoch": 1.726334285086756, + "grad_norm": 2.1643667221069336, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7093578577041626, + "num_tokens": 403717966.0, + "step": 15720 + }, + { + "epoch": 1.7264441027893698, + "grad_norm": 2.0800836086273193, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7157893180847168, + "num_tokens": 403747225.0, + "step": 15721 + }, + { + "epoch": 1.7265539204919833, + "grad_norm": 2.318488836288452, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7134044766426086, + "num_tokens": 403772398.0, + "step": 15722 + }, + { + "epoch": 1.7266637381945968, + "grad_norm": 2.194382429122925, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7121524810791016, + "num_tokens": 403801221.0, + "step": 15723 + }, + { + "epoch": 1.7267735558972106, + "grad_norm": 2.360400676727295, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.740268349647522, + "num_tokens": 403823526.0, + "step": 15724 + }, + { + "epoch": 1.7268833735998244, + "grad_norm": 2.3600966930389404, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7133181691169739, + "num_tokens": 403847802.0, + "step": 15725 + }, + { + "epoch": 1.7269931913024381, + "grad_norm": 2.084561586380005, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7304773330688477, + "num_tokens": 403874965.0, + "step": 15726 + }, + { + "epoch": 1.7271030090050516, + "grad_norm": 2.3875584602355957, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7354788780212402, + "num_tokens": 403897545.0, + "step": 15727 + }, + { + "epoch": 1.7272128267076652, + "grad_norm": 2.129693031311035, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7197630405426025, + "num_tokens": 403926024.0, + "step": 15728 + }, + { + "epoch": 1.727322644410279, + "grad_norm": 2.294354200363159, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7206562757492065, + "num_tokens": 403950177.0, + "step": 15729 + }, + { + "epoch": 1.7274324621128927, + "grad_norm": 2.393634796142578, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7192888259887695, + "num_tokens": 403973979.0, + "step": 15730 + }, + { + "epoch": 1.7275422798155062, + "grad_norm": 2.589444637298584, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7140651345252991, + "num_tokens": 403994384.0, + "step": 15731 + }, + { + "epoch": 1.7276520975181198, + "grad_norm": 2.2101664543151855, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.700231671333313, + "num_tokens": 404022434.0, + "step": 15732 + }, + { + "epoch": 1.7277619152207335, + "grad_norm": 2.0527570247650146, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6942915916442871, + "num_tokens": 404052129.0, + "step": 15733 + }, + { + "epoch": 1.7278717329233473, + "grad_norm": 1.969900131225586, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7318037748336792, + "num_tokens": 404082830.0, + "step": 15734 + }, + { + "epoch": 1.727981550625961, + "grad_norm": 2.2867705821990967, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7237319946289062, + "num_tokens": 404106691.0, + "step": 15735 + }, + { + "epoch": 1.7280913683285746, + "grad_norm": 2.0509703159332275, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7087361812591553, + "num_tokens": 404136581.0, + "step": 15736 + }, + { + "epoch": 1.728201186031188, + "grad_norm": 2.1077888011932373, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6857733130455017, + "num_tokens": 404167585.0, + "step": 15737 + }, + { + "epoch": 1.7283110037338019, + "grad_norm": 2.1401782035827637, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7008336782455444, + "num_tokens": 404198033.0, + "step": 15738 + }, + { + "epoch": 1.7284208214364156, + "grad_norm": 2.1944289207458496, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6941671371459961, + "num_tokens": 404224715.0, + "step": 15739 + }, + { + "epoch": 1.7285306391390292, + "grad_norm": 2.264529228210449, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7097630500793457, + "num_tokens": 404249840.0, + "step": 15740 + }, + { + "epoch": 1.728640456841643, + "grad_norm": 2.126073122024536, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7147301435470581, + "num_tokens": 404278108.0, + "step": 15741 + }, + { + "epoch": 1.7287502745442564, + "grad_norm": 2.383446455001831, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6956197023391724, + "num_tokens": 404301764.0, + "step": 15742 + }, + { + "epoch": 1.7288600922468702, + "grad_norm": 1.936922550201416, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.698452353477478, + "num_tokens": 404334827.0, + "step": 15743 + }, + { + "epoch": 1.728969909949484, + "grad_norm": 2.0354108810424805, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6946412324905396, + "num_tokens": 404368572.0, + "step": 15744 + }, + { + "epoch": 1.7290797276520975, + "grad_norm": 2.0968239307403564, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7158845663070679, + "num_tokens": 404398899.0, + "step": 15745 + }, + { + "epoch": 1.729189545354711, + "grad_norm": 2.2021331787109375, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7206042408943176, + "num_tokens": 404424703.0, + "step": 15746 + }, + { + "epoch": 1.7292993630573248, + "grad_norm": 2.2007803916931152, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7339110374450684, + "num_tokens": 404449517.0, + "step": 15747 + }, + { + "epoch": 1.7294091807599385, + "grad_norm": 2.084077835083008, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.704057514667511, + "num_tokens": 404477243.0, + "step": 15748 + }, + { + "epoch": 1.7295189984625523, + "grad_norm": 1.9559266567230225, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7057595252990723, + "num_tokens": 404508925.0, + "step": 15749 + }, + { + "epoch": 1.7296288161651658, + "grad_norm": 2.1453471183776855, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7181112766265869, + "num_tokens": 404536451.0, + "step": 15750 + }, + { + "epoch": 1.7297386338677794, + "grad_norm": 2.082825183868408, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7121851444244385, + "num_tokens": 404565430.0, + "step": 15751 + }, + { + "epoch": 1.7298484515703931, + "grad_norm": 1.9924800395965576, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6920900940895081, + "num_tokens": 404598281.0, + "step": 15752 + }, + { + "epoch": 1.7299582692730069, + "grad_norm": 2.0797839164733887, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6994156241416931, + "num_tokens": 404628211.0, + "step": 15753 + }, + { + "epoch": 1.7300680869756204, + "grad_norm": 2.086261510848999, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7328619956970215, + "num_tokens": 404656796.0, + "step": 15754 + }, + { + "epoch": 1.7301779046782342, + "grad_norm": 1.8988416194915771, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6913789510726929, + "num_tokens": 404690043.0, + "step": 15755 + }, + { + "epoch": 1.7302877223808477, + "grad_norm": 2.1345062255859375, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7310625910758972, + "num_tokens": 404716587.0, + "step": 15756 + }, + { + "epoch": 1.7303975400834615, + "grad_norm": 2.277261257171631, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7242838144302368, + "num_tokens": 404741412.0, + "step": 15757 + }, + { + "epoch": 1.7305073577860752, + "grad_norm": 2.2559287548065186, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7204697132110596, + "num_tokens": 404768351.0, + "step": 15758 + }, + { + "epoch": 1.7306171754886888, + "grad_norm": 2.0332541465759277, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6885111927986145, + "num_tokens": 404800911.0, + "step": 15759 + }, + { + "epoch": 1.7307269931913023, + "grad_norm": 2.395414113998413, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7278726100921631, + "num_tokens": 404824015.0, + "step": 15760 + }, + { + "epoch": 1.730836810893916, + "grad_norm": 2.1862776279449463, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7083959579467773, + "num_tokens": 404853602.0, + "step": 15761 + }, + { + "epoch": 1.7309466285965298, + "grad_norm": 2.2149155139923096, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7227436304092407, + "num_tokens": 404881644.0, + "step": 15762 + }, + { + "epoch": 1.7310564462991436, + "grad_norm": 2.5769314765930176, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7497672438621521, + "num_tokens": 404900789.0, + "step": 15763 + }, + { + "epoch": 1.731166264001757, + "grad_norm": 2.1032395362854004, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7122858762741089, + "num_tokens": 404930457.0, + "step": 15764 + }, + { + "epoch": 1.7312760817043706, + "grad_norm": 2.059154748916626, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7000802755355835, + "num_tokens": 404960959.0, + "step": 15765 + }, + { + "epoch": 1.7313858994069844, + "grad_norm": 2.3755722045898438, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7412758469581604, + "num_tokens": 404984774.0, + "step": 15766 + }, + { + "epoch": 1.7314957171095982, + "grad_norm": 2.110466957092285, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7372128367424011, + "num_tokens": 405010901.0, + "step": 15767 + }, + { + "epoch": 1.7316055348122117, + "grad_norm": 2.0794544219970703, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7162909507751465, + "num_tokens": 405044556.0, + "step": 15768 + }, + { + "epoch": 1.7317153525148252, + "grad_norm": 2.0698695182800293, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7034485340118408, + "num_tokens": 405078710.0, + "step": 15769 + }, + { + "epoch": 1.731825170217439, + "grad_norm": 2.4702646732330322, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7103887796401978, + "num_tokens": 405101742.0, + "step": 15770 + }, + { + "epoch": 1.7319349879200527, + "grad_norm": 2.3090484142303467, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7345969676971436, + "num_tokens": 405125635.0, + "step": 15771 + }, + { + "epoch": 1.7320448056226665, + "grad_norm": 2.291712760925293, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7534550428390503, + "num_tokens": 405150102.0, + "step": 15772 + }, + { + "epoch": 1.73215462332528, + "grad_norm": 2.065403461456299, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7127126455307007, + "num_tokens": 405179300.0, + "step": 15773 + }, + { + "epoch": 1.7322644410278936, + "grad_norm": 2.0252325534820557, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7217519879341125, + "num_tokens": 405211338.0, + "step": 15774 + }, + { + "epoch": 1.7323742587305073, + "grad_norm": 2.3252737522125244, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7048788070678711, + "num_tokens": 405235805.0, + "step": 15775 + }, + { + "epoch": 1.732484076433121, + "grad_norm": 2.2294504642486572, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7434976100921631, + "num_tokens": 405261109.0, + "step": 15776 + }, + { + "epoch": 1.7325938941357348, + "grad_norm": 2.0107343196868896, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7319759130477905, + "num_tokens": 405290232.0, + "step": 15777 + }, + { + "epoch": 1.7327037118383484, + "grad_norm": 2.4066989421844482, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7034310102462769, + "num_tokens": 405314117.0, + "step": 15778 + }, + { + "epoch": 1.732813529540962, + "grad_norm": 2.597909450531006, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7241288423538208, + "num_tokens": 405334488.0, + "step": 15779 + }, + { + "epoch": 1.7329233472435757, + "grad_norm": 2.455470561981201, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7120930552482605, + "num_tokens": 405357321.0, + "step": 15780 + }, + { + "epoch": 1.7330331649461894, + "grad_norm": 2.1470947265625, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7214950323104858, + "num_tokens": 405383966.0, + "step": 15781 + }, + { + "epoch": 1.733142982648803, + "grad_norm": 2.225435495376587, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7234790921211243, + "num_tokens": 405409449.0, + "step": 15782 + }, + { + "epoch": 1.7332528003514165, + "grad_norm": 2.275146245956421, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7369382381439209, + "num_tokens": 405433020.0, + "step": 15783 + }, + { + "epoch": 1.7333626180540302, + "grad_norm": 2.571540117263794, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7496159076690674, + "num_tokens": 405452193.0, + "step": 15784 + }, + { + "epoch": 1.733472435756644, + "grad_norm": 2.1129086017608643, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7410200834274292, + "num_tokens": 405478949.0, + "step": 15785 + }, + { + "epoch": 1.7335822534592578, + "grad_norm": 2.235302686691284, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7574683427810669, + "num_tokens": 405502049.0, + "step": 15786 + }, + { + "epoch": 1.7336920711618713, + "grad_norm": 2.314432382583618, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7119224071502686, + "num_tokens": 405525876.0, + "step": 15787 + }, + { + "epoch": 1.7338018888644848, + "grad_norm": 2.0079116821289062, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6909967660903931, + "num_tokens": 405558370.0, + "step": 15788 + }, + { + "epoch": 1.7339117065670986, + "grad_norm": 2.4155824184417725, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7318140268325806, + "num_tokens": 405580197.0, + "step": 15789 + }, + { + "epoch": 1.7340215242697123, + "grad_norm": 2.0264811515808105, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6967142820358276, + "num_tokens": 405613022.0, + "step": 15790 + }, + { + "epoch": 1.734131341972326, + "grad_norm": 2.3776445388793945, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7170315980911255, + "num_tokens": 405635336.0, + "step": 15791 + }, + { + "epoch": 1.7342411596749396, + "grad_norm": 2.1878929138183594, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7098040580749512, + "num_tokens": 405661598.0, + "step": 15792 + }, + { + "epoch": 1.7343509773775532, + "grad_norm": 2.1255085468292236, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7249225974082947, + "num_tokens": 405689989.0, + "step": 15793 + }, + { + "epoch": 1.734460795080167, + "grad_norm": 2.3420519828796387, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7117017507553101, + "num_tokens": 405714471.0, + "step": 15794 + }, + { + "epoch": 1.7345706127827807, + "grad_norm": 1.9315563440322876, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7012125253677368, + "num_tokens": 405749496.0, + "step": 15795 + }, + { + "epoch": 1.7346804304853942, + "grad_norm": 2.283649206161499, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7199913263320923, + "num_tokens": 405775405.0, + "step": 15796 + }, + { + "epoch": 1.7347902481880078, + "grad_norm": 2.5312771797180176, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7174098491668701, + "num_tokens": 405796198.0, + "step": 15797 + }, + { + "epoch": 1.7349000658906215, + "grad_norm": 2.155658483505249, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7325426340103149, + "num_tokens": 405823166.0, + "step": 15798 + }, + { + "epoch": 1.7350098835932353, + "grad_norm": 2.637213706970215, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7215020060539246, + "num_tokens": 405842728.0, + "step": 15799 + }, + { + "epoch": 1.735119701295849, + "grad_norm": 2.443676233291626, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7460494041442871, + "num_tokens": 405863028.0, + "step": 15800 + }, + { + "epoch": 1.7352295189984626, + "grad_norm": 2.0134403705596924, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.732296884059906, + "num_tokens": 405892273.0, + "step": 15801 + }, + { + "epoch": 1.735339336701076, + "grad_norm": 2.0792038440704346, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7236424684524536, + "num_tokens": 405921358.0, + "step": 15802 + }, + { + "epoch": 1.7354491544036899, + "grad_norm": 2.1641409397125244, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7324167490005493, + "num_tokens": 405949306.0, + "step": 15803 + }, + { + "epoch": 1.7355589721063036, + "grad_norm": 2.1393120288848877, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.726300835609436, + "num_tokens": 405976437.0, + "step": 15804 + }, + { + "epoch": 1.7356687898089171, + "grad_norm": 2.163532018661499, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.70365971326828, + "num_tokens": 406004001.0, + "step": 15805 + }, + { + "epoch": 1.735778607511531, + "grad_norm": 2.3951799869537354, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7144277691841125, + "num_tokens": 406027445.0, + "step": 15806 + }, + { + "epoch": 1.7358884252141444, + "grad_norm": 2.817880868911743, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7423091530799866, + "num_tokens": 406044082.0, + "step": 15807 + }, + { + "epoch": 1.7359982429167582, + "grad_norm": 2.435603380203247, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7337484955787659, + "num_tokens": 406066474.0, + "step": 15808 + }, + { + "epoch": 1.736108060619372, + "grad_norm": 2.2158889770507812, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7131001949310303, + "num_tokens": 406092285.0, + "step": 15809 + }, + { + "epoch": 1.7362178783219855, + "grad_norm": 2.1124956607818604, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7228211760520935, + "num_tokens": 406120230.0, + "step": 15810 + }, + { + "epoch": 1.736327696024599, + "grad_norm": 2.0422592163085938, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7212892770767212, + "num_tokens": 406151740.0, + "step": 15811 + }, + { + "epoch": 1.7364375137272128, + "grad_norm": 2.1044201850891113, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7187560796737671, + "num_tokens": 406180417.0, + "step": 15812 + }, + { + "epoch": 1.7365473314298265, + "grad_norm": 2.122980833053589, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7129172682762146, + "num_tokens": 406209472.0, + "step": 15813 + }, + { + "epoch": 1.7366571491324403, + "grad_norm": 2.0499608516693115, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.6983178853988647, + "num_tokens": 406239393.0, + "step": 15814 + }, + { + "epoch": 1.7367669668350538, + "grad_norm": 2.0600485801696777, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.726370632648468, + "num_tokens": 406267982.0, + "step": 15815 + }, + { + "epoch": 1.7368767845376674, + "grad_norm": 2.3760457038879395, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7176963090896606, + "num_tokens": 406292024.0, + "step": 15816 + }, + { + "epoch": 1.7369866022402811, + "grad_norm": 2.281005382537842, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.722774863243103, + "num_tokens": 406316736.0, + "step": 15817 + }, + { + "epoch": 1.7370964199428949, + "grad_norm": 2.3565945625305176, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7063226699829102, + "num_tokens": 406339661.0, + "step": 15818 + }, + { + "epoch": 1.7372062376455084, + "grad_norm": 2.560633420944214, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7316661477088928, + "num_tokens": 406359347.0, + "step": 15819 + }, + { + "epoch": 1.7373160553481222, + "grad_norm": 2.0428316593170166, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6950390338897705, + "num_tokens": 406392316.0, + "step": 15820 + }, + { + "epoch": 1.7374258730507357, + "grad_norm": 2.459777593612671, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7159550786018372, + "num_tokens": 406413423.0, + "step": 15821 + }, + { + "epoch": 1.7375356907533495, + "grad_norm": 2.0289385318756104, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7213959097862244, + "num_tokens": 406445822.0, + "step": 15822 + }, + { + "epoch": 1.7376455084559632, + "grad_norm": 2.2636144161224365, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.742549479007721, + "num_tokens": 406471098.0, + "step": 15823 + }, + { + "epoch": 1.7377553261585768, + "grad_norm": 2.2782251834869385, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7122379541397095, + "num_tokens": 406496608.0, + "step": 15824 + }, + { + "epoch": 1.7378651438611903, + "grad_norm": 2.4580636024475098, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7039594650268555, + "num_tokens": 406522058.0, + "step": 15825 + }, + { + "epoch": 1.737974961563804, + "grad_norm": 2.032787322998047, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7070072293281555, + "num_tokens": 406554323.0, + "step": 15826 + }, + { + "epoch": 1.7380847792664178, + "grad_norm": 2.1064884662628174, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7269392013549805, + "num_tokens": 406581593.0, + "step": 15827 + }, + { + "epoch": 1.7381945969690316, + "grad_norm": 2.347419023513794, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7194417715072632, + "num_tokens": 406607440.0, + "step": 15828 + }, + { + "epoch": 1.738304414671645, + "grad_norm": 2.063586711883545, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.72926926612854, + "num_tokens": 406634638.0, + "step": 15829 + }, + { + "epoch": 1.7384142323742586, + "grad_norm": 2.019144296646118, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.720420777797699, + "num_tokens": 406664656.0, + "step": 15830 + }, + { + "epoch": 1.7385240500768724, + "grad_norm": 2.3170032501220703, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7265058755874634, + "num_tokens": 406691031.0, + "step": 15831 + }, + { + "epoch": 1.7386338677794861, + "grad_norm": 2.2581827640533447, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7154483199119568, + "num_tokens": 406717263.0, + "step": 15832 + }, + { + "epoch": 1.7387436854820997, + "grad_norm": 2.4113166332244873, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7188968658447266, + "num_tokens": 406740314.0, + "step": 15833 + }, + { + "epoch": 1.7388535031847132, + "grad_norm": 2.522549629211426, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7359141111373901, + "num_tokens": 406761631.0, + "step": 15834 + }, + { + "epoch": 1.738963320887327, + "grad_norm": 2.268387794494629, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7043898105621338, + "num_tokens": 406788376.0, + "step": 15835 + }, + { + "epoch": 1.7390731385899407, + "grad_norm": 2.155449628829956, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7113239169120789, + "num_tokens": 406817116.0, + "step": 15836 + }, + { + "epoch": 1.7391829562925545, + "grad_norm": 2.1671948432922363, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7087926864624023, + "num_tokens": 406845150.0, + "step": 15837 + }, + { + "epoch": 1.739292773995168, + "grad_norm": 2.3574464321136475, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7257161140441895, + "num_tokens": 406870275.0, + "step": 15838 + }, + { + "epoch": 1.7394025916977816, + "grad_norm": 2.4348819255828857, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7213692665100098, + "num_tokens": 406892959.0, + "step": 15839 + }, + { + "epoch": 1.7395124094003953, + "grad_norm": 2.1222569942474365, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6929346323013306, + "num_tokens": 406920780.0, + "step": 15840 + }, + { + "epoch": 1.739622227103009, + "grad_norm": 2.4637789726257324, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7316334247589111, + "num_tokens": 406943373.0, + "step": 15841 + }, + { + "epoch": 1.7397320448056228, + "grad_norm": 2.1896133422851562, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7257115840911865, + "num_tokens": 406969136.0, + "step": 15842 + }, + { + "epoch": 1.7398418625082364, + "grad_norm": 2.1203768253326416, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7139273881912231, + "num_tokens": 406997724.0, + "step": 15843 + }, + { + "epoch": 1.73995168021085, + "grad_norm": 2.3006694316864014, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7161461114883423, + "num_tokens": 407024873.0, + "step": 15844 + }, + { + "epoch": 1.7400614979134637, + "grad_norm": 1.9117307662963867, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7136040925979614, + "num_tokens": 407056875.0, + "step": 15845 + }, + { + "epoch": 1.7401713156160774, + "grad_norm": 2.375281810760498, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7421653270721436, + "num_tokens": 407079198.0, + "step": 15846 + }, + { + "epoch": 1.740281133318691, + "grad_norm": 2.2621748447418213, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6924964785575867, + "num_tokens": 407104766.0, + "step": 15847 + }, + { + "epoch": 1.7403909510213045, + "grad_norm": 2.2876861095428467, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7201018333435059, + "num_tokens": 407130152.0, + "step": 15848 + }, + { + "epoch": 1.7405007687239182, + "grad_norm": 2.023965358734131, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.761529803276062, + "num_tokens": 407157836.0, + "step": 15849 + }, + { + "epoch": 1.740610586426532, + "grad_norm": 2.108699321746826, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7238333225250244, + "num_tokens": 407184131.0, + "step": 15850 + }, + { + "epoch": 1.7407204041291457, + "grad_norm": 2.0500054359436035, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7181330323219299, + "num_tokens": 407214512.0, + "step": 15851 + }, + { + "epoch": 1.7408302218317593, + "grad_norm": 2.274111270904541, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7128061056137085, + "num_tokens": 407240635.0, + "step": 15852 + }, + { + "epoch": 1.7409400395343728, + "grad_norm": 2.5836434364318848, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7201796770095825, + "num_tokens": 407262167.0, + "step": 15853 + }, + { + "epoch": 1.7410498572369866, + "grad_norm": 2.038593053817749, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7005085349082947, + "num_tokens": 407290657.0, + "step": 15854 + }, + { + "epoch": 1.7411596749396003, + "grad_norm": 2.2633745670318604, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.733798086643219, + "num_tokens": 407314857.0, + "step": 15855 + }, + { + "epoch": 1.741269492642214, + "grad_norm": 2.425166606903076, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7386325001716614, + "num_tokens": 407336547.0, + "step": 15856 + }, + { + "epoch": 1.7413793103448276, + "grad_norm": 2.2050838470458984, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7163712978363037, + "num_tokens": 407362905.0, + "step": 15857 + }, + { + "epoch": 1.7414891280474412, + "grad_norm": 2.1192266941070557, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7030468583106995, + "num_tokens": 407392183.0, + "step": 15858 + }, + { + "epoch": 1.741598945750055, + "grad_norm": 2.4749596118927, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7261812686920166, + "num_tokens": 407414012.0, + "step": 15859 + }, + { + "epoch": 1.7417087634526687, + "grad_norm": 2.1475980281829834, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.735542893409729, + "num_tokens": 407440241.0, + "step": 15860 + }, + { + "epoch": 1.7418185811552822, + "grad_norm": 2.3587992191314697, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7217953205108643, + "num_tokens": 407463209.0, + "step": 15861 + }, + { + "epoch": 1.7419283988578957, + "grad_norm": 2.9804742336273193, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7364351749420166, + "num_tokens": 407479471.0, + "step": 15862 + }, + { + "epoch": 1.7420382165605095, + "grad_norm": 2.1218793392181396, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.720429539680481, + "num_tokens": 407506526.0, + "step": 15863 + }, + { + "epoch": 1.7421480342631233, + "grad_norm": 1.9922637939453125, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7152962684631348, + "num_tokens": 407538476.0, + "step": 15864 + }, + { + "epoch": 1.742257851965737, + "grad_norm": 2.3608455657958984, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7186180949211121, + "num_tokens": 407561774.0, + "step": 15865 + }, + { + "epoch": 1.7423676696683505, + "grad_norm": 2.245044469833374, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7176914811134338, + "num_tokens": 407587111.0, + "step": 15866 + }, + { + "epoch": 1.742477487370964, + "grad_norm": 1.9696744680404663, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7044302225112915, + "num_tokens": 407621211.0, + "step": 15867 + }, + { + "epoch": 1.7425873050735778, + "grad_norm": 2.2258834838867188, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7273764610290527, + "num_tokens": 407645369.0, + "step": 15868 + }, + { + "epoch": 1.7426971227761916, + "grad_norm": 2.144526243209839, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6752868294715881, + "num_tokens": 407676047.0, + "step": 15869 + }, + { + "epoch": 1.7428069404788051, + "grad_norm": 2.5345888137817383, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7322779893875122, + "num_tokens": 407697789.0, + "step": 15870 + }, + { + "epoch": 1.742916758181419, + "grad_norm": 1.9939593076705933, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7325813174247742, + "num_tokens": 407727300.0, + "step": 15871 + }, + { + "epoch": 1.7430265758840324, + "grad_norm": 2.2462611198425293, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7230325937271118, + "num_tokens": 407752375.0, + "step": 15872 + }, + { + "epoch": 1.7431363935866462, + "grad_norm": 2.5208911895751953, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7099426984786987, + "num_tokens": 407774507.0, + "step": 15873 + }, + { + "epoch": 1.74324621128926, + "grad_norm": 2.409104585647583, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7333787679672241, + "num_tokens": 407796719.0, + "step": 15874 + }, + { + "epoch": 1.7433560289918735, + "grad_norm": 2.038252592086792, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7471393346786499, + "num_tokens": 407824911.0, + "step": 15875 + }, + { + "epoch": 1.743465846694487, + "grad_norm": 2.382415771484375, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7287552356719971, + "num_tokens": 407846220.0, + "step": 15876 + }, + { + "epoch": 1.7435756643971008, + "grad_norm": 2.1727919578552246, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7044104337692261, + "num_tokens": 407874767.0, + "step": 15877 + }, + { + "epoch": 1.7436854820997145, + "grad_norm": 2.3747551441192627, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7197792530059814, + "num_tokens": 407897994.0, + "step": 15878 + }, + { + "epoch": 1.7437952998023283, + "grad_norm": 2.365882396697998, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7177178859710693, + "num_tokens": 407921858.0, + "step": 15879 + }, + { + "epoch": 1.7439051175049418, + "grad_norm": 2.2270963191986084, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7140449285507202, + "num_tokens": 407948352.0, + "step": 15880 + }, + { + "epoch": 1.7440149352075554, + "grad_norm": 2.076176166534424, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7122780680656433, + "num_tokens": 407978435.0, + "step": 15881 + }, + { + "epoch": 1.744124752910169, + "grad_norm": 2.3353750705718994, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7089754939079285, + "num_tokens": 408003535.0, + "step": 15882 + }, + { + "epoch": 1.7442345706127829, + "grad_norm": 2.3735270500183105, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.711308479309082, + "num_tokens": 408027687.0, + "step": 15883 + }, + { + "epoch": 1.7443443883153964, + "grad_norm": 2.1284260749816895, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7086539268493652, + "num_tokens": 408055228.0, + "step": 15884 + }, + { + "epoch": 1.7444542060180102, + "grad_norm": 2.6272716522216797, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7444037199020386, + "num_tokens": 408073399.0, + "step": 15885 + }, + { + "epoch": 1.7445640237206237, + "grad_norm": 2.479388475418091, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7211599946022034, + "num_tokens": 408095485.0, + "step": 15886 + }, + { + "epoch": 1.7446738414232374, + "grad_norm": 2.289755344390869, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7114561200141907, + "num_tokens": 408120663.0, + "step": 15887 + }, + { + "epoch": 1.7447836591258512, + "grad_norm": 2.1792213916778564, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7336427569389343, + "num_tokens": 408148601.0, + "step": 15888 + }, + { + "epoch": 1.7448934768284647, + "grad_norm": 2.367150068283081, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7363248467445374, + "num_tokens": 408171484.0, + "step": 15889 + }, + { + "epoch": 1.7450032945310783, + "grad_norm": 2.228097677230835, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.746289849281311, + "num_tokens": 408195712.0, + "step": 15890 + }, + { + "epoch": 1.745113112233692, + "grad_norm": 2.1787731647491455, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7195006608963013, + "num_tokens": 408224940.0, + "step": 15891 + }, + { + "epoch": 1.7452229299363058, + "grad_norm": 2.399674892425537, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7117330431938171, + "num_tokens": 408248960.0, + "step": 15892 + }, + { + "epoch": 1.7453327476389195, + "grad_norm": 2.089641809463501, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6979119777679443, + "num_tokens": 408279425.0, + "step": 15893 + }, + { + "epoch": 1.745442565341533, + "grad_norm": 2.308400869369507, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7100040912628174, + "num_tokens": 408305026.0, + "step": 15894 + }, + { + "epoch": 1.7455523830441466, + "grad_norm": 2.1780526638031006, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7232714295387268, + "num_tokens": 408331513.0, + "step": 15895 + }, + { + "epoch": 1.7456622007467604, + "grad_norm": 2.1764447689056396, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7060674428939819, + "num_tokens": 408360016.0, + "step": 15896 + }, + { + "epoch": 1.7457720184493741, + "grad_norm": 2.14664626121521, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7459762096405029, + "num_tokens": 408385551.0, + "step": 15897 + }, + { + "epoch": 1.7458818361519877, + "grad_norm": 2.285025119781494, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7014394402503967, + "num_tokens": 408409116.0, + "step": 15898 + }, + { + "epoch": 1.7459916538546012, + "grad_norm": 2.1205813884735107, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7144032716751099, + "num_tokens": 408439164.0, + "step": 15899 + }, + { + "epoch": 1.746101471557215, + "grad_norm": 2.091089963912964, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7013486623764038, + "num_tokens": 408472183.0, + "step": 15900 + }, + { + "epoch": 1.7462112892598287, + "grad_norm": 2.30991268157959, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7332075834274292, + "num_tokens": 408494926.0, + "step": 15901 + }, + { + "epoch": 1.7463211069624425, + "grad_norm": 2.1771340370178223, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7113134264945984, + "num_tokens": 408521673.0, + "step": 15902 + }, + { + "epoch": 1.746430924665056, + "grad_norm": 2.077291250228882, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7089101672172546, + "num_tokens": 408551253.0, + "step": 15903 + }, + { + "epoch": 1.7465407423676695, + "grad_norm": 2.1791629791259766, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7229592800140381, + "num_tokens": 408578601.0, + "step": 15904 + }, + { + "epoch": 1.7466505600702833, + "grad_norm": 2.494143009185791, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7350955009460449, + "num_tokens": 408599329.0, + "step": 15905 + }, + { + "epoch": 1.746760377772897, + "grad_norm": 2.2229270935058594, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7105631828308105, + "num_tokens": 408627394.0, + "step": 15906 + }, + { + "epoch": 1.7468701954755108, + "grad_norm": 2.398132085800171, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7131024599075317, + "num_tokens": 408652283.0, + "step": 15907 + }, + { + "epoch": 1.7469800131781243, + "grad_norm": 2.296813726425171, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7279964685440063, + "num_tokens": 408675259.0, + "step": 15908 + }, + { + "epoch": 1.7470898308807379, + "grad_norm": 2.66577410697937, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7303718328475952, + "num_tokens": 408693582.0, + "step": 15909 + }, + { + "epoch": 1.7471996485833516, + "grad_norm": 2.5732409954071045, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7410599589347839, + "num_tokens": 408713057.0, + "step": 15910 + }, + { + "epoch": 1.7473094662859654, + "grad_norm": 1.9592114686965942, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7092931866645813, + "num_tokens": 408745559.0, + "step": 15911 + }, + { + "epoch": 1.747419283988579, + "grad_norm": 2.0833961963653564, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7300411462783813, + "num_tokens": 408771568.0, + "step": 15912 + }, + { + "epoch": 1.7475291016911925, + "grad_norm": 2.2156333923339844, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.6998171210289001, + "num_tokens": 408797665.0, + "step": 15913 + }, + { + "epoch": 1.7476389193938062, + "grad_norm": 2.2301435470581055, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7001776099205017, + "num_tokens": 408823361.0, + "step": 15914 + }, + { + "epoch": 1.74774873709642, + "grad_norm": 2.0113744735717773, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7005500793457031, + "num_tokens": 408856682.0, + "step": 15915 + }, + { + "epoch": 1.7478585547990337, + "grad_norm": 2.0460286140441895, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7050641775131226, + "num_tokens": 408885591.0, + "step": 15916 + }, + { + "epoch": 1.7479683725016473, + "grad_norm": 2.190497875213623, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7246435880661011, + "num_tokens": 408910193.0, + "step": 15917 + }, + { + "epoch": 1.7480781902042608, + "grad_norm": 2.2812047004699707, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7228049039840698, + "num_tokens": 408936216.0, + "step": 15918 + }, + { + "epoch": 1.7481880079068746, + "grad_norm": 2.04081130027771, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.726291298866272, + "num_tokens": 408966321.0, + "step": 15919 + }, + { + "epoch": 1.7482978256094883, + "grad_norm": 2.0299618244171143, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7174056768417358, + "num_tokens": 408994376.0, + "step": 15920 + }, + { + "epoch": 1.7484076433121019, + "grad_norm": 2.295450448989868, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7357462644577026, + "num_tokens": 409018904.0, + "step": 15921 + }, + { + "epoch": 1.7485174610147156, + "grad_norm": 2.140089988708496, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7237035036087036, + "num_tokens": 409045079.0, + "step": 15922 + }, + { + "epoch": 1.7486272787173291, + "grad_norm": 2.5529003143310547, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7114644050598145, + "num_tokens": 409066171.0, + "step": 15923 + }, + { + "epoch": 1.748737096419943, + "grad_norm": 2.5701940059661865, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7035565376281738, + "num_tokens": 409086373.0, + "step": 15924 + }, + { + "epoch": 1.7488469141225567, + "grad_norm": 2.2293860912323, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6972514390945435, + "num_tokens": 409114621.0, + "step": 15925 + }, + { + "epoch": 1.7489567318251702, + "grad_norm": 2.053158760070801, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7155055403709412, + "num_tokens": 409143687.0, + "step": 15926 + }, + { + "epoch": 1.7490665495277837, + "grad_norm": 2.526853322982788, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7247036099433899, + "num_tokens": 409165807.0, + "step": 15927 + }, + { + "epoch": 1.7491763672303975, + "grad_norm": 2.221946954727173, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7100789546966553, + "num_tokens": 409190899.0, + "step": 15928 + }, + { + "epoch": 1.7492861849330112, + "grad_norm": 2.1647353172302246, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7190686464309692, + "num_tokens": 409220195.0, + "step": 15929 + }, + { + "epoch": 1.749396002635625, + "grad_norm": 2.2816317081451416, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7069234848022461, + "num_tokens": 409245861.0, + "step": 15930 + }, + { + "epoch": 1.7495058203382385, + "grad_norm": 2.262784004211426, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7183754444122314, + "num_tokens": 409271163.0, + "step": 15931 + }, + { + "epoch": 1.749615638040852, + "grad_norm": 2.1683850288391113, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7210875153541565, + "num_tokens": 409298637.0, + "step": 15932 + }, + { + "epoch": 1.7497254557434658, + "grad_norm": 2.128784656524658, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7032233476638794, + "num_tokens": 409327962.0, + "step": 15933 + }, + { + "epoch": 1.7498352734460796, + "grad_norm": 2.043544292449951, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7036172151565552, + "num_tokens": 409359524.0, + "step": 15934 + }, + { + "epoch": 1.7499450911486931, + "grad_norm": 2.109508514404297, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7535066604614258, + "num_tokens": 409385976.0, + "step": 15935 + }, + { + "epoch": 1.7500549088513069, + "grad_norm": 2.1193861961364746, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.714626133441925, + "num_tokens": 409414523.0, + "step": 15936 + }, + { + "epoch": 1.7501647265539204, + "grad_norm": 2.356884479522705, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7042478322982788, + "num_tokens": 409438318.0, + "step": 15937 + }, + { + "epoch": 1.7502745442565342, + "grad_norm": 2.02191424369812, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7029663324356079, + "num_tokens": 409468512.0, + "step": 15938 + }, + { + "epoch": 1.750384361959148, + "grad_norm": 2.0440516471862793, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7472915053367615, + "num_tokens": 409495882.0, + "step": 15939 + }, + { + "epoch": 1.7504941796617615, + "grad_norm": 2.3338277339935303, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7288544178009033, + "num_tokens": 409520887.0, + "step": 15940 + }, + { + "epoch": 1.750603997364375, + "grad_norm": 2.1883113384246826, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.72606360912323, + "num_tokens": 409547083.0, + "step": 15941 + }, + { + "epoch": 1.7507138150669888, + "grad_norm": 2.4780337810516357, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7609944343566895, + "num_tokens": 409566284.0, + "step": 15942 + }, + { + "epoch": 1.7508236327696025, + "grad_norm": 2.3455021381378174, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7039504051208496, + "num_tokens": 409589278.0, + "step": 15943 + }, + { + "epoch": 1.7509334504722163, + "grad_norm": 2.0300028324127197, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7104442119598389, + "num_tokens": 409619283.0, + "step": 15944 + }, + { + "epoch": 1.7510432681748298, + "grad_norm": 1.9003300666809082, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.71141517162323, + "num_tokens": 409651092.0, + "step": 15945 + }, + { + "epoch": 1.7511530858774433, + "grad_norm": 2.4344186782836914, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7319925427436829, + "num_tokens": 409672842.0, + "step": 15946 + }, + { + "epoch": 1.751262903580057, + "grad_norm": 2.533889055252075, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7441209554672241, + "num_tokens": 409692757.0, + "step": 15947 + }, + { + "epoch": 1.7513727212826709, + "grad_norm": 2.0205423831939697, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.699675977230072, + "num_tokens": 409721380.0, + "step": 15948 + }, + { + "epoch": 1.7514825389852844, + "grad_norm": 2.228470802307129, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7173017859458923, + "num_tokens": 409747776.0, + "step": 15949 + }, + { + "epoch": 1.7515923566878981, + "grad_norm": 2.2919232845306396, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7028658390045166, + "num_tokens": 409774829.0, + "step": 15950 + }, + { + "epoch": 1.7517021743905117, + "grad_norm": 2.2655084133148193, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7106727361679077, + "num_tokens": 409800785.0, + "step": 15951 + }, + { + "epoch": 1.7518119920931254, + "grad_norm": 2.1895790100097656, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7342284321784973, + "num_tokens": 409825357.0, + "step": 15952 + }, + { + "epoch": 1.7519218097957392, + "grad_norm": 2.1757700443267822, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.710396409034729, + "num_tokens": 409853167.0, + "step": 15953 + }, + { + "epoch": 1.7520316274983527, + "grad_norm": 2.4251794815063477, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7262210249900818, + "num_tokens": 409875669.0, + "step": 15954 + }, + { + "epoch": 1.7521414452009663, + "grad_norm": 2.391231060028076, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7203239798545837, + "num_tokens": 409898151.0, + "step": 15955 + }, + { + "epoch": 1.75225126290358, + "grad_norm": 2.333667278289795, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7253760695457458, + "num_tokens": 409922680.0, + "step": 15956 + }, + { + "epoch": 1.7523610806061938, + "grad_norm": 2.0649654865264893, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7296311855316162, + "num_tokens": 409949926.0, + "step": 15957 + }, + { + "epoch": 1.7524708983088075, + "grad_norm": 2.167766571044922, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7303703427314758, + "num_tokens": 409974891.0, + "step": 15958 + }, + { + "epoch": 1.752580716011421, + "grad_norm": 2.037317991256714, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6875268816947937, + "num_tokens": 410007254.0, + "step": 15959 + }, + { + "epoch": 1.7526905337140346, + "grad_norm": 2.2357308864593506, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7255629897117615, + "num_tokens": 410033448.0, + "step": 15960 + }, + { + "epoch": 1.7528003514166484, + "grad_norm": 2.4219205379486084, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7245198488235474, + "num_tokens": 410054490.0, + "step": 15961 + }, + { + "epoch": 1.7529101691192621, + "grad_norm": 2.3969650268554688, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.6977053880691528, + "num_tokens": 410078313.0, + "step": 15962 + }, + { + "epoch": 1.7530199868218757, + "grad_norm": 1.9355700016021729, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.685539722442627, + "num_tokens": 410110583.0, + "step": 15963 + }, + { + "epoch": 1.7531298045244892, + "grad_norm": 1.9585617780685425, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7002536058425903, + "num_tokens": 410144183.0, + "step": 15964 + }, + { + "epoch": 1.753239622227103, + "grad_norm": 2.525296211242676, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7184118032455444, + "num_tokens": 410165187.0, + "step": 15965 + }, + { + "epoch": 1.7533494399297167, + "grad_norm": 2.243741512298584, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7181626558303833, + "num_tokens": 410189917.0, + "step": 15966 + }, + { + "epoch": 1.7534592576323305, + "grad_norm": 2.0471484661102295, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7222030162811279, + "num_tokens": 410220453.0, + "step": 15967 + }, + { + "epoch": 1.753569075334944, + "grad_norm": 2.2552638053894043, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7204092144966125, + "num_tokens": 410246641.0, + "step": 15968 + }, + { + "epoch": 1.7536788930375575, + "grad_norm": 2.111950397491455, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7046003341674805, + "num_tokens": 410275641.0, + "step": 15969 + }, + { + "epoch": 1.7537887107401713, + "grad_norm": 2.2489774227142334, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7239007949829102, + "num_tokens": 410300691.0, + "step": 15970 + }, + { + "epoch": 1.753898528442785, + "grad_norm": 2.338116407394409, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7280033826828003, + "num_tokens": 410324860.0, + "step": 15971 + }, + { + "epoch": 1.7540083461453988, + "grad_norm": 2.4696149826049805, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.720734179019928, + "num_tokens": 410347424.0, + "step": 15972 + }, + { + "epoch": 1.7541181638480123, + "grad_norm": 2.387111186981201, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7153000831604004, + "num_tokens": 410372195.0, + "step": 15973 + }, + { + "epoch": 1.7542279815506259, + "grad_norm": 2.516512393951416, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7128037214279175, + "num_tokens": 410395020.0, + "step": 15974 + }, + { + "epoch": 1.7543377992532396, + "grad_norm": 2.0504212379455566, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7383604049682617, + "num_tokens": 410423562.0, + "step": 15975 + }, + { + "epoch": 1.7544476169558534, + "grad_norm": 2.3980209827423096, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7072227597236633, + "num_tokens": 410447777.0, + "step": 15976 + }, + { + "epoch": 1.754557434658467, + "grad_norm": 2.244875907897949, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7235559225082397, + "num_tokens": 410472251.0, + "step": 15977 + }, + { + "epoch": 1.7546672523610805, + "grad_norm": 1.9903539419174194, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7026443481445312, + "num_tokens": 410503669.0, + "step": 15978 + }, + { + "epoch": 1.7547770700636942, + "grad_norm": 2.1656785011291504, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7137027978897095, + "num_tokens": 410530535.0, + "step": 15979 + }, + { + "epoch": 1.754886887766308, + "grad_norm": 2.2655978202819824, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7293954491615295, + "num_tokens": 410553863.0, + "step": 15980 + }, + { + "epoch": 1.7549967054689217, + "grad_norm": 2.543004035949707, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7403931021690369, + "num_tokens": 410574762.0, + "step": 15981 + }, + { + "epoch": 1.7551065231715353, + "grad_norm": 2.355431079864502, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7271562814712524, + "num_tokens": 410598712.0, + "step": 15982 + }, + { + "epoch": 1.7552163408741488, + "grad_norm": 2.285861015319824, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7269273996353149, + "num_tokens": 410623534.0, + "step": 15983 + }, + { + "epoch": 1.7553261585767626, + "grad_norm": 2.0082924365997314, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7335447072982788, + "num_tokens": 410652598.0, + "step": 15984 + }, + { + "epoch": 1.7554359762793763, + "grad_norm": 2.077521562576294, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6971935033798218, + "num_tokens": 410684702.0, + "step": 15985 + }, + { + "epoch": 1.7555457939819898, + "grad_norm": 2.721442222595215, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7375569343566895, + "num_tokens": 410704373.0, + "step": 15986 + }, + { + "epoch": 1.7556556116846036, + "grad_norm": 1.8498446941375732, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7068917751312256, + "num_tokens": 410737556.0, + "step": 15987 + }, + { + "epoch": 1.7557654293872171, + "grad_norm": 1.97994065284729, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7059887647628784, + "num_tokens": 410769889.0, + "step": 15988 + }, + { + "epoch": 1.755875247089831, + "grad_norm": 2.0217671394348145, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7307522296905518, + "num_tokens": 410799572.0, + "step": 15989 + }, + { + "epoch": 1.7559850647924446, + "grad_norm": 2.247199535369873, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6898099780082703, + "num_tokens": 410827634.0, + "step": 15990 + }, + { + "epoch": 1.7560948824950582, + "grad_norm": 2.5660338401794434, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7248622179031372, + "num_tokens": 410848804.0, + "step": 15991 + }, + { + "epoch": 1.7562047001976717, + "grad_norm": 2.1893138885498047, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7108677625656128, + "num_tokens": 410875633.0, + "step": 15992 + }, + { + "epoch": 1.7563145179002855, + "grad_norm": 2.357445478439331, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7275543212890625, + "num_tokens": 410898446.0, + "step": 15993 + }, + { + "epoch": 1.7564243356028992, + "grad_norm": 2.31598162651062, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7329918742179871, + "num_tokens": 410923403.0, + "step": 15994 + }, + { + "epoch": 1.756534153305513, + "grad_norm": 2.8237640857696533, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7412888407707214, + "num_tokens": 410947366.0, + "step": 15995 + }, + { + "epoch": 1.7566439710081265, + "grad_norm": 2.196737051010132, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.710496187210083, + "num_tokens": 410971817.0, + "step": 15996 + }, + { + "epoch": 1.75675378871074, + "grad_norm": 2.4834976196289062, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7290604114532471, + "num_tokens": 410993153.0, + "step": 15997 + }, + { + "epoch": 1.7568636064133538, + "grad_norm": 2.3488411903381348, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7365249395370483, + "num_tokens": 411015917.0, + "step": 15998 + }, + { + "epoch": 1.7569734241159676, + "grad_norm": 2.478337049484253, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7379847764968872, + "num_tokens": 411036545.0, + "step": 15999 + }, + { + "epoch": 1.757083241818581, + "grad_norm": 2.3805484771728516, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7377640008926392, + "num_tokens": 411059362.0, + "step": 16000 + }, + { + "epoch": 1.7571930595211949, + "grad_norm": 2.022217273712158, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6878302693367004, + "num_tokens": 411090296.0, + "step": 16001 + }, + { + "epoch": 1.7573028772238084, + "grad_norm": 2.4877161979675293, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7217832803726196, + "num_tokens": 411112271.0, + "step": 16002 + }, + { + "epoch": 1.7574126949264222, + "grad_norm": 2.2062160968780518, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7212854623794556, + "num_tokens": 411138846.0, + "step": 16003 + }, + { + "epoch": 1.757522512629036, + "grad_norm": 2.1107022762298584, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7094839215278625, + "num_tokens": 411166965.0, + "step": 16004 + }, + { + "epoch": 1.7576323303316495, + "grad_norm": 2.094770669937134, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7208811044692993, + "num_tokens": 411195233.0, + "step": 16005 + }, + { + "epoch": 1.757742148034263, + "grad_norm": 2.2006983757019043, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7237476110458374, + "num_tokens": 411221533.0, + "step": 16006 + }, + { + "epoch": 1.7578519657368767, + "grad_norm": 2.522412061691284, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7239576578140259, + "num_tokens": 411244436.0, + "step": 16007 + }, + { + "epoch": 1.7579617834394905, + "grad_norm": 2.2996163368225098, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7150543928146362, + "num_tokens": 411268342.0, + "step": 16008 + }, + { + "epoch": 1.7580716011421043, + "grad_norm": 2.320464611053467, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7030477523803711, + "num_tokens": 411293301.0, + "step": 16009 + }, + { + "epoch": 1.7581814188447178, + "grad_norm": 2.5313007831573486, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7081559300422668, + "num_tokens": 411314506.0, + "step": 16010 + }, + { + "epoch": 1.7582912365473313, + "grad_norm": 2.44864821434021, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7321016788482666, + "num_tokens": 411336460.0, + "step": 16011 + }, + { + "epoch": 1.758401054249945, + "grad_norm": 2.112097978591919, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.695989727973938, + "num_tokens": 411366613.0, + "step": 16012 + }, + { + "epoch": 1.7585108719525588, + "grad_norm": 2.419224262237549, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7286157608032227, + "num_tokens": 411388743.0, + "step": 16013 + }, + { + "epoch": 1.7586206896551724, + "grad_norm": 2.0610437393188477, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7463077306747437, + "num_tokens": 411416863.0, + "step": 16014 + }, + { + "epoch": 1.758730507357786, + "grad_norm": 2.0617566108703613, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7098280191421509, + "num_tokens": 411445893.0, + "step": 16015 + }, + { + "epoch": 1.7588403250603997, + "grad_norm": 2.145233631134033, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7075432538986206, + "num_tokens": 411473165.0, + "step": 16016 + }, + { + "epoch": 1.7589501427630134, + "grad_norm": 2.038710117340088, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.71506667137146, + "num_tokens": 411504113.0, + "step": 16017 + }, + { + "epoch": 1.7590599604656272, + "grad_norm": 2.228422164916992, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6996973752975464, + "num_tokens": 411532441.0, + "step": 16018 + }, + { + "epoch": 1.7591697781682407, + "grad_norm": 2.6107659339904785, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7410594820976257, + "num_tokens": 411552354.0, + "step": 16019 + }, + { + "epoch": 1.7592795958708543, + "grad_norm": 2.5017459392547607, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7275078892707825, + "num_tokens": 411573935.0, + "step": 16020 + }, + { + "epoch": 1.759389413573468, + "grad_norm": 2.465557336807251, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7462022304534912, + "num_tokens": 411593061.0, + "step": 16021 + }, + { + "epoch": 1.7594992312760818, + "grad_norm": 2.239828109741211, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7120170593261719, + "num_tokens": 411616833.0, + "step": 16022 + }, + { + "epoch": 1.7596090489786955, + "grad_norm": 2.0673365592956543, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7132775783538818, + "num_tokens": 411646984.0, + "step": 16023 + }, + { + "epoch": 1.759718866681309, + "grad_norm": 2.120426893234253, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7423046827316284, + "num_tokens": 411673061.0, + "step": 16024 + }, + { + "epoch": 1.7598286843839226, + "grad_norm": 2.220689296722412, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.70290207862854, + "num_tokens": 411700873.0, + "step": 16025 + }, + { + "epoch": 1.7599385020865363, + "grad_norm": 2.1998331546783447, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7103201746940613, + "num_tokens": 411727134.0, + "step": 16026 + }, + { + "epoch": 1.76004831978915, + "grad_norm": 2.6870083808898926, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7198999524116516, + "num_tokens": 411745498.0, + "step": 16027 + }, + { + "epoch": 1.7601581374917636, + "grad_norm": 1.8839048147201538, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.6979232430458069, + "num_tokens": 411780312.0, + "step": 16028 + }, + { + "epoch": 1.7602679551943772, + "grad_norm": 2.1021807193756104, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7161114811897278, + "num_tokens": 411810922.0, + "step": 16029 + }, + { + "epoch": 1.760377772896991, + "grad_norm": 2.140556812286377, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.717499852180481, + "num_tokens": 411839177.0, + "step": 16030 + }, + { + "epoch": 1.7604875905996047, + "grad_norm": 2.019880771636963, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7266539931297302, + "num_tokens": 411868597.0, + "step": 16031 + }, + { + "epoch": 1.7605974083022184, + "grad_norm": 2.2334365844726562, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7276166677474976, + "num_tokens": 411894080.0, + "step": 16032 + }, + { + "epoch": 1.760707226004832, + "grad_norm": 2.24433970451355, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7239251136779785, + "num_tokens": 411919747.0, + "step": 16033 + }, + { + "epoch": 1.7608170437074455, + "grad_norm": 2.168649435043335, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7148725390434265, + "num_tokens": 411948491.0, + "step": 16034 + }, + { + "epoch": 1.7609268614100593, + "grad_norm": 2.073193073272705, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.738972544670105, + "num_tokens": 411976521.0, + "step": 16035 + }, + { + "epoch": 1.761036679112673, + "grad_norm": 2.2550041675567627, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7206982374191284, + "num_tokens": 412002487.0, + "step": 16036 + }, + { + "epoch": 1.7611464968152868, + "grad_norm": 2.118889093399048, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7011928558349609, + "num_tokens": 412032606.0, + "step": 16037 + }, + { + "epoch": 1.7612563145179003, + "grad_norm": 2.1666200160980225, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.717412531375885, + "num_tokens": 412060420.0, + "step": 16038 + }, + { + "epoch": 1.7613661322205139, + "grad_norm": 2.133979320526123, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7101999521255493, + "num_tokens": 412087477.0, + "step": 16039 + }, + { + "epoch": 1.7614759499231276, + "grad_norm": 2.3055150508880615, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7003186941146851, + "num_tokens": 412111131.0, + "step": 16040 + }, + { + "epoch": 1.7615857676257414, + "grad_norm": 2.1227290630340576, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7224822044372559, + "num_tokens": 412138032.0, + "step": 16041 + }, + { + "epoch": 1.761695585328355, + "grad_norm": 2.237649917602539, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7342228293418884, + "num_tokens": 412161722.0, + "step": 16042 + }, + { + "epoch": 1.7618054030309684, + "grad_norm": 2.2830355167388916, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7168363332748413, + "num_tokens": 412189018.0, + "step": 16043 + }, + { + "epoch": 1.7619152207335822, + "grad_norm": 2.080946922302246, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7345359921455383, + "num_tokens": 412218026.0, + "step": 16044 + }, + { + "epoch": 1.762025038436196, + "grad_norm": 2.2054238319396973, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7052813768386841, + "num_tokens": 412244482.0, + "step": 16045 + }, + { + "epoch": 1.7621348561388097, + "grad_norm": 2.521693706512451, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7382959127426147, + "num_tokens": 412265969.0, + "step": 16046 + }, + { + "epoch": 1.7622446738414232, + "grad_norm": 2.2413763999938965, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7181047201156616, + "num_tokens": 412290763.0, + "step": 16047 + }, + { + "epoch": 1.7623544915440368, + "grad_norm": 2.360969305038452, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7017814517021179, + "num_tokens": 412312079.0, + "step": 16048 + }, + { + "epoch": 1.7624643092466505, + "grad_norm": 2.4998416900634766, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7279291152954102, + "num_tokens": 412333643.0, + "step": 16049 + }, + { + "epoch": 1.7625741269492643, + "grad_norm": 2.771379232406616, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7352707386016846, + "num_tokens": 412351409.0, + "step": 16050 + }, + { + "epoch": 1.7626839446518778, + "grad_norm": 2.361514091491699, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7233197689056396, + "num_tokens": 412373569.0, + "step": 16051 + }, + { + "epoch": 1.7627937623544916, + "grad_norm": 2.2612643241882324, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7244665622711182, + "num_tokens": 412397122.0, + "step": 16052 + }, + { + "epoch": 1.7629035800571051, + "grad_norm": 1.9006609916687012, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7467356324195862, + "num_tokens": 412426110.0, + "step": 16053 + }, + { + "epoch": 1.7630133977597189, + "grad_norm": 2.4907994270324707, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7024857997894287, + "num_tokens": 412448296.0, + "step": 16054 + }, + { + "epoch": 1.7631232154623326, + "grad_norm": 2.314192056655884, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7195342779159546, + "num_tokens": 412471460.0, + "step": 16055 + }, + { + "epoch": 1.7632330331649462, + "grad_norm": 2.1471333503723145, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7056406736373901, + "num_tokens": 412498939.0, + "step": 16056 + }, + { + "epoch": 1.7633428508675597, + "grad_norm": 2.204312324523926, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7352906465530396, + "num_tokens": 412525243.0, + "step": 16057 + }, + { + "epoch": 1.7634526685701735, + "grad_norm": 2.0464701652526855, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7398324608802795, + "num_tokens": 412554491.0, + "step": 16058 + }, + { + "epoch": 1.7635624862727872, + "grad_norm": 2.116811990737915, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7168359756469727, + "num_tokens": 412581358.0, + "step": 16059 + }, + { + "epoch": 1.763672303975401, + "grad_norm": 2.227778673171997, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7299652099609375, + "num_tokens": 412606879.0, + "step": 16060 + }, + { + "epoch": 1.7637821216780145, + "grad_norm": 2.3482651710510254, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7246525287628174, + "num_tokens": 412630540.0, + "step": 16061 + }, + { + "epoch": 1.763891939380628, + "grad_norm": 2.31366229057312, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7075562477111816, + "num_tokens": 412655758.0, + "step": 16062 + }, + { + "epoch": 1.7640017570832418, + "grad_norm": 2.3040850162506104, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7300682067871094, + "num_tokens": 412680334.0, + "step": 16063 + }, + { + "epoch": 1.7641115747858556, + "grad_norm": 2.3587069511413574, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7182049751281738, + "num_tokens": 412702953.0, + "step": 16064 + }, + { + "epoch": 1.764221392488469, + "grad_norm": 2.2044074535369873, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7313897609710693, + "num_tokens": 412728973.0, + "step": 16065 + }, + { + "epoch": 1.7643312101910829, + "grad_norm": 2.2775001525878906, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7148191928863525, + "num_tokens": 412753157.0, + "step": 16066 + }, + { + "epoch": 1.7644410278936964, + "grad_norm": 2.4375271797180176, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7331589460372925, + "num_tokens": 412775653.0, + "step": 16067 + }, + { + "epoch": 1.7645508455963101, + "grad_norm": 2.3916265964508057, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7210906744003296, + "num_tokens": 412799179.0, + "step": 16068 + }, + { + "epoch": 1.764660663298924, + "grad_norm": 1.9592506885528564, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.6971070766448975, + "num_tokens": 412830239.0, + "step": 16069 + }, + { + "epoch": 1.7647704810015374, + "grad_norm": 2.5513017177581787, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7374095916748047, + "num_tokens": 412850463.0, + "step": 16070 + }, + { + "epoch": 1.764880298704151, + "grad_norm": 2.2588250637054443, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7221422791481018, + "num_tokens": 412874644.0, + "step": 16071 + }, + { + "epoch": 1.7649901164067647, + "grad_norm": 2.2652809619903564, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7225744724273682, + "num_tokens": 412902769.0, + "step": 16072 + }, + { + "epoch": 1.7650999341093785, + "grad_norm": 2.5374505519866943, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7324768304824829, + "num_tokens": 412921213.0, + "step": 16073 + }, + { + "epoch": 1.7652097518119922, + "grad_norm": 2.387608051300049, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7223985195159912, + "num_tokens": 412944965.0, + "step": 16074 + }, + { + "epoch": 1.7653195695146058, + "grad_norm": 2.1497340202331543, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7042543292045593, + "num_tokens": 412972958.0, + "step": 16075 + }, + { + "epoch": 1.7654293872172193, + "grad_norm": 2.59977126121521, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.729088306427002, + "num_tokens": 412993169.0, + "step": 16076 + }, + { + "epoch": 1.765539204919833, + "grad_norm": 2.166887044906616, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7349787354469299, + "num_tokens": 413020130.0, + "step": 16077 + }, + { + "epoch": 1.7656490226224468, + "grad_norm": 2.09086537361145, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7389991283416748, + "num_tokens": 413049567.0, + "step": 16078 + }, + { + "epoch": 1.7657588403250604, + "grad_norm": 2.284454822540283, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7401487231254578, + "num_tokens": 413072516.0, + "step": 16079 + }, + { + "epoch": 1.765868658027674, + "grad_norm": 2.343247652053833, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7222661972045898, + "num_tokens": 413096451.0, + "step": 16080 + }, + { + "epoch": 1.7659784757302877, + "grad_norm": 2.3030033111572266, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7051823139190674, + "num_tokens": 413121695.0, + "step": 16081 + }, + { + "epoch": 1.7660882934329014, + "grad_norm": 2.7603328227996826, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7115064859390259, + "num_tokens": 413139257.0, + "step": 16082 + }, + { + "epoch": 1.7661981111355152, + "grad_norm": 2.1064231395721436, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7102186679840088, + "num_tokens": 413166594.0, + "step": 16083 + }, + { + "epoch": 1.7663079288381287, + "grad_norm": 2.3631839752197266, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7336616516113281, + "num_tokens": 413188644.0, + "step": 16084 + }, + { + "epoch": 1.7664177465407422, + "grad_norm": 2.2444939613342285, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7328816056251526, + "num_tokens": 413212905.0, + "step": 16085 + }, + { + "epoch": 1.766527564243356, + "grad_norm": 2.028733491897583, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.6991643309593201, + "num_tokens": 413244350.0, + "step": 16086 + }, + { + "epoch": 1.7666373819459698, + "grad_norm": 2.25899338722229, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7043399810791016, + "num_tokens": 413270066.0, + "step": 16087 + }, + { + "epoch": 1.7667471996485835, + "grad_norm": 2.0244977474212646, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6886815428733826, + "num_tokens": 413301437.0, + "step": 16088 + }, + { + "epoch": 1.766857017351197, + "grad_norm": 2.1679086685180664, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7180722951889038, + "num_tokens": 413326682.0, + "step": 16089 + }, + { + "epoch": 1.7669668350538106, + "grad_norm": 2.0796756744384766, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7283624410629272, + "num_tokens": 413354968.0, + "step": 16090 + }, + { + "epoch": 1.7670766527564243, + "grad_norm": 2.169987440109253, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7103879451751709, + "num_tokens": 413381812.0, + "step": 16091 + }, + { + "epoch": 1.767186470459038, + "grad_norm": 2.1197917461395264, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7127099633216858, + "num_tokens": 413408826.0, + "step": 16092 + }, + { + "epoch": 1.7672962881616516, + "grad_norm": 2.129234790802002, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7114536762237549, + "num_tokens": 413437200.0, + "step": 16093 + }, + { + "epoch": 1.7674061058642652, + "grad_norm": 2.1709671020507812, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.727698564529419, + "num_tokens": 413463189.0, + "step": 16094 + }, + { + "epoch": 1.767515923566879, + "grad_norm": 2.2512824535369873, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7246137261390686, + "num_tokens": 413489066.0, + "step": 16095 + }, + { + "epoch": 1.7676257412694927, + "grad_norm": 2.1299076080322266, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7199947834014893, + "num_tokens": 413515238.0, + "step": 16096 + }, + { + "epoch": 1.7677355589721064, + "grad_norm": 2.0637423992156982, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7172995805740356, + "num_tokens": 413543855.0, + "step": 16097 + }, + { + "epoch": 1.76784537667472, + "grad_norm": 2.080300807952881, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7304003238677979, + "num_tokens": 413571151.0, + "step": 16098 + }, + { + "epoch": 1.7679551943773335, + "grad_norm": 2.23917555809021, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7093137502670288, + "num_tokens": 413595822.0, + "step": 16099 + }, + { + "epoch": 1.7680650120799473, + "grad_norm": 2.257269859313965, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.733694851398468, + "num_tokens": 413621840.0, + "step": 16100 + }, + { + "epoch": 1.768174829782561, + "grad_norm": 2.5037903785705566, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.720572829246521, + "num_tokens": 413642840.0, + "step": 16101 + }, + { + "epoch": 1.7682846474851748, + "grad_norm": 2.070749521255493, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.724337637424469, + "num_tokens": 413671275.0, + "step": 16102 + }, + { + "epoch": 1.7683944651877883, + "grad_norm": 2.0755083560943604, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7052907347679138, + "num_tokens": 413701020.0, + "step": 16103 + }, + { + "epoch": 1.7685042828904018, + "grad_norm": 2.2062907218933105, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7242118716239929, + "num_tokens": 413727429.0, + "step": 16104 + }, + { + "epoch": 1.7686141005930156, + "grad_norm": 2.3269927501678467, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7265370488166809, + "num_tokens": 413751544.0, + "step": 16105 + }, + { + "epoch": 1.7687239182956294, + "grad_norm": 2.16296124458313, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7290911674499512, + "num_tokens": 413778074.0, + "step": 16106 + }, + { + "epoch": 1.768833735998243, + "grad_norm": 2.375232458114624, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6958290338516235, + "num_tokens": 413802887.0, + "step": 16107 + }, + { + "epoch": 1.7689435537008564, + "grad_norm": 2.229217767715454, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7273879051208496, + "num_tokens": 413828366.0, + "step": 16108 + }, + { + "epoch": 1.7690533714034702, + "grad_norm": 2.678765296936035, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7316621541976929, + "num_tokens": 413847356.0, + "step": 16109 + }, + { + "epoch": 1.769163189106084, + "grad_norm": 2.171006679534912, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.6999145746231079, + "num_tokens": 413874106.0, + "step": 16110 + }, + { + "epoch": 1.7692730068086977, + "grad_norm": 2.1933298110961914, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7158998847007751, + "num_tokens": 413900423.0, + "step": 16111 + }, + { + "epoch": 1.7693828245113112, + "grad_norm": 2.05857253074646, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7234379053115845, + "num_tokens": 413927153.0, + "step": 16112 + }, + { + "epoch": 1.7694926422139248, + "grad_norm": 2.4537971019744873, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7046011090278625, + "num_tokens": 413950434.0, + "step": 16113 + }, + { + "epoch": 1.7696024599165385, + "grad_norm": 2.4357428550720215, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7066045999526978, + "num_tokens": 413972383.0, + "step": 16114 + }, + { + "epoch": 1.7697122776191523, + "grad_norm": 2.2849981784820557, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.708132266998291, + "num_tokens": 413998611.0, + "step": 16115 + }, + { + "epoch": 1.7698220953217658, + "grad_norm": 2.5195109844207764, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7320581674575806, + "num_tokens": 414019796.0, + "step": 16116 + }, + { + "epoch": 1.7699319130243796, + "grad_norm": 2.4088985919952393, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7306358814239502, + "num_tokens": 414041740.0, + "step": 16117 + }, + { + "epoch": 1.7700417307269931, + "grad_norm": 2.2113473415374756, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7148084044456482, + "num_tokens": 414066795.0, + "step": 16118 + }, + { + "epoch": 1.7701515484296069, + "grad_norm": 2.1851236820220947, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7167186737060547, + "num_tokens": 414093877.0, + "step": 16119 + }, + { + "epoch": 1.7702613661322206, + "grad_norm": 2.157994508743286, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7154865860939026, + "num_tokens": 414121998.0, + "step": 16120 + }, + { + "epoch": 1.7703711838348342, + "grad_norm": 2.408623218536377, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7222908139228821, + "num_tokens": 414144635.0, + "step": 16121 + }, + { + "epoch": 1.7704810015374477, + "grad_norm": 2.2621641159057617, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7430025339126587, + "num_tokens": 414169840.0, + "step": 16122 + }, + { + "epoch": 1.7705908192400615, + "grad_norm": 2.583801507949829, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7435436844825745, + "num_tokens": 414189746.0, + "step": 16123 + }, + { + "epoch": 1.7707006369426752, + "grad_norm": 2.4730138778686523, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7350617051124573, + "num_tokens": 414210639.0, + "step": 16124 + }, + { + "epoch": 1.770810454645289, + "grad_norm": 2.1492624282836914, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.720564603805542, + "num_tokens": 414239312.0, + "step": 16125 + }, + { + "epoch": 1.7709202723479025, + "grad_norm": 1.896974802017212, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7177450656890869, + "num_tokens": 414273795.0, + "step": 16126 + }, + { + "epoch": 1.771030090050516, + "grad_norm": 2.0646626949310303, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7162178158760071, + "num_tokens": 414302936.0, + "step": 16127 + }, + { + "epoch": 1.7711399077531298, + "grad_norm": 2.008979320526123, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6941625475883484, + "num_tokens": 414333140.0, + "step": 16128 + }, + { + "epoch": 1.7712497254557436, + "grad_norm": 2.312321901321411, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7380657196044922, + "num_tokens": 414356803.0, + "step": 16129 + }, + { + "epoch": 1.771359543158357, + "grad_norm": 2.065025568008423, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.717778742313385, + "num_tokens": 414386642.0, + "step": 16130 + }, + { + "epoch": 1.7714693608609708, + "grad_norm": 2.29255747795105, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7295186519622803, + "num_tokens": 414411008.0, + "step": 16131 + }, + { + "epoch": 1.7715791785635844, + "grad_norm": 2.23710560798645, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7110954523086548, + "num_tokens": 414437281.0, + "step": 16132 + }, + { + "epoch": 1.7716889962661981, + "grad_norm": 2.4277775287628174, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7320154905319214, + "num_tokens": 414457846.0, + "step": 16133 + }, + { + "epoch": 1.771798813968812, + "grad_norm": 2.1444432735443115, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.725134015083313, + "num_tokens": 414485697.0, + "step": 16134 + }, + { + "epoch": 1.7719086316714254, + "grad_norm": 2.2848904132843018, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7386218309402466, + "num_tokens": 414510693.0, + "step": 16135 + }, + { + "epoch": 1.772018449374039, + "grad_norm": 2.376298189163208, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7174376249313354, + "num_tokens": 414533606.0, + "step": 16136 + }, + { + "epoch": 1.7721282670766527, + "grad_norm": 2.5675134658813477, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7304081320762634, + "num_tokens": 414554167.0, + "step": 16137 + }, + { + "epoch": 1.7722380847792665, + "grad_norm": 2.233452558517456, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7152352333068848, + "num_tokens": 414580844.0, + "step": 16138 + }, + { + "epoch": 1.7723479024818802, + "grad_norm": 2.380175828933716, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7427358627319336, + "num_tokens": 414603604.0, + "step": 16139 + }, + { + "epoch": 1.7724577201844938, + "grad_norm": 2.2541730403900146, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7215982675552368, + "num_tokens": 414628935.0, + "step": 16140 + }, + { + "epoch": 1.7725675378871073, + "grad_norm": 2.1553454399108887, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7356016635894775, + "num_tokens": 414654704.0, + "step": 16141 + }, + { + "epoch": 1.772677355589721, + "grad_norm": 2.6409342288970947, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7345330715179443, + "num_tokens": 414672319.0, + "step": 16142 + }, + { + "epoch": 1.7727871732923348, + "grad_norm": 2.0296225547790527, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7035936713218689, + "num_tokens": 414701100.0, + "step": 16143 + }, + { + "epoch": 1.7728969909949484, + "grad_norm": 2.3227717876434326, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7122828960418701, + "num_tokens": 414725206.0, + "step": 16144 + }, + { + "epoch": 1.7730068086975619, + "grad_norm": 2.0970458984375, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7315545082092285, + "num_tokens": 414752912.0, + "step": 16145 + }, + { + "epoch": 1.7731166264001756, + "grad_norm": 2.341649055480957, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7252328395843506, + "num_tokens": 414775950.0, + "step": 16146 + }, + { + "epoch": 1.7732264441027894, + "grad_norm": 2.266511917114258, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.721160888671875, + "num_tokens": 414801356.0, + "step": 16147 + }, + { + "epoch": 1.7733362618054032, + "grad_norm": 1.9751498699188232, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.698998749256134, + "num_tokens": 414833253.0, + "step": 16148 + }, + { + "epoch": 1.7734460795080167, + "grad_norm": 1.9534837007522583, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7078385353088379, + "num_tokens": 414866857.0, + "step": 16149 + }, + { + "epoch": 1.7735558972106302, + "grad_norm": 2.607802152633667, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.733107328414917, + "num_tokens": 414887238.0, + "step": 16150 + }, + { + "epoch": 1.773665714913244, + "grad_norm": 2.0693092346191406, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.733903169631958, + "num_tokens": 414916153.0, + "step": 16151 + }, + { + "epoch": 1.7737755326158577, + "grad_norm": 2.059680938720703, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7347228527069092, + "num_tokens": 414944438.0, + "step": 16152 + }, + { + "epoch": 1.7738853503184715, + "grad_norm": 2.444432258605957, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7118551731109619, + "num_tokens": 414968459.0, + "step": 16153 + }, + { + "epoch": 1.773995168021085, + "grad_norm": 2.2964138984680176, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7246568202972412, + "num_tokens": 414991806.0, + "step": 16154 + }, + { + "epoch": 1.7741049857236986, + "grad_norm": 2.0159895420074463, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6887824535369873, + "num_tokens": 415024860.0, + "step": 16155 + }, + { + "epoch": 1.7742148034263123, + "grad_norm": 2.2868759632110596, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7044038772583008, + "num_tokens": 415050201.0, + "step": 16156 + }, + { + "epoch": 1.774324621128926, + "grad_norm": 2.233550786972046, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.710345983505249, + "num_tokens": 415074360.0, + "step": 16157 + }, + { + "epoch": 1.7744344388315396, + "grad_norm": 1.992740273475647, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7242910861968994, + "num_tokens": 415105567.0, + "step": 16158 + }, + { + "epoch": 1.7745442565341532, + "grad_norm": 2.052556037902832, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7244529724121094, + "num_tokens": 415136846.0, + "step": 16159 + }, + { + "epoch": 1.774654074236767, + "grad_norm": 2.3064918518066406, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7254083752632141, + "num_tokens": 415160893.0, + "step": 16160 + }, + { + "epoch": 1.7747638919393807, + "grad_norm": 2.035287618637085, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7176512479782104, + "num_tokens": 415194371.0, + "step": 16161 + }, + { + "epoch": 1.7748737096419944, + "grad_norm": 2.5359973907470703, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7430989742279053, + "num_tokens": 415213996.0, + "step": 16162 + }, + { + "epoch": 1.774983527344608, + "grad_norm": 2.105942964553833, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7197561264038086, + "num_tokens": 415242723.0, + "step": 16163 + }, + { + "epoch": 1.7750933450472215, + "grad_norm": 2.2470755577087402, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7271938323974609, + "num_tokens": 415268073.0, + "step": 16164 + }, + { + "epoch": 1.7752031627498353, + "grad_norm": 1.7547369003295898, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7134594917297363, + "num_tokens": 415303925.0, + "step": 16165 + }, + { + "epoch": 1.775312980452449, + "grad_norm": 2.3785080909729004, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7085589170455933, + "num_tokens": 415328862.0, + "step": 16166 + }, + { + "epoch": 1.7754227981550625, + "grad_norm": 2.532656669616699, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7355182766914368, + "num_tokens": 415348850.0, + "step": 16167 + }, + { + "epoch": 1.7755326158576763, + "grad_norm": 2.370234727859497, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7101354598999023, + "num_tokens": 415373734.0, + "step": 16168 + }, + { + "epoch": 1.7756424335602898, + "grad_norm": 2.263277292251587, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7468823790550232, + "num_tokens": 415397428.0, + "step": 16169 + }, + { + "epoch": 1.7757522512629036, + "grad_norm": 2.302884101867676, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7224332094192505, + "num_tokens": 415421851.0, + "step": 16170 + }, + { + "epoch": 1.7758620689655173, + "grad_norm": 2.143911600112915, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7002178430557251, + "num_tokens": 415451563.0, + "step": 16171 + }, + { + "epoch": 1.7759718866681309, + "grad_norm": 2.0607569217681885, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7265055179595947, + "num_tokens": 415480539.0, + "step": 16172 + }, + { + "epoch": 1.7760817043707444, + "grad_norm": 2.182014226913452, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.718165397644043, + "num_tokens": 415506283.0, + "step": 16173 + }, + { + "epoch": 1.7761915220733582, + "grad_norm": 2.146118640899658, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7023437023162842, + "num_tokens": 415533429.0, + "step": 16174 + }, + { + "epoch": 1.776301339775972, + "grad_norm": 2.3711180686950684, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7242769002914429, + "num_tokens": 415556035.0, + "step": 16175 + }, + { + "epoch": 1.7764111574785857, + "grad_norm": 2.3010241985321045, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7303668856620789, + "num_tokens": 415580482.0, + "step": 16176 + }, + { + "epoch": 1.7765209751811992, + "grad_norm": 2.212925910949707, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7016487121582031, + "num_tokens": 415608120.0, + "step": 16177 + }, + { + "epoch": 1.7766307928838128, + "grad_norm": 2.322380542755127, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.723202109336853, + "num_tokens": 415631432.0, + "step": 16178 + }, + { + "epoch": 1.7767406105864265, + "grad_norm": 2.214245319366455, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.696777880191803, + "num_tokens": 415658834.0, + "step": 16179 + }, + { + "epoch": 1.7768504282890403, + "grad_norm": 2.1177315711975098, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7319076061248779, + "num_tokens": 415685740.0, + "step": 16180 + }, + { + "epoch": 1.7769602459916538, + "grad_norm": 2.283931255340576, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7167369723320007, + "num_tokens": 415710915.0, + "step": 16181 + }, + { + "epoch": 1.7770700636942676, + "grad_norm": 2.198920965194702, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7240668535232544, + "num_tokens": 415737822.0, + "step": 16182 + }, + { + "epoch": 1.777179881396881, + "grad_norm": 2.0720930099487305, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7105299234390259, + "num_tokens": 415765803.0, + "step": 16183 + }, + { + "epoch": 1.7772896990994949, + "grad_norm": 2.2231948375701904, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7180993556976318, + "num_tokens": 415790541.0, + "step": 16184 + }, + { + "epoch": 1.7773995168021086, + "grad_norm": 1.8835501670837402, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7019817233085632, + "num_tokens": 415826297.0, + "step": 16185 + }, + { + "epoch": 1.7775093345047221, + "grad_norm": 2.033517599105835, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7092366814613342, + "num_tokens": 415855741.0, + "step": 16186 + }, + { + "epoch": 1.7776191522073357, + "grad_norm": 2.070486068725586, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7317215204238892, + "num_tokens": 415882973.0, + "step": 16187 + }, + { + "epoch": 1.7777289699099494, + "grad_norm": 2.7395429611206055, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7437851428985596, + "num_tokens": 415900140.0, + "step": 16188 + }, + { + "epoch": 1.7778387876125632, + "grad_norm": 2.4597771167755127, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7235273122787476, + "num_tokens": 415920561.0, + "step": 16189 + }, + { + "epoch": 1.777948605315177, + "grad_norm": 2.3161680698394775, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7214595079421997, + "num_tokens": 415943317.0, + "step": 16190 + }, + { + "epoch": 1.7780584230177905, + "grad_norm": 2.153012990951538, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7264552712440491, + "num_tokens": 415973591.0, + "step": 16191 + }, + { + "epoch": 1.778168240720404, + "grad_norm": 2.467338800430298, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.727118968963623, + "num_tokens": 415996058.0, + "step": 16192 + }, + { + "epoch": 1.7782780584230178, + "grad_norm": 2.108447551727295, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6993194222450256, + "num_tokens": 416026482.0, + "step": 16193 + }, + { + "epoch": 1.7783878761256315, + "grad_norm": 2.2199971675872803, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.709182858467102, + "num_tokens": 416051784.0, + "step": 16194 + }, + { + "epoch": 1.778497693828245, + "grad_norm": 2.157001256942749, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7065234780311584, + "num_tokens": 416078230.0, + "step": 16195 + }, + { + "epoch": 1.7786075115308586, + "grad_norm": 2.0510785579681396, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.736942708492279, + "num_tokens": 416105104.0, + "step": 16196 + }, + { + "epoch": 1.7787173292334724, + "grad_norm": 2.058218479156494, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.6949077844619751, + "num_tokens": 416136233.0, + "step": 16197 + }, + { + "epoch": 1.7788271469360861, + "grad_norm": 2.335519552230835, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7269831895828247, + "num_tokens": 416160077.0, + "step": 16198 + }, + { + "epoch": 1.7789369646386999, + "grad_norm": 2.182149887084961, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6994545459747314, + "num_tokens": 416188612.0, + "step": 16199 + }, + { + "epoch": 1.7790467823413134, + "grad_norm": 2.2519805431365967, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7094190120697021, + "num_tokens": 416212834.0, + "step": 16200 + }, + { + "epoch": 1.779156600043927, + "grad_norm": 2.606025218963623, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7381196022033691, + "num_tokens": 416231432.0, + "step": 16201 + }, + { + "epoch": 1.7792664177465407, + "grad_norm": 2.4482438564300537, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7328733205795288, + "num_tokens": 416252666.0, + "step": 16202 + }, + { + "epoch": 1.7793762354491545, + "grad_norm": 2.251875877380371, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7140461206436157, + "num_tokens": 416278425.0, + "step": 16203 + }, + { + "epoch": 1.7794860531517682, + "grad_norm": 2.181532621383667, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7214127779006958, + "num_tokens": 416303047.0, + "step": 16204 + }, + { + "epoch": 1.7795958708543818, + "grad_norm": 1.9958559274673462, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7266741394996643, + "num_tokens": 416334112.0, + "step": 16205 + }, + { + "epoch": 1.7797056885569953, + "grad_norm": 2.2182974815368652, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7292979955673218, + "num_tokens": 416359442.0, + "step": 16206 + }, + { + "epoch": 1.779815506259609, + "grad_norm": 2.0811779499053955, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7077049612998962, + "num_tokens": 416389193.0, + "step": 16207 + }, + { + "epoch": 1.7799253239622228, + "grad_norm": 2.4822285175323486, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7193412780761719, + "num_tokens": 416411643.0, + "step": 16208 + }, + { + "epoch": 1.7800351416648363, + "grad_norm": 2.403355598449707, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7564759850502014, + "num_tokens": 416434310.0, + "step": 16209 + }, + { + "epoch": 1.7801449593674499, + "grad_norm": 2.3257439136505127, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7161275744438171, + "num_tokens": 416458574.0, + "step": 16210 + }, + { + "epoch": 1.7802547770700636, + "grad_norm": 2.229863166809082, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7243067622184753, + "num_tokens": 416485230.0, + "step": 16211 + }, + { + "epoch": 1.7803645947726774, + "grad_norm": 2.2499895095825195, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7412159442901611, + "num_tokens": 416509548.0, + "step": 16212 + }, + { + "epoch": 1.7804744124752911, + "grad_norm": 2.1650094985961914, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7115354537963867, + "num_tokens": 416536820.0, + "step": 16213 + }, + { + "epoch": 1.7805842301779047, + "grad_norm": 2.0316011905670166, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.702422559261322, + "num_tokens": 416568729.0, + "step": 16214 + }, + { + "epoch": 1.7806940478805182, + "grad_norm": 2.3527581691741943, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7147660851478577, + "num_tokens": 416593528.0, + "step": 16215 + }, + { + "epoch": 1.780803865583132, + "grad_norm": 2.814842462539673, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7360248565673828, + "num_tokens": 416611529.0, + "step": 16216 + }, + { + "epoch": 1.7809136832857457, + "grad_norm": 2.418630838394165, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.700013279914856, + "num_tokens": 416634162.0, + "step": 16217 + }, + { + "epoch": 1.7810235009883595, + "grad_norm": 2.2675704956054688, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7310839891433716, + "num_tokens": 416658365.0, + "step": 16218 + }, + { + "epoch": 1.781133318690973, + "grad_norm": 2.138747453689575, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7150688171386719, + "num_tokens": 416687878.0, + "step": 16219 + }, + { + "epoch": 1.7812431363935866, + "grad_norm": 2.0647709369659424, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7021399736404419, + "num_tokens": 416718155.0, + "step": 16220 + }, + { + "epoch": 1.7813529540962003, + "grad_norm": 2.4246301651000977, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7137660980224609, + "num_tokens": 416740328.0, + "step": 16221 + }, + { + "epoch": 1.781462771798814, + "grad_norm": 2.0748305320739746, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6897713541984558, + "num_tokens": 416769581.0, + "step": 16222 + }, + { + "epoch": 1.7815725895014276, + "grad_norm": 2.329246759414673, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7164838910102844, + "num_tokens": 416792091.0, + "step": 16223 + }, + { + "epoch": 1.7816824072040411, + "grad_norm": 2.1600451469421387, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7317154407501221, + "num_tokens": 416817090.0, + "step": 16224 + }, + { + "epoch": 1.781792224906655, + "grad_norm": 2.155877113342285, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7062947750091553, + "num_tokens": 416845105.0, + "step": 16225 + }, + { + "epoch": 1.7819020426092687, + "grad_norm": 2.6870384216308594, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7288556694984436, + "num_tokens": 416864509.0, + "step": 16226 + }, + { + "epoch": 1.7820118603118824, + "grad_norm": 2.109274387359619, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7244046926498413, + "num_tokens": 416892885.0, + "step": 16227 + }, + { + "epoch": 1.782121678014496, + "grad_norm": 2.2103636264801025, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7131603956222534, + "num_tokens": 416919578.0, + "step": 16228 + }, + { + "epoch": 1.7822314957171095, + "grad_norm": 2.127073049545288, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7358501553535461, + "num_tokens": 416946145.0, + "step": 16229 + }, + { + "epoch": 1.7823413134197232, + "grad_norm": 2.3985564708709717, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7329684495925903, + "num_tokens": 416969385.0, + "step": 16230 + }, + { + "epoch": 1.782451131122337, + "grad_norm": 2.3000526428222656, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7217943668365479, + "num_tokens": 416992504.0, + "step": 16231 + }, + { + "epoch": 1.7825609488249505, + "grad_norm": 2.436755418777466, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7057521343231201, + "num_tokens": 417015639.0, + "step": 16232 + }, + { + "epoch": 1.7826707665275643, + "grad_norm": 2.196871519088745, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7194820046424866, + "num_tokens": 417041402.0, + "step": 16233 + }, + { + "epoch": 1.7827805842301778, + "grad_norm": 2.4613656997680664, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7308915853500366, + "num_tokens": 417064413.0, + "step": 16234 + }, + { + "epoch": 1.7828904019327916, + "grad_norm": 2.21075439453125, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7213211059570312, + "num_tokens": 417090744.0, + "step": 16235 + }, + { + "epoch": 1.7830002196354053, + "grad_norm": 2.145573139190674, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6997007131576538, + "num_tokens": 417118852.0, + "step": 16236 + }, + { + "epoch": 1.7831100373380189, + "grad_norm": 2.1144137382507324, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7132478952407837, + "num_tokens": 417146218.0, + "step": 16237 + }, + { + "epoch": 1.7832198550406324, + "grad_norm": 2.1482393741607666, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.722819447517395, + "num_tokens": 417172637.0, + "step": 16238 + }, + { + "epoch": 1.7833296727432462, + "grad_norm": 2.2807180881500244, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.704164445400238, + "num_tokens": 417197470.0, + "step": 16239 + }, + { + "epoch": 1.78343949044586, + "grad_norm": 2.1943764686584473, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.7005419731140137, + "num_tokens": 417225045.0, + "step": 16240 + }, + { + "epoch": 1.7835493081484737, + "grad_norm": 2.529527425765991, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7162621021270752, + "num_tokens": 417247831.0, + "step": 16241 + }, + { + "epoch": 1.7836591258510872, + "grad_norm": 2.2531378269195557, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7292212247848511, + "num_tokens": 417274442.0, + "step": 16242 + }, + { + "epoch": 1.7837689435537007, + "grad_norm": 2.287017583847046, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7369526624679565, + "num_tokens": 417299867.0, + "step": 16243 + }, + { + "epoch": 1.7838787612563145, + "grad_norm": 2.4644296169281006, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7197437286376953, + "num_tokens": 417322387.0, + "step": 16244 + }, + { + "epoch": 1.7839885789589283, + "grad_norm": 2.251509666442871, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.737116813659668, + "num_tokens": 417346073.0, + "step": 16245 + }, + { + "epoch": 1.7840983966615418, + "grad_norm": 1.9409300088882446, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6914818286895752, + "num_tokens": 417378436.0, + "step": 16246 + }, + { + "epoch": 1.7842082143641556, + "grad_norm": 2.060487985610962, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7176923751831055, + "num_tokens": 417407159.0, + "step": 16247 + }, + { + "epoch": 1.784318032066769, + "grad_norm": 2.035611629486084, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7099195122718811, + "num_tokens": 417438427.0, + "step": 16248 + }, + { + "epoch": 1.7844278497693828, + "grad_norm": 2.166158437728882, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7319273948669434, + "num_tokens": 417467337.0, + "step": 16249 + }, + { + "epoch": 1.7845376674719966, + "grad_norm": 2.4667022228240967, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7280054092407227, + "num_tokens": 417488890.0, + "step": 16250 + }, + { + "epoch": 1.7846474851746101, + "grad_norm": 2.1730871200561523, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7350016832351685, + "num_tokens": 417515322.0, + "step": 16251 + }, + { + "epoch": 1.7847573028772237, + "grad_norm": 2.2342700958251953, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.723701000213623, + "num_tokens": 417540746.0, + "step": 16252 + }, + { + "epoch": 1.7848671205798374, + "grad_norm": 2.5071640014648438, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7402567267417908, + "num_tokens": 417561837.0, + "step": 16253 + }, + { + "epoch": 1.7849769382824512, + "grad_norm": 2.397298574447632, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7359691262245178, + "num_tokens": 417583270.0, + "step": 16254 + }, + { + "epoch": 1.785086755985065, + "grad_norm": 2.2932026386260986, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7514293193817139, + "num_tokens": 417604935.0, + "step": 16255 + }, + { + "epoch": 1.7851965736876785, + "grad_norm": 2.3218657970428467, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7066187858581543, + "num_tokens": 417630258.0, + "step": 16256 + }, + { + "epoch": 1.785306391390292, + "grad_norm": 2.171511173248291, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7304320335388184, + "num_tokens": 417655407.0, + "step": 16257 + }, + { + "epoch": 1.7854162090929058, + "grad_norm": 2.0857579708099365, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7332897186279297, + "num_tokens": 417684082.0, + "step": 16258 + }, + { + "epoch": 1.7855260267955195, + "grad_norm": 2.194441556930542, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7219272255897522, + "num_tokens": 417712409.0, + "step": 16259 + }, + { + "epoch": 1.785635844498133, + "grad_norm": 2.21565580368042, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7251392602920532, + "num_tokens": 417737709.0, + "step": 16260 + }, + { + "epoch": 1.7857456622007466, + "grad_norm": 2.2447876930236816, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7289479970932007, + "num_tokens": 417763714.0, + "step": 16261 + }, + { + "epoch": 1.7858554799033604, + "grad_norm": 2.302093982696533, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.715160608291626, + "num_tokens": 417790048.0, + "step": 16262 + }, + { + "epoch": 1.7859652976059741, + "grad_norm": 2.31048583984375, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7248761057853699, + "num_tokens": 417814695.0, + "step": 16263 + }, + { + "epoch": 1.7860751153085879, + "grad_norm": 2.3474793434143066, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7527831792831421, + "num_tokens": 417837541.0, + "step": 16264 + }, + { + "epoch": 1.7861849330112014, + "grad_norm": 2.071756601333618, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7343641519546509, + "num_tokens": 417866076.0, + "step": 16265 + }, + { + "epoch": 1.786294750713815, + "grad_norm": 2.522509813308716, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7360786199569702, + "num_tokens": 417888270.0, + "step": 16266 + }, + { + "epoch": 1.7864045684164287, + "grad_norm": 2.396444797515869, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7264684438705444, + "num_tokens": 417911654.0, + "step": 16267 + }, + { + "epoch": 1.7865143861190425, + "grad_norm": 2.213949680328369, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7140708565711975, + "num_tokens": 417937266.0, + "step": 16268 + }, + { + "epoch": 1.7866242038216562, + "grad_norm": 2.158670663833618, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7119695544242859, + "num_tokens": 417965016.0, + "step": 16269 + }, + { + "epoch": 1.7867340215242697, + "grad_norm": 2.657888650894165, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7398836612701416, + "num_tokens": 417982665.0, + "step": 16270 + }, + { + "epoch": 1.7868438392268833, + "grad_norm": 2.7320914268493652, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7277984619140625, + "num_tokens": 418001884.0, + "step": 16271 + }, + { + "epoch": 1.786953656929497, + "grad_norm": 2.192976951599121, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6969743371009827, + "num_tokens": 418026772.0, + "step": 16272 + }, + { + "epoch": 1.7870634746321108, + "grad_norm": 2.165473461151123, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7179132699966431, + "num_tokens": 418053252.0, + "step": 16273 + }, + { + "epoch": 1.7871732923347243, + "grad_norm": 2.3077821731567383, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7378338575363159, + "num_tokens": 418076425.0, + "step": 16274 + }, + { + "epoch": 1.7872831100373379, + "grad_norm": 2.2627480030059814, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7055642008781433, + "num_tokens": 418102581.0, + "step": 16275 + }, + { + "epoch": 1.7873929277399516, + "grad_norm": 2.1102585792541504, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7202965021133423, + "num_tokens": 418130633.0, + "step": 16276 + }, + { + "epoch": 1.7875027454425654, + "grad_norm": 2.318265914916992, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7115363478660583, + "num_tokens": 418153900.0, + "step": 16277 + }, + { + "epoch": 1.7876125631451791, + "grad_norm": 2.321068048477173, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7284624576568604, + "num_tokens": 418175139.0, + "step": 16278 + }, + { + "epoch": 1.7877223808477927, + "grad_norm": 2.018946409225464, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7282894849777222, + "num_tokens": 418204237.0, + "step": 16279 + }, + { + "epoch": 1.7878321985504062, + "grad_norm": 2.09416127204895, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6942166090011597, + "num_tokens": 418235575.0, + "step": 16280 + }, + { + "epoch": 1.78794201625302, + "grad_norm": 2.437452793121338, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7005407810211182, + "num_tokens": 418259753.0, + "step": 16281 + }, + { + "epoch": 1.7880518339556337, + "grad_norm": 2.4133501052856445, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7548917531967163, + "num_tokens": 418281393.0, + "step": 16282 + }, + { + "epoch": 1.7881616516582475, + "grad_norm": 2.089975595474243, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7174529433250427, + "num_tokens": 418309569.0, + "step": 16283 + }, + { + "epoch": 1.788271469360861, + "grad_norm": 2.3918681144714355, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7337010502815247, + "num_tokens": 418330278.0, + "step": 16284 + }, + { + "epoch": 1.7883812870634745, + "grad_norm": 2.3023383617401123, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7008893489837646, + "num_tokens": 418357392.0, + "step": 16285 + }, + { + "epoch": 1.7884911047660883, + "grad_norm": 2.2976791858673096, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7230697870254517, + "num_tokens": 418383250.0, + "step": 16286 + }, + { + "epoch": 1.788600922468702, + "grad_norm": 1.9098758697509766, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7159445285797119, + "num_tokens": 418417275.0, + "step": 16287 + }, + { + "epoch": 1.7887107401713156, + "grad_norm": 2.401017189025879, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7302378416061401, + "num_tokens": 418440292.0, + "step": 16288 + }, + { + "epoch": 1.7888205578739291, + "grad_norm": 2.1152241230010986, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7137706875801086, + "num_tokens": 418467084.0, + "step": 16289 + }, + { + "epoch": 1.7889303755765429, + "grad_norm": 2.292534351348877, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7110567092895508, + "num_tokens": 418492014.0, + "step": 16290 + }, + { + "epoch": 1.7890401932791566, + "grad_norm": 2.0412840843200684, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.694843053817749, + "num_tokens": 418525167.0, + "step": 16291 + }, + { + "epoch": 1.7891500109817704, + "grad_norm": 2.3826639652252197, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7142607569694519, + "num_tokens": 418547457.0, + "step": 16292 + }, + { + "epoch": 1.789259828684384, + "grad_norm": 2.3695318698883057, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7090311050415039, + "num_tokens": 418571557.0, + "step": 16293 + }, + { + "epoch": 1.7893696463869975, + "grad_norm": 2.4585046768188477, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7283828854560852, + "num_tokens": 418592176.0, + "step": 16294 + }, + { + "epoch": 1.7894794640896112, + "grad_norm": 2.629801034927368, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7071732878684998, + "num_tokens": 418613426.0, + "step": 16295 + }, + { + "epoch": 1.789589281792225, + "grad_norm": 2.052917242050171, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7030375003814697, + "num_tokens": 418642169.0, + "step": 16296 + }, + { + "epoch": 1.7896990994948385, + "grad_norm": 2.346747398376465, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7299008965492249, + "num_tokens": 418666645.0, + "step": 16297 + }, + { + "epoch": 1.7898089171974523, + "grad_norm": 2.2175984382629395, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6921654939651489, + "num_tokens": 418695852.0, + "step": 16298 + }, + { + "epoch": 1.7899187349000658, + "grad_norm": 2.107381582260132, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7083019018173218, + "num_tokens": 418723805.0, + "step": 16299 + }, + { + "epoch": 1.7900285526026796, + "grad_norm": 2.315924644470215, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7094210386276245, + "num_tokens": 418748412.0, + "step": 16300 + }, + { + "epoch": 1.7901383703052933, + "grad_norm": 2.116405725479126, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7263354063034058, + "num_tokens": 418775025.0, + "step": 16301 + }, + { + "epoch": 1.7902481880079069, + "grad_norm": 2.1059558391571045, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7280827760696411, + "num_tokens": 418803739.0, + "step": 16302 + }, + { + "epoch": 1.7903580057105204, + "grad_norm": 2.69838285446167, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7357484102249146, + "num_tokens": 418821412.0, + "step": 16303 + }, + { + "epoch": 1.7904678234131342, + "grad_norm": 2.117338180541992, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7062901854515076, + "num_tokens": 418851475.0, + "step": 16304 + }, + { + "epoch": 1.790577641115748, + "grad_norm": 2.154604911804199, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.725177526473999, + "num_tokens": 418878663.0, + "step": 16305 + }, + { + "epoch": 1.7906874588183617, + "grad_norm": 2.3971915245056152, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7392210364341736, + "num_tokens": 418901070.0, + "step": 16306 + }, + { + "epoch": 1.7907972765209752, + "grad_norm": 2.0505456924438477, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.714401125907898, + "num_tokens": 418930406.0, + "step": 16307 + }, + { + "epoch": 1.7909070942235887, + "grad_norm": 2.193697214126587, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7099584341049194, + "num_tokens": 418960625.0, + "step": 16308 + }, + { + "epoch": 1.7910169119262025, + "grad_norm": 2.281794309616089, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7234516143798828, + "num_tokens": 418984851.0, + "step": 16309 + }, + { + "epoch": 1.7911267296288162, + "grad_norm": 2.31209659576416, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7095023989677429, + "num_tokens": 419009474.0, + "step": 16310 + }, + { + "epoch": 1.7912365473314298, + "grad_norm": 2.621424674987793, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7343481183052063, + "num_tokens": 419028874.0, + "step": 16311 + }, + { + "epoch": 1.7913463650340435, + "grad_norm": 1.9907392263412476, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7146025896072388, + "num_tokens": 419059514.0, + "step": 16312 + }, + { + "epoch": 1.791456182736657, + "grad_norm": 2.3166587352752686, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7055346965789795, + "num_tokens": 419083038.0, + "step": 16313 + }, + { + "epoch": 1.7915660004392708, + "grad_norm": 2.4313011169433594, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7278659343719482, + "num_tokens": 419105560.0, + "step": 16314 + }, + { + "epoch": 1.7916758181418846, + "grad_norm": 2.3434970378875732, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7279080152511597, + "num_tokens": 419128662.0, + "step": 16315 + }, + { + "epoch": 1.7917856358444981, + "grad_norm": 2.8162715435028076, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7407550811767578, + "num_tokens": 419145706.0, + "step": 16316 + }, + { + "epoch": 1.7918954535471117, + "grad_norm": 2.462157726287842, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7232672572135925, + "num_tokens": 419167142.0, + "step": 16317 + }, + { + "epoch": 1.7920052712497254, + "grad_norm": 2.151637315750122, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6957899332046509, + "num_tokens": 419194109.0, + "step": 16318 + }, + { + "epoch": 1.7921150889523392, + "grad_norm": 1.9621667861938477, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7174843549728394, + "num_tokens": 419225770.0, + "step": 16319 + }, + { + "epoch": 1.792224906654953, + "grad_norm": 2.2245662212371826, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7244084477424622, + "num_tokens": 419253604.0, + "step": 16320 + }, + { + "epoch": 1.7923347243575665, + "grad_norm": 2.0427563190460205, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7133580446243286, + "num_tokens": 419281709.0, + "step": 16321 + }, + { + "epoch": 1.79244454206018, + "grad_norm": 2.196453332901001, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.698306679725647, + "num_tokens": 419310592.0, + "step": 16322 + }, + { + "epoch": 1.7925543597627938, + "grad_norm": 2.399782657623291, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.733471691608429, + "num_tokens": 419333873.0, + "step": 16323 + }, + { + "epoch": 1.7926641774654075, + "grad_norm": 2.1010429859161377, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7371994256973267, + "num_tokens": 419359996.0, + "step": 16324 + }, + { + "epoch": 1.792773995168021, + "grad_norm": 2.2798471450805664, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7263574600219727, + "num_tokens": 419386676.0, + "step": 16325 + }, + { + "epoch": 1.7928838128706346, + "grad_norm": 2.35578989982605, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6948226690292358, + "num_tokens": 419413580.0, + "step": 16326 + }, + { + "epoch": 1.7929936305732483, + "grad_norm": 2.2565174102783203, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.711815595626831, + "num_tokens": 419439433.0, + "step": 16327 + }, + { + "epoch": 1.793103448275862, + "grad_norm": 2.0449206829071045, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7334434986114502, + "num_tokens": 419467462.0, + "step": 16328 + }, + { + "epoch": 1.7932132659784759, + "grad_norm": 2.639725923538208, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.741667628288269, + "num_tokens": 419486205.0, + "step": 16329 + }, + { + "epoch": 1.7933230836810894, + "grad_norm": 2.2793586254119873, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7143380641937256, + "num_tokens": 419510738.0, + "step": 16330 + }, + { + "epoch": 1.793432901383703, + "grad_norm": 1.8808953762054443, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7012752294540405, + "num_tokens": 419546626.0, + "step": 16331 + }, + { + "epoch": 1.7935427190863167, + "grad_norm": 2.596874713897705, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7246453166007996, + "num_tokens": 419569176.0, + "step": 16332 + }, + { + "epoch": 1.7936525367889304, + "grad_norm": 2.378748893737793, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7314311861991882, + "num_tokens": 419592373.0, + "step": 16333 + }, + { + "epoch": 1.7937623544915442, + "grad_norm": 2.289703845977783, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7253652811050415, + "num_tokens": 419618422.0, + "step": 16334 + }, + { + "epoch": 1.7938721721941577, + "grad_norm": 2.3161556720733643, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7285540103912354, + "num_tokens": 419641591.0, + "step": 16335 + }, + { + "epoch": 1.7939819898967713, + "grad_norm": 2.2672078609466553, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7499547004699707, + "num_tokens": 419664836.0, + "step": 16336 + }, + { + "epoch": 1.794091807599385, + "grad_norm": 2.3969061374664307, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7317668199539185, + "num_tokens": 419686954.0, + "step": 16337 + }, + { + "epoch": 1.7942016253019988, + "grad_norm": 2.0364432334899902, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7225115895271301, + "num_tokens": 419714567.0, + "step": 16338 + }, + { + "epoch": 1.7943114430046123, + "grad_norm": 1.988243579864502, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6977115869522095, + "num_tokens": 419747003.0, + "step": 16339 + }, + { + "epoch": 1.7944212607072259, + "grad_norm": 2.3604977130889893, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7293236255645752, + "num_tokens": 419770075.0, + "step": 16340 + }, + { + "epoch": 1.7945310784098396, + "grad_norm": 2.601083993911743, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7185906171798706, + "num_tokens": 419790375.0, + "step": 16341 + }, + { + "epoch": 1.7946408961124534, + "grad_norm": 2.2337071895599365, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7195285558700562, + "num_tokens": 419817609.0, + "step": 16342 + }, + { + "epoch": 1.7947507138150671, + "grad_norm": 2.092437744140625, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7251248359680176, + "num_tokens": 419847007.0, + "step": 