commit b514380385bc2a3dc3fae6bdf834f2bce8b0ffca Author: ModelHub XC Date: Thu Apr 23 20:46:04 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_LoXv00.01 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..52373fe --- /dev/null +++ b/.gitattributes @@ -0,0 +1,36 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..59b8549 --- /dev/null +++ b/README.md @@ -0,0 +1,60 @@ +--- +base_model: Neelectric/Llama-3.1-8B-Instruct_LoX_k_6_a_1.25 +datasets: Neelectric/OpenR1-Math-220k_all_Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_LoXv00.01 +tags: +- generated_from_trainer +- open-r1 +- sft +- trl +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_LoXv00.01 + +This model is a fine-tuned version of [Neelectric/Llama-3.1-8B-Instruct_LoX_k_6_a_1.25](https://huggingface.co/Neelectric/Llama-3.1-8B-Instruct_LoX_k_6_a_1.25) on the [Neelectric/OpenR1-Math-220k_all_Llama3_4096toks](https://huggingface.co/datasets/Neelectric/OpenR1-Math-220k_all_Llama3_4096toks) dataset. +It has been trained using [TRL](https://github.com/huggingface/trl). + +## Quick start + +```python +from transformers import pipeline + +question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" +generator = pipeline("text-generation", model="Neelectric/Llama-3.1-8B-Instruct_LoXv00.01", device="cuda") +output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0] +print(output["generated_text"]) +``` + +## Training procedure + +[Visualize in Weights & Biases](https://wandb.ai/neelectric/open-r1_math/runs/n6p3ixwo) + + + +This model was trained with SFT. + +### Framework versions + +- TRL: 1.1.0.dev0 +- Transformers: 4.57.6 +- Pytorch: 2.9.0 +- Datasets: 4.8.4 +- 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..19268fd --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 5.62815163329864e+19, + "train_loss": 0.6996442060487161, + "train_runtime": 42634.2383, + "train_samples": 125770, + "train_samples_per_second": 8.85, + "train_steps_per_second": 0.553 +} \ 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..48680a0 --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf9dc94be4c6d7104bac6a9384f8e50b3732db5d649362c136caadf42ba48569 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..8ff9f33 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d0fd292ad3c63fa62249e44d558fc7a2dcf127d6c6805e8d37b6ccaded59f1d +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..20a6e09 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac1bef0bb8585e3a5af76af4a87de7920360a298d614a5d0f9c509a3e6721bec +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..24e05bd --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba73a2ef14f4893e6ababf5e632a4947236a40665ffd6efee9cdd66fcf92791f +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..19268fd --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 5.62815163329864e+19, + "train_loss": 0.6996442060487161, + "train_runtime": 42634.2383, + "train_samples": 125770, + "train_samples_per_second": 8.85, + "train_steps_per_second": 0.553 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..146a562 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,212290 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 23583, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00012721027859051011, + "grad_norm": 29.75, + "learning_rate": 0.0, + "loss": 1.2028, + "mean_token_accuracy": 0.687287449836731, + "num_tokens": 38493.0, + "step": 1 + }, + { + "epoch": 0.00025442055718102023, + "grad_norm": 30.75, + "learning_rate": 4.2390843577787196e-10, + "loss": 1.2597, + "mean_token_accuracy": 0.6652442216873169, + "num_tokens": 80419.0, + "step": 2 + }, + { + "epoch": 0.0003816308357715303, + "grad_norm": 28.25, + "learning_rate": 8.478168715557439e-10, + "loss": 1.1047, + "mean_token_accuracy": 0.7121957540512085, + "num_tokens": 118717.0, + "step": 3 + }, + { + "epoch": 0.0005088411143620405, + "grad_norm": 29.75, + "learning_rate": 1.271725307333616e-09, + "loss": 1.2414, + "mean_token_accuracy": 0.6777700185775757, + "num_tokens": 150155.0, + "step": 4 + }, + { + "epoch": 0.0006360513929525506, + "grad_norm": 28.875, + "learning_rate": 1.6956337431114878e-09, + "loss": 1.1954, + "mean_token_accuracy": 0.6812636852264404, + "num_tokens": 193616.0, + "step": 5 + }, + { + "epoch": 0.0007632616715430606, + "grad_norm": 30.375, + "learning_rate": 2.1195421788893596e-09, + "loss": 1.2119, + "mean_token_accuracy": 0.6868374347686768, + "num_tokens": 227640.0, + "step": 6 + }, + { + "epoch": 0.0008904719501335708, + "grad_norm": 29.5, + "learning_rate": 2.543450614667232e-09, + "loss": 1.2373, + "mean_token_accuracy": 0.6850466728210449, + "num_tokens": 265114.0, + "step": 7 + }, + { + "epoch": 0.001017682228724081, + "grad_norm": 28.75, + "learning_rate": 2.967359050445104e-09, + "loss": 1.1586, + "mean_token_accuracy": 0.6989333033561707, + "num_tokens": 299865.0, + "step": 8 + }, + { + "epoch": 0.001144892507314591, + "grad_norm": 29.5, + "learning_rate": 3.3912674862229757e-09, + "loss": 1.2321, + "mean_token_accuracy": 0.6764354705810547, + "num_tokens": 342063.0, + "step": 9 + }, + { + "epoch": 0.0012721027859051012, + "grad_norm": 34.25, + "learning_rate": 3.815175922000847e-09, + "loss": 1.3425, + "mean_token_accuracy": 0.6644104719161987, + "num_tokens": 374864.0, + "step": 10 + }, + { + "epoch": 0.0013993130644956112, + "grad_norm": 27.375, + "learning_rate": 4.239084357778719e-09, + "loss": 1.155, + "mean_token_accuracy": 0.6877422332763672, + "num_tokens": 416605.0, + "step": 11 + }, + { + "epoch": 0.0015265233430861213, + "grad_norm": 31.875, + "learning_rate": 4.662992793556591e-09, + "loss": 1.2857, + "mean_token_accuracy": 0.6711574792861938, + "num_tokens": 448798.0, + "step": 12 + }, + { + "epoch": 0.0016537336216766315, + "grad_norm": 35.75, + "learning_rate": 5.086901229334464e-09, + "loss": 1.3728, + "mean_token_accuracy": 0.6446583271026611, + "num_tokens": 480084.0, + "step": 13 + }, + { + "epoch": 0.0017809439002671415, + "grad_norm": 27.875, + "learning_rate": 5.510809665112336e-09, + "loss": 1.2149, + "mean_token_accuracy": 0.6816909313201904, + "num_tokens": 524543.0, + "step": 14 + }, + { + "epoch": 0.0019081541788576518, + "grad_norm": 29.0, + "learning_rate": 5.934718100890208e-09, + "loss": 1.1592, + "mean_token_accuracy": 0.6922162175178528, + "num_tokens": 563314.0, + "step": 15 + }, + { + "epoch": 0.002035364457448162, + "grad_norm": 32.5, + "learning_rate": 6.3586265366680796e-09, + "loss": 1.2888, + "mean_token_accuracy": 0.669562816619873, + "num_tokens": 598421.0, + "step": 16 + }, + { + "epoch": 0.002162574736038672, + "grad_norm": 34.0, + "learning_rate": 6.782534972445951e-09, + "loss": 1.3056, + "mean_token_accuracy": 0.6631807088851929, + "num_tokens": 634690.0, + "step": 17 + }, + { + "epoch": 0.002289785014629182, + "grad_norm": 32.25, + "learning_rate": 7.206443408223823e-09, + "loss": 1.3316, + "mean_token_accuracy": 0.6595425009727478, + "num_tokens": 674653.0, + "step": 18 + }, + { + "epoch": 0.0024169952932196924, + "grad_norm": 28.875, + "learning_rate": 7.630351844001695e-09, + "loss": 1.2155, + "mean_token_accuracy": 0.6842992305755615, + "num_tokens": 708238.0, + "step": 19 + }, + { + "epoch": 0.0025442055718102024, + "grad_norm": 27.75, + "learning_rate": 8.054260279779567e-09, + "loss": 1.1795, + "mean_token_accuracy": 0.6884785294532776, + "num_tokens": 749312.0, + "step": 20 + }, + { + "epoch": 0.0026714158504007124, + "grad_norm": 33.75, + "learning_rate": 8.478168715557438e-09, + "loss": 1.337, + "mean_token_accuracy": 0.6617263555526733, + "num_tokens": 783532.0, + "step": 21 + }, + { + "epoch": 0.0027986261289912225, + "grad_norm": 31.75, + "learning_rate": 8.902077151335311e-09, + "loss": 1.2745, + "mean_token_accuracy": 0.6728295087814331, + "num_tokens": 817429.0, + "step": 22 + }, + { + "epoch": 0.0029258364075817325, + "grad_norm": 28.5, + "learning_rate": 9.325985587113182e-09, + "loss": 1.176, + "mean_token_accuracy": 0.6932163238525391, + "num_tokens": 853964.0, + "step": 23 + }, + { + "epoch": 0.0030530466861722425, + "grad_norm": 30.0, + "learning_rate": 9.749894022891054e-09, + "loss": 1.2706, + "mean_token_accuracy": 0.6693286299705505, + "num_tokens": 885070.0, + "step": 24 + }, + { + "epoch": 0.003180256964762753, + "grad_norm": 36.5, + "learning_rate": 1.0173802458668929e-08, + "loss": 1.3737, + "mean_token_accuracy": 0.641587495803833, + "num_tokens": 926893.0, + "step": 25 + }, + { + "epoch": 0.003307467243353263, + "grad_norm": 31.375, + "learning_rate": 1.05977108944468e-08, + "loss": 1.2805, + "mean_token_accuracy": 0.6626410484313965, + "num_tokens": 964773.0, + "step": 26 + }, + { + "epoch": 0.003434677521943773, + "grad_norm": 29.25, + "learning_rate": 1.1021619330224672e-08, + "loss": 1.1839, + "mean_token_accuracy": 0.6923789978027344, + "num_tokens": 1002725.0, + "step": 27 + }, + { + "epoch": 0.003561887800534283, + "grad_norm": 32.75, + "learning_rate": 1.1445527766002543e-08, + "loss": 1.3351, + "mean_token_accuracy": 0.6538394689559937, + "num_tokens": 1040296.0, + "step": 28 + }, + { + "epoch": 0.003689098079124793, + "grad_norm": 28.125, + "learning_rate": 1.1869436201780416e-08, + "loss": 1.1696, + "mean_token_accuracy": 0.6941937804222107, + "num_tokens": 1081711.0, + "step": 29 + }, + { + "epoch": 0.0038163083577153036, + "grad_norm": 29.375, + "learning_rate": 1.2293344637558287e-08, + "loss": 1.2352, + "mean_token_accuracy": 0.6851239204406738, + "num_tokens": 1120556.0, + "step": 30 + }, + { + "epoch": 0.003943518636305814, + "grad_norm": 28.25, + "learning_rate": 1.2717253073336159e-08, + "loss": 1.1495, + "mean_token_accuracy": 0.6981396675109863, + "num_tokens": 1157723.0, + "step": 31 + }, + { + "epoch": 0.004070728914896324, + "grad_norm": 28.75, + "learning_rate": 1.314116150911403e-08, + "loss": 1.2123, + "mean_token_accuracy": 0.6823868155479431, + "num_tokens": 1197879.0, + "step": 32 + }, + { + "epoch": 0.004197939193486834, + "grad_norm": 29.75, + "learning_rate": 1.3565069944891903e-08, + "loss": 1.2206, + "mean_token_accuracy": 0.6850752830505371, + "num_tokens": 1237342.0, + "step": 33 + }, + { + "epoch": 0.004325149472077344, + "grad_norm": 29.75, + "learning_rate": 1.3988978380669775e-08, + "loss": 1.1978, + "mean_token_accuracy": 0.6881247162818909, + "num_tokens": 1280197.0, + "step": 34 + }, + { + "epoch": 0.004452359750667854, + "grad_norm": 28.5, + "learning_rate": 1.4412886816447646e-08, + "loss": 1.1701, + "mean_token_accuracy": 0.6932596564292908, + "num_tokens": 1318625.0, + "step": 35 + }, + { + "epoch": 0.004579570029258364, + "grad_norm": 32.25, + "learning_rate": 1.4836795252225519e-08, + "loss": 1.269, + "mean_token_accuracy": 0.6720812916755676, + "num_tokens": 1356868.0, + "step": 36 + }, + { + "epoch": 0.004706780307848874, + "grad_norm": 30.375, + "learning_rate": 1.526070368800339e-08, + "loss": 1.2207, + "mean_token_accuracy": 0.6806917190551758, + "num_tokens": 1394696.0, + "step": 37 + }, + { + "epoch": 0.004833990586439385, + "grad_norm": 29.375, + "learning_rate": 1.5684612123781262e-08, + "loss": 1.1979, + "mean_token_accuracy": 0.6867342591285706, + "num_tokens": 1438738.0, + "step": 38 + }, + { + "epoch": 0.004961200865029895, + "grad_norm": 27.625, + "learning_rate": 1.6108520559559135e-08, + "loss": 1.1297, + "mean_token_accuracy": 0.7074792981147766, + "num_tokens": 1475089.0, + "step": 39 + }, + { + "epoch": 0.005088411143620405, + "grad_norm": 29.25, + "learning_rate": 1.6532428995337004e-08, + "loss": 1.2259, + "mean_token_accuracy": 0.6831930875778198, + "num_tokens": 1514566.0, + "step": 40 + }, + { + "epoch": 0.005215621422210915, + "grad_norm": 29.75, + "learning_rate": 1.6956337431114877e-08, + "loss": 1.2474, + "mean_token_accuracy": 0.6790611743927002, + "num_tokens": 1552560.0, + "step": 41 + }, + { + "epoch": 0.005342831700801425, + "grad_norm": 30.375, + "learning_rate": 1.738024586689275e-08, + "loss": 1.2247, + "mean_token_accuracy": 0.6840267777442932, + "num_tokens": 1584759.0, + "step": 42 + }, + { + "epoch": 0.005470041979391935, + "grad_norm": 30.375, + "learning_rate": 1.7804154302670622e-08, + "loss": 1.2308, + "mean_token_accuracy": 0.6772377490997314, + "num_tokens": 1621825.0, + "step": 43 + }, + { + "epoch": 0.005597252257982445, + "grad_norm": 30.5, + "learning_rate": 1.8228062738448494e-08, + "loss": 1.2535, + "mean_token_accuracy": 0.6696719527244568, + "num_tokens": 1662946.0, + "step": 44 + }, + { + "epoch": 0.005724462536572955, + "grad_norm": 32.0, + "learning_rate": 1.8651971174226364e-08, + "loss": 1.2837, + "mean_token_accuracy": 0.6704589128494263, + "num_tokens": 1699433.0, + "step": 45 + }, + { + "epoch": 0.005851672815163465, + "grad_norm": 28.375, + "learning_rate": 1.9075879610004236e-08, + "loss": 1.1467, + "mean_token_accuracy": 0.6924684047698975, + "num_tokens": 1742812.0, + "step": 46 + }, + { + "epoch": 0.005978883093753975, + "grad_norm": 27.375, + "learning_rate": 1.949978804578211e-08, + "loss": 1.201, + "mean_token_accuracy": 0.6909250020980835, + "num_tokens": 1778725.0, + "step": 47 + }, + { + "epoch": 0.006106093372344485, + "grad_norm": 32.5, + "learning_rate": 1.9923696481559985e-08, + "loss": 1.2966, + "mean_token_accuracy": 0.66363924741745, + "num_tokens": 1816592.0, + "step": 48 + }, + { + "epoch": 0.006233303650934996, + "grad_norm": 29.25, + "learning_rate": 2.0347604917337857e-08, + "loss": 1.1376, + "mean_token_accuracy": 0.6998938322067261, + "num_tokens": 1859907.0, + "step": 49 + }, + { + "epoch": 0.006360513929525506, + "grad_norm": 26.75, + "learning_rate": 2.0771513353115727e-08, + "loss": 1.1186, + "mean_token_accuracy": 0.70852130651474, + "num_tokens": 1896627.0, + "step": 50 + }, + { + "epoch": 0.006487724208116016, + "grad_norm": 31.125, + "learning_rate": 2.11954217888936e-08, + "loss": 1.2865, + "mean_token_accuracy": 0.668583333492279, + "num_tokens": 1934041.0, + "step": 51 + }, + { + "epoch": 0.006614934486706526, + "grad_norm": 28.625, + "learning_rate": 2.1619330224671472e-08, + "loss": 1.1751, + "mean_token_accuracy": 0.6902894973754883, + "num_tokens": 1976482.0, + "step": 52 + }, + { + "epoch": 0.006742144765297036, + "grad_norm": 28.875, + "learning_rate": 2.2043238660449344e-08, + "loss": 1.2153, + "mean_token_accuracy": 0.6876301765441895, + "num_tokens": 2009224.0, + "step": 53 + }, + { + "epoch": 0.006869355043887546, + "grad_norm": 31.25, + "learning_rate": 2.2467147096227214e-08, + "loss": 1.302, + "mean_token_accuracy": 0.6653440594673157, + "num_tokens": 2049235.0, + "step": 54 + }, + { + "epoch": 0.006996565322478056, + "grad_norm": 32.5, + "learning_rate": 2.2891055532005086e-08, + "loss": 1.3172, + "mean_token_accuracy": 0.659644365310669, + "num_tokens": 2090260.0, + "step": 55 + }, + { + "epoch": 0.007123775601068566, + "grad_norm": 32.25, + "learning_rate": 2.331496396778296e-08, + "loss": 1.3244, + "mean_token_accuracy": 0.661136269569397, + "num_tokens": 2126686.0, + "step": 56 + }, + { + "epoch": 0.007250985879659076, + "grad_norm": 27.875, + "learning_rate": 2.373887240356083e-08, + "loss": 1.1504, + "mean_token_accuracy": 0.6963536739349365, + "num_tokens": 2171355.0, + "step": 57 + }, + { + "epoch": 0.007378196158249586, + "grad_norm": 27.875, + "learning_rate": 2.4162780839338704e-08, + "loss": 1.1337, + "mean_token_accuracy": 0.7024140357971191, + "num_tokens": 2211660.0, + "step": 58 + }, + { + "epoch": 0.007505406436840096, + "grad_norm": 34.0, + "learning_rate": 2.4586689275116573e-08, + "loss": 1.3553, + "mean_token_accuracy": 0.6570191979408264, + "num_tokens": 2244411.0, + "step": 59 + }, + { + "epoch": 0.007632616715430607, + "grad_norm": 29.25, + "learning_rate": 2.5010597710894446e-08, + "loss": 1.2091, + "mean_token_accuracy": 0.6762730479240417, + "num_tokens": 2285253.0, + "step": 60 + }, + { + "epoch": 0.007759826994021117, + "grad_norm": 28.375, + "learning_rate": 2.5434506146672318e-08, + "loss": 1.1626, + "mean_token_accuracy": 0.6921408772468567, + "num_tokens": 2328577.0, + "step": 61 + }, + { + "epoch": 0.007887037272611627, + "grad_norm": 27.625, + "learning_rate": 2.585841458245019e-08, + "loss": 1.2511, + "mean_token_accuracy": 0.6794916391372681, + "num_tokens": 2366524.0, + "step": 62 + }, + { + "epoch": 0.008014247551202136, + "grad_norm": 27.75, + "learning_rate": 2.628232301822806e-08, + "loss": 1.1453, + "mean_token_accuracy": 0.6973813772201538, + "num_tokens": 2408628.0, + "step": 63 + }, + { + "epoch": 0.008141457829792647, + "grad_norm": 28.375, + "learning_rate": 2.6706231454005933e-08, + "loss": 1.1962, + "mean_token_accuracy": 0.6793848872184753, + "num_tokens": 2451800.0, + "step": 64 + }, + { + "epoch": 0.008268668108383158, + "grad_norm": 33.75, + "learning_rate": 2.7130139889783805e-08, + "loss": 1.3404, + "mean_token_accuracy": 0.6563202142715454, + "num_tokens": 2481448.0, + "step": 65 + }, + { + "epoch": 0.008395878386973667, + "grad_norm": 28.875, + "learning_rate": 2.7554048325561678e-08, + "loss": 1.2075, + "mean_token_accuracy": 0.6841487884521484, + "num_tokens": 2526339.0, + "step": 66 + }, + { + "epoch": 0.008523088665564178, + "grad_norm": 29.875, + "learning_rate": 2.797795676133955e-08, + "loss": 1.2373, + "mean_token_accuracy": 0.671256422996521, + "num_tokens": 2570691.0, + "step": 67 + }, + { + "epoch": 0.008650298944154687, + "grad_norm": 29.125, + "learning_rate": 2.840186519711742e-08, + "loss": 1.1982, + "mean_token_accuracy": 0.6838635802268982, + "num_tokens": 2609207.0, + "step": 68 + }, + { + "epoch": 0.008777509222745198, + "grad_norm": 30.875, + "learning_rate": 2.8825773632895292e-08, + "loss": 1.2176, + "mean_token_accuracy": 0.6777025461196899, + "num_tokens": 2645494.0, + "step": 69 + }, + { + "epoch": 0.008904719501335707, + "grad_norm": 27.25, + "learning_rate": 2.9249682068673165e-08, + "loss": 1.1351, + "mean_token_accuracy": 0.7048642635345459, + "num_tokens": 2686258.0, + "step": 70 + }, + { + "epoch": 0.009031929779926218, + "grad_norm": 31.0, + "learning_rate": 2.9673590504451037e-08, + "loss": 1.2399, + "mean_token_accuracy": 0.6771516799926758, + "num_tokens": 2720544.0, + "step": 71 + }, + { + "epoch": 0.009159140058516728, + "grad_norm": 28.875, + "learning_rate": 3.0097498940228907e-08, + "loss": 1.1785, + "mean_token_accuracy": 0.6958903074264526, + "num_tokens": 2758068.0, + "step": 72 + }, + { + "epoch": 0.009286350337107238, + "grad_norm": 35.25, + "learning_rate": 3.052140737600678e-08, + "loss": 1.3416, + "mean_token_accuracy": 0.662199854850769, + "num_tokens": 2793342.0, + "step": 73 + }, + { + "epoch": 0.009413560615697748, + "grad_norm": 34.0, + "learning_rate": 3.094531581178465e-08, + "loss": 1.3659, + "mean_token_accuracy": 0.6518688201904297, + "num_tokens": 2828003.0, + "step": 74 + }, + { + "epoch": 0.009540770894288259, + "grad_norm": 28.75, + "learning_rate": 3.1369224247562524e-08, + "loss": 1.1565, + "mean_token_accuracy": 0.6964083909988403, + "num_tokens": 2874755.0, + "step": 75 + }, + { + "epoch": 0.00966798117287877, + "grad_norm": 29.625, + "learning_rate": 3.17931326833404e-08, + "loss": 1.209, + "mean_token_accuracy": 0.6815237998962402, + "num_tokens": 2909703.0, + "step": 76 + }, + { + "epoch": 0.009795191451469279, + "grad_norm": 25.625, + "learning_rate": 3.221704111911827e-08, + "loss": 1.0832, + "mean_token_accuracy": 0.7115992903709412, + "num_tokens": 2954020.0, + "step": 77 + }, + { + "epoch": 0.00992240173005979, + "grad_norm": 29.625, + "learning_rate": 3.264094955489614e-08, + "loss": 1.2568, + "mean_token_accuracy": 0.676813006401062, + "num_tokens": 2984894.0, + "step": 78 + }, + { + "epoch": 0.010049612008650299, + "grad_norm": 26.75, + "learning_rate": 3.306485799067401e-08, + "loss": 1.1199, + "mean_token_accuracy": 0.7107468843460083, + "num_tokens": 3017773.0, + "step": 79 + }, + { + "epoch": 0.01017682228724081, + "grad_norm": 26.625, + "learning_rate": 3.348876642645188e-08, + "loss": 1.0993, + "mean_token_accuracy": 0.7060564160346985, + "num_tokens": 3059739.0, + "step": 80 + }, + { + "epoch": 0.010304032565831319, + "grad_norm": 30.0, + "learning_rate": 3.391267486222975e-08, + "loss": 1.2489, + "mean_token_accuracy": 0.6756494045257568, + "num_tokens": 3094910.0, + "step": 81 + }, + { + "epoch": 0.01043124284442183, + "grad_norm": 31.125, + "learning_rate": 3.4336583298007626e-08, + "loss": 1.262, + "mean_token_accuracy": 0.6751126646995544, + "num_tokens": 3133884.0, + "step": 82 + }, + { + "epoch": 0.010558453123012339, + "grad_norm": 33.5, + "learning_rate": 3.47604917337855e-08, + "loss": 1.3377, + "mean_token_accuracy": 0.6497499942779541, + "num_tokens": 3166755.0, + "step": 83 + }, + { + "epoch": 0.01068566340160285, + "grad_norm": 31.75, + "learning_rate": 3.518440016956337e-08, + "loss": 1.2756, + "mean_token_accuracy": 0.6652512550354004, + "num_tokens": 3204825.0, + "step": 84 + }, + { + "epoch": 0.010812873680193359, + "grad_norm": 33.0, + "learning_rate": 3.5608308605341244e-08, + "loss": 1.3382, + "mean_token_accuracy": 0.6597685813903809, + "num_tokens": 3244036.0, + "step": 85 + }, + { + "epoch": 0.01094008395878387, + "grad_norm": 33.0, + "learning_rate": 3.6032217041119116e-08, + "loss": 1.2866, + "mean_token_accuracy": 0.6628565788269043, + "num_tokens": 3285969.0, + "step": 86 + }, + { + "epoch": 0.01106729423737438, + "grad_norm": 29.25, + "learning_rate": 3.645612547689699e-08, + "loss": 1.2307, + "mean_token_accuracy": 0.6810464859008789, + "num_tokens": 3327648.0, + "step": 87 + }, + { + "epoch": 0.01119450451596489, + "grad_norm": 28.875, + "learning_rate": 3.6880033912674855e-08, + "loss": 1.2268, + "mean_token_accuracy": 0.6782336831092834, + "num_tokens": 3367399.0, + "step": 88 + }, + { + "epoch": 0.0113217147945554, + "grad_norm": 30.625, + "learning_rate": 3.730394234845273e-08, + "loss": 1.2377, + "mean_token_accuracy": 0.6805353164672852, + "num_tokens": 3405402.0, + "step": 89 + }, + { + "epoch": 0.01144892507314591, + "grad_norm": 35.75, + "learning_rate": 3.77278507842306e-08, + "loss": 1.37, + "mean_token_accuracy": 0.6426373720169067, + "num_tokens": 3441791.0, + "step": 90 + }, + { + "epoch": 0.01157613535173642, + "grad_norm": 33.25, + "learning_rate": 3.815175922000847e-08, + "loss": 1.3032, + "mean_token_accuracy": 0.6690061688423157, + "num_tokens": 3480151.0, + "step": 91 + }, + { + "epoch": 0.01170334563032693, + "grad_norm": 29.125, + "learning_rate": 3.8575667655786345e-08, + "loss": 1.2322, + "mean_token_accuracy": 0.6792813539505005, + "num_tokens": 3516867.0, + "step": 92 + }, + { + "epoch": 0.01183055590891744, + "grad_norm": 30.25, + "learning_rate": 3.899957609156422e-08, + "loss": 1.2121, + "mean_token_accuracy": 0.6839563846588135, + "num_tokens": 3557466.0, + "step": 93 + }, + { + "epoch": 0.01195776618750795, + "grad_norm": 30.625, + "learning_rate": 3.94234845273421e-08, + "loss": 1.2462, + "mean_token_accuracy": 0.6767478585243225, + "num_tokens": 3596009.0, + "step": 94 + }, + { + "epoch": 0.012084976466098461, + "grad_norm": 29.875, + "learning_rate": 3.984739296311997e-08, + "loss": 1.2284, + "mean_token_accuracy": 0.679149329662323, + "num_tokens": 3631582.0, + "step": 95 + }, + { + "epoch": 0.01221218674468897, + "grad_norm": 30.625, + "learning_rate": 4.027130139889784e-08, + "loss": 1.2593, + "mean_token_accuracy": 0.6699120402336121, + "num_tokens": 3672855.0, + "step": 96 + }, + { + "epoch": 0.012339397023279481, + "grad_norm": 24.875, + "learning_rate": 4.0695209834675715e-08, + "loss": 1.1829, + "mean_token_accuracy": 0.6905128359794617, + "num_tokens": 3714100.0, + "step": 97 + }, + { + "epoch": 0.012466607301869992, + "grad_norm": 27.625, + "learning_rate": 4.111911827045358e-08, + "loss": 1.1728, + "mean_token_accuracy": 0.6945182085037231, + "num_tokens": 3752834.0, + "step": 98 + }, + { + "epoch": 0.012593817580460501, + "grad_norm": 27.75, + "learning_rate": 4.154302670623145e-08, + "loss": 1.1875, + "mean_token_accuracy": 0.6914156079292297, + "num_tokens": 3791513.0, + "step": 99 + }, + { + "epoch": 0.012721027859051012, + "grad_norm": 29.75, + "learning_rate": 4.1966935142009326e-08, + "loss": 1.2381, + "mean_token_accuracy": 0.6817656755447388, + "num_tokens": 3825085.0, + "step": 100 + }, + { + "epoch": 0.012848238137641521, + "grad_norm": 30.25, + "learning_rate": 4.23908435777872e-08, + "loss": 1.2452, + "mean_token_accuracy": 0.6809250116348267, + "num_tokens": 3866492.0, + "step": 101 + }, + { + "epoch": 0.012975448416232032, + "grad_norm": 31.5, + "learning_rate": 4.281475201356507e-08, + "loss": 1.2897, + "mean_token_accuracy": 0.6654545068740845, + "num_tokens": 3904053.0, + "step": 102 + }, + { + "epoch": 0.013102658694822541, + "grad_norm": 27.875, + "learning_rate": 4.3238660449342943e-08, + "loss": 1.1545, + "mean_token_accuracy": 0.7089903354644775, + "num_tokens": 3935118.0, + "step": 103 + }, + { + "epoch": 0.013229868973413052, + "grad_norm": 30.25, + "learning_rate": 4.3662568885120816e-08, + "loss": 1.2305, + "mean_token_accuracy": 0.6757335662841797, + "num_tokens": 3977899.0, + "step": 104 + }, + { + "epoch": 0.013357079252003561, + "grad_norm": 31.625, + "learning_rate": 4.408647732089869e-08, + "loss": 1.2747, + "mean_token_accuracy": 0.6714427471160889, + "num_tokens": 4019416.0, + "step": 105 + }, + { + "epoch": 0.013484289530594072, + "grad_norm": 29.5, + "learning_rate": 4.451038575667656e-08, + "loss": 1.2254, + "mean_token_accuracy": 0.680296778678894, + "num_tokens": 4063887.0, + "step": 106 + }, + { + "epoch": 0.013611499809184581, + "grad_norm": 33.5, + "learning_rate": 4.493429419245443e-08, + "loss": 1.3073, + "mean_token_accuracy": 0.6637348532676697, + "num_tokens": 4101309.0, + "step": 107 + }, + { + "epoch": 0.013738710087775092, + "grad_norm": 31.125, + "learning_rate": 4.53582026282323e-08, + "loss": 1.2663, + "mean_token_accuracy": 0.6681212186813354, + "num_tokens": 4136669.0, + "step": 108 + }, + { + "epoch": 0.013865920366365603, + "grad_norm": 24.625, + "learning_rate": 4.578211106401017e-08, + "loss": 1.1267, + "mean_token_accuracy": 0.7057394981384277, + "num_tokens": 4178583.0, + "step": 109 + }, + { + "epoch": 0.013993130644956112, + "grad_norm": 30.375, + "learning_rate": 4.6206019499788045e-08, + "loss": 1.2034, + "mean_token_accuracy": 0.6798733472824097, + "num_tokens": 4219552.0, + "step": 110 + }, + { + "epoch": 0.014120340923546623, + "grad_norm": 27.375, + "learning_rate": 4.662992793556592e-08, + "loss": 1.172, + "mean_token_accuracy": 0.6963104009628296, + "num_tokens": 4264487.0, + "step": 111 + }, + { + "epoch": 0.014247551202137132, + "grad_norm": 26.125, + "learning_rate": 4.705383637134379e-08, + "loss": 1.1266, + "mean_token_accuracy": 0.703598141670227, + "num_tokens": 4303015.0, + "step": 112 + }, + { + "epoch": 0.014374761480727643, + "grad_norm": 29.5, + "learning_rate": 4.747774480712166e-08, + "loss": 1.2224, + "mean_token_accuracy": 0.6757254600524902, + "num_tokens": 4345446.0, + "step": 113 + }, + { + "epoch": 0.014501971759318152, + "grad_norm": 29.375, + "learning_rate": 4.7901653242899535e-08, + "loss": 1.1533, + "mean_token_accuracy": 0.6975631713867188, + "num_tokens": 4383854.0, + "step": 114 + }, + { + "epoch": 0.014629182037908663, + "grad_norm": 30.25, + "learning_rate": 4.832556167867741e-08, + "loss": 1.2452, + "mean_token_accuracy": 0.6776224374771118, + "num_tokens": 4420750.0, + "step": 115 + }, + { + "epoch": 0.014756392316499172, + "grad_norm": 31.125, + "learning_rate": 4.8749470114455274e-08, + "loss": 1.2598, + "mean_token_accuracy": 0.6724961996078491, + "num_tokens": 4458761.0, + "step": 116 + }, + { + "epoch": 0.014883602595089683, + "grad_norm": 28.25, + "learning_rate": 4.9173378550233146e-08, + "loss": 1.1898, + "mean_token_accuracy": 0.6878355741500854, + "num_tokens": 4496547.0, + "step": 117 + }, + { + "epoch": 0.015010812873680193, + "grad_norm": 28.625, + "learning_rate": 4.959728698601102e-08, + "loss": 1.1924, + "mean_token_accuracy": 0.6959755420684814, + "num_tokens": 4533357.0, + "step": 118 + }, + { + "epoch": 0.015138023152270703, + "grad_norm": 28.5, + "learning_rate": 5.002119542178889e-08, + "loss": 1.1555, + "mean_token_accuracy": 0.6991840600967407, + "num_tokens": 4573570.0, + "step": 119 + }, + { + "epoch": 0.015265233430861214, + "grad_norm": 30.5, + "learning_rate": 5.0445103857566764e-08, + "loss": 1.2336, + "mean_token_accuracy": 0.6735767722129822, + "num_tokens": 4613195.0, + "step": 120 + }, + { + "epoch": 0.015392443709451724, + "grad_norm": 26.75, + "learning_rate": 5.0869012293344637e-08, + "loss": 1.106, + "mean_token_accuracy": 0.7066845893859863, + "num_tokens": 4657230.0, + "step": 121 + }, + { + "epoch": 0.015519653988042234, + "grad_norm": 29.875, + "learning_rate": 5.129292072912251e-08, + "loss": 1.2335, + "mean_token_accuracy": 0.6804032325744629, + "num_tokens": 4701436.0, + "step": 122 + }, + { + "epoch": 0.015646864266632744, + "grad_norm": 31.875, + "learning_rate": 5.171682916490038e-08, + "loss": 1.2413, + "mean_token_accuracy": 0.6732272505760193, + "num_tokens": 4741806.0, + "step": 123 + }, + { + "epoch": 0.015774074545223254, + "grad_norm": 31.5, + "learning_rate": 5.2140737600678254e-08, + "loss": 1.2799, + "mean_token_accuracy": 0.6625981330871582, + "num_tokens": 4780671.0, + "step": 124 + }, + { + "epoch": 0.015901284823813765, + "grad_norm": 28.25, + "learning_rate": 5.256464603645612e-08, + "loss": 1.1898, + "mean_token_accuracy": 0.6885746717453003, + "num_tokens": 4816879.0, + "step": 125 + }, + { + "epoch": 0.016028495102404273, + "grad_norm": 29.625, + "learning_rate": 5.298855447223399e-08, + "loss": 1.2726, + "mean_token_accuracy": 0.6686959266662598, + "num_tokens": 4858704.0, + "step": 126 + }, + { + "epoch": 0.016155705380994784, + "grad_norm": 29.25, + "learning_rate": 5.3412462908011865e-08, + "loss": 1.2175, + "mean_token_accuracy": 0.6816202998161316, + "num_tokens": 4896780.0, + "step": 127 + }, + { + "epoch": 0.016282915659585295, + "grad_norm": 30.625, + "learning_rate": 5.383637134378974e-08, + "loss": 1.2557, + "mean_token_accuracy": 0.6775717735290527, + "num_tokens": 4932108.0, + "step": 128 + }, + { + "epoch": 0.016410125938175806, + "grad_norm": 28.0, + "learning_rate": 5.426027977956761e-08, + "loss": 1.2069, + "mean_token_accuracy": 0.6855578422546387, + "num_tokens": 4970384.0, + "step": 129 + }, + { + "epoch": 0.016537336216766316, + "grad_norm": 31.75, + "learning_rate": 5.468418821534548e-08, + "loss": 1.244, + "mean_token_accuracy": 0.6807116866111755, + "num_tokens": 5005996.0, + "step": 130 + }, + { + "epoch": 0.016664546495356824, + "grad_norm": 27.25, + "learning_rate": 5.5108096651123356e-08, + "loss": 1.189, + "mean_token_accuracy": 0.6875318884849548, + "num_tokens": 5044208.0, + "step": 131 + }, + { + "epoch": 0.016791756773947335, + "grad_norm": 29.125, + "learning_rate": 5.553200508690123e-08, + "loss": 1.2138, + "mean_token_accuracy": 0.6799690127372742, + "num_tokens": 5085173.0, + "step": 132 + }, + { + "epoch": 0.016918967052537846, + "grad_norm": 31.75, + "learning_rate": 5.59559135226791e-08, + "loss": 1.2657, + "mean_token_accuracy": 0.6682533025741577, + "num_tokens": 5118440.0, + "step": 133 + }, + { + "epoch": 0.017046177331128357, + "grad_norm": 30.75, + "learning_rate": 5.637982195845697e-08, + "loss": 1.2804, + "mean_token_accuracy": 0.6665184497833252, + "num_tokens": 5155202.0, + "step": 134 + }, + { + "epoch": 0.017173387609718864, + "grad_norm": 27.875, + "learning_rate": 5.680373039423484e-08, + "loss": 1.2124, + "mean_token_accuracy": 0.6891870498657227, + "num_tokens": 5193240.0, + "step": 135 + }, + { + "epoch": 0.017300597888309375, + "grad_norm": 28.875, + "learning_rate": 5.722763883001271e-08, + "loss": 1.2225, + "mean_token_accuracy": 0.6826634407043457, + "num_tokens": 5230975.0, + "step": 136 + }, + { + "epoch": 0.017427808166899886, + "grad_norm": 32.25, + "learning_rate": 5.7651547265790585e-08, + "loss": 1.3442, + "mean_token_accuracy": 0.6557543873786926, + "num_tokens": 5264786.0, + "step": 137 + }, + { + "epoch": 0.017555018445490397, + "grad_norm": 29.0, + "learning_rate": 5.807545570156846e-08, + "loss": 1.2629, + "mean_token_accuracy": 0.679882824420929, + "num_tokens": 5301554.0, + "step": 138 + }, + { + "epoch": 0.017682228724080904, + "grad_norm": 30.375, + "learning_rate": 5.849936413734633e-08, + "loss": 1.2103, + "mean_token_accuracy": 0.6885099411010742, + "num_tokens": 5337356.0, + "step": 139 + }, + { + "epoch": 0.017809439002671415, + "grad_norm": 28.375, + "learning_rate": 5.89232725731242e-08, + "loss": 1.2462, + "mean_token_accuracy": 0.680656909942627, + "num_tokens": 5375767.0, + "step": 140 + }, + { + "epoch": 0.017936649281261926, + "grad_norm": 27.5, + "learning_rate": 5.9347181008902075e-08, + "loss": 1.1955, + "mean_token_accuracy": 0.6882649660110474, + "num_tokens": 5413683.0, + "step": 141 + }, + { + "epoch": 0.018063859559852437, + "grad_norm": 28.25, + "learning_rate": 5.977108944467995e-08, + "loss": 1.1638, + "mean_token_accuracy": 0.694760799407959, + "num_tokens": 5457750.0, + "step": 142 + }, + { + "epoch": 0.018191069838442948, + "grad_norm": 28.25, + "learning_rate": 6.019499788045781e-08, + "loss": 1.1811, + "mean_token_accuracy": 0.6888551712036133, + "num_tokens": 5495743.0, + "step": 143 + }, + { + "epoch": 0.018318280117033455, + "grad_norm": 29.5, + "learning_rate": 6.061890631623569e-08, + "loss": 1.1982, + "mean_token_accuracy": 0.6781882047653198, + "num_tokens": 5532198.0, + "step": 144 + }, + { + "epoch": 0.018445490395623966, + "grad_norm": 30.125, + "learning_rate": 6.104281475201356e-08, + "loss": 1.2554, + "mean_token_accuracy": 0.6783879995346069, + "num_tokens": 5568977.0, + "step": 145 + }, + { + "epoch": 0.018572700674214477, + "grad_norm": 29.875, + "learning_rate": 6.146672318779143e-08, + "loss": 1.2517, + "mean_token_accuracy": 0.6730386018753052, + "num_tokens": 5606229.0, + "step": 146 + }, + { + "epoch": 0.018699910952804988, + "grad_norm": 32.0, + "learning_rate": 6.18906316235693e-08, + "loss": 1.2521, + "mean_token_accuracy": 0.6699821949005127, + "num_tokens": 5649828.0, + "step": 147 + }, + { + "epoch": 0.018827121231395495, + "grad_norm": 27.5, + "learning_rate": 6.231454005934718e-08, + "loss": 1.1577, + "mean_token_accuracy": 0.6989731788635254, + "num_tokens": 5687433.0, + "step": 148 + }, + { + "epoch": 0.018954331509986006, + "grad_norm": 30.125, + "learning_rate": 6.273844849512505e-08, + "loss": 1.2436, + "mean_token_accuracy": 0.6842092275619507, + "num_tokens": 5723247.0, + "step": 149 + }, + { + "epoch": 0.019081541788576517, + "grad_norm": 31.0, + "learning_rate": 6.316235693090292e-08, + "loss": 1.2583, + "mean_token_accuracy": 0.6749197244644165, + "num_tokens": 5760305.0, + "step": 150 + }, + { + "epoch": 0.019208752067167028, + "grad_norm": 32.25, + "learning_rate": 6.35862653666808e-08, + "loss": 1.2419, + "mean_token_accuracy": 0.6765167713165283, + "num_tokens": 5800586.0, + "step": 151 + }, + { + "epoch": 0.01933596234575754, + "grad_norm": 30.625, + "learning_rate": 6.401017380245867e-08, + "loss": 1.2336, + "mean_token_accuracy": 0.6733064651489258, + "num_tokens": 5840351.0, + "step": 152 + }, + { + "epoch": 0.019463172624348046, + "grad_norm": 33.75, + "learning_rate": 6.443408223823654e-08, + "loss": 1.3203, + "mean_token_accuracy": 0.6592285633087158, + "num_tokens": 5880162.0, + "step": 153 + }, + { + "epoch": 0.019590382902938557, + "grad_norm": 33.25, + "learning_rate": 6.485799067401441e-08, + "loss": 1.3935, + "mean_token_accuracy": 0.6441901922225952, + "num_tokens": 5922213.0, + "step": 154 + }, + { + "epoch": 0.019717593181529068, + "grad_norm": 29.25, + "learning_rate": 6.528189910979228e-08, + "loss": 1.2045, + "mean_token_accuracy": 0.6831943988800049, + "num_tokens": 5957461.0, + "step": 155 + }, + { + "epoch": 0.01984480346011958, + "grad_norm": 28.375, + "learning_rate": 6.570580754557016e-08, + "loss": 1.1763, + "mean_token_accuracy": 0.6864670515060425, + "num_tokens": 5994675.0, + "step": 156 + }, + { + "epoch": 0.019972013738710086, + "grad_norm": 25.5, + "learning_rate": 6.612971598134802e-08, + "loss": 1.1, + "mean_token_accuracy": 0.7117395401000977, + "num_tokens": 6032514.0, + "step": 157 + }, + { + "epoch": 0.020099224017300597, + "grad_norm": 30.625, + "learning_rate": 6.655362441712589e-08, + "loss": 1.2692, + "mean_token_accuracy": 0.668988823890686, + "num_tokens": 6069160.0, + "step": 158 + }, + { + "epoch": 0.020226434295891108, + "grad_norm": 34.75, + "learning_rate": 6.697753285290376e-08, + "loss": 1.3672, + "mean_token_accuracy": 0.6466127634048462, + "num_tokens": 6106652.0, + "step": 159 + }, + { + "epoch": 0.02035364457448162, + "grad_norm": 31.25, + "learning_rate": 6.740144128868163e-08, + "loss": 1.2487, + "mean_token_accuracy": 0.6811245083808899, + "num_tokens": 6143070.0, + "step": 160 + }, + { + "epoch": 0.020480854853072127, + "grad_norm": 28.125, + "learning_rate": 6.78253497244595e-08, + "loss": 1.1853, + "mean_token_accuracy": 0.6832646131515503, + "num_tokens": 6186495.0, + "step": 161 + }, + { + "epoch": 0.020608065131662637, + "grad_norm": 30.5, + "learning_rate": 6.824925816023738e-08, + "loss": 1.224, + "mean_token_accuracy": 0.6827049851417542, + "num_tokens": 6222454.0, + "step": 162 + }, + { + "epoch": 0.02073527541025315, + "grad_norm": 33.0, + "learning_rate": 6.867316659601525e-08, + "loss": 1.277, + "mean_token_accuracy": 0.6792930960655212, + "num_tokens": 6253260.0, + "step": 163 + }, + { + "epoch": 0.02086248568884366, + "grad_norm": 28.0, + "learning_rate": 6.909707503179312e-08, + "loss": 1.1996, + "mean_token_accuracy": 0.6917874813079834, + "num_tokens": 6294773.0, + "step": 164 + }, + { + "epoch": 0.02098969596743417, + "grad_norm": 32.0, + "learning_rate": 6.9520983467571e-08, + "loss": 1.2687, + "mean_token_accuracy": 0.6672902703285217, + "num_tokens": 6335334.0, + "step": 165 + }, + { + "epoch": 0.021116906246024678, + "grad_norm": 28.75, + "learning_rate": 6.994489190334887e-08, + "loss": 1.1809, + "mean_token_accuracy": 0.6912703514099121, + "num_tokens": 6372255.0, + "step": 166 + }, + { + "epoch": 0.02124411652461519, + "grad_norm": 27.75, + "learning_rate": 7.036880033912674e-08, + "loss": 1.1371, + "mean_token_accuracy": 0.7015582323074341, + "num_tokens": 6411971.0, + "step": 167 + }, + { + "epoch": 0.0213713268032057, + "grad_norm": 28.5, + "learning_rate": 7.079270877490461e-08, + "loss": 1.2482, + "mean_token_accuracy": 0.6804158091545105, + "num_tokens": 6450366.0, + "step": 168 + }, + { + "epoch": 0.02149853708179621, + "grad_norm": 29.375, + "learning_rate": 7.121661721068249e-08, + "loss": 1.1994, + "mean_token_accuracy": 0.6911055445671082, + "num_tokens": 6482199.0, + "step": 169 + }, + { + "epoch": 0.021625747360386718, + "grad_norm": 31.125, + "learning_rate": 7.164052564646036e-08, + "loss": 1.2037, + "mean_token_accuracy": 0.688421905040741, + "num_tokens": 6519697.0, + "step": 170 + }, + { + "epoch": 0.02175295763897723, + "grad_norm": 31.375, + "learning_rate": 7.206443408223823e-08, + "loss": 1.2845, + "mean_token_accuracy": 0.6719388961791992, + "num_tokens": 6554687.0, + "step": 171 + }, + { + "epoch": 0.02188016791756774, + "grad_norm": 29.875, + "learning_rate": 7.24883425180161e-08, + "loss": 1.2212, + "mean_token_accuracy": 0.6763405799865723, + "num_tokens": 6593066.0, + "step": 172 + }, + { + "epoch": 0.02200737819615825, + "grad_norm": 27.125, + "learning_rate": 7.291225095379398e-08, + "loss": 1.1207, + "mean_token_accuracy": 0.7024766802787781, + "num_tokens": 6635484.0, + "step": 173 + }, + { + "epoch": 0.02213458847474876, + "grad_norm": 30.75, + "learning_rate": 7.333615938957185e-08, + "loss": 1.2602, + "mean_token_accuracy": 0.6658720970153809, + "num_tokens": 6677090.0, + "step": 174 + }, + { + "epoch": 0.02226179875333927, + "grad_norm": 27.5, + "learning_rate": 7.376006782534971e-08, + "loss": 1.1575, + "mean_token_accuracy": 0.7004204392433167, + "num_tokens": 6710250.0, + "step": 175 + }, + { + "epoch": 0.02238900903192978, + "grad_norm": 28.0, + "learning_rate": 7.418397626112758e-08, + "loss": 1.1644, + "mean_token_accuracy": 0.6936711668968201, + "num_tokens": 6752998.0, + "step": 176 + }, + { + "epoch": 0.02251621931052029, + "grad_norm": 31.75, + "learning_rate": 7.460788469690545e-08, + "loss": 1.2237, + "mean_token_accuracy": 0.6831626892089844, + "num_tokens": 6789568.0, + "step": 177 + }, + { + "epoch": 0.0226434295891108, + "grad_norm": 34.25, + "learning_rate": 7.503179313268333e-08, + "loss": 1.2933, + "mean_token_accuracy": 0.6670907139778137, + "num_tokens": 6822810.0, + "step": 178 + }, + { + "epoch": 0.02277063986770131, + "grad_norm": 28.0, + "learning_rate": 7.54557015684612e-08, + "loss": 1.1669, + "mean_token_accuracy": 0.6948963403701782, + "num_tokens": 6861598.0, + "step": 179 + }, + { + "epoch": 0.02289785014629182, + "grad_norm": 28.5, + "learning_rate": 7.587961000423907e-08, + "loss": 1.1975, + "mean_token_accuracy": 0.6852601170539856, + "num_tokens": 6900220.0, + "step": 180 + }, + { + "epoch": 0.02302506042488233, + "grad_norm": 34.75, + "learning_rate": 7.630351844001694e-08, + "loss": 1.3727, + "mean_token_accuracy": 0.6394256353378296, + "num_tokens": 6946983.0, + "step": 181 + }, + { + "epoch": 0.02315227070347284, + "grad_norm": 29.25, + "learning_rate": 7.672742687579482e-08, + "loss": 1.2345, + "mean_token_accuracy": 0.6800859570503235, + "num_tokens": 6986019.0, + "step": 182 + }, + { + "epoch": 0.02327948098206335, + "grad_norm": 28.125, + "learning_rate": 7.715133531157269e-08, + "loss": 1.1933, + "mean_token_accuracy": 0.6844331622123718, + "num_tokens": 7020722.0, + "step": 183 + }, + { + "epoch": 0.02340669126065386, + "grad_norm": 31.25, + "learning_rate": 7.757524374735056e-08, + "loss": 1.2907, + "mean_token_accuracy": 0.6645493507385254, + "num_tokens": 7062691.0, + "step": 184 + }, + { + "epoch": 0.02353390153924437, + "grad_norm": 30.875, + "learning_rate": 7.799915218312844e-08, + "loss": 1.2474, + "mean_token_accuracy": 0.6772946119308472, + "num_tokens": 7101090.0, + "step": 185 + }, + { + "epoch": 0.02366111181783488, + "grad_norm": 28.125, + "learning_rate": 7.842306061890631e-08, + "loss": 1.1787, + "mean_token_accuracy": 0.6859437227249146, + "num_tokens": 7139891.0, + "step": 186 + }, + { + "epoch": 0.023788322096425393, + "grad_norm": 31.125, + "learning_rate": 7.88469690546842e-08, + "loss": 1.2636, + "mean_token_accuracy": 0.674735963344574, + "num_tokens": 7179501.0, + "step": 187 + }, + { + "epoch": 0.0239155323750159, + "grad_norm": 30.25, + "learning_rate": 7.927087749046207e-08, + "loss": 1.2164, + "mean_token_accuracy": 0.6887587308883667, + "num_tokens": 7213809.0, + "step": 188 + }, + { + "epoch": 0.02404274265360641, + "grad_norm": 31.125, + "learning_rate": 7.969478592623994e-08, + "loss": 1.3211, + "mean_token_accuracy": 0.6563828587532043, + "num_tokens": 7254493.0, + "step": 189 + }, + { + "epoch": 0.024169952932196922, + "grad_norm": 28.125, + "learning_rate": 8.011869436201781e-08, + "loss": 1.1171, + "mean_token_accuracy": 0.7073049545288086, + "num_tokens": 7292643.0, + "step": 190 + }, + { + "epoch": 0.024297163210787433, + "grad_norm": 32.0, + "learning_rate": 8.054260279779568e-08, + "loss": 1.3071, + "mean_token_accuracy": 0.6647696495056152, + "num_tokens": 7329810.0, + "step": 191 + }, + { + "epoch": 0.02442437348937794, + "grad_norm": 30.625, + "learning_rate": 8.096651123357356e-08, + "loss": 1.2891, + "mean_token_accuracy": 0.6617994904518127, + "num_tokens": 7367630.0, + "step": 192 + }, + { + "epoch": 0.02455158376796845, + "grad_norm": 30.75, + "learning_rate": 8.139041966935143e-08, + "loss": 1.2521, + "mean_token_accuracy": 0.6702940464019775, + "num_tokens": 7411580.0, + "step": 193 + }, + { + "epoch": 0.024678794046558962, + "grad_norm": 29.5, + "learning_rate": 8.181432810512929e-08, + "loss": 1.1943, + "mean_token_accuracy": 0.6850152015686035, + "num_tokens": 7451631.0, + "step": 194 + }, + { + "epoch": 0.024806004325149473, + "grad_norm": 28.0, + "learning_rate": 8.223823654090716e-08, + "loss": 1.1702, + "mean_token_accuracy": 0.6924347281455994, + "num_tokens": 7493645.0, + "step": 195 + }, + { + "epoch": 0.024933214603739984, + "grad_norm": 26.75, + "learning_rate": 8.266214497668503e-08, + "loss": 1.126, + "mean_token_accuracy": 0.7067143321037292, + "num_tokens": 7538042.0, + "step": 196 + }, + { + "epoch": 0.02506042488233049, + "grad_norm": 32.5, + "learning_rate": 8.30860534124629e-08, + "loss": 1.2568, + "mean_token_accuracy": 0.6753851175308228, + "num_tokens": 7575374.0, + "step": 197 + }, + { + "epoch": 0.025187635160921002, + "grad_norm": 32.0, + "learning_rate": 8.350996184824078e-08, + "loss": 1.2788, + "mean_token_accuracy": 0.6743011474609375, + "num_tokens": 7606880.0, + "step": 198 + }, + { + "epoch": 0.025314845439511513, + "grad_norm": 28.75, + "learning_rate": 8.393387028401865e-08, + "loss": 1.1847, + "mean_token_accuracy": 0.688566267490387, + "num_tokens": 7644840.0, + "step": 199 + }, + { + "epoch": 0.025442055718102024, + "grad_norm": 27.875, + "learning_rate": 8.435777871979652e-08, + "loss": 1.2216, + "mean_token_accuracy": 0.6794540882110596, + "num_tokens": 7683856.0, + "step": 200 + }, + { + "epoch": 0.02556926599669253, + "grad_norm": 29.625, + "learning_rate": 8.47816871555744e-08, + "loss": 1.2247, + "mean_token_accuracy": 0.6879350543022156, + "num_tokens": 7715306.0, + "step": 201 + }, + { + "epoch": 0.025696476275283042, + "grad_norm": 27.0, + "learning_rate": 8.520559559135227e-08, + "loss": 1.1194, + "mean_token_accuracy": 0.7066353559494019, + "num_tokens": 7752442.0, + "step": 202 + }, + { + "epoch": 0.025823686553873553, + "grad_norm": 28.75, + "learning_rate": 8.562950402713014e-08, + "loss": 1.2301, + "mean_token_accuracy": 0.6744222044944763, + "num_tokens": 7799891.0, + "step": 203 + }, + { + "epoch": 0.025950896832464064, + "grad_norm": 30.375, + "learning_rate": 8.605341246290801e-08, + "loss": 1.2698, + "mean_token_accuracy": 0.6756240129470825, + "num_tokens": 7830696.0, + "step": 204 + }, + { + "epoch": 0.026078107111054575, + "grad_norm": 27.5, + "learning_rate": 8.647732089868589e-08, + "loss": 1.1914, + "mean_token_accuracy": 0.6912791728973389, + "num_tokens": 7870968.0, + "step": 205 + }, + { + "epoch": 0.026205317389645082, + "grad_norm": 29.5, + "learning_rate": 8.690122933446376e-08, + "loss": 1.2096, + "mean_token_accuracy": 0.6814894080162048, + "num_tokens": 7914418.0, + "step": 206 + }, + { + "epoch": 0.026332527668235593, + "grad_norm": 25.0, + "learning_rate": 8.732513777024163e-08, + "loss": 1.1148, + "mean_token_accuracy": 0.7108790874481201, + "num_tokens": 7954544.0, + "step": 207 + }, + { + "epoch": 0.026459737946826104, + "grad_norm": 29.75, + "learning_rate": 8.77490462060195e-08, + "loss": 1.2164, + "mean_token_accuracy": 0.6865906715393066, + "num_tokens": 7997373.0, + "step": 208 + }, + { + "epoch": 0.026586948225416615, + "grad_norm": 29.125, + "learning_rate": 8.817295464179738e-08, + "loss": 1.1796, + "mean_token_accuracy": 0.6868748664855957, + "num_tokens": 8033314.0, + "step": 209 + }, + { + "epoch": 0.026714158504007122, + "grad_norm": 30.75, + "learning_rate": 8.859686307757525e-08, + "loss": 1.2243, + "mean_token_accuracy": 0.6817008852958679, + "num_tokens": 8069362.0, + "step": 210 + }, + { + "epoch": 0.026841368782597633, + "grad_norm": 31.375, + "learning_rate": 8.902077151335312e-08, + "loss": 1.2687, + "mean_token_accuracy": 0.6660625338554382, + "num_tokens": 8109126.0, + "step": 211 + }, + { + "epoch": 0.026968579061188144, + "grad_norm": 33.25, + "learning_rate": 8.944467994913098e-08, + "loss": 1.3695, + "mean_token_accuracy": 0.654360294342041, + "num_tokens": 8143083.0, + "step": 212 + }, + { + "epoch": 0.027095789339778655, + "grad_norm": 26.375, + "learning_rate": 8.986858838490885e-08, + "loss": 1.1028, + "mean_token_accuracy": 0.7059556245803833, + "num_tokens": 8176636.0, + "step": 213 + }, + { + "epoch": 0.027222999618369163, + "grad_norm": 27.75, + "learning_rate": 9.029249682068673e-08, + "loss": 1.1928, + "mean_token_accuracy": 0.6869773268699646, + "num_tokens": 8220262.0, + "step": 214 + }, + { + "epoch": 0.027350209896959674, + "grad_norm": 30.875, + "learning_rate": 9.07164052564646e-08, + "loss": 1.2668, + "mean_token_accuracy": 0.6777706742286682, + "num_tokens": 8256303.0, + "step": 215 + }, + { + "epoch": 0.027477420175550184, + "grad_norm": 25.25, + "learning_rate": 9.114031369224247e-08, + "loss": 1.1413, + "mean_token_accuracy": 0.7055786848068237, + "num_tokens": 8293888.0, + "step": 216 + }, + { + "epoch": 0.027604630454140695, + "grad_norm": 28.875, + "learning_rate": 9.156422212802034e-08, + "loss": 1.1462, + "mean_token_accuracy": 0.6972092390060425, + "num_tokens": 8335368.0, + "step": 217 + }, + { + "epoch": 0.027731840732731206, + "grad_norm": 30.0, + "learning_rate": 9.198813056379822e-08, + "loss": 1.2258, + "mean_token_accuracy": 0.687328577041626, + "num_tokens": 8370338.0, + "step": 218 + }, + { + "epoch": 0.027859051011321714, + "grad_norm": 28.875, + "learning_rate": 9.241203899957609e-08, + "loss": 1.1756, + "mean_token_accuracy": 0.6946353912353516, + "num_tokens": 8409904.0, + "step": 219 + }, + { + "epoch": 0.027986261289912225, + "grad_norm": 34.25, + "learning_rate": 9.283594743535396e-08, + "loss": 1.4228, + "mean_token_accuracy": 0.640252947807312, + "num_tokens": 8446579.0, + "step": 220 + }, + { + "epoch": 0.028113471568502735, + "grad_norm": 27.25, + "learning_rate": 9.325985587113183e-08, + "loss": 1.1488, + "mean_token_accuracy": 0.7039964199066162, + "num_tokens": 8483663.0, + "step": 221 + }, + { + "epoch": 0.028240681847093246, + "grad_norm": 30.125, + "learning_rate": 9.368376430690971e-08, + "loss": 1.2246, + "mean_token_accuracy": 0.6873204708099365, + "num_tokens": 8516346.0, + "step": 222 + }, + { + "epoch": 0.028367892125683754, + "grad_norm": 29.0, + "learning_rate": 9.410767274268758e-08, + "loss": 1.1844, + "mean_token_accuracy": 0.6881027221679688, + "num_tokens": 8556914.0, + "step": 223 + }, + { + "epoch": 0.028495102404274265, + "grad_norm": 27.625, + "learning_rate": 9.453158117846545e-08, + "loss": 1.1916, + "mean_token_accuracy": 0.6917071342468262, + "num_tokens": 8596548.0, + "step": 224 + }, + { + "epoch": 0.028622312682864776, + "grad_norm": 29.875, + "learning_rate": 9.495548961424333e-08, + "loss": 1.2484, + "mean_token_accuracy": 0.6810131669044495, + "num_tokens": 8634366.0, + "step": 225 + }, + { + "epoch": 0.028749522961455286, + "grad_norm": 29.0, + "learning_rate": 9.53793980500212e-08, + "loss": 1.1745, + "mean_token_accuracy": 0.6973158121109009, + "num_tokens": 8666480.0, + "step": 226 + }, + { + "epoch": 0.028876733240045797, + "grad_norm": 33.75, + "learning_rate": 9.580330648579907e-08, + "loss": 1.3912, + "mean_token_accuracy": 0.6479523181915283, + "num_tokens": 8705880.0, + "step": 227 + }, + { + "epoch": 0.029003943518636305, + "grad_norm": 30.75, + "learning_rate": 9.622721492157694e-08, + "loss": 1.1917, + "mean_token_accuracy": 0.6894640922546387, + "num_tokens": 8743772.0, + "step": 228 + }, + { + "epoch": 0.029131153797226816, + "grad_norm": 31.375, + "learning_rate": 9.665112335735482e-08, + "loss": 1.2224, + "mean_token_accuracy": 0.6783393621444702, + "num_tokens": 8775919.0, + "step": 229 + }, + { + "epoch": 0.029258364075817327, + "grad_norm": 26.5, + "learning_rate": 9.707503179313267e-08, + "loss": 1.1201, + "mean_token_accuracy": 0.7099014520645142, + "num_tokens": 8809336.0, + "step": 230 + }, + { + "epoch": 0.029385574354407838, + "grad_norm": 26.5, + "learning_rate": 9.749894022891055e-08, + "loss": 1.188, + "mean_token_accuracy": 0.686130702495575, + "num_tokens": 8845952.0, + "step": 231 + }, + { + "epoch": 0.029512784632998345, + "grad_norm": 28.875, + "learning_rate": 9.792284866468842e-08, + "loss": 1.2091, + "mean_token_accuracy": 0.6749844551086426, + "num_tokens": 8889801.0, + "step": 232 + }, + { + "epoch": 0.029639994911588856, + "grad_norm": 29.5, + "learning_rate": 9.834675710046629e-08, + "loss": 1.195, + "mean_token_accuracy": 0.6866620182991028, + "num_tokens": 8925429.0, + "step": 233 + }, + { + "epoch": 0.029767205190179367, + "grad_norm": 30.875, + "learning_rate": 9.877066553624416e-08, + "loss": 1.2247, + "mean_token_accuracy": 0.676378607749939, + "num_tokens": 8963360.0, + "step": 234 + }, + { + "epoch": 0.029894415468769878, + "grad_norm": 28.125, + "learning_rate": 9.919457397202204e-08, + "loss": 1.2329, + "mean_token_accuracy": 0.6798755526542664, + "num_tokens": 8998314.0, + "step": 235 + }, + { + "epoch": 0.030021625747360385, + "grad_norm": 32.5, + "learning_rate": 9.961848240779991e-08, + "loss": 1.3699, + "mean_token_accuracy": 0.6511887907981873, + "num_tokens": 9034804.0, + "step": 236 + }, + { + "epoch": 0.030148836025950896, + "grad_norm": 29.125, + "learning_rate": 1.0004239084357778e-07, + "loss": 1.2401, + "mean_token_accuracy": 0.681018590927124, + "num_tokens": 9070840.0, + "step": 237 + }, + { + "epoch": 0.030276046304541407, + "grad_norm": 28.125, + "learning_rate": 1.0046629927935566e-07, + "loss": 1.2139, + "mean_token_accuracy": 0.6908760666847229, + "num_tokens": 9109370.0, + "step": 238 + }, + { + "epoch": 0.030403256583131918, + "grad_norm": 23.375, + "learning_rate": 1.0089020771513353e-07, + "loss": 1.0798, + "mean_token_accuracy": 0.7163187265396118, + "num_tokens": 9148191.0, + "step": 239 + }, + { + "epoch": 0.03053046686172243, + "grad_norm": 29.125, + "learning_rate": 1.013141161509114e-07, + "loss": 1.2188, + "mean_token_accuracy": 0.6909701824188232, + "num_tokens": 9182923.0, + "step": 240 + }, + { + "epoch": 0.030657677140312936, + "grad_norm": 29.375, + "learning_rate": 1.0173802458668927e-07, + "loss": 1.2242, + "mean_token_accuracy": 0.6773542165756226, + "num_tokens": 9215882.0, + "step": 241 + }, + { + "epoch": 0.030784887418903447, + "grad_norm": 31.25, + "learning_rate": 1.0216193302246715e-07, + "loss": 1.2795, + "mean_token_accuracy": 0.660506010055542, + "num_tokens": 9249718.0, + "step": 242 + }, + { + "epoch": 0.030912097697493958, + "grad_norm": 27.25, + "learning_rate": 1.0258584145824502e-07, + "loss": 1.1346, + "mean_token_accuracy": 0.7045959830284119, + "num_tokens": 9291146.0, + "step": 243 + }, + { + "epoch": 0.03103930797608447, + "grad_norm": 26.75, + "learning_rate": 1.0300974989402289e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6962883472442627, + "num_tokens": 9326403.0, + "step": 244 + }, + { + "epoch": 0.031166518254674976, + "grad_norm": 29.625, + "learning_rate": 1.0343365832980076e-07, + "loss": 1.221, + "mean_token_accuracy": 0.6781588792800903, + "num_tokens": 9368491.0, + "step": 245 + }, + { + "epoch": 0.03129372853326549, + "grad_norm": 28.0, + "learning_rate": 1.0385756676557864e-07, + "loss": 1.2087, + "mean_token_accuracy": 0.6863057017326355, + "num_tokens": 9409041.0, + "step": 246 + }, + { + "epoch": 0.031420938811855995, + "grad_norm": 27.125, + "learning_rate": 1.0428147520135651e-07, + "loss": 1.1527, + "mean_token_accuracy": 0.6984075307846069, + "num_tokens": 9448591.0, + "step": 247 + }, + { + "epoch": 0.03154814909044651, + "grad_norm": 32.0, + "learning_rate": 1.0470538363713437e-07, + "loss": 1.2464, + "mean_token_accuracy": 0.6663109064102173, + "num_tokens": 9486604.0, + "step": 248 + }, + { + "epoch": 0.031675359369037016, + "grad_norm": 29.125, + "learning_rate": 1.0512929207291224e-07, + "loss": 1.231, + "mean_token_accuracy": 0.6716802716255188, + "num_tokens": 9525285.0, + "step": 249 + }, + { + "epoch": 0.03180256964762753, + "grad_norm": 29.125, + "learning_rate": 1.0555320050869011e-07, + "loss": 1.2055, + "mean_token_accuracy": 0.6861064434051514, + "num_tokens": 9557898.0, + "step": 250 + }, + { + "epoch": 0.03192977992621804, + "grad_norm": 34.75, + "learning_rate": 1.0597710894446799e-07, + "loss": 1.364, + "mean_token_accuracy": 0.6477567553520203, + "num_tokens": 9590438.0, + "step": 251 + }, + { + "epoch": 0.032056990204808546, + "grad_norm": 26.875, + "learning_rate": 1.0640101738024586e-07, + "loss": 1.1325, + "mean_token_accuracy": 0.7028815746307373, + "num_tokens": 9631402.0, + "step": 252 + }, + { + "epoch": 0.03218420048339906, + "grad_norm": 28.25, + "learning_rate": 1.0682492581602373e-07, + "loss": 1.1617, + "mean_token_accuracy": 0.697769284248352, + "num_tokens": 9663702.0, + "step": 253 + }, + { + "epoch": 0.03231141076198957, + "grad_norm": 28.125, + "learning_rate": 1.072488342518016e-07, + "loss": 1.1401, + "mean_token_accuracy": 0.70120769739151, + "num_tokens": 9698969.0, + "step": 254 + }, + { + "epoch": 0.03243862104058008, + "grad_norm": 28.25, + "learning_rate": 1.0767274268757948e-07, + "loss": 1.2033, + "mean_token_accuracy": 0.6868095397949219, + "num_tokens": 9738075.0, + "step": 255 + }, + { + "epoch": 0.03256583131917059, + "grad_norm": 28.5, + "learning_rate": 1.0809665112335735e-07, + "loss": 1.1947, + "mean_token_accuracy": 0.686470091342926, + "num_tokens": 9782675.0, + "step": 256 + }, + { + "epoch": 0.0326930415977611, + "grad_norm": 27.375, + "learning_rate": 1.0852055955913522e-07, + "loss": 1.0722, + "mean_token_accuracy": 0.7204323410987854, + "num_tokens": 9820585.0, + "step": 257 + }, + { + "epoch": 0.03282025187635161, + "grad_norm": 30.125, + "learning_rate": 1.089444679949131e-07, + "loss": 1.2478, + "mean_token_accuracy": 0.6746362447738647, + "num_tokens": 9860693.0, + "step": 258 + }, + { + "epoch": 0.03294746215494212, + "grad_norm": 26.375, + "learning_rate": 1.0936837643069097e-07, + "loss": 1.1418, + "mean_token_accuracy": 0.6977858543395996, + "num_tokens": 9902629.0, + "step": 259 + }, + { + "epoch": 0.03307467243353263, + "grad_norm": 26.375, + "learning_rate": 1.0979228486646884e-07, + "loss": 1.1214, + "mean_token_accuracy": 0.7053110599517822, + "num_tokens": 9937304.0, + "step": 260 + }, + { + "epoch": 0.03320188271212314, + "grad_norm": 32.0, + "learning_rate": 1.1021619330224671e-07, + "loss": 1.3182, + "mean_token_accuracy": 0.6525719165802002, + "num_tokens": 9976911.0, + "step": 261 + }, + { + "epoch": 0.03332909299071365, + "grad_norm": 29.625, + "learning_rate": 1.1064010173802458e-07, + "loss": 1.2315, + "mean_token_accuracy": 0.6783162355422974, + "num_tokens": 10015740.0, + "step": 262 + }, + { + "epoch": 0.03345630326930416, + "grad_norm": 26.875, + "learning_rate": 1.1106401017380246e-07, + "loss": 1.1307, + "mean_token_accuracy": 0.702600359916687, + "num_tokens": 10053054.0, + "step": 263 + }, + { + "epoch": 0.03358351354789467, + "grad_norm": 31.625, + "learning_rate": 1.1148791860958033e-07, + "loss": 1.3369, + "mean_token_accuracy": 0.6486976146697998, + "num_tokens": 10082088.0, + "step": 264 + }, + { + "epoch": 0.03371072382648518, + "grad_norm": 29.75, + "learning_rate": 1.119118270453582e-07, + "loss": 1.2619, + "mean_token_accuracy": 0.6745377779006958, + "num_tokens": 10121122.0, + "step": 265 + }, + { + "epoch": 0.03383793410507569, + "grad_norm": 28.75, + "learning_rate": 1.1233573548113607e-07, + "loss": 1.2343, + "mean_token_accuracy": 0.6760827898979187, + "num_tokens": 10161017.0, + "step": 266 + }, + { + "epoch": 0.0339651443836662, + "grad_norm": 27.25, + "learning_rate": 1.1275964391691393e-07, + "loss": 1.144, + "mean_token_accuracy": 0.6955097317695618, + "num_tokens": 10204913.0, + "step": 267 + }, + { + "epoch": 0.03409235466225671, + "grad_norm": 28.0, + "learning_rate": 1.131835523526918e-07, + "loss": 1.2359, + "mean_token_accuracy": 0.6765199899673462, + "num_tokens": 10245154.0, + "step": 268 + }, + { + "epoch": 0.03421956494084722, + "grad_norm": 30.875, + "learning_rate": 1.1360746078846968e-07, + "loss": 1.2633, + "mean_token_accuracy": 0.671457052230835, + "num_tokens": 10283006.0, + "step": 269 + }, + { + "epoch": 0.03434677521943773, + "grad_norm": 30.25, + "learning_rate": 1.1403136922424755e-07, + "loss": 1.2352, + "mean_token_accuracy": 0.6770668029785156, + "num_tokens": 10316689.0, + "step": 270 + }, + { + "epoch": 0.03447398549802824, + "grad_norm": 29.375, + "learning_rate": 1.1445527766002542e-07, + "loss": 1.1878, + "mean_token_accuracy": 0.682701587677002, + "num_tokens": 10358145.0, + "step": 271 + }, + { + "epoch": 0.03460119577661875, + "grad_norm": 33.5, + "learning_rate": 1.148791860958033e-07, + "loss": 1.3679, + "mean_token_accuracy": 0.646016001701355, + "num_tokens": 10393424.0, + "step": 272 + }, + { + "epoch": 0.034728406055209264, + "grad_norm": 29.0, + "learning_rate": 1.1530309453158117e-07, + "loss": 1.2172, + "mean_token_accuracy": 0.6823354959487915, + "num_tokens": 10433560.0, + "step": 273 + }, + { + "epoch": 0.03485561633379977, + "grad_norm": 28.875, + "learning_rate": 1.1572700296735904e-07, + "loss": 1.2271, + "mean_token_accuracy": 0.6872689723968506, + "num_tokens": 10472024.0, + "step": 274 + }, + { + "epoch": 0.03498282661239028, + "grad_norm": 31.875, + "learning_rate": 1.1615091140313691e-07, + "loss": 1.2999, + "mean_token_accuracy": 0.6656768321990967, + "num_tokens": 10510874.0, + "step": 275 + }, + { + "epoch": 0.03511003689098079, + "grad_norm": 26.5, + "learning_rate": 1.1657481983891479e-07, + "loss": 1.1628, + "mean_token_accuracy": 0.6936239004135132, + "num_tokens": 10552411.0, + "step": 276 + }, + { + "epoch": 0.0352372471695713, + "grad_norm": 31.75, + "learning_rate": 1.1699872827469266e-07, + "loss": 1.217, + "mean_token_accuracy": 0.6831883192062378, + "num_tokens": 10591537.0, + "step": 277 + }, + { + "epoch": 0.03536445744816181, + "grad_norm": 31.25, + "learning_rate": 1.1742263671047053e-07, + "loss": 1.2257, + "mean_token_accuracy": 0.67851722240448, + "num_tokens": 10632228.0, + "step": 278 + }, + { + "epoch": 0.03549166772675232, + "grad_norm": 30.0, + "learning_rate": 1.178465451462484e-07, + "loss": 1.286, + "mean_token_accuracy": 0.66211998462677, + "num_tokens": 10672829.0, + "step": 279 + }, + { + "epoch": 0.03561887800534283, + "grad_norm": 31.625, + "learning_rate": 1.1827045358202628e-07, + "loss": 1.3053, + "mean_token_accuracy": 0.6703018546104431, + "num_tokens": 10706146.0, + "step": 280 + }, + { + "epoch": 0.035746088283933344, + "grad_norm": 24.875, + "learning_rate": 1.1869436201780415e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.711277425289154, + "num_tokens": 10744112.0, + "step": 281 + }, + { + "epoch": 0.03587329856252385, + "grad_norm": 27.5, + "learning_rate": 1.1911827045358202e-07, + "loss": 1.179, + "mean_token_accuracy": 0.6999143958091736, + "num_tokens": 10783726.0, + "step": 282 + }, + { + "epoch": 0.03600050884111436, + "grad_norm": 30.125, + "learning_rate": 1.195421788893599e-07, + "loss": 1.2798, + "mean_token_accuracy": 0.671887993812561, + "num_tokens": 10824476.0, + "step": 283 + }, + { + "epoch": 0.036127719119704874, + "grad_norm": 29.375, + "learning_rate": 1.1996608732513778e-07, + "loss": 1.2379, + "mean_token_accuracy": 0.6812333464622498, + "num_tokens": 10859997.0, + "step": 284 + }, + { + "epoch": 0.03625492939829538, + "grad_norm": 28.375, + "learning_rate": 1.2038999576091563e-07, + "loss": 1.1824, + "mean_token_accuracy": 0.6883386373519897, + "num_tokens": 10898800.0, + "step": 285 + }, + { + "epoch": 0.036382139676885895, + "grad_norm": 26.25, + "learning_rate": 1.208139041966935e-07, + "loss": 1.1355, + "mean_token_accuracy": 0.7122577428817749, + "num_tokens": 10937462.0, + "step": 286 + }, + { + "epoch": 0.0365093499554764, + "grad_norm": 30.625, + "learning_rate": 1.2123781263247137e-07, + "loss": 1.258, + "mean_token_accuracy": 0.6742825508117676, + "num_tokens": 10970739.0, + "step": 287 + }, + { + "epoch": 0.03663656023406691, + "grad_norm": 33.75, + "learning_rate": 1.2166172106824924e-07, + "loss": 1.3702, + "mean_token_accuracy": 0.6611979603767395, + "num_tokens": 11003000.0, + "step": 288 + }, + { + "epoch": 0.036763770512657425, + "grad_norm": 28.75, + "learning_rate": 1.2208562950402712e-07, + "loss": 1.1838, + "mean_token_accuracy": 0.6896117925643921, + "num_tokens": 11039665.0, + "step": 289 + }, + { + "epoch": 0.03689098079124793, + "grad_norm": 29.0, + "learning_rate": 1.22509537939805e-07, + "loss": 1.1969, + "mean_token_accuracy": 0.6826200485229492, + "num_tokens": 11078368.0, + "step": 290 + }, + { + "epoch": 0.03701819106983844, + "grad_norm": 28.875, + "learning_rate": 1.2293344637558286e-07, + "loss": 1.2048, + "mean_token_accuracy": 0.6786288022994995, + "num_tokens": 11122654.0, + "step": 291 + }, + { + "epoch": 0.037145401348428954, + "grad_norm": 27.0, + "learning_rate": 1.2335735481136073e-07, + "loss": 1.1605, + "mean_token_accuracy": 0.6917229890823364, + "num_tokens": 11162622.0, + "step": 292 + }, + { + "epoch": 0.03727261162701946, + "grad_norm": 28.0, + "learning_rate": 1.237812632471386e-07, + "loss": 1.1892, + "mean_token_accuracy": 0.6897135972976685, + "num_tokens": 11202564.0, + "step": 293 + }, + { + "epoch": 0.037399821905609976, + "grad_norm": 30.0, + "learning_rate": 1.2420517168291648e-07, + "loss": 1.2242, + "mean_token_accuracy": 0.6702020764350891, + "num_tokens": 11243422.0, + "step": 294 + }, + { + "epoch": 0.03752703218420048, + "grad_norm": 28.375, + "learning_rate": 1.2462908011869435e-07, + "loss": 1.1814, + "mean_token_accuracy": 0.6826740503311157, + "num_tokens": 11280867.0, + "step": 295 + }, + { + "epoch": 0.03765424246279099, + "grad_norm": 27.875, + "learning_rate": 1.2505298855447223e-07, + "loss": 1.116, + "mean_token_accuracy": 0.7053584456443787, + "num_tokens": 11318454.0, + "step": 296 + }, + { + "epoch": 0.037781452741381505, + "grad_norm": 33.0, + "learning_rate": 1.254768969902501e-07, + "loss": 1.3692, + "mean_token_accuracy": 0.6524268984794617, + "num_tokens": 11354721.0, + "step": 297 + }, + { + "epoch": 0.03790866301997201, + "grad_norm": 32.0, + "learning_rate": 1.2590080542602797e-07, + "loss": 1.3132, + "mean_token_accuracy": 0.6651740074157715, + "num_tokens": 11385938.0, + "step": 298 + }, + { + "epoch": 0.03803587329856253, + "grad_norm": 29.375, + "learning_rate": 1.2632471386180584e-07, + "loss": 1.2455, + "mean_token_accuracy": 0.6720113754272461, + "num_tokens": 11424911.0, + "step": 299 + }, + { + "epoch": 0.038163083577153034, + "grad_norm": 27.75, + "learning_rate": 1.2674862229758372e-07, + "loss": 1.1835, + "mean_token_accuracy": 0.6949384212493896, + "num_tokens": 11460580.0, + "step": 300 + }, + { + "epoch": 0.03829029385574354, + "grad_norm": 27.875, + "learning_rate": 1.271725307333616e-07, + "loss": 1.1606, + "mean_token_accuracy": 0.6969467997550964, + "num_tokens": 11501031.0, + "step": 301 + }, + { + "epoch": 0.038417504134334056, + "grad_norm": 29.5, + "learning_rate": 1.2759643916913946e-07, + "loss": 1.1897, + "mean_token_accuracy": 0.6907870769500732, + "num_tokens": 11544053.0, + "step": 302 + }, + { + "epoch": 0.03854471441292456, + "grad_norm": 31.0, + "learning_rate": 1.2802034760491733e-07, + "loss": 1.2533, + "mean_token_accuracy": 0.6725675463676453, + "num_tokens": 11576944.0, + "step": 303 + }, + { + "epoch": 0.03867192469151508, + "grad_norm": 30.5, + "learning_rate": 1.284442560406952e-07, + "loss": 1.2838, + "mean_token_accuracy": 0.6582578420639038, + "num_tokens": 11619735.0, + "step": 304 + }, + { + "epoch": 0.038799134970105585, + "grad_norm": 33.25, + "learning_rate": 1.2886816447647308e-07, + "loss": 1.3754, + "mean_token_accuracy": 0.6436046361923218, + "num_tokens": 11655025.0, + "step": 305 + }, + { + "epoch": 0.03892634524869609, + "grad_norm": 27.25, + "learning_rate": 1.2929207291225095e-07, + "loss": 1.1751, + "mean_token_accuracy": 0.690939724445343, + "num_tokens": 11691970.0, + "step": 306 + }, + { + "epoch": 0.03905355552728661, + "grad_norm": 27.5, + "learning_rate": 1.2971598134802882e-07, + "loss": 1.2075, + "mean_token_accuracy": 0.6849638223648071, + "num_tokens": 11729561.0, + "step": 307 + }, + { + "epoch": 0.039180765805877114, + "grad_norm": 30.375, + "learning_rate": 1.301398897838067e-07, + "loss": 1.2772, + "mean_token_accuracy": 0.665428638458252, + "num_tokens": 11765707.0, + "step": 308 + }, + { + "epoch": 0.03930797608446762, + "grad_norm": 29.5, + "learning_rate": 1.3056379821958457e-07, + "loss": 1.2489, + "mean_token_accuracy": 0.6808481216430664, + "num_tokens": 11806223.0, + "step": 309 + }, + { + "epoch": 0.039435186363058136, + "grad_norm": 28.875, + "learning_rate": 1.3098770665536244e-07, + "loss": 1.1932, + "mean_token_accuracy": 0.6920714378356934, + "num_tokens": 11845477.0, + "step": 310 + }, + { + "epoch": 0.039562396641648644, + "grad_norm": 28.375, + "learning_rate": 1.3141161509114031e-07, + "loss": 1.2009, + "mean_token_accuracy": 0.6876450777053833, + "num_tokens": 11877973.0, + "step": 311 + }, + { + "epoch": 0.03968960692023916, + "grad_norm": 29.125, + "learning_rate": 1.3183552352691819e-07, + "loss": 1.2203, + "mean_token_accuracy": 0.6828383207321167, + "num_tokens": 11912093.0, + "step": 312 + }, + { + "epoch": 0.039816817198829665, + "grad_norm": 28.875, + "learning_rate": 1.3225943196269603e-07, + "loss": 1.2207, + "mean_token_accuracy": 0.6856380701065063, + "num_tokens": 11952541.0, + "step": 313 + }, + { + "epoch": 0.03994402747742017, + "grad_norm": 28.375, + "learning_rate": 1.3268334039847393e-07, + "loss": 1.1839, + "mean_token_accuracy": 0.688917338848114, + "num_tokens": 11990414.0, + "step": 314 + }, + { + "epoch": 0.04007123775601069, + "grad_norm": 34.5, + "learning_rate": 1.3310724883425178e-07, + "loss": 1.3697, + "mean_token_accuracy": 0.6420711278915405, + "num_tokens": 12025925.0, + "step": 315 + }, + { + "epoch": 0.040198448034601195, + "grad_norm": 29.875, + "learning_rate": 1.3353115727002968e-07, + "loss": 1.2431, + "mean_token_accuracy": 0.6704326272010803, + "num_tokens": 12068879.0, + "step": 316 + }, + { + "epoch": 0.04032565831319171, + "grad_norm": 27.875, + "learning_rate": 1.3395506570580752e-07, + "loss": 1.1758, + "mean_token_accuracy": 0.691962480545044, + "num_tokens": 12111931.0, + "step": 317 + }, + { + "epoch": 0.040452868591782216, + "grad_norm": 33.75, + "learning_rate": 1.3437897414158542e-07, + "loss": 1.3903, + "mean_token_accuracy": 0.6466199159622192, + "num_tokens": 12140431.0, + "step": 318 + }, + { + "epoch": 0.040580078870372724, + "grad_norm": 26.875, + "learning_rate": 1.3480288257736327e-07, + "loss": 1.1816, + "mean_token_accuracy": 0.6924995183944702, + "num_tokens": 12173932.0, + "step": 319 + }, + { + "epoch": 0.04070728914896324, + "grad_norm": 28.125, + "learning_rate": 1.3522679101314117e-07, + "loss": 1.1543, + "mean_token_accuracy": 0.6902577877044678, + "num_tokens": 12210040.0, + "step": 320 + }, + { + "epoch": 0.040834499427553746, + "grad_norm": 28.125, + "learning_rate": 1.35650699448919e-07, + "loss": 1.1939, + "mean_token_accuracy": 0.6882185935974121, + "num_tokens": 12248556.0, + "step": 321 + }, + { + "epoch": 0.04096170970614425, + "grad_norm": 27.875, + "learning_rate": 1.360746078846969e-07, + "loss": 1.1781, + "mean_token_accuracy": 0.6896816492080688, + "num_tokens": 12285664.0, + "step": 322 + }, + { + "epoch": 0.04108891998473477, + "grad_norm": 33.25, + "learning_rate": 1.3649851632047476e-07, + "loss": 1.2679, + "mean_token_accuracy": 0.6667803525924683, + "num_tokens": 12326964.0, + "step": 323 + }, + { + "epoch": 0.041216130263325275, + "grad_norm": 24.375, + "learning_rate": 1.3692242475625266e-07, + "loss": 1.0969, + "mean_token_accuracy": 0.7146943211555481, + "num_tokens": 12366541.0, + "step": 324 + }, + { + "epoch": 0.04134334054191579, + "grad_norm": 28.875, + "learning_rate": 1.373463331920305e-07, + "loss": 1.2102, + "mean_token_accuracy": 0.6909917593002319, + "num_tokens": 12405664.0, + "step": 325 + }, + { + "epoch": 0.0414705508205063, + "grad_norm": 30.75, + "learning_rate": 1.377702416278084e-07, + "loss": 1.3083, + "mean_token_accuracy": 0.6588596105575562, + "num_tokens": 12445039.0, + "step": 326 + }, + { + "epoch": 0.041597761099096804, + "grad_norm": 26.5, + "learning_rate": 1.3819415006358625e-07, + "loss": 1.1013, + "mean_token_accuracy": 0.7075461149215698, + "num_tokens": 12485481.0, + "step": 327 + }, + { + "epoch": 0.04172497137768732, + "grad_norm": 26.0, + "learning_rate": 1.3861805849936415e-07, + "loss": 1.1519, + "mean_token_accuracy": 0.7005997896194458, + "num_tokens": 12530272.0, + "step": 328 + }, + { + "epoch": 0.041852181656277826, + "grad_norm": 31.375, + "learning_rate": 1.39041966935142e-07, + "loss": 1.2782, + "mean_token_accuracy": 0.6637688875198364, + "num_tokens": 12565854.0, + "step": 329 + }, + { + "epoch": 0.04197939193486834, + "grad_norm": 29.0, + "learning_rate": 1.394658753709199e-07, + "loss": 1.196, + "mean_token_accuracy": 0.6846940517425537, + "num_tokens": 12607313.0, + "step": 330 + }, + { + "epoch": 0.04210660221345885, + "grad_norm": 30.125, + "learning_rate": 1.3988978380669774e-07, + "loss": 1.2814, + "mean_token_accuracy": 0.6674593687057495, + "num_tokens": 12642057.0, + "step": 331 + }, + { + "epoch": 0.042233812492049355, + "grad_norm": 29.5, + "learning_rate": 1.403136922424756e-07, + "loss": 1.2364, + "mean_token_accuracy": 0.6720514297485352, + "num_tokens": 12684118.0, + "step": 332 + }, + { + "epoch": 0.04236102277063987, + "grad_norm": 27.875, + "learning_rate": 1.4073760067825348e-07, + "loss": 1.1925, + "mean_token_accuracy": 0.681996762752533, + "num_tokens": 12721945.0, + "step": 333 + }, + { + "epoch": 0.04248823304923038, + "grad_norm": 27.5, + "learning_rate": 1.4116150911403136e-07, + "loss": 1.1905, + "mean_token_accuracy": 0.6823940277099609, + "num_tokens": 12764175.0, + "step": 334 + }, + { + "epoch": 0.04261544332782089, + "grad_norm": 26.75, + "learning_rate": 1.4158541754980923e-07, + "loss": 1.1481, + "mean_token_accuracy": 0.6939631700515747, + "num_tokens": 12801600.0, + "step": 335 + }, + { + "epoch": 0.0427426536064114, + "grad_norm": 29.25, + "learning_rate": 1.420093259855871e-07, + "loss": 1.1957, + "mean_token_accuracy": 0.6946434378623962, + "num_tokens": 12843584.0, + "step": 336 + }, + { + "epoch": 0.042869863885001906, + "grad_norm": 31.0, + "learning_rate": 1.4243323442136497e-07, + "loss": 1.2532, + "mean_token_accuracy": 0.6739703416824341, + "num_tokens": 12878105.0, + "step": 337 + }, + { + "epoch": 0.04299707416359242, + "grad_norm": 31.125, + "learning_rate": 1.4285714285714285e-07, + "loss": 1.2632, + "mean_token_accuracy": 0.6734256744384766, + "num_tokens": 12911712.0, + "step": 338 + }, + { + "epoch": 0.04312428444218293, + "grad_norm": 27.75, + "learning_rate": 1.4328105129292072e-07, + "loss": 1.1845, + "mean_token_accuracy": 0.689749538898468, + "num_tokens": 12951303.0, + "step": 339 + }, + { + "epoch": 0.043251494720773435, + "grad_norm": 28.75, + "learning_rate": 1.437049597286986e-07, + "loss": 1.1959, + "mean_token_accuracy": 0.6866641044616699, + "num_tokens": 12990544.0, + "step": 340 + }, + { + "epoch": 0.04337870499936395, + "grad_norm": 31.0, + "learning_rate": 1.4412886816447646e-07, + "loss": 1.2992, + "mean_token_accuracy": 0.6639156341552734, + "num_tokens": 13023766.0, + "step": 341 + }, + { + "epoch": 0.04350591527795446, + "grad_norm": 26.0, + "learning_rate": 1.4455277660025434e-07, + "loss": 1.1677, + "mean_token_accuracy": 0.7017995715141296, + "num_tokens": 13056337.0, + "step": 342 + }, + { + "epoch": 0.04363312555654497, + "grad_norm": 29.0, + "learning_rate": 1.449766850360322e-07, + "loss": 1.2358, + "mean_token_accuracy": 0.6761553883552551, + "num_tokens": 13097345.0, + "step": 343 + }, + { + "epoch": 0.04376033583513548, + "grad_norm": 30.5, + "learning_rate": 1.4540059347181008e-07, + "loss": 1.2722, + "mean_token_accuracy": 0.6717910170555115, + "num_tokens": 13127332.0, + "step": 344 + }, + { + "epoch": 0.043887546113725986, + "grad_norm": 30.0, + "learning_rate": 1.4582450190758795e-07, + "loss": 1.251, + "mean_token_accuracy": 0.671883225440979, + "num_tokens": 13170760.0, + "step": 345 + }, + { + "epoch": 0.0440147563923165, + "grad_norm": 28.0, + "learning_rate": 1.4624841034336583e-07, + "loss": 1.2318, + "mean_token_accuracy": 0.6800907850265503, + "num_tokens": 13207061.0, + "step": 346 + }, + { + "epoch": 0.04414196667090701, + "grad_norm": 26.375, + "learning_rate": 1.466723187791437e-07, + "loss": 1.1171, + "mean_token_accuracy": 0.7040960192680359, + "num_tokens": 13247903.0, + "step": 347 + }, + { + "epoch": 0.04426917694949752, + "grad_norm": 26.125, + "learning_rate": 1.4709622721492157e-07, + "loss": 1.1428, + "mean_token_accuracy": 0.6959642171859741, + "num_tokens": 13287193.0, + "step": 348 + }, + { + "epoch": 0.04439638722808803, + "grad_norm": 29.5, + "learning_rate": 1.4752013565069942e-07, + "loss": 1.2048, + "mean_token_accuracy": 0.688086211681366, + "num_tokens": 13323112.0, + "step": 349 + }, + { + "epoch": 0.04452359750667854, + "grad_norm": 27.125, + "learning_rate": 1.4794404408647732e-07, + "loss": 1.163, + "mean_token_accuracy": 0.6981509923934937, + "num_tokens": 13359653.0, + "step": 350 + }, + { + "epoch": 0.04465080778526905, + "grad_norm": 26.375, + "learning_rate": 1.4836795252225516e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6987067461013794, + "num_tokens": 13400571.0, + "step": 351 + }, + { + "epoch": 0.04477801806385956, + "grad_norm": 26.75, + "learning_rate": 1.4879186095803306e-07, + "loss": 1.0972, + "mean_token_accuracy": 0.7021986246109009, + "num_tokens": 13437656.0, + "step": 352 + }, + { + "epoch": 0.04490522834245007, + "grad_norm": 27.375, + "learning_rate": 1.492157693938109e-07, + "loss": 1.1987, + "mean_token_accuracy": 0.6842859983444214, + "num_tokens": 13474688.0, + "step": 353 + }, + { + "epoch": 0.04503243862104058, + "grad_norm": 24.625, + "learning_rate": 1.496396778295888e-07, + "loss": 1.0511, + "mean_token_accuracy": 0.7176826000213623, + "num_tokens": 13518256.0, + "step": 354 + }, + { + "epoch": 0.04515964889963109, + "grad_norm": 28.875, + "learning_rate": 1.5006358626536665e-07, + "loss": 1.2358, + "mean_token_accuracy": 0.6812875270843506, + "num_tokens": 13554315.0, + "step": 355 + }, + { + "epoch": 0.0452868591782216, + "grad_norm": 30.5, + "learning_rate": 1.5048749470114455e-07, + "loss": 1.2726, + "mean_token_accuracy": 0.6701937317848206, + "num_tokens": 13594388.0, + "step": 356 + }, + { + "epoch": 0.04541406945681211, + "grad_norm": 26.25, + "learning_rate": 1.509114031369224e-07, + "loss": 1.1518, + "mean_token_accuracy": 0.6961113214492798, + "num_tokens": 13633704.0, + "step": 357 + }, + { + "epoch": 0.04554127973540262, + "grad_norm": 29.5, + "learning_rate": 1.513353115727003e-07, + "loss": 1.244, + "mean_token_accuracy": 0.6674226522445679, + "num_tokens": 13671183.0, + "step": 358 + }, + { + "epoch": 0.04566849001399313, + "grad_norm": 28.375, + "learning_rate": 1.5175922000847814e-07, + "loss": 1.17, + "mean_token_accuracy": 0.6923094987869263, + "num_tokens": 13709640.0, + "step": 359 + }, + { + "epoch": 0.04579570029258364, + "grad_norm": 28.125, + "learning_rate": 1.5218312844425604e-07, + "loss": 1.1716, + "mean_token_accuracy": 0.6830503940582275, + "num_tokens": 13751114.0, + "step": 360 + }, + { + "epoch": 0.045922910571174154, + "grad_norm": 26.125, + "learning_rate": 1.526070368800339e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.7087699174880981, + "num_tokens": 13786448.0, + "step": 361 + }, + { + "epoch": 0.04605012084976466, + "grad_norm": 26.75, + "learning_rate": 1.530309453158118e-07, + "loss": 1.12, + "mean_token_accuracy": 0.7064129114151001, + "num_tokens": 13827663.0, + "step": 362 + }, + { + "epoch": 0.04617733112835517, + "grad_norm": 31.0, + "learning_rate": 1.5345485375158963e-07, + "loss": 1.2429, + "mean_token_accuracy": 0.6761044859886169, + "num_tokens": 13866078.0, + "step": 363 + }, + { + "epoch": 0.04630454140694568, + "grad_norm": 23.125, + "learning_rate": 1.5387876218736753e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.7246973514556885, + "num_tokens": 13909767.0, + "step": 364 + }, + { + "epoch": 0.04643175168553619, + "grad_norm": 25.875, + "learning_rate": 1.5430267062314538e-07, + "loss": 1.2021, + "mean_token_accuracy": 0.6861128807067871, + "num_tokens": 13948937.0, + "step": 365 + }, + { + "epoch": 0.0465589619641267, + "grad_norm": 31.625, + "learning_rate": 1.5472657905892328e-07, + "loss": 1.2682, + "mean_token_accuracy": 0.6685630083084106, + "num_tokens": 13984039.0, + "step": 366 + }, + { + "epoch": 0.04668617224271721, + "grad_norm": 29.125, + "learning_rate": 1.5515048749470113e-07, + "loss": 1.2185, + "mean_token_accuracy": 0.6796393394470215, + "num_tokens": 14018162.0, + "step": 367 + }, + { + "epoch": 0.04681338252130772, + "grad_norm": 29.0, + "learning_rate": 1.55574395930479e-07, + "loss": 1.2004, + "mean_token_accuracy": 0.6820600032806396, + "num_tokens": 14056493.0, + "step": 368 + }, + { + "epoch": 0.046940592799898234, + "grad_norm": 30.125, + "learning_rate": 1.5599830436625687e-07, + "loss": 1.2083, + "mean_token_accuracy": 0.6834849715232849, + "num_tokens": 14097530.0, + "step": 369 + }, + { + "epoch": 0.04706780307848874, + "grad_norm": 27.375, + "learning_rate": 1.5642221280203474e-07, + "loss": 1.2116, + "mean_token_accuracy": 0.6849920749664307, + "num_tokens": 14136240.0, + "step": 370 + }, + { + "epoch": 0.04719501335707925, + "grad_norm": 25.25, + "learning_rate": 1.5684612123781262e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.7121806144714355, + "num_tokens": 14171010.0, + "step": 371 + }, + { + "epoch": 0.04732222363566976, + "grad_norm": 26.375, + "learning_rate": 1.572700296735905e-07, + "loss": 1.1678, + "mean_token_accuracy": 0.6992442607879639, + "num_tokens": 14206985.0, + "step": 372 + }, + { + "epoch": 0.04744943391426027, + "grad_norm": 29.0, + "learning_rate": 1.576939381093684e-07, + "loss": 1.2314, + "mean_token_accuracy": 0.6759426593780518, + "num_tokens": 14236728.0, + "step": 373 + }, + { + "epoch": 0.047576644192850785, + "grad_norm": 27.0, + "learning_rate": 1.5811784654514623e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.7058377265930176, + "num_tokens": 14270758.0, + "step": 374 + }, + { + "epoch": 0.04770385447144129, + "grad_norm": 29.125, + "learning_rate": 1.5854175498092413e-07, + "loss": 1.2458, + "mean_token_accuracy": 0.6687042713165283, + "num_tokens": 14315002.0, + "step": 375 + }, + { + "epoch": 0.0478310647500318, + "grad_norm": 28.375, + "learning_rate": 1.5896566341670198e-07, + "loss": 1.2092, + "mean_token_accuracy": 0.6747889518737793, + "num_tokens": 14353385.0, + "step": 376 + }, + { + "epoch": 0.047958275028622314, + "grad_norm": 29.75, + "learning_rate": 1.5938957185247988e-07, + "loss": 1.278, + "mean_token_accuracy": 0.6616823673248291, + "num_tokens": 14391702.0, + "step": 377 + }, + { + "epoch": 0.04808548530721282, + "grad_norm": 29.5, + "learning_rate": 1.5981348028825772e-07, + "loss": 1.2502, + "mean_token_accuracy": 0.6725956201553345, + "num_tokens": 14427876.0, + "step": 378 + }, + { + "epoch": 0.048212695585803336, + "grad_norm": 35.0, + "learning_rate": 1.6023738872403562e-07, + "loss": 1.4152, + "mean_token_accuracy": 0.6350142955780029, + "num_tokens": 14460349.0, + "step": 379 + }, + { + "epoch": 0.048339905864393844, + "grad_norm": 24.375, + "learning_rate": 1.6066129715981347e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.7111331820487976, + "num_tokens": 14497763.0, + "step": 380 + }, + { + "epoch": 0.04846711614298435, + "grad_norm": 26.375, + "learning_rate": 1.6108520559559137e-07, + "loss": 1.1107, + "mean_token_accuracy": 0.7066468000411987, + "num_tokens": 14537172.0, + "step": 381 + }, + { + "epoch": 0.048594326421574865, + "grad_norm": 32.0, + "learning_rate": 1.6150911403136921e-07, + "loss": 1.3147, + "mean_token_accuracy": 0.6726958155632019, + "num_tokens": 14565515.0, + "step": 382 + }, + { + "epoch": 0.04872153670016537, + "grad_norm": 27.0, + "learning_rate": 1.619330224671471e-07, + "loss": 1.1974, + "mean_token_accuracy": 0.6817454099655151, + "num_tokens": 14608104.0, + "step": 383 + }, + { + "epoch": 0.04884874697875588, + "grad_norm": 26.375, + "learning_rate": 1.6235693090292496e-07, + "loss": 1.2143, + "mean_token_accuracy": 0.6831420660018921, + "num_tokens": 14645328.0, + "step": 384 + }, + { + "epoch": 0.048975957257346395, + "grad_norm": 34.0, + "learning_rate": 1.6278083933870286e-07, + "loss": 1.3611, + "mean_token_accuracy": 0.6398276090621948, + "num_tokens": 14678792.0, + "step": 385 + }, + { + "epoch": 0.0491031675359369, + "grad_norm": 28.25, + "learning_rate": 1.632047477744807e-07, + "loss": 1.202, + "mean_token_accuracy": 0.6863372921943665, + "num_tokens": 14715095.0, + "step": 386 + }, + { + "epoch": 0.049230377814527417, + "grad_norm": 26.625, + "learning_rate": 1.6362865621025858e-07, + "loss": 1.1829, + "mean_token_accuracy": 0.6843962669372559, + "num_tokens": 14753641.0, + "step": 387 + }, + { + "epoch": 0.049357588093117924, + "grad_norm": 28.125, + "learning_rate": 1.6405256464603645e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.6864273548126221, + "num_tokens": 14788816.0, + "step": 388 + }, + { + "epoch": 0.04948479837170843, + "grad_norm": 30.25, + "learning_rate": 1.6447647308181432e-07, + "loss": 1.2839, + "mean_token_accuracy": 0.6684393286705017, + "num_tokens": 14821607.0, + "step": 389 + }, + { + "epoch": 0.049612008650298946, + "grad_norm": 30.625, + "learning_rate": 1.649003815175922e-07, + "loss": 1.2739, + "mean_token_accuracy": 0.6662248969078064, + "num_tokens": 14861298.0, + "step": 390 + }, + { + "epoch": 0.04973921892888945, + "grad_norm": 26.25, + "learning_rate": 1.6532428995337007e-07, + "loss": 1.1302, + "mean_token_accuracy": 0.698701024055481, + "num_tokens": 14899951.0, + "step": 391 + }, + { + "epoch": 0.04986642920747997, + "grad_norm": 26.125, + "learning_rate": 1.6574819838914794e-07, + "loss": 1.1172, + "mean_token_accuracy": 0.7071030735969543, + "num_tokens": 14937169.0, + "step": 392 + }, + { + "epoch": 0.049993639486070475, + "grad_norm": 27.75, + "learning_rate": 1.661721068249258e-07, + "loss": 1.2005, + "mean_token_accuracy": 0.6835173964500427, + "num_tokens": 14975549.0, + "step": 393 + }, + { + "epoch": 0.05012084976466098, + "grad_norm": 27.25, + "learning_rate": 1.6659601526070368e-07, + "loss": 1.2035, + "mean_token_accuracy": 0.6858507394790649, + "num_tokens": 15016630.0, + "step": 394 + }, + { + "epoch": 0.0502480600432515, + "grad_norm": 28.875, + "learning_rate": 1.6701992369648156e-07, + "loss": 1.2027, + "mean_token_accuracy": 0.6812106370925903, + "num_tokens": 15057822.0, + "step": 395 + }, + { + "epoch": 0.050375270321842004, + "grad_norm": 28.25, + "learning_rate": 1.6744383213225943e-07, + "loss": 1.165, + "mean_token_accuracy": 0.6795510649681091, + "num_tokens": 15094385.0, + "step": 396 + }, + { + "epoch": 0.05050248060043251, + "grad_norm": 24.0, + "learning_rate": 1.678677405680373e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.7158093452453613, + "num_tokens": 15130969.0, + "step": 397 + }, + { + "epoch": 0.050629690879023026, + "grad_norm": 27.5, + "learning_rate": 1.6829164900381518e-07, + "loss": 1.1623, + "mean_token_accuracy": 0.6916706562042236, + "num_tokens": 15166586.0, + "step": 398 + }, + { + "epoch": 0.05075690115761353, + "grad_norm": 25.125, + "learning_rate": 1.6871555743959305e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.7074527740478516, + "num_tokens": 15209603.0, + "step": 399 + }, + { + "epoch": 0.05088411143620405, + "grad_norm": 25.5, + "learning_rate": 1.6913946587537092e-07, + "loss": 1.1178, + "mean_token_accuracy": 0.6984015107154846, + "num_tokens": 15251542.0, + "step": 400 + }, + { + "epoch": 0.051011321714794555, + "grad_norm": 26.0, + "learning_rate": 1.695633743111488e-07, + "loss": 1.1674, + "mean_token_accuracy": 0.6920915842056274, + "num_tokens": 15288940.0, + "step": 401 + }, + { + "epoch": 0.05113853199338506, + "grad_norm": 27.0, + "learning_rate": 1.6998728274692667e-07, + "loss": 1.1244, + "mean_token_accuracy": 0.7069745659828186, + "num_tokens": 15321965.0, + "step": 402 + }, + { + "epoch": 0.05126574227197558, + "grad_norm": 26.0, + "learning_rate": 1.7041119118270454e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.7104476094245911, + "num_tokens": 15362895.0, + "step": 403 + }, + { + "epoch": 0.051392952550566084, + "grad_norm": 27.25, + "learning_rate": 1.7083509961848238e-07, + "loss": 1.251, + "mean_token_accuracy": 0.6726882457733154, + "num_tokens": 15396675.0, + "step": 404 + }, + { + "epoch": 0.0515201628291566, + "grad_norm": 26.0, + "learning_rate": 1.7125900805426028e-07, + "loss": 1.1987, + "mean_token_accuracy": 0.6783266067504883, + "num_tokens": 15443070.0, + "step": 405 + }, + { + "epoch": 0.051647373107747106, + "grad_norm": 26.875, + "learning_rate": 1.7168291649003813e-07, + "loss": 1.1563, + "mean_token_accuracy": 0.6956391334533691, + "num_tokens": 15476747.0, + "step": 406 + }, + { + "epoch": 0.051774583386337614, + "grad_norm": 26.875, + "learning_rate": 1.7210682492581603e-07, + "loss": 1.1653, + "mean_token_accuracy": 0.6945924758911133, + "num_tokens": 15514866.0, + "step": 407 + }, + { + "epoch": 0.05190179366492813, + "grad_norm": 25.125, + "learning_rate": 1.7253073336159387e-07, + "loss": 1.1309, + "mean_token_accuracy": 0.7020200490951538, + "num_tokens": 15554245.0, + "step": 408 + }, + { + "epoch": 0.052029003943518635, + "grad_norm": 28.875, + "learning_rate": 1.7295464179737177e-07, + "loss": 1.1936, + "mean_token_accuracy": 0.6867693662643433, + "num_tokens": 15593958.0, + "step": 409 + }, + { + "epoch": 0.05215621422210915, + "grad_norm": 26.0, + "learning_rate": 1.7337855023314962e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.7085069417953491, + "num_tokens": 15632567.0, + "step": 410 + }, + { + "epoch": 0.05228342450069966, + "grad_norm": 25.625, + "learning_rate": 1.7380245866892752e-07, + "loss": 1.1324, + "mean_token_accuracy": 0.7040749788284302, + "num_tokens": 15668255.0, + "step": 411 + }, + { + "epoch": 0.052410634779290165, + "grad_norm": 25.875, + "learning_rate": 1.7422636710470536e-07, + "loss": 1.1985, + "mean_token_accuracy": 0.6876128911972046, + "num_tokens": 15705435.0, + "step": 412 + }, + { + "epoch": 0.05253784505788068, + "grad_norm": 30.25, + "learning_rate": 1.7465027554048326e-07, + "loss": 1.2607, + "mean_token_accuracy": 0.6698974370956421, + "num_tokens": 15744657.0, + "step": 413 + }, + { + "epoch": 0.05266505533647119, + "grad_norm": 27.375, + "learning_rate": 1.750741839762611e-07, + "loss": 1.1661, + "mean_token_accuracy": 0.6975141763687134, + "num_tokens": 15777374.0, + "step": 414 + }, + { + "epoch": 0.052792265615061694, + "grad_norm": 24.625, + "learning_rate": 1.75498092412039e-07, + "loss": 1.0785, + "mean_token_accuracy": 0.7088586688041687, + "num_tokens": 15814047.0, + "step": 415 + }, + { + "epoch": 0.05291947589365221, + "grad_norm": 30.375, + "learning_rate": 1.7592200084781686e-07, + "loss": 1.2653, + "mean_token_accuracy": 0.6718404293060303, + "num_tokens": 15849110.0, + "step": 416 + }, + { + "epoch": 0.053046686172242716, + "grad_norm": 25.375, + "learning_rate": 1.7634590928359475e-07, + "loss": 1.1642, + "mean_token_accuracy": 0.6903318762779236, + "num_tokens": 15888910.0, + "step": 417 + }, + { + "epoch": 0.05317389645083323, + "grad_norm": 27.375, + "learning_rate": 1.767698177193726e-07, + "loss": 1.1592, + "mean_token_accuracy": 0.6921442747116089, + "num_tokens": 15925604.0, + "step": 418 + }, + { + "epoch": 0.05330110672942374, + "grad_norm": 25.75, + "learning_rate": 1.771937261551505e-07, + "loss": 1.1127, + "mean_token_accuracy": 0.709343671798706, + "num_tokens": 15961401.0, + "step": 419 + }, + { + "epoch": 