16343 + }, + { + "epoch": 1.7948605315176807, + "grad_norm": 2.0934019088745117, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7240113615989685, + "num_tokens": 419876387.0, + "step": 16344 + }, + { + "epoch": 1.7949703492202942, + "grad_norm": 2.086874485015869, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7148748636245728, + "num_tokens": 419908757.0, + "step": 16345 + }, + { + "epoch": 1.795080166922908, + "grad_norm": 3.0120630264282227, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7353546619415283, + "num_tokens": 419924266.0, + "step": 16346 + }, + { + "epoch": 1.7951899846255217, + "grad_norm": 2.2776241302490234, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7345105409622192, + "num_tokens": 419948922.0, + "step": 16347 + }, + { + "epoch": 1.7952998023281352, + "grad_norm": 2.249499797821045, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6937153339385986, + "num_tokens": 419975282.0, + "step": 16348 + }, + { + "epoch": 1.795409620030749, + "grad_norm": 2.2967913150787354, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7248943448066711, + "num_tokens": 420000074.0, + "step": 16349 + }, + { + "epoch": 1.7955194377333625, + "grad_norm": 2.426889657974243, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7142716646194458, + "num_tokens": 420023897.0, + "step": 16350 + }, + { + "epoch": 1.7956292554359763, + "grad_norm": 2.3201634883880615, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6960813999176025, + "num_tokens": 420049915.0, + "step": 16351 + }, + { + "epoch": 1.79573907313859, + "grad_norm": 2.187239646911621, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7508264780044556, + "num_tokens": 420075350.0, + "step": 16352 + }, + { + "epoch": 1.7958488908412036, + "grad_norm": 2.4495646953582764, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7375352382659912, + "num_tokens": 420100434.0, + "step": 16353 + }, + { + "epoch": 1.7959587085438171, + "grad_norm": 2.4937331676483154, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.716371476650238, + "num_tokens": 420123768.0, + "step": 16354 + }, + { + "epoch": 1.7960685262464309, + "grad_norm": 2.0752248764038086, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7374519109725952, + "num_tokens": 420153091.0, + "step": 16355 + }, + { + "epoch": 1.7961783439490446, + "grad_norm": 1.9910414218902588, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7140849828720093, + "num_tokens": 420185392.0, + "step": 16356 + }, + { + "epoch": 1.7962881616516584, + "grad_norm": 2.4531190395355225, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7243971228599548, + "num_tokens": 420208507.0, + "step": 16357 + }, + { + "epoch": 1.796397979354272, + "grad_norm": 1.8655686378479004, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7152981162071228, + "num_tokens": 420243207.0, + "step": 16358 + }, + { + "epoch": 1.7965077970568855, + "grad_norm": 2.1131770610809326, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6974618434906006, + "num_tokens": 420272641.0, + "step": 16359 + }, + { + "epoch": 1.7966176147594992, + "grad_norm": 2.1915957927703857, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7104998826980591, + "num_tokens": 420297973.0, + "step": 16360 + }, + { + "epoch": 1.796727432462113, + "grad_norm": 2.3603198528289795, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7497390508651733, + "num_tokens": 420321781.0, + "step": 16361 + }, + { + "epoch": 1.7968372501647265, + "grad_norm": 2.3809733390808105, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7132660150527954, + "num_tokens": 420344298.0, + "step": 16362 + }, + { + "epoch": 1.7969470678673403, + "grad_norm": 2.4807558059692383, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7206760048866272, + "num_tokens": 420367724.0, + "step": 16363 + }, + { + "epoch": 1.7970568855699538, + "grad_norm": 2.2110395431518555, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7353255748748779, + "num_tokens": 420392946.0, + "step": 16364 + }, + { + "epoch": 1.7971667032725676, + "grad_norm": 1.841049075126648, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7088209390640259, + "num_tokens": 420428038.0, + "step": 16365 + }, + { + "epoch": 1.7972765209751813, + "grad_norm": 2.2695930004119873, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7292650938034058, + "num_tokens": 420453784.0, + "step": 16366 + }, + { + "epoch": 1.7973863386777948, + "grad_norm": 2.1928529739379883, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.71982342004776, + "num_tokens": 420478786.0, + "step": 16367 + }, + { + "epoch": 1.7974961563804084, + "grad_norm": 2.2048871517181396, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7074512243270874, + "num_tokens": 420509656.0, + "step": 16368 + }, + { + "epoch": 1.7976059740830221, + "grad_norm": 2.156675100326538, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7304514646530151, + "num_tokens": 420536387.0, + "step": 16369 + }, + { + "epoch": 1.797715791785636, + "grad_norm": 2.339440107345581, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7316356897354126, + "num_tokens": 420559723.0, + "step": 16370 + }, + { + "epoch": 1.7978256094882497, + "grad_norm": 2.0404293537139893, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7263022661209106, + "num_tokens": 420588799.0, + "step": 16371 + }, + { + "epoch": 1.7979354271908632, + "grad_norm": 2.5062077045440674, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7129548192024231, + "num_tokens": 420609781.0, + "step": 16372 + }, + { + "epoch": 1.7980452448934767, + "grad_norm": 1.9303454160690308, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.726994514465332, + "num_tokens": 420641688.0, + "step": 16373 + }, + { + "epoch": 1.7981550625960905, + "grad_norm": 2.657613754272461, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7181185483932495, + "num_tokens": 420662084.0, + "step": 16374 + }, + { + "epoch": 1.7982648802987042, + "grad_norm": 2.2205464839935303, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7528935074806213, + "num_tokens": 420686300.0, + "step": 16375 + }, + { + "epoch": 1.7983746980013178, + "grad_norm": 2.1629512310028076, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7399957180023193, + "num_tokens": 420712497.0, + "step": 16376 + }, + { + "epoch": 1.7984845157039313, + "grad_norm": 2.3037281036376953, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7232520580291748, + "num_tokens": 420736000.0, + "step": 16377 + }, + { + "epoch": 1.798594333406545, + "grad_norm": 2.3213021755218506, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7277477979660034, + "num_tokens": 420759938.0, + "step": 16378 + }, + { + "epoch": 1.7987041511091588, + "grad_norm": 1.9964933395385742, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6972461938858032, + "num_tokens": 420790840.0, + "step": 16379 + }, + { + "epoch": 1.7988139688117726, + "grad_norm": 2.1191353797912598, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7113499045372009, + "num_tokens": 420818655.0, + "step": 16380 + }, + { + "epoch": 1.7989237865143861, + "grad_norm": 1.9856153726577759, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7256011962890625, + "num_tokens": 420851491.0, + "step": 16381 + }, + { + "epoch": 1.7990336042169996, + "grad_norm": 2.127963066101074, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7311281561851501, + "num_tokens": 420878790.0, + "step": 16382 + }, + { + "epoch": 1.7991434219196134, + "grad_norm": 2.0868122577667236, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7101728916168213, + "num_tokens": 420909852.0, + "step": 16383 + }, + { + "epoch": 1.7992532396222272, + "grad_norm": 2.0848300457000732, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.726377546787262, + "num_tokens": 420937174.0, + "step": 16384 + }, + { + "epoch": 1.799363057324841, + "grad_norm": 2.32114577293396, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7036236524581909, + "num_tokens": 420960997.0, + "step": 16385 + }, + { + "epoch": 1.7994728750274545, + "grad_norm": 1.9701178073883057, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7080931663513184, + "num_tokens": 420993154.0, + "step": 16386 + }, + { + "epoch": 1.799582692730068, + "grad_norm": 2.538262367248535, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6976260542869568, + "num_tokens": 421017020.0, + "step": 16387 + }, + { + "epoch": 1.7996925104326817, + "grad_norm": 2.5385541915893555, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7459383606910706, + "num_tokens": 421036574.0, + "step": 16388 + }, + { + "epoch": 1.7998023281352955, + "grad_norm": 2.455864906311035, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7293356657028198, + "num_tokens": 421056932.0, + "step": 16389 + }, + { + "epoch": 1.799912145837909, + "grad_norm": 2.161966323852539, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.707025408744812, + "num_tokens": 421083135.0, + "step": 16390 + }, + { + "epoch": 1.8000219635405226, + "grad_norm": 2.1000499725341797, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7445964813232422, + "num_tokens": 421108555.0, + "step": 16391 + }, + { + "epoch": 1.8001317812431363, + "grad_norm": 2.1898000240325928, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7128372192382812, + "num_tokens": 421134595.0, + "step": 16392 + }, + { + "epoch": 1.80024159894575, + "grad_norm": 2.1901590824127197, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6953024864196777, + "num_tokens": 421160437.0, + "step": 16393 + }, + { + "epoch": 1.8003514166483638, + "grad_norm": 1.9417507648468018, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.6994637250900269, + "num_tokens": 421192169.0, + "step": 16394 + }, + { + "epoch": 1.8004612343509774, + "grad_norm": 2.031296491622925, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.721126914024353, + "num_tokens": 421221836.0, + "step": 16395 + }, + { + "epoch": 1.800571052053591, + "grad_norm": 2.3940396308898926, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7174990177154541, + "num_tokens": 421246923.0, + "step": 16396 + }, + { + "epoch": 1.8006808697562047, + "grad_norm": 2.23443603515625, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7203077077865601, + "num_tokens": 421273730.0, + "step": 16397 + }, + { + "epoch": 1.8007906874588184, + "grad_norm": 2.2874162197113037, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7316663265228271, + "num_tokens": 421298742.0, + "step": 16398 + }, + { + "epoch": 1.8009005051614322, + "grad_norm": 2.255970001220703, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7388380765914917, + "num_tokens": 421322528.0, + "step": 16399 + }, + { + "epoch": 1.8010103228640457, + "grad_norm": 2.1147680282592773, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7089366912841797, + "num_tokens": 421349217.0, + "step": 16400 + }, + { + "epoch": 1.8011201405666593, + "grad_norm": 2.588560104370117, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7468881011009216, + "num_tokens": 421368832.0, + "step": 16401 + }, + { + "epoch": 1.801229958269273, + "grad_norm": 2.1714699268341064, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7209469079971313, + "num_tokens": 421397420.0, + "step": 16402 + }, + { + "epoch": 1.8013397759718868, + "grad_norm": 2.2716689109802246, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7370884418487549, + "num_tokens": 421421767.0, + "step": 16403 + }, + { + "epoch": 1.8014495936745003, + "grad_norm": 2.5517070293426514, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.725696325302124, + "num_tokens": 421441888.0, + "step": 16404 + }, + { + "epoch": 1.8015594113771138, + "grad_norm": 2.3970816135406494, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7211129069328308, + "num_tokens": 421464973.0, + "step": 16405 + }, + { + "epoch": 1.8016692290797276, + "grad_norm": 2.031796455383301, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7230416536331177, + "num_tokens": 421493974.0, + "step": 16406 + }, + { + "epoch": 1.8017790467823414, + "grad_norm": 2.2469630241394043, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.708321213722229, + "num_tokens": 421522239.0, + "step": 16407 + }, + { + "epoch": 1.801888864484955, + "grad_norm": 2.379633903503418, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7158422470092773, + "num_tokens": 421546859.0, + "step": 16408 + }, + { + "epoch": 1.8019986821875686, + "grad_norm": 2.1777093410491943, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7035754323005676, + "num_tokens": 421576591.0, + "step": 16409 + }, + { + "epoch": 1.8021084998901822, + "grad_norm": 2.1005494594573975, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7285876870155334, + "num_tokens": 421605125.0, + "step": 16410 + }, + { + "epoch": 1.802218317592796, + "grad_norm": 2.1965417861938477, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7247294187545776, + "num_tokens": 421633835.0, + "step": 16411 + }, + { + "epoch": 1.8023281352954097, + "grad_norm": 2.048039436340332, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7302685379981995, + "num_tokens": 421662452.0, + "step": 16412 + }, + { + "epoch": 1.8024379529980232, + "grad_norm": 2.1012368202209473, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7249554991722107, + "num_tokens": 421690994.0, + "step": 16413 + }, + { + "epoch": 1.802547770700637, + "grad_norm": 2.3328540325164795, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6984771490097046, + "num_tokens": 421715948.0, + "step": 16414 + }, + { + "epoch": 1.8026575884032505, + "grad_norm": 1.9060035943984985, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6862583160400391, + "num_tokens": 421754208.0, + "step": 16415 + }, + { + "epoch": 1.8027674061058643, + "grad_norm": 2.0467610359191895, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7198367714881897, + "num_tokens": 421783231.0, + "step": 16416 + }, + { + "epoch": 1.802877223808478, + "grad_norm": 1.9149806499481201, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7127314209938049, + "num_tokens": 421817356.0, + "step": 16417 + }, + { + "epoch": 1.8029870415110916, + "grad_norm": 2.254997730255127, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7072587013244629, + "num_tokens": 421842247.0, + "step": 16418 + }, + { + "epoch": 1.803096859213705, + "grad_norm": 2.119684934616089, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7352911233901978, + "num_tokens": 421868331.0, + "step": 16419 + }, + { + "epoch": 1.8032066769163189, + "grad_norm": 2.278804302215576, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6979812979698181, + "num_tokens": 421892762.0, + "step": 16420 + }, + { + "epoch": 1.8033164946189326, + "grad_norm": 2.5719709396362305, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6997419595718384, + "num_tokens": 421914695.0, + "step": 16421 + }, + { + "epoch": 1.8034263123215464, + "grad_norm": 2.233050584793091, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7114788293838501, + "num_tokens": 421939701.0, + "step": 16422 + }, + { + "epoch": 1.80353613002416, + "grad_norm": 2.3140220642089844, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7176812291145325, + "num_tokens": 421962934.0, + "step": 16423 + }, + { + "epoch": 1.8036459477267734, + "grad_norm": 2.031036853790283, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6832724809646606, + "num_tokens": 421993022.0, + "step": 16424 + }, + { + "epoch": 1.8037557654293872, + "grad_norm": 2.2397685050964355, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6992530822753906, + "num_tokens": 422021122.0, + "step": 16425 + }, + { + "epoch": 1.803865583132001, + "grad_norm": 2.389101505279541, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7122654318809509, + "num_tokens": 422044166.0, + "step": 16426 + }, + { + "epoch": 1.8039754008346145, + "grad_norm": 2.1219820976257324, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7320331931114197, + "num_tokens": 422070929.0, + "step": 16427 + }, + { + "epoch": 1.8040852185372283, + "grad_norm": 2.003977060317993, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.741295576095581, + "num_tokens": 422098217.0, + "step": 16428 + }, + { + "epoch": 1.8041950362398418, + "grad_norm": 2.5740692615509033, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7658401727676392, + "num_tokens": 422117040.0, + "step": 16429 + }, + { + "epoch": 1.8043048539424555, + "grad_norm": 2.2225873470306396, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7196612358093262, + "num_tokens": 422143396.0, + "step": 16430 + }, + { + "epoch": 1.8044146716450693, + "grad_norm": 1.9572625160217285, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7026495933532715, + "num_tokens": 422176568.0, + "step": 16431 + }, + { + "epoch": 1.8045244893476828, + "grad_norm": 2.282526731491089, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7284479737281799, + "num_tokens": 422199734.0, + "step": 16432 + }, + { + "epoch": 1.8046343070502964, + "grad_norm": 2.0042505264282227, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7338177561759949, + "num_tokens": 422229092.0, + "step": 16433 + }, + { + "epoch": 1.8047441247529101, + "grad_norm": 2.13348650932312, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7264728546142578, + "num_tokens": 422256293.0, + "step": 16434 + }, + { + "epoch": 1.8048539424555239, + "grad_norm": 2.0562288761138916, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7136359214782715, + "num_tokens": 422286675.0, + "step": 16435 + }, + { + "epoch": 1.8049637601581376, + "grad_norm": 2.7002992630004883, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7495710849761963, + "num_tokens": 422305230.0, + "step": 16436 + }, + { + "epoch": 1.8050735778607512, + "grad_norm": 2.265239715576172, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7203887701034546, + "num_tokens": 422329354.0, + "step": 16437 + }, + { + "epoch": 1.8051833955633647, + "grad_norm": 2.025294303894043, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6987041234970093, + "num_tokens": 422361221.0, + "step": 16438 + }, + { + "epoch": 1.8052932132659785, + "grad_norm": 2.333115816116333, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7276903390884399, + "num_tokens": 422385002.0, + "step": 16439 + }, + { + "epoch": 1.8054030309685922, + "grad_norm": 2.2029998302459717, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.681236207485199, + "num_tokens": 422413482.0, + "step": 16440 + }, + { + "epoch": 1.8055128486712058, + "grad_norm": 2.3551337718963623, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6973085403442383, + "num_tokens": 422438318.0, + "step": 16441 + }, + { + "epoch": 1.8056226663738193, + "grad_norm": 2.3409814834594727, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7252976894378662, + "num_tokens": 422463425.0, + "step": 16442 + }, + { + "epoch": 1.805732484076433, + "grad_norm": 2.0794479846954346, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7042943239212036, + "num_tokens": 422495883.0, + "step": 16443 + }, + { + "epoch": 1.8058423017790468, + "grad_norm": 2.237227201461792, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7148829698562622, + "num_tokens": 422523324.0, + "step": 16444 + }, + { + "epoch": 1.8059521194816606, + "grad_norm": 2.6097750663757324, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7294301986694336, + "num_tokens": 422543952.0, + "step": 16445 + }, + { + "epoch": 1.806061937184274, + "grad_norm": 2.2231979370117188, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7279524803161621, + "num_tokens": 422569805.0, + "step": 16446 + }, + { + "epoch": 1.8061717548868876, + "grad_norm": 2.1953585147857666, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7263011932373047, + "num_tokens": 422596500.0, + "step": 16447 + }, + { + "epoch": 1.8062815725895014, + "grad_norm": 2.3837127685546875, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7385315299034119, + "num_tokens": 422619530.0, + "step": 16448 + }, + { + "epoch": 1.8063913902921152, + "grad_norm": 2.014425277709961, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7075828313827515, + "num_tokens": 422649491.0, + "step": 16449 + }, + { + "epoch": 1.806501207994729, + "grad_norm": 2.3671531677246094, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7396804094314575, + "num_tokens": 422673103.0, + "step": 16450 + }, + { + "epoch": 1.8066110256973424, + "grad_norm": 2.2310709953308105, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7037770748138428, + "num_tokens": 422698628.0, + "step": 16451 + }, + { + "epoch": 1.806720843399956, + "grad_norm": 2.275186538696289, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7212880849838257, + "num_tokens": 422723330.0, + "step": 16452 + }, + { + "epoch": 1.8068306611025697, + "grad_norm": 2.024016857147217, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7392823696136475, + "num_tokens": 422751646.0, + "step": 16453 + }, + { + "epoch": 1.8069404788051835, + "grad_norm": 2.2531309127807617, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7279433608055115, + "num_tokens": 422776717.0, + "step": 16454 + }, + { + "epoch": 1.807050296507797, + "grad_norm": 2.5314643383026123, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7295886874198914, + "num_tokens": 422797919.0, + "step": 16455 + }, + { + "epoch": 1.8071601142104106, + "grad_norm": 2.053370714187622, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7141474485397339, + "num_tokens": 422827457.0, + "step": 16456 + }, + { + "epoch": 1.8072699319130243, + "grad_norm": 2.278982639312744, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7145105004310608, + "num_tokens": 422852598.0, + "step": 16457 + }, + { + "epoch": 1.807379749615638, + "grad_norm": 2.134312868118286, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7057381272315979, + "num_tokens": 422881362.0, + "step": 16458 + }, + { + "epoch": 1.8074895673182518, + "grad_norm": 1.9570823907852173, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.741649329662323, + "num_tokens": 422911265.0, + "step": 16459 + }, + { + "epoch": 1.8075993850208654, + "grad_norm": 2.573840379714966, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7523049116134644, + "num_tokens": 422930537.0, + "step": 16460 + }, + { + "epoch": 1.807709202723479, + "grad_norm": 2.278754234313965, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7191734910011292, + "num_tokens": 422957667.0, + "step": 16461 + }, + { + "epoch": 1.8078190204260927, + "grad_norm": 2.394425868988037, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7427668571472168, + "num_tokens": 422979806.0, + "step": 16462 + }, + { + "epoch": 1.8079288381287064, + "grad_norm": 1.8916939496994019, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.715839147567749, + "num_tokens": 423013584.0, + "step": 16463 + }, + { + "epoch": 1.8080386558313202, + "grad_norm": 2.2880165576934814, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7147655487060547, + "num_tokens": 423036396.0, + "step": 16464 + }, + { + "epoch": 1.8081484735339337, + "grad_norm": 2.1104865074157715, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7256501913070679, + "num_tokens": 423063037.0, + "step": 16465 + }, + { + "epoch": 1.8082582912365472, + "grad_norm": 2.1599647998809814, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7066037654876709, + "num_tokens": 423090300.0, + "step": 16466 + }, + { + "epoch": 1.808368108939161, + "grad_norm": 1.892102599143982, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7168233394622803, + "num_tokens": 423124908.0, + "step": 16467 + }, + { + "epoch": 1.8084779266417748, + "grad_norm": 2.0704989433288574, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7031012177467346, + "num_tokens": 423156486.0, + "step": 16468 + }, + { + "epoch": 1.8085877443443883, + "grad_norm": 2.0668866634368896, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7263541221618652, + "num_tokens": 423186565.0, + "step": 16469 + }, + { + "epoch": 1.8086975620470018, + "grad_norm": 2.3474743366241455, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7253588438034058, + "num_tokens": 423211557.0, + "step": 16470 + }, + { + "epoch": 1.8088073797496156, + "grad_norm": 1.9291170835494995, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7097991704940796, + "num_tokens": 423244548.0, + "step": 16471 + }, + { + "epoch": 1.8089171974522293, + "grad_norm": 2.324695110321045, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7016487717628479, + "num_tokens": 423268990.0, + "step": 16472 + }, + { + "epoch": 1.809027015154843, + "grad_norm": 2.3775224685668945, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7226595282554626, + "num_tokens": 423293745.0, + "step": 16473 + }, + { + "epoch": 1.8091368328574566, + "grad_norm": 2.431044340133667, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6983921527862549, + "num_tokens": 423318258.0, + "step": 16474 + }, + { + "epoch": 1.8092466505600702, + "grad_norm": 2.2447237968444824, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7202306389808655, + "num_tokens": 423342609.0, + "step": 16475 + }, + { + "epoch": 1.809356468262684, + "grad_norm": 2.2403416633605957, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7044734954833984, + "num_tokens": 423366956.0, + "step": 16476 + }, + { + "epoch": 1.8094662859652977, + "grad_norm": 1.9080382585525513, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6891655921936035, + "num_tokens": 423400914.0, + "step": 16477 + }, + { + "epoch": 1.8095761036679112, + "grad_norm": 2.5089728832244873, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7199833393096924, + "num_tokens": 423422806.0, + "step": 16478 + }, + { + "epoch": 1.809685921370525, + "grad_norm": 2.2369790077209473, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7240493297576904, + "num_tokens": 423446814.0, + "step": 16479 + }, + { + "epoch": 1.8097957390731385, + "grad_norm": 2.47639799118042, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7390170097351074, + "num_tokens": 423469377.0, + "step": 16480 + }, + { + "epoch": 1.8099055567757523, + "grad_norm": 2.449493408203125, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7191833257675171, + "num_tokens": 423492217.0, + "step": 16481 + }, + { + "epoch": 1.810015374478366, + "grad_norm": 2.6144580841064453, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7233462333679199, + "num_tokens": 423512691.0, + "step": 16482 + }, + { + "epoch": 1.8101251921809796, + "grad_norm": 2.0762531757354736, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7298852205276489, + "num_tokens": 423542221.0, + "step": 16483 + }, + { + "epoch": 1.810235009883593, + "grad_norm": 2.4515109062194824, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7315517067909241, + "num_tokens": 423565174.0, + "step": 16484 + }, + { + "epoch": 1.8103448275862069, + "grad_norm": 2.2523162364959717, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7187854647636414, + "num_tokens": 423590609.0, + "step": 16485 + }, + { + "epoch": 1.8104546452888206, + "grad_norm": 2.4156851768493652, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7331220507621765, + "num_tokens": 423613625.0, + "step": 16486 + }, + { + "epoch": 1.8105644629914344, + "grad_norm": 2.3198111057281494, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7093658447265625, + "num_tokens": 423638249.0, + "step": 16487 + }, + { + "epoch": 1.810674280694048, + "grad_norm": 2.553279161453247, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7235591411590576, + "num_tokens": 423658625.0, + "step": 16488 + }, + { + "epoch": 1.8107840983966614, + "grad_norm": 1.9160752296447754, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.717247724533081, + "num_tokens": 423693491.0, + "step": 16489 + }, + { + "epoch": 1.8108939160992752, + "grad_norm": 2.0900440216064453, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7105563879013062, + "num_tokens": 423721733.0, + "step": 16490 + }, + { + "epoch": 1.811003733801889, + "grad_norm": 2.0077285766601562, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7160285711288452, + "num_tokens": 423750139.0, + "step": 16491 + }, + { + "epoch": 1.8111135515045025, + "grad_norm": 2.2866158485412598, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7128628492355347, + "num_tokens": 423774302.0, + "step": 16492 + }, + { + "epoch": 1.8112233692071162, + "grad_norm": 2.209965467453003, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7245428562164307, + "num_tokens": 423800420.0, + "step": 16493 + }, + { + "epoch": 1.8113331869097298, + "grad_norm": 2.721304178237915, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7334548234939575, + "num_tokens": 423819062.0, + "step": 16494 + }, + { + "epoch": 1.8114430046123435, + "grad_norm": 2.421968936920166, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7109025120735168, + "num_tokens": 423842314.0, + "step": 16495 + }, + { + "epoch": 1.8115528223149573, + "grad_norm": 2.0753607749938965, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7299834489822388, + "num_tokens": 423872546.0, + "step": 16496 + }, + { + "epoch": 1.8116626400175708, + "grad_norm": 2.537733554840088, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7219628095626831, + "num_tokens": 423892722.0, + "step": 16497 + }, + { + "epoch": 1.8117724577201844, + "grad_norm": 2.4230456352233887, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7362667322158813, + "num_tokens": 423913897.0, + "step": 16498 + }, + { + "epoch": 1.8118822754227981, + "grad_norm": 1.9987397193908691, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7222785949707031, + "num_tokens": 423944721.0, + "step": 16499 + }, + { + "epoch": 1.8119920931254119, + "grad_norm": 2.2415833473205566, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7303202748298645, + "num_tokens": 423969584.0, + "step": 16500 + }, + { + "epoch": 1.8121019108280256, + "grad_norm": 2.1014163494110107, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7177205085754395, + "num_tokens": 423997103.0, + "step": 16501 + }, + { + "epoch": 1.8122117285306392, + "grad_norm": 3.07755184173584, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7176008820533752, + "num_tokens": 424021495.0, + "step": 16502 + }, + { + "epoch": 1.8123215462332527, + "grad_norm": 1.8925001621246338, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7069344520568848, + "num_tokens": 424056302.0, + "step": 16503 + }, + { + "epoch": 1.8124313639358665, + "grad_norm": 2.4300119876861572, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7251890897750854, + "num_tokens": 424079878.0, + "step": 16504 + }, + { + "epoch": 1.8125411816384802, + "grad_norm": 2.1762442588806152, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.6999949216842651, + "num_tokens": 424110079.0, + "step": 16505 + }, + { + "epoch": 1.8126509993410937, + "grad_norm": 2.3572726249694824, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6997946500778198, + "num_tokens": 424136678.0, + "step": 16506 + }, + { + "epoch": 1.8127608170437073, + "grad_norm": 2.3227505683898926, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7042944431304932, + "num_tokens": 424161728.0, + "step": 16507 + }, + { + "epoch": 1.812870634746321, + "grad_norm": 2.6594598293304443, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.719671368598938, + "num_tokens": 424181894.0, + "step": 16508 + }, + { + "epoch": 1.8129804524489348, + "grad_norm": 2.166571855545044, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7146972417831421, + "num_tokens": 424208938.0, + "step": 16509 + }, + { + "epoch": 1.8130902701515486, + "grad_norm": 1.951856017112732, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7192720174789429, + "num_tokens": 424243509.0, + "step": 16510 + }, + { + "epoch": 1.813200087854162, + "grad_norm": 2.487236261367798, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7186311483383179, + "num_tokens": 424265922.0, + "step": 16511 + }, + { + "epoch": 1.8133099055567756, + "grad_norm": 2.3782076835632324, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7763712406158447, + "num_tokens": 424286490.0, + "step": 16512 + }, + { + "epoch": 1.8134197232593894, + "grad_norm": 2.3574275970458984, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7152186036109924, + "num_tokens": 424310800.0, + "step": 16513 + }, + { + "epoch": 1.8135295409620031, + "grad_norm": 2.172523260116577, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7121429443359375, + "num_tokens": 424338293.0, + "step": 16514 + }, + { + "epoch": 1.813639358664617, + "grad_norm": 2.060314893722534, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7329266667366028, + "num_tokens": 424369492.0, + "step": 16515 + }, + { + "epoch": 1.8137491763672304, + "grad_norm": 2.31017804145813, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7232446670532227, + "num_tokens": 424393487.0, + "step": 16516 + }, + { + "epoch": 1.813858994069844, + "grad_norm": 2.6434214115142822, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7236459255218506, + "num_tokens": 424413597.0, + "step": 16517 + }, + { + "epoch": 1.8139688117724577, + "grad_norm": 2.1504805088043213, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7284588813781738, + "num_tokens": 424439569.0, + "step": 16518 + }, + { + "epoch": 1.8140786294750715, + "grad_norm": 2.1863536834716797, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.709342360496521, + "num_tokens": 424466125.0, + "step": 16519 + }, + { + "epoch": 1.814188447177685, + "grad_norm": 2.346346616744995, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7289945483207703, + "num_tokens": 424489312.0, + "step": 16520 + }, + { + "epoch": 1.8142982648802986, + "grad_norm": 2.205772638320923, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7433520555496216, + "num_tokens": 424512873.0, + "step": 16521 + }, + { + "epoch": 1.8144080825829123, + "grad_norm": 2.0577242374420166, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7089296579360962, + "num_tokens": 424542150.0, + "step": 16522 + }, + { + "epoch": 1.814517900285526, + "grad_norm": 2.4120850563049316, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7407251596450806, + "num_tokens": 424563594.0, + "step": 16523 + }, + { + "epoch": 1.8146277179881398, + "grad_norm": 2.578530788421631, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7224876880645752, + "num_tokens": 424584817.0, + "step": 16524 + }, + { + "epoch": 1.8147375356907534, + "grad_norm": 2.482340097427368, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7309823632240295, + "num_tokens": 424604755.0, + "step": 16525 + }, + { + "epoch": 1.814847353393367, + "grad_norm": 2.5760960578918457, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7286885976791382, + "num_tokens": 424625581.0, + "step": 16526 + }, + { + "epoch": 1.8149571710959806, + "grad_norm": 2.2103164196014404, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7011691331863403, + "num_tokens": 424653627.0, + "step": 16527 + }, + { + "epoch": 1.8150669887985944, + "grad_norm": 2.5175909996032715, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7258406281471252, + "num_tokens": 424674265.0, + "step": 16528 + }, + { + "epoch": 1.815176806501208, + "grad_norm": 2.087425708770752, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7131519317626953, + "num_tokens": 424702589.0, + "step": 16529 + }, + { + "epoch": 1.8152866242038217, + "grad_norm": 2.163470506668091, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7331439256668091, + "num_tokens": 424730164.0, + "step": 16530 + }, + { + "epoch": 1.8153964419064352, + "grad_norm": 2.5749855041503906, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7093830108642578, + "num_tokens": 424751786.0, + "step": 16531 + }, + { + "epoch": 1.815506259609049, + "grad_norm": 1.983416199684143, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7113946080207825, + "num_tokens": 424783154.0, + "step": 16532 + }, + { + "epoch": 1.8156160773116627, + "grad_norm": 2.232682943344116, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7226698398590088, + "num_tokens": 424808288.0, + "step": 16533 + }, + { + "epoch": 1.8157258950142763, + "grad_norm": 2.0476346015930176, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7062870264053345, + "num_tokens": 424836859.0, + "step": 16534 + }, + { + "epoch": 1.8158357127168898, + "grad_norm": 2.134498119354248, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7208974957466125, + "num_tokens": 424864729.0, + "step": 16535 + }, + { + "epoch": 1.8159455304195036, + "grad_norm": 2.362088918685913, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6866470575332642, + "num_tokens": 424889551.0, + "step": 16536 + }, + { + "epoch": 1.8160553481221173, + "grad_norm": 2.071873426437378, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7067148089408875, + "num_tokens": 424920993.0, + "step": 16537 + }, + { + "epoch": 1.816165165824731, + "grad_norm": 2.332125663757324, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7195425033569336, + "num_tokens": 424945162.0, + "step": 16538 + }, + { + "epoch": 1.8162749835273446, + "grad_norm": 2.1164486408233643, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.714051365852356, + "num_tokens": 424975276.0, + "step": 16539 + }, + { + "epoch": 1.8163848012299582, + "grad_norm": 2.374943733215332, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7326568961143494, + "num_tokens": 424997442.0, + "step": 16540 + }, + { + "epoch": 1.816494618932572, + "grad_norm": 1.9801708459854126, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7130846381187439, + "num_tokens": 425032998.0, + "step": 16541 + }, + { + "epoch": 1.8166044366351857, + "grad_norm": 2.413594961166382, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.72922682762146, + "num_tokens": 425055196.0, + "step": 16542 + }, + { + "epoch": 1.8167142543377992, + "grad_norm": 2.290614604949951, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7225824594497681, + "num_tokens": 425080939.0, + "step": 16543 + }, + { + "epoch": 1.816824072040413, + "grad_norm": 1.9175187349319458, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.694023847579956, + "num_tokens": 425115227.0, + "step": 16544 + }, + { + "epoch": 1.8169338897430265, + "grad_norm": 2.0226752758026123, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7045201063156128, + "num_tokens": 425146096.0, + "step": 16545 + }, + { + "epoch": 1.8170437074456403, + "grad_norm": 2.3948190212249756, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7248307466506958, + "num_tokens": 425168685.0, + "step": 16546 + }, + { + "epoch": 1.817153525148254, + "grad_norm": 2.0947206020355225, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7160464525222778, + "num_tokens": 425196777.0, + "step": 16547 + }, + { + "epoch": 1.8172633428508675, + "grad_norm": 2.4936487674713135, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7062187194824219, + "num_tokens": 425219921.0, + "step": 16548 + }, + { + "epoch": 1.817373160553481, + "grad_norm": 2.1609139442443848, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7173107266426086, + "num_tokens": 425246182.0, + "step": 16549 + }, + { + "epoch": 1.8174829782560948, + "grad_norm": 2.497652292251587, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7230172753334045, + "num_tokens": 425268522.0, + "step": 16550 + }, + { + "epoch": 1.8175927959587086, + "grad_norm": 2.0463149547576904, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7096116542816162, + "num_tokens": 425298852.0, + "step": 16551 + }, + { + "epoch": 1.8177026136613224, + "grad_norm": 2.40505313873291, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.72969651222229, + "num_tokens": 425322135.0, + "step": 16552 + }, + { + "epoch": 1.8178124313639359, + "grad_norm": 2.017939329147339, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7309153079986572, + "num_tokens": 425352592.0, + "step": 16553 + }, + { + "epoch": 1.8179222490665494, + "grad_norm": 2.199172258377075, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7534111142158508, + "num_tokens": 425375690.0, + "step": 16554 + }, + { + "epoch": 1.8180320667691632, + "grad_norm": 2.532505989074707, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7018991112709045, + "num_tokens": 425400947.0, + "step": 16555 + }, + { + "epoch": 1.818141884471777, + "grad_norm": 2.3202102184295654, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7065613269805908, + "num_tokens": 425425963.0, + "step": 16556 + }, + { + "epoch": 1.8182517021743905, + "grad_norm": 2.4945826530456543, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7286013960838318, + "num_tokens": 425446174.0, + "step": 16557 + }, + { + "epoch": 1.818361519877004, + "grad_norm": 2.0199544429779053, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6912705898284912, + "num_tokens": 425478244.0, + "step": 16558 + }, + { + "epoch": 1.8184713375796178, + "grad_norm": 2.2441954612731934, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7279343605041504, + "num_tokens": 425503803.0, + "step": 16559 + }, + { + "epoch": 1.8185811552822315, + "grad_norm": 2.224761486053467, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.705727756023407, + "num_tokens": 425530075.0, + "step": 16560 + }, + { + "epoch": 1.8186909729848453, + "grad_norm": 2.2863802909851074, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7225377559661865, + "num_tokens": 425553458.0, + "step": 16561 + }, + { + "epoch": 1.8188007906874588, + "grad_norm": 2.3426008224487305, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7137892246246338, + "num_tokens": 425576832.0, + "step": 16562 + }, + { + "epoch": 1.8189106083900723, + "grad_norm": 2.4507594108581543, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7088167667388916, + "num_tokens": 425601236.0, + "step": 16563 + }, + { + "epoch": 1.819020426092686, + "grad_norm": 2.0849764347076416, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.715154230594635, + "num_tokens": 425630647.0, + "step": 16564 + }, + { + "epoch": 1.8191302437952999, + "grad_norm": 2.3031697273254395, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7267087697982788, + "num_tokens": 425654123.0, + "step": 16565 + }, + { + "epoch": 1.8192400614979136, + "grad_norm": 2.5069386959075928, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7519298195838928, + "num_tokens": 425675715.0, + "step": 16566 + }, + { + "epoch": 1.8193498792005272, + "grad_norm": 2.056469202041626, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7279234528541565, + "num_tokens": 425704976.0, + "step": 16567 + }, + { + "epoch": 1.8194596969031407, + "grad_norm": 2.2687790393829346, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7219975590705872, + "num_tokens": 425727900.0, + "step": 16568 + }, + { + "epoch": 1.8195695146057544, + "grad_norm": 2.265430450439453, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7236334085464478, + "num_tokens": 425751139.0, + "step": 16569 + }, + { + "epoch": 1.8196793323083682, + "grad_norm": 2.534726619720459, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7063074111938477, + "num_tokens": 425772755.0, + "step": 16570 + }, + { + "epoch": 1.8197891500109817, + "grad_norm": 2.8005576133728027, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.720626711845398, + "num_tokens": 425790655.0, + "step": 16571 + }, + { + "epoch": 1.8198989677135953, + "grad_norm": 2.295593023300171, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6956690549850464, + "num_tokens": 425816024.0, + "step": 16572 + }, + { + "epoch": 1.820008785416209, + "grad_norm": 2.438209056854248, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7313069105148315, + "num_tokens": 425837677.0, + "step": 16573 + }, + { + "epoch": 1.8201186031188228, + "grad_norm": 2.213620901107788, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7086809873580933, + "num_tokens": 425863913.0, + "step": 16574 + }, + { + "epoch": 1.8202284208214365, + "grad_norm": 2.1594276428222656, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7272793054580688, + "num_tokens": 425889365.0, + "step": 16575 + }, + { + "epoch": 1.82033823852405, + "grad_norm": 2.50506329536438, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7169971466064453, + "num_tokens": 425909991.0, + "step": 16576 + }, + { + "epoch": 1.8204480562266636, + "grad_norm": 2.411062717437744, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7199802994728088, + "num_tokens": 425935041.0, + "step": 16577 + }, + { + "epoch": 1.8205578739292774, + "grad_norm": 2.4089858531951904, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7043013572692871, + "num_tokens": 425958040.0, + "step": 16578 + }, + { + "epoch": 1.8206676916318911, + "grad_norm": 2.078624963760376, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7029839754104614, + "num_tokens": 425986843.0, + "step": 16579 + }, + { + "epoch": 1.8207775093345049, + "grad_norm": 2.2841525077819824, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7368571758270264, + "num_tokens": 426010289.0, + "step": 16580 + }, + { + "epoch": 1.8208873270371184, + "grad_norm": 2.5573246479034424, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7492128610610962, + "num_tokens": 426030099.0, + "step": 16581 + }, + { + "epoch": 1.820997144739732, + "grad_norm": 2.4459481239318848, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7255594730377197, + "num_tokens": 426052380.0, + "step": 16582 + }, + { + "epoch": 1.8211069624423457, + "grad_norm": 2.1536800861358643, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7341214418411255, + "num_tokens": 426080150.0, + "step": 16583 + }, + { + "epoch": 1.8212167801449595, + "grad_norm": 2.073483943939209, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6845889091491699, + "num_tokens": 426110100.0, + "step": 16584 + }, + { + "epoch": 1.821326597847573, + "grad_norm": 2.295809030532837, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7329224348068237, + "num_tokens": 426134011.0, + "step": 16585 + }, + { + "epoch": 1.8214364155501865, + "grad_norm": 2.1002628803253174, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.718481719493866, + "num_tokens": 426162662.0, + "step": 16586 + }, + { + "epoch": 1.8215462332528003, + "grad_norm": 2.2681171894073486, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7390401363372803, + "num_tokens": 426187289.0, + "step": 16587 + }, + { + "epoch": 1.821656050955414, + "grad_norm": 1.9548044204711914, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7061624526977539, + "num_tokens": 426221742.0, + "step": 16588 + }, + { + "epoch": 1.8217658686580278, + "grad_norm": 2.0921239852905273, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7198469042778015, + "num_tokens": 426250543.0, + "step": 16589 + }, + { + "epoch": 1.8218756863606413, + "grad_norm": 2.2493162155151367, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.725894570350647, + "num_tokens": 426274972.0, + "step": 16590 + }, + { + "epoch": 1.8219855040632549, + "grad_norm": 1.9953868389129639, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7076238393783569, + "num_tokens": 426304041.0, + "step": 16591 + }, + { + "epoch": 1.8220953217658686, + "grad_norm": 2.1642067432403564, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7240935564041138, + "num_tokens": 426330935.0, + "step": 16592 + }, + { + "epoch": 1.8222051394684824, + "grad_norm": 1.9250099658966064, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7183547019958496, + "num_tokens": 426363877.0, + "step": 16593 + }, + { + "epoch": 1.822314957171096, + "grad_norm": 2.0131847858428955, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7140192985534668, + "num_tokens": 426395296.0, + "step": 16594 + }, + { + "epoch": 1.8224247748737097, + "grad_norm": 2.3352715969085693, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7296711206436157, + "num_tokens": 426418563.0, + "step": 16595 + }, + { + "epoch": 1.8225345925763232, + "grad_norm": 2.2421462535858154, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.716280996799469, + "num_tokens": 426445005.0, + "step": 16596 + }, + { + "epoch": 1.822644410278937, + "grad_norm": 2.2088782787323, + "learning_rate": 1e-06, + "loss": 1.0949, + "mean_token_accuracy": 0.6840066909790039, + "num_tokens": 426473801.0, + "step": 16597 + }, + { + "epoch": 1.8227542279815507, + "grad_norm": 2.216769218444824, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7137932777404785, + "num_tokens": 426499711.0, + "step": 16598 + }, + { + "epoch": 1.8228640456841643, + "grad_norm": 2.073347568511963, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7234530448913574, + "num_tokens": 426529238.0, + "step": 16599 + }, + { + "epoch": 1.8229738633867778, + "grad_norm": 2.196462392807007, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7158478498458862, + "num_tokens": 426554369.0, + "step": 16600 + }, + { + "epoch": 1.8230836810893916, + "grad_norm": 2.1483864784240723, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7523103356361389, + "num_tokens": 426582328.0, + "step": 16601 + }, + { + "epoch": 1.8231934987920053, + "grad_norm": 2.606607675552368, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7160818576812744, + "num_tokens": 426603141.0, + "step": 16602 + }, + { + "epoch": 1.823303316494619, + "grad_norm": 2.114125967025757, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7261505722999573, + "num_tokens": 426630709.0, + "step": 16603 + }, + { + "epoch": 1.8234131341972326, + "grad_norm": 2.208390235900879, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7262759208679199, + "num_tokens": 426656205.0, + "step": 16604 + }, + { + "epoch": 1.8235229518998461, + "grad_norm": 2.1348559856414795, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7068197131156921, + "num_tokens": 426683617.0, + "step": 16605 + }, + { + "epoch": 1.82363276960246, + "grad_norm": 2.265777111053467, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7286138534545898, + "num_tokens": 426708176.0, + "step": 16606 + }, + { + "epoch": 1.8237425873050737, + "grad_norm": 2.4030981063842773, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7101515531539917, + "num_tokens": 426731225.0, + "step": 16607 + }, + { + "epoch": 1.8238524050076872, + "grad_norm": 2.139899492263794, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7056450843811035, + "num_tokens": 426757828.0, + "step": 16608 + }, + { + "epoch": 1.823962222710301, + "grad_norm": 2.4824554920196533, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7152888178825378, + "num_tokens": 426779153.0, + "step": 16609 + }, + { + "epoch": 1.8240720404129145, + "grad_norm": 2.2062695026397705, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7200114727020264, + "num_tokens": 426804564.0, + "step": 16610 + }, + { + "epoch": 1.8241818581155282, + "grad_norm": 2.5838279724121094, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6948257684707642, + "num_tokens": 426825811.0, + "step": 16611 + }, + { + "epoch": 1.824291675818142, + "grad_norm": 2.3275868892669678, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7277241945266724, + "num_tokens": 426849559.0, + "step": 16612 + }, + { + "epoch": 1.8244014935207555, + "grad_norm": 2.1372292041778564, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7260572910308838, + "num_tokens": 426877621.0, + "step": 16613 + }, + { + "epoch": 1.824511311223369, + "grad_norm": 2.0714399814605713, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7198706865310669, + "num_tokens": 426905431.0, + "step": 16614 + }, + { + "epoch": 1.8246211289259828, + "grad_norm": 2.1360676288604736, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7458363175392151, + "num_tokens": 426932288.0, + "step": 16615 + }, + { + "epoch": 1.8247309466285966, + "grad_norm": 2.128178358078003, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7361997365951538, + "num_tokens": 426958022.0, + "step": 16616 + }, + { + "epoch": 1.8248407643312103, + "grad_norm": 2.158529043197632, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7182095050811768, + "num_tokens": 426985677.0, + "step": 16617 + }, + { + "epoch": 1.8249505820338239, + "grad_norm": 2.049534559249878, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7029531002044678, + "num_tokens": 427017748.0, + "step": 16618 + }, + { + "epoch": 1.8250603997364374, + "grad_norm": 2.013751745223999, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7006791830062866, + "num_tokens": 427048883.0, + "step": 16619 + }, + { + "epoch": 1.8251702174390512, + "grad_norm": 2.1934802532196045, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.725734531879425, + "num_tokens": 427076610.0, + "step": 16620 + }, + { + "epoch": 1.825280035141665, + "grad_norm": 2.337157964706421, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7100443840026855, + "num_tokens": 427100066.0, + "step": 16621 + }, + { + "epoch": 1.8253898528442785, + "grad_norm": 2.2016780376434326, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7324050664901733, + "num_tokens": 427125871.0, + "step": 16622 + }, + { + "epoch": 1.825499670546892, + "grad_norm": 2.486480712890625, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7175264358520508, + "num_tokens": 427149219.0, + "step": 16623 + }, + { + "epoch": 1.8256094882495058, + "grad_norm": 2.5185608863830566, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7596566677093506, + "num_tokens": 427168478.0, + "step": 16624 + }, + { + "epoch": 1.8257193059521195, + "grad_norm": 2.242910146713257, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7233703136444092, + "num_tokens": 427195195.0, + "step": 16625 + }, + { + "epoch": 1.8258291236547333, + "grad_norm": 2.3262758255004883, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7081504464149475, + "num_tokens": 427221175.0, + "step": 16626 + }, + { + "epoch": 1.8259389413573468, + "grad_norm": 2.147152900695801, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7104300260543823, + "num_tokens": 427248938.0, + "step": 16627 + }, + { + "epoch": 1.8260487590599603, + "grad_norm": 2.1562089920043945, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.732488751411438, + "num_tokens": 427275752.0, + "step": 16628 + }, + { + "epoch": 1.826158576762574, + "grad_norm": 2.0940144062042236, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7212027907371521, + "num_tokens": 427304229.0, + "step": 16629 + }, + { + "epoch": 1.8262683944651878, + "grad_norm": 2.185748338699341, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7121949791908264, + "num_tokens": 427330132.0, + "step": 16630 + }, + { + "epoch": 1.8263782121678016, + "grad_norm": 2.2042155265808105, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7046419382095337, + "num_tokens": 427359371.0, + "step": 16631 + }, + { + "epoch": 1.8264880298704151, + "grad_norm": 2.3476145267486572, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7057366371154785, + "num_tokens": 427383609.0, + "step": 16632 + }, + { + "epoch": 1.8265978475730287, + "grad_norm": 2.0632107257843018, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7162380814552307, + "num_tokens": 427411520.0, + "step": 16633 + }, + { + "epoch": 1.8267076652756424, + "grad_norm": 2.0455079078674316, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7203230857849121, + "num_tokens": 427442956.0, + "step": 16634 + }, + { + "epoch": 1.8268174829782562, + "grad_norm": 2.4178786277770996, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7192544937133789, + "num_tokens": 427466812.0, + "step": 16635 + }, + { + "epoch": 1.8269273006808697, + "grad_norm": 2.4135425090789795, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.714881181716919, + "num_tokens": 427489597.0, + "step": 16636 + }, + { + "epoch": 1.8270371183834833, + "grad_norm": 2.2878880500793457, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7035232782363892, + "num_tokens": 427513864.0, + "step": 16637 + }, + { + "epoch": 1.827146936086097, + "grad_norm": 2.0220510959625244, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7109106779098511, + "num_tokens": 427544979.0, + "step": 16638 + }, + { + "epoch": 1.8272567537887108, + "grad_norm": 2.0491507053375244, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.730951726436615, + "num_tokens": 427572502.0, + "step": 16639 + }, + { + "epoch": 1.8273665714913245, + "grad_norm": 2.003633975982666, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7057443261146545, + "num_tokens": 427603833.0, + "step": 16640 + }, + { + "epoch": 1.827476389193938, + "grad_norm": 2.5606658458709717, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7464781999588013, + "num_tokens": 427622002.0, + "step": 16641 + }, + { + "epoch": 1.8275862068965516, + "grad_norm": 2.252601146697998, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.734283983707428, + "num_tokens": 427646116.0, + "step": 16642 + }, + { + "epoch": 1.8276960245991654, + "grad_norm": 2.3094677925109863, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7080386281013489, + "num_tokens": 427671948.0, + "step": 16643 + }, + { + "epoch": 1.8278058423017791, + "grad_norm": 1.9605162143707275, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6907718181610107, + "num_tokens": 427705390.0, + "step": 16644 + }, + { + "epoch": 1.8279156600043929, + "grad_norm": 2.5244576930999756, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7583986520767212, + "num_tokens": 427726296.0, + "step": 16645 + }, + { + "epoch": 1.8280254777070064, + "grad_norm": 2.1257669925689697, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7385880351066589, + "num_tokens": 427753434.0, + "step": 16646 + }, + { + "epoch": 1.82813529540962, + "grad_norm": 2.420485258102417, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7118139266967773, + "num_tokens": 427775328.0, + "step": 16647 + }, + { + "epoch": 1.8282451131122337, + "grad_norm": 2.3186709880828857, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7105672359466553, + "num_tokens": 427796348.0, + "step": 16648 + }, + { + "epoch": 1.8283549308148475, + "grad_norm": 2.1801626682281494, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7269726395606995, + "num_tokens": 427821597.0, + "step": 16649 + }, + { + "epoch": 1.828464748517461, + "grad_norm": 2.4126667976379395, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7026171684265137, + "num_tokens": 427843990.0, + "step": 16650 + }, + { + "epoch": 1.8285745662200745, + "grad_norm": 2.138852834701538, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7172670364379883, + "num_tokens": 427872070.0, + "step": 16651 + }, + { + "epoch": 1.8286843839226883, + "grad_norm": 2.1698555946350098, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7194256782531738, + "num_tokens": 427900642.0, + "step": 16652 + }, + { + "epoch": 1.828794201625302, + "grad_norm": 2.1987459659576416, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7319203615188599, + "num_tokens": 427925915.0, + "step": 16653 + }, + { + "epoch": 1.8289040193279158, + "grad_norm": 2.3733510971069336, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7158346176147461, + "num_tokens": 427950809.0, + "step": 16654 + }, + { + "epoch": 1.8290138370305293, + "grad_norm": 2.1324920654296875, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7212322950363159, + "num_tokens": 