0.053428317008014245, + "grad_norm": 28.125, + "learning_rate": 1.7761763459092835e-07, + "loss": 1.1741, + "mean_token_accuracy": 0.6897167563438416, + "num_tokens": 16001029.0, + "step": 420 + }, + { + "epoch": 0.05355552728660476, + "grad_norm": 25.25, + "learning_rate": 1.7804154302670624e-07, + "loss": 1.1253, + "mean_token_accuracy": 0.7049593925476074, + "num_tokens": 16036014.0, + "step": 421 + }, + { + "epoch": 0.05368273756519527, + "grad_norm": 25.5, + "learning_rate": 1.784654514624841e-07, + "loss": 1.1171, + "mean_token_accuracy": 0.7007932662963867, + "num_tokens": 16075373.0, + "step": 422 + }, + { + "epoch": 0.05380994784378578, + "grad_norm": 27.0, + "learning_rate": 1.7888935989826196e-07, + "loss": 1.1739, + "mean_token_accuracy": 0.6859757304191589, + "num_tokens": 16111352.0, + "step": 423 + }, + { + "epoch": 0.05393715812237629, + "grad_norm": 27.25, + "learning_rate": 1.7931326833403984e-07, + "loss": 1.1969, + "mean_token_accuracy": 0.6791154146194458, + "num_tokens": 16148826.0, + "step": 424 + }, + { + "epoch": 0.054064368400966796, + "grad_norm": 22.0, + "learning_rate": 1.797371767698177e-07, + "loss": 0.9879, + "mean_token_accuracy": 0.7310689687728882, + "num_tokens": 16193491.0, + "step": 425 + }, + { + "epoch": 0.05419157867955731, + "grad_norm": 25.375, + "learning_rate": 1.8016108520559558e-07, + "loss": 1.1322, + "mean_token_accuracy": 0.7008160352706909, + "num_tokens": 16235595.0, + "step": 426 + }, + { + "epoch": 0.05431878895814782, + "grad_norm": 28.125, + "learning_rate": 1.8058499364137345e-07, + "loss": 1.2376, + "mean_token_accuracy": 0.6800603866577148, + "num_tokens": 16274105.0, + "step": 427 + }, + { + "epoch": 0.054445999236738325, + "grad_norm": 29.75, + "learning_rate": 1.8100890207715133e-07, + "loss": 1.2789, + "mean_token_accuracy": 0.6552425622940063, + "num_tokens": 16315886.0, + "step": 428 + }, + { + "epoch": 0.05457320951532884, + "grad_norm": 27.875, + "learning_rate": 1.814328105129292e-07, + "loss": 1.2006, + "mean_token_accuracy": 0.6847518682479858, + "num_tokens": 16353021.0, + "step": 429 + }, + { + "epoch": 0.05470041979391935, + "grad_norm": 28.75, + "learning_rate": 1.8185671894870707e-07, + "loss": 1.224, + "mean_token_accuracy": 0.6791858673095703, + "num_tokens": 16380887.0, + "step": 430 + }, + { + "epoch": 0.05482763007250986, + "grad_norm": 29.875, + "learning_rate": 1.8228062738448494e-07, + "loss": 1.2619, + "mean_token_accuracy": 0.6678520441055298, + "num_tokens": 16421066.0, + "step": 431 + }, + { + "epoch": 0.05495484035110037, + "grad_norm": 28.0, + "learning_rate": 1.8270453582026282e-07, + "loss": 1.2142, + "mean_token_accuracy": 0.6812165975570679, + "num_tokens": 16457197.0, + "step": 432 + }, + { + "epoch": 0.055082050629690876, + "grad_norm": 24.0, + "learning_rate": 1.831284442560407e-07, + "loss": 1.0896, + "mean_token_accuracy": 0.7064093351364136, + "num_tokens": 16497385.0, + "step": 433 + }, + { + "epoch": 0.05520926090828139, + "grad_norm": 27.75, + "learning_rate": 1.8355235269181856e-07, + "loss": 1.1523, + "mean_token_accuracy": 0.6893447637557983, + "num_tokens": 16540099.0, + "step": 434 + }, + { + "epoch": 0.0553364711868719, + "grad_norm": 28.0, + "learning_rate": 1.8397626112759643e-07, + "loss": 1.1853, + "mean_token_accuracy": 0.6923507452011108, + "num_tokens": 16572760.0, + "step": 435 + }, + { + "epoch": 0.05546368146546241, + "grad_norm": 23.75, + "learning_rate": 1.844001695633743e-07, + "loss": 1.0772, + "mean_token_accuracy": 0.710911750793457, + "num_tokens": 16608549.0, + "step": 436 + }, + { + "epoch": 0.05559089174405292, + "grad_norm": 26.25, + "learning_rate": 1.8482407799915218e-07, + "loss": 1.164, + "mean_token_accuracy": 0.6978640556335449, + "num_tokens": 16643800.0, + "step": 437 + }, + { + "epoch": 0.05571810202264343, + "grad_norm": 26.25, + "learning_rate": 1.8524798643493005e-07, + "loss": 1.2648, + "mean_token_accuracy": 0.6646109223365784, + "num_tokens": 16677948.0, + "step": 438 + }, + { + "epoch": 0.05584531230123394, + "grad_norm": 26.5, + "learning_rate": 1.8567189487070792e-07, + "loss": 1.1883, + "mean_token_accuracy": 0.6889941096305847, + "num_tokens": 16716905.0, + "step": 439 + }, + { + "epoch": 0.05597252257982445, + "grad_norm": 24.0, + "learning_rate": 1.8609580330648577e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.7139890193939209, + "num_tokens": 16756158.0, + "step": 440 + }, + { + "epoch": 0.05609973285841496, + "grad_norm": 28.625, + "learning_rate": 1.8651971174226367e-07, + "loss": 1.2162, + "mean_token_accuracy": 0.6785497069358826, + "num_tokens": 16794261.0, + "step": 441 + }, + { + "epoch": 0.05622694313700547, + "grad_norm": 26.0, + "learning_rate": 1.8694362017804152e-07, + "loss": 1.1454, + "mean_token_accuracy": 0.6949807405471802, + "num_tokens": 16834953.0, + "step": 442 + }, + { + "epoch": 0.05635415341559598, + "grad_norm": 24.75, + "learning_rate": 1.8736752861381941e-07, + "loss": 1.0713, + "mean_token_accuracy": 0.7109888792037964, + "num_tokens": 16873954.0, + "step": 443 + }, + { + "epoch": 0.05648136369418649, + "grad_norm": 28.0, + "learning_rate": 1.8779143704959726e-07, + "loss": 1.1673, + "mean_token_accuracy": 0.6933863162994385, + "num_tokens": 16906018.0, + "step": 444 + }, + { + "epoch": 0.056608573972777, + "grad_norm": 28.75, + "learning_rate": 1.8821534548537516e-07, + "loss": 1.2449, + "mean_token_accuracy": 0.6748523712158203, + "num_tokens": 16938945.0, + "step": 445 + }, + { + "epoch": 0.05673578425136751, + "grad_norm": 25.625, + "learning_rate": 1.88639253921153e-07, + "loss": 1.1341, + "mean_token_accuracy": 0.7016510963439941, + "num_tokens": 16982214.0, + "step": 446 + }, + { + "epoch": 0.05686299452995802, + "grad_norm": 27.25, + "learning_rate": 1.890631623569309e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.6935659646987915, + "num_tokens": 17024367.0, + "step": 447 + }, + { + "epoch": 0.05699020480854853, + "grad_norm": 27.75, + "learning_rate": 1.8948707079270875e-07, + "loss": 1.2251, + "mean_token_accuracy": 0.6742690801620483, + "num_tokens": 17057451.0, + "step": 448 + }, + { + "epoch": 0.057117415087139044, + "grad_norm": 25.5, + "learning_rate": 1.8991097922848665e-07, + "loss": 1.1239, + "mean_token_accuracy": 0.7018045783042908, + "num_tokens": 17094633.0, + "step": 449 + }, + { + "epoch": 0.05724462536572955, + "grad_norm": 26.25, + "learning_rate": 1.903348876642645e-07, + "loss": 1.175, + "mean_token_accuracy": 0.6842432022094727, + "num_tokens": 17135892.0, + "step": 450 + }, + { + "epoch": 0.05737183564432006, + "grad_norm": 24.875, + "learning_rate": 1.907587961000424e-07, + "loss": 1.0952, + "mean_token_accuracy": 0.707603394985199, + "num_tokens": 17173881.0, + "step": 451 + }, + { + "epoch": 0.05749904592291057, + "grad_norm": 25.25, + "learning_rate": 1.9118270453582024e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.6887599229812622, + "num_tokens": 17215668.0, + "step": 452 + }, + { + "epoch": 0.05762625620150108, + "grad_norm": 27.875, + "learning_rate": 1.9160661297159814e-07, + "loss": 1.2075, + "mean_token_accuracy": 0.6812747716903687, + "num_tokens": 17254339.0, + "step": 453 + }, + { + "epoch": 0.057753466480091595, + "grad_norm": 27.5, + "learning_rate": 1.9203052140737599e-07, + "loss": 1.2152, + "mean_token_accuracy": 0.67823725938797, + "num_tokens": 17288858.0, + "step": 454 + }, + { + "epoch": 0.0578806767586821, + "grad_norm": 24.875, + "learning_rate": 1.9245442984315389e-07, + "loss": 1.1257, + "mean_token_accuracy": 0.7010714411735535, + "num_tokens": 17330601.0, + "step": 455 + }, + { + "epoch": 0.05800788703727261, + "grad_norm": 23.875, + "learning_rate": 1.9287833827893173e-07, + "loss": 1.1474, + "mean_token_accuracy": 0.6926630735397339, + "num_tokens": 17367797.0, + "step": 456 + }, + { + "epoch": 0.058135097315863124, + "grad_norm": 26.75, + "learning_rate": 1.9330224671470963e-07, + "loss": 1.1804, + "mean_token_accuracy": 0.6793103218078613, + "num_tokens": 17403206.0, + "step": 457 + }, + { + "epoch": 0.05826230759445363, + "grad_norm": 27.125, + "learning_rate": 1.9372615515048748e-07, + "loss": 1.2464, + "mean_token_accuracy": 0.6766704320907593, + "num_tokens": 17437207.0, + "step": 458 + }, + { + "epoch": 0.05838951787304414, + "grad_norm": 25.75, + "learning_rate": 1.9415006358626535e-07, + "loss": 1.1582, + "mean_token_accuracy": 0.6936004161834717, + "num_tokens": 17473335.0, + "step": 459 + }, + { + "epoch": 0.05851672815163465, + "grad_norm": 26.625, + "learning_rate": 1.9457397202204322e-07, + "loss": 1.158, + "mean_token_accuracy": 0.6967023015022278, + "num_tokens": 17505104.0, + "step": 460 + }, + { + "epoch": 0.05864393843022516, + "grad_norm": 24.5, + "learning_rate": 1.949978804578211e-07, + "loss": 1.1285, + "mean_token_accuracy": 0.7036781311035156, + "num_tokens": 17539052.0, + "step": 461 + }, + { + "epoch": 0.058771148708815675, + "grad_norm": 24.75, + "learning_rate": 1.9542178889359897e-07, + "loss": 1.1741, + "mean_token_accuracy": 0.6891838312149048, + "num_tokens": 17578942.0, + "step": 462 + }, + { + "epoch": 0.05889835898740618, + "grad_norm": 25.25, + "learning_rate": 1.9584569732937684e-07, + "loss": 1.1899, + "mean_token_accuracy": 0.6949295997619629, + "num_tokens": 17611931.0, + "step": 463 + }, + { + "epoch": 0.05902556926599669, + "grad_norm": 24.0, + "learning_rate": 1.962696057651547e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.7233564853668213, + "num_tokens": 17648764.0, + "step": 464 + }, + { + "epoch": 0.059152779544587204, + "grad_norm": 30.125, + "learning_rate": 1.9669351420093258e-07, + "loss": 1.2439, + "mean_token_accuracy": 0.6758604645729065, + "num_tokens": 17683099.0, + "step": 465 + }, + { + "epoch": 0.05927998982317771, + "grad_norm": 25.125, + "learning_rate": 1.9711742263671046e-07, + "loss": 1.1466, + "mean_token_accuracy": 0.699210524559021, + "num_tokens": 17727120.0, + "step": 466 + }, + { + "epoch": 0.059407200101768226, + "grad_norm": 25.0, + "learning_rate": 1.9754133107248833e-07, + "loss": 1.1273, + "mean_token_accuracy": 0.6931002140045166, + "num_tokens": 17761676.0, + "step": 467 + }, + { + "epoch": 0.059534410380358734, + "grad_norm": 26.0, + "learning_rate": 1.979652395082662e-07, + "loss": 1.1881, + "mean_token_accuracy": 0.6807807087898254, + "num_tokens": 17802311.0, + "step": 468 + }, + { + "epoch": 0.05966162065894924, + "grad_norm": 23.875, + "learning_rate": 1.9838914794404408e-07, + "loss": 1.0502, + "mean_token_accuracy": 0.715239942073822, + "num_tokens": 17839048.0, + "step": 469 + }, + { + "epoch": 0.059788830937539755, + "grad_norm": 23.75, + "learning_rate": 1.9881305637982195e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.7020200490951538, + "num_tokens": 17881510.0, + "step": 470 + }, + { + "epoch": 0.05991604121613026, + "grad_norm": 27.375, + "learning_rate": 1.9923696481559982e-07, + "loss": 1.2326, + "mean_token_accuracy": 0.6809006929397583, + "num_tokens": 17915080.0, + "step": 471 + }, + { + "epoch": 0.06004325149472077, + "grad_norm": 26.875, + "learning_rate": 1.996608732513777e-07, + "loss": 1.2083, + "mean_token_accuracy": 0.6753402352333069, + "num_tokens": 17957972.0, + "step": 472 + }, + { + "epoch": 0.060170461773311285, + "grad_norm": 24.25, + "learning_rate": 2.0008478168715557e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.7143442034721375, + "num_tokens": 17998008.0, + "step": 473 + }, + { + "epoch": 0.06029767205190179, + "grad_norm": 22.875, + "learning_rate": 2.0050869012293344e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.7130613327026367, + "num_tokens": 18032427.0, + "step": 474 + }, + { + "epoch": 0.060424882330492306, + "grad_norm": 25.25, + "learning_rate": 2.009325985587113e-07, + "loss": 1.1168, + "mean_token_accuracy": 0.7034891843795776, + "num_tokens": 18069848.0, + "step": 475 + }, + { + "epoch": 0.060552092609082814, + "grad_norm": 29.625, + "learning_rate": 2.0135650699448918e-07, + "loss": 1.253, + "mean_token_accuracy": 0.664897084236145, + "num_tokens": 18109240.0, + "step": 476 + }, + { + "epoch": 0.06067930288767332, + "grad_norm": 26.875, + "learning_rate": 2.0178041543026706e-07, + "loss": 1.1981, + "mean_token_accuracy": 0.685088038444519, + "num_tokens": 18146849.0, + "step": 477 + }, + { + "epoch": 0.060806513166263836, + "grad_norm": 28.125, + "learning_rate": 2.022043238660449e-07, + "loss": 1.2589, + "mean_token_accuracy": 0.6646258234977722, + "num_tokens": 18186872.0, + "step": 478 + }, + { + "epoch": 0.06093372344485434, + "grad_norm": 29.125, + "learning_rate": 2.026282323018228e-07, + "loss": 1.2378, + "mean_token_accuracy": 0.6793026328086853, + "num_tokens": 18222299.0, + "step": 479 + }, + { + "epoch": 0.06106093372344486, + "grad_norm": 24.75, + "learning_rate": 2.0305214073760065e-07, + "loss": 1.205, + "mean_token_accuracy": 0.6858837604522705, + "num_tokens": 18261645.0, + "step": 480 + }, + { + "epoch": 0.061188144002035365, + "grad_norm": 27.25, + "learning_rate": 2.0347604917337855e-07, + "loss": 1.2154, + "mean_token_accuracy": 0.6818623542785645, + "num_tokens": 18292970.0, + "step": 481 + }, + { + "epoch": 0.06131535428062587, + "grad_norm": 23.25, + "learning_rate": 2.038999576091564e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.7125015258789062, + "num_tokens": 18331133.0, + "step": 482 + }, + { + "epoch": 0.06144256455921639, + "grad_norm": 28.25, + "learning_rate": 2.043238660449343e-07, + "loss": 1.2267, + "mean_token_accuracy": 0.6676434278488159, + "num_tokens": 18368905.0, + "step": 483 + }, + { + "epoch": 0.061569774837806894, + "grad_norm": 25.875, + "learning_rate": 2.0474777448071214e-07, + "loss": 1.1269, + "mean_token_accuracy": 0.6996800303459167, + "num_tokens": 18406362.0, + "step": 484 + }, + { + "epoch": 0.0616969851163974, + "grad_norm": 27.75, + "learning_rate": 2.0517168291649004e-07, + "loss": 1.2091, + "mean_token_accuracy": 0.6825910210609436, + "num_tokens": 18439246.0, + "step": 485 + }, + { + "epoch": 0.061824195394987916, + "grad_norm": 28.875, + "learning_rate": 2.0559559135226788e-07, + "loss": 1.2117, + "mean_token_accuracy": 0.6758161187171936, + "num_tokens": 18473408.0, + "step": 486 + }, + { + "epoch": 0.06195140567357842, + "grad_norm": 24.625, + "learning_rate": 2.0601949978804578e-07, + "loss": 1.0951, + "mean_token_accuracy": 0.7072306871414185, + "num_tokens": 18513246.0, + "step": 487 + }, + { + "epoch": 0.06207861595216894, + "grad_norm": 22.875, + "learning_rate": 2.0644340822382363e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.7111845016479492, + "num_tokens": 18549125.0, + "step": 488 + }, + { + "epoch": 0.062205826230759445, + "grad_norm": 25.25, + "learning_rate": 2.0686731665960153e-07, + "loss": 1.1443, + "mean_token_accuracy": 0.7027828097343445, + "num_tokens": 18585306.0, + "step": 489 + }, + { + "epoch": 0.06233303650934995, + "grad_norm": 24.5, + "learning_rate": 2.0729122509537937e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.7121580243110657, + "num_tokens": 18624378.0, + "step": 490 + }, + { + "epoch": 0.06246024678794047, + "grad_norm": 26.375, + "learning_rate": 2.0771513353115727e-07, + "loss": 1.1752, + "mean_token_accuracy": 0.6881194710731506, + "num_tokens": 18660814.0, + "step": 491 + }, + { + "epoch": 0.06258745706653097, + "grad_norm": 23.625, + "learning_rate": 2.0813904196693512e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.7198270559310913, + "num_tokens": 18702129.0, + "step": 492 + }, + { + "epoch": 0.06271466734512149, + "grad_norm": 26.125, + "learning_rate": 2.0856295040271302e-07, + "loss": 1.2145, + "mean_token_accuracy": 0.6783840656280518, + "num_tokens": 18737496.0, + "step": 493 + }, + { + "epoch": 0.06284187762371199, + "grad_norm": 23.625, + "learning_rate": 2.0898685883849086e-07, + "loss": 1.1047, + "mean_token_accuracy": 0.7083240747451782, + "num_tokens": 18779332.0, + "step": 494 + }, + { + "epoch": 0.0629690879023025, + "grad_norm": 23.625, + "learning_rate": 2.0941076727426874e-07, + "loss": 1.1118, + "mean_token_accuracy": 0.7047414779663086, + "num_tokens": 18817976.0, + "step": 495 + }, + { + "epoch": 0.06309629818089302, + "grad_norm": 22.625, + "learning_rate": 2.098346757100466e-07, + "loss": 1.0604, + "mean_token_accuracy": 0.7145814895629883, + "num_tokens": 18862860.0, + "step": 496 + }, + { + "epoch": 0.06322350845948353, + "grad_norm": 27.125, + "learning_rate": 2.1025858414582448e-07, + "loss": 1.1783, + "mean_token_accuracy": 0.6855897903442383, + "num_tokens": 18899025.0, + "step": 497 + }, + { + "epoch": 0.06335071873807403, + "grad_norm": 26.25, + "learning_rate": 2.1068249258160238e-07, + "loss": 1.1698, + "mean_token_accuracy": 0.6838425993919373, + "num_tokens": 18933118.0, + "step": 498 + }, + { + "epoch": 0.06347792901666455, + "grad_norm": 24.625, + "learning_rate": 2.1110640101738023e-07, + "loss": 1.1251, + "mean_token_accuracy": 0.704369068145752, + "num_tokens": 18969165.0, + "step": 499 + }, + { + "epoch": 0.06360513929525506, + "grad_norm": 26.75, + "learning_rate": 2.1153030945315813e-07, + "loss": 1.1409, + "mean_token_accuracy": 0.6929699182510376, + "num_tokens": 19003882.0, + "step": 500 + }, + { + "epoch": 0.06373234957384556, + "grad_norm": 24.375, + "learning_rate": 2.1195421788893597e-07, + "loss": 1.1353, + "mean_token_accuracy": 0.7014108300209045, + "num_tokens": 19037540.0, + "step": 501 + }, + { + "epoch": 0.06385955985243608, + "grad_norm": 23.75, + "learning_rate": 2.1237812632471387e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.6987220048904419, + "num_tokens": 19077270.0, + "step": 502 + }, + { + "epoch": 0.06398677013102659, + "grad_norm": 29.0, + "learning_rate": 2.1280203476049172e-07, + "loss": 1.2592, + "mean_token_accuracy": 0.6610092520713806, + "num_tokens": 19117579.0, + "step": 503 + }, + { + "epoch": 0.06411398040961709, + "grad_norm": 26.0, + "learning_rate": 2.1322594319626962e-07, + "loss": 1.1893, + "mean_token_accuracy": 0.6812867522239685, + "num_tokens": 19156599.0, + "step": 504 + }, + { + "epoch": 0.0642411906882076, + "grad_norm": 27.125, + "learning_rate": 2.1364985163204746e-07, + "loss": 1.2075, + "mean_token_accuracy": 0.6837060451507568, + "num_tokens": 19187367.0, + "step": 505 + }, + { + "epoch": 0.06436840096679812, + "grad_norm": 23.75, + "learning_rate": 2.1407376006782536e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.7124614119529724, + "num_tokens": 19223916.0, + "step": 506 + }, + { + "epoch": 0.06449561124538863, + "grad_norm": 24.0, + "learning_rate": 2.144976685036032e-07, + "loss": 1.1068, + "mean_token_accuracy": 0.7088305354118347, + "num_tokens": 19260336.0, + "step": 507 + }, + { + "epoch": 0.06462282152397913, + "grad_norm": 26.5, + "learning_rate": 2.149215769393811e-07, + "loss": 1.2026, + "mean_token_accuracy": 0.6813916563987732, + "num_tokens": 19299895.0, + "step": 508 + }, + { + "epoch": 0.06475003180256965, + "grad_norm": 25.375, + "learning_rate": 2.1534548537515895e-07, + "loss": 1.1231, + "mean_token_accuracy": 0.7011598944664001, + "num_tokens": 19336614.0, + "step": 509 + }, + { + "epoch": 0.06487724208116016, + "grad_norm": 26.0, + "learning_rate": 2.1576939381093685e-07, + "loss": 1.1845, + "mean_token_accuracy": 0.6874239444732666, + "num_tokens": 19378683.0, + "step": 510 + }, + { + "epoch": 0.06500445235975066, + "grad_norm": 25.0, + "learning_rate": 2.161933022467147e-07, + "loss": 1.1273, + "mean_token_accuracy": 0.6972498297691345, + "num_tokens": 19422598.0, + "step": 511 + }, + { + "epoch": 0.06513166263834118, + "grad_norm": 25.25, + "learning_rate": 2.166172106824926e-07, + "loss": 1.128, + "mean_token_accuracy": 0.7019932270050049, + "num_tokens": 19466972.0, + "step": 512 + }, + { + "epoch": 0.06525887291693169, + "grad_norm": 25.875, + "learning_rate": 2.1704111911827044e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.691651463508606, + "num_tokens": 19503809.0, + "step": 513 + }, + { + "epoch": 0.0653860831955222, + "grad_norm": 27.5, + "learning_rate": 2.1746502755404831e-07, + "loss": 1.2091, + "mean_token_accuracy": 0.6842914819717407, + "num_tokens": 19545713.0, + "step": 514 + }, + { + "epoch": 0.06551329347411271, + "grad_norm": 26.5, + "learning_rate": 2.178889359898262e-07, + "loss": 1.1961, + "mean_token_accuracy": 0.6865599155426025, + "num_tokens": 19582272.0, + "step": 515 + }, + { + "epoch": 0.06564050375270322, + "grad_norm": 26.0, + "learning_rate": 2.1831284442560406e-07, + "loss": 1.1915, + "mean_token_accuracy": 0.6864949464797974, + "num_tokens": 19625574.0, + "step": 516 + }, + { + "epoch": 0.06576771403129372, + "grad_norm": 26.125, + "learning_rate": 2.1873675286138193e-07, + "loss": 1.1775, + "mean_token_accuracy": 0.6775625944137573, + "num_tokens": 19667791.0, + "step": 517 + }, + { + "epoch": 0.06589492430988424, + "grad_norm": 28.5, + "learning_rate": 2.191606612971598e-07, + "loss": 1.2631, + "mean_token_accuracy": 0.6683406829833984, + "num_tokens": 19708289.0, + "step": 518 + }, + { + "epoch": 0.06602213458847475, + "grad_norm": 25.75, + "learning_rate": 2.1958456973293768e-07, + "loss": 1.184, + "mean_token_accuracy": 0.6738008260726929, + "num_tokens": 19747444.0, + "step": 519 + }, + { + "epoch": 0.06614934486706527, + "grad_norm": 22.375, + "learning_rate": 2.2000847816871555e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.7247603535652161, + "num_tokens": 19787061.0, + "step": 520 + }, + { + "epoch": 0.06627655514565577, + "grad_norm": 24.125, + "learning_rate": 2.2043238660449342e-07, + "loss": 1.0937, + "mean_token_accuracy": 0.7034186124801636, + "num_tokens": 19826893.0, + "step": 521 + }, + { + "epoch": 0.06640376542424628, + "grad_norm": 22.875, + "learning_rate": 2.208562950402713e-07, + "loss": 1.0795, + "mean_token_accuracy": 0.7117700576782227, + "num_tokens": 19864285.0, + "step": 522 + }, + { + "epoch": 0.0665309757028368, + "grad_norm": 27.375, + "learning_rate": 2.2128020347604917e-07, + "loss": 1.1514, + "mean_token_accuracy": 0.6918144822120667, + "num_tokens": 19903806.0, + "step": 523 + }, + { + "epoch": 0.0666581859814273, + "grad_norm": 22.625, + "learning_rate": 2.2170411191182704e-07, + "loss": 1.0671, + "mean_token_accuracy": 0.7116342782974243, + "num_tokens": 19950382.0, + "step": 524 + }, + { + "epoch": 0.06678539626001781, + "grad_norm": 25.0, + "learning_rate": 2.221280203476049e-07, + "loss": 1.1521, + "mean_token_accuracy": 0.6916357278823853, + "num_tokens": 19984765.0, + "step": 525 + }, + { + "epoch": 0.06691260653860832, + "grad_norm": 24.125, + "learning_rate": 2.2255192878338279e-07, + "loss": 1.1898, + "mean_token_accuracy": 0.6927103996276855, + "num_tokens": 20021081.0, + "step": 526 + }, + { + "epoch": 0.06703981681719882, + "grad_norm": 24.625, + "learning_rate": 2.2297583721916066e-07, + "loss": 1.1837, + "mean_token_accuracy": 0.6893125772476196, + "num_tokens": 20055860.0, + "step": 527 + }, + { + "epoch": 0.06716702709578934, + "grad_norm": 24.25, + "learning_rate": 2.2339974565493853e-07, + "loss": 1.1083, + "mean_token_accuracy": 0.7067604064941406, + "num_tokens": 20095023.0, + "step": 528 + }, + { + "epoch": 0.06729423737437985, + "grad_norm": 25.125, + "learning_rate": 2.238236540907164e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6944122314453125, + "num_tokens": 20134620.0, + "step": 529 + }, + { + "epoch": 0.06742144765297035, + "grad_norm": 25.5, + "learning_rate": 2.2424756252649428e-07, + "loss": 1.1785, + "mean_token_accuracy": 0.6861029863357544, + "num_tokens": 20169690.0, + "step": 530 + }, + { + "epoch": 0.06754865793156087, + "grad_norm": 26.25, + "learning_rate": 2.2467147096227215e-07, + "loss": 1.1838, + "mean_token_accuracy": 0.6882814168930054, + "num_tokens": 20203609.0, + "step": 531 + }, + { + "epoch": 0.06767586821015138, + "grad_norm": 26.75, + "learning_rate": 2.2509537939805002e-07, + "loss": 1.1967, + "mean_token_accuracy": 0.6859830021858215, + "num_tokens": 20234616.0, + "step": 532 + }, + { + "epoch": 0.0678030784887419, + "grad_norm": 24.125, + "learning_rate": 2.2551928783382787e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.7003049254417419, + "num_tokens": 20271634.0, + "step": 533 + }, + { + "epoch": 0.0679302887673324, + "grad_norm": 25.25, + "learning_rate": 2.2594319626960577e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.699425458908081, + "num_tokens": 20309683.0, + "step": 534 + }, + { + "epoch": 0.06805749904592291, + "grad_norm": 21.25, + "learning_rate": 2.263671047053836e-07, + "loss": 1.0071, + "mean_token_accuracy": 0.7190131545066833, + "num_tokens": 20346013.0, + "step": 535 + }, + { + "epoch": 0.06818470932451343, + "grad_norm": 24.125, + "learning_rate": 2.267910131411615e-07, + "loss": 1.119, + "mean_token_accuracy": 0.7018981575965881, + "num_tokens": 20387613.0, + "step": 536 + }, + { + "epoch": 0.06831191960310393, + "grad_norm": 25.875, + "learning_rate": 2.2721492157693936e-07, + "loss": 1.181, + "mean_token_accuracy": 0.6887447237968445, + "num_tokens": 20417579.0, + "step": 537 + }, + { + "epoch": 0.06843912988169444, + "grad_norm": 21.625, + "learning_rate": 2.2763883001271726e-07, + "loss": 1.0632, + "mean_token_accuracy": 0.7159764766693115, + "num_tokens": 20455100.0, + "step": 538 + }, + { + "epoch": 0.06856634016028496, + "grad_norm": 26.5, + "learning_rate": 2.280627384484951e-07, + "loss": 1.187, + "mean_token_accuracy": 0.682623028755188, + "num_tokens": 20496458.0, + "step": 539 + }, + { + "epoch": 0.06869355043887546, + "grad_norm": 25.5, + "learning_rate": 2.28486646884273e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6977063417434692, + "num_tokens": 20533758.0, + "step": 540 + }, + { + "epoch": 0.06882076071746597, + "grad_norm": 26.0, + "learning_rate": 2.2891055532005085e-07, + "loss": 1.2086, + "mean_token_accuracy": 0.6828888058662415, + "num_tokens": 20567259.0, + "step": 541 + }, + { + "epoch": 0.06894797099605648, + "grad_norm": 26.375, + "learning_rate": 2.2933446375582875e-07, + "loss": 1.1945, + "mean_token_accuracy": 0.6846005916595459, + "num_tokens": 20599333.0, + "step": 542 + }, + { + "epoch": 0.06907518127464699, + "grad_norm": 22.875, + "learning_rate": 2.297583721916066e-07, + "loss": 1.0559, + "mean_token_accuracy": 0.7191632986068726, + "num_tokens": 20640195.0, + "step": 543 + }, + { + "epoch": 0.0692023915532375, + "grad_norm": 22.0, + "learning_rate": 2.301822806273845e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.7136548757553101, + "num_tokens": 20682617.0, + "step": 544 + }, + { + "epoch": 0.06932960183182801, + "grad_norm": 26.125, + "learning_rate": 2.3060618906316234e-07, + "loss": 1.2032, + "mean_token_accuracy": 0.6828359961509705, + "num_tokens": 20718032.0, + "step": 545 + }, + { + "epoch": 0.06945681211041853, + "grad_norm": 24.625, + "learning_rate": 2.3103009749894024e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6994045376777649, + "num_tokens": 20755190.0, + "step": 546 + }, + { + "epoch": 0.06958402238900903, + "grad_norm": 25.875, + "learning_rate": 2.3145400593471808e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.6887730360031128, + "num_tokens": 20799088.0, + "step": 547 + }, + { + "epoch": 0.06971123266759954, + "grad_norm": 29.5, + "learning_rate": 2.3187791437049598e-07, + "loss": 1.2746, + "mean_token_accuracy": 0.6587653160095215, + "num_tokens": 20831819.0, + "step": 548 + }, + { + "epoch": 0.06983844294619006, + "grad_norm": 27.0, + "learning_rate": 2.3230182280627383e-07, + "loss": 1.2026, + "mean_token_accuracy": 0.6877304911613464, + "num_tokens": 20861291.0, + "step": 549 + }, + { + "epoch": 0.06996565322478056, + "grad_norm": 23.5, + "learning_rate": 2.327257312420517e-07, + "loss": 1.0916, + "mean_token_accuracy": 0.7143093347549438, + "num_tokens": 20897446.0, + "step": 550 + }, + { + "epoch": 0.07009286350337107, + "grad_norm": 24.5, + "learning_rate": 2.3314963967782957e-07, + "loss": 1.0967, + "mean_token_accuracy": 0.7023328542709351, + "num_tokens": 20934423.0, + "step": 551 + }, + { + "epoch": 0.07022007378196159, + "grad_norm": 24.125, + "learning_rate": 2.3357354811360745e-07, + "loss": 1.112, + "mean_token_accuracy": 0.703018307685852, + "num_tokens": 20978394.0, + "step": 552 + }, + { + "epoch": 0.07034728406055209, + "grad_norm": 26.125, + "learning_rate": 2.3399745654938532e-07, + "loss": 1.1975, + "mean_token_accuracy": 0.684022068977356, + "num_tokens": 21010669.0, + "step": 553 + }, + { + "epoch": 0.0704744943391426, + "grad_norm": 23.5, + "learning_rate": 2.344213649851632e-07, + "loss": 1.1196, + "mean_token_accuracy": 0.7049205303192139, + "num_tokens": 21047696.0, + "step": 554 + }, + { + "epoch": 0.07060170461773312, + "grad_norm": 27.75, + "learning_rate": 2.3484527342094106e-07, + "loss": 1.2271, + "mean_token_accuracy": 0.6742489337921143, + "num_tokens": 21091437.0, + "step": 555 + }, + { + "epoch": 0.07072891489632362, + "grad_norm": 26.125, + "learning_rate": 2.3526918185671894e-07, + "loss": 1.1888, + "mean_token_accuracy": 0.6863871216773987, + "num_tokens": 21128880.0, + "step": 556 + }, + { + "epoch": 0.07085612517491413, + "grad_norm": 23.875, + "learning_rate": 2.356930902924968e-07, + "loss": 1.1159, + "mean_token_accuracy": 0.7016575932502747, + "num_tokens": 21168523.0, + "step": 557 + }, + { + "epoch": 0.07098333545350465, + "grad_norm": 27.0, + "learning_rate": 2.3611699872827468e-07, + "loss": 1.2573, + "mean_token_accuracy": 0.6652576327323914, + "num_tokens": 21211376.0, + "step": 558 + }, + { + "epoch": 0.07111054573209516, + "grad_norm": 24.25, + "learning_rate": 2.3654090716405255e-07, + "loss": 1.1329, + "mean_token_accuracy": 0.6981450319290161, + "num_tokens": 21249162.0, + "step": 559 + }, + { + "epoch": 0.07123775601068566, + "grad_norm": 27.25, + "learning_rate": 2.3696481559983043e-07, + "loss": 1.2243, + "mean_token_accuracy": 0.6716225147247314, + "num_tokens": 21282351.0, + "step": 560 + }, + { + "epoch": 0.07136496628927617, + "grad_norm": 27.5, + "learning_rate": 2.373887240356083e-07, + "loss": 1.1809, + "mean_token_accuracy": 0.6844885945320129, + "num_tokens": 21319508.0, + "step": 561 + }, + { + "epoch": 0.07149217656786669, + "grad_norm": 22.75, + "learning_rate": 2.3781263247138617e-07, + "loss": 1.0856, + "mean_token_accuracy": 0.7055985927581787, + "num_tokens": 21357111.0, + "step": 562 + }, + { + "epoch": 0.07161938684645719, + "grad_norm": 24.5, + "learning_rate": 2.3823654090716404e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.6981772184371948, + "num_tokens": 21393222.0, + "step": 563 + }, + { + "epoch": 0.0717465971250477, + "grad_norm": 22.75, + "learning_rate": 2.386604493429419e-07, + "loss": 1.0599, + "mean_token_accuracy": 0.7176359295845032, + "num_tokens": 21429595.0, + "step": 564 + }, + { + "epoch": 0.07187380740363822, + "grad_norm": 27.75, + "learning_rate": 2.390843577787198e-07, + "loss": 1.2259, + "mean_token_accuracy": 0.6690883636474609, + "num_tokens": 21467341.0, + "step": 565 + }, + { + "epoch": 0.07200101768222872, + "grad_norm": 22.0, + "learning_rate": 2.3950826621449766e-07, + "loss": 0.9746, + "mean_token_accuracy": 0.7331970930099487, + "num_tokens": 21510927.0, + "step": 566 + }, + { + "epoch": 0.07212822796081923, + "grad_norm": 24.375, + "learning_rate": 2.3993217465027556e-07, + "loss": 1.1081, + "mean_token_accuracy": 0.6991102695465088, + "num_tokens": 21551604.0, + "step": 567 + }, + { + "epoch": 0.07225543823940975, + "grad_norm": 22.625, + "learning_rate": 2.403560830860534e-07, + "loss": 1.1584, + "mean_token_accuracy": 0.6946207284927368, + "num_tokens": 21591871.0, + "step": 568 + }, + { + "epoch": 0.07238264851800025, + "grad_norm": 25.125, + "learning_rate": 2.4077999152183125e-07, + "loss": 1.1774, + "mean_token_accuracy": 0.6783366799354553, + "num_tokens": 21629810.0, + "step": 569 + }, + { + "epoch": 0.07250985879659076, + "grad_norm": 22.125, + "learning_rate": 2.4120389995760915e-07, + "loss": 1.0389, + "mean_token_accuracy": 0.7231581211090088, + "num_tokens": 21668706.0, + "step": 570 + }, + { + "epoch": 0.07263706907518128, + "grad_norm": 23.625, + "learning_rate": 2.41627808393387e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6975125670433044, + "num_tokens": 21712725.0, + "step": 571 + }, + { + "epoch": 0.07276427935377179, + "grad_norm": 26.25, + "learning_rate": 2.420517168291649e-07, + "loss": 1.1629, + "mean_token_accuracy": 0.6979041695594788, + "num_tokens": 21746086.0, + "step": 572 + }, + { + "epoch": 0.07289148963236229, + "grad_norm": 22.75, + "learning_rate": 2.4247562526494274e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.7048013210296631, + "num_tokens": 21779573.0, + "step": 573 + }, + { + "epoch": 0.0730186999109528, + "grad_norm": 24.0, + "learning_rate": 2.4289953370072064e-07, + "loss": 1.161, + "mean_token_accuracy": 0.6897958517074585, + "num_tokens": 21818766.0, + "step": 574 + }, + { + "epoch": 0.07314591018954332, + "grad_norm": 23.75, + "learning_rate": 2.433234421364985e-07, + "loss": 1.0977, + "mean_token_accuracy": 0.7076700925827026, + "num_tokens": 21855539.0, + "step": 575 + }, + { + "epoch": 0.07327312046813382, + "grad_norm": 23.625, + "learning_rate": 2.437473505722764e-07, + "loss": 1.0506, + "mean_token_accuracy": 0.7191532254219055, + "num_tokens": 21892567.0, + "step": 576 + }, + { + "epoch": 0.07340033074672433, + "grad_norm": 23.0, + "learning_rate": 2.4417125900805423e-07, + "loss": 1.0934, + "mean_token_accuracy": 0.7027009129524231, + "num_tokens": 21928607.0, + "step": 577 + }, + { + "epoch": 0.07352754102531485, + "grad_norm": 25.75, + "learning_rate": 2.4459516744383213e-07, + "loss": 1.1661, + "mean_token_accuracy": 0.688184380531311, + "num_tokens": 21963781.0, + "step": 578 + }, + { + "epoch": 0.07365475130390535, + "grad_norm": 23.125, + "learning_rate": 2.4501907587961e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.7073572874069214, + "num_tokens": 21999140.0, + "step": 579 + }, + { + "epoch": 0.07378196158249586, + "grad_norm": 27.5, + "learning_rate": 2.454429843153879e-07, + "loss": 1.2343, + "mean_token_accuracy": 0.6706569194793701, + "num_tokens": 22038703.0, + "step": 580 + }, + { + "epoch": 0.07390917186108638, + "grad_norm": 28.0, + "learning_rate": 2.458668927511657e-07, + "loss": 1.1775, + "mean_token_accuracy": 0.6771315336227417, + "num_tokens": 22071052.0, + "step": 581 + }, + { + "epoch": 0.07403638213967688, + "grad_norm": 22.5, + "learning_rate": 2.462908011869436e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.7183138132095337, + "num_tokens": 22109629.0, + "step": 582 + }, + { + "epoch": 0.0741635924182674, + "grad_norm": 21.875, + "learning_rate": 2.4671470962272147e-07, + "loss": 1.0563, + "mean_token_accuracy": 0.715839684009552, + "num_tokens": 22151544.0, + "step": 583 + }, + { + "epoch": 0.07429080269685791, + "grad_norm": 22.125, + "learning_rate": 2.4713861805849937e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.7201766967773438, + "num_tokens": 22189241.0, + "step": 584 + }, + { + "epoch": 0.07441801297544842, + "grad_norm": 21.875, + "learning_rate": 2.475625264942772e-07, + "loss": 1.0489, + "mean_token_accuracy": 0.7140419483184814, + "num_tokens": 22229138.0, + "step": 585 + }, + { + "epoch": 0.07454522325403892, + "grad_norm": 24.0, + "learning_rate": 2.479864349300551e-07, + "loss": 1.1269, + "mean_token_accuracy": 0.6967936754226685, + "num_tokens": 22264930.0, + "step": 586 + }, + { + "epoch": 0.07467243353262944, + "grad_norm": 22.75, + "learning_rate": 2.4841034336583296e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.7113279700279236, + "num_tokens": 22297162.0, + "step": 587 + }, + { + "epoch": 0.07479964381121995, + "grad_norm": 25.375, + "learning_rate": 2.488342518016108e-07, + "loss": 1.1645, + "mean_token_accuracy": 0.690231442451477, + "num_tokens": 22334779.0, + "step": 588 + }, + { + "epoch": 0.07492685408981045, + "grad_norm": 22.25, + "learning_rate": 2.492581602373887e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.7162479162216187, + "num_tokens": 22372820.0, + "step": 589 + }, + { + "epoch": 0.07505406436840097, + "grad_norm": 23.5, + "learning_rate": 2.4968206867316655e-07, + "loss": 1.1115, + "mean_token_accuracy": 0.6963298320770264, + "num_tokens": 22417841.0, + "step": 590 + }, + { + "epoch": 0.07518127464699148, + "grad_norm": 22.125, + "learning_rate": 2.5010597710894445e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.7262516617774963, + "num_tokens": 22449133.0, + "step": 591 + }, + { + "epoch": 0.07530848492558198, + "grad_norm": 23.375, + "learning_rate": 2.505298855447223e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.7031003832817078, + "num_tokens": 22491145.0, + "step": 592 + }, + { + "epoch": 0.0754356952041725, + "grad_norm": 22.375, + "learning_rate": 2.509537939805002e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.7094952464103699, + "num_tokens": 22531738.0, + "step": 593 + }, + { + "epoch": 0.07556290548276301, + "grad_norm": 22.625, + "learning_rate": 2.513777024162781e-07, + "loss": 1.1039, + "mean_token_accuracy": 0.7086392641067505, + "num_tokens": 22563005.0, + "step": 594 + }, + { + "epoch": 0.07569011576135352, + "grad_norm": 23.25, + "learning_rate": 2.5180161085205594e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.7190022468566895, + "num_tokens": 22601333.0, + "step": 595 + }, + { + "epoch": 0.07581732603994402, + "grad_norm": 23.75, + "learning_rate": 2.522255192878338e-07, + "loss": 1.1298, + "mean_token_accuracy": 0.6940996646881104, + "num_tokens": 22639528.0, + "step": 596 + }, + { + "epoch": 0.07594453631853454, + "grad_norm": 23.125, + "learning_rate": 2.526494277236117e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.7044841051101685, + "num_tokens": 22675290.0, + "step": 597 + }, + { + "epoch": 0.07607174659712505, + "grad_norm": 23.125, + "learning_rate": 2.530733361593896e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.7075134515762329, + "num_tokens": 22712823.0, + "step": 598 + }, + { + "epoch": 0.07619895687571555, + "grad_norm": 23.125, + "learning_rate": 2.5349724459516743e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.7049279808998108, + "num_tokens": 22750916.0, + "step": 599 + }, + { + "epoch": 0.07632616715430607, + "grad_norm": 22.125, + "learning_rate": 2.539211530309453e-07, + "loss": 1.0585, + "mean_token_accuracy": 0.7151417136192322, + "num_tokens": 22782537.0, + "step": 600 + }, + { + "epoch": 0.07645337743289658, + "grad_norm": 25.0, + "learning_rate": 2.543450614667232e-07, + "loss": 1.2022, + "mean_token_accuracy": 0.6816546320915222, + "num_tokens": 22814805.0, + "step": 601 + }, + { + "epoch": 0.07658058771148708, + "grad_norm": 22.875, + "learning_rate": 2.547689699025011e-07, + "loss": 1.1321, + "mean_token_accuracy": 0.6904566287994385, + "num_tokens": 22852654.0, + "step": 602 + }, + { + "epoch": 0.0767077979900776, + "grad_norm": 23.375, + "learning_rate": 2.551928783382789e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.7054063081741333, + "num_tokens": 22894205.0, + "step": 603 + }, + { + "epoch": 0.07683500826866811, + "grad_norm": 23.875, + "learning_rate": 2.5561678677405677e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.7095728516578674, + "num_tokens": 22924095.0, + "step": 604 + }, + { + "epoch": 0.07696221854725861, + "grad_norm": 24.875, + "learning_rate": 2.5604069520983467e-07, + "loss": 1.1868, + "mean_token_accuracy": 0.6770404577255249, + "num_tokens": 22960857.0, + "step": 605 + }, + { + "epoch": 0.07708942882584913, + "grad_norm": 21.625, + "learning_rate": 2.564646036456125e-07, + "loss": 1.0623, + "mean_token_accuracy": 0.7071213722229004, + "num_tokens": 23004670.0, + "step": 606 + }, + { + "epoch": 0.07721663910443964, + "grad_norm": 19.5, + "learning_rate": 2.568885120813904e-07, + "loss": 1.0477, + "mean_token_accuracy": 0.7144882082939148, + "num_tokens": 23041999.0, + "step": 607 + }, + { + "epoch": 0.07734384938303016, + "grad_norm": 21.375, + "learning_rate": 2.5731242051716826e-07, + "loss": 1.0627, + "mean_token_accuracy": 0.7149096131324768, + "num_tokens": 23080704.0, + "step": 608 + }, + { + "epoch": 0.07747105966162066, + "grad_norm": 24.25, + "learning_rate": 2.5773632895294616e-07, + "loss": 1.1287, + "mean_token_accuracy": 0.6957448124885559, + "num_tokens": 23118662.0, + "step": 609 + }, + { + "epoch": 0.07759826994021117, + "grad_norm": 20.75, + "learning_rate": 2.58160237388724e-07, + "loss": 1.0431, + "mean_token_accuracy": 0.7206839919090271, + "num_tokens": 23150367.0, + "step": 610 + }, + { + "epoch": 0.07772548021880168, + "grad_norm": 23.25, + "learning_rate": 2.585841458245019e-07, + "loss": 1.1039, + "mean_token_accuracy": 0.7029403448104858, + "num_tokens": 23185836.0, + "step": 611 + }, + { + "epoch": 0.07785269049739219, + "grad_norm": 21.125, + "learning_rate": 2.5900805426027975e-07, + "loss": 1.0199, + "mean_token_accuracy": 0.7240686416625977, + "num_tokens": 23219346.0, + "step": 612 + }, + { + "epoch": 0.0779799007759827, + "grad_norm": 24.125, + "learning_rate": 2.5943196269605765e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.7001389265060425, + "num_tokens": 23257569.0, + "step": 613 + }, + { + "epoch": 0.07810711105457321, + "grad_norm": 19.5, + "learning_rate": 2.598558711318355e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.7139310836791992, + "num_tokens": 23302944.0, + "step": 614 + }, + { + "epoch": 0.07823432133316371, + "grad_norm": 23.375, + "learning_rate": 2.602797795676134e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.7078394889831543, + "num_tokens": 23341660.0, + "step": 615 + }, + { + "epoch": 0.07836153161175423, + "grad_norm": 23.375, + "learning_rate": 2.6070368800339124e-07, + "loss": 1.1207, + "mean_token_accuracy": 0.6960120797157288, + "num_tokens": 23375956.0, + "step": 616 + }, + { + "epoch": 0.07848874189034474, + "grad_norm": 23.125, + "learning_rate": 2.6112759643916914e-07, + "loss": 1.0606, + "mean_token_accuracy": 0.7144505381584167, + "num_tokens": 23407568.0, + "step": 617 + }, + { + "epoch": 0.07861595216893524, + "grad_norm": 24.0, + "learning_rate": 2.61551504874947e-07, + "loss": 1.1388, + "mean_token_accuracy": 0.6955732107162476, + "num_tokens": 23440261.0, + "step": 618 + }, + { + "epoch": 0.07874316244752576, + "grad_norm": 20.5, + "learning_rate": 2.619754133107249e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.7127429246902466, + "num_tokens": 23481706.0, + "step": 619 + }, + { + "epoch": 0.07887037272611627, + "grad_norm": 22.25, + "learning_rate": 2.623993217465028e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6910877823829651, + "num_tokens": 23515410.0, + "step": 620 + }, + { + "epoch": 0.07899758300470679, + "grad_norm": 22.125, + "learning_rate": 2.6282323018228063e-07, + "loss": 1.1215, + "mean_token_accuracy": 0.7011175155639648, + "num_tokens": 23554627.0, + "step": 621 + }, + { + "epoch": 0.07912479328329729, + "grad_norm": 23.375, + "learning_rate": 2.632471386180585e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6930450797080994, + "num_tokens": 23593400.0, + "step": 622 + }, + { + "epoch": 0.0792520035618878, + "grad_norm": 24.25, + "learning_rate": 2.6367104705383637e-07, + "loss": 1.1484, + "mean_token_accuracy": 0.6883034706115723, + "num_tokens": 23628242.0, + "step": 623 + }, + { + "epoch": 0.07937921384047832, + "grad_norm": 21.0, + "learning_rate": 2.6409495548961427e-07, + "loss": 1.1037, + "mean_token_accuracy": 0.7050097584724426, + "num_tokens": 23671041.0, + "step": 624 + }, + { + "epoch": 0.07950642411906882, + "grad_norm": 22.625, + "learning_rate": 2.6451886392539206e-07, + "loss": 1.1342, + "mean_token_accuracy": 0.6968657970428467, + "num_tokens": 23701904.0, + "step": 625 + }, + { + "epoch": 0.07963363439765933, + "grad_norm": 21.5, + "learning_rate": 2.6494277236116996e-07, + "loss": 1.044, + "mean_token_accuracy": 0.7130603790283203, + "num_tokens": 23742374.0, + "step": 626 + }, + { + "epoch": 0.07976084467624985, + "grad_norm": 25.25, + "learning_rate": 2.6536668079694786e-07, + "loss": 1.1945, + "mean_token_accuracy": 0.6782009601593018, + "num_tokens": 23780216.0, + "step": 627 + }, + { + "epoch": 0.07988805495484035, + "grad_norm": 19.75, + "learning_rate": 2.6579058923272576e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.7360508441925049, + "num_tokens": 23813641.0, + "step": 628 + }, + { + "epoch": 0.08001526523343086, + "grad_norm": 24.0, + "learning_rate": 2.6621449766850356e-07, + "loss": 1.108, + "mean_token_accuracy": 0.7016698718070984, + "num_tokens": 23850189.0, + "step": 629 + }, + { + "epoch": 0.08014247551202137, + "grad_norm": 20.875, + "learning_rate": 2.6663840610428145e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.7198824882507324, + "num_tokens": 23884410.0, + "step": 630 + }, + { + "epoch": 0.08026968579061187, + "grad_norm": 21.0, + "learning_rate": 2.6706231454005935e-07, + "loss": 1.0249, + "mean_token_accuracy": 0.719190239906311, + "num_tokens": 23930139.0, + "step": 631 + }, + { + "epoch": 0.08039689606920239, + "grad_norm": 19.625, + "learning_rate": 2.6748622297583725e-07, + "loss": 0.998, + "mean_token_accuracy": 0.7277332544326782, + "num_tokens": 23973124.0, + "step": 632 + }, + { + "epoch": 0.0805241063477929, + "grad_norm": 22.375, + "learning_rate": 2.6791013141161505e-07, + "loss": 1.0477, + "mean_token_accuracy": 0.7165235877037048, + "num_tokens": 24008676.0, + "step": 633 + }, + { + "epoch": 0.08065131662638342, + "grad_norm": 21.0, + "learning_rate": 2.6833403984739294e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.7089120745658875, + "num_tokens": 24044880.0, + "step": 634 + }, + { + "epoch": 0.08077852690497392, + "grad_norm": 19.375, + "learning_rate": 2.6875794828317084e-07, + "loss": 1.0091, + "mean_token_accuracy": 0.7268365621566772, + "num_tokens": 24090140.0, + "step": 635 + }, + { + "epoch": 0.08090573718356443, + "grad_norm": 24.125, + "learning_rate": 2.6918185671894874e-07, + "loss": 1.1931, + "mean_token_accuracy": 0.6850985288619995, + "num_tokens": 24123311.0, + "step": 636 + }, + { + "epoch": 0.08103294746215495, + "grad_norm": 24.375, + "learning_rate": 2.6960576515472654e-07, + "loss": 1.2018, + "mean_token_accuracy": 0.6691579818725586, + "num_tokens": 24164019.0, + "step": 637 + }, + { + "epoch": 0.08116015774074545, + "grad_norm": 19.0, + "learning_rate": 2.7002967359050443e-07, + "loss": 1.0019, + "mean_token_accuracy": 0.7281860113143921, + "num_tokens": 24205986.0, + "step": 638 + }, + { + "epoch": 0.08128736801933596, + "grad_norm": 21.0, + "learning_rate": 2.7045358202628233e-07, + "loss": 1.0786, + "mean_token_accuracy": 0.696865439414978, + "num_tokens": 24251956.0, + "step": 639 + }, + { + "epoch": 0.08141457829792648, + "grad_norm": 21.25, + "learning_rate": 2.7087749046206023e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.7141708135604858, + "num_tokens": 24284990.0, + "step": 640 + }, + { + "epoch": 0.08154178857651698, + "grad_norm": 20.875, + "learning_rate": 2.71301398897838e-07, + "loss": 1.0726, + "mean_token_accuracy": 0.7054905891418457, + "num_tokens": 24323305.0, + "step": 641 + }, + { + "epoch": 0.08166899885510749, + "grad_norm": 17.75, + "learning_rate": 2.717253073336159e-07, + "loss": 0.9392, + "mean_token_accuracy": 0.7429755926132202, + "num_tokens": 24365847.0, + "step": 642 + }, + { + "epoch": 0.081796209133698, + "grad_norm": 19.375, + "learning_rate": 2.721492157693938e-07, + "loss": 0.9697, + "mean_token_accuracy": 0.7330651879310608, + "num_tokens": 24407269.0, + "step": 643 + }, + { + "epoch": 0.0819234194122885, + "grad_norm": 20.125, + "learning_rate": 2.7257312420517167e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.7168303728103638, + "num_tokens": 24444990.0, + "step": 644 + }, + { + "epoch": 0.08205062969087902, + "grad_norm": 19.875, + "learning_rate": 2.729970326409495e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.7081500291824341, + "num_tokens": 24487293.0, + "step": 645 + }, + { + "epoch": 0.08217783996946953, + "grad_norm": 22.25, + "learning_rate": 2.734209410767274e-07, + "loss": 1.0804, + "mean_token_accuracy": 0.7025428414344788, + "num_tokens": 24521403.0, + "step": 646 + }, + { + "epoch": 0.08230505024806005, + "grad_norm": 20.75, + "learning_rate": 2.738448495125053e-07, + "loss": 1.0451, + "mean_token_accuracy": 0.7140856981277466, + "num_tokens": 24562740.0, + "step": 647 + }, + { + "epoch": 0.08243226052665055, + "grad_norm": 21.0, + "learning_rate": 2.7426875794828316e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.7043890953063965, + "num_tokens": 24597507.0, + "step": 648 + }, + { + "epoch": 0.08255947080524106, + "grad_norm": 19.5, + "learning_rate": 2.74692666384061e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.7278171181678772, + "num_tokens": 24635629.0, + "step": 649 + }, + { + "epoch": 0.08268668108383158, + "grad_norm": 19.625, + "learning_rate": 2.751165748198389e-07, + "loss": 1.0277, + "mean_token_accuracy": 0.7130093574523926, + "num_tokens": 24676087.0, + "step": 650 + }, + { + "epoch": 0.08281389136242208, + "grad_norm": 18.75, + "learning_rate": 2.755404832556168e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.7133986949920654, + "num_tokens": 24713652.0, + "step": 651 + }, + { + "epoch": 0.0829411016410126, + "grad_norm": 20.25, + "learning_rate": 2.7596439169139465e-07, + "loss": 1.0621, + "mean_token_accuracy": 0.7144383192062378, + "num_tokens": 24750130.0, + "step": 652 + }, + { + "epoch": 0.08306831191960311, + "grad_norm": 22.875, + "learning_rate": 2.763883001271725e-07, + "loss": 1.1282, + "mean_token_accuracy": 0.6923073530197144, + "num_tokens": 24787337.0, + "step": 653 + }, + { + "epoch": 0.08319552219819361, + "grad_norm": 20.5, + "learning_rate": 2.768122085629504e-07, + "loss": 1.1201, + "mean_token_accuracy": 0.6939257383346558, + "num_tokens": 24822736.0, + "step": 654 + }, + { + "epoch": 0.08332273247678412, + "grad_norm": 20.125, + "learning_rate": 2.772361169987283e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.7173677682876587, + "num_tokens": 24858743.0, + "step": 655 + }, + { + "epoch": 0.08344994275537464, + "grad_norm": 19.25, + "learning_rate": 2.7766002543450614e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.7151784896850586, + "num_tokens": 24901286.0, + "step": 656 + }, + { + "epoch": 0.08357715303396514, + "grad_norm": 20.25, + "learning_rate": 2.78083933870284e-07, + "loss": 1.1165, + "mean_token_accuracy": 0.6937843561172485, + "num_tokens": 24943457.0, + "step": 657 + }, + { + "epoch": 0.08370436331255565, + "grad_norm": 19.5, + "learning_rate": 2.785078423060619e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.706733763217926, + "num_tokens": 24979247.0, + "step": 658 + }, + { + "epoch": 0.08383157359114617, + "grad_norm": 20.25, + "learning_rate": 2.789317507418398e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.7077597379684448, + "num_tokens": 25017456.0, + "step": 659 + }, + { + "epoch": 0.08395878386973668, + "grad_norm": 19.375, + "learning_rate": 2.7935565917761763e-07, + "loss": 0.9897, + "mean_token_accuracy": 0.7248193025588989, + "num_tokens": 25054682.0, + "step": 660 + }, + { + "epoch": 0.08408599414832718, + "grad_norm": 22.25, + "learning_rate": 2.797795676133955e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6984315514564514, + "num_tokens": 25092501.0, + "step": 661 + }, + { + "epoch": 0.0842132044269177, + "grad_norm": 15.1875, + "learning_rate": 2.802034760491734e-07, + "loss": 0.8673, + "mean_token_accuracy": 0.7615010142326355, + "num_tokens": 25129253.0, + "step": 662 + }, + { + "epoch": 0.08434041470550821, + "grad_norm": 17.625, + "learning_rate": 2.806273844849512e-07, + "loss": 0.9503, + "mean_token_accuracy": 0.7385420799255371, + "num_tokens": 25169119.0, + "step": 663 + }, + { + "epoch": 0.08446762498409871, + "grad_norm": 19.5, + "learning_rate": 2.810512929207291e-07, + "loss": 1.057, + "mean_token_accuracy": 0.7035218477249146, + "num_tokens": 25207654.0, + "step": 664 + }, + { + "epoch": 0.08459483526268922, + "grad_norm": 17.5, + "learning_rate": 2.8147520135650697e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.7294244766235352, + "num_tokens": 25243258.0, + "step": 665 + }, + { + "epoch": 0.08472204554127974, + "grad_norm": 19.5, + "learning_rate": 2.8189910979228487e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.7157796621322632, + "num_tokens": 25281301.0, + "step": 666 + }, + { + "epoch": 0.08484925581987024, + "grad_norm": 19.875, + "learning_rate": 2.823230182280627e-07, + "loss": 0.9903, + "mean_token_accuracy": 0.7379879951477051, + "num_tokens": 25313500.0, + "step": 667 + }, + { + "epoch": 0.08497646609846075, + "grad_norm": 18.375, + "learning_rate": 2.827469266638406e-07, + "loss": 1.007, + "mean_token_accuracy": 0.7222340106964111, + "num_tokens": 25355121.0, + "step": 668 + }, + { + "epoch": 0.08510367637705127, + "grad_norm": 18.875, + "learning_rate": 2.8317083509961846e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.7221139073371887, + "num_tokens": 25398868.0, + "step": 669 + }, + { + "epoch": 0.08523088665564178, + "grad_norm": 18.375, + "learning_rate": 2.8359474353539636e-07, + "loss": 0.9963, + "mean_token_accuracy": 0.7236101627349854, + "num_tokens": 25438670.0, + "step": 670 + }, + { + "epoch": 0.08535809693423228, + "grad_norm": 18.625, + "learning_rate": 2.840186519711742e-07, + "loss": 1.0559, + "mean_token_accuracy": 0.7095853686332703, + "num_tokens": 25479243.0, + "step": 671 + }, + { + "epoch": 0.0854853072128228, + "grad_norm": 20.125, + "learning_rate": 2.844425604069521e-07, + "loss": 1.058, + "mean_token_accuracy": 0.7066758275032043, + "num_tokens": 25514615.0, + "step": 672 + }, + { + "epoch": 0.08561251749141331, + "grad_norm": 19.75, + "learning_rate": 2.8486646884272995e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.7166709899902344, + "num_tokens": 25552753.0, + "step": 673 + }, + { + "epoch": 0.08573972777000381, + "grad_norm": 18.0, + "learning_rate": 2.8529037727850785e-07, + "loss": 0.9327, + "mean_token_accuracy": 0.7393825650215149, + "num_tokens": 25587415.0, + "step": 674 + }, + { + "epoch": 0.08586693804859433, + "grad_norm": 18.125, + "learning_rate": 2.857142857142857e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7321558594703674, + "num_tokens": 25625962.0, + "step": 675 + }, + { + "epoch": 0.08599414832718484, + "grad_norm": 20.0, + "learning_rate": 2.861381941500636e-07, + "loss": 1.056, + "mean_token_accuracy": 0.7076576352119446, + "num_tokens": 25669680.0, + "step": 676 + }, + { + "epoch": 0.08612135860577534, + "grad_norm": 20.0, + "learning_rate": 2.8656210258584144e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6949584484100342, + "num_tokens": 25709221.0, + "step": 677 + }, + { + "epoch": 0.08624856888436586, + "grad_norm": 21.5, + "learning_rate": 2.869860110216193e-07, + "loss": 1.1348, + "mean_token_accuracy": 0.6892749667167664, + "num_tokens": 25741880.0, + "step": 678 + }, + { + "epoch": 0.08637577916295637, + "grad_norm": 18.25, + "learning_rate": 2.874099194573972e-07, + "loss": 0.991, + "mean_token_accuracy": 0.7265719175338745, + "num_tokens": 25786282.0, + "step": 679 + }, + { + "epoch": 0.08650298944154687, + "grad_norm": 21.0, + "learning_rate": 2.878338278931751e-07, + "loss": 1.1039, + "mean_token_accuracy": 0.7057605981826782, + "num_tokens": 25820229.0, + "step": 680 + }, + { + "epoch": 0.08663019972013739, + "grad_norm": 17.625, + "learning_rate": 2.8825773632895293e-07, + "loss": 0.9745, + "mean_token_accuracy": 0.7317899465560913, + "num_tokens": 25856465.0, + "step": 681 + }, + { + "epoch": 0.0867574099987279, + "grad_norm": 16.875, + "learning_rate": 2.886816447647308e-07, + "loss": 0.9565, + "mean_token_accuracy": 0.7362920045852661, + "num_tokens": 25897407.0, + "step": 682 + }, + { + "epoch": 0.08688462027731841, + "grad_norm": 17.75, + "learning_rate": 2.891055532005087e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.7272186279296875, + "num_tokens": 25938825.0, + "step": 683 + }, + { + "epoch": 0.08701183055590891, + "grad_norm": 18.0, + "learning_rate": 2.8952946163628657e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.7116289734840393, + "num_tokens": 25974944.0, + "step": 684 + }, + { + "epoch": 0.08713904083449943, + "grad_norm": 18.125, + "learning_rate": 2.899533700720644e-07, + "loss": 0.999, + "mean_token_accuracy": 0.7228718996047974, + "num_tokens": 26012864.0, + "step": 685 + }, + { + "epoch": 0.08726625111308994, + "grad_norm": 20.0, + "learning_rate": 2.9037727850784227e-07, + "loss": 1.0769, + "mean_token_accuracy": 0.7049542665481567, + "num_tokens": 26049779.0, + "step": 686 + }, + { + "epoch": 0.08739346139168044, + "grad_norm": 16.75, + "learning_rate": 2.9080118694362016e-07, + "loss": 1.0, + "mean_token_accuracy": 0.7252973318099976, + "num_tokens": 26089095.0, + "step": 687 + }, + { + "epoch": 0.08752067167027096, + "grad_norm": 17.75, + "learning_rate": 2.9122509537939806e-07, + "loss": 0.9771, + "mean_token_accuracy": 0.7279123067855835, + "num_tokens": 26132553.0, + "step": 688 + }, + { + "epoch": 0.08764788194886147, + "grad_norm": 19.125, + "learning_rate": 2.916490038151759e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.7015843391418457, + "num_tokens": 26171301.0, + "step": 689 + }, + { + "epoch": 0.08777509222745197, + "grad_norm": 18.25, + "learning_rate": 2.9207291225095376e-07, + "loss": 1.0409, + "mean_token_accuracy": 0.7114174365997314, + "num_tokens": 26217871.0, + "step": 690 + }, + { + "epoch": 0.08790230250604249, + "grad_norm": 18.0, + "learning_rate": 2.9249682068673166e-07, + "loss": 1.0166, + "mean_token_accuracy": 0.7191122770309448, + "num_tokens": 26255927.0, + "step": 691 + }, + { + "epoch": 0.088029512784633, + "grad_norm": 17.25, + "learning_rate": 2.9292072912250955e-07, + "loss": 0.9834, + "mean_token_accuracy": 0.7335360050201416, + "num_tokens": 26296041.0, + "step": 692 + }, + { + "epoch": 0.0881567230632235, + "grad_norm": 18.125, + "learning_rate": 2.933446375582874e-07, + "loss": 1.0939, + "mean_token_accuracy": 0.6993444561958313, + "num_tokens": 26333785.0, + "step": 693 + }, + { + "epoch": 0.08828393334181402, + "grad_norm": 16.875, + "learning_rate": 2.9376854599406525e-07, + "loss": 0.9708, + "mean_token_accuracy": 0.7300194501876831, + "num_tokens": 26368837.0, + "step": 694 + }, + { + "epoch": 0.08841114362040453, + "grad_norm": 19.875, + "learning_rate": 2.9419245442984315e-07, + "loss": 1.0993, + "mean_token_accuracy": 0.6984378099441528, + "num_tokens": 26402151.0, + "step": 695 + }, + { + "epoch": 0.08853835389899505, + "grad_norm": 20.25, + "learning_rate": 2.9461636286562104e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.7159765958786011, + "num_tokens": 26435550.0, + "step": 696 + }, + { + "epoch": 0.08866556417758555, + "grad_norm": 16.625, + "learning_rate": 2.9504027130139884e-07, + "loss": 0.9617, + "mean_token_accuracy": 0.7368009090423584, + "num_tokens": 26479299.0, + "step": 697 + }, + { + "epoch": 0.08879277445617606, + "grad_norm": 16.875, + "learning_rate": 2.9546417973717674e-07, + "loss": 1.0234, + "mean_token_accuracy": 0.7159767150878906, + "num_tokens": 26519380.0, + "step": 698 + }, + { + "epoch": 0.08891998473476657, + "grad_norm": 16.5, + "learning_rate": 2.9588808817295464e-07, + "loss": 0.9342, + "mean_token_accuracy": 0.7382280826568604, + "num_tokens": 26557306.0, + "step": 699 + }, + { + "epoch": 0.08904719501335707, + "grad_norm": 19.125, + "learning_rate": 2.9631199660873253e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.7099152207374573, + "num_tokens": 26593549.0, + "step": 700 + }, + { + "epoch": 0.08917440529194759, + "grad_norm": 16.375, + "learning_rate": 2.9673590504451033e-07, + "loss": 0.935, + "mean_token_accuracy": 0.7434406280517578, + "num_tokens": 26629926.0, + "step": 701 + }, + { + "epoch": 0.0893016155705381, + "grad_norm": 16.625, + "learning_rate": 2.9715981348028823e-07, + "loss": 0.9715, + "mean_token_accuracy": 0.7353770136833191, + "num_tokens": 26668652.0, + "step": 702 + }, + { + "epoch": 0.0894288258491286, + "grad_norm": 17.375, + "learning_rate": 2.975837219160661e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.7122847437858582, + "num_tokens": 26708084.0, + "step": 703 + }, + { + "epoch": 0.08955603612771912, + "grad_norm": 17.375, + "learning_rate": 2.98007630351844e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.7198034524917603, + "num_tokens": 26750134.0, + "step": 704 + }, + { + "epoch": 0.08968324640630963, + "grad_norm": 18.25, + "learning_rate": 2.984315387876218e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.7158739566802979, + "num_tokens": 26785410.0, + "step": 705 + }, + { + "epoch": 0.08981045668490013, + "grad_norm": 17.5, + "learning_rate": 2.988554472233997e-07, + "loss": 1.0219, + "mean_token_accuracy": 0.7212008833885193, + "num_tokens": 26825226.0, + "step": 706 + }, + { + "epoch": 0.08993766696349065, + "grad_norm": 16.75, + "learning_rate": 2.992793556591776e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.7216747403144836, + "num_tokens": 