427979360.0, + "step": 16655 + }, + { + "epoch": 1.8291236547331429, + "grad_norm": 1.9649395942687988, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7136930227279663, + "num_tokens": 428010002.0, + "step": 16656 + }, + { + "epoch": 1.8292334724357566, + "grad_norm": 2.45198917388916, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7266581058502197, + "num_tokens": 428030984.0, + "step": 16657 + }, + { + "epoch": 1.8293432901383704, + "grad_norm": 2.453174352645874, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7038529515266418, + "num_tokens": 428056743.0, + "step": 16658 + }, + { + "epoch": 1.829453107840984, + "grad_norm": 2.84218692779541, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7205060124397278, + "num_tokens": 428074685.0, + "step": 16659 + }, + { + "epoch": 1.8295629255435977, + "grad_norm": 2.274282932281494, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7182163000106812, + "num_tokens": 428103261.0, + "step": 16660 + }, + { + "epoch": 1.8296727432462112, + "grad_norm": 2.4012482166290283, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.716294527053833, + "num_tokens": 428127988.0, + "step": 16661 + }, + { + "epoch": 1.829782560948825, + "grad_norm": 2.040269374847412, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7172071933746338, + "num_tokens": 428156305.0, + "step": 16662 + }, + { + "epoch": 1.8298923786514387, + "grad_norm": 2.141646385192871, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7336220741271973, + "num_tokens": 428181527.0, + "step": 16663 + }, + { + "epoch": 1.8300021963540523, + "grad_norm": 2.2992806434631348, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7084335684776306, + "num_tokens": 428205251.0, + "step": 16664 + }, + { + "epoch": 1.8301120140566658, + "grad_norm": 2.32612943649292, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7128674983978271, + "num_tokens": 428230176.0, + "step": 16665 + }, + { + "epoch": 1.8302218317592795, + "grad_norm": 2.099682331085205, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7153306603431702, + "num_tokens": 428258908.0, + "step": 16666 + }, + { + "epoch": 1.8303316494618933, + "grad_norm": 2.045063018798828, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7420853972434998, + "num_tokens": 428289421.0, + "step": 16667 + }, + { + "epoch": 1.830441467164507, + "grad_norm": 2.2552380561828613, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7559807300567627, + "num_tokens": 428313388.0, + "step": 16668 + }, + { + "epoch": 1.8305512848671206, + "grad_norm": 2.2100484371185303, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7300107479095459, + "num_tokens": 428337487.0, + "step": 16669 + }, + { + "epoch": 1.8306611025697341, + "grad_norm": 2.5370583534240723, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7268314361572266, + "num_tokens": 428357998.0, + "step": 16670 + }, + { + "epoch": 1.830770920272348, + "grad_norm": 2.0018625259399414, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7251124978065491, + "num_tokens": 428386570.0, + "step": 16671 + }, + { + "epoch": 1.8308807379749616, + "grad_norm": 2.130910873413086, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7200525403022766, + "num_tokens": 428417135.0, + "step": 16672 + }, + { + "epoch": 1.8309905556775752, + "grad_norm": 1.9832491874694824, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.6956894993782043, + "num_tokens": 428448606.0, + "step": 16673 + }, + { + "epoch": 1.831100373380189, + "grad_norm": 2.5959908962249756, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7310843467712402, + "num_tokens": 428470055.0, + "step": 16674 + }, + { + "epoch": 1.8312101910828025, + "grad_norm": 2.3992676734924316, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7482835650444031, + "num_tokens": 428492683.0, + "step": 16675 + }, + { + "epoch": 1.8313200087854162, + "grad_norm": 2.4849915504455566, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7109174728393555, + "num_tokens": 428516799.0, + "step": 16676 + }, + { + "epoch": 1.83142982648803, + "grad_norm": 2.258690595626831, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7259998321533203, + "num_tokens": 428542993.0, + "step": 16677 + }, + { + "epoch": 1.8315396441906435, + "grad_norm": 2.552966356277466, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7109015583992004, + "num_tokens": 428564835.0, + "step": 16678 + }, + { + "epoch": 1.831649461893257, + "grad_norm": 2.3546478748321533, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7155397534370422, + "num_tokens": 428588528.0, + "step": 16679 + }, + { + "epoch": 1.8317592795958708, + "grad_norm": 2.0354137420654297, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7331937551498413, + "num_tokens": 428617897.0, + "step": 16680 + }, + { + "epoch": 1.8318690972984846, + "grad_norm": 2.2160701751708984, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7142291069030762, + "num_tokens": 428643216.0, + "step": 16681 + }, + { + "epoch": 1.8319789150010983, + "grad_norm": 2.371328592300415, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7397368550300598, + "num_tokens": 428666942.0, + "step": 16682 + }, + { + "epoch": 1.8320887327037119, + "grad_norm": 2.23252272605896, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7390958070755005, + "num_tokens": 428689675.0, + "step": 16683 + }, + { + "epoch": 1.8321985504063254, + "grad_norm": 2.16595458984375, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6955040693283081, + "num_tokens": 428717878.0, + "step": 16684 + }, + { + "epoch": 1.8323083681089392, + "grad_norm": 2.4729385375976562, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7172399759292603, + "num_tokens": 428739624.0, + "step": 16685 + }, + { + "epoch": 1.832418185811553, + "grad_norm": 2.2947680950164795, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7242748141288757, + "num_tokens": 428763062.0, + "step": 16686 + }, + { + "epoch": 1.8325280035141664, + "grad_norm": 2.1076958179473877, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7096582651138306, + "num_tokens": 428791502.0, + "step": 16687 + }, + { + "epoch": 1.83263782121678, + "grad_norm": 2.3968894481658936, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7378630638122559, + "num_tokens": 428813534.0, + "step": 16688 + }, + { + "epoch": 1.8327476389193937, + "grad_norm": 2.345303535461426, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.726305365562439, + "num_tokens": 428835352.0, + "step": 16689 + }, + { + "epoch": 1.8328574566220075, + "grad_norm": 2.334778070449829, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7178324460983276, + "num_tokens": 428861789.0, + "step": 16690 + }, + { + "epoch": 1.8329672743246213, + "grad_norm": 2.241485595703125, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7054440975189209, + "num_tokens": 428886935.0, + "step": 16691 + }, + { + "epoch": 1.8330770920272348, + "grad_norm": 2.0381321907043457, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7130520939826965, + "num_tokens": 428915443.0, + "step": 16692 + }, + { + "epoch": 1.8331869097298483, + "grad_norm": 2.142481803894043, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.730067253112793, + "num_tokens": 428943674.0, + "step": 16693 + }, + { + "epoch": 1.833296727432462, + "grad_norm": 1.941787838935852, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7083508968353271, + "num_tokens": 428976643.0, + "step": 16694 + }, + { + "epoch": 1.8334065451350758, + "grad_norm": 2.111668348312378, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7317211627960205, + "num_tokens": 429002517.0, + "step": 16695 + }, + { + "epoch": 1.8335163628376896, + "grad_norm": 2.621908187866211, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7269162535667419, + "num_tokens": 429022095.0, + "step": 16696 + }, + { + "epoch": 1.8336261805403031, + "grad_norm": 2.0443708896636963, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7019991874694824, + "num_tokens": 429053047.0, + "step": 16697 + }, + { + "epoch": 1.8337359982429167, + "grad_norm": 2.4398396015167236, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7415319681167603, + "num_tokens": 429075830.0, + "step": 16698 + }, + { + "epoch": 1.8338458159455304, + "grad_norm": 2.1570591926574707, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7218546867370605, + "num_tokens": 429104997.0, + "step": 16699 + }, + { + "epoch": 1.8339556336481442, + "grad_norm": 2.277052164077759, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7264876365661621, + "num_tokens": 429131144.0, + "step": 16700 + }, + { + "epoch": 1.8340654513507577, + "grad_norm": 2.1637747287750244, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7169561982154846, + "num_tokens": 429156471.0, + "step": 16701 + }, + { + "epoch": 1.8341752690533712, + "grad_norm": 2.647496461868286, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7240511178970337, + "num_tokens": 429177108.0, + "step": 16702 + }, + { + "epoch": 1.834285086755985, + "grad_norm": 2.249082088470459, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7151121497154236, + "num_tokens": 429202887.0, + "step": 16703 + }, + { + "epoch": 1.8343949044585988, + "grad_norm": 2.0958144664764404, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7276139259338379, + "num_tokens": 429230710.0, + "step": 16704 + }, + { + "epoch": 1.8345047221612125, + "grad_norm": 2.0334622859954834, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7119887471199036, + "num_tokens": 429262515.0, + "step": 16705 + }, + { + "epoch": 1.834614539863826, + "grad_norm": 2.2231802940368652, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.720572829246521, + "num_tokens": 429286023.0, + "step": 16706 + }, + { + "epoch": 1.8347243575664396, + "grad_norm": 2.4109909534454346, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.732948899269104, + "num_tokens": 429308819.0, + "step": 16707 + }, + { + "epoch": 1.8348341752690533, + "grad_norm": 2.2069010734558105, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7129188179969788, + "num_tokens": 429336845.0, + "step": 16708 + }, + { + "epoch": 1.834943992971667, + "grad_norm": 2.2218058109283447, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7061526775360107, + "num_tokens": 429363442.0, + "step": 16709 + }, + { + "epoch": 1.8350538106742806, + "grad_norm": 2.2486374378204346, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7129439115524292, + "num_tokens": 429388335.0, + "step": 16710 + }, + { + "epoch": 1.8351636283768944, + "grad_norm": 2.4844202995300293, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7363331317901611, + "num_tokens": 429409217.0, + "step": 16711 + }, + { + "epoch": 1.835273446079508, + "grad_norm": 2.4017763137817383, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7235667705535889, + "num_tokens": 429432475.0, + "step": 16712 + }, + { + "epoch": 1.8353832637821217, + "grad_norm": 2.4358325004577637, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7227985858917236, + "num_tokens": 429455702.0, + "step": 16713 + }, + { + "epoch": 1.8354930814847354, + "grad_norm": 2.396425247192383, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7114846110343933, + "num_tokens": 429479227.0, + "step": 16714 + }, + { + "epoch": 1.835602899187349, + "grad_norm": 2.0517349243164062, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7216901779174805, + "num_tokens": 429507586.0, + "step": 16715 + }, + { + "epoch": 1.8357127168899625, + "grad_norm": 2.1711807250976562, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7158159017562866, + "num_tokens": 429533765.0, + "step": 16716 + }, + { + "epoch": 1.8358225345925763, + "grad_norm": 2.081956386566162, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7211865782737732, + "num_tokens": 429564564.0, + "step": 16717 + }, + { + "epoch": 1.83593235229519, + "grad_norm": 2.163090705871582, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7332092523574829, + "num_tokens": 429592443.0, + "step": 16718 + }, + { + "epoch": 1.8360421699978038, + "grad_norm": 2.0055534839630127, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7018921375274658, + "num_tokens": 429627621.0, + "step": 16719 + }, + { + "epoch": 1.8361519877004173, + "grad_norm": 2.197190999984741, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7376620769500732, + "num_tokens": 429651755.0, + "step": 16720 + }, + { + "epoch": 1.8362618054030309, + "grad_norm": 2.2150673866271973, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.736518144607544, + "num_tokens": 429677246.0, + "step": 16721 + }, + { + "epoch": 1.8363716231056446, + "grad_norm": 2.0810000896453857, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7068369388580322, + "num_tokens": 429705654.0, + "step": 16722 + }, + { + "epoch": 1.8364814408082584, + "grad_norm": 2.173689365386963, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6981434226036072, + "num_tokens": 429732607.0, + "step": 16723 + }, + { + "epoch": 1.836591258510872, + "grad_norm": 2.1694133281707764, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7140651345252991, + "num_tokens": 429762045.0, + "step": 16724 + }, + { + "epoch": 1.8367010762134857, + "grad_norm": 2.414989709854126, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7445142865180969, + "num_tokens": 429783927.0, + "step": 16725 + }, + { + "epoch": 1.8368108939160992, + "grad_norm": 2.6296215057373047, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7334306240081787, + "num_tokens": 429803414.0, + "step": 16726 + }, + { + "epoch": 1.836920711618713, + "grad_norm": 2.267652988433838, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7314957976341248, + "num_tokens": 429827901.0, + "step": 16727 + }, + { + "epoch": 1.8370305293213267, + "grad_norm": 2.578115940093994, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7541948556900024, + "num_tokens": 429847373.0, + "step": 16728 + }, + { + "epoch": 1.8371403470239402, + "grad_norm": 2.3531429767608643, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7355328798294067, + "num_tokens": 429871763.0, + "step": 16729 + }, + { + "epoch": 1.8372501647265538, + "grad_norm": 2.0491044521331787, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.725203275680542, + "num_tokens": 429903685.0, + "step": 16730 + }, + { + "epoch": 1.8373599824291675, + "grad_norm": 2.5559728145599365, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7277052998542786, + "num_tokens": 429924667.0, + "step": 16731 + }, + { + "epoch": 1.8374698001317813, + "grad_norm": 2.131844997406006, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7108625173568726, + "num_tokens": 429951698.0, + "step": 16732 + }, + { + "epoch": 1.837579617834395, + "grad_norm": 2.195603847503662, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7194186449050903, + "num_tokens": 429976509.0, + "step": 16733 + }, + { + "epoch": 1.8376894355370086, + "grad_norm": 2.5465352535247803, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7097926735877991, + "num_tokens": 429998418.0, + "step": 16734 + }, + { + "epoch": 1.8377992532396221, + "grad_norm": 2.205312490463257, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7229536771774292, + "num_tokens": 430022321.0, + "step": 16735 + }, + { + "epoch": 1.8379090709422359, + "grad_norm": 2.327495813369751, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7257454991340637, + "num_tokens": 430047212.0, + "step": 16736 + }, + { + "epoch": 1.8380188886448496, + "grad_norm": 2.0651211738586426, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7151705026626587, + "num_tokens": 430077143.0, + "step": 16737 + }, + { + "epoch": 1.8381287063474632, + "grad_norm": 2.680293560028076, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7308303117752075, + "num_tokens": 430097251.0, + "step": 16738 + }, + { + "epoch": 1.838238524050077, + "grad_norm": 2.036863327026367, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7190271615982056, + "num_tokens": 430126807.0, + "step": 16739 + }, + { + "epoch": 1.8383483417526905, + "grad_norm": 2.195432424545288, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.712539553642273, + "num_tokens": 430154106.0, + "step": 16740 + }, + { + "epoch": 1.8384581594553042, + "grad_norm": 2.095043659210205, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7265269160270691, + "num_tokens": 430183161.0, + "step": 16741 + }, + { + "epoch": 1.838567977157918, + "grad_norm": 2.3671159744262695, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7413632869720459, + "num_tokens": 430204426.0, + "step": 16742 + }, + { + "epoch": 1.8386777948605315, + "grad_norm": 2.2124881744384766, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7040078639984131, + "num_tokens": 430231193.0, + "step": 16743 + }, + { + "epoch": 1.838787612563145, + "grad_norm": 2.3729560375213623, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7249774932861328, + "num_tokens": 430253668.0, + "step": 16744 + }, + { + "epoch": 1.8388974302657588, + "grad_norm": 3.0998728275299072, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7550000548362732, + "num_tokens": 430268929.0, + "step": 16745 + }, + { + "epoch": 1.8390072479683726, + "grad_norm": 2.293942928314209, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6955962181091309, + "num_tokens": 430297598.0, + "step": 16746 + }, + { + "epoch": 1.8391170656709863, + "grad_norm": 2.1564924716949463, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7258546352386475, + "num_tokens": 430323510.0, + "step": 16747 + }, + { + "epoch": 1.8392268833735999, + "grad_norm": 2.105902910232544, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7278182506561279, + "num_tokens": 430353944.0, + "step": 16748 + }, + { + "epoch": 1.8393367010762134, + "grad_norm": 2.1448476314544678, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7266618013381958, + "num_tokens": 430381051.0, + "step": 16749 + }, + { + "epoch": 1.8394465187788271, + "grad_norm": 2.4803152084350586, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.739660382270813, + "num_tokens": 430400720.0, + "step": 16750 + }, + { + "epoch": 1.839556336481441, + "grad_norm": 2.0910720825195312, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6983940601348877, + "num_tokens": 430431301.0, + "step": 16751 + }, + { + "epoch": 1.8396661541840544, + "grad_norm": 2.141649007797241, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7051401734352112, + "num_tokens": 430458733.0, + "step": 16752 + }, + { + "epoch": 1.839775971886668, + "grad_norm": 2.2797789573669434, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7133216857910156, + "num_tokens": 430484612.0, + "step": 16753 + }, + { + "epoch": 1.8398857895892817, + "grad_norm": 2.2469632625579834, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7272001504898071, + "num_tokens": 430510671.0, + "step": 16754 + }, + { + "epoch": 1.8399956072918955, + "grad_norm": 2.6072490215301514, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7404091358184814, + "num_tokens": 430529689.0, + "step": 16755 + }, + { + "epoch": 1.8401054249945092, + "grad_norm": 2.3213884830474854, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7329666614532471, + "num_tokens": 430552311.0, + "step": 16756 + }, + { + "epoch": 1.8402152426971228, + "grad_norm": 2.098674774169922, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7259455919265747, + "num_tokens": 430578407.0, + "step": 16757 + }, + { + "epoch": 1.8403250603997363, + "grad_norm": 2.2612192630767822, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7136862277984619, + "num_tokens": 430605581.0, + "step": 16758 + }, + { + "epoch": 1.84043487810235, + "grad_norm": 2.4123833179473877, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7209001779556274, + "num_tokens": 430629760.0, + "step": 16759 + }, + { + "epoch": 1.8405446958049638, + "grad_norm": 2.378401756286621, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7089134454727173, + "num_tokens": 430652919.0, + "step": 16760 + }, + { + "epoch": 1.8406545135075776, + "grad_norm": 1.9216444492340088, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7130797505378723, + "num_tokens": 430685784.0, + "step": 16761 + }, + { + "epoch": 1.8407643312101911, + "grad_norm": 2.268561363220215, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.713580846786499, + "num_tokens": 430711555.0, + "step": 16762 + }, + { + "epoch": 1.8408741489128047, + "grad_norm": 2.1713759899139404, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.721170961856842, + "num_tokens": 430738646.0, + "step": 16763 + }, + { + "epoch": 1.8409839666154184, + "grad_norm": 2.2158584594726562, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7121729850769043, + "num_tokens": 430766952.0, + "step": 16764 + }, + { + "epoch": 1.8410937843180322, + "grad_norm": 2.5085289478302, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7307379245758057, + "num_tokens": 430789080.0, + "step": 16765 + }, + { + "epoch": 1.8412036020206457, + "grad_norm": 2.150641918182373, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.72236168384552, + "num_tokens": 430815473.0, + "step": 16766 + }, + { + "epoch": 1.8413134197232592, + "grad_norm": 1.965610384941101, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7320669889450073, + "num_tokens": 430847609.0, + "step": 16767 + }, + { + "epoch": 1.841423237425873, + "grad_norm": 2.3980565071105957, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7261167764663696, + "num_tokens": 430868096.0, + "step": 16768 + }, + { + "epoch": 1.8415330551284868, + "grad_norm": 2.1485395431518555, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7138563990592957, + "num_tokens": 430897550.0, + "step": 16769 + }, + { + "epoch": 1.8416428728311005, + "grad_norm": 2.066734552383423, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7097852826118469, + "num_tokens": 430928966.0, + "step": 16770 + }, + { + "epoch": 1.841752690533714, + "grad_norm": 2.2075693607330322, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7261778116226196, + "num_tokens": 430955316.0, + "step": 16771 + }, + { + "epoch": 1.8418625082363276, + "grad_norm": 2.4558420181274414, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.719254732131958, + "num_tokens": 430977280.0, + "step": 16772 + }, + { + "epoch": 1.8419723259389413, + "grad_norm": 2.347905397415161, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7065407037734985, + "num_tokens": 431003694.0, + "step": 16773 + }, + { + "epoch": 1.842082143641555, + "grad_norm": 2.077885866165161, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7499940395355225, + "num_tokens": 431028730.0, + "step": 16774 + }, + { + "epoch": 1.8421919613441686, + "grad_norm": 1.9888288974761963, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7175981998443604, + "num_tokens": 431058906.0, + "step": 16775 + }, + { + "epoch": 1.8423017790467824, + "grad_norm": 2.351069688796997, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7554723024368286, + "num_tokens": 431082496.0, + "step": 16776 + }, + { + "epoch": 1.842411596749396, + "grad_norm": 1.9823709726333618, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7269790768623352, + "num_tokens": 431113138.0, + "step": 16777 + }, + { + "epoch": 1.8425214144520097, + "grad_norm": 2.200312614440918, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7335214614868164, + "num_tokens": 431138314.0, + "step": 16778 + }, + { + "epoch": 1.8426312321546234, + "grad_norm": 2.360086441040039, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7122343182563782, + "num_tokens": 431164144.0, + "step": 16779 + }, + { + "epoch": 1.842741049857237, + "grad_norm": 2.371419668197632, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.728054404258728, + "num_tokens": 431189599.0, + "step": 16780 + }, + { + "epoch": 1.8428508675598505, + "grad_norm": 2.0306029319763184, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.728973388671875, + "num_tokens": 431217388.0, + "step": 16781 + }, + { + "epoch": 1.8429606852624643, + "grad_norm": 2.0904693603515625, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7044341564178467, + "num_tokens": 431244970.0, + "step": 16782 + }, + { + "epoch": 1.843070502965078, + "grad_norm": 2.209690809249878, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7191163301467896, + "num_tokens": 431271428.0, + "step": 16783 + }, + { + "epoch": 1.8431803206676918, + "grad_norm": 2.4240143299102783, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7354264259338379, + "num_tokens": 431293108.0, + "step": 16784 + }, + { + "epoch": 1.8432901383703053, + "grad_norm": 2.42933988571167, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7302771210670471, + "num_tokens": 431316704.0, + "step": 16785 + }, + { + "epoch": 1.8433999560729188, + "grad_norm": 2.135451316833496, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7234783172607422, + "num_tokens": 431343702.0, + "step": 16786 + }, + { + "epoch": 1.8435097737755326, + "grad_norm": 2.136167287826538, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.6981117725372314, + "num_tokens": 431371534.0, + "step": 16787 + }, + { + "epoch": 1.8436195914781464, + "grad_norm": 2.2800352573394775, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7290433645248413, + "num_tokens": 431396160.0, + "step": 16788 + }, + { + "epoch": 1.84372940918076, + "grad_norm": 2.223240375518799, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7347460389137268, + "num_tokens": 431420865.0, + "step": 16789 + }, + { + "epoch": 1.8438392268833736, + "grad_norm": 2.2981765270233154, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7354477643966675, + "num_tokens": 431444967.0, + "step": 16790 + }, + { + "epoch": 1.8439490445859872, + "grad_norm": 2.190361499786377, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.736519992351532, + "num_tokens": 431471906.0, + "step": 16791 + }, + { + "epoch": 1.844058862288601, + "grad_norm": 2.2622692584991455, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7279601097106934, + "num_tokens": 431495137.0, + "step": 16792 + }, + { + "epoch": 1.8441686799912147, + "grad_norm": 2.3589375019073486, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7090080380439758, + "num_tokens": 431519064.0, + "step": 16793 + }, + { + "epoch": 1.8442784976938282, + "grad_norm": 2.3309712409973145, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7306018471717834, + "num_tokens": 431543199.0, + "step": 16794 + }, + { + "epoch": 1.8443883153964418, + "grad_norm": 2.051517963409424, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.708646297454834, + "num_tokens": 431573260.0, + "step": 16795 + }, + { + "epoch": 1.8444981330990555, + "grad_norm": 2.1620073318481445, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7002419233322144, + "num_tokens": 431599878.0, + "step": 16796 + }, + { + "epoch": 1.8446079508016693, + "grad_norm": 1.9079031944274902, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7207971215248108, + "num_tokens": 431632036.0, + "step": 16797 + }, + { + "epoch": 1.844717768504283, + "grad_norm": 2.1410317420959473, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7009098529815674, + "num_tokens": 431662390.0, + "step": 16798 + }, + { + "epoch": 1.8448275862068966, + "grad_norm": 2.409000873565674, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7208398580551147, + "num_tokens": 431684069.0, + "step": 16799 + }, + { + "epoch": 1.84493740390951, + "grad_norm": 2.160684585571289, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.694082498550415, + "num_tokens": 431712512.0, + "step": 16800 + }, + { + "epoch": 1.8450472216121239, + "grad_norm": 2.4561519622802734, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6965245604515076, + "num_tokens": 431735586.0, + "step": 16801 + }, + { + "epoch": 1.8451570393147376, + "grad_norm": 2.4195640087127686, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7246652841567993, + "num_tokens": 431756950.0, + "step": 16802 + }, + { + "epoch": 1.8452668570173512, + "grad_norm": 2.432725191116333, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7188985347747803, + "num_tokens": 431779519.0, + "step": 16803 + }, + { + "epoch": 1.8453766747199647, + "grad_norm": 2.175382137298584, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7228056788444519, + "num_tokens": 431806298.0, + "step": 16804 + }, + { + "epoch": 1.8454864924225785, + "grad_norm": 2.1765010356903076, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7113505601882935, + "num_tokens": 431832512.0, + "step": 16805 + }, + { + "epoch": 1.8455963101251922, + "grad_norm": 2.0175228118896484, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.714888334274292, + "num_tokens": 431863013.0, + "step": 16806 + }, + { + "epoch": 1.845706127827806, + "grad_norm": 2.2602314949035645, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7207356691360474, + "num_tokens": 431888276.0, + "step": 16807 + }, + { + "epoch": 1.8458159455304195, + "grad_norm": 2.277409315109253, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7318348288536072, + "num_tokens": 431914625.0, + "step": 16808 + }, + { + "epoch": 1.845925763233033, + "grad_norm": 2.0468814373016357, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7254732847213745, + "num_tokens": 431942701.0, + "step": 16809 + }, + { + "epoch": 1.8460355809356468, + "grad_norm": 2.071963310241699, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7074071168899536, + "num_tokens": 431971778.0, + "step": 16810 + }, + { + "epoch": 1.8461453986382605, + "grad_norm": 2.634584903717041, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7461729049682617, + "num_tokens": 431991230.0, + "step": 16811 + }, + { + "epoch": 1.8462552163408743, + "grad_norm": 2.111584186553955, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7101471424102783, + "num_tokens": 432018744.0, + "step": 16812 + }, + { + "epoch": 1.8463650340434878, + "grad_norm": 2.3090622425079346, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7322007417678833, + "num_tokens": 432040149.0, + "step": 16813 + }, + { + "epoch": 1.8464748517461014, + "grad_norm": 2.3001933097839355, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7150853872299194, + "num_tokens": 432065449.0, + "step": 16814 + }, + { + "epoch": 1.8465846694487151, + "grad_norm": 2.184767961502075, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7119184732437134, + "num_tokens": 432093951.0, + "step": 16815 + }, + { + "epoch": 1.846694487151329, + "grad_norm": 2.16861891746521, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7276148796081543, + "num_tokens": 432120059.0, + "step": 16816 + }, + { + "epoch": 1.8468043048539424, + "grad_norm": 2.1147353649139404, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7301255464553833, + "num_tokens": 432149745.0, + "step": 16817 + }, + { + "epoch": 1.846914122556556, + "grad_norm": 2.7726519107818604, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7295194864273071, + "num_tokens": 432167228.0, + "step": 16818 + }, + { + "epoch": 1.8470239402591697, + "grad_norm": 2.1613643169403076, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7235015630722046, + "num_tokens": 432194015.0, + "step": 16819 + }, + { + "epoch": 1.8471337579617835, + "grad_norm": 2.5501339435577393, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7313397526741028, + "num_tokens": 432214878.0, + "step": 16820 + }, + { + "epoch": 1.8472435756643972, + "grad_norm": 2.3221235275268555, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7069336175918579, + "num_tokens": 432240558.0, + "step": 16821 + }, + { + "epoch": 1.8473533933670108, + "grad_norm": 2.1676883697509766, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7393075823783875, + "num_tokens": 432264966.0, + "step": 16822 + }, + { + "epoch": 1.8474632110696243, + "grad_norm": 2.0023269653320312, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7218626141548157, + "num_tokens": 432294351.0, + "step": 16823 + }, + { + "epoch": 1.847573028772238, + "grad_norm": 2.0666189193725586, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.737391471862793, + "num_tokens": 432322211.0, + "step": 16824 + }, + { + "epoch": 1.8476828464748518, + "grad_norm": 1.9607462882995605, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7043775916099548, + "num_tokens": 432354160.0, + "step": 16825 + }, + { + "epoch": 1.8477926641774656, + "grad_norm": 2.0444419384002686, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7390128374099731, + "num_tokens": 432382797.0, + "step": 16826 + }, + { + "epoch": 1.847902481880079, + "grad_norm": 2.191143035888672, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7185707092285156, + "num_tokens": 432408026.0, + "step": 16827 + }, + { + "epoch": 1.8480122995826926, + "grad_norm": 2.546653985977173, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7333633899688721, + "num_tokens": 432427870.0, + "step": 16828 + }, + { + "epoch": 1.8481221172853064, + "grad_norm": 2.083815574645996, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7336020469665527, + "num_tokens": 432454170.0, + "step": 16829 + }, + { + "epoch": 1.8482319349879202, + "grad_norm": 2.33536434173584, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7387512922286987, + "num_tokens": 432477035.0, + "step": 16830 + }, + { + "epoch": 1.8483417526905337, + "grad_norm": 1.9766309261322021, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7104846239089966, + "num_tokens": 432509074.0, + "step": 16831 + }, + { + "epoch": 1.8484515703931472, + "grad_norm": 2.2285454273223877, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.718804121017456, + "num_tokens": 432534121.0, + "step": 16832 + }, + { + "epoch": 1.848561388095761, + "grad_norm": 2.1267824172973633, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.678434431552887, + "num_tokens": 432564246.0, + "step": 16833 + }, + { + "epoch": 1.8486712057983747, + "grad_norm": 2.086979866027832, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7158915996551514, + "num_tokens": 432593248.0, + "step": 16834 + }, + { + "epoch": 1.8487810235009885, + "grad_norm": 2.2371060848236084, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7195388078689575, + "num_tokens": 432618433.0, + "step": 16835 + }, + { + "epoch": 1.848890841203602, + "grad_norm": 2.0408079624176025, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6986345052719116, + "num_tokens": 432650119.0, + "step": 16836 + }, + { + "epoch": 1.8490006589062156, + "grad_norm": 2.1824746131896973, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7023062705993652, + "num_tokens": 432678414.0, + "step": 16837 + }, + { + "epoch": 1.8491104766088293, + "grad_norm": 2.5521962642669678, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7342671155929565, + "num_tokens": 432699183.0, + "step": 16838 + }, + { + "epoch": 1.849220294311443, + "grad_norm": 2.3655171394348145, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7255443930625916, + "num_tokens": 432724667.0, + "step": 16839 + }, + { + "epoch": 1.8493301120140566, + "grad_norm": 1.9049723148345947, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7032925486564636, + "num_tokens": 432759330.0, + "step": 16840 + }, + { + "epoch": 1.8494399297166704, + "grad_norm": 2.553342819213867, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7491408586502075, + "num_tokens": 432778669.0, + "step": 16841 + }, + { + "epoch": 1.849549747419284, + "grad_norm": 1.9033747911453247, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6988351941108704, + "num_tokens": 432815291.0, + "step": 16842 + }, + { + "epoch": 1.8496595651218977, + "grad_norm": 2.189687728881836, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7239810824394226, + "num_tokens": 432840121.0, + "step": 16843 + }, + { + "epoch": 1.8497693828245114, + "grad_norm": 2.5314979553222656, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.740836501121521, + "num_tokens": 432860817.0, + "step": 16844 + }, + { + "epoch": 1.849879200527125, + "grad_norm": 1.9601428508758545, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7273192405700684, + "num_tokens": 432891234.0, + "step": 16845 + }, + { + "epoch": 1.8499890182297385, + "grad_norm": 2.1313114166259766, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6972031593322754, + "num_tokens": 432918859.0, + "step": 16846 + }, + { + "epoch": 1.8500988359323522, + "grad_norm": 2.1926231384277344, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7003370523452759, + "num_tokens": 432948549.0, + "step": 16847 + }, + { + "epoch": 1.850208653634966, + "grad_norm": 2.2410247325897217, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7173442244529724, + "num_tokens": 432974852.0, + "step": 16848 + }, + { + "epoch": 1.8503184713375798, + "grad_norm": 2.487548828125, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.726413369178772, + "num_tokens": 432996163.0, + "step": 16849 + }, + { + "epoch": 1.8504282890401933, + "grad_norm": 2.132058620452881, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7279884219169617, + "num_tokens": 433025294.0, + "step": 16850 + }, + { + "epoch": 1.8505381067428068, + "grad_norm": 2.0481226444244385, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.721835196018219, + "num_tokens": 433053387.0, + "step": 16851 + }, + { + "epoch": 1.8506479244454206, + "grad_norm": 2.403115749359131, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7400987148284912, + "num_tokens": 433074835.0, + "step": 16852 + }, + { + "epoch": 1.8507577421480343, + "grad_norm": 2.3240506649017334, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7237576246261597, + "num_tokens": 433097657.0, + "step": 16853 + }, + { + "epoch": 1.8508675598506479, + "grad_norm": 2.2996838092803955, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7227245569229126, + "num_tokens": 433122913.0, + "step": 16854 + }, + { + "epoch": 1.8509773775532616, + "grad_norm": 2.4374310970306396, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7093088030815125, + "num_tokens": 433147458.0, + "step": 16855 + }, + { + "epoch": 1.8510871952558752, + "grad_norm": 2.184840202331543, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.717055082321167, + "num_tokens": 433175394.0, + "step": 16856 + }, + { + "epoch": 1.851197012958489, + "grad_norm": 2.1458334922790527, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7267049551010132, + "num_tokens": 433201358.0, + "step": 16857 + }, + { + "epoch": 1.8513068306611027, + "grad_norm": 2.2365167140960693, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6978633403778076, + "num_tokens": 433227971.0, + "step": 16858 + }, + { + "epoch": 1.8514166483637162, + "grad_norm": 2.438520908355713, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7196193933486938, + "num_tokens": 433250621.0, + "step": 16859 + }, + { + "epoch": 1.8515264660663298, + "grad_norm": 2.405043840408325, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7378701567649841, + "num_tokens": 433273693.0, + "step": 16860 + }, + { + "epoch": 1.8516362837689435, + "grad_norm": 2.237539529800415, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7386729717254639, + "num_tokens": 433298405.0, + "step": 16861 + }, + { + "epoch": 1.8517461014715573, + "grad_norm": 2.317600965499878, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7307876348495483, + "num_tokens": 433322372.0, + "step": 16862 + }, + { + "epoch": 1.851855919174171, + "grad_norm": 2.1193857192993164, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.717540442943573, + "num_tokens": 433348995.0, + "step": 16863 + }, + { + "epoch": 1.8519657368767846, + "grad_norm": 2.3580596446990967, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.742853045463562, + "num_tokens": 433370381.0, + "step": 16864 + }, + { + "epoch": 1.852075554579398, + "grad_norm": 2.099299669265747, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7100108861923218, + "num_tokens": 433401109.0, + "step": 16865 + }, + { + "epoch": 1.8521853722820119, + "grad_norm": 2.258077383041382, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7078124284744263, + "num_tokens": 433426699.0, + "step": 16866 + }, + { + "epoch": 1.8522951899846256, + "grad_norm": 1.9822646379470825, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7153725624084473, + "num_tokens": 433459085.0, + "step": 16867 + }, + { + "epoch": 1.8524050076872391, + "grad_norm": 2.2077553272247314, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7234033942222595, + "num_tokens": 433484852.0, + "step": 16868 + }, + { + "epoch": 1.8525148253898527, + "grad_norm": 2.19097900390625, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7187846899032593, + "num_tokens": 433512362.0, + "step": 16869 + }, + { + "epoch": 1.8526246430924664, + "grad_norm": 1.9292563199996948, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7390344142913818, + "num_tokens": 433542762.0, + "step": 16870 + }, + { + "epoch": 1.8527344607950802, + "grad_norm": 2.1239724159240723, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7174123525619507, + "num_tokens": 433570245.0, + "step": 16871 + }, + { + "epoch": 1.852844278497694, + "grad_norm": 2.1259427070617676, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7210584282875061, + "num_tokens": 433599641.0, + "step": 16872 + }, + { + "epoch": 1.8529540962003075, + "grad_norm": 2.2534985542297363, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7177552580833435, + "num_tokens": 433625187.0, + "step": 16873 + }, + { + "epoch": 1.853063913902921, + "grad_norm": 2.3567404747009277, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7173362970352173, + "num_tokens": 433650074.0, + "step": 16874 + }, + { + "epoch": 1.8531737316055348, + "grad_norm": 2.248762607574463, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7172861099243164, + "num_tokens": 433676541.0, + "step": 16875 + }, + { + "epoch": 1.8532835493081485, + "grad_norm": 2.3253066539764404, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7304701805114746, + "num_tokens": 433700799.0, + "step": 16876 + }, + { + "epoch": 1.8533933670107623, + "grad_norm": 2.352900266647339, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7084609866142273, + "num_tokens": 433725597.0, + "step": 16877 + }, + { + "epoch": 1.8535031847133758, + "grad_norm": 2.1884875297546387, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7585318684577942, + "num_tokens": 433750630.0, + "step": 16878 + }, + { + "epoch": 1.8536130024159894, + "grad_norm": 1.934769868850708, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7284591794013977, + "num_tokens": 433782452.0, + "step": 16879 + }, + { + "epoch": 1.8537228201186031, + "grad_norm": 2.3533670902252197, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7370526790618896, + "num_tokens": 433806440.0, + "step": 16880 + }, + { + "epoch": 1.8538326378212169, + "grad_norm": 2.259045124053955, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.697982907295227, + "num_tokens": 433831926.0, + "step": 16881 + }, + { + "epoch": 1.8539424555238304, + "grad_norm": 2.148143768310547, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7166610360145569, + "num_tokens": 433859372.0, + "step": 16882 + }, + { + "epoch": 1.854052273226444, + "grad_norm": 2.051753520965576, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7169749736785889, + "num_tokens": 433888447.0, + "step": 16883 + }, + { + "epoch": 1.8541620909290577, + "grad_norm": 2.4437530040740967, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.739208459854126, + "num_tokens": 433909144.0, + "step": 16884 + }, + { + "epoch": 1.8542719086316715, + "grad_norm": 2.372058629989624, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.745682954788208, + "num_tokens": 433930519.0, + "step": 16885 + }, + { + "epoch": 1.8543817263342852, + "grad_norm": 2.478069543838501, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7423353791236877, + "num_tokens": 433951497.0, + "step": 16886 + }, + { + "epoch": 1.8544915440368988, + "grad_norm": 2.2916438579559326, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7148669958114624, + "num_tokens": 433974819.0, + "step": 16887 + }, + { + "epoch": 1.8546013617395123, + "grad_norm": 2.220754384994507, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7092413306236267, + "num_tokens": 434003188.0, + "step": 16888 + }, + { + "epoch": 1.854711179442126, + "grad_norm": 2.181082010269165, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7187158465385437, + "num_tokens": 434028939.0, + "step": 16889 + }, + { + "epoch": 1.8548209971447398, + "grad_norm": 2.5202925205230713, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7403889894485474, + "num_tokens": 434049183.0, + "step": 16890 + }, + { + "epoch": 1.8549308148473536, + "grad_norm": 2.167954206466675, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.696586012840271, + "num_tokens": 434078641.0, + "step": 16891 + }, + { + "epoch": 1.855040632549967, + "grad_norm": 2.4854536056518555, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7352015972137451, + "num_tokens": 434097897.0, + "step": 16892 + }, + { + "epoch": 1.8551504502525806, + "grad_norm": 2.5314290523529053, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.712074875831604, + "num_tokens": 434119656.0, + "step": 16893 + }, + { + "epoch": 1.8552602679551944, + "grad_norm": 2.1918869018554688, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6906461119651794, + "num_tokens": 434147952.0, + "step": 16894 + }, + { + "epoch": 1.8553700856578081, + "grad_norm": 2.467454433441162, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7293922305107117, + "num_tokens": 434169856.0, + "step": 16895 + }, + { + "epoch": 1.8554799033604217, + "grad_norm": 2.331202507019043, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7259196043014526, + "num_tokens": 434194219.0, + "step": 16896 + }, + { + "epoch": 1.8555897210630352, + "grad_norm": 2.26939058303833, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7090524435043335, + "num_tokens": 434220393.0, + "step": 16897 + }, + { + "epoch": 1.855699538765649, + "grad_norm": 2.3855502605438232, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.734252393245697, + "num_tokens": 434242825.0, + "step": 16898 + }, + { + "epoch": 1.8558093564682627, + "grad_norm": 2.1701607704162598, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6928403377532959, + "num_tokens": 434273311.0, + "step": 16899 + }, + { + "epoch": 1.8559191741708765, + "grad_norm": 2.3337247371673584, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7058854103088379, + "num_tokens": 434296419.0, + "step": 16900 + }, + { + "epoch": 1.85602899187349, + "grad_norm": 2.138418436050415, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7322248816490173, + "num_tokens": 434325190.0, + "step": 16901 + }, + { + "epoch": 1.8561388095761036, + "grad_norm": 2.321776866912842, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7215330600738525, + "num_tokens": 434347846.0, + "step": 16902 + }, + { + "epoch": 1.8562486272787173, + "grad_norm": 2.4558002948760986, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7022077441215515, + "num_tokens": 434371777.0, + "step": 16903 + }, + { + "epoch": 1.856358444981331, + "grad_norm": 2.099290370941162, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7125911712646484, + "num_tokens": 434400298.0, + "step": 16904 + }, + { + "epoch": 1.8564682626839446, + "grad_norm": 2.0574023723602295, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7260940670967102, + "num_tokens": 434430847.0, + "step": 16905 + }, + { + "epoch": 1.8565780803865584, + "grad_norm": 2.220153570175171, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7106672525405884, + "num_tokens": 434457156.0, + "step": 16906 + }, + { + "epoch": 1.856687898089172, + "grad_norm": 2.332597255706787, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7315426468849182, + "num_tokens": 434481486.0, + "step": 16907 + }, + { + "epoch": 1.8567977157917857, + "grad_norm": 2.211111545562744, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7292914390563965, + "num_tokens": 434507102.0, + "step": 16908 + }, + { + "epoch": 1.8569075334943994, + "grad_norm": 2.63839054107666, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7011846303939819, + "num_tokens": 434529486.0, + "step": 16909 + }, + { + "epoch": 1.857017351197013, + "grad_norm": 1.9893546104431152, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7157684564590454, + "num_tokens": 434561477.0, + "step": 16910 + }, + { + "epoch": 1.8571271688996265, + "grad_norm": 2.173978567123413, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7010318040847778, + "num_tokens": 434589660.0, + "step": 16911 + }, + { + "epoch": 1.8572369866022402, + "grad_norm": 2.350337028503418, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7373667359352112, + "num_tokens": 434611385.0, + "step": 16912 + }, + { + "epoch": 1.857346804304854, + "grad_norm": 2.2033581733703613, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7075444459915161, + "num_tokens": 434638419.0, + "step": 16913 + }, + { + "epoch": 1.8574566220074678, + "grad_norm": 2.348496198654175, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7211036086082458, + "num_tokens": 434661175.0, + "step": 16914 + }, + { + "epoch": 1.8575664397100813, + "grad_norm": 2.1805036067962646, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7438188791275024, + "num_tokens": 434686054.0, + "step": 16915 + }, + { + "epoch": 1.8576762574126948, + "grad_norm": 2.437041759490967, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7193565368652344, + "num_tokens": 434709193.0, + "step": 16916 + }, + { + "epoch": 1.8577860751153086, + "grad_norm": 2.115588426589966, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6912795901298523, + "num_tokens": 434738168.0, + "step": 16917 + }, + { + "epoch": 1.8578958928179223, + "grad_norm": 2.050424575805664, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7062311768531799, + "num_tokens": 434769567.0, + "step": 16918 + }, + { + "epoch": 1.8580057105205359, + "grad_norm": 2.341963052749634, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7031151652336121, + "num_tokens": 434792749.0, + "step": 16919 + }, + { + "epoch": 1.8581155282231496, + "grad_norm": 2.5943713188171387, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7358534336090088, + "num_tokens": 434812818.0, + "step": 16920 + }, + { + "epoch": 1.8582253459257632, + "grad_norm": 2.088576316833496, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6920274496078491, + "num_tokens": 434842745.0, + "step": 16921 + }, + { + "epoch": 1.858335163628377, + "grad_norm": 2.1639461517333984, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7033201456069946, + "num_tokens": 434871498.0, + "step": 16922 + }, + { + "epoch": 1.8584449813309907, + "grad_norm": 2.5413174629211426, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7343283891677856, + "num_tokens": 434892404.0, + "step": 16923 + }, + { + "epoch": 1.8585547990336042, + "grad_norm": 2.269116163253784, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7246645092964172, + "num_tokens": 434916763.0, + "step": 16924 + }, + { + "epoch": 1.8586646167362177, + "grad_norm": 2.0499274730682373, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7316943407058716, + "num_tokens": 434944991.0, + "step": 16925 + }, + { + "epoch": 1.8587744344388315, + "grad_norm": 1.992562174797058, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7253150939941406, + "num_tokens": 434974880.0, + "step": 16926 + }, + { + "epoch": 1.8588842521414453, + "grad_norm": 2.188997268676758, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7124404311180115, + "num_tokens": 435000259.0, + "step": 16927 + }, + { + "epoch": 1.858994069844059, + "grad_norm": 2.3731510639190674, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7320191860198975, + "num_tokens": 435023153.0, + "step": 16928 + }, + { + "epoch": 1.8591038875466726, + "grad_norm": 2.128160238265991, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6865246295928955, + "num_tokens": 435053410.0, + "step": 16929 + }, + { + "epoch": 1.859213705249286, + "grad_norm": 2.301215887069702, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7055331468582153, + "num_tokens": 435078121.0, + "step": 16930 + }, + { + "epoch": 1.8593235229518998, + "grad_norm": 2.729623317718506, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7460307478904724, + "num_tokens": 435097603.0, + "step": 16931 + }, + { + "epoch": 1.8594333406545136, + "grad_norm": 2.5766706466674805, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7252671122550964, + "num_tokens": 435119079.0, + "step": 16932 + }, + { + "epoch": 1.8595431583571271, + "grad_norm": 2.2378084659576416, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.719819188117981, + "num_tokens": 435145135.0, + "step": 16933 + }, + { + "epoch": 1.8596529760597407, + "grad_norm": 2.1494550704956055, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7123498916625977, + "num_tokens": 435172642.0, + "step": 16934 + }, + { + "epoch": 1.8597627937623544, + "grad_norm": 2.2137057781219482, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7447425127029419, + "num_tokens": 435196575.0, + "step": 16935 + }, + { + "epoch": 1.8598726114649682, + "grad_norm": 2.174933433532715, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7028862237930298, + "num_tokens": 435223376.0, + "step": 16936 + }, + { + "epoch": 1.859982429167582, + "grad_norm": 2.4413485527038574, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.720634937286377, + "num_tokens": 435246129.0, + "step": 16937 + }, + { + "epoch": 1.8600922468701955, + "grad_norm": 2.4926586151123047, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.722507894039154, + "num_tokens": 435268014.0, + "step": 16938 + }, + { + "epoch": 1.860202064572809, + "grad_norm": 2.525740385055542, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7517036199569702, + "num_tokens": 435287535.0, + "step": 16939 + }, + { + "epoch": 1.8603118822754228, + "grad_norm": 2.0586423873901367, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.718587338924408, + "num_tokens": 435315960.0, + "step": 16940 + }, + { + "epoch": 1.8604216999780365, + "grad_norm": 2.19288969039917, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7173329591751099, + "num_tokens": 435341378.0, + "step": 16941 + }, + { + "epoch": 1.8605315176806503, + "grad_norm": 2.221343517303467, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7313913106918335, + "num_tokens": 435365817.0, + "step": 16942 + }, + { + "epoch": 1.8606413353832638, + "grad_norm": 1.9488409757614136, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7241960763931274, + "num_tokens": 435397460.0, + "step": 16943 + }, + { + "epoch": 1.8607511530858774, + "grad_norm": 2.3302688598632812, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7334867715835571, + "num_tokens": 435421216.0, + "step": 16944 + }, + { + "epoch": 1.860860970788491, + "grad_norm": 2.193897008895874, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7202252149581909, + "num_tokens": 435448163.0, + "step": 16945 + }, + { + "epoch": 1.8609707884911049, + "grad_norm": 2.2254743576049805, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7175277471542358, + "num_tokens": 435473741.0, + "step": 16946 + }, + { + "epoch": 1.8610806061937184, + "grad_norm": 2.228503942489624, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7035282254219055, + "num_tokens": 435501513.0, + "step": 16947 + }, + { + "epoch": 1.861190423896332, + "grad_norm": 2.2522659301757812, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7246437072753906, + "num_tokens": 435526529.0, + "step": 16948 + }, + { + "epoch": 1.8613002415989457, + "grad_norm": 1.852895736694336, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7203547358512878, + "num_tokens": 435561409.0, + "step": 16949 + }, + { + "epoch": 1.8614100593015594, + "grad_norm": 1.9204181432724, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6922940015792847, + "num_tokens": 435595529.0, + "step": 16950 + }, + { + "epoch": 1.8615198770041732, + "grad_norm": 2.250014543533325, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.722671389579773, + "num_tokens": 435620664.0, + "step": 16951 + }, + { + "epoch": 1.8616296947067867, + "grad_norm": 2.4017863273620605, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7284355163574219, + "num_tokens": 435642630.0, + "step": 16952 + }, + { + "epoch": 1.8617395124094003, + "grad_norm": 2.3231120109558105, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7032412886619568, + "num_tokens": 435666549.0, + "step": 16953 + }, + { + "epoch": 1.861849330112014, + "grad_norm": 2.203141212463379, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7135213613510132, + "num_tokens": 435692546.0, + "step": 16954 + }, + { + "epoch": 1.8619591478146278, + "grad_norm": 2.313504695892334, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7345374822616577, + "num_tokens": 435715507.0, + "step": 16955 + }, + { + "epoch": 1.8620689655172413, + "grad_norm": 2.0985093116760254, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.6962358951568604, + "num_tokens": 435743989.0, + "step": 16956 + }, + { + "epoch": 1.862178783219855, + "grad_norm": 2.045637845993042, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7018964290618896, + "num_tokens": 435771863.0, + "step": 16957 + }, + { + "epoch": 1.8622886009224686, + "grad_norm": 2.202568292617798, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6971070170402527, + "num_tokens": 435801277.0, + "step": 16958 + }, + { + "epoch": 1.8623984186250824, + "grad_norm": 2.1202566623687744, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7094756960868835, + "num_tokens": 435830848.0, + "step": 16959 + }, + { + "epoch": 1.8625082363276961, + "grad_norm": 1.893303394317627, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7128355503082275, + "num_tokens": 435863585.0, + "step": 16960 + }, + { + "epoch": 1.8626180540303097, + "grad_norm": 2.18206787109375, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7224055528640747, + "num_tokens": 435888343.0, + "step": 16961 + }, + { + "epoch": 1.8627278717329232, + "grad_norm": 2.4538135528564453, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7159662842750549, + "num_tokens": 435910345.0, + "step": 16962 + }, + { + "epoch": 1.862837689435537, + "grad_norm": 2.205566167831421, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6921948790550232, + "num_tokens": 435937422.0, + "step": 16963 + }, + { + "epoch": 1.8629475071381507, + "grad_norm": 2.484501361846924, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7069160342216492, + "num_tokens": 435960473.0, + "step": 16964 + }, + { + "epoch": 1.8630573248407645, + "grad_norm": 2.0565426349639893, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7068991661071777, + "num_tokens": 435991275.0, + "step": 16965 + }, + { + "epoch": 1.863167142543378, + "grad_norm": 2.117493152618408, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7339804172515869, + "num_tokens": 436018515.0, + "step": 16966 + }, + { + "epoch": 1.8632769602459915, + "grad_norm": 2.69193959236145, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7272461652755737, + "num_tokens": 