26865401.0, + "step": 707 + }, + { + "epoch": 0.09006487724208116, + "grad_norm": 17.0, + "learning_rate": 2.997032640949555e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.7163669466972351, + "num_tokens": 26905365.0, + "step": 708 + }, + { + "epoch": 0.09019208752067168, + "grad_norm": 16.75, + "learning_rate": 3.001271725307333e-07, + "loss": 0.9545, + "mean_token_accuracy": 0.7366237044334412, + "num_tokens": 26941348.0, + "step": 709 + }, + { + "epoch": 0.09031929779926218, + "grad_norm": 18.625, + "learning_rate": 3.005510809665112e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.7196816205978394, + "num_tokens": 26975803.0, + "step": 710 + }, + { + "epoch": 0.09044650807785269, + "grad_norm": 15.5, + "learning_rate": 3.009749894022891e-07, + "loss": 0.9185, + "mean_token_accuracy": 0.7443243265151978, + "num_tokens": 27014931.0, + "step": 711 + }, + { + "epoch": 0.0905737183564432, + "grad_norm": 17.875, + "learning_rate": 3.01398897838067e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.7209705710411072, + "num_tokens": 27055505.0, + "step": 712 + }, + { + "epoch": 0.0907009286350337, + "grad_norm": 15.5625, + "learning_rate": 3.018228062738448e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.7205471992492676, + "num_tokens": 27092121.0, + "step": 713 + }, + { + "epoch": 0.09082813891362422, + "grad_norm": 15.5625, + "learning_rate": 3.022467147096227e-07, + "loss": 0.941, + "mean_token_accuracy": 0.7411490678787231, + "num_tokens": 27129827.0, + "step": 714 + }, + { + "epoch": 0.09095534919221474, + "grad_norm": 15.6875, + "learning_rate": 3.026706231454006e-07, + "loss": 0.9568, + "mean_token_accuracy": 0.733454167842865, + "num_tokens": 27174977.0, + "step": 715 + }, + { + "epoch": 0.09108255947080524, + "grad_norm": 17.625, + "learning_rate": 3.0309453158117844e-07, + "loss": 0.9869, + "mean_token_accuracy": 0.7301174998283386, + "num_tokens": 27211472.0, + "step": 716 + }, + { + "epoch": 0.09120976974939575, + "grad_norm": 18.625, + "learning_rate": 3.035184400169563e-07, + "loss": 1.117, + "mean_token_accuracy": 0.6996034383773804, + "num_tokens": 27249160.0, + "step": 717 + }, + { + "epoch": 0.09133698002798626, + "grad_norm": 15.4375, + "learning_rate": 3.039423484527342e-07, + "loss": 0.9661, + "mean_token_accuracy": 0.7333890199661255, + "num_tokens": 27287946.0, + "step": 718 + }, + { + "epoch": 0.09146419030657676, + "grad_norm": 17.625, + "learning_rate": 3.043662568885121e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.7113898396492004, + "num_tokens": 27323388.0, + "step": 719 + }, + { + "epoch": 0.09159140058516728, + "grad_norm": 19.25, + "learning_rate": 3.0479016532428993e-07, + "loss": 1.1027, + "mean_token_accuracy": 0.6970309019088745, + "num_tokens": 27357524.0, + "step": 720 + }, + { + "epoch": 0.0917186108637578, + "grad_norm": 17.25, + "learning_rate": 3.052140737600678e-07, + "loss": 0.9679, + "mean_token_accuracy": 0.7340580224990845, + "num_tokens": 27392037.0, + "step": 721 + }, + { + "epoch": 0.09184582114234831, + "grad_norm": 15.625, + "learning_rate": 3.056379821958457e-07, + "loss": 0.9425, + "mean_token_accuracy": 0.7403529286384583, + "num_tokens": 27429149.0, + "step": 722 + }, + { + "epoch": 0.09197303142093881, + "grad_norm": 16.875, + "learning_rate": 3.060618906316236e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.7270947694778442, + "num_tokens": 27465023.0, + "step": 723 + }, + { + "epoch": 0.09210024169952932, + "grad_norm": 15.1875, + "learning_rate": 3.064857990674014e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7331424951553345, + "num_tokens": 27501268.0, + "step": 724 + }, + { + "epoch": 0.09222745197811984, + "grad_norm": 17.0, + "learning_rate": 3.0690970750317927e-07, + "loss": 1.0307, + "mean_token_accuracy": 0.7160599231719971, + "num_tokens": 27539937.0, + "step": 725 + }, + { + "epoch": 0.09235466225671034, + "grad_norm": 16.0, + "learning_rate": 3.0733361593895717e-07, + "loss": 0.9932, + "mean_token_accuracy": 0.7300741672515869, + "num_tokens": 27574576.0, + "step": 726 + }, + { + "epoch": 0.09248187253530085, + "grad_norm": 15.5625, + "learning_rate": 3.0775752437473507e-07, + "loss": 0.9607, + "mean_token_accuracy": 0.7314765453338623, + "num_tokens": 27616036.0, + "step": 727 + }, + { + "epoch": 0.09260908281389137, + "grad_norm": 17.0, + "learning_rate": 3.081814328105129e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.7195746898651123, + "num_tokens": 27662040.0, + "step": 728 + }, + { + "epoch": 0.09273629309248187, + "grad_norm": 16.75, + "learning_rate": 3.0860534124629076e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.7086097002029419, + "num_tokens": 27705136.0, + "step": 729 + }, + { + "epoch": 0.09286350337107238, + "grad_norm": 16.375, + "learning_rate": 3.0902924968206866e-07, + "loss": 1.0321, + "mean_token_accuracy": 0.7186294794082642, + "num_tokens": 27740564.0, + "step": 730 + }, + { + "epoch": 0.0929907136496629, + "grad_norm": 16.625, + "learning_rate": 3.0945315811784656e-07, + "loss": 0.9606, + "mean_token_accuracy": 0.7313696146011353, + "num_tokens": 27774788.0, + "step": 731 + }, + { + "epoch": 0.0931179239282534, + "grad_norm": 18.25, + "learning_rate": 3.098770665536244e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6842482089996338, + "num_tokens": 27818300.0, + "step": 732 + }, + { + "epoch": 0.09324513420684391, + "grad_norm": 15.25, + "learning_rate": 3.1030097498940225e-07, + "loss": 0.9338, + "mean_token_accuracy": 0.7420231699943542, + "num_tokens": 27857739.0, + "step": 733 + }, + { + "epoch": 0.09337234448543442, + "grad_norm": 17.0, + "learning_rate": 3.1072488342518015e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.7126227617263794, + "num_tokens": 27893897.0, + "step": 734 + }, + { + "epoch": 0.09349955476402494, + "grad_norm": 16.875, + "learning_rate": 3.11148791860958e-07, + "loss": 0.9911, + "mean_token_accuracy": 0.72606360912323, + "num_tokens": 27930511.0, + "step": 735 + }, + { + "epoch": 0.09362676504261544, + "grad_norm": 16.75, + "learning_rate": 3.115727002967359e-07, + "loss": 1.0303, + "mean_token_accuracy": 0.7151440382003784, + "num_tokens": 27961128.0, + "step": 736 + }, + { + "epoch": 0.09375397532120595, + "grad_norm": 16.125, + "learning_rate": 3.1199660873251374e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.7272268533706665, + "num_tokens": 28000748.0, + "step": 737 + }, + { + "epoch": 0.09388118559979647, + "grad_norm": 15.375, + "learning_rate": 3.1242051716829164e-07, + "loss": 1.0164, + "mean_token_accuracy": 0.7131921052932739, + "num_tokens": 28046455.0, + "step": 738 + }, + { + "epoch": 0.09400839587838697, + "grad_norm": 15.8125, + "learning_rate": 3.128444256040695e-07, + "loss": 0.9862, + "mean_token_accuracy": 0.7294607162475586, + "num_tokens": 28087026.0, + "step": 739 + }, + { + "epoch": 0.09413560615697748, + "grad_norm": 14.375, + "learning_rate": 3.132683340398474e-07, + "loss": 0.8861, + "mean_token_accuracy": 0.7508013248443604, + "num_tokens": 28126382.0, + "step": 740 + }, + { + "epoch": 0.094262816435568, + "grad_norm": 16.75, + "learning_rate": 3.1369224247562523e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7353183627128601, + "num_tokens": 28161198.0, + "step": 741 + }, + { + "epoch": 0.0943900267141585, + "grad_norm": 13.4375, + "learning_rate": 3.1411615091140313e-07, + "loss": 0.8807, + "mean_token_accuracy": 0.7556292414665222, + "num_tokens": 28204612.0, + "step": 742 + }, + { + "epoch": 0.09451723699274901, + "grad_norm": 16.125, + "learning_rate": 3.14540059347181e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.728678822517395, + "num_tokens": 28241050.0, + "step": 743 + }, + { + "epoch": 0.09464444727133953, + "grad_norm": 13.9375, + "learning_rate": 3.149639677829589e-07, + "loss": 0.9129, + "mean_token_accuracy": 0.7479708790779114, + "num_tokens": 28278032.0, + "step": 744 + }, + { + "epoch": 0.09477165754993004, + "grad_norm": 16.125, + "learning_rate": 3.153878762187368e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.7218714952468872, + "num_tokens": 28315599.0, + "step": 745 + }, + { + "epoch": 0.09489886782852054, + "grad_norm": 17.0, + "learning_rate": 3.158117846545146e-07, + "loss": 1.0338, + "mean_token_accuracy": 0.7191723585128784, + "num_tokens": 28358504.0, + "step": 746 + }, + { + "epoch": 0.09502607810711106, + "grad_norm": 15.5, + "learning_rate": 3.1623569309029247e-07, + "loss": 0.9635, + "mean_token_accuracy": 0.7364542484283447, + "num_tokens": 28389626.0, + "step": 747 + }, + { + "epoch": 0.09515328838570157, + "grad_norm": 14.9375, + "learning_rate": 3.1665960152607037e-07, + "loss": 0.9326, + "mean_token_accuracy": 0.7388137578964233, + "num_tokens": 28430361.0, + "step": 748 + }, + { + "epoch": 0.09528049866429207, + "grad_norm": 15.375, + "learning_rate": 3.1708350996184826e-07, + "loss": 0.9784, + "mean_token_accuracy": 0.7288641929626465, + "num_tokens": 28469792.0, + "step": 749 + }, + { + "epoch": 0.09540770894288259, + "grad_norm": 15.375, + "learning_rate": 3.175074183976261e-07, + "loss": 0.9543, + "mean_token_accuracy": 0.7341081500053406, + "num_tokens": 28507569.0, + "step": 750 + }, + { + "epoch": 0.0955349192214731, + "grad_norm": 17.25, + "learning_rate": 3.1793132683340396e-07, + "loss": 1.0084, + "mean_token_accuracy": 0.7256824970245361, + "num_tokens": 28542336.0, + "step": 751 + }, + { + "epoch": 0.0956621295000636, + "grad_norm": 14.3125, + "learning_rate": 3.1835523526918186e-07, + "loss": 0.918, + "mean_token_accuracy": 0.7465086579322815, + "num_tokens": 28578587.0, + "step": 752 + }, + { + "epoch": 0.09578933977865411, + "grad_norm": 17.0, + "learning_rate": 3.1877914370495975e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.7171278595924377, + "num_tokens": 28617505.0, + "step": 753 + }, + { + "epoch": 0.09591655005724463, + "grad_norm": 15.25, + "learning_rate": 3.1920305214073755e-07, + "loss": 1.0264, + "mean_token_accuracy": 0.7191239595413208, + "num_tokens": 28658526.0, + "step": 754 + }, + { + "epoch": 0.09604376033583513, + "grad_norm": 15.25, + "learning_rate": 3.1962696057651545e-07, + "loss": 0.9433, + "mean_token_accuracy": 0.7368427515029907, + "num_tokens": 28698825.0, + "step": 755 + }, + { + "epoch": 0.09617097061442564, + "grad_norm": 16.5, + "learning_rate": 3.2005086901229335e-07, + "loss": 1.026, + "mean_token_accuracy": 0.7193849086761475, + "num_tokens": 28733238.0, + "step": 756 + }, + { + "epoch": 0.09629818089301616, + "grad_norm": 16.375, + "learning_rate": 3.2047477744807125e-07, + "loss": 1.0219, + "mean_token_accuracy": 0.7232450842857361, + "num_tokens": 28771552.0, + "step": 757 + }, + { + "epoch": 0.09642539117160667, + "grad_norm": 17.25, + "learning_rate": 3.2089868588384904e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.7108624577522278, + "num_tokens": 28810827.0, + "step": 758 + }, + { + "epoch": 0.09655260145019717, + "grad_norm": 14.25, + "learning_rate": 3.2132259431962694e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.7490453124046326, + "num_tokens": 28849017.0, + "step": 759 + }, + { + "epoch": 0.09667981172878769, + "grad_norm": 15.125, + "learning_rate": 3.2174650275540484e-07, + "loss": 0.9846, + "mean_token_accuracy": 0.7327885031700134, + "num_tokens": 28887659.0, + "step": 760 + }, + { + "epoch": 0.0968070220073782, + "grad_norm": 14.625, + "learning_rate": 3.2217041119118274e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7350621819496155, + "num_tokens": 28928983.0, + "step": 761 + }, + { + "epoch": 0.0969342322859687, + "grad_norm": 15.875, + "learning_rate": 3.2259431962696053e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.723263680934906, + "num_tokens": 28962195.0, + "step": 762 + }, + { + "epoch": 0.09706144256455922, + "grad_norm": 15.5, + "learning_rate": 3.2301822806273843e-07, + "loss": 0.9754, + "mean_token_accuracy": 0.7262634038925171, + "num_tokens": 28997351.0, + "step": 763 + }, + { + "epoch": 0.09718865284314973, + "grad_norm": 15.3125, + "learning_rate": 3.2344213649851633e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.7234402894973755, + "num_tokens": 29040988.0, + "step": 764 + }, + { + "epoch": 0.09731586312174023, + "grad_norm": 13.1875, + "learning_rate": 3.238660449342942e-07, + "loss": 0.9395, + "mean_token_accuracy": 0.7367449998855591, + "num_tokens": 29081873.0, + "step": 765 + }, + { + "epoch": 0.09744307340033075, + "grad_norm": 14.5, + "learning_rate": 3.24289953370072e-07, + "loss": 0.8965, + "mean_token_accuracy": 0.7424623966217041, + "num_tokens": 29115360.0, + "step": 766 + }, + { + "epoch": 0.09757028367892126, + "grad_norm": 13.9375, + "learning_rate": 3.247138618058499e-07, + "loss": 0.9407, + "mean_token_accuracy": 0.7298731803894043, + "num_tokens": 29160139.0, + "step": 767 + }, + { + "epoch": 0.09769749395751176, + "grad_norm": 16.25, + "learning_rate": 3.251377702416278e-07, + "loss": 0.979, + "mean_token_accuracy": 0.7268524169921875, + "num_tokens": 29202872.0, + "step": 768 + }, + { + "epoch": 0.09782470423610228, + "grad_norm": 13.9375, + "learning_rate": 3.255616786774057e-07, + "loss": 0.9125, + "mean_token_accuracy": 0.7471508979797363, + "num_tokens": 29246965.0, + "step": 769 + }, + { + "epoch": 0.09795191451469279, + "grad_norm": 12.3125, + "learning_rate": 3.259855871131835e-07, + "loss": 0.8351, + "mean_token_accuracy": 0.7655847072601318, + "num_tokens": 29289531.0, + "step": 770 + }, + { + "epoch": 0.0980791247932833, + "grad_norm": 14.0625, + "learning_rate": 3.264094955489614e-07, + "loss": 0.9111, + "mean_token_accuracy": 0.7435864210128784, + "num_tokens": 29330484.0, + "step": 771 + }, + { + "epoch": 0.0982063350718738, + "grad_norm": 13.8125, + "learning_rate": 3.268334039847393e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.735177755355835, + "num_tokens": 29369595.0, + "step": 772 + }, + { + "epoch": 0.09833354535046432, + "grad_norm": 12.9375, + "learning_rate": 3.2725731242051715e-07, + "loss": 0.9199, + "mean_token_accuracy": 0.7475470304489136, + "num_tokens": 29411856.0, + "step": 773 + }, + { + "epoch": 0.09846075562905483, + "grad_norm": 13.9375, + "learning_rate": 3.27681220856295e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.7479563355445862, + "num_tokens": 29454754.0, + "step": 774 + }, + { + "epoch": 0.09858796590764533, + "grad_norm": 14.625, + "learning_rate": 3.281051292920729e-07, + "loss": 0.9472, + "mean_token_accuracy": 0.7373055219650269, + "num_tokens": 29489131.0, + "step": 775 + }, + { + "epoch": 0.09871517618623585, + "grad_norm": 15.0, + "learning_rate": 3.285290377278508e-07, + "loss": 0.9966, + "mean_token_accuracy": 0.7204103469848633, + "num_tokens": 29528292.0, + "step": 776 + }, + { + "epoch": 0.09884238646482636, + "grad_norm": 13.4375, + "learning_rate": 3.2895294616362864e-07, + "loss": 0.9012, + "mean_token_accuracy": 0.7488977313041687, + "num_tokens": 29564878.0, + "step": 777 + }, + { + "epoch": 0.09896959674341686, + "grad_norm": 14.0625, + "learning_rate": 3.293768545994065e-07, + "loss": 0.9384, + "mean_token_accuracy": 0.7324604988098145, + "num_tokens": 29602826.0, + "step": 778 + }, + { + "epoch": 0.09909680702200738, + "grad_norm": 16.125, + "learning_rate": 3.298007630351844e-07, + "loss": 1.0173, + "mean_token_accuracy": 0.7183084487915039, + "num_tokens": 29640323.0, + "step": 779 + }, + { + "epoch": 0.09922401730059789, + "grad_norm": 12.75, + "learning_rate": 3.302246714709623e-07, + "loss": 0.896, + "mean_token_accuracy": 0.7519168853759766, + "num_tokens": 29683302.0, + "step": 780 + }, + { + "epoch": 0.09935122757918839, + "grad_norm": 14.1875, + "learning_rate": 3.3064857990674013e-07, + "loss": 0.9549, + "mean_token_accuracy": 0.7405475378036499, + "num_tokens": 29721085.0, + "step": 781 + }, + { + "epoch": 0.0994784378577789, + "grad_norm": 14.25, + "learning_rate": 3.31072488342518e-07, + "loss": 0.8989, + "mean_token_accuracy": 0.7504743337631226, + "num_tokens": 29753810.0, + "step": 782 + }, + { + "epoch": 0.09960564813636942, + "grad_norm": 13.75, + "learning_rate": 3.314963967782959e-07, + "loss": 0.9311, + "mean_token_accuracy": 0.7417940497398376, + "num_tokens": 29796488.0, + "step": 783 + }, + { + "epoch": 0.09973285841495994, + "grad_norm": 14.875, + "learning_rate": 3.319203052140738e-07, + "loss": 0.9176, + "mean_token_accuracy": 0.7463815212249756, + "num_tokens": 29838181.0, + "step": 784 + }, + { + "epoch": 0.09986006869355044, + "grad_norm": 14.3125, + "learning_rate": 3.323442136498516e-07, + "loss": 0.9218, + "mean_token_accuracy": 0.7481790781021118, + "num_tokens": 29868893.0, + "step": 785 + }, + { + "epoch": 0.09998727897214095, + "grad_norm": 17.75, + "learning_rate": 3.3276812208562947e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.7304205894470215, + "num_tokens": 29903571.0, + "step": 786 + }, + { + "epoch": 0.10011448925073146, + "grad_norm": 14.8125, + "learning_rate": 3.3319203052140737e-07, + "loss": 1.0259, + "mean_token_accuracy": 0.7242917418479919, + "num_tokens": 29940438.0, + "step": 787 + }, + { + "epoch": 0.10024169952932196, + "grad_norm": 13.3125, + "learning_rate": 3.336159389571852e-07, + "loss": 0.9093, + "mean_token_accuracy": 0.7546390295028687, + "num_tokens": 29976538.0, + "step": 788 + }, + { + "epoch": 0.10036890980791248, + "grad_norm": 15.1875, + "learning_rate": 3.340398473929631e-07, + "loss": 0.9849, + "mean_token_accuracy": 0.7261306643486023, + "num_tokens": 30019814.0, + "step": 789 + }, + { + "epoch": 0.100496120086503, + "grad_norm": 13.75, + "learning_rate": 3.3446375582874096e-07, + "loss": 0.9275, + "mean_token_accuracy": 0.7414413690567017, + "num_tokens": 30057578.0, + "step": 790 + }, + { + "epoch": 0.1006233303650935, + "grad_norm": 15.0625, + "learning_rate": 3.3488766426451886e-07, + "loss": 1.0045, + "mean_token_accuracy": 0.7301353216171265, + "num_tokens": 30094869.0, + "step": 791 + }, + { + "epoch": 0.10075054064368401, + "grad_norm": 13.875, + "learning_rate": 3.353115727002967e-07, + "loss": 0.9609, + "mean_token_accuracy": 0.7279514670372009, + "num_tokens": 30136446.0, + "step": 792 + }, + { + "epoch": 0.10087775092227452, + "grad_norm": 14.5625, + "learning_rate": 3.357354811360746e-07, + "loss": 0.9578, + "mean_token_accuracy": 0.7276961803436279, + "num_tokens": 30179226.0, + "step": 793 + }, + { + "epoch": 0.10100496120086502, + "grad_norm": 16.0, + "learning_rate": 3.3615938957185245e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7282083630561829, + "num_tokens": 30213890.0, + "step": 794 + }, + { + "epoch": 0.10113217147945554, + "grad_norm": 19.75, + "learning_rate": 3.3658329800763035e-07, + "loss": 1.0532, + "mean_token_accuracy": 0.7055599689483643, + "num_tokens": 30245691.0, + "step": 795 + }, + { + "epoch": 0.10125938175804605, + "grad_norm": 20.25, + "learning_rate": 3.370072064434082e-07, + "loss": 0.948, + "mean_token_accuracy": 0.7295295596122742, + "num_tokens": 30281738.0, + "step": 796 + }, + { + "epoch": 0.10138659203663657, + "grad_norm": 18.75, + "learning_rate": 3.374311148791861e-07, + "loss": 0.8997, + "mean_token_accuracy": 0.7482744455337524, + "num_tokens": 30320707.0, + "step": 797 + }, + { + "epoch": 0.10151380231522707, + "grad_norm": 19.125, + "learning_rate": 3.3785502331496394e-07, + "loss": 0.978, + "mean_token_accuracy": 0.7287864685058594, + "num_tokens": 30351379.0, + "step": 798 + }, + { + "epoch": 0.10164101259381758, + "grad_norm": 15.5, + "learning_rate": 3.3827893175074184e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.7214280366897583, + "num_tokens": 30382395.0, + "step": 799 + }, + { + "epoch": 0.1017682228724081, + "grad_norm": 17.375, + "learning_rate": 3.387028401865197e-07, + "loss": 0.9563, + "mean_token_accuracy": 0.7441190481185913, + "num_tokens": 30419758.0, + "step": 800 + }, + { + "epoch": 0.1018954331509986, + "grad_norm": 16.375, + "learning_rate": 3.391267486222976e-07, + "loss": 0.9256, + "mean_token_accuracy": 0.7401406168937683, + "num_tokens": 30458115.0, + "step": 801 + }, + { + "epoch": 0.10202264342958911, + "grad_norm": 19.625, + "learning_rate": 3.3955065705807543e-07, + "loss": 1.0187, + "mean_token_accuracy": 0.7149002552032471, + "num_tokens": 30489679.0, + "step": 802 + }, + { + "epoch": 0.10214985370817962, + "grad_norm": 17.625, + "learning_rate": 3.3997456549385333e-07, + "loss": 0.9114, + "mean_token_accuracy": 0.7498874664306641, + "num_tokens": 30524874.0, + "step": 803 + }, + { + "epoch": 0.10227706398677013, + "grad_norm": 15.8125, + "learning_rate": 3.403984739296312e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.7175202369689941, + "num_tokens": 30558354.0, + "step": 804 + }, + { + "epoch": 0.10240427426536064, + "grad_norm": 27.125, + "learning_rate": 3.408223823654091e-07, + "loss": 0.9265, + "mean_token_accuracy": 0.7450395226478577, + "num_tokens": 30594827.0, + "step": 805 + }, + { + "epoch": 0.10253148454395115, + "grad_norm": 21.75, + "learning_rate": 3.412462908011869e-07, + "loss": 0.9253, + "mean_token_accuracy": 0.7472800016403198, + "num_tokens": 30635259.0, + "step": 806 + }, + { + "epoch": 0.10265869482254165, + "grad_norm": 24.0, + "learning_rate": 3.4167019923696477e-07, + "loss": 0.8961, + "mean_token_accuracy": 0.7529244422912598, + "num_tokens": 30681850.0, + "step": 807 + }, + { + "epoch": 0.10278590510113217, + "grad_norm": 12.8125, + "learning_rate": 3.4209410767274267e-07, + "loss": 0.8877, + "mean_token_accuracy": 0.7565872669219971, + "num_tokens": 30721302.0, + "step": 808 + }, + { + "epoch": 0.10291311537972268, + "grad_norm": 27.375, + "learning_rate": 3.4251801610852057e-07, + "loss": 0.9932, + "mean_token_accuracy": 0.7223988175392151, + "num_tokens": 30761975.0, + "step": 809 + }, + { + "epoch": 0.1030403256583132, + "grad_norm": 15.75, + "learning_rate": 3.429419245442984e-07, + "loss": 0.9661, + "mean_token_accuracy": 0.7320435643196106, + "num_tokens": 30797754.0, + "step": 810 + }, + { + "epoch": 0.1031675359369037, + "grad_norm": 42.25, + "learning_rate": 3.4336583298007626e-07, + "loss": 0.9206, + "mean_token_accuracy": 0.747933030128479, + "num_tokens": 30834183.0, + "step": 811 + }, + { + "epoch": 0.10329474621549421, + "grad_norm": 44.75, + "learning_rate": 3.4378974141585416e-07, + "loss": 0.9115, + "mean_token_accuracy": 0.74497389793396, + "num_tokens": 30874160.0, + "step": 812 + }, + { + "epoch": 0.10342195649408473, + "grad_norm": 43.25, + "learning_rate": 3.4421364985163206e-07, + "loss": 0.8407, + "mean_token_accuracy": 0.7654283046722412, + "num_tokens": 30916827.0, + "step": 813 + }, + { + "epoch": 0.10354916677267523, + "grad_norm": 45.25, + "learning_rate": 3.446375582874099e-07, + "loss": 0.942, + "mean_token_accuracy": 0.7482165694236755, + "num_tokens": 30960063.0, + "step": 814 + }, + { + "epoch": 0.10367637705126574, + "grad_norm": 28.125, + "learning_rate": 3.4506146672318775e-07, + "loss": 0.8873, + "mean_token_accuracy": 0.7544404864311218, + "num_tokens": 30995172.0, + "step": 815 + }, + { + "epoch": 0.10380358732985626, + "grad_norm": 16.875, + "learning_rate": 3.4548537515896565e-07, + "loss": 0.9787, + "mean_token_accuracy": 0.7301729917526245, + "num_tokens": 31035089.0, + "step": 816 + }, + { + "epoch": 0.10393079760844676, + "grad_norm": 22.125, + "learning_rate": 3.4590928359474355e-07, + "loss": 0.9205, + "mean_token_accuracy": 0.7422080039978027, + "num_tokens": 31074836.0, + "step": 817 + }, + { + "epoch": 0.10405800788703727, + "grad_norm": 24.75, + "learning_rate": 3.463331920305214e-07, + "loss": 0.9508, + "mean_token_accuracy": 0.7348617911338806, + "num_tokens": 31114700.0, + "step": 818 + }, + { + "epoch": 0.10418521816562779, + "grad_norm": 13.8125, + "learning_rate": 3.4675710046629924e-07, + "loss": 0.9132, + "mean_token_accuracy": 0.7506334781646729, + "num_tokens": 31148258.0, + "step": 819 + }, + { + "epoch": 0.1043124284442183, + "grad_norm": 60.75, + "learning_rate": 3.4718100890207714e-07, + "loss": 0.9389, + "mean_token_accuracy": 0.7401403188705444, + "num_tokens": 31185720.0, + "step": 820 + }, + { + "epoch": 0.1044396387228088, + "grad_norm": 54.75, + "learning_rate": 3.4760491733785504e-07, + "loss": 0.9696, + "mean_token_accuracy": 0.7299767732620239, + "num_tokens": 31220353.0, + "step": 821 + }, + { + "epoch": 0.10456684900139931, + "grad_norm": 13.5, + "learning_rate": 3.480288257736329e-07, + "loss": 0.9428, + "mean_token_accuracy": 0.7400366067886353, + "num_tokens": 31256477.0, + "step": 822 + }, + { + "epoch": 0.10469405927998983, + "grad_norm": 13.625, + "learning_rate": 3.4845273420941073e-07, + "loss": 0.927, + "mean_token_accuracy": 0.7501624226570129, + "num_tokens": 31294365.0, + "step": 823 + }, + { + "epoch": 0.10482126955858033, + "grad_norm": 40.5, + "learning_rate": 3.4887664264518863e-07, + "loss": 0.9083, + "mean_token_accuracy": 0.7448686957359314, + "num_tokens": 31329357.0, + "step": 824 + }, + { + "epoch": 0.10494847983717084, + "grad_norm": 24.375, + "learning_rate": 3.4930055108096653e-07, + "loss": 0.8868, + "mean_token_accuracy": 0.7566778659820557, + "num_tokens": 31372180.0, + "step": 825 + }, + { + "epoch": 0.10507569011576136, + "grad_norm": 31.625, + "learning_rate": 3.497244595167443e-07, + "loss": 0.8945, + "mean_token_accuracy": 0.7528958320617676, + "num_tokens": 31414866.0, + "step": 826 + }, + { + "epoch": 0.10520290039435186, + "grad_norm": 13.375, + "learning_rate": 3.501483679525222e-07, + "loss": 0.9772, + "mean_token_accuracy": 0.7358534932136536, + "num_tokens": 31450583.0, + "step": 827 + }, + { + "epoch": 0.10533011067294237, + "grad_norm": 14.0625, + "learning_rate": 3.505722763883001e-07, + "loss": 0.9104, + "mean_token_accuracy": 0.7505537271499634, + "num_tokens": 31490371.0, + "step": 828 + }, + { + "epoch": 0.10545732095153289, + "grad_norm": 11.6875, + "learning_rate": 3.50996184824078e-07, + "loss": 0.9565, + "mean_token_accuracy": 0.7372041940689087, + "num_tokens": 31525951.0, + "step": 829 + }, + { + "epoch": 0.10558453123012339, + "grad_norm": 20.875, + "learning_rate": 3.514200932598558e-07, + "loss": 0.9873, + "mean_token_accuracy": 0.7318968176841736, + "num_tokens": 31570109.0, + "step": 830 + }, + { + "epoch": 0.1057117415087139, + "grad_norm": 17.625, + "learning_rate": 3.518440016956337e-07, + "loss": 0.9096, + "mean_token_accuracy": 0.7532470226287842, + "num_tokens": 31610569.0, + "step": 831 + }, + { + "epoch": 0.10583895178730442, + "grad_norm": 13.0625, + "learning_rate": 3.522679101314116e-07, + "loss": 0.9085, + "mean_token_accuracy": 0.7550406455993652, + "num_tokens": 31653585.0, + "step": 832 + }, + { + "epoch": 0.10596616206589493, + "grad_norm": 30.75, + "learning_rate": 3.526918185671895e-07, + "loss": 0.9298, + "mean_token_accuracy": 0.7453006505966187, + "num_tokens": 31696353.0, + "step": 833 + }, + { + "epoch": 0.10609337234448543, + "grad_norm": 13.125, + "learning_rate": 3.531157270029673e-07, + "loss": 0.8923, + "mean_token_accuracy": 0.7587699294090271, + "num_tokens": 31730620.0, + "step": 834 + }, + { + "epoch": 0.10622058262307595, + "grad_norm": 16.5, + "learning_rate": 3.535396354387452e-07, + "loss": 0.8581, + "mean_token_accuracy": 0.7648332118988037, + "num_tokens": 31770538.0, + "step": 835 + }, + { + "epoch": 0.10634779290166646, + "grad_norm": 48.75, + "learning_rate": 3.539635438745231e-07, + "loss": 0.9277, + "mean_token_accuracy": 0.747955858707428, + "num_tokens": 31808687.0, + "step": 836 + }, + { + "epoch": 0.10647500318025696, + "grad_norm": 14.0625, + "learning_rate": 3.54387452310301e-07, + "loss": 0.8978, + "mean_token_accuracy": 0.7564042806625366, + "num_tokens": 31852310.0, + "step": 837 + }, + { + "epoch": 0.10660221345884748, + "grad_norm": 21.125, + "learning_rate": 3.548113607460788e-07, + "loss": 0.8644, + "mean_token_accuracy": 0.7682621479034424, + "num_tokens": 31887897.0, + "step": 838 + }, + { + "epoch": 0.10672942373743799, + "grad_norm": 36.0, + "learning_rate": 3.552352691818567e-07, + "loss": 0.8886, + "mean_token_accuracy": 0.757681131362915, + "num_tokens": 31926159.0, + "step": 839 + }, + { + "epoch": 0.10685663401602849, + "grad_norm": 30.75, + "learning_rate": 3.556591776176346e-07, + "loss": 0.8528, + "mean_token_accuracy": 0.7697405815124512, + "num_tokens": 31968432.0, + "step": 840 + }, + { + "epoch": 0.106983844294619, + "grad_norm": 46.75, + "learning_rate": 3.560830860534125e-07, + "loss": 0.9386, + "mean_token_accuracy": 0.7512098550796509, + "num_tokens": 32000949.0, + "step": 841 + }, + { + "epoch": 0.10711105457320952, + "grad_norm": 64.5, + "learning_rate": 3.565069944891903e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.7319236993789673, + "num_tokens": 32033862.0, + "step": 842 + }, + { + "epoch": 0.10723826485180002, + "grad_norm": 42.0, + "learning_rate": 3.569309029249682e-07, + "loss": 0.921, + "mean_token_accuracy": 0.7527355551719666, + "num_tokens": 32068623.0, + "step": 843 + }, + { + "epoch": 0.10736547513039053, + "grad_norm": 24.875, + "learning_rate": 3.573548113607461e-07, + "loss": 0.9032, + "mean_token_accuracy": 0.7514212131500244, + "num_tokens": 32106658.0, + "step": 844 + }, + { + "epoch": 0.10749268540898105, + "grad_norm": 71.0, + "learning_rate": 3.577787197965239e-07, + "loss": 0.951, + "mean_token_accuracy": 0.7385581731796265, + "num_tokens": 32142031.0, + "step": 845 + }, + { + "epoch": 0.10761989568757156, + "grad_norm": 68.0, + "learning_rate": 3.5820262823230177e-07, + "loss": 0.8217, + "mean_token_accuracy": 0.776104211807251, + "num_tokens": 32182686.0, + "step": 846 + }, + { + "epoch": 0.10774710596616206, + "grad_norm": 91.0, + "learning_rate": 3.5862653666807967e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.7305594086647034, + "num_tokens": 32218027.0, + "step": 847 + }, + { + "epoch": 0.10787431624475258, + "grad_norm": 25.625, + "learning_rate": 3.5905044510385757e-07, + "loss": 0.9019, + "mean_token_accuracy": 0.7540241479873657, + "num_tokens": 32257854.0, + "step": 848 + }, + { + "epoch": 0.10800152652334309, + "grad_norm": 28.375, + "learning_rate": 3.594743535396354e-07, + "loss": 0.8826, + "mean_token_accuracy": 0.7579610347747803, + "num_tokens": 32295019.0, + "step": 849 + }, + { + "epoch": 0.10812873680193359, + "grad_norm": 30.25, + "learning_rate": 3.5989826197541326e-07, + "loss": 0.9445, + "mean_token_accuracy": 0.7412285804748535, + "num_tokens": 32335652.0, + "step": 850 + }, + { + "epoch": 0.1082559470805241, + "grad_norm": 63.25, + "learning_rate": 3.6032217041119116e-07, + "loss": 0.883, + "mean_token_accuracy": 0.753654956817627, + "num_tokens": 32377249.0, + "step": 851 + }, + { + "epoch": 0.10838315735911462, + "grad_norm": 21.5, + "learning_rate": 3.6074607884696906e-07, + "loss": 0.8293, + "mean_token_accuracy": 0.7767292261123657, + "num_tokens": 32422467.0, + "step": 852 + }, + { + "epoch": 0.10851036763770512, + "grad_norm": 22.375, + "learning_rate": 3.611699872827469e-07, + "loss": 0.8519, + "mean_token_accuracy": 0.7659552097320557, + "num_tokens": 32462120.0, + "step": 853 + }, + { + "epoch": 0.10863757791629564, + "grad_norm": 28.625, + "learning_rate": 3.6159389571852475e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.7285845279693604, + "num_tokens": 32498484.0, + "step": 854 + }, + { + "epoch": 0.10876478819488615, + "grad_norm": 31.75, + "learning_rate": 3.6201780415430265e-07, + "loss": 0.8851, + "mean_token_accuracy": 0.7587168216705322, + "num_tokens": 32541536.0, + "step": 855 + }, + { + "epoch": 0.10889199847347665, + "grad_norm": 49.25, + "learning_rate": 3.6244171259008055e-07, + "loss": 0.9444, + "mean_token_accuracy": 0.7487439513206482, + "num_tokens": 32579750.0, + "step": 856 + }, + { + "epoch": 0.10901920875206716, + "grad_norm": 36.75, + "learning_rate": 3.628656210258584e-07, + "loss": 0.9369, + "mean_token_accuracy": 0.7456364631652832, + "num_tokens": 32617518.0, + "step": 857 + }, + { + "epoch": 0.10914641903065768, + "grad_norm": 65.0, + "learning_rate": 3.6328952946163624e-07, + "loss": 0.8824, + "mean_token_accuracy": 0.7626596689224243, + "num_tokens": 32653931.0, + "step": 858 + }, + { + "epoch": 0.1092736293092482, + "grad_norm": 40.75, + "learning_rate": 3.6371343789741414e-07, + "loss": 0.8376, + "mean_token_accuracy": 0.7719643712043762, + "num_tokens": 32693797.0, + "step": 859 + }, + { + "epoch": 0.1094008395878387, + "grad_norm": 127.0, + "learning_rate": 3.6413734633319204e-07, + "loss": 0.9424, + "mean_token_accuracy": 0.7443575263023376, + "num_tokens": 32732432.0, + "step": 860 + }, + { + "epoch": 0.10952804986642921, + "grad_norm": 18.25, + "learning_rate": 3.645612547689699e-07, + "loss": 0.832, + "mean_token_accuracy": 0.7678720951080322, + "num_tokens": 32769225.0, + "step": 861 + }, + { + "epoch": 0.10965526014501972, + "grad_norm": 54.25, + "learning_rate": 3.6498516320474773e-07, + "loss": 0.848, + "mean_token_accuracy": 0.7703560590744019, + "num_tokens": 32812027.0, + "step": 862 + }, + { + "epoch": 0.10978247042361022, + "grad_norm": 27.75, + "learning_rate": 3.6540907164052563e-07, + "loss": 0.8242, + "mean_token_accuracy": 0.7746438384056091, + "num_tokens": 32849334.0, + "step": 863 + }, + { + "epoch": 0.10990968070220074, + "grad_norm": 67.5, + "learning_rate": 3.658329800763035e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7400662899017334, + "num_tokens": 32888422.0, + "step": 864 + }, + { + "epoch": 0.11003689098079125, + "grad_norm": 18.75, + "learning_rate": 3.662568885120814e-07, + "loss": 0.9113, + "mean_token_accuracy": 0.7505604028701782, + "num_tokens": 32922672.0, + "step": 865 + }, + { + "epoch": 0.11016410125938175, + "grad_norm": 15.8125, + "learning_rate": 3.666807969478592e-07, + "loss": 0.9126, + "mean_token_accuracy": 0.7472258806228638, + "num_tokens": 32961694.0, + "step": 866 + }, + { + "epoch": 0.11029131153797227, + "grad_norm": 43.5, + "learning_rate": 3.671047053836371e-07, + "loss": 0.9085, + "mean_token_accuracy": 0.7493160963058472, + "num_tokens": 32998595.0, + "step": 867 + }, + { + "epoch": 0.11041852181656278, + "grad_norm": 55.25, + "learning_rate": 3.6752861381941497e-07, + "loss": 0.8524, + "mean_token_accuracy": 0.7648178339004517, + "num_tokens": 33035884.0, + "step": 868 + }, + { + "epoch": 0.11054573209515328, + "grad_norm": 65.0, + "learning_rate": 3.6795252225519287e-07, + "loss": 0.9359, + "mean_token_accuracy": 0.7466700077056885, + "num_tokens": 33073335.0, + "step": 869 + }, + { + "epoch": 0.1106729423737438, + "grad_norm": 45.0, + "learning_rate": 3.6837643069097077e-07, + "loss": 0.8032, + "mean_token_accuracy": 0.7802207469940186, + "num_tokens": 33111443.0, + "step": 870 + }, + { + "epoch": 0.11080015265233431, + "grad_norm": 71.0, + "learning_rate": 3.688003391267486e-07, + "loss": 0.8676, + "mean_token_accuracy": 0.7659070491790771, + "num_tokens": 33156362.0, + "step": 871 + }, + { + "epoch": 0.11092736293092482, + "grad_norm": 46.75, + "learning_rate": 3.6922424756252646e-07, + "loss": 0.8148, + "mean_token_accuracy": 0.7738468050956726, + "num_tokens": 33204968.0, + "step": 872 + }, + { + "epoch": 0.11105457320951533, + "grad_norm": 37.25, + "learning_rate": 3.6964815599830436e-07, + "loss": 0.9244, + "mean_token_accuracy": 0.7448877692222595, + "num_tokens": 33239265.0, + "step": 873 + }, + { + "epoch": 0.11118178348810584, + "grad_norm": 19.625, + "learning_rate": 3.7007206443408226e-07, + "loss": 0.8136, + "mean_token_accuracy": 0.778551459312439, + "num_tokens": 33275897.0, + "step": 874 + }, + { + "epoch": 0.11130899376669635, + "grad_norm": 20.625, + "learning_rate": 3.704959728698601e-07, + "loss": 0.9785, + "mean_token_accuracy": 0.7310295701026917, + "num_tokens": 33313966.0, + "step": 875 + }, + { + "epoch": 0.11143620404528685, + "grad_norm": 38.0, + "learning_rate": 3.7091988130563795e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7593320608139038, + "num_tokens": 33352909.0, + "step": 876 + }, + { + "epoch": 0.11156341432387737, + "grad_norm": 18.0, + "learning_rate": 3.7134378974141585e-07, + "loss": 0.8468, + "mean_token_accuracy": 0.7681145071983337, + "num_tokens": 33400120.0, + "step": 877 + }, + { + "epoch": 0.11169062460246788, + "grad_norm": 38.0, + "learning_rate": 3.7176769817719375e-07, + "loss": 0.9167, + "mean_token_accuracy": 0.7469961643218994, + "num_tokens": 33436703.0, + "step": 878 + }, + { + "epoch": 0.11181783488105838, + "grad_norm": 13.5, + "learning_rate": 3.7219160661297154e-07, + "loss": 0.8061, + "mean_token_accuracy": 0.7822992205619812, + "num_tokens": 33473515.0, + "step": 879 + }, + { + "epoch": 0.1119450451596489, + "grad_norm": 38.0, + "learning_rate": 3.7261551504874944e-07, + "loss": 0.8973, + "mean_token_accuracy": 0.7583902478218079, + "num_tokens": 33507659.0, + "step": 880 + }, + { + "epoch": 0.11207225543823941, + "grad_norm": 27.0, + "learning_rate": 3.7303942348452734e-07, + "loss": 0.8269, + "mean_token_accuracy": 0.7748005986213684, + "num_tokens": 33542430.0, + "step": 881 + }, + { + "epoch": 0.11219946571682991, + "grad_norm": 70.5, + "learning_rate": 3.7346333192030524e-07, + "loss": 0.8053, + "mean_token_accuracy": 0.781152069568634, + "num_tokens": 33582579.0, + "step": 882 + }, + { + "epoch": 0.11232667599542043, + "grad_norm": 38.75, + "learning_rate": 3.7388724035608303e-07, + "loss": 0.8351, + "mean_token_accuracy": 0.7725771069526672, + "num_tokens": 33625928.0, + "step": 883 + }, + { + "epoch": 0.11245388627401094, + "grad_norm": 30.75, + "learning_rate": 3.7431114879186093e-07, + "loss": 0.9136, + "mean_token_accuracy": 0.749354898929596, + "num_tokens": 33661325.0, + "step": 884 + }, + { + "epoch": 0.11258109655260146, + "grad_norm": 23.125, + "learning_rate": 3.7473505722763883e-07, + "loss": 0.9446, + "mean_token_accuracy": 0.7429124116897583, + "num_tokens": 33696472.0, + "step": 885 + }, + { + "epoch": 0.11270830683119196, + "grad_norm": 39.0, + "learning_rate": 3.7515896566341673e-07, + "loss": 0.8848, + "mean_token_accuracy": 0.7540959119796753, + "num_tokens": 33732637.0, + "step": 886 + }, + { + "epoch": 0.11283551710978247, + "grad_norm": 14.125, + "learning_rate": 3.755828740991945e-07, + "loss": 0.8851, + "mean_token_accuracy": 0.7555031776428223, + "num_tokens": 33775106.0, + "step": 887 + }, + { + "epoch": 0.11296272738837299, + "grad_norm": 27.125, + "learning_rate": 3.760067825349724e-07, + "loss": 0.8472, + "mean_token_accuracy": 0.7699581980705261, + "num_tokens": 33811297.0, + "step": 888 + }, + { + "epoch": 0.11308993766696349, + "grad_norm": 54.75, + "learning_rate": 3.764306909707503e-07, + "loss": 0.7899, + "mean_token_accuracy": 0.7797701954841614, + "num_tokens": 33854038.0, + "step": 889 + }, + { + "epoch": 0.113217147945554, + "grad_norm": 16.5, + "learning_rate": 3.768545994065282e-07, + "loss": 0.895, + "mean_token_accuracy": 0.7563177347183228, + "num_tokens": 33884929.0, + "step": 890 + }, + { + "epoch": 0.11334435822414451, + "grad_norm": 30.25, + "learning_rate": 3.77278507842306e-07, + "loss": 0.8708, + "mean_token_accuracy": 0.763077974319458, + "num_tokens": 33926065.0, + "step": 891 + }, + { + "epoch": 0.11347156850273502, + "grad_norm": 96.0, + "learning_rate": 3.777024162780839e-07, + "loss": 0.8131, + "mean_token_accuracy": 0.7724713683128357, + "num_tokens": 33965557.0, + "step": 892 + }, + { + "epoch": 0.11359877878132553, + "grad_norm": 10.25, + "learning_rate": 3.781263247138618e-07, + "loss": 0.8382, + "mean_token_accuracy": 0.7716077566146851, + "num_tokens": 34004132.0, + "step": 893 + }, + { + "epoch": 0.11372598905991604, + "grad_norm": 10.5625, + "learning_rate": 3.785502331496397e-07, + "loss": 0.8777, + "mean_token_accuracy": 0.7546240091323853, + "num_tokens": 34039781.0, + "step": 894 + }, + { + "epoch": 0.11385319933850654, + "grad_norm": 33.25, + "learning_rate": 3.789741415854175e-07, + "loss": 0.9129, + "mean_token_accuracy": 0.7487689256668091, + "num_tokens": 34080487.0, + "step": 895 + }, + { + "epoch": 0.11398040961709706, + "grad_norm": 43.75, + "learning_rate": 3.793980500211954e-07, + "loss": 0.8684, + "mean_token_accuracy": 0.7630884647369385, + "num_tokens": 34119913.0, + "step": 896 + }, + { + "epoch": 0.11410761989568757, + "grad_norm": 67.5, + "learning_rate": 3.798219584569733e-07, + "loss": 0.7887, + "mean_token_accuracy": 0.782180905342102, + "num_tokens": 34155811.0, + "step": 897 + }, + { + "epoch": 0.11423483017427809, + "grad_norm": 30.0, + "learning_rate": 3.8024586689275115e-07, + "loss": 0.9473, + "mean_token_accuracy": 0.746902346611023, + "num_tokens": 34195022.0, + "step": 898 + }, + { + "epoch": 0.11436204045286859, + "grad_norm": 16.125, + "learning_rate": 3.80669775328529e-07, + "loss": 0.8811, + "mean_token_accuracy": 0.7587298154830933, + "num_tokens": 34229182.0, + "step": 899 + }, + { + "epoch": 0.1144892507314591, + "grad_norm": 32.25, + "learning_rate": 3.810936837643069e-07, + "loss": 0.7835, + "mean_token_accuracy": 0.7799924612045288, + "num_tokens": 34266931.0, + "step": 900 + }, + { + "epoch": 0.11461646101004962, + "grad_norm": 26.125, + "learning_rate": 3.815175922000848e-07, + "loss": 0.8068, + "mean_token_accuracy": 0.7752554416656494, + "num_tokens": 34301705.0, + "step": 901 + }, + { + "epoch": 0.11474367128864012, + "grad_norm": 61.0, + "learning_rate": 3.8194150063586264e-07, + "loss": 0.8992, + "mean_token_accuracy": 0.7567087411880493, + "num_tokens": 34340975.0, + "step": 902 + }, + { + "epoch": 0.11487088156723063, + "grad_norm": 54.25, + "learning_rate": 3.823654090716405e-07, + "loss": 0.8535, + "mean_token_accuracy": 0.7635543346405029, + "num_tokens": 34380024.0, + "step": 903 + }, + { + "epoch": 0.11499809184582115, + "grad_norm": 27.25, + "learning_rate": 3.827893175074184e-07, + "loss": 0.8716, + "mean_token_accuracy": 0.7665048837661743, + "num_tokens": 34419654.0, + "step": 904 + }, + { + "epoch": 0.11512530212441165, + "grad_norm": 58.25, + "learning_rate": 3.832132259431963e-07, + "loss": 0.8107, + "mean_token_accuracy": 0.7778184413909912, + "num_tokens": 34453034.0, + "step": 905 + }, + { + "epoch": 0.11525251240300216, + "grad_norm": 28.5, + "learning_rate": 3.8363713437897413e-07, + "loss": 0.9449, + "mean_token_accuracy": 0.7427191734313965, + "num_tokens": 34493694.0, + "step": 906 + }, + { + "epoch": 0.11537972268159268, + "grad_norm": 10.8125, + "learning_rate": 3.8406104281475197e-07, + "loss": 0.9136, + "mean_token_accuracy": 0.7521924376487732, + "num_tokens": 34531636.0, + "step": 907 + }, + { + "epoch": 0.11550693296018319, + "grad_norm": 9.5, + "learning_rate": 3.8448495125052987e-07, + "loss": 0.8928, + "mean_token_accuracy": 0.7523755431175232, + "num_tokens": 34570851.0, + "step": 908 + }, + { + "epoch": 0.11563414323877369, + "grad_norm": 33.75, + "learning_rate": 3.8490885968630777e-07, + "loss": 0.8296, + "mean_token_accuracy": 0.772234320640564, + "num_tokens": 34613314.0, + "step": 909 + }, + { + "epoch": 0.1157613535173642, + "grad_norm": 24.75, + "learning_rate": 3.853327681220856e-07, + "loss": 0.8422, + "mean_token_accuracy": 0.7680240869522095, + "num_tokens": 34656701.0, + "step": 910 + }, + { + "epoch": 0.11588856379595472, + "grad_norm": 46.75, + "learning_rate": 3.8575667655786346e-07, + "loss": 0.8503, + "mean_token_accuracy": 0.7667496800422668, + "num_tokens": 34688595.0, + "step": 911 + }, + { + "epoch": 0.11601577407454522, + "grad_norm": 15.875, + "learning_rate": 3.8618058499364136e-07, + "loss": 0.8476, + "mean_token_accuracy": 0.7683393955230713, + "num_tokens": 34726609.0, + "step": 912 + }, + { + "epoch": 0.11614298435313573, + "grad_norm": 15.125, + "learning_rate": 3.8660449342941926e-07, + "loss": 0.9263, + "mean_token_accuracy": 0.7471234798431396, + "num_tokens": 34769213.0, + "step": 913 + }, + { + "epoch": 0.11627019463172625, + "grad_norm": 12.875, + "learning_rate": 3.870284018651971e-07, + "loss": 0.8957, + "mean_token_accuracy": 0.7553977966308594, + "num_tokens": 34806722.0, + "step": 914 + }, + { + "epoch": 0.11639740491031675, + "grad_norm": 14.25, + "learning_rate": 3.8745231030097495e-07, + "loss": 0.8264, + "mean_token_accuracy": 0.773391604423523, + "num_tokens": 34846408.0, + "step": 915 + }, + { + "epoch": 0.11652461518890726, + "grad_norm": 34.25, + "learning_rate": 3.8787621873675285e-07, + "loss": 0.814, + "mean_token_accuracy": 0.7778081893920898, + "num_tokens": 34883791.0, + "step": 916 + }, + { + "epoch": 0.11665182546749778, + "grad_norm": 12.25, + "learning_rate": 3.883001271725307e-07, + "loss": 0.7956, + "mean_token_accuracy": 0.7797029614448547, + "num_tokens": 34922768.0, + "step": 917 + }, + { + "epoch": 0.11677903574608828, + "grad_norm": 46.5, + "learning_rate": 3.887240356083086e-07, + "loss": 0.7784, + "mean_token_accuracy": 0.7824025750160217, + "num_tokens": 34962184.0, + "step": 918 + }, + { + "epoch": 0.11690624602467879, + "grad_norm": 53.5, + "learning_rate": 3.8914794404408644e-07, + "loss": 0.866, + "mean_token_accuracy": 0.7606185078620911, + "num_tokens": 34991455.0, + "step": 919 + }, + { + "epoch": 0.1170334563032693, + "grad_norm": 53.0, + "learning_rate": 3.8957185247986434e-07, + "loss": 0.9088, + "mean_token_accuracy": 0.7530273795127869, + "num_tokens": 35025044.0, + "step": 920 + }, + { + "epoch": 0.11716066658185982, + "grad_norm": 50.5, + "learning_rate": 3.899957609156422e-07, + "loss": 0.8316, + "mean_token_accuracy": 0.7710975408554077, + "num_tokens": 35064564.0, + "step": 921 + }, + { + "epoch": 0.11728787686045032, + "grad_norm": 41.25, + "learning_rate": 3.904196693514201e-07, + "loss": 0.84, + "mean_token_accuracy": 0.7699521780014038, + "num_tokens": 35109786.0, + "step": 922 + }, + { + "epoch": 0.11741508713904084, + "grad_norm": 89.0, + "learning_rate": 3.9084357778719793e-07, + "loss": 0.9418, + "mean_token_accuracy": 0.7413361668586731, + "num_tokens": 35148351.0, + "step": 923 + }, + { + "epoch": 0.11754229741763135, + "grad_norm": 10.625, + "learning_rate": 3.9126748622297583e-07, + "loss": 0.9042, + "mean_token_accuracy": 0.7531908750534058, + "num_tokens": 35183464.0, + "step": 924 + }, + { + "epoch": 0.11766950769622185, + "grad_norm": 28.0, + "learning_rate": 3.916913946587537e-07, + "loss": 0.8962, + "mean_token_accuracy": 0.7536771893501282, + "num_tokens": 35218561.0, + "step": 925 + }, + { + "epoch": 0.11779671797481236, + "grad_norm": 10.4375, + "learning_rate": 3.921153030945316e-07, + "loss": 0.8846, + "mean_token_accuracy": 0.7557656764984131, + "num_tokens": 35257740.0, + "step": 926 + }, + { + "epoch": 0.11792392825340288, + "grad_norm": 53.0, + "learning_rate": 3.925392115303094e-07, + "loss": 0.8379, + "mean_token_accuracy": 0.770429253578186, + "num_tokens": 35291632.0, + "step": 927 + }, + { + "epoch": 0.11805113853199338, + "grad_norm": 12.375, + "learning_rate": 3.929631199660873e-07, + "loss": 0.8366, + "mean_token_accuracy": 0.7647106647491455, + "num_tokens": 35327686.0, + "step": 928 + }, + { + "epoch": 0.1181783488105839, + "grad_norm": 42.5, + "learning_rate": 3.9338702840186517e-07, + "loss": 0.8443, + "mean_token_accuracy": 0.762473464012146, + "num_tokens": 35364480.0, + "step": 929 + }, + { + "epoch": 0.11830555908917441, + "grad_norm": 56.75, + "learning_rate": 3.9381093683764307e-07, + "loss": 0.8326, + "mean_token_accuracy": 0.768363356590271, + "num_tokens": 35406017.0, + "step": 930 + }, + { + "epoch": 0.11843276936776491, + "grad_norm": 16.375, + "learning_rate": 3.942348452734209e-07, + "loss": 0.922, + "mean_token_accuracy": 0.7489050626754761, + "num_tokens": 35448447.0, + "step": 931 + }, + { + "epoch": 0.11855997964635542, + "grad_norm": 9.0, + "learning_rate": 3.946587537091988e-07, + "loss": 0.8509, + "mean_token_accuracy": 0.7647194266319275, + "num_tokens": 35488120.0, + "step": 932 + }, + { + "epoch": 0.11868718992494594, + "grad_norm": 18.625, + "learning_rate": 3.9508266214497666e-07, + "loss": 0.7935, + "mean_token_accuracy": 0.7790551781654358, + "num_tokens": 35531759.0, + "step": 933 + }, + { + "epoch": 0.11881440020353645, + "grad_norm": 10.1875, + "learning_rate": 3.9550657058075456e-07, + "loss": 0.8011, + "mean_token_accuracy": 0.7803207039833069, + "num_tokens": 35571476.0, + "step": 934 + }, + { + "epoch": 0.11894161048212695, + "grad_norm": 8.125, + "learning_rate": 3.959304790165324e-07, + "loss": 0.8368, + "mean_token_accuracy": 0.7677301168441772, + "num_tokens": 35608440.0, + "step": 935 + }, + { + "epoch": 0.11906882076071747, + "grad_norm": 60.5, + "learning_rate": 3.9635438745231025e-07, + "loss": 0.8593, + "mean_token_accuracy": 0.764519214630127, + "num_tokens": 35641861.0, + "step": 936 + }, + { + "epoch": 0.11919603103930798, + "grad_norm": 16.75, + "learning_rate": 3.9677829588808815e-07, + "loss": 0.8769, + "mean_token_accuracy": 0.7527868747711182, + "num_tokens": 35684626.0, + "step": 937 + }, + { + "epoch": 0.11932324131789848, + "grad_norm": 16.375, + "learning_rate": 3.9720220432386605e-07, + "loss": 0.8844, + "mean_token_accuracy": 0.753171443939209, + "num_tokens": 35718125.0, + "step": 938 + }, + { + "epoch": 0.119450451596489, + "grad_norm": 16.375, + "learning_rate": 3.976261127596439e-07, + "loss": 0.8217, + "mean_token_accuracy": 0.7767395973205566, + "num_tokens": 35752818.0, + "step": 939 + }, + { + "epoch": 0.11957766187507951, + "grad_norm": 64.0, + "learning_rate": 3.9805002119542174e-07, + "loss": 0.8044, + "mean_token_accuracy": 0.7785148024559021, + "num_tokens": 35787097.0, + "step": 940 + }, + { + "epoch": 0.11970487215367001, + "grad_norm": 60.0, + "learning_rate": 3.9847392963119964e-07, + "loss": 0.9177, + "mean_token_accuracy": 0.7502143979072571, + "num_tokens": 35825887.0, + "step": 941 + }, + { + "epoch": 0.11983208243226053, + "grad_norm": 40.75, + "learning_rate": 3.9889783806697754e-07, + "loss": 0.8148, + "mean_token_accuracy": 0.7749588489532471, + "num_tokens": 35863592.0, + "step": 942 + }, + { + "epoch": 0.11995929271085104, + "grad_norm": 47.75, + "learning_rate": 3.993217465027554e-07, + "loss": 0.9101, + "mean_token_accuracy": 0.7516124248504639, + "num_tokens": 35904483.0, + "step": 943 + }, + { + "epoch": 0.12008650298944154, + "grad_norm": 105.0, + "learning_rate": 3.9974565493853323e-07, + "loss": 0.7985, + "mean_token_accuracy": 0.7794484496116638, + "num_tokens": 35938662.0, + "step": 944 + }, + { + "epoch": 0.12021371326803205, + "grad_norm": 83.5, + "learning_rate": 4.0016956337431113e-07, + "loss": 0.8654, + "mean_token_accuracy": 0.761763870716095, + "num_tokens": 35975176.0, + "step": 945 + }, + { + "epoch": 0.12034092354662257, + "grad_norm": 20.75, + "learning_rate": 4.0059347181008903e-07, + "loss": 0.8614, + "mean_token_accuracy": 0.7603275775909424, + "num_tokens": 36008893.0, + "step": 946 + }, + { + "epoch": 0.12046813382521308, + "grad_norm": 24.75, + "learning_rate": 4.010173802458669e-07, + "loss": 0.7695, + "mean_token_accuracy": 0.7859340906143188, + "num_tokens": 36049882.0, + "step": 947 + }, + { + "epoch": 0.12059534410380358, + "grad_norm": 44.0, + "learning_rate": 4.014412886816447e-07, + "loss": 0.9189, + "mean_token_accuracy": 0.7467974424362183, + "num_tokens": 36090615.0, + "step": 948 + }, + { + "epoch": 0.1207225543823941, + "grad_norm": 31.125, + "learning_rate": 4.018651971174226e-07, + "loss": 0.8398, + "mean_token_accuracy": 0.7637832164764404, + "num_tokens": 36128159.0, + "step": 949 + }, + { + "epoch": 0.12084976466098461, + "grad_norm": 52.25, + "learning_rate": 4.022891055532005e-07, + "loss": 0.8295, + "mean_token_accuracy": 0.7685191035270691, + "num_tokens": 36172851.0, + "step": 950 + }, + { + "epoch": 0.12097697493957511, + "grad_norm": 12.5, + "learning_rate": 4.0271301398897837e-07, + "loss": 0.9087, + "mean_token_accuracy": 0.7559993267059326, + "num_tokens": 36214282.0, + "step": 951 + }, + { + "epoch": 0.12110418521816563, + "grad_norm": 78.5, + "learning_rate": 4.031369224247562e-07, + "loss": 0.8304, + "mean_token_accuracy": 0.7675397992134094, + "num_tokens": 36248619.0, + "step": 952 + }, + { + "epoch": 0.12123139549675614, + "grad_norm": 12.1875, + "learning_rate": 4.035608308605341e-07, + "loss": 0.8714, + "mean_token_accuracy": 0.7575130462646484, + "num_tokens": 36289070.0, + "step": 953 + }, + { + "epoch": 0.12135860577534664, + "grad_norm": 42.0, + "learning_rate": 4.03984739296312e-07, + "loss": 0.8531, + "mean_token_accuracy": 0.7670208811759949, + "num_tokens": 36330883.0, + "step": 954 + }, + { + "epoch": 0.12148581605393716, + "grad_norm": 16.375, + "learning_rate": 4.044086477320898e-07, + "loss": 0.8402, + "mean_token_accuracy": 0.768088698387146, + "num_tokens": 36371507.0, + "step": 955 + }, + { + "epoch": 0.12161302633252767, + "grad_norm": 11.25, + "learning_rate": 4.048325561678677e-07, + "loss": 0.8761, + "mean_token_accuracy": 0.7569968104362488, + "num_tokens": 36407262.0, + "step": 956 + }, + { + "epoch": 0.12174023661111817, + "grad_norm": 32.75, + "learning_rate": 4.052564646036456e-07, + "loss": 0.8513, + "mean_token_accuracy": 0.7623845934867859, + "num_tokens": 36448089.0, + "step": 957 + }, + { + "epoch": 0.12186744688970869, + "grad_norm": 12.3125, + "learning_rate": 4.056803730394235e-07, + "loss": 0.8739, + "mean_token_accuracy": 0.7542451620101929, + "num_tokens": 36484022.0, + "step": 958 + }, + { + "epoch": 0.1219946571682992, + "grad_norm": 34.0, + "learning_rate": 4.061042814752013e-07, + "loss": 0.8515, + "mean_token_accuracy": 0.7659925818443298, + "num_tokens": 36519385.0, + "step": 959 + }, + { + "epoch": 0.12212186744688971, + "grad_norm": 12.5625, + "learning_rate": 4.065281899109792e-07, + "loss": 0.8617, + "mean_token_accuracy": 0.7600671052932739, + "num_tokens": 36551943.0, + "step": 960 + }, + { + "epoch": 0.12224907772548022, + "grad_norm": 43.5, + "learning_rate": 4.069520983467571e-07, + "loss": 0.756, + "mean_token_accuracy": 0.7874608039855957, + "num_tokens": 36590927.0, + "step": 961 + }, + { + "epoch": 0.12237628800407073, + "grad_norm": 36.5, + "learning_rate": 4.07376006782535e-07, + "loss": 0.8472, + "mean_token_accuracy": 0.7646480798721313, + "num_tokens": 36625199.0, + "step": 962 + }, + { + "epoch": 0.12250349828266124, + "grad_norm": 15.8125, + "learning_rate": 4.077999152183128e-07, + "loss": 0.8002, + "mean_token_accuracy": 0.7745091915130615, + "num_tokens": 36658981.0, + "step": 963 + }, + { + "epoch": 0.12263070856125174, + "grad_norm": 38.75, + "learning_rate": 4.082238236540907e-07, + "loss": 0.8345, + "mean_token_accuracy": 0.7713778018951416, + "num_tokens": 36704214.0, + "step": 964 + }, + { + "epoch": 0.12275791883984226, + "grad_norm": 18.0, + "learning_rate": 4.086477320898686e-07, + "loss": 0.7987, + "mean_token_accuracy": 0.7766036987304688, + "num_tokens": 36744782.0, + "step": 965 + }, + { + "epoch": 0.12288512911843277, + "grad_norm": 78.0, + "learning_rate": 4.090716405256465e-07, + "loss": 0.85, + "mean_token_accuracy": 0.7625743746757507, + "num_tokens": 36783645.0, + "step": 966 + }, + { + "epoch": 0.12301233939702327, + "grad_norm": 20.75, + "learning_rate": 4.094955489614243e-07, + "loss": 0.8254, + "mean_token_accuracy": 0.7678741216659546, + "num_tokens": 36817539.0, + "step": 967 + }, + { + "epoch": 0.12313954967561379, + "grad_norm": 26.125, + "learning_rate": 4.099194573972022e-07, + "loss": 0.8091, + "mean_token_accuracy": 0.7742983102798462, + "num_tokens": 36859144.0, + "step": 968 + }, + { + "epoch": 0.1232667599542043, + "grad_norm": 13.5, + "learning_rate": 4.1034336583298007e-07, + "loss": 0.8597, + "mean_token_accuracy": 0.7607051134109497, + "num_tokens": 36900833.0, + "step": 969 + }, + { + "epoch": 0.1233939702327948, + "grad_norm": 11.8125, + "learning_rate": 4.1076727426875797e-07, + "loss": 0.8415, + "mean_token_accuracy": 0.7644674181938171, + "num_tokens": 36940935.0, + "step": 970 + }, + { + "epoch": 0.12352118051138532, + "grad_norm": 61.75, + "learning_rate": 4.1119118270453577e-07, + "loss": 0.9343, + "mean_token_accuracy": 0.7363735437393188, + "num_tokens": 36979874.0, + "step": 971 + }, + { + "epoch": 0.12364839078997583, + "grad_norm": 96.0, + "learning_rate": 4.1161509114031366e-07, + "loss": 0.7915, + "mean_token_accuracy": 0.779619574546814, + "num_tokens": 37017917.0, + "step": 972 + }, + { + "epoch": 0.12377560106856635, + "grad_norm": 92.0, + "learning_rate": 4.1203899957609156e-07, + "loss": 0.8721, + "mean_token_accuracy": 0.7616618275642395, + "num_tokens": 37054914.0, + "step": 973 + }, + { + "epoch": 0.12390281134715685, + "grad_norm": 82.5, + "learning_rate": 4.124629080118694e-07, + "loss": 0.8323, + "mean_token_accuracy": 0.7660294771194458, + "num_tokens": 37088277.0, + "step": 974 + }, + { + "epoch": 0.12403002162574736, + "grad_norm": 17.25, + "learning_rate": 4.1288681644764726e-07, + "loss": 0.8667, + "mean_token_accuracy": 0.7559898495674133, + "num_tokens": 37126963.0, + "step": 975 + }, + { + "epoch": 0.12415723190433788, + "grad_norm": 14.125, + "learning_rate": 4.1331072488342515e-07, + "loss": 0.884, + "mean_token_accuracy": 0.755911111831665, + "num_tokens": 37161948.0, + "step": 976 + }, + { + "epoch": 0.12428444218292838, + "grad_norm": 15.125, + "learning_rate": 4.1373463331920305e-07, + "loss": 0.7636, + "mean_token_accuracy": 0.7857329845428467, + "num_tokens": 37198552.0, + "step": 977 + }, + { + "epoch": 0.12441165246151889, + "grad_norm": 9.6875, + "learning_rate": 4.141585417549809e-07, + "loss": 0.8588, + "mean_token_accuracy": 0.7572212219238281, + "num_tokens": 37235990.0, + "step": 978 + }, + { + "epoch": 0.1245388627401094, + "grad_norm": 75.5, + "learning_rate": 4.1458245019075875e-07, + "loss": 0.8153, + "mean_token_accuracy": 0.7698215246200562, + "num_tokens": 37269664.0, + "step": 979 + }, + { + "epoch": 0.1246660730186999, + "grad_norm": 87.5, + "learning_rate": 4.1500635862653664e-07, + "loss": 0.8484, + "mean_token_accuracy": 0.7636063098907471, + "num_tokens": 37308647.0, + "step": 980 + }, + { + "epoch": 0.12479328329729042, + "grad_norm": 118.0, + "learning_rate": 4.1543026706231454e-07, + "loss": 0.8065, + "mean_token_accuracy": 0.774264931678772, + "num_tokens": 37345953.0, + "step": 981 + }, + { + "epoch": 0.12492049357588093, + "grad_norm": 39.5, + "learning_rate": 4.158541754980924e-07, + "loss": 0.7824, + "mean_token_accuracy": 0.7850643992424011, + "num_tokens": 37382735.0, + "step": 982 + }, + { + "epoch": 0.12504770385447145, + "grad_norm": 108.0, + "learning_rate": 4.1627808393387024e-07, + "loss": 0.8681, + "mean_token_accuracy": 0.7520871162414551, + "num_tokens": 37419605.0, + "step": 983 + }, + { + "epoch": 0.12517491413306195, + "grad_norm": 13.0, + "learning_rate": 4.1670199236964813e-07, + "loss": 0.8782, + "mean_token_accuracy": 0.7516127824783325, + "num_tokens": 37455395.0, + "step": 984 + }, + { + "epoch": 0.12530212441165245, + "grad_norm": 20.75, + "learning_rate": 4.1712590080542603e-07, + "loss": 0.7826, + "mean_token_accuracy": 0.7836606502532959, + "num_tokens": 37492208.0, + "step": 985 + }, + { + "epoch": 0.12542933469024298, + "grad_norm": 30.875, + "learning_rate": 4.175498092412039e-07, + "loss": 0.8344, + "mean_token_accuracy": 0.7684052586555481, + "num_tokens": 37533675.0, + "step": 986 + }, + { + "epoch": 0.12555654496883348, + "grad_norm": 26.625, + "learning_rate": 4.179737176769817e-07, + "loss": 0.8037, + "mean_token_accuracy": 0.7788405418395996, + "num_tokens": 37572792.0, + "step": 987 + }, + { + "epoch": 0.12568375524742398, + "grad_norm": 15.375, + "learning_rate": 4.183976261127596e-07, + "loss": 0.8038, + "mean_token_accuracy": 0.7750339508056641, + "num_tokens": 37611481.0, + "step": 988 + }, + { + "epoch": 0.1258109655260145, + "grad_norm": 21.75, + "learning_rate": 4.1882153454853747e-07, + "loss": 0.8948, + "mean_token_accuracy": 0.753101110458374, + "num_tokens": 37648253.0, + "step": 989 + }, + { + "epoch": 0.125938175804605, + "grad_norm": 9.625, + "learning_rate": 4.1924544298431537e-07, + "loss": 0.8003, + "mean_token_accuracy": 0.7769972085952759, + "num_tokens": 37687086.0, + "step": 990 + }, + { + "epoch": 0.12606538608319554, + "grad_norm": 92.5, + "learning_rate": 4.196693514200932e-07, + "loss": 0.7896, + "mean_token_accuracy": 0.7794468402862549, + "num_tokens": 37730928.0, + "step": 991 + }, + { + "epoch": 0.12619259636178604, + "grad_norm": 12.75, + "learning_rate": 4.200932598558711e-07, + "loss": 0.8387, + "mean_token_accuracy": 0.7611480951309204, + "num_tokens": 37771857.0, + "step": 992 + }, + { + "epoch": 0.12631980664037654, + "grad_norm": 35.25, + "learning_rate": 4.2051716829164896e-07, + "loss": 0.8142, + "mean_token_accuracy": 0.7755063772201538, + "num_tokens": 37817517.0, + "step": 993 + }, + { + "epoch": 0.12644701691896706, + "grad_norm": 56.5, + "learning_rate": 4.2094107672742686e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7513904571533203, + "num_tokens": 37855891.0, + "step": 994 + }, + { + "epoch": 0.12657422719755757, + "grad_norm": 39.75, + "learning_rate": 4.2136498516320476e-07, + "loss": 0.8286, + "mean_token_accuracy": 0.7734410762786865, + "num_tokens": 37888719.0, + "step": 995 + }, + { + "epoch": 0.12670143747614807, + "grad_norm": 63.75, + "learning_rate": 