436037647.0, + "step": 16967 + }, + { + "epoch": 1.8633867779486053, + "grad_norm": 2.240187883377075, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7089592218399048, + "num_tokens": 436065370.0, + "step": 16968 + }, + { + "epoch": 1.863496595651219, + "grad_norm": 1.9876830577850342, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6954078078269958, + "num_tokens": 436096150.0, + "step": 16969 + }, + { + "epoch": 1.8636064133538326, + "grad_norm": 2.077953338623047, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6960462927818298, + "num_tokens": 436126470.0, + "step": 16970 + }, + { + "epoch": 1.8637162310564463, + "grad_norm": 2.3305420875549316, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7155725955963135, + "num_tokens": 436150357.0, + "step": 16971 + }, + { + "epoch": 1.8638260487590599, + "grad_norm": 2.240137815475464, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7290237545967102, + "num_tokens": 436175024.0, + "step": 16972 + }, + { + "epoch": 1.8639358664616736, + "grad_norm": 2.305931568145752, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7311506867408752, + "num_tokens": 436197995.0, + "step": 16973 + }, + { + "epoch": 1.8640456841642874, + "grad_norm": 2.091068983078003, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7081271409988403, + "num_tokens": 436227532.0, + "step": 16974 + }, + { + "epoch": 1.864155501866901, + "grad_norm": 2.0822503566741943, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7170013785362244, + "num_tokens": 436256446.0, + "step": 16975 + }, + { + "epoch": 1.8642653195695145, + "grad_norm": 2.0357680320739746, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7206146717071533, + "num_tokens": 436284163.0, + "step": 16976 + }, + { + "epoch": 1.8643751372721282, + "grad_norm": 2.2759060859680176, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7232862710952759, + "num_tokens": 436307380.0, + "step": 16977 + }, + { + "epoch": 1.864484954974742, + "grad_norm": 2.0947115421295166, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7052087783813477, + "num_tokens": 436336856.0, + "step": 16978 + }, + { + "epoch": 1.8645947726773557, + "grad_norm": 2.7195749282836914, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7410646080970764, + "num_tokens": 436354762.0, + "step": 16979 + }, + { + "epoch": 1.8647045903799693, + "grad_norm": 2.191931962966919, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7051573395729065, + "num_tokens": 436381422.0, + "step": 16980 + }, + { + "epoch": 1.8648144080825828, + "grad_norm": 2.072535514831543, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7265093326568604, + "num_tokens": 436408062.0, + "step": 16981 + }, + { + "epoch": 1.8649242257851966, + "grad_norm": 2.353534698486328, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7077279090881348, + "num_tokens": 436433017.0, + "step": 16982 + }, + { + "epoch": 1.8650340434878103, + "grad_norm": 2.416149854660034, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.711521565914154, + "num_tokens": 436455699.0, + "step": 16983 + }, + { + "epoch": 1.8651438611904239, + "grad_norm": 2.27163028717041, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7284209728240967, + "num_tokens": 436480432.0, + "step": 16984 + }, + { + "epoch": 1.8652536788930374, + "grad_norm": 2.245697498321533, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7244784832000732, + "num_tokens": 436504446.0, + "step": 16985 + }, + { + "epoch": 1.8653634965956511, + "grad_norm": 2.168229579925537, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6998780369758606, + "num_tokens": 436533842.0, + "step": 16986 + }, + { + "epoch": 1.865473314298265, + "grad_norm": 2.2097532749176025, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7185303568840027, + "num_tokens": 436562297.0, + "step": 16987 + }, + { + "epoch": 1.8655831320008787, + "grad_norm": 2.341336250305176, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7083561420440674, + "num_tokens": 436588131.0, + "step": 16988 + }, + { + "epoch": 1.8656929497034922, + "grad_norm": 2.349421977996826, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7217448949813843, + "num_tokens": 436612613.0, + "step": 16989 + }, + { + "epoch": 1.8658027674061057, + "grad_norm": 2.4788191318511963, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7222198247909546, + "num_tokens": 436634282.0, + "step": 16990 + }, + { + "epoch": 1.8659125851087195, + "grad_norm": 2.2857022285461426, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7161887884140015, + "num_tokens": 436661003.0, + "step": 16991 + }, + { + "epoch": 1.8660224028113332, + "grad_norm": 2.1892006397247314, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7048430442810059, + "num_tokens": 436690013.0, + "step": 16992 + }, + { + "epoch": 1.866132220513947, + "grad_norm": 2.5248477458953857, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7350367307662964, + "num_tokens": 436711365.0, + "step": 16993 + }, + { + "epoch": 1.8662420382165605, + "grad_norm": 2.157580614089966, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7279832363128662, + "num_tokens": 436738142.0, + "step": 16994 + }, + { + "epoch": 1.866351855919174, + "grad_norm": 2.5844292640686035, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7446017861366272, + "num_tokens": 436758100.0, + "step": 16995 + }, + { + "epoch": 1.8664616736217878, + "grad_norm": 2.367910861968994, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7310816049575806, + "num_tokens": 436781808.0, + "step": 16996 + }, + { + "epoch": 1.8665714913244016, + "grad_norm": 2.402353525161743, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7297720313072205, + "num_tokens": 436804142.0, + "step": 16997 + }, + { + "epoch": 1.8666813090270151, + "grad_norm": 2.1808347702026367, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7196604609489441, + "num_tokens": 436831361.0, + "step": 16998 + }, + { + "epoch": 1.8667911267296287, + "grad_norm": 2.5923619270324707, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7396117448806763, + "num_tokens": 436851540.0, + "step": 16999 + }, + { + "epoch": 1.8669009444322424, + "grad_norm": 2.092702865600586, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7525756359100342, + "num_tokens": 436879653.0, + "step": 17000 + }, + { + "epoch": 1.8670107621348562, + "grad_norm": 2.0918643474578857, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7136763334274292, + "num_tokens": 436907428.0, + "step": 17001 + }, + { + "epoch": 1.86712057983747, + "grad_norm": 2.3945536613464355, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7507915496826172, + "num_tokens": 436928391.0, + "step": 17002 + }, + { + "epoch": 1.8672303975400835, + "grad_norm": 2.2238805294036865, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7384018301963806, + "num_tokens": 436953116.0, + "step": 17003 + }, + { + "epoch": 1.867340215242697, + "grad_norm": 2.2067301273345947, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7155913710594177, + "num_tokens": 436978252.0, + "step": 17004 + }, + { + "epoch": 1.8674500329453108, + "grad_norm": 2.1916093826293945, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.737873911857605, + "num_tokens": 437002687.0, + "step": 17005 + }, + { + "epoch": 1.8675598506479245, + "grad_norm": 2.2269725799560547, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7147582769393921, + "num_tokens": 437026544.0, + "step": 17006 + }, + { + "epoch": 1.8676696683505383, + "grad_norm": 2.3930017948150635, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7195831537246704, + "num_tokens": 437050905.0, + "step": 17007 + }, + { + "epoch": 1.8677794860531518, + "grad_norm": 2.3857431411743164, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7229914665222168, + "num_tokens": 437073709.0, + "step": 17008 + }, + { + "epoch": 1.8678893037557653, + "grad_norm": 2.5533883571624756, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7170125246047974, + "num_tokens": 437093969.0, + "step": 17009 + }, + { + "epoch": 1.867999121458379, + "grad_norm": 2.3420536518096924, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7144919633865356, + "num_tokens": 437118028.0, + "step": 17010 + }, + { + "epoch": 1.8681089391609929, + "grad_norm": 2.3134336471557617, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7295284271240234, + "num_tokens": 437141930.0, + "step": 17011 + }, + { + "epoch": 1.8682187568636064, + "grad_norm": 2.2217602729797363, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7274767756462097, + "num_tokens": 437168198.0, + "step": 17012 + }, + { + "epoch": 1.86832857456622, + "grad_norm": 2.1527910232543945, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7198346853256226, + "num_tokens": 437195902.0, + "step": 17013 + }, + { + "epoch": 1.8684383922688337, + "grad_norm": 2.3432304859161377, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7172436714172363, + "num_tokens": 437220649.0, + "step": 17014 + }, + { + "epoch": 1.8685482099714474, + "grad_norm": 2.3842718601226807, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7140940427780151, + "num_tokens": 437243715.0, + "step": 17015 + }, + { + "epoch": 1.8686580276740612, + "grad_norm": 2.245950937271118, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7071869373321533, + "num_tokens": 437269940.0, + "step": 17016 + }, + { + "epoch": 1.8687678453766747, + "grad_norm": 2.005363702774048, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7122379541397095, + "num_tokens": 437300324.0, + "step": 17017 + }, + { + "epoch": 1.8688776630792883, + "grad_norm": 2.106672763824463, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7215609550476074, + "num_tokens": 437328603.0, + "step": 17018 + }, + { + "epoch": 1.868987480781902, + "grad_norm": 1.94968843460083, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7042160630226135, + "num_tokens": 437362052.0, + "step": 17019 + }, + { + "epoch": 1.8690972984845158, + "grad_norm": 2.3781440258026123, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7539675235748291, + "num_tokens": 437384324.0, + "step": 17020 + }, + { + "epoch": 1.8692071161871293, + "grad_norm": 2.1644399166107178, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7220808863639832, + "num_tokens": 437409802.0, + "step": 17021 + }, + { + "epoch": 1.869316933889743, + "grad_norm": 2.131035804748535, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7254610061645508, + "num_tokens": 437435480.0, + "step": 17022 + }, + { + "epoch": 1.8694267515923566, + "grad_norm": 2.1502597332000732, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7297278642654419, + "num_tokens": 437461682.0, + "step": 17023 + }, + { + "epoch": 1.8695365692949704, + "grad_norm": 2.20707631111145, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7170153856277466, + "num_tokens": 437488126.0, + "step": 17024 + }, + { + "epoch": 1.8696463869975841, + "grad_norm": 2.4320223331451416, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7404393553733826, + "num_tokens": 437511205.0, + "step": 17025 + }, + { + "epoch": 1.8697562047001977, + "grad_norm": 2.2677087783813477, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7357348203659058, + "num_tokens": 437534166.0, + "step": 17026 + }, + { + "epoch": 1.8698660224028112, + "grad_norm": 2.3198912143707275, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6995803713798523, + "num_tokens": 437559471.0, + "step": 17027 + }, + { + "epoch": 1.869975840105425, + "grad_norm": 2.371021032333374, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7275718450546265, + "num_tokens": 437581666.0, + "step": 17028 + }, + { + "epoch": 1.8700856578080387, + "grad_norm": 2.2853879928588867, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6995512247085571, + "num_tokens": 437607618.0, + "step": 17029 + }, + { + "epoch": 1.8701954755106525, + "grad_norm": 2.2733302116394043, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7128045558929443, + "num_tokens": 437634321.0, + "step": 17030 + }, + { + "epoch": 1.870305293213266, + "grad_norm": 2.9264628887176514, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7521674633026123, + "num_tokens": 437649576.0, + "step": 17031 + }, + { + "epoch": 1.8704151109158795, + "grad_norm": 2.185088872909546, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7264949083328247, + "num_tokens": 437674366.0, + "step": 17032 + }, + { + "epoch": 1.8705249286184933, + "grad_norm": 2.452261209487915, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7307741045951843, + "num_tokens": 437696654.0, + "step": 17033 + }, + { + "epoch": 1.870634746321107, + "grad_norm": 2.100053310394287, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7220229506492615, + "num_tokens": 437725791.0, + "step": 17034 + }, + { + "epoch": 1.8707445640237206, + "grad_norm": 2.2579386234283447, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7191624641418457, + "num_tokens": 437751364.0, + "step": 17035 + }, + { + "epoch": 1.8708543817263343, + "grad_norm": 2.166414499282837, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.705974817276001, + "num_tokens": 437779170.0, + "step": 17036 + }, + { + "epoch": 1.8709641994289479, + "grad_norm": 2.1406610012054443, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6954995393753052, + "num_tokens": 437806226.0, + "step": 17037 + }, + { + "epoch": 1.8710740171315616, + "grad_norm": 2.434128999710083, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7401736974716187, + "num_tokens": 437828807.0, + "step": 17038 + }, + { + "epoch": 1.8711838348341754, + "grad_norm": 2.140343427658081, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7251561880111694, + "num_tokens": 437854250.0, + "step": 17039 + }, + { + "epoch": 1.871293652536789, + "grad_norm": 2.2385592460632324, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7017430067062378, + "num_tokens": 437880389.0, + "step": 17040 + }, + { + "epoch": 1.8714034702394025, + "grad_norm": 2.427267551422119, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7341623306274414, + "num_tokens": 437902664.0, + "step": 17041 + }, + { + "epoch": 1.8715132879420162, + "grad_norm": 2.260082244873047, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7363757491111755, + "num_tokens": 437927578.0, + "step": 17042 + }, + { + "epoch": 1.87162310564463, + "grad_norm": 2.132814407348633, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7411688566207886, + "num_tokens": 437953169.0, + "step": 17043 + }, + { + "epoch": 1.8717329233472437, + "grad_norm": 2.1377365589141846, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7176154851913452, + "num_tokens": 437979095.0, + "step": 17044 + }, + { + "epoch": 1.8718427410498573, + "grad_norm": 2.3404529094696045, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7558707594871521, + "num_tokens": 438002295.0, + "step": 17045 + }, + { + "epoch": 1.8719525587524708, + "grad_norm": 2.054898738861084, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7036567330360413, + "num_tokens": 438036584.0, + "step": 17046 + }, + { + "epoch": 1.8720623764550846, + "grad_norm": 2.3612561225891113, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7255582809448242, + "num_tokens": 438058570.0, + "step": 17047 + }, + { + "epoch": 1.8721721941576983, + "grad_norm": 2.216615915298462, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7258229851722717, + "num_tokens": 438084020.0, + "step": 17048 + }, + { + "epoch": 1.8722820118603118, + "grad_norm": 2.1772027015686035, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7044853568077087, + "num_tokens": 438111694.0, + "step": 17049 + }, + { + "epoch": 1.8723918295629254, + "grad_norm": 2.293178081512451, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7229125499725342, + "num_tokens": 438137802.0, + "step": 17050 + }, + { + "epoch": 1.8725016472655391, + "grad_norm": 2.2740941047668457, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7267552614212036, + "num_tokens": 438163456.0, + "step": 17051 + }, + { + "epoch": 1.872611464968153, + "grad_norm": 2.172154426574707, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7007616758346558, + "num_tokens": 438191494.0, + "step": 17052 + }, + { + "epoch": 1.8727212826707667, + "grad_norm": 2.086228847503662, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7208577990531921, + "num_tokens": 438220037.0, + "step": 17053 + }, + { + "epoch": 1.8728311003733802, + "grad_norm": 2.1575510501861572, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7033877372741699, + "num_tokens": 438247707.0, + "step": 17054 + }, + { + "epoch": 1.8729409180759937, + "grad_norm": 2.149653434753418, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7107926607131958, + "num_tokens": 438276023.0, + "step": 17055 + }, + { + "epoch": 1.8730507357786075, + "grad_norm": 2.2990264892578125, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7180555462837219, + "num_tokens": 438299987.0, + "step": 17056 + }, + { + "epoch": 1.8731605534812212, + "grad_norm": 2.1387550830841064, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7224283218383789, + "num_tokens": 438328330.0, + "step": 17057 + }, + { + "epoch": 1.873270371183835, + "grad_norm": 2.7859370708465576, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7476880550384521, + "num_tokens": 438345618.0, + "step": 17058 + }, + { + "epoch": 1.8733801888864485, + "grad_norm": 2.654313325881958, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7328664064407349, + "num_tokens": 438364869.0, + "step": 17059 + }, + { + "epoch": 1.873490006589062, + "grad_norm": 2.0033798217773438, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7105205655097961, + "num_tokens": 438396917.0, + "step": 17060 + }, + { + "epoch": 1.8735998242916758, + "grad_norm": 1.9876819849014282, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.688391387462616, + "num_tokens": 438431837.0, + "step": 17061 + }, + { + "epoch": 1.8737096419942896, + "grad_norm": 2.392151117324829, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7352268695831299, + "num_tokens": 438453752.0, + "step": 17062 + }, + { + "epoch": 1.8738194596969031, + "grad_norm": 2.3342175483703613, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.714529275894165, + "num_tokens": 438478531.0, + "step": 17063 + }, + { + "epoch": 1.8739292773995166, + "grad_norm": 2.115818500518799, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7060994505882263, + "num_tokens": 438508154.0, + "step": 17064 + }, + { + "epoch": 1.8740390951021304, + "grad_norm": 2.189490556716919, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7482702732086182, + "num_tokens": 438533170.0, + "step": 17065 + }, + { + "epoch": 1.8741489128047442, + "grad_norm": 2.5154876708984375, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7345795035362244, + "num_tokens": 438553491.0, + "step": 17066 + }, + { + "epoch": 1.874258730507358, + "grad_norm": 2.1825814247131348, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6967716217041016, + "num_tokens": 438580091.0, + "step": 17067 + }, + { + "epoch": 1.8743685482099715, + "grad_norm": 2.100212812423706, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7279875874519348, + "num_tokens": 438609390.0, + "step": 17068 + }, + { + "epoch": 1.874478365912585, + "grad_norm": 2.1516711711883545, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7301605939865112, + "num_tokens": 438636147.0, + "step": 17069 + }, + { + "epoch": 1.8745881836151987, + "grad_norm": 2.149883985519409, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.730836808681488, + "num_tokens": 438662860.0, + "step": 17070 + }, + { + "epoch": 1.8746980013178125, + "grad_norm": 2.304412603378296, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7215702533721924, + "num_tokens": 438689074.0, + "step": 17071 + }, + { + "epoch": 1.8748078190204263, + "grad_norm": 2.0644912719726562, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.707187294960022, + "num_tokens": 438720186.0, + "step": 17072 + }, + { + "epoch": 1.8749176367230398, + "grad_norm": 2.2270703315734863, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.721795380115509, + "num_tokens": 438745186.0, + "step": 17073 + }, + { + "epoch": 1.8750274544256533, + "grad_norm": 2.6780805587768555, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7126739025115967, + "num_tokens": 438764999.0, + "step": 17074 + }, + { + "epoch": 1.875137272128267, + "grad_norm": 2.3604719638824463, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7266740798950195, + "num_tokens": 438788054.0, + "step": 17075 + }, + { + "epoch": 1.8752470898308808, + "grad_norm": 2.283864974975586, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7147158980369568, + "num_tokens": 438813105.0, + "step": 17076 + }, + { + "epoch": 1.8753569075334944, + "grad_norm": 2.2736735343933105, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7198766469955444, + "num_tokens": 438838079.0, + "step": 17077 + }, + { + "epoch": 1.875466725236108, + "grad_norm": 2.292975664138794, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7145149111747742, + "num_tokens": 438862812.0, + "step": 17078 + }, + { + "epoch": 1.8755765429387217, + "grad_norm": 2.072957754135132, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7295023202896118, + "num_tokens": 438894257.0, + "step": 17079 + }, + { + "epoch": 1.8756863606413354, + "grad_norm": 2.3880927562713623, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7410921454429626, + "num_tokens": 438918891.0, + "step": 17080 + }, + { + "epoch": 1.8757961783439492, + "grad_norm": 2.4034717082977295, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7213519811630249, + "num_tokens": 438944964.0, + "step": 17081 + }, + { + "epoch": 1.8759059960465627, + "grad_norm": 2.2149498462677, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.727762758731842, + "num_tokens": 438972576.0, + "step": 17082 + }, + { + "epoch": 1.8760158137491763, + "grad_norm": 2.589803457260132, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7241756319999695, + "num_tokens": 438993288.0, + "step": 17083 + }, + { + "epoch": 1.87612563145179, + "grad_norm": 2.167178153991699, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7250804901123047, + "num_tokens": 439018516.0, + "step": 17084 + }, + { + "epoch": 1.8762354491544038, + "grad_norm": 2.1953277587890625, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.729865550994873, + "num_tokens": 439045834.0, + "step": 17085 + }, + { + "epoch": 1.8763452668570173, + "grad_norm": 2.0470542907714844, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7123783826828003, + "num_tokens": 439077719.0, + "step": 17086 + }, + { + "epoch": 1.876455084559631, + "grad_norm": 2.3610291481018066, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7304253578186035, + "num_tokens": 439101804.0, + "step": 17087 + }, + { + "epoch": 1.8765649022622446, + "grad_norm": 2.453141927719116, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.725472092628479, + "num_tokens": 439122506.0, + "step": 17088 + }, + { + "epoch": 1.8766747199648584, + "grad_norm": 2.1441102027893066, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.715116024017334, + "num_tokens": 439149896.0, + "step": 17089 + }, + { + "epoch": 1.876784537667472, + "grad_norm": 2.1996681690216064, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7162661552429199, + "num_tokens": 439174543.0, + "step": 17090 + }, + { + "epoch": 1.8768943553700856, + "grad_norm": 2.257814645767212, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7422950267791748, + "num_tokens": 439197667.0, + "step": 17091 + }, + { + "epoch": 1.8770041730726992, + "grad_norm": 2.1376216411590576, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7179531455039978, + "num_tokens": 439224821.0, + "step": 17092 + }, + { + "epoch": 1.877113990775313, + "grad_norm": 2.124056816101074, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6909171342849731, + "num_tokens": 439251900.0, + "step": 17093 + }, + { + "epoch": 1.8772238084779267, + "grad_norm": 2.653649091720581, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7200325727462769, + "num_tokens": 439273186.0, + "step": 17094 + }, + { + "epoch": 1.8773336261805404, + "grad_norm": 2.140198230743408, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7430918216705322, + "num_tokens": 439299649.0, + "step": 17095 + }, + { + "epoch": 1.877443443883154, + "grad_norm": 2.192577838897705, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7211939692497253, + "num_tokens": 439325218.0, + "step": 17096 + }, + { + "epoch": 1.8775532615857675, + "grad_norm": 2.0854549407958984, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7295453548431396, + "num_tokens": 439351904.0, + "step": 17097 + }, + { + "epoch": 1.8776630792883813, + "grad_norm": 2.3749265670776367, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7374860048294067, + "num_tokens": 439373727.0, + "step": 17098 + }, + { + "epoch": 1.877772896990995, + "grad_norm": 2.2100770473480225, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7300788164138794, + "num_tokens": 439398206.0, + "step": 17099 + }, + { + "epoch": 1.8778827146936086, + "grad_norm": 1.9030157327651978, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7074512839317322, + "num_tokens": 439432729.0, + "step": 17100 + }, + { + "epoch": 1.8779925323962223, + "grad_norm": 2.360318422317505, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7093885540962219, + "num_tokens": 439457562.0, + "step": 17101 + }, + { + "epoch": 1.8781023500988359, + "grad_norm": 2.2680888175964355, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7151195406913757, + "num_tokens": 439482720.0, + "step": 17102 + }, + { + "epoch": 1.8782121678014496, + "grad_norm": 2.1383872032165527, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6949318647384644, + "num_tokens": 439510575.0, + "step": 17103 + }, + { + "epoch": 1.8783219855040634, + "grad_norm": 2.5821969509124756, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7467793226242065, + "num_tokens": 439530142.0, + "step": 17104 + }, + { + "epoch": 1.878431803206677, + "grad_norm": 2.13590669631958, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7053695917129517, + "num_tokens": 439558570.0, + "step": 17105 + }, + { + "epoch": 1.8785416209092904, + "grad_norm": 1.9208879470825195, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7218770980834961, + "num_tokens": 439592871.0, + "step": 17106 + }, + { + "epoch": 1.8786514386119042, + "grad_norm": 2.430961847305298, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7236431837081909, + "num_tokens": 439617384.0, + "step": 17107 + }, + { + "epoch": 1.878761256314518, + "grad_norm": 2.540708303451538, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7270151376724243, + "num_tokens": 439637007.0, + "step": 17108 + }, + { + "epoch": 1.8788710740171317, + "grad_norm": 2.0494279861450195, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7217165231704712, + "num_tokens": 439664686.0, + "step": 17109 + }, + { + "epoch": 1.8789808917197452, + "grad_norm": 1.8911983966827393, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7129896879196167, + "num_tokens": 439698332.0, + "step": 17110 + }, + { + "epoch": 1.8790907094223588, + "grad_norm": 2.0210018157958984, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.705839991569519, + "num_tokens": 439727298.0, + "step": 17111 + }, + { + "epoch": 1.8792005271249725, + "grad_norm": 2.3475940227508545, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.729560375213623, + "num_tokens": 439751563.0, + "step": 17112 + }, + { + "epoch": 1.8793103448275863, + "grad_norm": 2.169297933578491, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7350534796714783, + "num_tokens": 439777071.0, + "step": 17113 + }, + { + "epoch": 1.8794201625301998, + "grad_norm": 2.297705888748169, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7013206481933594, + "num_tokens": 439803503.0, + "step": 17114 + }, + { + "epoch": 1.8795299802328134, + "grad_norm": 2.5180447101593018, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7266802787780762, + "num_tokens": 439823877.0, + "step": 17115 + }, + { + "epoch": 1.8796397979354271, + "grad_norm": 1.9865971803665161, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7168734073638916, + "num_tokens": 439854986.0, + "step": 17116 + }, + { + "epoch": 1.8797496156380409, + "grad_norm": 2.431057929992676, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7092148065567017, + "num_tokens": 439877231.0, + "step": 17117 + }, + { + "epoch": 1.8798594333406546, + "grad_norm": 2.386488437652588, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7495784163475037, + "num_tokens": 439898541.0, + "step": 17118 + }, + { + "epoch": 1.8799692510432682, + "grad_norm": 2.238795757293701, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7044175863265991, + "num_tokens": 439925280.0, + "step": 17119 + }, + { + "epoch": 1.8800790687458817, + "grad_norm": 2.41377329826355, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7227568626403809, + "num_tokens": 439947912.0, + "step": 17120 + }, + { + "epoch": 1.8801888864484955, + "grad_norm": 2.5773191452026367, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7438604235649109, + "num_tokens": 439966436.0, + "step": 17121 + }, + { + "epoch": 1.8802987041511092, + "grad_norm": 1.9807566404342651, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7271435260772705, + "num_tokens": 439995873.0, + "step": 17122 + }, + { + "epoch": 1.880408521853723, + "grad_norm": 2.1714959144592285, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7112420797348022, + "num_tokens": 440022606.0, + "step": 17123 + }, + { + "epoch": 1.8805183395563365, + "grad_norm": 2.2044854164123535, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7140533328056335, + "num_tokens": 440048301.0, + "step": 17124 + }, + { + "epoch": 1.88062815725895, + "grad_norm": 2.0610382556915283, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.708138108253479, + "num_tokens": 440077595.0, + "step": 17125 + }, + { + "epoch": 1.8807379749615638, + "grad_norm": 1.976246953010559, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7155115008354187, + "num_tokens": 440108631.0, + "step": 17126 + }, + { + "epoch": 1.8808477926641776, + "grad_norm": 2.3805644512176514, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7242037057876587, + "num_tokens": 440131708.0, + "step": 17127 + }, + { + "epoch": 1.880957610366791, + "grad_norm": 2.0520405769348145, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7178789377212524, + "num_tokens": 440163667.0, + "step": 17128 + }, + { + "epoch": 1.8810674280694046, + "grad_norm": 2.2587602138519287, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7344177961349487, + "num_tokens": 440189310.0, + "step": 17129 + }, + { + "epoch": 1.8811772457720184, + "grad_norm": 2.3454840183258057, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7405500411987305, + "num_tokens": 440210292.0, + "step": 17130 + }, + { + "epoch": 1.8812870634746321, + "grad_norm": 2.284703493118286, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7201651930809021, + "num_tokens": 440236586.0, + "step": 17131 + }, + { + "epoch": 1.881396881177246, + "grad_norm": 2.277597188949585, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7151871919631958, + "num_tokens": 440261491.0, + "step": 17132 + }, + { + "epoch": 1.8815066988798594, + "grad_norm": 2.1256330013275146, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7454839944839478, + "num_tokens": 440288001.0, + "step": 17133 + }, + { + "epoch": 1.881616516582473, + "grad_norm": 2.097552537918091, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.712814211845398, + "num_tokens": 440315833.0, + "step": 17134 + }, + { + "epoch": 1.8817263342850867, + "grad_norm": 2.294196605682373, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7358419299125671, + "num_tokens": 440339860.0, + "step": 17135 + }, + { + "epoch": 1.8818361519877005, + "grad_norm": 2.371553421020508, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7385985851287842, + "num_tokens": 440362034.0, + "step": 17136 + }, + { + "epoch": 1.881945969690314, + "grad_norm": 2.3919544219970703, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7253013849258423, + "num_tokens": 440386143.0, + "step": 17137 + }, + { + "epoch": 1.8820557873929278, + "grad_norm": 2.550426721572876, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7481061220169067, + "num_tokens": 440406176.0, + "step": 17138 + }, + { + "epoch": 1.8821656050955413, + "grad_norm": 2.328935384750366, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7369290590286255, + "num_tokens": 440430102.0, + "step": 17139 + }, + { + "epoch": 1.882275422798155, + "grad_norm": 2.366910696029663, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7011167407035828, + "num_tokens": 440455931.0, + "step": 17140 + }, + { + "epoch": 1.8823852405007688, + "grad_norm": 2.3492980003356934, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7272549867630005, + "num_tokens": 440480664.0, + "step": 17141 + }, + { + "epoch": 1.8824950582033824, + "grad_norm": 1.8823988437652588, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7404898405075073, + "num_tokens": 440512087.0, + "step": 17142 + }, + { + "epoch": 1.882604875905996, + "grad_norm": 2.2095093727111816, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7277250289916992, + "num_tokens": 440538104.0, + "step": 17143 + }, + { + "epoch": 1.8827146936086097, + "grad_norm": 2.3426103591918945, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7275869846343994, + "num_tokens": 440563431.0, + "step": 17144 + }, + { + "epoch": 1.8828245113112234, + "grad_norm": 2.2138168811798096, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7128397822380066, + "num_tokens": 440588686.0, + "step": 17145 + }, + { + "epoch": 1.8829343290138372, + "grad_norm": 2.0903286933898926, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7301578521728516, + "num_tokens": 440616992.0, + "step": 17146 + }, + { + "epoch": 1.8830441467164507, + "grad_norm": 2.2992055416107178, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6997504234313965, + "num_tokens": 440642900.0, + "step": 17147 + }, + { + "epoch": 1.8831539644190642, + "grad_norm": 2.0360519886016846, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7050744891166687, + "num_tokens": 440673424.0, + "step": 17148 + }, + { + "epoch": 1.883263782121678, + "grad_norm": 2.21711802482605, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7383899688720703, + "num_tokens": 440696368.0, + "step": 17149 + }, + { + "epoch": 1.8833735998242918, + "grad_norm": 2.1365203857421875, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7159157395362854, + "num_tokens": 440724049.0, + "step": 17150 + }, + { + "epoch": 1.8834834175269053, + "grad_norm": 2.3006858825683594, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7289135456085205, + "num_tokens": 440748072.0, + "step": 17151 + }, + { + "epoch": 1.883593235229519, + "grad_norm": 2.2263331413269043, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7418813109397888, + "num_tokens": 440772138.0, + "step": 17152 + }, + { + "epoch": 1.8837030529321326, + "grad_norm": 2.533928871154785, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7240316867828369, + "num_tokens": 440792760.0, + "step": 17153 + }, + { + "epoch": 1.8838128706347463, + "grad_norm": 2.286254405975342, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7275984287261963, + "num_tokens": 440816152.0, + "step": 17154 + }, + { + "epoch": 1.88392268833736, + "grad_norm": 2.242973804473877, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7242527008056641, + "num_tokens": 440840650.0, + "step": 17155 + }, + { + "epoch": 1.8840325060399736, + "grad_norm": 2.267162561416626, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7367818355560303, + "num_tokens": 440863128.0, + "step": 17156 + }, + { + "epoch": 1.8841423237425872, + "grad_norm": 2.3013837337493896, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7077456712722778, + "num_tokens": 440887862.0, + "step": 17157 + }, + { + "epoch": 1.884252141445201, + "grad_norm": 2.204645872116089, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7261217832565308, + "num_tokens": 440913188.0, + "step": 17158 + }, + { + "epoch": 1.8843619591478147, + "grad_norm": 2.520902156829834, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7303595542907715, + "num_tokens": 440933835.0, + "step": 17159 + }, + { + "epoch": 1.8844717768504284, + "grad_norm": 2.162673234939575, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7334491610527039, + "num_tokens": 440960328.0, + "step": 17160 + }, + { + "epoch": 1.884581594553042, + "grad_norm": 2.4523673057556152, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7195042371749878, + "num_tokens": 440981377.0, + "step": 17161 + }, + { + "epoch": 1.8846914122556555, + "grad_norm": 2.717977523803711, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7328248023986816, + "num_tokens": 441003700.0, + "step": 17162 + }, + { + "epoch": 1.8848012299582693, + "grad_norm": 2.1763784885406494, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7366994619369507, + "num_tokens": 441029983.0, + "step": 17163 + }, + { + "epoch": 1.884911047660883, + "grad_norm": 2.127218246459961, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7242060899734497, + "num_tokens": 441057922.0, + "step": 17164 + }, + { + "epoch": 1.8850208653634966, + "grad_norm": 2.0241355895996094, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7085764408111572, + "num_tokens": 441089733.0, + "step": 17165 + }, + { + "epoch": 1.88513068306611, + "grad_norm": 2.1833221912384033, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7165732383728027, + "num_tokens": 441116434.0, + "step": 17166 + }, + { + "epoch": 1.8852405007687238, + "grad_norm": 2.31980299949646, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.724388599395752, + "num_tokens": 441140829.0, + "step": 17167 + }, + { + "epoch": 1.8853503184713376, + "grad_norm": 2.244687795639038, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7374094724655151, + "num_tokens": 441165242.0, + "step": 17168 + }, + { + "epoch": 1.8854601361739514, + "grad_norm": 2.2205116748809814, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7283540368080139, + "num_tokens": 441191159.0, + "step": 17169 + }, + { + "epoch": 1.885569953876565, + "grad_norm": 2.0330910682678223, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7121662497520447, + "num_tokens": 441219602.0, + "step": 17170 + }, + { + "epoch": 1.8856797715791784, + "grad_norm": 2.0524682998657227, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7098837494850159, + "num_tokens": 441249946.0, + "step": 17171 + }, + { + "epoch": 1.8857895892817922, + "grad_norm": 2.345780372619629, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7341561317443848, + "num_tokens": 441272319.0, + "step": 17172 + }, + { + "epoch": 1.885899406984406, + "grad_norm": 2.1281402111053467, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7027721405029297, + "num_tokens": 441298309.0, + "step": 17173 + }, + { + "epoch": 1.8860092246870197, + "grad_norm": 2.6614887714385986, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7241002321243286, + "num_tokens": 441319071.0, + "step": 17174 + }, + { + "epoch": 1.8861190423896332, + "grad_norm": 2.047879695892334, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6899941563606262, + "num_tokens": 441349575.0, + "step": 17175 + }, + { + "epoch": 1.8862288600922468, + "grad_norm": 2.2106282711029053, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7053156495094299, + "num_tokens": 441375860.0, + "step": 17176 + }, + { + "epoch": 1.8863386777948605, + "grad_norm": 2.1266775131225586, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.713486909866333, + "num_tokens": 441402519.0, + "step": 17177 + }, + { + "epoch": 1.8864484954974743, + "grad_norm": 2.084137201309204, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.71607506275177, + "num_tokens": 441431239.0, + "step": 17178 + }, + { + "epoch": 1.8865583132000878, + "grad_norm": 2.144507646560669, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7238215208053589, + "num_tokens": 441457664.0, + "step": 17179 + }, + { + "epoch": 1.8866681309027014, + "grad_norm": 1.9025293588638306, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.72584068775177, + "num_tokens": 441489520.0, + "step": 17180 + }, + { + "epoch": 1.8867779486053151, + "grad_norm": 2.123518705368042, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6908848881721497, + "num_tokens": 441521952.0, + "step": 17181 + }, + { + "epoch": 1.8868877663079289, + "grad_norm": 1.9883308410644531, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.701120913028717, + "num_tokens": 441554864.0, + "step": 17182 + }, + { + "epoch": 1.8869975840105426, + "grad_norm": 2.552521228790283, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7333171367645264, + "num_tokens": 441575225.0, + "step": 17183 + }, + { + "epoch": 1.8871074017131562, + "grad_norm": 2.1643974781036377, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.741205096244812, + "num_tokens": 441601819.0, + "step": 17184 + }, + { + "epoch": 1.8872172194157697, + "grad_norm": 2.2674202919006348, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6954840421676636, + "num_tokens": 441628742.0, + "step": 17185 + }, + { + "epoch": 1.8873270371183835, + "grad_norm": 2.526245594024658, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7311331629753113, + "num_tokens": 441649644.0, + "step": 17186 + }, + { + "epoch": 1.8874368548209972, + "grad_norm": 2.2341816425323486, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7187213897705078, + "num_tokens": 441675281.0, + "step": 17187 + }, + { + "epoch": 1.887546672523611, + "grad_norm": 2.119220018386841, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.729687511920929, + "num_tokens": 441702288.0, + "step": 17188 + }, + { + "epoch": 1.8876564902262245, + "grad_norm": 2.46751070022583, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7371476292610168, + "num_tokens": 441723857.0, + "step": 17189 + }, + { + "epoch": 1.887766307928838, + "grad_norm": 2.307013988494873, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6962541341781616, + "num_tokens": 441749378.0, + "step": 17190 + }, + { + "epoch": 1.8878761256314518, + "grad_norm": 2.174684762954712, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7302262187004089, + "num_tokens": 441776052.0, + "step": 17191 + }, + { + "epoch": 1.8879859433340656, + "grad_norm": 1.967908263206482, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7169022560119629, + "num_tokens": 441808516.0, + "step": 17192 + }, + { + "epoch": 1.888095761036679, + "grad_norm": 2.2839064598083496, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7326350808143616, + "num_tokens": 441833422.0, + "step": 17193 + }, + { + "epoch": 1.8882055787392926, + "grad_norm": 2.1305882930755615, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.720069169998169, + "num_tokens": 441861096.0, + "step": 17194 + }, + { + "epoch": 1.8883153964419064, + "grad_norm": 2.2609660625457764, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7111802101135254, + "num_tokens": 441885743.0, + "step": 17195 + }, + { + "epoch": 1.8884252141445201, + "grad_norm": 2.1301825046539307, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7280195355415344, + "num_tokens": 441912093.0, + "step": 17196 + }, + { + "epoch": 1.888535031847134, + "grad_norm": 2.4553139209747314, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.734100341796875, + "num_tokens": 441932176.0, + "step": 17197 + }, + { + "epoch": 1.8886448495497474, + "grad_norm": 2.32438063621521, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7109524011611938, + "num_tokens": 441957126.0, + "step": 17198 + }, + { + "epoch": 1.888754667252361, + "grad_norm": 2.2986903190612793, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6903114318847656, + "num_tokens": 441986578.0, + "step": 17199 + }, + { + "epoch": 1.8888644849549747, + "grad_norm": 2.108037233352661, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7189347743988037, + "num_tokens": 442014899.0, + "step": 17200 + }, + { + "epoch": 1.8889743026575885, + "grad_norm": 2.4026801586151123, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7108845710754395, + "num_tokens": 442038611.0, + "step": 17201 + }, + { + "epoch": 1.889084120360202, + "grad_norm": 2.42073917388916, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7414061427116394, + "num_tokens": 442059556.0, + "step": 17202 + }, + { + "epoch": 1.8891939380628158, + "grad_norm": 2.145110607147217, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6922369003295898, + "num_tokens": 442088835.0, + "step": 17203 + }, + { + "epoch": 1.8893037557654293, + "grad_norm": 2.337923526763916, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7400968074798584, + "num_tokens": 442111865.0, + "step": 17204 + }, + { + "epoch": 1.889413573468043, + "grad_norm": 2.07596755027771, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7156603336334229, + "num_tokens": 442142913.0, + "step": 17205 + }, + { + "epoch": 1.8895233911706568, + "grad_norm": 2.446467399597168, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.708059549331665, + "num_tokens": 442165603.0, + "step": 17206 + }, + { + "epoch": 1.8896332088732704, + "grad_norm": 2.2020604610443115, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7417510747909546, + "num_tokens": 442191090.0, + "step": 17207 + }, + { + "epoch": 1.8897430265758839, + "grad_norm": 2.1991753578186035, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.728361964225769, + "num_tokens": 442216707.0, + "step": 17208 + }, + { + "epoch": 1.8898528442784976, + "grad_norm": 2.282721996307373, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7418052554130554, + "num_tokens": 442241802.0, + "step": 17209 + }, + { + "epoch": 1.8899626619811114, + "grad_norm": 1.9517542123794556, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.707850456237793, + "num_tokens": 442273832.0, + "step": 17210 + }, + { + "epoch": 1.8900724796837252, + "grad_norm": 2.3996455669403076, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.72632896900177, + "num_tokens": 442298732.0, + "step": 17211 + }, + { + "epoch": 1.8901822973863387, + "grad_norm": 2.141286611557007, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.734356164932251, + "num_tokens": 442324475.0, + "step": 17212 + }, + { + "epoch": 1.8902921150889522, + "grad_norm": 2.426004648208618, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.73850017786026, + "num_tokens": 442345473.0, + "step": 17213 + }, + { + "epoch": 1.890401932791566, + "grad_norm": 2.533384323120117, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7345409393310547, + "num_tokens": 442367222.0, + "step": 17214 + }, + { + "epoch": 1.8905117504941797, + "grad_norm": 2.3356926441192627, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7184792757034302, + "num_tokens": 442390042.0, + "step": 17215 + }, + { + "epoch": 1.8906215681967933, + "grad_norm": 2.3717784881591797, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7384194135665894, + "num_tokens": 442412443.0, + "step": 17216 + }, + { + "epoch": 1.890731385899407, + "grad_norm": 2.3320319652557373, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7227250337600708, + "num_tokens": 442437816.0, + "step": 17217 + }, + { + "epoch": 1.8908412036020206, + "grad_norm": 2.368628978729248, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7299314141273499, + "num_tokens": 442459283.0, + "step": 17218 + }, + { + "epoch": 1.8909510213046343, + "grad_norm": 2.2534191608428955, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7226341962814331, + "num_tokens": 442482900.0, + "step": 17219 + }, + { + "epoch": 1.891060839007248, + "grad_norm": 2.2175393104553223, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7239517569541931, + "num_tokens": 442507801.0, + "step": 17220 + }, + { + "epoch": 1.8911706567098616, + "grad_norm": 2.1260321140289307, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7214590311050415, + "num_tokens": 442535134.0, + "step": 17221 + }, + { + "epoch": 1.8912804744124752, + "grad_norm": 2.298520088195801, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.727888822555542, + "num_tokens": 442559322.0, + "step": 17222 + }, + { + "epoch": 1.891390292115089, + "grad_norm": 2.2843310832977295, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.723477840423584, + "num_tokens": 442584326.0, + "step": 17223 + }, + { + "epoch": 1.8915001098177027, + "grad_norm": 2.0462887287139893, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7007470726966858, + "num_tokens": 442613059.0, + "step": 17224 + }, + { + "epoch": 1.8916099275203164, + "grad_norm": 2.2630577087402344, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7095367312431335, + "num_tokens": 442641484.0, + "step": 17225 + }, + { + "epoch": 1.89171974522293, + "grad_norm": 2.6688520908355713, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7159783244132996, + "num_tokens": 442664012.0, + "step": 17226 + }, + { + "epoch": 1.8918295629255435, + "grad_norm": 2.074740171432495, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7081092596054077, + "num_tokens": 442693085.0, + "step": 17227 + }, + { + "epoch": 1.8919393806281573, + "grad_norm": 2.2959771156311035, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7144067287445068, + "num_tokens": 442716246.0, + "step": 17228 + }, + { + "epoch": 1.892049198330771, + "grad_norm": 2.1869466304779053, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7124668955802917, + "num_tokens": 442741594.0, + "step": 17229 + }, + { + "epoch": 1.8921590160333845, + "grad_norm": 2.3434176445007324, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7392978072166443, + "num_tokens": 442764516.0, + "step": 17230 + }, + { + "epoch": 1.892268833735998, + "grad_norm": 2.540498971939087, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7267934083938599, + "num_tokens": 442785587.0, + "step": 17231 + }, + { + "epoch": 1.8923786514386118, + "grad_norm": 2.0580198764801025, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7237323522567749, + "num_tokens": 442815003.0, + "step": 17232 + }, + { + "epoch": 1.8924884691412256, + "grad_norm": 2.235522508621216, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7404120564460754, + "num_tokens": 442838825.0, + "step": 17233 + }, + { + "epoch": 1.8925982868438394, + "grad_norm": 2.0628058910369873, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7475948333740234, + "num_tokens": 442865392.0, + "step": 17234 + }, + { + "epoch": 1.8927081045464529, + "grad_norm": 2.183246612548828, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7227177619934082, + "num_tokens": 442894864.0, + "step": 17235 + }, + { + "epoch": 1.8928179222490664, + "grad_norm": 2.218571901321411, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7401386499404907, + "num_tokens": 442919255.0, + "step": 17236 + }, + { + "epoch": 1.8929277399516802, + "grad_norm": 2.1527516841888428, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7386632561683655, + "num_tokens": 442947559.0, + "step": 17237 + }, + { + "epoch": 1.893037557654294, + "grad_norm": 2.237093210220337, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7127673625946045, + "num_tokens": 442973297.0, + "step": 17238 + }, + { + "epoch": 1.8931473753569077, + "grad_norm": 2.303521156311035, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7256292104721069, + "num_tokens": 442997847.0, + "step": 17239 + }, + { + "epoch": 1.8932571930595212, + "grad_norm": 2.016627788543701, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7014477252960205, + "num_tokens": 443029662.0, + "step": 17240 + }, + { + "epoch": 1.8933670107621348, + "grad_norm": 2.2490646839141846, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7421154379844666, + "num_tokens": 443052499.0, + "step": 17241 + }, + { + "epoch": 1.8934768284647485, + "grad_norm": 2.2131457328796387, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7069264054298401, + "num_tokens": 443081090.0, + "step": 17242 + }, + { + "epoch": 1.8935866461673623, + "grad_norm": 2.386103630065918, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7335749268531799, + "num_tokens": 443104493.0, + "step": 17243 + }, + { + "epoch": 1.8936964638699758, + "grad_norm": 2.2131459712982178, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7270574569702148, + "num_tokens": 443128793.0, + "step": 17244 + }, + { + "epoch": 1.8938062815725893, + "grad_norm": 2.3319473266601562, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7148870229721069, + "num_tokens": 443153118.0, + "step": 17245 + }, + { + "epoch": 1.893916099275203, + "grad_norm": 2.2420260906219482, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7257421612739563, + "num_tokens": 443179411.0, + "step": 17246 + }, + { + "epoch": 1.8940259169778169, + "grad_norm": 2.026332378387451, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.701430082321167, + "num_tokens": 443211528.0, + "step": 17247 + }, + { + "epoch": 1.8941357346804306, + "grad_norm": 2.2262003421783447, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7117682099342346, + "num_tokens": 443239218.0, + "step": 17248 + }, + { + "epoch": 1.8942455523830442, + "grad_norm": 2.3131773471832275, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7374110221862793, + "num_tokens": 443263325.0, + "step": 17249 + }, + { + "epoch": 1.8943553700856577, + "grad_norm": 2.114560604095459, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7200911045074463, + "num_tokens": 443288953.0, + "step": 17250 + }, + { + "epoch": 1.8944651877882714, + "grad_norm": 2.0537655353546143, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7205768823623657, + "num_tokens": 443318812.0, + "step": 17251 + }, + { + "epoch": 1.8945750054908852, + "grad_norm": 2.1021475791931152, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7020227909088135, + "num_tokens": 443348166.0, + "step": 17252 + }, + { + "epoch": 1.894684823193499, + "grad_norm": 2.064159870147705, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7208224534988403, + "num_tokens": 443378015.0, + "step": 17253 + }, + { + "epoch": 1.8947946408961125, + "grad_norm": 2.3613643646240234, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7647217512130737, + "num_tokens": 443399738.0, + "step": 17254 + }, + { + "epoch": 1.894904458598726, + "grad_norm": 2.332331418991089, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7202755212783813, + "num_tokens": 443423352.0, + "step": 17255 + }, + { + "epoch": 1.8950142763013398, + "grad_norm": 2.063361883163452, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.72503662109375, + "num_tokens": 443452929.0, + "step": 17256 + }, + { + "epoch": 1.8951240940039535, + "grad_norm": 2.3167386054992676, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7184538841247559, + "num_tokens": 443476258.0, + "step": 17257 + }, + { + "epoch": 1.895233911706567, + "grad_norm": 2.1596312522888184, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7226983308792114, + "num_tokens": 443502238.0, + "step": 17258 + }, + { + "epoch": 1.8953437294091806, + "grad_norm": 2.3241941928863525, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7273274660110474, + "num_tokens": 443526271.0, + "step": 17259 + }, + { + "epoch": 1.8954535471117944, + "grad_norm": 2.2636213302612305, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7309474349021912, + "num_tokens": 443551496.0, + "step": 17260 + }, + { + "epoch": 1.8955633648144081, + "grad_norm": 2.1213173866271973, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7113000154495239, + "num_tokens": 443580053.0, + "step": 17261 + }, + { + "epoch": 1.8956731825170219, + "grad_norm": 2.3160481452941895, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.747433066368103, + "num_tokens": 443601541.0, + "step": 17262 + }, + { + "epoch": 1.8957830002196354, + "grad_norm": 2.2503793239593506, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6974420547485352, + "num_tokens": 443626232.0, + "step": 17263 + }, + { + "epoch": 1.895892817922249, + "grad_norm": 2.377558469772339, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7134218215942383, + "num_tokens": 443650542.0, + "step": 17264 + }, + { + "epoch": 1.8960026356248627, + "grad_norm": 2.1063287258148193, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7207820415496826, + "num_tokens": 443678269.0, + "step": 17265 + }, + { + "epoch": 1.8961124533274765, + "grad_norm": 2.0396506786346436, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7106917500495911, + "num_tokens": 443707177.0, + "step": 17266 + }, + { + "epoch": 1.89622227103009, + "grad_norm": 2.3560924530029297, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7189223766326904, + "num_tokens": 443729722.0, + "step": 17267 + }, + { + "epoch": 1.8963320887327038, + "grad_norm": 2.204543352127075, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7219224572181702, + "num_tokens": 443755265.0, + "step": 17268 + }, + { + "epoch": 1.8964419064353173, + "grad_norm": 2.4165284633636475, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7066965103149414, + "num_tokens": 443779904.0, + "step": 17269 + }, + { + "epoch": 1.896551724137931, + "grad_norm": 2.052945852279663, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7391921877861023, + "num_tokens": 443808592.0, + "step": 17270 + }, + { + "epoch": 1.8966615418405448, + "grad_norm": 2.3662054538726807, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.701166570186615, + "num_tokens": 443832860.0, + "step": 17271 + }, + { + "epoch": 1.8967713595431583, + "grad_norm": 2.289001941680908, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7325310111045837, + "num_tokens": 443857474.0, + "step": 17272 + }, + { + "epoch": 1.8968811772457719, + "grad_norm": 2.159485101699829, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7269266843795776, + "num_tokens": 443883989.0, + "step": 17273 + }, + { + "epoch": 1.8969909949483856, + "grad_norm": 2.269852876663208, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7296249866485596, + "num_tokens": 443908035.0, + "step": 17274 + }, + { + "epoch": 1.8971008126509994, + "grad_norm": 2.5718917846679688, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7199241518974304, + "num_tokens": 443930042.0, + "step": 17275 + }, + { + "epoch": 1.8972106303536131, + "grad_norm": 2.3148794174194336, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7218109369277954, + "num_tokens": 443954541.0, + "step": 17276 + }, + { + "epoch": 1.8973204480562267, + "grad_norm": 2.6300861835479736, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7145426273345947, + "num_tokens": 443975798.0, + "step": 17277 + }, + { + "epoch": 1.8974302657588402, + "grad_norm": 1.9895033836364746, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6998553276062012, + "num_tokens": 444006249.0, + "step": 17278 + }, + { + "epoch": 1.897540083461454, + "grad_norm": 2.208055019378662, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7345336675643921, + "num_tokens": 444031907.0, + "step": 17279 + }, + { + "epoch": 1.8976499011640677, + "grad_norm": 2.1246397495269775, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.74525386095047, + "num_tokens": 444058331.0, + "step": 17280 + }, + { + "epoch": 1.8977597188666813, + "grad_norm": 2.558769702911377, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7343899011611938, + "num_tokens": 444077835.0, + "step": 17281 + }, + { + "epoch": 1.897869536569295, + "grad_norm": 2.1600394248962402, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7178802490234375, + "num_tokens": 444105135.0, + "step": 17282 + }, + { + "epoch": 1.8979793542719086, + "grad_norm": 2.2958483695983887, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7167344689369202, + "num_tokens": 444129721.0, + "step": 17283 + }, + { + "epoch": 1.8980891719745223, + "grad_norm": 2.3086278438568115, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.709882915019989, + "num_tokens": 444157982.0, + "step": 17284 + }, + { + "epoch": 1.898198989677136, + "grad_norm": 2.2712271213531494, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7167714834213257, + "num_tokens": 444183019.0, + "step": 17285 + }, + { + "epoch": 1.8983088073797496, + "grad_norm": 2.124584674835205, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.727376401424408, + "num_tokens": 444210893.0, + "step": 17286 + }, + { + "epoch": 1.8984186250823631, + "grad_norm": 2.0728302001953125, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7331881523132324, + "num_tokens": 444239395.0, + "step": 17287 + }, + { + "epoch": 1.898528442784977, + "grad_norm": 2.2646703720092773, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7275280952453613, + "num_tokens": 444265903.0, + "step": 17288 + }, + { + "epoch": 1.8986382604875907, + "grad_norm": 2.0653302669525146, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7211899161338806, + "num_tokens": 444297165.0, + "step": 17289 + }, + { + "epoch": 1.8987480781902044, + "grad_norm": 2.1400184631347656, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7067181468009949, + "num_tokens": 444327504.0, + "step": 17290 + }, + { + "epoch": 1.898857895892818, + "grad_norm": 2.4395599365234375, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7424381971359253, + "num_tokens": 444349880.0, + "step": 17291 + }, + { + "epoch": 1.8989677135954315, + "grad_norm": 2.5039076805114746, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.736869752407074, + "num_tokens": 444370411.0, + "step": 17292 + }, + { + "epoch": 1.8990775312980452, + "grad_norm": 2.1682519912719727, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7112917900085449, + "num_tokens": 444398089.0, + "step": 17293 + }, + { + "epoch": 1.899187349000659, + "grad_norm": 2.7867519855499268, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7346795201301575, + "num_tokens": 444416400.0, + "step": 17294 + }, + { + "epoch": 1.8992971667032725, + "grad_norm": 2.1230552196502686, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7158449292182922, + "num_tokens": 444447573.0, + "step": 17295 + }, + { + "epoch": 1.899406984405886, + "grad_norm": 2.4783377647399902, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7231080532073975, + "num_tokens": 444469572.0, + "step": 17296 + }, + { + "epoch": 1.8995168021084998, + "grad_norm": 2.631747245788574, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7427985072135925, + "num_tokens": 444488687.0, + "step": 17297 + }, + { + "epoch": 1.8996266198111136, + "grad_norm": 2.5912623405456543, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7372776865959167, + "num_tokens": 444508243.0, + "step": 17298 + }, + { + "epoch": 1.8997364375137273, + "grad_norm": 2.7263002395629883, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7326339483261108, + "num_tokens": 444527627.0, + "step": 17299 + }, + { + "epoch": 1.8998462552163409, + "grad_norm": 2.108323574066162, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7011916637420654, + "num_tokens": 444557851.0, + "step": 17300 + }, + { + "epoch": 1.8999560729189544, + "grad_norm": 2.1198880672454834, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7161626815795898, + "num_tokens": 444585439.0, + "step": 17301 + }, + { + "epoch": 1.9000658906215682, + "grad_norm": 2.3316726684570312, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7102392911911011, + "num_tokens": 444609655.0, + "step": 17302 + }, + { + "epoch": 1.900175708324182, + "grad_norm": 2.0824313163757324, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7228244543075562, + "num_tokens": 444638784.0, + "step": 17303 + }, + { + "epoch": 1.9002855260267957, + "grad_norm": 2.0746021270751953, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7124489545822144, + "num_tokens": 444668031.0, + "step": 17304 + }, + { + "epoch": 1.9003953437294092, + "grad_norm": 2.1989567279815674, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7245290875434875, + "num_tokens": 444693604.0, + "step": 17305 + }, + { + "epoch": 1.9005051614320227, + "grad_norm": 2.3767096996307373, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7108893394470215, + "num_tokens": 444717485.0, + "step": 17306 + }, + { + "epoch": 1.9006149791346365, + "grad_norm": 2.1986806392669678, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7489906549453735, + "num_tokens": 444740775.0, + "step": 17307 + }, + { + "epoch": 1.9007247968372503, + "grad_norm": 2.104980707168579, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.739276111125946, + "num_tokens": 444767606.0, + "step": 17308 + }, + { + "epoch": 1.9008346145398638, + "grad_norm": 2.2188782691955566, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7074687480926514, + "num_tokens": 444794845.0, + "step": 17309 + }, + { + "epoch": 1.9009444322424773, + "grad_norm": 2.052913188934326, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7275406122207642, + "num_tokens": 444824148.0, + "step": 17310 + }, + { + "epoch": 1.901054249945091, + "grad_norm": 2.108839750289917, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7084642052650452, + "num_tokens": 444852752.0, + "step": 17311 + }, + { + "epoch": 1.9011640676477048, + "grad_norm": 2.2170298099517822, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7389167547225952, + "num_tokens": 444878107.0, + "step": 17312 + }, + { + "epoch": 1.9012738853503186, + "grad_norm": 2.3548476696014404, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7186512351036072, + "num_tokens": 444900927.0, + "step": 17313 + }, + { + "epoch": 1.9013837030529321, + "grad_norm": 2.174445629119873, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7158641219139099, + "num_tokens": 444929449.0, + "step": 17314 + }, + { + "epoch": 1.9014935207555457, + "grad_norm": 2.428427219390869, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7339845895767212, + "num_tokens": 444951471.0, + "step": 17315 + }, + { + "epoch": 1.9016033384581594, + "grad_norm": 2.4819369316101074, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7398216724395752, + "num_tokens": 444972948.0, + "step": 17316 + }, + { + "epoch": 1.9017131561607732, + "grad_norm": 2.122281074523926, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7174734473228455, + "num_tokens": 445001156.0, + "step": 17317 + }, + { + "epoch": 1.9018229738633867, + "grad_norm": 2.3522114753723145, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7163584232330322, + "num_tokens": 445024462.0, + "step": 17318 + }, + { + "epoch": 1.9019327915660005, + "grad_norm": 2.4559326171875, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7298785448074341, + "num_tokens": 445046022.0, + "step": 17319 + }, + { + "epoch": 1.902042609268614, + "grad_norm": 2.047976493835449, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7226516008377075, + "num_tokens": 445074683.0, + "step": 17320 + }, + { + "epoch": 1.9021524269712278, + "grad_norm": 2.3278872966766357, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7264363169670105, + "num_tokens": 445097585.0, + "step": 17321 + }, + { + "epoch": 1.9022622446738415, + "grad_norm": 2.3941104412078857, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7212046384811401, + "num_tokens": 445120711.0, + "step": 17322 + }, + { + "epoch": 1.902372062376455, + "grad_norm": 2.047666072845459, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7010297775268555, + "num_tokens": 445151724.0, + "step": 17323 + }, + { + "epoch": 1.9024818800790686, + "grad_norm": 2.279019832611084, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7338109016418457, + "num_tokens": 445176530.0, + "step": 17324 + }, + { + "epoch": 1.9025916977816824, + "grad_norm": 2.2281551361083984, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7417744994163513, + "num_tokens": 445200229.0, + "step": 17325 + }, + { + "epoch": 1.9027015154842961, + "grad_norm": 2.378661632537842, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7209346294403076, + "num_tokens": 445222784.0, + "step": 17326 + }, + { + "epoch": 1.9028113331869099, + "grad_norm": 2.067951202392578, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7128795385360718, + "num_tokens": 445252179.0, + "step": 17327 + }, + { + "epoch": 1.9029211508895234, + "grad_norm": 2.14817214012146, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7197049856185913, + "num_tokens": 445278938.0, + "step": 17328 + }, + { + "epoch": 1.903030968592137, + "grad_norm": 2.658773183822632, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7035056352615356, + "num_tokens": 445301316.0, + "step": 17329 + }, + { + "epoch": 1.9031407862947507, + "grad_norm": 2.1054189205169678, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7329961061477661, + "num_tokens": 445327221.0, + "step": 17330 + }, + { + "epoch": 1.9032506039973645, + "grad_norm": 2.006964921951294, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7119057178497314, + "num_tokens": 445359770.0, + "step": 17331 + }, + { + "epoch": 1.903360421699978, + "grad_norm": 2.257431983947754, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7194546461105347, + "num_tokens": 445384800.0, + "step": 17332 + }, + { + "epoch": 1.9034702394025917, + "grad_norm": 2.0845131874084473, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6957346200942993, + "num_tokens": 445414100.0, + "step": 17333 + }, + { + "epoch": 1.9035800571052053, + "grad_norm": 2.1270360946655273, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7258576154708862, + "num_tokens": 445443137.0, + "step": 17334 + }, + { + "epoch": 1.903689874807819, + "grad_norm": 2.211301565170288, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7194554805755615, + "num_tokens": 445467855.0, + "step": 17335 + }, + { + "epoch": 1.9037996925104328, + "grad_norm": 2.206784963607788, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.704166054725647, + "num_tokens": 445493032.0, + "step": 17336 + }, + { + "epoch": 1.9039095102130463, + "grad_norm": 2.201159954071045, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7221301794052124, + "num_tokens": 445520105.0, + "step": 17337 + }, + { + "epoch": 1.9040193279156599, + "grad_norm": 2.465151786804199, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7440333366394043, + "num_tokens": 445542065.0, + "step": 17338 + }, + { + "epoch": 1.9041291456182736, + "grad_norm": 1.8786673545837402, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.708642840385437, + "num_tokens": 445577976.0, + "step": 17339 + }, + { + "epoch": 1.9042389633208874, + "grad_norm": 2.3616726398468018, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7242279052734375, + "num_tokens": 445601323.0, + "step": 17340 + }, + { + "epoch": 1.9043487810235011, + "grad_norm": 2.069857120513916, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7064071297645569, + "num_tokens": 445630792.0, + "step": 17341 + }, + { + "epoch": 1.9044585987261147, + "grad_norm": 1.8568025827407837, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7363038659095764, + "num_tokens": 445666378.0, + "step": 17342 + }, + { + "epoch": 1.9045684164287282, + "grad_norm": 2.2224037647247314, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7243626117706299, + "num_tokens": 445690826.0, + "step": 17343 + }, + { + "epoch": 1.904678234131342, + "grad_norm": 2.0685534477233887, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7301419377326965, + "num_tokens": 445717066.0, + "step": 17344 + }, + { + "epoch": 1.9047880518339557, + "grad_norm": 2.1848926544189453, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7333028316497803, + "num_tokens": 445742610.0, + "step": 17345 + }, + { + "epoch": 1.9048978695365693, + "grad_norm": 2.166905403137207, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7190741896629333, + "num_tokens": 445769818.0, + "step": 17346 + }, + { + "epoch": 1.9050076872391828, + "grad_norm": 2.4260594844818115, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7369744181632996, + "num_tokens": 445790881.0, + "step": 17347 + }, + { + "epoch": 1.9051175049417965, + "grad_norm": 2.6305437088012695, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7196539640426636, + "num_tokens": 445811377.0, + "step": 17348 + }, + { + "epoch": 1.9052273226444103, + "grad_norm": 2.2605268955230713, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.717252254486084, + "num_tokens": 445836290.0, + "step": 17349 + }, + { + "epoch": 1.905337140347024, + "grad_norm": 2.190876007080078, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7202920913696289, + "num_tokens": 445861403.0, + "step": 17350 + }, + { + "epoch": 1.9054469580496376, + "grad_norm": 2.4723827838897705, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7310658693313599, + "num_tokens": 445881943.0, + "step": 17351 + }, + { + "epoch": 1.9055567757522511, + "grad_norm": 2.0480308532714844, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7258802652359009, + "num_tokens": 445912402.0, + "step": 17352 + }, + { + "epoch": 1.9056665934548649, + "grad_norm": 2.2836883068084717, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7206000685691833, + "num_tokens": 445938443.0, + "step": 17353 + }, + { + "epoch": 1.9057764111574786, + "grad_norm": 2.356762409210205, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7303439974784851, + "num_tokens": 445961539.0, + "step": 17354 + }, + { + "epoch": 1.9058862288600924, + "grad_norm": 1.998653531074524, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6853405833244324, + "num_tokens": 445993618.0, + "step": 17355 + }, + { + "epoch": 1.905996046562706, + "grad_norm": 2.534134864807129, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7485975027084351, + "num_tokens": 446013965.0, + "step": 17356 + }, + { + "epoch": 1.9061058642653195, + "grad_norm": 2.3734216690063477, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7401329278945923, + "num_tokens": 446034976.0, + "step": 17357 + }, + { + "epoch": 1.9062156819679332, + "grad_norm": 2.2537360191345215, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7003687620162964, + "num_tokens": 446062345.0, + "step": 17358 + }, + { + "epoch": 1.906325499670547, + "grad_norm": 2.2506065368652344, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7395790815353394, + "num_tokens": 446085915.0, + "step": 17359 + }, + { + "epoch": 1.9064353173731605, + "grad_norm": 2.176365852355957, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7275968790054321, + "num_tokens": 446112572.0, + "step": 17360 + }, + { + "epoch": 1.906545135075774, + "grad_norm": 2.2801356315612793, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7413173913955688, + "num_tokens": 446138442.0, + "step": 17361 + }, + { + "epoch": 1.9066549527783878, + "grad_norm": 2.356846809387207, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7218714952468872, + "num_tokens": 446163988.0, + "step": 17362 + }, + { + "epoch": 1.9067647704810016, + "grad_norm": 2.5298781394958496, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7261326313018799, + "num_tokens": 446185882.0, + "step": 17363 + }, + { + "epoch": 1.9068745881836153, + "grad_norm": 2.319176197052002, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.73634934425354, + "num_tokens": 446209149.0, + "step": 17364 + }, + { + "epoch": 1.9069844058862289, + "grad_norm": 2.1992270946502686, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7131165266036987, + "num_tokens": 446237374.0, + "step": 17365 + }, + { + "epoch": 1.9070942235888424, + "grad_norm": 2.141627550125122, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7449575662612915, + "num_tokens": 446262657.0, + "step": 17366 + }, + { + "epoch": 1.9072040412914562, + "grad_norm": 2.1581759452819824, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7021451592445374, + "num_tokens": 446292115.0, + "step": 17367 + }, + { + "epoch": 1.90731385899407, + "grad_norm": 2.1707851886749268, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7207763195037842, + "num_tokens": 446319451.0, + "step": 17368 + }, + { + "epoch": 1.9074236766966837, + "grad_norm": 2.2134358882904053, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7394275069236755, + "num_tokens": 446343682.0, + "step": 17369 + }, + { + "epoch": 1.9075334943992972, + "grad_norm": 2.1468257904052734, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7059187889099121, + "num_tokens": 446371547.0, + "step": 17370 + }, + { + "epoch": 1.9076433121019107, + "grad_norm": 1.8865747451782227, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7210729122161865, + "num_tokens": 446407037.0, + "step": 17371 + }, + { + "epoch": 1.9077531298045245, + "grad_norm": 2.092811107635498, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6864307522773743, + "num_tokens": 446437609.0, + "step": 17372 + }, + { + "epoch": 1.9078629475071383, + "grad_norm": 2.1246962547302246, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7030041217803955, + "num_tokens": 446465176.0, + "step": 17373 + }, + { + "epoch": 1.9079727652097518, + "grad_norm": 2.251056432723999, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7325377464294434, + "num_tokens": 446489205.0, + "step": 17374 + }, + { + "epoch": 1.9080825829123653, + "grad_norm": 2.384073257446289, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7191604971885681, + "num_tokens": 446512179.0, + "step": 17375 + }, + { + "epoch": 1.908192400614979, + "grad_norm": 2.167834520339966, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.731124758720398, + "num_tokens": 446537373.0, + "step": 17376 + }, + { + "epoch": 1.9083022183175928, + "grad_norm": 2.246455669403076, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7173765897750854, + "num_tokens": 446563739.0, + "step": 17377 + }, + { + "epoch": 1.9084120360202066, + "grad_norm": 2.291102170944214, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7369793653488159, + "num_tokens": 446588027.0, + "step": 17378 + }, + { + "epoch": 1.9085218537228201, + "grad_norm": 2.268629312515259, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.717544674873352, + "num_tokens": 446612251.0, + "step": 17379 + }, + { + "epoch": 1.9086316714254337, + "grad_norm": 2.186152219772339, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7162884473800659, + "num_tokens": 446638883.0, + "step": 17380 + }, + { + "epoch": 1.9087414891280474, + "grad_norm": 2.330467939376831, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7257205247879028, + "num_tokens": 446662363.0, + "step": 17381 + }, + { + "epoch": 1.9088513068306612, + "grad_norm": 2.476015567779541, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7215918898582458, + "num_tokens": 446684236.0, + "step": 17382 + }, + { + "epoch": 1.9089611245332747, + "grad_norm": 2.228391647338867, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7347884178161621, + "num_tokens": 446708793.0, + "step": 17383 + }, + { + "epoch": 1.9090709422358885, + "grad_norm": 2.6275243759155273, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7440213561058044, + "num_tokens": 446726607.0, + "step": 17384 + }, + { + "epoch": 1.909180759938502, + "grad_norm": 2.2350332736968994, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7204967141151428, + "num_tokens": 446750456.0, + "step": 17385 + }, + { + "epoch": 1.9092905776411158, + "grad_norm": 2.091057777404785, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7014958262443542, + "num_tokens": 446780854.0, + "step": 17386 + }, + { + "epoch": 1.9094003953437295, + "grad_norm": 2.301318645477295, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7223309874534607, + "num_tokens": 446805922.0, + "step": 17387 + }, + { + "epoch": 1.909510213046343, + "grad_norm": 2.3623571395874023, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.717186689376831, + "num_tokens": 446829884.0, + "step": 17388 + }, + { + "epoch": 1.9096200307489566, + "grad_norm": 2.21030592918396, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7164048552513123, + "num_tokens": 446856344.0, + "step": 17389 + }, + { + "epoch": 1.9097298484515703, + "grad_norm": 2.511474609375, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7387544512748718, + "num_tokens": 446877827.0, + "step": 17390 + }, + { + "epoch": 1.909839666154184, + "grad_norm": 2.428837299346924, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7197006940841675, + "num_tokens": 446901428.0, + "step": 17391 + }, + { + "epoch": 1.9099494838567979, + "grad_norm": 1.9858404397964478, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7136166095733643, + "num_tokens": 446932329.0, + "step": 17392 + }, + { + "epoch": 1.9100593015594114, + "grad_norm": 2.6332695484161377, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7159274816513062, + "num_tokens": 446951408.0, + "step": 17393 + }, + { + "epoch": 1.910169119262025, + "grad_norm": 2.261845350265503, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7121590375900269, + "num_tokens": 446978210.0, + "step": 17394 + }, + { + "epoch": 1.9102789369646387, + "grad_norm": 2.1876869201660156, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.731746256351471, + "num_tokens": 447003182.0, + "step": 17395 + }, + { + "epoch": 1.9103887546672524, + "grad_norm": 2.0140199661254883, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7149993777275085, + "num_tokens": 447033888.0, + "step": 17396 + }, + { + "epoch": 1.910498572369866, + "grad_norm": 2.365760326385498, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7582377195358276, + "num_tokens": 447055451.0, + "step": 17397 + }, + { + "epoch": 1.9106083900724797, + "grad_norm": 2.0350842475891113, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7108968496322632, + "num_tokens": 447086151.0, + "step": 17398 + }, + { + "epoch": 1.9107182077750933, + "grad_norm": 2.3692667484283447, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7380924820899963, + "num_tokens": 447110569.0, + "step": 17399 + }, + { + "epoch": 1.910828025477707, + "grad_norm": 2.149984121322632, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7322911620140076, + "num_tokens": 447135824.0, + "step": 17400 + }, + { + "epoch": 1.9109378431803208, + "grad_norm": 2.165698289871216, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.721495509147644, + "num_tokens": 447162231.0, + "step": 17401 + }, + { + "epoch": 1.9110476608829343, + "grad_norm": 2.070852279663086, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7162042260169983, + "num_tokens": 447189517.0, + "step": 17402 + }, + { + "epoch": 1.9111574785855479, + "grad_norm": 2.725466251373291, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7250639200210571, + "num_tokens": 447208112.0, + "step": 17403 + }, + { + "epoch": 1.9112672962881616, + "grad_norm": 2.0904996395111084, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7195223569869995, + "num_tokens": 447237403.0, + "step": 17404 + }, + { + "epoch": 1.9113771139907754, + "grad_norm": 2.4672114849090576, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7310501337051392, + "num_tokens": 447258679.0, + "step": 17405 + }, + { + "epoch": 1.9114869316933891, + "grad_norm": 2.496270179748535, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7340213656425476, + "num_tokens": 447281516.0, + "step": 17406 + }, + { + "epoch": 1.9115967493960027, + "grad_norm": 2.3886594772338867, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7340095043182373, + "num_tokens": 447302126.0, + "step": 17407 + }, + { + "epoch": 1.9117065670986162, + "grad_norm": 2.34378719329834, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7255263328552246, + "num_tokens": 447324920.0, + "step": 17408 + }, + { + "epoch": 1.91181638480123, + "grad_norm": 2.2060208320617676, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7216925621032715, + "num_tokens": 447353905.0, + "step": 17409 + }, + { + "epoch": 1.9119262025038437, + "grad_norm": 2.0620293617248535, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7121833562850952, + "num_tokens": 447382996.0, + "step": 17410 + }, + { + "epoch": 1.9120360202064572, + "grad_norm": 2.3597772121429443, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7097779512405396, + "num_tokens": 447405368.0, + "step": 17411 + }, + { + "epoch": 1.9121458379090708, + "grad_norm": 2.195622444152832, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7328276634216309, + "num_tokens": 447431444.0, + "step": 17412 + }, + { + "epoch": 1.9122556556116845, + "grad_norm": 2.0646705627441406, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7503384947776794, + "num_tokens": 447458115.0, + "step": 17413 + }, + { + "epoch": 1.9123654733142983, + "grad_norm": 2.360334634780884, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7327970266342163, + "num_tokens": 447481866.0, + "step": 17414 + }, + { + "epoch": 1.912475291016912, + "grad_norm": 2.216588258743286, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6999707221984863, + "num_tokens": 447508395.0, + "step": 17415 + }, + { + "epoch": 1.9125851087195256, + "grad_norm": 2.345255136489868, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.723504364490509, + "num_tokens": 447532614.0, + "step": 17416 + }, + { + "epoch": 1.9126949264221391, + "grad_norm": 2.1329903602600098, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7396010756492615, + "num_tokens": 447558997.0, + "step": 17417 + }, + { + "epoch": 1.9128047441247529, + "grad_norm": 2.2940101623535156, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7367701530456543, + "num_tokens": 447583279.0, + "step": 17418 + }, + { + "epoch": 1.9129145618273666, + "grad_norm": 2.1301732063293457, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7433502674102783, + "num_tokens": 447609240.0, + "step": 17419 + }, + { + "epoch": 1.9130243795299804, + "grad_norm": 2.4986624717712402, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.731661856174469, + "num_tokens": 447631055.0, + "step": 17420 + }, + { + "epoch": 1.913134197232594, + "grad_norm": 2.3018271923065186, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7362546920776367, + "num_tokens": 447655256.0, + "step": 17421 + }, + { + "epoch": 1.9132440149352075, + "grad_norm": 2.270333766937256, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7161318063735962, + "num_tokens": 447682507.0, + "step": 17422 + }, + { + "epoch": 1.9133538326378212, + "grad_norm": 2.4414310455322266, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7151129245758057, + "num_tokens": 447707723.0, + "step": 17423 + }, + { + "epoch": 1.913463650340435, + "grad_norm": 2.4168715476989746, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.721406102180481, + "num_tokens": 447731913.0, + "step": 17424 + }, + { + "epoch": 1.9135734680430485, + "grad_norm": 2.14821720123291, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7226591110229492, + "num_tokens": 447759390.0, + "step": 17425 + }, + { + "epoch": 1.913683285745662, + "grad_norm": 2.2671616077423096, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7294848561286926, + "num_tokens": 447783830.0, + "step": 17426 + }, + { + "epoch": 1.9137931034482758, + "grad_norm": 2.3419620990753174, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7201009392738342, + "num_tokens": 447807377.0, + "step": 17427 + }, + { + "epoch": 1.9139029211508896, + "grad_norm": 1.8775222301483154, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7153558731079102, + "num_tokens": 447842512.0, + "step": 17428 + }, + { + "epoch": 1.9140127388535033, + "grad_norm": 2.5128989219665527, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7223362922668457, + "num_tokens": 447863704.0, + "step": 17429 + }, + { + "epoch": 1.9141225565561168, + "grad_norm": 2.240913152694702, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7093490362167358, + "num_tokens": 447890840.0, + "step": 17430 + }, + { + "epoch": 1.9142323742587304, + "grad_norm": 2.407214403152466, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7164795994758606, + "num_tokens": 447913508.0, + "step": 17431 + }, + { + "epoch": 1.9143421919613441, + "grad_norm": 2.1757755279541016, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7168418169021606, + "num_tokens": 447940030.0, + "step": 17432 + }, + { + "epoch": 1.914452009663958, + "grad_norm": 2.083186149597168, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7479141354560852, + "num_tokens": 447967713.0, + "step": 17433 + }, + { + "epoch": 1.9145618273665717, + "grad_norm": 2.4282050132751465, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7362052202224731, + "num_tokens": 447989552.0, + "step": 17434 + }, + { + "epoch": 1.9146716450691852, + "grad_norm": 2.4282145500183105, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7336851954460144, + "num_tokens": 448010700.0, + "step": 17435 + }, + { + "epoch": 1.9147814627717987, + "grad_norm": 2.3026185035705566, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.701947033405304, + "num_tokens": 448035771.0, + "step": 17436 + }, + { + "epoch": 1.9148912804744125, + "grad_norm": 2.3783721923828125, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6986762285232544, + "num_tokens": 448061676.0, + "step": 17437 + }, + { + "epoch": 1.9150010981770262, + "grad_norm": 2.3023881912231445, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7065219879150391, + "num_tokens": 448087940.0, + "step": 17438 + }, + { + "epoch": 1.9151109158796398, + "grad_norm": 2.2020785808563232, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7058275938034058, + "num_tokens": 448116681.0, + "step": 17439 + }, + { + "epoch": 1.9152207335822533, + "grad_norm": 2.4851019382476807, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7316125631332397, + "num_tokens": 448139000.0, + "step": 17440 + }, + { + "epoch": 1.915330551284867, + "grad_norm": 2.016874313354492, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7277106046676636, + "num_tokens": 448168786.0, + "step": 17441 + }, + { + "epoch": 1.9154403689874808, + "grad_norm": 2.2668871879577637, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7465802431106567, + "num_tokens": 448192971.0, + "step": 17442 + }, + { + "epoch": 1.9155501866900946, + "grad_norm": 2.027517795562744, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7096837759017944, + "num_tokens": 448225881.0, + "step": 17443 + }, + { + "epoch": 1.9156600043927081, + "grad_norm": 2.322659969329834, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7243655323982239, + "num_tokens": 448250238.0, + "step": 17444 + }, + { + "epoch": 1.9157698220953217, + "grad_norm": 2.300673007965088, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7158203721046448, + "num_tokens": 448276365.0, + "step": 17445 + }, + { + "epoch": 1.9158796397979354, + "grad_norm": 2.426788091659546, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7366963028907776, + "num_tokens": 448298387.0, + "step": 17446 + }, + { + "epoch": 1.9159894575005492, + "grad_norm": 2.2014882564544678, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7058809995651245, + "num_tokens": 448325735.0, + "step": 17447 + }, + { + "epoch": 1.9160992752031627, + "grad_norm": 2.061469316482544, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7044095993041992, + "num_tokens": 448355388.0, + "step": 17448 + }, + { + "epoch": 1.9162090929057765, + "grad_norm": 2.0734879970550537, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7289339900016785, + "num_tokens": 448384661.0, + "step": 17449 + }, + { + "epoch": 1.91631891060839, + "grad_norm": 2.0549826622009277, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7119558453559875, + "num_tokens": 448413952.0, + "step": 17450 + }, + { + "epoch": 1.9164287283110037, + "grad_norm": 2.0562942028045654, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7242767810821533, + "num_tokens": 448442521.0, + "step": 17451 + }, + { + "epoch": 1.9165385460136175, + "grad_norm": 2.307774782180786, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6983771324157715, + "num_tokens": 448468572.0, + "step": 17452 + }, + { + "epoch": 1.916648363716231, + "grad_norm": 2.209879159927368, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7154697775840759, + "num_tokens": 448493212.0, + "step": 17453 + }, + { + "epoch": 1.9167581814188446, + "grad_norm": 2.2369086742401123, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.724173903465271, + "num_tokens": 448518827.0, + "step": 17454 + }, + { + "epoch": 1.9168679991214583, + "grad_norm": 2.176881790161133, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.719171404838562, + "num_tokens": 448545794.0, + "step": 17455 + }, + { + "epoch": 1.916977816824072, + "grad_norm": 2.169574737548828, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7160453796386719, + "num_tokens": 448572155.0, + "step": 17456 + }, + { + "epoch": 1.9170876345266858, + "grad_norm": 2.394862174987793, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7365802526473999, + "num_tokens": 448594129.0, + "step": 17457 + }, + { + "epoch": 1.9171974522292994, + "grad_norm": 2.4439969062805176, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7030285596847534, + "num_tokens": 448619491.0, + "step": 17458 + }, + { + "epoch": 1.917307269931913, + "grad_norm": 2.1480460166931152, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7147740125656128, + "num_tokens": 448647166.0, + "step": 17459 + }, + { + "epoch": 1.9174170876345267, + "grad_norm": 1.9498116970062256, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7116392254829407, + "num_tokens": 448680643.0, + "step": 17460 + }, + { + "epoch": 1.9175269053371404, + "grad_norm": 2.3585102558135986, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7379446029663086, + "num_tokens": 448705057.0, + "step": 17461 + }, + { + "epoch": 1.917636723039754, + "grad_norm": 2.147148847579956, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7224094867706299, + "num_tokens": 448732442.0, + "step": 17462 + }, + { + "epoch": 1.9177465407423677, + "grad_norm": 2.2204864025115967, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7325634956359863, + "num_tokens": 448757267.0, + "step": 17463 + }, + { + "epoch": 1.9178563584449813, + "grad_norm": 2.4511756896972656, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7086697816848755, + "num_tokens": 448779794.0, + "step": 17464 + }, + { + "epoch": 1.917966176147595, + "grad_norm": 2.1312437057495117, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.731601357460022, + "num_tokens": 448805771.0, + "step": 17465 + }, + { + "epoch": 1.9180759938502088, + "grad_norm": 2.0436878204345703, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6980576515197754, + "num_tokens": 448836976.0, + "step": 17466 + }, + { + "epoch": 1.9181858115528223, + "grad_norm": 2.1370038986206055, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7263800501823425, + "num_tokens": 448864008.0, + "step": 17467 + }, + { + "epoch": 1.9182956292554358, + "grad_norm": 2.3044450283050537, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7110446691513062, + "num_tokens": 448889269.0, + "step": 17468 + }, + { + "epoch": 1.9184054469580496, + "grad_norm": 2.3175201416015625, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7236687541007996, + "num_tokens": 448912584.0, + "step": 17469 + }, + { + "epoch": 1.9185152646606634, + "grad_norm": 2.2564620971679688, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7143998146057129, + "num_tokens": 448936682.0, + "step": 17470 + }, + { + "epoch": 1.9186250823632771, + "grad_norm": 2.1898207664489746, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6957966685295105, + "num_tokens": 448963192.0, + "step": 17471 + }, + { + "epoch": 1.9187349000658906, + "grad_norm": 2.5334572792053223, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.709354043006897, + "num_tokens": 448985646.0, + "step": 17472 + }, + { + "epoch": 1.9188447177685042, + "grad_norm": 2.3045287132263184, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7619518041610718, + "num_tokens": 449007855.0, + "step": 17473 + }, + { + "epoch": 1.918954535471118, + "grad_norm": 2.2019593715667725, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7199530601501465, + "num_tokens": 449034546.0, + "step": 17474 + }, + { + "epoch": 1.9190643531737317, + "grad_norm": 2.082578659057617, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7355821132659912, + "num_tokens": 449062158.0, + "step": 17475 + }, + { + "epoch": 1.9191741708763452, + "grad_norm": 1.8964989185333252, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7160375118255615, + "num_tokens": 449094513.0, + "step": 17476 + }, + { + "epoch": 1.9192839885789588, + "grad_norm": 2.4089486598968506, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7382155656814575, + "num_tokens": 449116173.0, + "step": 17477 + }, + { + "epoch": 1.9193938062815725, + "grad_norm": 2.0038223266601562, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6994503140449524, + "num_tokens": 449146857.0, + "step": 17478 + }, + { + "epoch": 1.9195036239841863, + "grad_norm": 1.9213835000991821, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7206679582595825, + "num_tokens": 449177474.0, + "step": 17479 + }, + { + "epoch": 1.9196134416868, + "grad_norm": 2.5111536979675293, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.73153156042099, + "num_tokens": 449196952.0, + "step": 17480 + }, + { + "epoch": 1.9197232593894136, + "grad_norm": 2.6643335819244385, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7437558174133301, + "num_tokens": 449215742.0, + "step": 17481 + }, + { + "epoch": 1.919833077092027, + "grad_norm": 2.0789496898651123, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7117518782615662, + "num_tokens": 449245027.0, + "step": 17482 + }, + { + "epoch": 1.9199428947946409, + "grad_norm": 2.3460495471954346, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7419072389602661, + "num_tokens": 449266747.0, + "step": 17483 + }, + { + "epoch": 1.9200527124972546, + "grad_norm": 2.1263108253479004, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7058286666870117, + "num_tokens": 449295421.0, + "step": 17484 + }, + { + "epoch": 1.9201625301998684, + "grad_norm": 2.0516357421875, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7039377689361572, + "num_tokens": 449324398.0, + "step": 17485 + }, + { + "epoch": 1.920272347902482, + "grad_norm": 2.159494400024414, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7067471742630005, + "num_tokens": 449353163.0, + "step": 17486 + }, + { + "epoch": 1.9203821656050954, + "grad_norm": 2.499664068222046, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7565092444419861, + "num_tokens": 449371739.0, + "step": 17487 + }, + { + "epoch": 1.9204919833077092, + "grad_norm": 2.3234052658081055, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7031739950180054, + "num_tokens": 449394885.0, + "step": 17488 + }, + { + "epoch": 1.920601801010323, + "grad_norm": 2.021890878677368, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7225803136825562, + "num_tokens": 449427373.0, + "step": 17489 + }, + { + "epoch": 1.9207116187129365, + "grad_norm": 2.520352363586426, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7154141664505005, + "num_tokens": 449451645.0, + "step": 17490 + }, + { + "epoch": 1.92082143641555, + "grad_norm": 2.4171977043151855, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.708701491355896, + "num_tokens": 449476690.0, + "step": 17491 + }, + { + "epoch": 1.9209312541181638, + "grad_norm": 2.4793360233306885, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7244082093238831, + "num_tokens": 449496918.0, + "step": 17492 + }, + { + "epoch": 1.9210410718207775, + "grad_norm": 2.13989520072937, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.717821478843689, + "num_tokens": 449526873.0, + "step": 17493 + }, + { + "epoch": 1.9211508895233913, + "grad_norm": 2.627467393875122, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7450635433197021, + "num_tokens": 449547453.0, + "step": 17494 + }, + { + "epoch": 1.9212607072260048, + "grad_norm": 2.200110673904419, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7299834489822388, + "num_tokens": 449572751.0, + "step": 17495 + }, + { + "epoch": 1.9213705249286184, + "grad_norm": 2.1200995445251465, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.723598837852478, + "num_tokens": 449604301.0, + "step": 17496 + }, + { + "epoch": 1.9214803426312321, + "grad_norm": 2.0623605251312256, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7491687536239624, + "num_tokens": 449632171.0, + "step": 17497 + }, + { + "epoch": 1.9215901603338459, + "grad_norm": 2.045830488204956, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7206480503082275, + "num_tokens": 449661414.0, + "step": 17498 + }, + { + "epoch": 1.9216999780364594, + "grad_norm": 2.128284215927124, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7079795002937317, + "num_tokens": 449690531.0, + "step": 17499 + }, + { + "epoch": 1.9218097957390732, + "grad_norm": 2.229936361312866, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.6974706649780273, + "num_tokens": 449717713.0, + "step": 17500 + }, + { + "epoch": 1.9219196134416867, + "grad_norm": 2.3030147552490234, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.702924907207489, + "num_tokens": 449743011.0, + "step": 17501 + }, + { + "epoch": 1.9220294311443005, + "grad_norm": 2.603652000427246, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7119885683059692, + "num_tokens": 449762882.0, + "step": 17502 + }, + { + "epoch": 1.9221392488469142, + "grad_norm": 2.415463447570801, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7350132465362549, + "num_tokens": 449785535.0, + "step": 17503 + }, + { + "epoch": 1.9222490665495278, + "grad_norm": 2.1941516399383545, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7081514000892639, + "num_tokens": 449812820.0, + "step": 17504 + }, + { + "epoch": 1.9223588842521413, + "grad_norm": 2.3215887546539307, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7281089425086975, + "num_tokens": 449835874.0, + "step": 17505 + }, + { + "epoch": 1.922468701954755, + "grad_norm": 2.0870394706726074, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6903800964355469, + "num_tokens": 449866390.0, + "step": 17506 + }, + { + "epoch": 1.9225785196573688, + "grad_norm": 2.4020004272460938, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7260761260986328, + "num_tokens": 449887637.0, + "step": 17507 + }, + { + "epoch": 1.9226883373599826, + "grad_norm": 2.5747292041778564, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.758846640586853, + "num_tokens": 449906238.0, + "step": 17508 + }, + { + "epoch": 1.922798155062596, + "grad_norm": 2.5832695960998535, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.724237322807312, + "num_tokens": 449926401.0, + "step": 17509 + }, + { + "epoch": 1.9229079727652096, + "grad_norm": 2.1710822582244873, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7224550247192383, + "num_tokens": 449953848.0, + "step": 17510 + }, + { + "epoch": 1.9230177904678234, + "grad_norm": 2.5485222339630127, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7450504302978516, + "num_tokens": 449975025.0, + "step": 17511 + }, + { + "epoch": 1.9231276081704372, + "grad_norm": 1.9446882009506226, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7043856382369995, + "num_tokens": 450008518.0, + "step": 17512 + }, + { + "epoch": 1.9232374258730507, + "grad_norm": 2.0649783611297607, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7120600342750549, + "num_tokens": 450037751.0, + "step": 17513 + }, + { + "epoch": 1.9233472435756644, + "grad_norm": 2.3792479038238525, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7129294872283936, + "num_tokens": 450061105.0, + "step": 17514 + }, + { + "epoch": 1.923457061278278, + "grad_norm": 2.260971784591675, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6877437829971313, + "num_tokens": 450088251.0, + "step": 17515 + }, + { + "epoch": 1.9235668789808917, + "grad_norm": 2.2608253955841064, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.733857274055481, + "num_tokens": 450112245.0, + "step": 17516 + }, + { + "epoch": 1.9236766966835055, + "grad_norm": 1.9982974529266357, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7281383275985718, + "num_tokens": 450142358.0, + "step": 17517 + }, + { + "epoch": 1.923786514386119, + "grad_norm": 2.4603471755981445, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7326381802558899, + "num_tokens": 450163640.0, + "step": 17518 + }, + { + "epoch": 1.9238963320887326, + "grad_norm": 2.207303762435913, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.707924485206604, + "num_tokens": 450189272.0, + "step": 17519 + }, + { + "epoch": 1.9240061497913463, + "grad_norm": 2.5199081897735596, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7327405214309692, + "num_tokens": 450209815.0, + "step": 17520 + }, + { + "epoch": 1.92411596749396, + "grad_norm": 2.197498321533203, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7102011442184448, + "num_tokens": 450236249.0, + "step": 17521 + }, + { + "epoch": 1.9242257851965738, + "grad_norm": 2.247960090637207, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7048107385635376, + "num_tokens": 450264560.0, + "step": 17522 + }, + { + "epoch": 1.9243356028991874, + "grad_norm": 2.286393880844116, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7197664976119995, + "num_tokens": 450289599.0, + "step": 17523 + }, + { + "epoch": 1.924445420601801, + "grad_norm": 2.210232734680176, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7113547325134277, + "num_tokens": 450316507.0, + "step": 17524 + }, + { + "epoch": 1.9245552383044147, + "grad_norm": 2.1736841201782227, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7217082977294922, + "num_tokens": 450342876.0, + "step": 17525 + }, + { + "epoch": 1.9246650560070284, + "grad_norm": 2.348177671432495, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7344771027565002, + "num_tokens": 450366367.0, + "step": 17526 + }, + { + "epoch": 1.924774873709642, + "grad_norm": 2.624330520629883, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7200073003768921, + "num_tokens": 450385936.0, + "step": 17527 + }, + { + "epoch": 1.9248846914122557, + "grad_norm": 2.384606122970581, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7266563177108765, + "num_tokens": 450409599.0, + "step": 17528 + }, + { + "epoch": 1.9249945091148692, + "grad_norm": 2.2756898403167725, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7244994044303894, + "num_tokens": 450435749.0, + "step": 17529 + }, + { + "epoch": 1.925104326817483, + "grad_norm": 2.4433114528656006, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7016801834106445, + "num_tokens": 450459265.0, + "step": 17530 + }, + { + "epoch": 1.9252141445200968, + "grad_norm": 2.3671181201934814, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7207682132720947, + "num_tokens": 450483751.0, + "step": 17531 + }, + { + "epoch": 1.9253239622227103, + "grad_norm": 2.200340986251831, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7523772120475769, + "num_tokens": 450510886.0, + "step": 17532 + }, + { + "epoch": 1.9254337799253238, + "grad_norm": 1.8933665752410889, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7047528028488159, + "num_tokens": 450542832.0, + "step": 17533 + }, + { + "epoch": 1.9255435976279376, + "grad_norm": 2.3416523933410645, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7360296249389648, + "num_tokens": 450565481.0, + "step": 17534 + }, + { + "epoch": 1.9256534153305513, + "grad_norm": 2.2146317958831787, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7081519365310669, + "num_tokens": 450592853.0, + "step": 17535 + }, + { + "epoch": 1.925763233033165, + "grad_norm": 2.1811273097991943, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7080785036087036, + "num_tokens": 450620169.0, + "step": 17536 + }, + { + "epoch": 1.9258730507357786, + "grad_norm": 2.1862964630126953, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7227952480316162, + "num_tokens": 450647980.0, + "step": 17537 + }, + { + "epoch": 1.9259828684383922, + "grad_norm": 2.101128578186035, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7007116079330444, + "num_tokens": 450677737.0, + "step": 17538 + }, + { + "epoch": 1.926092686141006, + "grad_norm": 2.1490166187286377, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7360941171646118, + "num_tokens": 450703942.0, + "step": 17539 + }, + { + "epoch": 1.9262025038436197, + "grad_norm": 2.232626438140869, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7279939651489258, + "num_tokens": 450730437.0, + "step": 17540 + }, + { + "epoch": 1.9263123215462332, + "grad_norm": 2.213179349899292, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7234671115875244, + "num_tokens": 450756118.0, + "step": 17541 + }, + { + "epoch": 1.9264221392488468, + "grad_norm": 2.362572431564331, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7395570874214172, + "num_tokens": 450781619.0, + "step": 17542 + }, + { + "epoch": 1.9265319569514605, + "grad_norm": 2.178924798965454, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7004359364509583, + "num_tokens": 450812197.0, + "step": 17543 + }, + { + "epoch": 1.9266417746540743, + "grad_norm": 2.4817070960998535, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7131235003471375, + "num_tokens": 450834246.0, + "step": 17544 + }, + { + "epoch": 1.926751592356688, + "grad_norm": 2.474977493286133, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7357127666473389, + "num_tokens": 450855650.0, + "step": 17545 + }, + { + "epoch": 1.9268614100593016, + "grad_norm": 2.0303432941436768, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7129186987876892, + "num_tokens": 450885517.0, + "step": 17546 + }, + { + "epoch": 1.926971227761915, + "grad_norm": 2.3782870769500732, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7399769425392151, + "num_tokens": 450908683.0, + "step": 17547 + }, + { + "epoch": 1.9270810454645289, + "grad_norm": 2.058061361312866, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6918580532073975, + "num_tokens": 450939313.0, + "step": 17548 + }, + { + "epoch": 1.9271908631671426, + "grad_norm": 2.243587017059326, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7021417617797852, + "num_tokens": 450965759.0, + "step": 17549 + }, + { + "epoch": 1.9273006808697564, + "grad_norm": 2.2810699939727783, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7191108465194702, + "num_tokens": 450991249.0, + "step": 17550 + }, + { + "epoch": 1.92741049857237, + "grad_norm": 2.007833480834961, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7232376337051392, + "num_tokens": 451022578.0, + "step": 17551 + }, + { + "epoch": 1.9275203162749834, + "grad_norm": 2.2032477855682373, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6950966119766235, + "num_tokens": 451049760.0, + "step": 17552 + }, + { + "epoch": 1.9276301339775972, + "grad_norm": 2.2306950092315674, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7195041179656982, + "num_tokens": 451075972.0, + "step": 17553 + }, + { + "epoch": 1.927739951680211, + "grad_norm": 2.368147850036621, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7293835282325745, + "num_tokens": 451097638.0, + "step": 17554 + }, + { + "epoch": 1.9278497693828245, + "grad_norm": 2.0714972019195557, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6990432143211365, + "num_tokens": 451125375.0, + "step": 17555 + }, + { + "epoch": 1.927959587085438, + "grad_norm": 2.3432047367095947, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7181710004806519, + "num_tokens": 451149891.0, + "step": 17556 + }, + { + "epoch": 1.9280694047880518, + "grad_norm": 2.1604018211364746, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.706862211227417, + "num_tokens": 451176907.0, + "step": 17557 + }, + { + "epoch": 1.9281792224906655, + "grad_norm": 2.2535312175750732, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.729228138923645, + "num_tokens": 451203139.0, + "step": 17558 + }, + { + "epoch": 1.9282890401932793, + "grad_norm": 2.198495626449585, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7098857164382935, + "num_tokens": 451230098.0, + "step": 17559 + }, + { + "epoch": 1.9283988578958928, + "grad_norm": 2.5983502864837646, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.738247811794281, + "num_tokens": 451249536.0, + "step": 17560 + }, + { + "epoch": 1.9285086755985064, + "grad_norm": 2.069809675216675, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6975513696670532, + "num_tokens": 451281544.0, + "step": 17561 + }, + { + "epoch": 1.9286184933011201, + "grad_norm": 2.0216562747955322, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.689663827419281, + "num_tokens": 451311328.0, + "step": 17562 + }, + { + "epoch": 1.9287283110037339, + "grad_norm": 2.2531888484954834, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7115519642829895, + "num_tokens": 451338730.0, + "step": 17563 + }, + { + "epoch": 1.9288381287063474, + "grad_norm": 2.477083206176758, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.722041666507721, + "num_tokens": 451362801.0, + "step": 17564 + }, + { + "epoch": 1.9289479464089612, + "grad_norm": 2.099142551422119, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7325794100761414, + "num_tokens": 451390102.0, + "step": 17565 + }, + { + "epoch": 1.9290577641115747, + "grad_norm": 2.253857374191284, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7161175012588501, + "num_tokens": 451413449.0, + "step": 17566 + }, + { + "epoch": 1.9291675818141885, + "grad_norm": 2.1502323150634766, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7034748792648315, + "num_tokens": 451441705.0, + "step": 17567 + }, + { + "epoch": 1.9292773995168022, + "grad_norm": 2.3990914821624756, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7175438404083252, + "num_tokens": 451462896.0, + "step": 17568 + }, + { + "epoch": 1.9293872172194158, + "grad_norm": 2.3652455806732178, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7187921404838562, + "num_tokens": 451484451.0, + "step": 17569 + }, + { + "epoch": 1.9294970349220293, + "grad_norm": 2.3294100761413574, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7060419917106628, + "num_tokens": 451509704.0, + "step": 17570 + }, + { + "epoch": 1.929606852624643, + "grad_norm": 2.3703413009643555, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7234455347061157, + "num_tokens": 451534066.0, + "step": 17571 + }, + { + "epoch": 1.9297166703272568, + "grad_norm": 2.1661031246185303, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7002384662628174, + "num_tokens": 451562752.0, + "step": 17572 + }, + { + "epoch": 1.9298264880298706, + "grad_norm": 2.387908935546875, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7297328114509583, + "num_tokens": 451585062.0, + "step": 17573 + }, + { + "epoch": 1.929936305732484, + "grad_norm": 2.6174490451812744, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7431706190109253, + "num_tokens": 451604966.0, + "step": 17574 + }, + { + "epoch": 1.9300461234350976, + "grad_norm": 2.405653953552246, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7038146257400513, + "num_tokens": 451628242.0, + "step": 17575 + }, + { + "epoch": 1.9301559411377114, + "grad_norm": 2.4489457607269287, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7289286851882935, + "num_tokens": 451649133.0, + "step": 17576 + }, + { + "epoch": 1.9302657588403251, + "grad_norm": 2.59509539604187, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7243754267692566, + "num_tokens": 451670017.0, + "step": 17577 + }, + { + "epoch": 1.9303755765429387, + "grad_norm": 2.26387882232666, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7324265241622925, + "num_tokens": 451696072.0, + "step": 17578 + }, + { + "epoch": 1.9304853942455524, + "grad_norm": 2.0956411361694336, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7198697924613953, + "num_tokens": 451723076.0, + "step": 17579 + }, + { + "epoch": 1.930595211948166, + "grad_norm": 2.4647252559661865, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7171190977096558, + "num_tokens": 451744973.0, + "step": 17580 + }, + { + "epoch": 1.9307050296507797, + "grad_norm": 2.153383255004883, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7063940763473511, + "num_tokens": 451773283.0, + "step": 17581 + }, + { + "epoch": 1.9308148473533935, + "grad_norm": 2.331190347671509, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7279237508773804, + "num_tokens": 451797451.0, + "step": 17582 + }, + { + "epoch": 1.930924665056007, + "grad_norm": 2.3221752643585205, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7021617889404297, + "num_tokens": 451822169.0, + "step": 17583 + }, + { + "epoch": 1.9310344827586206, + "grad_norm": 2.5513217449188232, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7313574552536011, + "num_tokens": 451842928.0, + "step": 17584 + }, + { + "epoch": 1.9311443004612343, + "grad_norm": 2.312371253967285, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6903075575828552, + "num_tokens": 451868395.0, + "step": 17585 + }, + { + "epoch": 1.931254118163848, + "grad_norm": 2.1737608909606934, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7407518029212952, + "num_tokens": 451894502.0, + "step": 17586 + }, + { + "epoch": 1.9313639358664618, + "grad_norm": 2.142022132873535, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7213485836982727, + "num_tokens": 451919492.0, + "step": 17587 + }, + { + "epoch": 1.9314737535690754, + "grad_norm": 2.159912109375, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.714564323425293, + "num_tokens": 451948245.0, + "step": 17588 + }, + { + "epoch": 1.931583571271689, + "grad_norm": 2.563586473464966, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.732421875, + "num_tokens": 451968830.0, + "step": 17589 + }, + { + "epoch": 1.9316933889743026, + "grad_norm": 2.311380624771118, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7253612875938416, + "num_tokens": 451993653.0, + "step": 17590 + }, + { + "epoch": 1.9318032066769164, + "grad_norm": 2.263827323913574, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7139350771903992, + "num_tokens": 452019513.0, + "step": 17591 + }, + { + "epoch": 1.93191302437953, + "grad_norm": 2.041860580444336, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7262364625930786, + "num_tokens": 452050207.0, + "step": 17592 + }, + { + "epoch": 1.9320228420821435, + "grad_norm": 2.163421154022217, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7252069711685181, + "num_tokens": 452076540.0, + "step": 17593 + }, + { + "epoch": 1.9321326597847572, + "grad_norm": 2.5968384742736816, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7232985496520996, + "num_tokens": 452096674.0, + "step": 17594 + }, + { + "epoch": 1.932242477487371, + "grad_norm": 2.1427643299102783, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7169545888900757, + "num_tokens": 452122731.0, + "step": 17595 + }, + { + "epoch": 1.9323522951899847, + "grad_norm": 1.9841605424880981, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7452414035797119, + "num_tokens": 452152122.0, + "step": 17596 + }, + { + "epoch": 1.9324621128925983, + "grad_norm": 2.2111074924468994, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7350505590438843, + "num_tokens": 452178980.0, + "step": 17597 + }, + { + "epoch": 1.9325719305952118, + "grad_norm": 2.335692882537842, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7519798278808594, + "num_tokens": 452201153.0, + "step": 17598 + }, + { + "epoch": 1.9326817482978256, + "grad_norm": 1.9661333560943604, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7139444947242737, + "num_tokens": 452234059.0, + "step": 17599 + }, + { + "epoch": 1.9327915660004393, + "grad_norm": 2.038867235183716, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7180202007293701, + "num_tokens": 452262971.0, + "step": 17600 + }, + { + "epoch": 1.932901383703053, + "grad_norm": 2.4621920585632324, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7308177947998047, + "num_tokens": 452285083.0, + "step": 17601 + }, + { + "epoch": 1.9330112014056666, + "grad_norm": 2.2561416625976562, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7286876440048218, + "num_tokens": 452310505.0, + "step": 17602 + }, + { + "epoch": 1.9331210191082802, + "grad_norm": 2.1600210666656494, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7336509823799133, + "num_tokens": 452338982.0, + "step": 17603 + }, + { + "epoch": 1.933230836810894, + "grad_norm": 2.4132578372955322, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7141944169998169, + "num_tokens": 452363016.0, + "step": 17604 + }, + { + "epoch": 1.9333406545135077, + "grad_norm": 1.970209002494812, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7056330442428589, + "num_tokens": 452395794.0, + "step": 17605 + }, + { + "epoch": 1.9334504722161212, + "grad_norm": 2.342162609100342, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7063655853271484, + "num_tokens": 452419992.0, + "step": 17606 + }, + { + "epoch": 1.9335602899187347, + "grad_norm": 2.2335283756256104, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7015471458435059, + "num_tokens": 452448238.0, + "step": 17607 + }, + { + "epoch": 1.9336701076213485, + "grad_norm": 2.570781946182251, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7336722612380981, + "num_tokens": 452469222.0, + "step": 17608 + }, + { + "epoch": 1.9337799253239623, + "grad_norm": 2.1898915767669678, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7150145769119263, + "num_tokens": 452496445.0, + "step": 17609 + }, + { + "epoch": 1.933889743026576, + "grad_norm": 2.1295313835144043, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7091735601425171, + "num_tokens": 452526348.0, + "step": 17610 + }, + { + "epoch": 1.9339995607291895, + "grad_norm": 2.2540225982666016, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7111856341362, + "num_tokens": 452553480.0, + "step": 17611 + }, + { + "epoch": 1.934109378431803, + "grad_norm": 2.365163564682007, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7264077663421631, + "num_tokens": 452575841.0, + "step": 17612 + }, + { + "epoch": 1.9342191961344168, + "grad_norm": 2.3207833766937256, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7307640910148621, + "num_tokens": 452597970.0, + "step": 17613 + }, + { + "epoch": 1.9343290138370306, + "grad_norm": 2.280237913131714, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.737032949924469, + "num_tokens": 452621126.0, + "step": 17614 + }, + { + "epoch": 1.9344388315396444, + "grad_norm": 2.165057897567749, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7492110133171082, + "num_tokens": 452646192.0, + "step": 17615 + }, + { + "epoch": 1.934548649242258, + "grad_norm": 2.390681266784668, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7403824329376221, + "num_tokens": 452668174.0, + "step": 17616 + }, + { + "epoch": 1.9346584669448714, + "grad_norm": 2.1058924198150635, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7107535600662231, + "num_tokens": 452696227.0, + "step": 17617 + }, + { + "epoch": 1.9347682846474852, + "grad_norm": 2.24613618850708, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7126856446266174, + "num_tokens": 452723145.0, + "step": 17618 + }, + { + "epoch": 1.934878102350099, + "grad_norm": 2.212306499481201, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7181015610694885, + "num_tokens": 452749896.0, + "step": 17619 + }, + { + "epoch": 1.9349879200527125, + "grad_norm": 2.481534957885742, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7328471541404724, + "num_tokens": 452770889.0, + "step": 17620 + }, + { + "epoch": 1.935097737755326, + "grad_norm": 2.299133062362671, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7099766135215759, + "num_tokens": 452797462.0, + "step": 17621 + }, + { + "epoch": 1.9352075554579398, + "grad_norm": 2.146939277648926, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.703106164932251, + "num_tokens": 452825873.0, + "step": 17622 + }, + { + "epoch": 1.9353173731605535, + "grad_norm": 2.4578781127929688, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7215659618377686, + "num_tokens": 452846530.0, + "step": 17623 + }, + { + "epoch": 1.9354271908631673, + "grad_norm": 2.589810609817505, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7321324944496155, + "num_tokens": 452866313.0, + "step": 17624 + }, + { + "epoch": 1.9355370085657808, + "grad_norm": 2.1648740768432617, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7112546563148499, + "num_tokens": 452893379.0, + "step": 17625 + }, + { + "epoch": 1.9356468262683943, + "grad_norm": 2.2750484943389893, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7352502346038818, + "num_tokens": 452917554.0, + "step": 17626 + }, + { + "epoch": 1.935756643971008, + "grad_norm": 2.268995761871338, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.698167085647583, + "num_tokens": 452943935.0, + "step": 17627 + }, + { + "epoch": 1.9358664616736219, + "grad_norm": 2.137268304824829, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7138581275939941, + "num_tokens": 452970908.0, + "step": 17628 + }, + { + "epoch": 1.9359762793762354, + "grad_norm": 2.1572189331054688, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7285735607147217, + "num_tokens": 452996556.0, + "step": 17629 + }, + { + "epoch": 1.9360860970788492, + "grad_norm": 2.344810724258423, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7328572869300842, + "num_tokens": 453021299.0, + "step": 17630 + }, + { + "epoch": 1.9361959147814627, + "grad_norm": 2.007707357406616, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6996847987174988, + "num_tokens": 453053199.0, + "step": 17631 + }, + { + "epoch": 1.9363057324840764, + "grad_norm": 2.476036310195923, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7111415266990662, + "num_tokens": 453076759.0, + "step": 17632 + }, + { + "epoch": 1.9364155501866902, + "grad_norm": 2.207331895828247, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.753501832485199, + "num_tokens": 453099484.0, + "step": 17633 + }, + { + "epoch": 1.9365253678893037, + "grad_norm": 2.011727809906006, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.724179744720459, + "num_tokens": 453132285.0, + "step": 17634 + }, + { + "epoch": 1.9366351855919173, + "grad_norm": 1.922492504119873, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7131187319755554, + "num_tokens": 453164026.0, + "step": 17635 + }, + { + "epoch": 1.936745003294531, + "grad_norm": 2.099025249481201, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.717820405960083, + "num_tokens": 453190425.0, + "step": 17636 + }, + { + "epoch": 1.9368548209971448, + "grad_norm": 2.3457467555999756, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7173699140548706, + "num_tokens": 453212739.0, + "step": 17637 + }, + { + "epoch": 1.9369646386997585, + "grad_norm": 2.566779851913452, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7473841905593872, + "num_tokens": 453231792.0, + "step": 17638 + }, + { + "epoch": 1.937074456402372, + "grad_norm": 2.041067123413086, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7390506863594055, + "num_tokens": 453257783.0, + "step": 17639 + }, + { + "epoch": 1.9371842741049856, + "grad_norm": 2.3261938095092773, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7054155468940735, + "num_tokens": 453283621.0, + "step": 17640 + }, + { + "epoch": 1.9372940918075994, + "grad_norm": 2.267491340637207, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6975946426391602, + "num_tokens": 453311688.0, + "step": 17641 + }, + { + "epoch": 1.9374039095102131, + "grad_norm": 2.1371994018554688, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7117407321929932, + "num_tokens": 453338477.0, + "step": 17642 + }, + { + "epoch": 1.9375137272128267, + "grad_norm": 2.327606439590454, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7014800310134888, + "num_tokens": 453363008.0, + "step": 17643 + }, + { + "epoch": 1.9376235449154404, + "grad_norm": 2.162261962890625, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7145622968673706, + "num_tokens": 453390415.0, + "step": 17644 + }, + { + "epoch": 1.937733362618054, + "grad_norm": 2.2287559509277344, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7310288548469543, + "num_tokens": 453415021.0, + "step": 17645 + }, + { + "epoch": 1.9378431803206677, + "grad_norm": 1.8949297666549683, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7141109704971313, + "num_tokens": 453448127.0, + "step": 17646 + }, + { + "epoch": 1.9379529980232815, + "grad_norm": 2.182673692703247, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7314624786376953, + "num_tokens": 453474977.0, + "step": 17647 + }, + { + "epoch": 1.938062815725895, + "grad_norm": 2.1200692653656006, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7223094701766968, + "num_tokens": 453503069.0, + "step": 17648 + }, + { + "epoch": 1.9381726334285085, + "grad_norm": 2.3059022426605225, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.721327543258667, + "num_tokens": 453526375.0, + "step": 17649 + }, + { + "epoch": 1.9382824511311223, + "grad_norm": 2.268932580947876, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7359033226966858, + "num_tokens": 453550422.0, + "step": 17650 + }, + { + "epoch": 1.938392268833736, + "grad_norm": 2.304445505142212, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.721829354763031, + "num_tokens": 453572788.0, + "step": 17651 + }, + { + "epoch": 1.9385020865363498, + "grad_norm": 2.428401231765747, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7246653437614441, + "num_tokens": 453596177.0, + "step": 17652 + }, + { + "epoch": 1.9386119042389633, + "grad_norm": 2.3838629722595215, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.715488612651825, + "num_tokens": 453619322.0, + "step": 17653 + }, + { + "epoch": 1.9387217219415769, + "grad_norm": 2.248016834259033, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7338649034500122, + "num_tokens": 453645062.0, + "step": 17654 + }, + { + "epoch": 1.9388315396441906, + "grad_norm": 2.153513193130493, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7065350413322449, + "num_tokens": 453673561.0, + "step": 17655 + }, + { + "epoch": 1.9389413573468044, + "grad_norm": 2.586405038833618, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7410849332809448, + "num_tokens": 453693851.0, + "step": 17656 + }, + { + "epoch": 1.939051175049418, + "grad_norm": 2.3032357692718506, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7028840780258179, + "num_tokens": 453718423.0, + "step": 17657 + }, + { + "epoch": 1.9391609927520315, + "grad_norm": 2.4978885650634766, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7104144096374512, + "num_tokens": 453739767.0, + "step": 17658 + }, + { + "epoch": 1.9392708104546452, + "grad_norm": 1.8276939392089844, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7231154441833496, + "num_tokens": 453771730.0, + "step": 17659 + }, + { + "epoch": 1.939380628157259, + "grad_norm": 2.0709075927734375, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7031822204589844, + "num_tokens": 453802302.0, + "step": 17660 + }, + { + "epoch": 1.9394904458598727, + "grad_norm": 2.3007028102874756, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7420197129249573, + "num_tokens": 453825037.0, + "step": 17661 + }, + { + "epoch": 1.9396002635624863, + "grad_norm": 2.4479808807373047, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7238848209381104, + "num_tokens": 453846376.0, + "step": 17662 + }, + { + "epoch": 1.9397100812650998, + "grad_norm": 2.485790729522705, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7615822553634644, + "num_tokens": 453866461.0, + "step": 17663 + }, + { + "epoch": 1.9398198989677136, + "grad_norm": 2.4302315711975098, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7566490769386292, + "num_tokens": 453885820.0, + "step": 17664 + }, + { + "epoch": 1.9399297166703273, + "grad_norm": 2.3295419216156006, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7044090032577515, + "num_tokens": 453909830.0, + "step": 17665 + }, + { + "epoch": 1.940039534372941, + "grad_norm": 2.077585458755493, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7387571334838867, + "num_tokens": 453938347.0, + "step": 17666 + }, + { + "epoch": 1.9401493520755546, + "grad_norm": 2.1746137142181396, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7123806476593018, + "num_tokens": 453965899.0, + "step": 17667 + }, + { + "epoch": 1.9402591697781681, + "grad_norm": 2.45575213432312, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.722501814365387, + "num_tokens": 453988343.0, + "step": 17668 + }, + { + "epoch": 1.940368987480782, + "grad_norm": 2.4826455116271973, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7431613206863403, + "num_tokens": 454010522.0, + "step": 17669 + }, + { + "epoch": 1.9404788051833957, + "grad_norm": 2.5983388423919678, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7245882749557495, + "num_tokens": 454030753.0, + "step": 17670 + }, + { + "epoch": 1.9405886228860092, + "grad_norm": 2.1590397357940674, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7052014470100403, + "num_tokens": 454057578.0, + "step": 17671 + }, + { + "epoch": 1.9406984405886227, + "grad_norm": 2.308985948562622, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7193624377250671, + "num_tokens": 454081574.0, + "step": 17672 + }, + { + "epoch": 1.9408082582912365, + "grad_norm": 2.171384572982788, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7035433650016785, + "num_tokens": 454107719.0, + "step": 17673 + }, + { + "epoch": 1.9409180759938502, + "grad_norm": 2.0024595260620117, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7053813934326172, + "num_tokens": 454139417.0, + "step": 17674 + }, + { + "epoch": 1.941027893696464, + "grad_norm": 2.1190245151519775, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.704122006893158, + "num_tokens": 454167608.0, + "step": 17675 + }, + { + "epoch": 1.9411377113990775, + "grad_norm": 2.3710644245147705, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7284226417541504, + "num_tokens": 454189244.0, + "step": 17676 + }, + { + "epoch": 1.941247529101691, + "grad_norm": 2.1868948936462402, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7245941162109375, + "num_tokens": 454215798.0, + "step": 17677 + }, + { + "epoch": 1.9413573468043048, + "grad_norm": 2.1332879066467285, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6927155256271362, + "num_tokens": 454245279.0, + "step": 17678 + }, + { + "epoch": 1.9414671645069186, + "grad_norm": 2.4390125274658203, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7557141184806824, + "num_tokens": 454264395.0, + "step": 17679 + }, + { + "epoch": 1.9415769822095323, + "grad_norm": 2.1264424324035645, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7261419296264648, + "num_tokens": 454290964.0, + "step": 17680 + }, + { + "epoch": 1.9416867999121459, + "grad_norm": 2.4420816898345947, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7245211601257324, + "num_tokens": 454312637.0, + "step": 17681 + }, + { + "epoch": 1.9417966176147594, + "grad_norm": 2.4068174362182617, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7322025895118713, + "num_tokens": 454333232.0, + "step": 17682 + }, + { + "epoch": 1.9419064353173732, + "grad_norm": 2.2353270053863525, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7165252566337585, + "num_tokens": 454362986.0, + "step": 17683 + }, + { + "epoch": 1.942016253019987, + "grad_norm": 2.360690116882324, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7336217164993286, + "num_tokens": 454385757.0, + "step": 17684 + }, + { + "epoch": 1.9421260707226005, + "grad_norm": 2.1916961669921875, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.751444935798645, + "num_tokens": 454410769.0, + "step": 17685 + }, + { + "epoch": 1.942235888425214, + "grad_norm": 2.218289852142334, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7316122055053711, + "num_tokens": 454439028.0, + "step": 17686 + }, + { + "epoch": 1.9423457061278278, + "grad_norm": 2.394275426864624, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7244454622268677, + "num_tokens": 454463027.0, + "step": 17687 + }, + { + "epoch": 1.9424555238304415, + "grad_norm": 2.0292725563049316, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7092509269714355, + "num_tokens": 454492596.0, + "step": 17688 + }, + { + "epoch": 1.9425653415330553, + "grad_norm": 2.6275482177734375, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7296733856201172, + "num_tokens": 454512182.0, + "step": 17689 + }, + { + "epoch": 1.9426751592356688, + "grad_norm": 2.0657453536987305, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.724183201789856, + "num_tokens": 454542181.0, + "step": 17690 + }, + { + "epoch": 1.9427849769382823, + "grad_norm": 2.397535562515259, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7342371344566345, + "num_tokens": 454564765.0, + "step": 17691 + }, + { + "epoch": 1.942894794640896, + "grad_norm": 2.0312325954437256, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7216591238975525, + "num_tokens": 454596141.0, + "step": 17692 + }, + { + "epoch": 1.9430046123435099, + "grad_norm": 2.0850846767425537, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7100274562835693, + "num_tokens": 454625425.0, + "step": 17693 + }, + { + "epoch": 1.9431144300461234, + "grad_norm": 1.9097261428833008, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7277713418006897, + "num_tokens": 454657324.0, + "step": 17694 + }, + { + "epoch": 1.9432242477487371, + "grad_norm": 2.366962194442749, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7317195534706116, + "num_tokens": 454678309.0, + "step": 17695 + }, + { + "epoch": 1.9433340654513507, + "grad_norm": 2.7391788959503174, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7203402519226074, + "num_tokens": 454696449.0, + "step": 17696 + }, + { + "epoch": 1.9434438831539644, + "grad_norm": 2.3385825157165527, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7230993509292603, + "num_tokens": 454720826.0, + "step": 17697 + }, + { + "epoch": 1.9435537008565782, + "grad_norm": 2.2924816608428955, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7133733034133911, + "num_tokens": 454746855.0, + "step": 17698 + }, + { + "epoch": 1.9436635185591917, + "grad_norm": 2.169294595718384, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7426004409790039, + "num_tokens": 454770588.0, + "step": 17699 + }, + { + "epoch": 1.9437733362618053, + "grad_norm": 2.070631980895996, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7038794755935669, + "num_tokens": 454800817.0, + "step": 17700 + }, + { + "epoch": 1.943883153964419, + "grad_norm": 2.32973575592041, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7198482751846313, + "num_tokens": 454823494.0, + "step": 17701 + }, + { + "epoch": 1.9439929716670328, + "grad_norm": 2.0721824169158936, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7512180805206299, + "num_tokens": 454849885.0, + "step": 17702 + }, + { + "epoch": 1.9441027893696465, + "grad_norm": 2.101022958755493, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7094641327857971, + "num_tokens": 454878355.0, + "step": 17703 + }, + { + "epoch": 1.94421260707226, + "grad_norm": 2.2828116416931152, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.708715558052063, + "num_tokens": 454905041.0, + "step": 17704 + }, + { + "epoch": 1.9443224247748736, + "grad_norm": 2.1736032962799072, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7243458032608032, + "num_tokens": 454930714.0, + "step": 17705 + }, + { + "epoch": 1.9444322424774874, + "grad_norm": 2.579801082611084, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7325031757354736, + "num_tokens": 454949941.0, + "step": 17706 + }, + { + "epoch": 1.9445420601801011, + "grad_norm": 2.6144630908966064, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7326192855834961, + "num_tokens": 454970240.0, + "step": 17707 + }, + { + "epoch": 1.9446518778827147, + "grad_norm": 2.1369383335113525, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7022671103477478, + "num_tokens": 455000466.0, + "step": 17708 + }, + { + "epoch": 1.9447616955853284, + "grad_norm": 2.2705609798431396, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7362795472145081, + "num_tokens": 455024439.0, + "step": 17709 + }, + { + "epoch": 1.944871513287942, + "grad_norm": 2.1877286434173584, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7105790972709656, + "num_tokens": 455050743.0, + "step": 17710 + }, + { + "epoch": 1.9449813309905557, + "grad_norm": 2.264772653579712, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7200360298156738, + "num_tokens": 455075405.0, + "step": 17711 + }, + { + "epoch": 1.9450911486931695, + "grad_norm": 2.2940638065338135, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7195345759391785, + "num_tokens": 455100190.0, + "step": 17712 + }, + { + "epoch": 1.945200966395783, + "grad_norm": 2.202064037322998, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7061023116111755, + "num_tokens": 455127328.0, + "step": 17713 + }, + { + "epoch": 1.9453107840983965, + "grad_norm": 1.9206275939941406, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7152547836303711, + "num_tokens": 455158897.0, + "step": 17714 + }, + { + "epoch": 1.9454206018010103, + "grad_norm": 2.28524112701416, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7136965990066528, + "num_tokens": 455182376.0, + "step": 17715 + }, + { + "epoch": 1.945530419503624, + "grad_norm": 2.4339444637298584, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7237423062324524, + "num_tokens": 455204924.0, + "step": 17716 + }, + { + "epoch": 1.9456402372062378, + "grad_norm": 2.020080804824829, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7062749266624451, + "num_tokens": 455235855.0, + "step": 17717 + }, + { + "epoch": 1.9457500549088513, + "grad_norm": 2.5243337154388428, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7250746488571167, + "num_tokens": 455257064.0, + "step": 17718 + }, + { + "epoch": 1.9458598726114649, + "grad_norm": 2.0754992961883545, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7100844979286194, + "num_tokens": 455286207.0, + "step": 17719 + }, + { + "epoch": 1.9459696903140786, + "grad_norm": 2.59236216545105, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7258695363998413, + "num_tokens": 455305849.0, + "step": 17720 + }, + { + "epoch": 1.9460795080166924, + "grad_norm": 2.2811481952667236, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7082067131996155, + "num_tokens": 455332140.0, + "step": 17721 + }, + { + "epoch": 1.946189325719306, + "grad_norm": 1.9844237565994263, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7201581001281738, + "num_tokens": 455362646.0, + "step": 17722 + }, + { + "epoch": 1.9462991434219195, + "grad_norm": 2.7233645915985107, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7339655160903931, + "num_tokens": 455380808.0, + "step": 17723 + }, + { + "epoch": 1.9464089611245332, + "grad_norm": 2.0643773078918457, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.742652952671051, + "num_tokens": 455407342.0, + "step": 17724 + }, + { + "epoch": 1.946518778827147, + "grad_norm": 2.4885332584381104, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7378345131874084, + "num_tokens": 455428589.0, + "step": 17725 + }, + { + "epoch": 1.9466285965297607, + "grad_norm": 2.386011838912964, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7266629934310913, + "num_tokens": 455451583.0, + "step": 17726 + }, + { + "epoch": 1.9467384142323743, + "grad_norm": 2.140043258666992, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7434971332550049, + "num_tokens": 455476453.0, + "step": 17727 + }, + { + "epoch": 1.9468482319349878, + "grad_norm": 2.2527804374694824, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7208739519119263, + "num_tokens": 455501884.0, + "step": 17728 + }, + { + "epoch": 1.9469580496376016, + "grad_norm": 2.181034564971924, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.734153151512146, + "num_tokens": 455527490.0, + "step": 17729 + }, + { + "epoch": 1.9470678673402153, + "grad_norm": 2.8630752563476562, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7341467142105103, + "num_tokens": 455544937.0, + "step": 17730 + }, + { + "epoch": 1.947177685042829, + "grad_norm": 2.4042506217956543, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7066890001296997, + "num_tokens": 455570186.0, + "step": 17731 + }, + { + "epoch": 1.9472875027454426, + "grad_norm": 2.5061686038970947, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7292200326919556, + "num_tokens": 455591853.0, + "step": 17732 + }, + { + "epoch": 1.9473973204480561, + "grad_norm": 2.287013053894043, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7242545485496521, + "num_tokens": 455617232.0, + "step": 17733 + }, + { + "epoch": 1.94750713815067, + "grad_norm": 2.3847341537475586, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7688043117523193, + "num_tokens": 455638562.0, + "step": 17734 + }, + { + "epoch": 1.9476169558532836, + "grad_norm": 2.577191114425659, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7604628801345825, + "num_tokens": 455656527.0, + "step": 17735 + }, + { + "epoch": 1.9477267735558972, + "grad_norm": 2.2350122928619385, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7130439877510071, + "num_tokens": 455681331.0, + "step": 17736 + }, + { + "epoch": 1.9478365912585107, + "grad_norm": 2.563157558441162, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7236495614051819, + "num_tokens": 455702739.0, + "step": 17737 + }, + { + "epoch": 1.9479464089611245, + "grad_norm": 1.9725205898284912, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7052398324012756, + "num_tokens": 455733038.0, + "step": 17738 + }, + { + "epoch": 1.9480562266637382, + "grad_norm": 2.2928273677825928, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7242339849472046, + "num_tokens": 455758273.0, + "step": 17739 + }, + { + "epoch": 1.948166044366352, + "grad_norm": 2.384337902069092, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7116335034370422, + "num_tokens": 455781566.0, + "step": 17740 + }, + { + "epoch": 1.9482758620689655, + "grad_norm": 2.1603565216064453, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7336645722389221, + "num_tokens": 455806355.0, + "step": 17741 + }, + { + "epoch": 1.948385679771579, + "grad_norm": 2.185673952102661, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7173769474029541, + "num_tokens": 455832243.0, + "step": 17742 + }, + { + "epoch": 1.9484954974741928, + "grad_norm": 2.404484272003174, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7437372803688049, + "num_tokens": 455854843.0, + "step": 17743 + }, + { + "epoch": 1.9486053151768066, + "grad_norm": 2.2017123699188232, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.734634518623352, + "num_tokens": 455882790.0, + "step": 17744 + }, + { + "epoch": 1.94871513287942, + "grad_norm": 2.18578839302063, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7364992499351501, + "num_tokens": 455906894.0, + "step": 17745 + }, + { + "epoch": 1.9488249505820339, + "grad_norm": 2.1293489933013916, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7280418872833252, + "num_tokens": 455934183.0, + "step": 17746 + }, + { + "epoch": 1.9489347682846474, + "grad_norm": 2.10305118560791, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7294555306434631, + "num_tokens": 455962147.0, + "step": 17747 + }, + { + "epoch": 1.9490445859872612, + "grad_norm": 2.324338436126709, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7232384085655212, + "num_tokens": 455985970.0, + "step": 17748 + }, + { + "epoch": 1.949154403689875, + "grad_norm": 2.4548020362854004, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7300685048103333, + "num_tokens": 456007706.0, + "step": 17749 + }, + { + "epoch": 1.9492642213924884, + "grad_norm": 1.912907361984253, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7376277446746826, + "num_tokens": 456040244.0, + "step": 17750 + }, + { + "epoch": 1.949374039095102, + "grad_norm": 2.2382664680480957, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7237725257873535, + "num_tokens": 456067643.0, + "step": 17751 + }, + { + "epoch": 1.9494838567977157, + "grad_norm": 2.558098316192627, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7450190782546997, + "num_tokens": 456087088.0, + "step": 17752 + }, + { + "epoch": 1.9495936745003295, + "grad_norm": 2.0937206745147705, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7068667411804199, + "num_tokens": 456117704.0, + "step": 17753 + }, + { + "epoch": 1.9497034922029433, + "grad_norm": 2.4410455226898193, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7410786747932434, + "num_tokens": 456137822.0, + "step": 17754 + }, + { + "epoch": 1.9498133099055568, + "grad_norm": 2.1726303100585938, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7153131365776062, + "num_tokens": 456162859.0, + "step": 17755 + }, + { + "epoch": 1.9499231276081703, + "grad_norm": 1.897357702255249, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7075427770614624, + "num_tokens": 456196828.0, + "step": 17756 + }, + { + "epoch": 1.950032945310784, + "grad_norm": 2.569516181945801, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7454887628555298, + "num_tokens": 456215736.0, + "step": 17757 + }, + { + "epoch": 1.9501427630133978, + "grad_norm": 2.042024850845337, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7252314686775208, + "num_tokens": 456245585.0, + "step": 17758 + }, + { + "epoch": 1.9502525807160114, + "grad_norm": 2.343069076538086, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7080380916595459, + "num_tokens": 456268420.0, + "step": 17759 + }, + { + "epoch": 1.9503623984186251, + "grad_norm": 2.2856414318084717, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6910280585289001, + "num_tokens": 456294571.0, + "step": 17760 + }, + { + "epoch": 1.9504722161212387, + "grad_norm": 2.468047618865967, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7357081174850464, + "num_tokens": 456316039.0, + "step": 17761 + }, + { + "epoch": 1.9505820338238524, + "grad_norm": 2.2453725337982178, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7053884863853455, + "num_tokens": 456340921.0, + "step": 17762 + }, + { + "epoch": 1.9506918515264662, + "grad_norm": 1.9435347318649292, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7077980041503906, + "num_tokens": 456373384.0, + "step": 17763 + }, + { + "epoch": 1.9508016692290797, + "grad_norm": 2.1899125576019287, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7185068130493164, + "num_tokens": 456401104.0, + "step": 17764 + }, + { + "epoch": 1.9509114869316933, + "grad_norm": 2.0698630809783936, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7094769477844238, + "num_tokens": 456429273.0, + "step": 17765 + }, + { + "epoch": 1.951021304634307, + "grad_norm": 2.214862108230591, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7415739893913269, + "num_tokens": 456454847.0, + "step": 17766 + }, + { + "epoch": 1.9511311223369208, + "grad_norm": 2.3669450283050537, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7402133345603943, + "num_tokens": 456477821.0, + "step": 17767 + }, + { + "epoch": 1.9512409400395345, + "grad_norm": 2.4059431552886963, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7062734365463257, + "num_tokens": 456502933.0, + "step": 17768 + }, + { + "epoch": 1.951350757742148, + "grad_norm": 2.4003148078918457, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7350742816925049, + "num_tokens": 456525817.0, + "step": 17769 + }, + { + "epoch": 1.9514605754447616, + "grad_norm": 2.18558406829834, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7186048030853271, + "num_tokens": 456553495.0, + "step": 17770 + }, + { + "epoch": 1.9515703931473753, + "grad_norm": 2.362748146057129, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7229516506195068, + "num_tokens": 456577663.0, + "step": 17771 + }, + { + "epoch": 1.951680210849989, + "grad_norm": 1.9829163551330566, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7251599431037903, + "num_tokens": 456610063.0, + "step": 17772 + }, + { + "epoch": 1.9517900285526026, + "grad_norm": 2.3904974460601807, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7587370872497559, + "num_tokens": 456630228.0, + "step": 17773 + }, + { + "epoch": 1.9518998462552162, + "grad_norm": 2.004215955734253, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7443989515304565, + "num_tokens": 456658599.0, + "step": 17774 + }, + { + "epoch": 1.95200966395783, + "grad_norm": 2.22296404838562, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7290807962417603, + "num_tokens": 456683813.0, + "step": 17775 + }, + { + "epoch": 1.9521194816604437, + "grad_norm": 1.9715570211410522, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6889523267745972, + "num_tokens": 456716058.0, + "step": 17776 + }, + { + "epoch": 1.9522292993630574, + "grad_norm": 2.8161845207214355, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7438864707946777, + "num_tokens": 456732917.0, + "step": 17777 + }, + { + "epoch": 1.952339117065671, + "grad_norm": 2.024369716644287, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.739478588104248, + "num_tokens": 456762701.0, + "step": 17778 + }, + { + "epoch": 1.9524489347682845, + "grad_norm": 2.083305835723877, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7225239276885986, + "num_tokens": 456790270.0, + "step": 17779 + }, + { + "epoch": 1.9525587524708983, + "grad_norm": 2.1675209999084473, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7240720987319946, + "num_tokens": 456816184.0, + "step": 17780 + }, + { + "epoch": 1.952668570173512, + "grad_norm": 2.542747735977173, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7254108786582947, + "num_tokens": 456837353.0, + "step": 17781 + }, + { + "epoch": 1.9527783878761258, + "grad_norm": 2.1798315048217773, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7368992567062378, + "num_tokens": 456863864.0, + "step": 17782 + }, + { + "epoch": 1.9528882055787393, + "grad_norm": 2.5714259147644043, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7465564608573914, + "num_tokens": 456884796.0, + "step": 17783 + }, + { + "epoch": 1.9529980232813529, + "grad_norm": 2.0486068725585938, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7143024802207947, + "num_tokens": 456916013.0, + "step": 17784 + }, + { + "epoch": 1.9531078409839666, + "grad_norm": 2.378176212310791, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7504527568817139, + "num_tokens": 456939087.0, + "step": 17785 + }, + { + "epoch": 1.9532176586865804, + "grad_norm": 2.5956153869628906, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7085355520248413, + "num_tokens": 456965293.0, + "step": 17786 + }, + { + "epoch": 1.953327476389194, + "grad_norm": 2.2243566513061523, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7413201928138733, + "num_tokens": 456991183.0, + "step": 17787 + }, + { + "epoch": 1.9534372940918074, + "grad_norm": 1.9946480989456177, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7270958423614502, + "num_tokens": 457022758.0, + "step": 17788 + }, + { + "epoch": 1.9535471117944212, + "grad_norm": 1.990399956703186, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7117791175842285, + "num_tokens": 457052964.0, + "step": 17789 + }, + { + "epoch": 1.953656929497035, + "grad_norm": 2.0374903678894043, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.704788088798523, + "num_tokens": 457082399.0, + "step": 17790 + }, + { + "epoch": 1.9537667471996487, + "grad_norm": 2.4554378986358643, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7404172420501709, + "num_tokens": 457104972.0, + "step": 17791 + }, + { + "epoch": 1.9538765649022622, + "grad_norm": 2.2093467712402344, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.6985456347465515, + "num_tokens": 457133557.0, + "step": 17792 + }, + { + "epoch": 1.9539863826048758, + "grad_norm": 2.163039207458496, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.721164882183075, + "num_tokens": 457161174.0, + "step": 17793 + }, + { + "epoch": 1.9540962003074895, + "grad_norm": 2.249263286590576, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7390694618225098, + "num_tokens": 457183032.0, + "step": 17794 + }, + { + "epoch": 1.9542060180101033, + "grad_norm": 2.3254270553588867, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7416924834251404, + "num_tokens": 457207308.0, + "step": 17795 + }, + { + "epoch": 1.954315835712717, + "grad_norm": 2.0089149475097656, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7228646874427795, + "num_tokens": 457238934.0, + "step": 17796 + }, + { + "epoch": 1.9544256534153306, + "grad_norm": 2.15781831741333, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7028248310089111, + "num_tokens": 457267295.0, + "step": 17797 + }, + { + "epoch": 1.9545354711179441, + "grad_norm": 2.4469664096832275, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.740807056427002, + "num_tokens": 457290254.0, + "step": 17798 + }, + { + "epoch": 1.9546452888205579, + "grad_norm": 2.277250289916992, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7271215915679932, + "num_tokens": 457315122.0, + "step": 17799 + }, + { + "epoch": 1.9547551065231716, + "grad_norm": 2.0626492500305176, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6910874843597412, + "num_tokens": 457345683.0, + "step": 17800 + }, + { + "epoch": 1.9548649242257852, + "grad_norm": 2.3317136764526367, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7118121981620789, + "num_tokens": 457372672.0, + "step": 17801 + }, + { + "epoch": 1.9549747419283987, + "grad_norm": 2.2042534351348877, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7067217826843262, + "num_tokens": 457400855.0, + "step": 17802 + }, + { + "epoch": 1.9550845596310125, + "grad_norm": 2.344763994216919, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7385902404785156, + "num_tokens": 457424487.0, + "step": 17803 + }, + { + "epoch": 1.9551943773336262, + "grad_norm": 2.094653844833374, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7114474773406982, + "num_tokens": 457453980.0, + "step": 17804 + }, + { + "epoch": 1.95530419503624, + "grad_norm": 2.2764062881469727, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7159119844436646, + "num_tokens": 457481150.0, + "step": 17805 + }, + { + "epoch": 1.9554140127388535, + "grad_norm": 2.262723922729492, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7002648115158081, + "num_tokens": 457507959.0, + "step": 17806 + }, + { + "epoch": 1.955523830441467, + "grad_norm": 2.1198837757110596, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.697472095489502, + "num_tokens": 457538236.0, + "step": 17807 + }, + { + "epoch": 1.9556336481440808, + "grad_norm": 2.408704996109009, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7251460552215576, + "num_tokens": 457561130.0, + "step": 17808 + }, + { + "epoch": 1.9557434658466946, + "grad_norm": 2.1998589038848877, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7292025089263916, + "num_tokens": 457587471.0, + "step": 17809 + }, + { + "epoch": 1.955853283549308, + "grad_norm": 2.19270658493042, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7065167427062988, + "num_tokens": 457615010.0, + "step": 17810 + }, + { + "epoch": 1.9559631012519219, + "grad_norm": 2.5980682373046875, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7260324954986572, + "num_tokens": 457634110.0, + "step": 17811 + }, + { + "epoch": 1.9560729189545354, + "grad_norm": 2.10768461227417, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7096550464630127, + "num_tokens": 457662416.0, + "step": 17812 + }, + { + "epoch": 1.9561827366571491, + "grad_norm": 2.2774837017059326, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6958699822425842, + "num_tokens": 457690476.0, + "step": 17813 + }, + { + "epoch": 1.956292554359763, + "grad_norm": 2.1691274642944336, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7127556800842285, + "num_tokens": 457716365.0, + "step": 17814 + }, + { + "epoch": 1.9564023720623764, + "grad_norm": 2.363715410232544, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7238158583641052, + "num_tokens": 457739092.0, + "step": 17815 + }, + { + "epoch": 1.95651218976499, + "grad_norm": 2.2701587677001953, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6980152130126953, + "num_tokens": 457764874.0, + "step": 17816 + }, + { + "epoch": 1.9566220074676037, + "grad_norm": 2.187497138977051, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7226781249046326, + "num_tokens": 457790201.0, + "step": 17817 + }, + { + "epoch": 1.9567318251702175, + "grad_norm": 2.3570783138275146, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7314233183860779, + "num_tokens": 457813916.0, + "step": 17818 + }, + { + "epoch": 1.9568416428728312, + "grad_norm": 2.2671711444854736, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7228196859359741, + "num_tokens": 457838569.0, + "step": 17819 + }, + { + "epoch": 1.9569514605754448, + "grad_norm": 2.047914743423462, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7020970582962036, + "num_tokens": 457872153.0, + "step": 17820 + }, + { + "epoch": 1.9570612782780583, + "grad_norm": 2.5661206245422363, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7527711987495422, + "num_tokens": 457891665.0, + "step": 17821 + }, + { + "epoch": 1.957171095980672, + "grad_norm": 2.30350661277771, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7233669757843018, + "num_tokens": 457914257.0, + "step": 17822 + }, + { + "epoch": 1.9572809136832858, + "grad_norm": 2.5762178897857666, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7484040856361389, + "num_tokens": 457933552.0, + "step": 17823 + }, + { + "epoch": 1.9573907313858994, + "grad_norm": 2.4304184913635254, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.744038999080658, + "num_tokens": 457954846.0, + "step": 17824 + }, + { + "epoch": 1.9575005490885131, + "grad_norm": 2.4587531089782715, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7208050489425659, + "num_tokens": 457975754.0, + "step": 17825 + }, + { + "epoch": 1.9576103667911267, + "grad_norm": 2.501469135284424, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7464779615402222, + "num_tokens": 457996246.0, + "step": 17826 + }, + { + "epoch": 1.9577201844937404, + "grad_norm": 2.1373796463012695, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7357661724090576, + "num_tokens": 458022248.0, + "step": 17827 + }, + { + "epoch": 1.9578300021963542, + "grad_norm": 2.4754083156585693, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.6995784044265747, + "num_tokens": 458044899.0, + "step": 17828 + }, + { + "epoch": 1.9579398198989677, + "grad_norm": 2.0759763717651367, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7072631120681763, + "num_tokens": 458073376.0, + "step": 17829 + }, + { + "epoch": 1.9580496376015812, + "grad_norm": 2.3660786151885986, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7364376783370972, + "num_tokens": 458095965.0, + "step": 17830 + }, + { + "epoch": 1.958159455304195, + "grad_norm": 2.30981707572937, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7037923336029053, + "num_tokens": 458122072.0, + "step": 17831 + }, + { + "epoch": 1.9582692730068088, + "grad_norm": 2.23298716545105, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7362407445907593, + "num_tokens": 458145595.0, + "step": 17832 + }, + { + "epoch": 1.9583790907094225, + "grad_norm": 1.9308830499649048, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7224213480949402, + "num_tokens": 458179600.0, + "step": 17833 + }, + { + "epoch": 1.958488908412036, + "grad_norm": 2.107419729232788, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7325747013092041, + "num_tokens": 458207497.0, + "step": 17834 + }, + { + "epoch": 1.9585987261146496, + "grad_norm": 2.117588520050049, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7174826860427856, + "num_tokens": 458234760.0, + "step": 17835 + }, + { + "epoch": 1.9587085438172633, + "grad_norm": 2.345043897628784, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.724445104598999, + "num_tokens": 458259247.0, + "step": 17836 + }, + { + "epoch": 1.958818361519877, + "grad_norm": 2.0600881576538086, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7129684686660767, + "num_tokens": 458289204.0, + "step": 17837 + }, + { + "epoch": 1.9589281792224906, + "grad_norm": 2.058288097381592, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7133548259735107, + "num_tokens": 458318763.0, + "step": 17838 + }, + { + "epoch": 1.9590379969251042, + "grad_norm": 2.2474570274353027, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7280194163322449, + "num_tokens": 458345420.0, + "step": 17839 + }, + { + "epoch": 1.959147814627718, + "grad_norm": 2.0814576148986816, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7268584966659546, + "num_tokens": 458371267.0, + "step": 17840 + }, + { + "epoch": 1.9592576323303317, + "grad_norm": 2.1192328929901123, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.74127197265625, + "num_tokens": 458399406.0, + "step": 17841 + }, + { + "epoch": 1.9593674500329454, + "grad_norm": 2.1566479206085205, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7097378969192505, + "num_tokens": 458428502.0, + "step": 17842 + }, + { + "epoch": 1.959477267735559, + "grad_norm": 2.2996108531951904, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7159149050712585, + "num_tokens": 458453774.0, + "step": 17843 + }, + { + "epoch": 1.9595870854381725, + "grad_norm": 2.227064847946167, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7108060121536255, + "num_tokens": 458480167.0, + "step": 17844 + }, + { + "epoch": 1.9596969031407863, + "grad_norm": 2.0864193439483643, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7084190845489502, + "num_tokens": 458508245.0, + "step": 17845 + }, + { + "epoch": 1.9598067208434, + "grad_norm": 2.3284707069396973, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7073034644126892, + "num_tokens": 458531825.0, + "step": 17846 + }, + { + "epoch": 1.9599165385460138, + "grad_norm": 2.6299548149108887, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7249977588653564, + "num_tokens": 458551063.0, + "step": 17847 + }, + { + "epoch": 1.9600263562486273, + "grad_norm": 2.3962504863739014, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7121301889419556, + "num_tokens": 458574508.0, + "step": 17848 + }, + { + "epoch": 1.9601361739512408, + "grad_norm": 2.310084581375122, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7302943468093872, + "num_tokens": 458598294.0, + "step": 17849 + }, + { + "epoch": 1.9602459916538546, + "grad_norm": 2.552865505218506, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7402967214584351, + "num_tokens": 458618322.0, + "step": 17850 + }, + { + "epoch": 1.9603558093564684, + "grad_norm": 2.2971551418304443, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7260355949401855, + "num_tokens": 458642618.0, + "step": 17851 + }, + { + "epoch": 1.960465627059082, + "grad_norm": 2.040503978729248, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6861430406570435, + "num_tokens": 458673143.0, + "step": 17852 + }, + { + "epoch": 1.9605754447616954, + "grad_norm": 2.42041015625, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7142986059188843, + "num_tokens": 458695331.0, + "step": 17853 + }, + { + "epoch": 1.9606852624643092, + "grad_norm": 2.2907814979553223, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7144845128059387, + "num_tokens": 458719745.0, + "step": 17854 + }, + { + "epoch": 1.960795080166923, + "grad_norm": 2.295681953430176, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7154470682144165, + "num_tokens": 458744952.0, + "step": 17855 + }, + { + "epoch": 1.9609048978695367, + "grad_norm": 2.2693328857421875, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7182608842849731, + "num_tokens": 458770213.0, + "step": 17856 + }, + { + "epoch": 1.9610147155721502, + "grad_norm": 2.0036206245422363, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7025604248046875, + "num_tokens": 458801692.0, + "step": 17857 + }, + { + "epoch": 1.9611245332747638, + "grad_norm": 2.201263189315796, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7408446669578552, + "num_tokens": 458828021.0, + "step": 17858 + }, + { + "epoch": 1.9612343509773775, + "grad_norm": 2.271765947341919, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7103630304336548, + "num_tokens": 458855393.0, + "step": 17859 + }, + { + "epoch": 1.9613441686799913, + "grad_norm": 2.584531545639038, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7215113639831543, + "num_tokens": 458878071.0, + "step": 17860 + }, + { + "epoch": 1.961453986382605, + "grad_norm": 2.1857950687408447, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7219173908233643, + "num_tokens": 458905067.0, + "step": 17861 + }, + { + "epoch": 1.9615638040852186, + "grad_norm": 2.1321463584899902, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7130478620529175, + "num_tokens": 458933384.0, + "step": 17862 + }, + { + "epoch": 1.9616736217878321, + "grad_norm": 2.2442073822021484, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7179692983627319, + "num_tokens": 458959582.0, + "step": 17863 + }, + { + "epoch": 1.9617834394904459, + "grad_norm": 2.1379857063293457, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7035080194473267, + "num_tokens": 458988564.0, + "step": 17864 + }, + { + "epoch": 1.9618932571930596, + "grad_norm": 2.4196832180023193, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7071214318275452, + "num_tokens": 459012410.0, + "step": 17865 + }, + { + "epoch": 1.9620030748956732, + "grad_norm": 2.2752392292022705, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7035750150680542, + "num_tokens": 459038868.0, + "step": 17866 + }, + { + "epoch": 1.9621128925982867, + "grad_norm": 2.0536766052246094, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.723422646522522, + "num_tokens": 459065932.0, + "step": 17867 + }, + { + "epoch": 1.9622227103009005, + "grad_norm": 2.1533560752868652, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7023031711578369, + "num_tokens": 459092941.0, + "step": 17868 + }, + { + "epoch": 1.9623325280035142, + "grad_norm": 2.5492827892303467, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7453863620758057, + "num_tokens": 459112898.0, + "step": 17869 + }, + { + "epoch": 1.962442345706128, + "grad_norm": 1.8882395029067993, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7170710563659668, + "num_tokens": 459145903.0, + "step": 17870 + }, + { + "epoch": 1.9625521634087415, + "grad_norm": 2.3517720699310303, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7248374223709106, + "num_tokens": 459168791.0, + "step": 17871 + }, + { + "epoch": 1.962661981111355, + "grad_norm": 2.1207685470581055, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.731471061706543, + "num_tokens": 459196157.0, + "step": 17872 + }, + { + "epoch": 1.9627717988139688, + "grad_norm": 2.1273975372314453, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7275964021682739, + "num_tokens": 459225326.0, + "step": 17873 + }, + { + "epoch": 1.9628816165165826, + "grad_norm": 2.255657911300659, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7052795886993408, + "num_tokens": 459251827.0, + "step": 17874 + }, + { + "epoch": 1.962991434219196, + "grad_norm": 2.3420212268829346, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7155330777168274, + "num_tokens": 459277607.0, + "step": 17875 + }, + { + "epoch": 1.9631012519218098, + "grad_norm": 2.8900513648986816, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7289321422576904, + "num_tokens": 459295944.0, + "step": 17876 + }, + { + "epoch": 1.9632110696244234, + "grad_norm": 2.5739758014678955, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7510679960250854, + "num_tokens": 459317082.0, + "step": 17877 + }, + { + "epoch": 1.9633208873270371, + "grad_norm": 2.4880356788635254, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7088000774383545, + "num_tokens": 459341413.0, + "step": 17878 + }, + { + "epoch": 1.963430705029651, + "grad_norm": 2.666886568069458, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7312653660774231, + "num_tokens": 459359844.0, + "step": 17879 + }, + { + "epoch": 1.9635405227322644, + "grad_norm": 2.4194464683532715, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7206042408943176, + "num_tokens": 459382762.0, + "step": 17880 + }, + { + "epoch": 1.963650340434878, + "grad_norm": 2.3528811931610107, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7366880178451538, + "num_tokens": 459406858.0, + "step": 17881 + }, + { + "epoch": 1.9637601581374917, + "grad_norm": 2.1099014282226562, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7238399982452393, + "num_tokens": 459435689.0, + "step": 17882 + }, + { + "epoch": 1.9638699758401055, + "grad_norm": 2.086430788040161, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.726401686668396, + "num_tokens": 459463412.0, + "step": 17883 + }, + { + "epoch": 1.9639797935427192, + "grad_norm": 2.1835150718688965, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7321068048477173, + "num_tokens": 459488325.0, + "step": 17884 + }, + { + "epoch": 1.9640896112453328, + "grad_norm": 2.1650190353393555, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6928157806396484, + "num_tokens": 459516638.0, + "step": 17885 + }, + { + "epoch": 1.9641994289479463, + "grad_norm": 2.5568857192993164, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7263974547386169, + "num_tokens": 459537882.0, + "step": 17886 + }, + { + "epoch": 1.96430924665056, + "grad_norm": 2.5296719074249268, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7127589583396912, + "num_tokens": 459559048.0, + "step": 17887 + }, + { + "epoch": 1.9644190643531738, + "grad_norm": 2.4907045364379883, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7500379085540771, + "num_tokens": 459579153.0, + "step": 17888 + }, + { + "epoch": 1.9645288820557874, + "grad_norm": 2.2294793128967285, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7146536707878113, + "num_tokens": 459603821.0, + "step": 17889 + }, + { + "epoch": 1.964638699758401, + "grad_norm": 2.475010633468628, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7270798683166504, + "num_tokens": 459624526.0, + "step": 17890 + }, + { + "epoch": 1.9647485174610146, + "grad_norm": 2.395963668823242, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7324261665344238, + "num_tokens": 459649013.0, + "step": 17891 + }, + { + "epoch": 1.9648583351636284, + "grad_norm": 2.3788697719573975, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7225288152694702, + "num_tokens": 459672229.0, + "step": 17892 + }, + { + "epoch": 1.9649681528662422, + "grad_norm": 2.3423056602478027, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7087263464927673, + "num_tokens": 459696359.0, + "step": 17893 + }, + { + "epoch": 1.9650779705688557, + "grad_norm": 2.2705254554748535, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7127360105514526, + "num_tokens": 459720173.0, + "step": 17894 + }, + { + "epoch": 1.9651877882714692, + "grad_norm": 2.1916511058807373, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.717605471611023, + "num_tokens": 459745185.0, + "step": 17895 + }, + { + "epoch": 1.965297605974083, + "grad_norm": 2.4224703311920166, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7292804718017578, + "num_tokens": 459768532.0, + "step": 17896 + }, + { + "epoch": 1.9654074236766967, + "grad_norm": 2.3518810272216797, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7364404797554016, + "num_tokens": 459792973.0, + "step": 17897 + }, + { + "epoch": 1.9655172413793105, + "grad_norm": 2.178680419921875, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7167155742645264, + "num_tokens": 459819099.0, + "step": 17898 + }, + { + "epoch": 1.965627059081924, + "grad_norm": 2.4150595664978027, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7473328709602356, + "num_tokens": 459840652.0, + "step": 17899 + }, + { + "epoch": 1.9657368767845376, + "grad_norm": 2.4960315227508545, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7057230472564697, + "num_tokens": 459862814.0, + "step": 17900 + }, + { + "epoch": 1.9658466944871513, + "grad_norm": 2.188030481338501, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7248163223266602, + "num_tokens": 459889926.0, + "step": 17901 + }, + { + "epoch": 1.965956512189765, + "grad_norm": 2.206803321838379, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.735085666179657, + "num_tokens": 459915494.0, + "step": 17902 + }, + { + "epoch": 1.9660663298923786, + "grad_norm": 2.253121852874756, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7241570353507996, + "num_tokens": 459940569.0, + "step": 17903 + }, + { + "epoch": 1.9661761475949922, + "grad_norm": 2.58675479888916, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7136499881744385, + "num_tokens": 459961970.0, + "step": 17904 + }, + { + "epoch": 1.966285965297606, + "grad_norm": 1.8538905382156372, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6897594928741455, + "num_tokens": 459999233.0, + "step": 17905 + }, + { + "epoch": 1.9663957830002197, + "grad_norm": 2.222842216491699, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7576613426208496, + "num_tokens": 460023419.0, + "step": 17906 + }, + { + "epoch": 1.9665056007028334, + "grad_norm": 2.311079263687134, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7419040203094482, + "num_tokens": 460046453.0, + "step": 17907 + }, + { + "epoch": 1.966615418405447, + "grad_norm": 2.266090154647827, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7196621894836426, + "num_tokens": 460071168.0, + "step": 17908 + }, + { + "epoch": 1.9667252361080605, + "grad_norm": 2.06396484375, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7143278121948242, + "num_tokens": 460098974.0, + "step": 17909 + }, + { + "epoch": 1.9668350538106742, + "grad_norm": 1.946154236793518, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6865195631980896, + "num_tokens": 460134076.0, + "step": 17910 + }, + { + "epoch": 1.966944871513288, + "grad_norm": 2.2874996662139893, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.738904595375061, + "num_tokens": 460160781.0, + "step": 17911 + }, + { + "epoch": 1.9670546892159018, + "grad_norm": 2.2232558727264404, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7151912450790405, + "num_tokens": 460186950.0, + "step": 17912 + }, + { + "epoch": 1.9671645069185153, + "grad_norm": 2.038719892501831, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7638072967529297, + "num_tokens": 460211531.0, + "step": 17913 + }, + { + "epoch": 1.9672743246211288, + "grad_norm": 2.2598876953125, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7353196144104004, + "num_tokens": 460236747.0, + "step": 17914 + }, + { + "epoch": 1.9673841423237426, + "grad_norm": 2.267669439315796, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7057253122329712, + "num_tokens": 460261325.0, + "step": 17915 + }, + { + "epoch": 1.9674939600263563, + "grad_norm": 2.223533868789673, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7215845584869385, + "num_tokens": 460287252.0, + "step": 17916 + }, + { + "epoch": 1.9676037777289699, + "grad_norm": 2.160090684890747, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7303354740142822, + "num_tokens": 460314445.0, + "step": 17917 + }, + { + "epoch": 1.9677135954315834, + "grad_norm": 2.1477813720703125, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7073837518692017, + "num_tokens": 460342463.0, + "step": 17918 + }, + { + "epoch": 1.9678234131341972, + "grad_norm": 2.232463836669922, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7400391697883606, + "num_tokens": 460366112.0, + "step": 17919 + }, + { + "epoch": 1.967933230836811, + "grad_norm": 2.4730048179626465, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7436054944992065, + "num_tokens": 460387435.0, + "step": 17920 + }, + { + "epoch": 1.9680430485394247, + "grad_norm": 2.0712106227874756, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7640619277954102, + "num_tokens": 460412786.0, + "step": 17921 + }, + { + "epoch": 1.9681528662420382, + "grad_norm": 2.455724000930786, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7372048497200012, + "num_tokens": 460433570.0, + "step": 17922 + }, + { + "epoch": 1.9682626839446518, + "grad_norm": 2.0028016567230225, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7174630165100098, + "num_tokens": 460463807.0, + "step": 17923 + }, + { + "epoch": 1.9683725016472655, + "grad_norm": 2.216402769088745, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7267168164253235, + "num_tokens": 460491884.0, + "step": 17924 + }, + { + "epoch": 1.9684823193498793, + "grad_norm": 2.191718578338623, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7381418347358704, + "num_tokens": 460514912.0, + "step": 17925 + }, + { + "epoch": 1.9685921370524928, + "grad_norm": 2.2975881099700928, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7447296977043152, + "num_tokens": 460538226.0, + "step": 17926 + }, + { + "epoch": 1.9687019547551066, + "grad_norm": 2.1802780628204346, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6886608600616455, + "num_tokens": 460568089.0, + "step": 17927 + }, + { + "epoch": 1.96881177245772, + "grad_norm": 2.2845191955566406, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7251887917518616, + "num_tokens": 460593044.0, + "step": 17928 + }, + { + "epoch": 1.9689215901603339, + "grad_norm": 1.9553401470184326, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7040456533432007, + "num_tokens": 460625099.0, + "step": 17929 + }, + { + "epoch": 1.9690314078629476, + "grad_norm": 2.175278902053833, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7563951015472412, + "num_tokens": 460650738.0, + "step": 17930 + }, + { + "epoch": 1.9691412255655611, + "grad_norm": 2.2694804668426514, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7211951017379761, + "num_tokens": 460675880.0, + "step": 17931 + }, + { + "epoch": 1.9692510432681747, + "grad_norm": 2.5000545978546143, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7040514945983887, + "num_tokens": 460701916.0, + "step": 17932 + }, + { + "epoch": 1.9693608609707884, + "grad_norm": 2.731154441833496, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7436134219169617, + "num_tokens": 460722095.0, + "step": 17933 + }, + { + "epoch": 1.9694706786734022, + "grad_norm": 1.9666224718093872, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7099746465682983, + "num_tokens": 460752852.0, + "step": 17934 + }, + { + "epoch": 1.969580496376016, + "grad_norm": 2.2247397899627686, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7105140686035156, + "num_tokens": 460778776.0, + "step": 17935 + }, + { + "epoch": 1.9696903140786295, + "grad_norm": 2.0930657386779785, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7206207513809204, + "num_tokens": 460807573.0, + "step": 17936 + }, + { + "epoch": 1.969800131781243, + "grad_norm": 2.063883066177368, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7253170013427734, + "num_tokens": 460836606.0, + "step": 17937 + }, + { + "epoch": 1.9699099494838568, + "grad_norm": 2.344224452972412, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7233489751815796, + "num_tokens": 460860231.0, + "step": 17938 + }, + { + "epoch": 1.9700197671864705, + "grad_norm": 2.6480610370635986, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7410752177238464, + "num_tokens": 460879374.0, + "step": 17939 + }, + { + "epoch": 1.970129584889084, + "grad_norm": 2.353205680847168, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7218620777130127, + "num_tokens": 460902971.0, + "step": 17940 + }, + { + "epoch": 1.9702394025916978, + "grad_norm": 2.1464452743530273, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7266583442687988, + "num_tokens": 460930001.0, + "step": 17941 + }, + { + "epoch": 1.9703492202943114, + "grad_norm": 2.3074939250946045, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7492296099662781, + "num_tokens": 460952137.0, + "step": 17942 + }, + { + "epoch": 1.9704590379969251, + "grad_norm": 2.342609167098999, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.743375837802887, + "num_tokens": 460974829.0, + "step": 17943 + }, + { + "epoch": 1.9705688556995389, + "grad_norm": 2.3381547927856445, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7264423370361328, + "num_tokens": 460999275.0, + "step": 17944 + }, + { + "epoch": 1.9706786734021524, + "grad_norm": 2.1062464714050293, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7209149599075317, + "num_tokens": 461029017.0, + "step": 17945 + }, + { + "epoch": 1.970788491104766, + "grad_norm": 2.4491472244262695, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7168477773666382, + "num_tokens": 461051867.0, + "step": 17946 + }, + { + "epoch": 1.9708983088073797, + "grad_norm": 2.119046211242676, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7082172632217407, + "num_tokens": 461080385.0, + "step": 17947 + }, + { + "epoch": 1.9710081265099935, + "grad_norm": 2.2391481399536133, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7200652360916138, + "num_tokens": 461106118.0, + "step": 17948 + }, + { + "epoch": 1.9711179442126072, + "grad_norm": 1.9601092338562012, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7035421133041382, + "num_tokens": 461137670.0, + "step": 17949 + }, + { + "epoch": 1.9712277619152208, + "grad_norm": 2.155447244644165, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7101935148239136, + "num_tokens": 461164351.0, + "step": 17950 + }, + { + "epoch": 1.9713375796178343, + "grad_norm": 2.347896099090576, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.736911416053772, + "num_tokens": 461186305.0, + "step": 17951 + }, + { + "epoch": 1.971447397320448, + "grad_norm": 2.115508794784546, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7136761546134949, + "num_tokens": 461214638.0, + "step": 17952 + }, + { + "epoch": 1.9715572150230618, + "grad_norm": 2.1108932495117188, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.71480792760849, + "num_tokens": 461242694.0, + "step": 17953 + }, + { + "epoch": 1.9716670327256753, + "grad_norm": 2.006624221801758, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7229639291763306, + "num_tokens": 461272433.0, + "step": 17954 + }, + { + "epoch": 1.9717768504282889, + "grad_norm": 2.226872205734253, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.724817156791687, + "num_tokens": 461298131.0, + "step": 17955 + }, + { + "epoch": 1.9718866681309026, + "grad_norm": 2.148545980453491, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7162413001060486, + "num_tokens": 461327522.0, + "step": 17956 + }, + { + "epoch": 1.9719964858335164, + "grad_norm": 2.027649402618408, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7033474445343018, + "num_tokens": 461355346.0, + "step": 17957 + }, + { + "epoch": 1.9721063035361301, + "grad_norm": 2.1704211235046387, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7058570384979248, + "num_tokens": 461382216.0, + "step": 17958 + }, + { + "epoch": 1.9722161212387437, + "grad_norm": 2.0767548084259033, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.708750307559967, + "num_tokens": 461412021.0, + "step": 17959 + }, + { + "epoch": 1.9723259389413572, + "grad_norm": 2.2400195598602295, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6918754577636719, + "num_tokens": 461440613.0, + "step": 17960 + }, + { + "epoch": 1.972435756643971, + "grad_norm": 2.195356607437134, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7245301008224487, + "num_tokens": 461467593.0, + "step": 17961 + }, + { + "epoch": 1.9725455743465847, + "grad_norm": 2.057710886001587, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6991156339645386, + "num_tokens": 461499933.0, + "step": 17962 + }, + { + "epoch": 1.9726553920491985, + "grad_norm": 2.1308815479278564, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7276978492736816, + "num_tokens": 461526709.0, + "step": 17963 + }, + { + "epoch": 1.972765209751812, + "grad_norm": 2.178074836730957, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7157537937164307, + "num_tokens": 461551089.0, + "step": 17964 + }, + { + "epoch": 1.9728750274544256, + "grad_norm": 2.2127857208251953, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.71270751953125, + "num_tokens": 461577840.0, + "step": 17965 + }, + { + "epoch": 1.9729848451570393, + "grad_norm": 2.066589832305908, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7072741985321045, + "num_tokens": 461606006.0, + "step": 17966 + }, + { + "epoch": 1.973094662859653, + "grad_norm": 2.449090003967285, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7158467769622803, + "num_tokens": 461628500.0, + "step": 17967 + }, + { + "epoch": 1.9732044805622666, + "grad_norm": 1.9525052309036255, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7214598059654236, + "num_tokens": 461659753.0, + "step": 17968 + }, + { + "epoch": 1.9733142982648801, + "grad_norm": 2.3194985389709473, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.729844331741333, + "num_tokens": 461683892.0, + "step": 17969 + }, + { + "epoch": 1.973424115967494, + "grad_norm": 2.5097787380218506, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7184698581695557, + "num_tokens": 461705047.0, + "step": 17970 + }, + { + "epoch": 1.9735339336701077, + "grad_norm": 2.183058023452759, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.740397572517395, + "num_tokens": 461732139.0, + "step": 17971 + }, + { + "epoch": 1.9736437513727214, + "grad_norm": 2.3210809230804443, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7221167683601379, + "num_tokens": 461756372.0, + "step": 17972 + }, + { + "epoch": 1.973753569075335, + "grad_norm": 2.3287723064422607, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7198973894119263, + "num_tokens": 461778677.0, + "step": 17973 + }, + { + "epoch": 1.9738633867779485, + "grad_norm": 1.946312427520752, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7400437593460083, + "num_tokens": 461807390.0, + "step": 17974 + }, + { + "epoch": 1.9739732044805622, + "grad_norm": 2.367947816848755, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7464134693145752, + "num_tokens": 461830481.0, + "step": 17975 + }, + { + "epoch": 1.974083022183176, + "grad_norm": 2.227139472961426, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7198632955551147, + "num_tokens": 461855651.0, + "step": 17976 + }, + { + "epoch": 1.9741928398857898, + "grad_norm": 2.4655110836029053, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.709126353263855, + "num_tokens": 461878228.0, + "step": 17977 + }, + { + "epoch": 1.9743026575884033, + "grad_norm": 2.4382333755493164, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7399489879608154, + "num_tokens": 461900458.0, + "step": 17978 + }, + { + "epoch": 1.9744124752910168, + "grad_norm": 2.4806203842163086, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.751017153263092, + "num_tokens": 461920623.0, + "step": 17979 + }, + { + "epoch": 1.9745222929936306, + "grad_norm": 2.1516366004943848, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7138501405715942, + "num_tokens": 461947232.0, + "step": 17980 + }, + { + "epoch": 1.9746321106962443, + "grad_norm": 2.2324647903442383, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7106600999832153, + "num_tokens": 461973112.0, + "step": 17981 + }, + { + "epoch": 1.9747419283988579, + "grad_norm": 2.2721381187438965, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7409055233001709, + "num_tokens": 461997339.0, + "step": 17982 + }, + { + "epoch": 1.9748517461014714, + "grad_norm": 2.365976333618164, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7416148781776428, + "num_tokens": 462019328.0, + "step": 17983 + }, + { + "epoch": 1.9749615638040852, + "grad_norm": 2.2510197162628174, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6940345764160156, + "num_tokens": 462045572.0, + "step": 17984 + }, + { + "epoch": 1.975071381506699, + "grad_norm": 2.143068313598633, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7265082597732544, + "num_tokens": 462071980.0, + "step": 17985 + }, + { + "epoch": 1.9751811992093127, + "grad_norm": 2.257233142852783, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7143785357475281, + "num_tokens": 462096966.0, + "step": 17986 + }, + { + "epoch": 1.9752910169119262, + "grad_norm": 2.046217203140259, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6991258859634399, + "num_tokens": 462125570.0, + "step": 17987 + }, + { + "epoch": 1.9754008346145397, + "grad_norm": 2.18650484085083, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7252031564712524, + "num_tokens": 462151156.0, + "step": 17988 + }, + { + "epoch": 1.9755106523171535, + "grad_norm": 2.2810802459716797, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7456482648849487, + "num_tokens": 462173916.0, + "step": 17989 + }, + { + "epoch": 1.9756204700197673, + "grad_norm": 2.5008656978607178, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7345306873321533, + "num_tokens": 462194931.0, + "step": 17990 + }, + { + "epoch": 1.9757302877223808, + "grad_norm": 2.511669397354126, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7189052104949951, + "num_tokens": 462216459.0, + "step": 17991 + }, + { + "epoch": 1.9758401054249946, + "grad_norm": 2.4479305744171143, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7281283140182495, + "num_tokens": 462238400.0, + "step": 17992 + }, + { + "epoch": 1.975949923127608, + "grad_norm": 2.522399425506592, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7353276610374451, + "num_tokens": 462258469.0, + "step": 17993 + }, + { + "epoch": 1.9760597408302218, + "grad_norm": 2.264742851257324, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7060703039169312, + "num_tokens": 462284911.0, + "step": 17994 + }, + { + "epoch": 1.9761695585328356, + "grad_norm": 2.4283735752105713, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7132038474082947, + "num_tokens": 462307839.0, + "step": 17995 + }, + { + "epoch": 1.9762793762354491, + "grad_norm": 2.098907470703125, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7344846725463867, + "num_tokens": 462335865.0, + "step": 17996 + }, + { + "epoch": 1.9763891939380627, + "grad_norm": 2.1416733264923096, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7197555899620056, + "num_tokens": 462361872.0, + "step": 17997 + }, + { + "epoch": 1.9764990116406764, + "grad_norm": 2.089292526245117, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7049432992935181, + "num_tokens": 462391306.0, + "step": 17998 + }, + { + "epoch": 1.9766088293432902, + "grad_norm": 2.1650640964508057, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7322999835014343, + "num_tokens": 462417557.0, + "step": 17999 + }, + { + "epoch": 1.976718647045904, + "grad_norm": 2.484825611114502, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7212819457054138, + "num_tokens": 462438533.0, + "step": 18000 + }, + { + "epoch": 1.9768284647485175, + "grad_norm": 2.427269220352173, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7063970565795898, + "num_tokens": 462462653.0, + "step": 18001 + }, + { + "epoch": 1.976938282451131, + "grad_norm": 2.265183687210083, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7305852174758911, + "num_tokens": 462486239.0, + "step": 18002 + }, + { + "epoch": 1.9770481001537448, + "grad_norm": 2.230597496032715, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7185332179069519, + "num_tokens": 462509836.0, + "step": 18003 + }, + { + "epoch": 1.9771579178563585, + "grad_norm": 2.2247180938720703, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.702971875667572, + "num_tokens": 462536621.0, + "step": 18004 + }, + { + "epoch": 1.977267735558972, + "grad_norm": 2.264225482940674, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7335833311080933, + "num_tokens": 462563172.0, + "step": 18005 + }, + { + "epoch": 1.9773775532615858, + "grad_norm": 2.508418083190918, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7315133810043335, + "num_tokens": 462585638.0, + "step": 18006 + }, + { + "epoch": 1.9774873709641994, + "grad_norm": 2.2215375900268555, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7170637249946594, + "num_tokens": 462612516.0, + "step": 18007 + }, + { + "epoch": 1.977597188666813, + "grad_norm": 2.0138444900512695, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6857616901397705, + "num_tokens": 462642870.0, + "step": 18008 + }, + { + "epoch": 1.9777070063694269, + "grad_norm": 2.132253408432007, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6884022951126099, + "num_tokens": 462672768.0, + "step": 18009 + }, + { + "epoch": 1.9778168240720404, + "grad_norm": 2.3855037689208984, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7215192317962646, + "num_tokens": 462695974.0, + "step": 18010 + }, + { + "epoch": 1.977926641774654, + "grad_norm": 2.231715202331543, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7135974168777466, + "num_tokens": 462721613.0, + "step": 18011 + }, + { + "epoch": 1.9780364594772677, + "grad_norm": 2.3940680027008057, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7208023071289062, + "num_tokens": 462744580.0, + "step": 18012 + }, + { + "epoch": 1.9781462771798815, + "grad_norm": 2.2351150512695312, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7308815121650696, + "num_tokens": 462767651.0, + "step": 18013 + }, + { + "epoch": 1.9782560948824952, + "grad_norm": 2.3649466037750244, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7116042375564575, + "num_tokens": 462789197.0, + "step": 18014 + }, + { + "epoch": 1.9783659125851087, + "grad_norm": 2.2866644859313965, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7112140655517578, + "num_tokens": 462815231.0, + "step": 18015 + }, + { + "epoch": 1.9784757302877223, + "grad_norm": 2.475651979446411, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7135306000709534, + "num_tokens": 462835963.0, + "step": 18016 + }, + { + "epoch": 1.978585547990336, + "grad_norm": 2.275520086288452, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7069090604782104, + "num_tokens": 462862475.0, + "step": 18017 + }, + { + "epoch": 1.9786953656929498, + "grad_norm": 2.0400755405426025, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7268890738487244, + "num_tokens": 462890444.0, + "step": 18018 + }, + { + "epoch": 1.9788051833955633, + "grad_norm": 2.1556923389434814, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7079237103462219, + "num_tokens": 462918486.0, + "step": 18019 + }, + { + "epoch": 1.9789150010981769, + "grad_norm": 2.3041365146636963, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7460241317749023, + "num_tokens": 462942109.0, + "step": 18020 + }, + { + "epoch": 1.9790248188007906, + "grad_norm": 2.066999912261963, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7367634177207947, + "num_tokens": 462969468.0, + "step": 18021 + }, + { + "epoch": 1.9791346365034044, + "grad_norm": 2.1792850494384766, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7129489779472351, + "num_tokens": 462996673.0, + "step": 18022 + }, + { + "epoch": 1.9792444542060181, + "grad_norm": 2.2094430923461914, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7235553860664368, + "num_tokens": 463020896.0, + "step": 18023 + }, + { + "epoch": 1.9793542719086317, + "grad_norm": 2.5134434700012207, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7113075256347656, + "num_tokens": 463042694.0, + "step": 18024 + }, + { + "epoch": 1.9794640896112452, + "grad_norm": 2.220137119293213, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7217441201210022, + "num_tokens": 463068286.0, + "step": 18025 + }, + { + "epoch": 1.979573907313859, + "grad_norm": 2.2075021266937256, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6992034912109375, + "num_tokens": 463096235.0, + "step": 18026 + }, + { + "epoch": 1.9796837250164727, + "grad_norm": 2.4263319969177246, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7308484315872192, + "num_tokens": 463118897.0, + "step": 18027 + }, + { + "epoch": 1.9797935427190865, + "grad_norm": 2.1126132011413574, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7208095788955688, + "num_tokens": 463146995.0, + "step": 18028 + }, + { + "epoch": 1.9799033604217, + "grad_norm": 2.395142078399658, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7572630047798157, + "num_tokens": 463168980.0, + "step": 18029 + }, + { + "epoch": 1.9800131781243135, + "grad_norm": 2.166464328765869, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7502635717391968, + "num_tokens": 463194992.0, + "step": 18030 + }, + { + "epoch": 1.9801229958269273, + "grad_norm": 2.617074489593506, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7374933958053589, + "num_tokens": 463216363.0, + "step": 18031 + }, + { + "epoch": 1.980232813529541, + "grad_norm": 2.2386534214019775, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7195165157318115, + "num_tokens": 463241665.0, + "step": 18032 + }, + { + "epoch": 1.9803426312321546, + "grad_norm": 2.3242883682250977, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7263427376747131, + "num_tokens": 463264586.0, + "step": 18033 + }, + { + "epoch": 1.9804524489347681, + "grad_norm": 2.337170124053955, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7366052865982056, + "num_tokens": 463287686.0, + "step": 18034 + }, + { + "epoch": 1.9805622666373819, + "grad_norm": 2.401273250579834, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7088162302970886, + "num_tokens": 463310383.0, + "step": 18035 + }, + { + "epoch": 1.9806720843399956, + "grad_norm": 2.330070734024048, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7041558027267456, + "num_tokens": 463337138.0, + "step": 18036 + }, + { + "epoch": 1.9807819020426094, + "grad_norm": 2.3583898544311523, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.717032790184021, + "num_tokens": 463361202.0, + "step": 18037 + }, + { + "epoch": 1.980891719745223, + "grad_norm": 2.2331111431121826, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6942247152328491, + "num_tokens": 463389231.0, + "step": 18038 + }, + { + "epoch": 1.9810015374478365, + "grad_norm": 2.1722798347473145, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7131596207618713, + "num_tokens": 463416644.0, + "step": 18039 + }, + { + "epoch": 1.9811113551504502, + "grad_norm": 2.3106420040130615, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7204588055610657, + "num_tokens": 463440315.0, + "step": 18040 + }, + { + "epoch": 1.981221172853064, + "grad_norm": 2.136472702026367, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7140712738037109, + "num_tokens": 463469039.0, + "step": 18041 + }, + { + "epoch": 1.9813309905556777, + "grad_norm": 2.371994733810425, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7461157441139221, + "num_tokens": 463491217.0, + "step": 18042 + }, + { + "epoch": 1.9814408082582913, + "grad_norm": 2.310638666152954, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7390307784080505, + "num_tokens": 463515976.0, + "step": 18043 + }, + { + "epoch": 1.9815506259609048, + "grad_norm": 2.5660884380340576, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.714038610458374, + "num_tokens": 463537048.0, + "step": 18044 + }, + { + "epoch": 1.9816604436635186, + "grad_norm": 2.248721122741699, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7150432467460632, + "num_tokens": 463562436.0, + "step": 18045 + }, + { + "epoch": 1.9817702613661323, + "grad_norm": 2.121488571166992, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7127320766448975, + "num_tokens": 463589393.0, + "step": 18046 + }, + { + "epoch": 1.9818800790687459, + "grad_norm": 2.1609277725219727, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7291260957717896, + "num_tokens": 463616673.0, + "step": 18047 + }, + { + "epoch": 1.9819898967713594, + "grad_norm": 2.094428300857544, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.730344295501709, + "num_tokens": 463643542.0, + "step": 18048 + }, + { + "epoch": 1.9820997144739732, + "grad_norm": 2.050079107284546, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6941026449203491, + "num_tokens": 463674435.0, + "step": 18049 + }, + { + "epoch": 1.982209532176587, + "grad_norm": 2.370452404022217, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7367917895317078, + "num_tokens": 463697333.0, + "step": 18050 + }, + { + "epoch": 1.9823193498792007, + "grad_norm": 1.9914240837097168, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7112935781478882, + "num_tokens": 463728277.0, + "step": 18051 + }, + { + "epoch": 1.9824291675818142, + "grad_norm": 2.415560245513916, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7145955562591553, + "num_tokens": 463751146.0, + "step": 18052 + }, + { + "epoch": 1.9825389852844277, + "grad_norm": 2.3657565116882324, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7303454875946045, + "num_tokens": 463773587.0, + "step": 18053 + }, + { + "epoch": 1.9826488029870415, + "grad_norm": 2.233306884765625, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7364534139633179, + "num_tokens": 463797997.0, + "step": 18054 + }, + { + "epoch": 1.9827586206896552, + "grad_norm": 2.112631320953369, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7071701288223267, + "num_tokens": 463827093.0, + "step": 18055 + }, + { + "epoch": 1.9828684383922688, + "grad_norm": 2.175718069076538, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7222217321395874, + "num_tokens": 463853439.0, + "step": 18056 + }, + { + "epoch": 1.9829782560948825, + "grad_norm": 2.345673084259033, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7075827121734619, + "num_tokens": 463878589.0, + "step": 18057 + }, + { + "epoch": 1.983088073797496, + "grad_norm": 2.1829397678375244, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7360560894012451, + "num_tokens": 463904283.0, + "step": 18058 + }, + { + "epoch": 1.9831978915001098, + "grad_norm": 2.2408735752105713, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7077473402023315, + "num_tokens": 463932466.0, + "step": 18059 + }, + { + "epoch": 1.9833077092027236, + "grad_norm": 2.425212860107422, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7169268727302551, + "num_tokens": 463954526.0, + "step": 18060 + }, + { + "epoch": 1.9834175269053371, + "grad_norm": 1.957319736480713, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7145064473152161, + "num_tokens": 463987781.0, + "step": 18061 + }, + { + "epoch": 1.9835273446079507, + "grad_norm": 2.0172536373138428, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7218371629714966, + "num_tokens": 464019698.0, + "step": 18062 + }, + { + "epoch": 1.9836371623105644, + "grad_norm": 2.5188510417938232, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7075717449188232, + "num_tokens": 464042284.0, + "step": 18063 + }, + { + "epoch": 1.9837469800131782, + "grad_norm": 2.187854766845703, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7424206733703613, + "num_tokens": 464067702.0, + "step": 18064 + }, + { + "epoch": 1.983856797715792, + "grad_norm": 2.126795530319214, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7205862998962402, + "num_tokens": 464094307.0, + "step": 18065 + }, + { + "epoch": 1.9839666154184055, + "grad_norm": 2.095745086669922, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7382528781890869, + "num_tokens": 464121805.0, + "step": 18066 + }, + { + "epoch": 1.984076433121019, + "grad_norm": 2.2143895626068115, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7225656509399414, + "num_tokens": 464146207.0, + "step": 18067 + }, + { + "epoch": 1.9841862508236328, + "grad_norm": 2.150388717651367, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7637024521827698, + "num_tokens": 464169299.0, + "step": 18068 + }, + { + "epoch": 1.9842960685262465, + "grad_norm": 2.705697774887085, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7245743274688721, + "num_tokens": 464190357.0, + "step": 18069 + }, + { + "epoch": 1.98440588622886, + "grad_norm": 2.212310791015625, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7414721846580505, + "num_tokens": 464215122.0, + "step": 18070 + }, + { + "epoch": 1.9845157039314738, + "grad_norm": 2.33396577835083, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.7056254148483276, + "num_tokens": 464240697.0, + "step": 18071 + }, + { + "epoch": 1.9846255216340873, + "grad_norm": 1.8520898818969727, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7018779516220093, + "num_tokens": 464275882.0, + "step": 18072 + }, + { + "epoch": 1.984735339336701, + "grad_norm": 2.4583513736724854, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7237844467163086, + "num_tokens": 464299114.0, + "step": 18073 + }, + { + "epoch": 1.9848451570393149, + "grad_norm": 2.1311824321746826, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7207002639770508, + "num_tokens": 464326446.0, + "step": 18074 + }, + { + "epoch": 1.9849549747419284, + "grad_norm": 2.5913636684417725, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7325298190116882, + "num_tokens": 464346079.0, + "step": 18075 + }, + { + "epoch": 1.985064792444542, + "grad_norm": 2.8623850345611572, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7611484527587891, + "num_tokens": 464362384.0, + "step": 18076 + }, + { + "epoch": 1.9851746101471557, + "grad_norm": 2.2238128185272217, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7340590357780457, + "num_tokens": 464387824.0, + "step": 18077 + }, + { + "epoch": 1.9852844278497694, + "grad_norm": 2.3438687324523926, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7221688032150269, + "num_tokens": 464412359.0, + "step": 18078 + }, + { + "epoch": 1.9853942455523832, + "grad_norm": 2.167379379272461, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7207779884338379, + "num_tokens": 464440160.0, + "step": 18079 + }, + { + "epoch": 1.9855040632549967, + "grad_norm": 2.3012290000915527, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7411856651306152, + "num_tokens": 464463837.0, + "step": 18080 + }, + { + "epoch": 1.9856138809576103, + "grad_norm": 2.457693576812744, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.714749813079834, + "num_tokens": 464491356.0, + "step": 18081 + }, + { + "epoch": 1.985723698660224, + "grad_norm": 2.526113510131836, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7100107073783875, + "num_tokens": 464514637.0, + "step": 18082 + }, + { + "epoch": 1.9858335163628378, + "grad_norm": 2.081007719039917, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7127655744552612, + "num_tokens": 464544898.0, + "step": 18083 + }, + { + "epoch": 1.9859433340654513, + "grad_norm": 2.1425843238830566, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7356938123703003, + "num_tokens": 464571248.0, + "step": 18084 + }, + { + "epoch": 1.9860531517680649, + "grad_norm": 2.086278200149536, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7083695530891418, + "num_tokens": 464600461.0, + "step": 18085 + }, + { + "epoch": 1.9861629694706786, + "grad_norm": 2.246445894241333, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7363051176071167, + "num_tokens": 464624558.0, + "step": 18086 + }, + { + "epoch": 1.9862727871732924, + "grad_norm": 2.336923360824585, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7480883002281189, + "num_tokens": 464648786.0, + "step": 18087 + }, + { + "epoch": 1.9863826048759061, + "grad_norm": 2.173410177230835, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.688084602355957, + "num_tokens": 464675525.0, + "step": 18088 + }, + { + "epoch": 1.9864924225785197, + "grad_norm": 2.363198757171631, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7276335954666138, + "num_tokens": 464698317.0, + "step": 18089 + }, + { + "epoch": 1.9866022402811332, + "grad_norm": 2.301973342895508, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7205609083175659, + "num_tokens": 464722410.0, + "step": 18090 + }, + { + "epoch": 1.986712057983747, + "grad_norm": 1.9296234846115112, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6933977007865906, + "num_tokens": 464754585.0, + "step": 18091 + }, + { + "epoch": 1.9868218756863607, + "grad_norm": 2.251476526260376, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7421589493751526, + "num_tokens": 464779021.0, + "step": 18092 + }, + { + "epoch": 1.9869316933889745, + "grad_norm": 2.2078182697296143, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7184262275695801, + "num_tokens": 464805512.0, + "step": 18093 + }, + { + "epoch": 1.987041511091588, + "grad_norm": 2.0133142471313477, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7135590314865112, + "num_tokens": 464836891.0, + "step": 18094 + }, + { + "epoch": 1.9871513287942015, + "grad_norm": 2.291827917098999, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7182571291923523, + "num_tokens": 464861895.0, + "step": 18095 + }, + { + "epoch": 1.9872611464968153, + "grad_norm": 2.342468738555908, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7066001296043396, + "num_tokens": 464884870.0, + "step": 18096 + }, + { + "epoch": 1.987370964199429, + "grad_norm": 2.428727149963379, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.726536214351654, + "num_tokens": 464908108.0, + "step": 18097 + }, + { + "epoch": 1.9874807819020426, + "grad_norm": 2.349231719970703, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7072762250900269, + "num_tokens": 464931340.0, + "step": 18098 + }, + { + "epoch": 1.9875905996046561, + "grad_norm": 2.1412503719329834, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7199554443359375, + "num_tokens": 464956772.0, + "step": 18099 + }, + { + "epoch": 1.9877004173072699, + "grad_norm": 2.117600440979004, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7134188413619995, + "num_tokens": 464987170.0, + "step": 18100 + }, + { + "epoch": 1.9878102350098836, + "grad_norm": 2.122606039047241, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.719218373298645, + "num_tokens": 465015876.0, + "step": 18101 + }, + { + "epoch": 1.9879200527124974, + "grad_norm": 2.3612287044525146, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7614398002624512, + "num_tokens": 465037599.0, + "step": 18102 + }, + { + "epoch": 1.988029870415111, + "grad_norm": 2.1078641414642334, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7174440026283264, + "num_tokens": 465067957.0, + "step": 18103 + }, + { + "epoch": 1.9881396881177245, + "grad_norm": 2.136823892593384, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7258110046386719, + "num_tokens": 465095164.0, + "step": 18104 + }, + { + "epoch": 1.9882495058203382, + "grad_norm": 2.224961042404175, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7569306492805481, + "num_tokens": 465119803.0, + "step": 18105 + }, + { + "epoch": 1.988359323522952, + "grad_norm": 2.400202512741089, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7340782880783081, + "num_tokens": 465144044.0, + "step": 18106 + }, + { + "epoch": 1.9884691412255655, + "grad_norm": 2.5714712142944336, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.708289623260498, + "num_tokens": 465165422.0, + "step": 18107 + }, + { + "epoch": 1.9885789589281793, + "grad_norm": 2.18719744682312, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7345753908157349, + "num_tokens": 465190341.0, + "step": 18108 + }, + { + "epoch": 1.9886887766307928, + "grad_norm": 2.4962542057037354, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7379316687583923, + "num_tokens": 465209325.0, + "step": 18109 + }, + { + "epoch": 1.9887985943334066, + "grad_norm": 2.1964776515960693, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7194658517837524, + "num_tokens": 465241202.0, + "step": 18110 + }, + { + "epoch": 1.9889084120360203, + "grad_norm": 2.3504788875579834, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7516571283340454, + "num_tokens": 465261979.0, + "step": 18111 + }, + { + "epoch": 1.9890182297386338, + "grad_norm": 2.147343158721924, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7057923078536987, + "num_tokens": 465288201.0, + "step": 18112 + }, + { + "epoch": 1.9891280474412474, + "grad_norm": 1.9098200798034668, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7240282297134399, + "num_tokens": 465320899.0, + "step": 18113 + }, + { + "epoch": 1.9892378651438611, + "grad_norm": 2.234393358230591, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7099891901016235, + "num_tokens": 465344473.0, + "step": 18114 + }, + { + "epoch": 1.989347682846475, + "grad_norm": 2.4567484855651855, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7320153713226318, + "num_tokens": 465365763.0, + "step": 18115 + }, + { + "epoch": 1.9894575005490887, + "grad_norm": 2.4729738235473633, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7263336777687073, + "num_tokens": 465387915.0, + "step": 18116 + }, + { + "epoch": 1.9895673182517022, + "grad_norm": 1.8688558340072632, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7031339406967163, + "num_tokens": 465422771.0, + "step": 18117 + }, + { + "epoch": 1.9896771359543157, + "grad_norm": 2.5863850116729736, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.733039379119873, + "num_tokens": 465442445.0, + "step": 18118 + }, + { + "epoch": 1.9897869536569295, + "grad_norm": 2.223052740097046, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7345458269119263, + "num_tokens": 465467646.0, + "step": 18119 + }, + { + "epoch": 1.9898967713595432, + "grad_norm": 2.020238161087036, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7011033296585083, + "num_tokens": 465497412.0, + "step": 18120 + }, + { + "epoch": 1.9900065890621568, + "grad_norm": 2.2622416019439697, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.719819188117981, + "num_tokens": 465520084.0, + "step": 18121 + }, + { + "epoch": 1.9901164067647705, + "grad_norm": 2.1629791259765625, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7171214818954468, + "num_tokens": 465546931.0, + "step": 18122 + }, + { + "epoch": 1.990226224467384, + "grad_norm": 2.1848535537719727, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7344121932983398, + "num_tokens": 465573135.0, + "step": 18123 + }, + { + "epoch": 1.9903360421699978, + "grad_norm": 1.936282753944397, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7229822874069214, + "num_tokens": 465604038.0, + "step": 18124 + }, + { + "epoch": 1.9904458598726116, + "grad_norm": 2.381695032119751, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7377314567565918, + "num_tokens": 465626040.0, + "step": 18125 + }, + { + "epoch": 1.9905556775752251, + "grad_norm": 2.2904164791107178, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7449390888214111, + "num_tokens": 465649421.0, + "step": 18126 + }, + { + "epoch": 1.9906654952778386, + "grad_norm": 2.190183401107788, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7139469385147095, + "num_tokens": 465675536.0, + "step": 18127 + }, + { + "epoch": 1.9907753129804524, + "grad_norm": 2.303276538848877, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7156394720077515, + "num_tokens": 465699371.0, + "step": 18128 + }, + { + "epoch": 1.9908851306830662, + "grad_norm": 2.1090734004974365, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7149443626403809, + "num_tokens": 465729467.0, + "step": 18129 + }, + { + "epoch": 1.99099494838568, + "grad_norm": 2.3311257362365723, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7114011645317078, + "num_tokens": 465754230.0, + "step": 18130 + }, + { + "epoch": 1.9911047660882935, + "grad_norm": 2.3186347484588623, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7245450615882874, + "num_tokens": 465777639.0, + "step": 18131 + }, + { + "epoch": 1.991214583790907, + "grad_norm": 2.1685168743133545, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7116088271141052, + "num_tokens": 465804544.0, + "step": 18132 + }, + { + "epoch": 1.9913244014935207, + "grad_norm": 2.5102314949035645, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7587827444076538, + "num_tokens": 465823416.0, + "step": 18133 + }, + { + "epoch": 1.9914342191961345, + "grad_norm": 2.1803317070007324, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7390269041061401, + "num_tokens": 465848448.0, + "step": 18134 + }, + { + "epoch": 1.991544036898748, + "grad_norm": 2.2690727710723877, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7111470699310303, + "num_tokens": 465873436.0, + "step": 18135 + }, + { + "epoch": 1.9916538546013618, + "grad_norm": 2.5881829261779785, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7510489225387573, + "num_tokens": 465895227.0, + "step": 18136 + }, + { + "epoch": 1.9917636723039753, + "grad_norm": 2.3097128868103027, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.690079927444458, + "num_tokens": 465920434.0, + "step": 18137 + }, + { + "epoch": 1.991873490006589, + "grad_norm": 2.3158071041107178, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.73856520652771, + "num_tokens": 465943143.0, + "step": 18138 + }, + { + "epoch": 1.9919833077092028, + "grad_norm": 2.1051573753356934, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7103118896484375, + "num_tokens": 465972034.0, + "step": 18139 + }, + { + "epoch": 1.9920931254118164, + "grad_norm": 2.104947566986084, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7078595161437988, + "num_tokens": 466002025.0, + "step": 18140 + }, + { + "epoch": 1.99220294311443, + "grad_norm": 1.9444036483764648, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6987276077270508, + "num_tokens": 466035154.0, + "step": 18141 + }, + { + "epoch": 1.9923127608170437, + "grad_norm": 2.0619513988494873, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6932362914085388, + "num_tokens": 466064246.0, + "step": 18142 + }, + { + "epoch": 1.9924225785196574, + "grad_norm": 2.332089424133301, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7383403778076172, + "num_tokens": 466087216.0, + "step": 18143 + }, + { + "epoch": 1.9925323962222712, + "grad_norm": 2.191098690032959, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7334403991699219, + "num_tokens": 466112303.0, + "step": 18144 + }, + { + "epoch": 1.9926422139248847, + "grad_norm": 2.1801395416259766, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7281506657600403, + "num_tokens": 466138142.0, + "step": 18145 + }, + { + "epoch": 1.9927520316274983, + "grad_norm": 2.0930421352386475, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7016110420227051, + "num_tokens": 466170185.0, + "step": 18146 + }, + { + "epoch": 1.992861849330112, + "grad_norm": 2.534656524658203, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7188408374786377, + "num_tokens": 466191411.0, + "step": 18147 + }, + { + "epoch": 1.9929716670327258, + "grad_norm": 2.2945199012756348, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.707697331905365, + "num_tokens": 466216066.0, + "step": 18148 + }, + { + "epoch": 1.9930814847353393, + "grad_norm": 2.107564926147461, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7058327198028564, + "num_tokens": 466245755.0, + "step": 18149 + }, + { + "epoch": 1.9931913024379528, + "grad_norm": 1.9441125392913818, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7071616649627686, + "num_tokens": 466277648.0, + "step": 18150 + }, + { + "epoch": 1.9933011201405666, + "grad_norm": 2.134956121444702, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7381126880645752, + "num_tokens": 466302493.0, + "step": 18151 + }, + { + "epoch": 1.9934109378431804, + "grad_norm": 1.982885479927063, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7283052206039429, + "num_tokens": 466330344.0, + "step": 18152 + }, + { + "epoch": 1.993520755545794, + "grad_norm": 1.9262595176696777, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7104325890541077, + "num_tokens": 466362058.0, + "step": 18153 + }, + { + "epoch": 1.9936305732484076, + "grad_norm": 2.0131778717041016, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6981642842292786, + "num_tokens": 466392748.0, + "step": 18154 + }, + { + "epoch": 1.9937403909510212, + "grad_norm": 2.388817310333252, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7220312356948853, + "num_tokens": 466416583.0, + "step": 18155 + }, + { + "epoch": 1.993850208653635, + "grad_norm": 2.2781729698181152, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7504202127456665, + "num_tokens": 466439280.0, + "step": 18156 + }, + { + "epoch": 1.9939600263562487, + "grad_norm": 2.186371088027954, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7144253849983215, + "num_tokens": 466468919.0, + "step": 18157 + }, + { + "epoch": 1.9940698440588625, + "grad_norm": 2.5366625785827637, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7318325042724609, + "num_tokens": 466489726.0, + "step": 18158 + }, + { + "epoch": 1.994179661761476, + "grad_norm": 2.1203415393829346, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.729749321937561, + "num_tokens": 466515573.0, + "step": 18159 + }, + { + "epoch": 1.9942894794640895, + "grad_norm": 2.1647794246673584, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7227855920791626, + "num_tokens": 466542013.0, + "step": 18160 + }, + { + "epoch": 1.9943992971667033, + "grad_norm": 2.463775634765625, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7052803039550781, + "num_tokens": 466564851.0, + "step": 18161 + }, + { + "epoch": 1.994509114869317, + "grad_norm": 2.3017079830169678, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7326422333717346, + "num_tokens": 466588704.0, + "step": 18162 + }, + { + "epoch": 1.9946189325719306, + "grad_norm": 2.256345272064209, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7211287617683411, + "num_tokens": 466613113.0, + "step": 18163 + }, + { + "epoch": 1.994728750274544, + "grad_norm": 2.4445321559906006, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7308934926986694, + "num_tokens": 466635795.0, + "step": 18164 + }, + { + "epoch": 1.9948385679771579, + "grad_norm": 2.421149969100952, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7328939437866211, + "num_tokens": 466658914.0, + "step": 18165 + }, + { + "epoch": 1.9949483856797716, + "grad_norm": 2.453174114227295, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7138367891311646, + "num_tokens": 466681712.0, + "step": 18166 + }, + { + "epoch": 1.9950582033823854, + "grad_norm": 2.1430463790893555, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7170034050941467, + "num_tokens": 466708575.0, + "step": 18167 + }, + { + "epoch": 1.995168021084999, + "grad_norm": 2.161884307861328, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7166060209274292, + "num_tokens": 466735592.0, + "step": 18168 + }, + { + "epoch": 1.9952778387876124, + "grad_norm": 2.257960796356201, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7054735422134399, + "num_tokens": 466762355.0, + "step": 18169 + }, + { + "epoch": 1.9953876564902262, + "grad_norm": 1.9802747964859009, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7013136744499207, + "num_tokens": 466794132.0, + "step": 18170 + }, + { + "epoch": 1.99549747419284, + "grad_norm": 2.2053956985473633, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7379405498504639, + "num_tokens": 466820757.0, + "step": 18171 + }, + { + "epoch": 1.9956072918954535, + "grad_norm": 2.4116523265838623, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7346206307411194, + "num_tokens": 466841834.0, + "step": 18172 + }, + { + "epoch": 1.9957171095980673, + "grad_norm": 2.543302297592163, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7353529930114746, + "num_tokens": 466863151.0, + "step": 18173 + }, + { + "epoch": 1.9958269273006808, + "grad_norm": 2.12131929397583, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7170572876930237, + "num_tokens": 466890628.0, + "step": 18174 + }, + { + "epoch": 1.9959367450032945, + "grad_norm": 2.383084774017334, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7172135710716248, + "num_tokens": 466916542.0, + "step": 18175 + }, + { + "epoch": 1.9960465627059083, + "grad_norm": 2.086819648742676, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7154118418693542, + "num_tokens": 466947956.0, + "step": 18176 + }, + { + "epoch": 1.9961563804085218, + "grad_norm": 2.3685107231140137, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7349948287010193, + "num_tokens": 466971640.0, + "step": 18177 + }, + { + "epoch": 1.9962661981111354, + "grad_norm": 2.193124294281006, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7259374260902405, + "num_tokens": 466995841.0, + "step": 18178 + }, + { + "epoch": 1.9963760158137491, + "grad_norm": 2.0644664764404297, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6920484900474548, + "num_tokens": 467025205.0, + "step": 18179 + }, + { + "epoch": 1.9964858335163629, + "grad_norm": 2.2045395374298096, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7130942344665527, + "num_tokens": 467051098.0, + "step": 18180 + }, + { + "epoch": 1.9965956512189766, + "grad_norm": 2.1174750328063965, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.6995404958724976, + "num_tokens": 467078979.0, + "step": 18181 + }, + { + "epoch": 1.9967054689215902, + "grad_norm": 2.0226752758026123, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7096934914588928, + "num_tokens": 467107587.0, + "step": 18182 + }, + { + "epoch": 1.9968152866242037, + "grad_norm": 2.001476526260376, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7303271889686584, + "num_tokens": 467137500.0, + "step": 18183 + }, + { + "epoch": 1.9969251043268175, + "grad_norm": 2.36568284034729, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7241260409355164, + "num_tokens": 467159631.0, + "step": 18184 + }, + { + "epoch": 1.9970349220294312, + "grad_norm": 2.4773247241973877, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7123476266860962, + "num_tokens": 467181296.0, + "step": 18185 + }, + { + "epoch": 1.9971447397320448, + "grad_norm": 2.09427809715271, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7111095190048218, + "num_tokens": 467209541.0, + "step": 18186 + }, + { + "epoch": 1.9972545574346585, + "grad_norm": 2.019425630569458, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7224204540252686, + "num_tokens": 467241243.0, + "step": 18187 + }, + { + "epoch": 1.997364375137272, + "grad_norm": 2.6060733795166016, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.737326443195343, + "num_tokens": 467262611.0, + "step": 18188 + }, + { + "epoch": 1.9974741928398858, + "grad_norm": 2.400103807449341, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7322056293487549, + "num_tokens": 467286383.0, + "step": 18189 + }, + { + "epoch": 1.9975840105424996, + "grad_norm": 2.2246735095977783, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7285124063491821, + "num_tokens": 467313088.0, + "step": 18190 + }, + { + "epoch": 1.997693828245113, + "grad_norm": 2.1116597652435303, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6980311274528503, + "num_tokens": 467342324.0, + "step": 18191 + }, + { + "epoch": 1.9978036459477266, + "grad_norm": 2.508584499359131, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7224026322364807, + "num_tokens": 467363493.0, + "step": 18192 + }, + { + "epoch": 1.9979134636503404, + "grad_norm": 2.1280722618103027, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7039735317230225, + "num_tokens": 467392044.0, + "step": 18193 + }, + { + "epoch": 1.9980232813529542, + "grad_norm": 2.119785785675049, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6913914680480957, + "num_tokens": 467421657.0, + "step": 18194 + }, + { + "epoch": 1.998133099055568, + "grad_norm": 2.224449872970581, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.735645592212677, + "num_tokens": 467446447.0, + "step": 18195 + }, + { + "epoch": 1.9982429167581814, + "grad_norm": 2.045576810836792, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7094264030456543, + "num_tokens": 467476509.0, + "step": 18196 + }, + { + "epoch": 1.998352734460795, + "grad_norm": 2.4321041107177734, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7138139009475708, + "num_tokens": 467500173.0, + "step": 18197 + }, + { + "epoch": 1.9984625521634087, + "grad_norm": 2.375444173812866, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7464569807052612, + "num_tokens": 467522736.0, + "step": 18198 + }, + { + "epoch": 1.9985723698660225, + "grad_norm": 2.3280587196350098, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7332707643508911, + "num_tokens": 467546146.0, + "step": 18199 + }, + { + "epoch": 1.998682187568636, + "grad_norm": 2.5334765911102295, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7147592306137085, + "num_tokens": 467567305.0, + "step": 18200 + }, + { + "epoch": 1.9987920052712496, + "grad_norm": 2.2884347438812256, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7407345771789551, + "num_tokens": 467590707.0, + "step": 18201 + }, + { + "epoch": 1.9989018229738633, + "grad_norm": 2.1917037963867188, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7002819180488586, + "num_tokens": 467620389.0, + "step": 18202 + }, + { + "epoch": 1.999011640676477, + "grad_norm": 2.371270179748535, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7552691698074341, + "num_tokens": 467642317.0, + "step": 18203 + }, + { + "epoch": 1.9991214583790908, + "grad_norm": 2.4881298542022705, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7488099932670593, + "num_tokens": 467663036.0, + "step": 18204 + }, + { + "epoch": 1.9992312760817044, + "grad_norm": 2.2525205612182617, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7230950593948364, + "num_tokens": 467687371.0, + "step": 18205 + }, + { + "epoch": 1.999341093784318, + "grad_norm": 2.2573606967926025, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7025941610336304, + "num_tokens": 467714296.0, + "step": 18206 + }, + { + "epoch": 1.9994509114869317, + "grad_norm": 2.4093058109283447, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7342577576637268, + "num_tokens": 467737537.0, + "step": 18207 + }, + { + "epoch": 1.9995607291895454, + "grad_norm": 2.197787284851074, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7498769164085388, + "num_tokens": 467762210.0, + "step": 18208 + }, + { + "epoch": 1.9996705468921592, + "grad_norm": 1.8845958709716797, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7160588502883911, + "num_tokens": 467796199.0, + "step": 18209 + }, + { + "epoch": 1.9997803645947727, + "grad_norm": 2.4534101486206055, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7365316152572632, + "num_tokens": 467817911.0, + "step": 18210 + }, + { + "epoch": 1.9998901822973862, + "grad_norm": 2.289191246032715, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7229564189910889, + "num_tokens": 467844122.0, + "step": 18211 + }, + { + "epoch": 2.0, + "grad_norm": 2.2008893489837646, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7493396997451782, + "num_tokens": 467867976.0, + "step": 18212 + }, + { + "epoch": 2.0, + "step": 18212, + "total_flos": 2.1067883493711675e+19, + "train_loss": 0.9522557386704232, + "train_runtime": 21515.8506, + "train_samples_per_second": 13.543, + "train_steps_per_second": 0.846 + } + ], + "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": 2.1067883493711675e+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..37f3f4e --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84263341608f3a2207d7786e4670e8e8b62d2ff914ae9bf4194ae4f65e7299a0 +size 13329