4.217888935989826e-07, + "loss": 0.8397, + "mean_token_accuracy": 0.7636210918426514, + "num_tokens": 37932968.0, + "step": 996 + }, + { + "epoch": 0.1268286477547386, + "grad_norm": 45.0, + "learning_rate": 4.2221280203476045e-07, + "loss": 0.8353, + "mean_token_accuracy": 0.7667267322540283, + "num_tokens": 37974297.0, + "step": 997 + }, + { + "epoch": 0.1269558580333291, + "grad_norm": 13.875, + "learning_rate": 4.2263671047053835e-07, + "loss": 0.8542, + "mean_token_accuracy": 0.7612994909286499, + "num_tokens": 38020294.0, + "step": 998 + }, + { + "epoch": 0.1270830683119196, + "grad_norm": 42.25, + "learning_rate": 4.2306061890631625e-07, + "loss": 0.825, + "mean_token_accuracy": 0.7729141116142273, + "num_tokens": 38063322.0, + "step": 999 + }, + { + "epoch": 0.12721027859051012, + "grad_norm": 10.5, + "learning_rate": 4.234845273420941e-07, + "loss": 0.9549, + "mean_token_accuracy": 0.7378584146499634, + "num_tokens": 38101052.0, + "step": 1000 + }, + { + "epoch": 0.12733748886910062, + "grad_norm": 8.125, + "learning_rate": 4.2390843577787194e-07, + "loss": 0.8434, + "mean_token_accuracy": 0.7616710662841797, + "num_tokens": 38145472.0, + "step": 1001 + }, + { + "epoch": 0.12746469914769112, + "grad_norm": 42.25, + "learning_rate": 4.2433234421364984e-07, + "loss": 0.8262, + "mean_token_accuracy": 0.7678482532501221, + "num_tokens": 38182662.0, + "step": 1002 + }, + { + "epoch": 0.12759190942628165, + "grad_norm": 23.625, + "learning_rate": 4.2475625264942774e-07, + "loss": 0.7903, + "mean_token_accuracy": 0.7778724431991577, + "num_tokens": 38224735.0, + "step": 1003 + }, + { + "epoch": 0.12771911970487215, + "grad_norm": 24.375, + "learning_rate": 4.251801610852056e-07, + "loss": 0.8736, + "mean_token_accuracy": 0.759963870048523, + "num_tokens": 38255847.0, + "step": 1004 + }, + { + "epoch": 0.12784632998346265, + "grad_norm": 43.75, + "learning_rate": 4.2560406952098343e-07, + "loss": 0.8792, + "mean_token_accuracy": 0.7592425346374512, + "num_tokens": 38296921.0, + "step": 1005 + }, + { + "epoch": 0.12797354026205318, + "grad_norm": 10.1875, + "learning_rate": 4.2602797795676133e-07, + "loss": 0.8904, + "mean_token_accuracy": 0.7543291449546814, + "num_tokens": 38336408.0, + "step": 1006 + }, + { + "epoch": 0.12810075054064368, + "grad_norm": 15.875, + "learning_rate": 4.2645188639253923e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.785287618637085, + "num_tokens": 38372809.0, + "step": 1007 + }, + { + "epoch": 0.12822796081923418, + "grad_norm": 29.75, + "learning_rate": 4.26875794828317e-07, + "loss": 0.8301, + "mean_token_accuracy": 0.7672501802444458, + "num_tokens": 38413312.0, + "step": 1008 + }, + { + "epoch": 0.1283551710978247, + "grad_norm": 35.25, + "learning_rate": 4.272997032640949e-07, + "loss": 0.8758, + "mean_token_accuracy": 0.7570757865905762, + "num_tokens": 38457710.0, + "step": 1009 + }, + { + "epoch": 0.1284823813764152, + "grad_norm": 47.0, + "learning_rate": 4.277236116998728e-07, + "loss": 0.8188, + "mean_token_accuracy": 0.7754043340682983, + "num_tokens": 38499630.0, + "step": 1010 + }, + { + "epoch": 0.1286095916550057, + "grad_norm": 12.0, + "learning_rate": 4.281475201356507e-07, + "loss": 0.8304, + "mean_token_accuracy": 0.7701732516288757, + "num_tokens": 38538133.0, + "step": 1011 + }, + { + "epoch": 0.12873680193359624, + "grad_norm": 45.25, + "learning_rate": 4.285714285714285e-07, + "loss": 0.8761, + "mean_token_accuracy": 0.7566054463386536, + "num_tokens": 38576816.0, + "step": 1012 + }, + { + "epoch": 0.12886401221218674, + "grad_norm": 70.5, + "learning_rate": 4.289953370072064e-07, + "loss": 0.8195, + "mean_token_accuracy": 0.7698135375976562, + "num_tokens": 38615270.0, + "step": 1013 + }, + { + "epoch": 0.12899122249077727, + "grad_norm": 30.375, + "learning_rate": 4.294192454429843e-07, + "loss": 0.8763, + "mean_token_accuracy": 0.7594831585884094, + "num_tokens": 38654173.0, + "step": 1014 + }, + { + "epoch": 0.12911843276936777, + "grad_norm": 43.5, + "learning_rate": 4.298431538787622e-07, + "loss": 0.7071, + "mean_token_accuracy": 0.8023419380187988, + "num_tokens": 38695758.0, + "step": 1015 + }, + { + "epoch": 0.12924564304795827, + "grad_norm": 18.375, + "learning_rate": 4.3026706231454e-07, + "loss": 0.9066, + "mean_token_accuracy": 0.7452760934829712, + "num_tokens": 38729524.0, + "step": 1016 + }, + { + "epoch": 0.1293728533265488, + "grad_norm": 44.25, + "learning_rate": 4.306909707503179e-07, + "loss": 0.8313, + "mean_token_accuracy": 0.7717088460922241, + "num_tokens": 38767500.0, + "step": 1017 + }, + { + "epoch": 0.1295000636051393, + "grad_norm": 43.5, + "learning_rate": 4.311148791860958e-07, + "loss": 0.8382, + "mean_token_accuracy": 0.7631324529647827, + "num_tokens": 38798683.0, + "step": 1018 + }, + { + "epoch": 0.1296272738837298, + "grad_norm": 18.125, + "learning_rate": 4.315387876218737e-07, + "loss": 0.8956, + "mean_token_accuracy": 0.7542533874511719, + "num_tokens": 38833653.0, + "step": 1019 + }, + { + "epoch": 0.12975448416232033, + "grad_norm": 10.5625, + "learning_rate": 4.319626960576515e-07, + "loss": 0.8765, + "mean_token_accuracy": 0.7564222812652588, + "num_tokens": 38869428.0, + "step": 1020 + }, + { + "epoch": 0.12988169444091083, + "grad_norm": 11.5, + "learning_rate": 4.323866044934294e-07, + "loss": 0.7928, + "mean_token_accuracy": 0.7772771120071411, + "num_tokens": 38909502.0, + "step": 1021 + }, + { + "epoch": 0.13000890471950133, + "grad_norm": 29.75, + "learning_rate": 4.328105129292073e-07, + "loss": 0.8273, + "mean_token_accuracy": 0.7687222957611084, + "num_tokens": 38951730.0, + "step": 1022 + }, + { + "epoch": 0.13013611499809186, + "grad_norm": 50.0, + "learning_rate": 4.332344213649852e-07, + "loss": 0.8099, + "mean_token_accuracy": 0.7749766111373901, + "num_tokens": 38988981.0, + "step": 1023 + }, + { + "epoch": 0.13026332527668236, + "grad_norm": 92.5, + "learning_rate": 4.33658329800763e-07, + "loss": 0.8226, + "mean_token_accuracy": 0.7733371257781982, + "num_tokens": 39026289.0, + "step": 1024 + }, + { + "epoch": 0.13039053555527286, + "grad_norm": 83.5, + "learning_rate": 4.340822382365409e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7731224298477173, + "num_tokens": 39069113.0, + "step": 1025 + }, + { + "epoch": 0.13051774583386339, + "grad_norm": 36.5, + "learning_rate": 4.345061466723188e-07, + "loss": 0.7734, + "mean_token_accuracy": 0.7815261483192444, + "num_tokens": 39112991.0, + "step": 1026 + }, + { + "epoch": 0.13064495611245389, + "grad_norm": 21.5, + "learning_rate": 4.3493005510809663e-07, + "loss": 0.8707, + "mean_token_accuracy": 0.7557674646377563, + "num_tokens": 39149593.0, + "step": 1027 + }, + { + "epoch": 0.1307721663910444, + "grad_norm": 9.6875, + "learning_rate": 4.353539635438745e-07, + "loss": 0.8046, + "mean_token_accuracy": 0.7761911749839783, + "num_tokens": 39186570.0, + "step": 1028 + }, + { + "epoch": 0.13089937666963491, + "grad_norm": 11.375, + "learning_rate": 4.357778719796524e-07, + "loss": 0.8993, + "mean_token_accuracy": 0.7460259199142456, + "num_tokens": 39220666.0, + "step": 1029 + }, + { + "epoch": 0.13102658694822542, + "grad_norm": 33.0, + "learning_rate": 4.362017804154303e-07, + "loss": 0.7923, + "mean_token_accuracy": 0.7820510268211365, + "num_tokens": 39258372.0, + "step": 1030 + }, + { + "epoch": 0.13115379722681592, + "grad_norm": 114.0, + "learning_rate": 4.366256888512081e-07, + "loss": 0.795, + "mean_token_accuracy": 0.7789103984832764, + "num_tokens": 39293647.0, + "step": 1031 + }, + { + "epoch": 0.13128100750540644, + "grad_norm": 72.5, + "learning_rate": 4.3704959728698597e-07, + "loss": 0.8251, + "mean_token_accuracy": 0.7688651084899902, + "num_tokens": 39332811.0, + "step": 1032 + }, + { + "epoch": 0.13140821778399694, + "grad_norm": 20.5, + "learning_rate": 4.3747350572276386e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7638885974884033, + "num_tokens": 39365971.0, + "step": 1033 + }, + { + "epoch": 0.13153542806258745, + "grad_norm": 27.875, + "learning_rate": 4.3789741415854176e-07, + "loss": 0.8497, + "mean_token_accuracy": 0.7652407884597778, + "num_tokens": 39403178.0, + "step": 1034 + }, + { + "epoch": 0.13166263834117797, + "grad_norm": 86.0, + "learning_rate": 4.383213225943196e-07, + "loss": 0.8067, + "mean_token_accuracy": 0.7719363570213318, + "num_tokens": 39441303.0, + "step": 1035 + }, + { + "epoch": 0.13178984861976847, + "grad_norm": 39.5, + "learning_rate": 4.3874523103009746e-07, + "loss": 0.7661, + "mean_token_accuracy": 0.7868616580963135, + "num_tokens": 39477359.0, + "step": 1036 + }, + { + "epoch": 0.13191705889835897, + "grad_norm": 60.5, + "learning_rate": 4.3916913946587536e-07, + "loss": 0.7796, + "mean_token_accuracy": 0.7768011093139648, + "num_tokens": 39511687.0, + "step": 1037 + }, + { + "epoch": 0.1320442691769495, + "grad_norm": 61.0, + "learning_rate": 4.3959304790165325e-07, + "loss": 0.8059, + "mean_token_accuracy": 0.7720578908920288, + "num_tokens": 39547453.0, + "step": 1038 + }, + { + "epoch": 0.13217147945554, + "grad_norm": 71.5, + "learning_rate": 4.400169563374311e-07, + "loss": 0.7974, + "mean_token_accuracy": 0.7732676267623901, + "num_tokens": 39588934.0, + "step": 1039 + }, + { + "epoch": 0.13229868973413053, + "grad_norm": 65.0, + "learning_rate": 4.4044086477320895e-07, + "loss": 0.9158, + "mean_token_accuracy": 0.7461695671081543, + "num_tokens": 39627638.0, + "step": 1040 + }, + { + "epoch": 0.13242590001272103, + "grad_norm": 40.5, + "learning_rate": 4.4086477320898685e-07, + "loss": 0.8162, + "mean_token_accuracy": 0.7682374715805054, + "num_tokens": 39661696.0, + "step": 1041 + }, + { + "epoch": 0.13255311029131153, + "grad_norm": 35.0, + "learning_rate": 4.4128868164476474e-07, + "loss": 0.8097, + "mean_token_accuracy": 0.7761013507843018, + "num_tokens": 39702357.0, + "step": 1042 + }, + { + "epoch": 0.13268032056990206, + "grad_norm": 16.375, + "learning_rate": 4.417125900805426e-07, + "loss": 0.787, + "mean_token_accuracy": 0.7812826037406921, + "num_tokens": 39744960.0, + "step": 1043 + }, + { + "epoch": 0.13280753084849256, + "grad_norm": 9.9375, + "learning_rate": 4.4213649851632044e-07, + "loss": 0.8561, + "mean_token_accuracy": 0.7616996765136719, + "num_tokens": 39785394.0, + "step": 1044 + }, + { + "epoch": 0.13293474112708306, + "grad_norm": 72.0, + "learning_rate": 4.4256040695209834e-07, + "loss": 0.853, + "mean_token_accuracy": 0.7628990411758423, + "num_tokens": 39822698.0, + "step": 1045 + }, + { + "epoch": 0.1330619514056736, + "grad_norm": 22.25, + "learning_rate": 4.429843153878762e-07, + "loss": 0.8571, + "mean_token_accuracy": 0.7595803737640381, + "num_tokens": 39856874.0, + "step": 1046 + }, + { + "epoch": 0.1331891616842641, + "grad_norm": 14.25, + "learning_rate": 4.434082238236541e-07, + "loss": 0.8833, + "mean_token_accuracy": 0.7538712024688721, + "num_tokens": 39899683.0, + "step": 1047 + }, + { + "epoch": 0.1333163719628546, + "grad_norm": 16.5, + "learning_rate": 4.4383213225943193e-07, + "loss": 0.8479, + "mean_token_accuracy": 0.7659793496131897, + "num_tokens": 39940565.0, + "step": 1048 + }, + { + "epoch": 0.13344358224144512, + "grad_norm": 26.625, + "learning_rate": 4.442560406952098e-07, + "loss": 0.8452, + "mean_token_accuracy": 0.7586374878883362, + "num_tokens": 39979010.0, + "step": 1049 + }, + { + "epoch": 0.13357079252003562, + "grad_norm": 9.6875, + "learning_rate": 4.4467994913098767e-07, + "loss": 0.7934, + "mean_token_accuracy": 0.7794747948646545, + "num_tokens": 40021151.0, + "step": 1050 + }, + { + "epoch": 0.13369800279862612, + "grad_norm": 38.0, + "learning_rate": 4.4510385756676557e-07, + "loss": 0.8738, + "mean_token_accuracy": 0.7549093961715698, + "num_tokens": 40059324.0, + "step": 1051 + }, + { + "epoch": 0.13382521307721665, + "grad_norm": 19.25, + "learning_rate": 4.455277660025434e-07, + "loss": 0.7985, + "mean_token_accuracy": 0.7755637168884277, + "num_tokens": 40104577.0, + "step": 1052 + }, + { + "epoch": 0.13395242335580715, + "grad_norm": 13.0, + "learning_rate": 4.459516744383213e-07, + "loss": 0.8502, + "mean_token_accuracy": 0.7623399496078491, + "num_tokens": 40139734.0, + "step": 1053 + }, + { + "epoch": 0.13407963363439765, + "grad_norm": 8.875, + "learning_rate": 4.4637558287409916e-07, + "loss": 0.8336, + "mean_token_accuracy": 0.7689833641052246, + "num_tokens": 40180724.0, + "step": 1054 + }, + { + "epoch": 0.13420684391298818, + "grad_norm": 8.4375, + "learning_rate": 4.4679949130987706e-07, + "loss": 0.8172, + "mean_token_accuracy": 0.7722484469413757, + "num_tokens": 40220475.0, + "step": 1055 + }, + { + "epoch": 0.13433405419157868, + "grad_norm": 66.0, + "learning_rate": 4.472233997456549e-07, + "loss": 0.8466, + "mean_token_accuracy": 0.7627736330032349, + "num_tokens": 40258208.0, + "step": 1056 + }, + { + "epoch": 0.13446126447016918, + "grad_norm": 47.5, + "learning_rate": 4.476473081814328e-07, + "loss": 0.8258, + "mean_token_accuracy": 0.767270565032959, + "num_tokens": 40299127.0, + "step": 1057 + }, + { + "epoch": 0.1345884747487597, + "grad_norm": 22.25, + "learning_rate": 4.4807121661721065e-07, + "loss": 0.7734, + "mean_token_accuracy": 0.783748984336853, + "num_tokens": 40339331.0, + "step": 1058 + }, + { + "epoch": 0.1347156850273502, + "grad_norm": 64.5, + "learning_rate": 4.4849512505298855e-07, + "loss": 0.8297, + "mean_token_accuracy": 0.7638365626335144, + "num_tokens": 40379386.0, + "step": 1059 + }, + { + "epoch": 0.1348428953059407, + "grad_norm": 22.875, + "learning_rate": 4.489190334887664e-07, + "loss": 0.8731, + "mean_token_accuracy": 0.7585963010787964, + "num_tokens": 40419868.0, + "step": 1060 + }, + { + "epoch": 0.13497010558453124, + "grad_norm": 83.5, + "learning_rate": 4.493429419245443e-07, + "loss": 0.8451, + "mean_token_accuracy": 0.7648396492004395, + "num_tokens": 40455346.0, + "step": 1061 + }, + { + "epoch": 0.13509731586312174, + "grad_norm": 9.0, + "learning_rate": 4.4976685036032214e-07, + "loss": 0.7955, + "mean_token_accuracy": 0.7778990864753723, + "num_tokens": 40490087.0, + "step": 1062 + }, + { + "epoch": 0.13522452614171224, + "grad_norm": 32.5, + "learning_rate": 4.5019075879610004e-07, + "loss": 0.882, + "mean_token_accuracy": 0.750700056552887, + "num_tokens": 40524799.0, + "step": 1063 + }, + { + "epoch": 0.13535173642030277, + "grad_norm": 16.5, + "learning_rate": 4.506146672318779e-07, + "loss": 0.8147, + "mean_token_accuracy": 0.7689935564994812, + "num_tokens": 40563740.0, + "step": 1064 + }, + { + "epoch": 0.13547894669889327, + "grad_norm": 31.5, + "learning_rate": 4.5103857566765573e-07, + "loss": 0.8888, + "mean_token_accuracy": 0.7538616061210632, + "num_tokens": 40604743.0, + "step": 1065 + }, + { + "epoch": 0.1356061569774838, + "grad_norm": 21.5, + "learning_rate": 4.5146248410343363e-07, + "loss": 0.7793, + "mean_token_accuracy": 0.7805470824241638, + "num_tokens": 40640909.0, + "step": 1066 + }, + { + "epoch": 0.1357333672560743, + "grad_norm": 20.875, + "learning_rate": 4.5188639253921153e-07, + "loss": 0.7829, + "mean_token_accuracy": 0.7800710201263428, + "num_tokens": 40678930.0, + "step": 1067 + }, + { + "epoch": 0.1358605775346648, + "grad_norm": 42.75, + "learning_rate": 4.523103009749894e-07, + "loss": 0.7828, + "mean_token_accuracy": 0.7798633575439453, + "num_tokens": 40715609.0, + "step": 1068 + }, + { + "epoch": 0.13598778781325532, + "grad_norm": 96.5, + "learning_rate": 4.527342094107672e-07, + "loss": 0.7965, + "mean_token_accuracy": 0.7745041251182556, + "num_tokens": 40760533.0, + "step": 1069 + }, + { + "epoch": 0.13611499809184582, + "grad_norm": 60.25, + "learning_rate": 4.531581178465451e-07, + "loss": 0.8274, + "mean_token_accuracy": 0.7686758637428284, + "num_tokens": 40800039.0, + "step": 1070 + }, + { + "epoch": 0.13624220837043632, + "grad_norm": 39.25, + "learning_rate": 4.53582026282323e-07, + "loss": 0.7979, + "mean_token_accuracy": 0.7778035402297974, + "num_tokens": 40840207.0, + "step": 1071 + }, + { + "epoch": 0.13636941864902685, + "grad_norm": 15.5625, + "learning_rate": 4.5400593471810087e-07, + "loss": 0.8962, + "mean_token_accuracy": 0.7496604919433594, + "num_tokens": 40879559.0, + "step": 1072 + }, + { + "epoch": 0.13649662892761735, + "grad_norm": 32.75, + "learning_rate": 4.544298431538787e-07, + "loss": 0.8045, + "mean_token_accuracy": 0.7758808135986328, + "num_tokens": 40919608.0, + "step": 1073 + }, + { + "epoch": 0.13662383920620785, + "grad_norm": 29.0, + "learning_rate": 4.548537515896566e-07, + "loss": 0.8964, + "mean_token_accuracy": 0.7500971555709839, + "num_tokens": 40953648.0, + "step": 1074 + }, + { + "epoch": 0.13675104948479838, + "grad_norm": 79.0, + "learning_rate": 4.552776600254345e-07, + "loss": 0.8635, + "mean_token_accuracy": 0.7564100623130798, + "num_tokens": 40989675.0, + "step": 1075 + }, + { + "epoch": 0.13687825976338888, + "grad_norm": 37.25, + "learning_rate": 4.5570156846121236e-07, + "loss": 0.7833, + "mean_token_accuracy": 0.7800505757331848, + "num_tokens": 41027992.0, + "step": 1076 + }, + { + "epoch": 0.13700547004197938, + "grad_norm": 133.0, + "learning_rate": 4.561254768969902e-07, + "loss": 0.7944, + "mean_token_accuracy": 0.7732223868370056, + "num_tokens": 41066925.0, + "step": 1077 + }, + { + "epoch": 0.1371326803205699, + "grad_norm": 96.5, + "learning_rate": 4.565493853327681e-07, + "loss": 0.8066, + "mean_token_accuracy": 0.7760273218154907, + "num_tokens": 41105090.0, + "step": 1078 + }, + { + "epoch": 0.1372598905991604, + "grad_norm": 103.0, + "learning_rate": 4.56973293768546e-07, + "loss": 0.8618, + "mean_token_accuracy": 0.7609249949455261, + "num_tokens": 41141924.0, + "step": 1079 + }, + { + "epoch": 0.1373871008777509, + "grad_norm": 130.0, + "learning_rate": 4.573972022043238e-07, + "loss": 0.9265, + "mean_token_accuracy": 0.7398228645324707, + "num_tokens": 41181512.0, + "step": 1080 + }, + { + "epoch": 0.13751431115634144, + "grad_norm": 84.0, + "learning_rate": 4.578211106401017e-07, + "loss": 0.8558, + "mean_token_accuracy": 0.7597695589065552, + "num_tokens": 41224015.0, + "step": 1081 + }, + { + "epoch": 0.13764152143493194, + "grad_norm": 44.0, + "learning_rate": 4.582450190758796e-07, + "loss": 0.8445, + "mean_token_accuracy": 0.7631808519363403, + "num_tokens": 41265568.0, + "step": 1082 + }, + { + "epoch": 0.13776873171352244, + "grad_norm": 8.625, + "learning_rate": 4.586689275116575e-07, + "loss": 0.7479, + "mean_token_accuracy": 0.7891853451728821, + "num_tokens": 41303595.0, + "step": 1083 + }, + { + "epoch": 0.13789594199211297, + "grad_norm": 12.0625, + "learning_rate": 4.590928359474353e-07, + "loss": 0.8873, + "mean_token_accuracy": 0.7533103823661804, + "num_tokens": 41340119.0, + "step": 1084 + }, + { + "epoch": 0.13802315227070347, + "grad_norm": 9.1875, + "learning_rate": 4.595167443832132e-07, + "loss": 0.8232, + "mean_token_accuracy": 0.770919680595398, + "num_tokens": 41380184.0, + "step": 1085 + }, + { + "epoch": 0.13815036254929397, + "grad_norm": 26.125, + "learning_rate": 4.599406528189911e-07, + "loss": 0.751, + "mean_token_accuracy": 0.786048412322998, + "num_tokens": 41419038.0, + "step": 1086 + }, + { + "epoch": 0.1382775728278845, + "grad_norm": 109.5, + "learning_rate": 4.60364561254769e-07, + "loss": 0.8178, + "mean_token_accuracy": 0.7673383951187134, + "num_tokens": 41455767.0, + "step": 1087 + }, + { + "epoch": 0.138404783106475, + "grad_norm": 12.6875, + "learning_rate": 4.607884696905468e-07, + "loss": 0.8078, + "mean_token_accuracy": 0.7748676538467407, + "num_tokens": 41497493.0, + "step": 1088 + }, + { + "epoch": 0.1385319933850655, + "grad_norm": 36.5, + "learning_rate": 4.612123781263247e-07, + "loss": 0.8208, + "mean_token_accuracy": 0.7716234922409058, + "num_tokens": 41542940.0, + "step": 1089 + }, + { + "epoch": 0.13865920366365603, + "grad_norm": 57.5, + "learning_rate": 4.616362865621026e-07, + "loss": 0.7326, + "mean_token_accuracy": 0.7952468395233154, + "num_tokens": 41579302.0, + "step": 1090 + }, + { + "epoch": 0.13878641394224653, + "grad_norm": 11.1875, + "learning_rate": 4.620601949978805e-07, + "loss": 0.845, + "mean_token_accuracy": 0.7649937272071838, + "num_tokens": 41617555.0, + "step": 1091 + }, + { + "epoch": 0.13891362422083706, + "grad_norm": 82.5, + "learning_rate": 4.6248410343365827e-07, + "loss": 0.911, + "mean_token_accuracy": 0.7420212030410767, + "num_tokens": 41654660.0, + "step": 1092 + }, + { + "epoch": 0.13904083449942756, + "grad_norm": 26.0, + "learning_rate": 4.6290801186943617e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7773160934448242, + "num_tokens": 41693773.0, + "step": 1093 + }, + { + "epoch": 0.13916804477801806, + "grad_norm": 18.75, + "learning_rate": 4.6333192030521407e-07, + "loss": 0.8037, + "mean_token_accuracy": 0.7768373489379883, + "num_tokens": 41728864.0, + "step": 1094 + }, + { + "epoch": 0.13929525505660859, + "grad_norm": 71.5, + "learning_rate": 4.6375582874099196e-07, + "loss": 0.7812, + "mean_token_accuracy": 0.7777161002159119, + "num_tokens": 41770596.0, + "step": 1095 + }, + { + "epoch": 0.1394224653351991, + "grad_norm": 10.4375, + "learning_rate": 4.6417973717676976e-07, + "loss": 0.7956, + "mean_token_accuracy": 0.7790030241012573, + "num_tokens": 41806585.0, + "step": 1096 + }, + { + "epoch": 0.1395496756137896, + "grad_norm": 10.375, + "learning_rate": 4.6460364561254766e-07, + "loss": 0.8447, + "mean_token_accuracy": 0.7618249654769897, + "num_tokens": 41845488.0, + "step": 1097 + }, + { + "epoch": 0.13967688589238011, + "grad_norm": 33.25, + "learning_rate": 4.6502755404832556e-07, + "loss": 0.7908, + "mean_token_accuracy": 0.7756761312484741, + "num_tokens": 41880674.0, + "step": 1098 + }, + { + "epoch": 0.13980409617097062, + "grad_norm": 49.75, + "learning_rate": 4.654514624841034e-07, + "loss": 0.8157, + "mean_token_accuracy": 0.7695437669754028, + "num_tokens": 41913020.0, + "step": 1099 + }, + { + "epoch": 0.13993130644956112, + "grad_norm": 81.0, + "learning_rate": 4.6587537091988125e-07, + "loss": 0.8164, + "mean_token_accuracy": 0.7731223106384277, + "num_tokens": 41948483.0, + "step": 1100 + }, + { + "epoch": 0.14005851672815164, + "grad_norm": 34.75, + "learning_rate": 4.6629927935565915e-07, + "loss": 0.7801, + "mean_token_accuracy": 0.7827013731002808, + "num_tokens": 41985659.0, + "step": 1101 + }, + { + "epoch": 0.14018572700674214, + "grad_norm": 84.5, + "learning_rate": 4.6672318779143705e-07, + "loss": 0.9088, + "mean_token_accuracy": 0.7502357363700867, + "num_tokens": 42027746.0, + "step": 1102 + }, + { + "epoch": 0.14031293728533265, + "grad_norm": 82.5, + "learning_rate": 4.671470962272149e-07, + "loss": 0.8626, + "mean_token_accuracy": 0.750427782535553, + "num_tokens": 42060903.0, + "step": 1103 + }, + { + "epoch": 0.14044014756392317, + "grad_norm": 15.5, + "learning_rate": 4.6757100466299274e-07, + "loss": 0.8214, + "mean_token_accuracy": 0.7705267071723938, + "num_tokens": 42102722.0, + "step": 1104 + }, + { + "epoch": 0.14056735784251367, + "grad_norm": 45.25, + "learning_rate": 4.6799491309877064e-07, + "loss": 0.8459, + "mean_token_accuracy": 0.7640585899353027, + "num_tokens": 42143418.0, + "step": 1105 + }, + { + "epoch": 0.14069456812110417, + "grad_norm": 76.5, + "learning_rate": 4.6841882153454854e-07, + "loss": 0.7961, + "mean_token_accuracy": 0.7728015184402466, + "num_tokens": 42177313.0, + "step": 1106 + }, + { + "epoch": 0.1408217783996947, + "grad_norm": 89.0, + "learning_rate": 4.688427299703264e-07, + "loss": 0.7548, + "mean_token_accuracy": 0.7821512222290039, + "num_tokens": 42214158.0, + "step": 1107 + }, + { + "epoch": 0.1409489886782852, + "grad_norm": 11.0, + "learning_rate": 4.6926663840610423e-07, + "loss": 0.8154, + "mean_token_accuracy": 0.7716019153594971, + "num_tokens": 42254189.0, + "step": 1108 + }, + { + "epoch": 0.1410761989568757, + "grad_norm": 84.0, + "learning_rate": 4.6969054684188213e-07, + "loss": 0.8473, + "mean_token_accuracy": 0.7573111653327942, + "num_tokens": 42296853.0, + "step": 1109 + }, + { + "epoch": 0.14120340923546623, + "grad_norm": 13.4375, + "learning_rate": 4.7011445527766003e-07, + "loss": 0.8262, + "mean_token_accuracy": 0.7627798914909363, + "num_tokens": 42331983.0, + "step": 1110 + }, + { + "epoch": 0.14133061951405673, + "grad_norm": 101.0, + "learning_rate": 4.7053836371343787e-07, + "loss": 0.8292, + "mean_token_accuracy": 0.7655702829360962, + "num_tokens": 42370496.0, + "step": 1111 + }, + { + "epoch": 0.14145782979264723, + "grad_norm": 12.125, + "learning_rate": 4.709622721492157e-07, + "loss": 0.8112, + "mean_token_accuracy": 0.771621584892273, + "num_tokens": 42406489.0, + "step": 1112 + }, + { + "epoch": 0.14158504007123776, + "grad_norm": 10.0625, + "learning_rate": 4.713861805849936e-07, + "loss": 0.8019, + "mean_token_accuracy": 0.7721503973007202, + "num_tokens": 42440679.0, + "step": 1113 + }, + { + "epoch": 0.14171225034982826, + "grad_norm": 11.25, + "learning_rate": 4.718100890207715e-07, + "loss": 0.8959, + "mean_token_accuracy": 0.7489395141601562, + "num_tokens": 42476699.0, + "step": 1114 + }, + { + "epoch": 0.1418394606284188, + "grad_norm": 25.0, + "learning_rate": 4.7223399745654936e-07, + "loss": 0.8487, + "mean_token_accuracy": 0.7621477246284485, + "num_tokens": 42513348.0, + "step": 1115 + }, + { + "epoch": 0.1419666709070093, + "grad_norm": 32.5, + "learning_rate": 4.726579058923272e-07, + "loss": 0.8229, + "mean_token_accuracy": 0.7719043493270874, + "num_tokens": 42550633.0, + "step": 1116 + }, + { + "epoch": 0.1420938811855998, + "grad_norm": 14.0625, + "learning_rate": 4.730818143281051e-07, + "loss": 0.8378, + "mean_token_accuracy": 0.7661926746368408, + "num_tokens": 42593649.0, + "step": 1117 + }, + { + "epoch": 0.14222109146419032, + "grad_norm": 20.625, + "learning_rate": 4.7350572276388295e-07, + "loss": 0.7812, + "mean_token_accuracy": 0.7788561582565308, + "num_tokens": 42632379.0, + "step": 1118 + }, + { + "epoch": 0.14234830174278082, + "grad_norm": 75.5, + "learning_rate": 4.7392963119966085e-07, + "loss": 0.8237, + "mean_token_accuracy": 0.7681070566177368, + "num_tokens": 42670319.0, + "step": 1119 + }, + { + "epoch": 0.14247551202137132, + "grad_norm": 79.5, + "learning_rate": 4.7435353963543875e-07, + "loss": 0.8091, + "mean_token_accuracy": 0.775574803352356, + "num_tokens": 42712680.0, + "step": 1120 + }, + { + "epoch": 0.14260272229996185, + "grad_norm": 96.0, + "learning_rate": 4.747774480712166e-07, + "loss": 0.8369, + "mean_token_accuracy": 0.7687269449234009, + "num_tokens": 42748241.0, + "step": 1121 + }, + { + "epoch": 0.14272993257855235, + "grad_norm": 102.5, + "learning_rate": 4.7520135650699444e-07, + "loss": 0.8695, + "mean_token_accuracy": 0.7589048147201538, + "num_tokens": 42783208.0, + "step": 1122 + }, + { + "epoch": 0.14285714285714285, + "grad_norm": 73.5, + "learning_rate": 4.7562526494277234e-07, + "loss": 0.8232, + "mean_token_accuracy": 0.7701683640480042, + "num_tokens": 42818570.0, + "step": 1123 + }, + { + "epoch": 0.14298435313573338, + "grad_norm": 47.5, + "learning_rate": 4.7604917337855024e-07, + "loss": 0.7828, + "mean_token_accuracy": 0.7808473706245422, + "num_tokens": 42852891.0, + "step": 1124 + }, + { + "epoch": 0.14311156341432388, + "grad_norm": 63.5, + "learning_rate": 4.764730818143281e-07, + "loss": 0.8209, + "mean_token_accuracy": 0.7624667882919312, + "num_tokens": 42887359.0, + "step": 1125 + }, + { + "epoch": 0.14323877369291438, + "grad_norm": 27.25, + "learning_rate": 4.768969902501059e-07, + "loss": 0.851, + "mean_token_accuracy": 0.7572923302650452, + "num_tokens": 42929308.0, + "step": 1126 + }, + { + "epoch": 0.1433659839715049, + "grad_norm": 28.625, + "learning_rate": 4.773208986858838e-07, + "loss": 0.7997, + "mean_token_accuracy": 0.7694792151451111, + "num_tokens": 42969373.0, + "step": 1127 + }, + { + "epoch": 0.1434931942500954, + "grad_norm": 14.75, + "learning_rate": 4.777448071216617e-07, + "loss": 0.8032, + "mean_token_accuracy": 0.7709816694259644, + "num_tokens": 43010133.0, + "step": 1128 + }, + { + "epoch": 0.1436204045286859, + "grad_norm": 112.5, + "learning_rate": 4.781687155574396e-07, + "loss": 0.8538, + "mean_token_accuracy": 0.7593593597412109, + "num_tokens": 43053743.0, + "step": 1129 + }, + { + "epoch": 0.14374761480727644, + "grad_norm": 54.5, + "learning_rate": 4.785926239932175e-07, + "loss": 0.7755, + "mean_token_accuracy": 0.7816210389137268, + "num_tokens": 43094125.0, + "step": 1130 + }, + { + "epoch": 0.14387482508586694, + "grad_norm": 13.75, + "learning_rate": 4.790165324289953e-07, + "loss": 0.8184, + "mean_token_accuracy": 0.7703747749328613, + "num_tokens": 43132399.0, + "step": 1131 + }, + { + "epoch": 0.14400203536445744, + "grad_norm": 81.5, + "learning_rate": 4.794404408647732e-07, + "loss": 0.8528, + "mean_token_accuracy": 0.7598587870597839, + "num_tokens": 43169036.0, + "step": 1132 + }, + { + "epoch": 0.14412924564304797, + "grad_norm": 174.0, + "learning_rate": 4.798643493005511e-07, + "loss": 0.7394, + "mean_token_accuracy": 0.7888901829719543, + "num_tokens": 43209397.0, + "step": 1133 + }, + { + "epoch": 0.14425645592163847, + "grad_norm": 10.0, + "learning_rate": 4.80288257736329e-07, + "loss": 0.8628, + "mean_token_accuracy": 0.7585143446922302, + "num_tokens": 43246788.0, + "step": 1134 + }, + { + "epoch": 0.14438366620022897, + "grad_norm": 77.5, + "learning_rate": 4.807121661721068e-07, + "loss": 0.7509, + "mean_token_accuracy": 0.785740077495575, + "num_tokens": 43282774.0, + "step": 1135 + }, + { + "epoch": 0.1445108764788195, + "grad_norm": 138.0, + "learning_rate": 4.811360746078847e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7675570249557495, + "num_tokens": 43322303.0, + "step": 1136 + }, + { + "epoch": 0.14463808675741, + "grad_norm": 72.5, + "learning_rate": 4.815599830436625e-07, + "loss": 0.8247, + "mean_token_accuracy": 0.7687124013900757, + "num_tokens": 43354611.0, + "step": 1137 + }, + { + "epoch": 0.1447652970360005, + "grad_norm": 90.0, + "learning_rate": 4.819838914794405e-07, + "loss": 0.7974, + "mean_token_accuracy": 0.7765135765075684, + "num_tokens": 43390640.0, + "step": 1138 + }, + { + "epoch": 0.14489250731459102, + "grad_norm": 38.5, + "learning_rate": 4.824077999152183e-07, + "loss": 0.8327, + "mean_token_accuracy": 0.7605582475662231, + "num_tokens": 43424513.0, + "step": 1139 + }, + { + "epoch": 0.14501971759318152, + "grad_norm": 18.25, + "learning_rate": 4.828317083509962e-07, + "loss": 0.8315, + "mean_token_accuracy": 0.7673456072807312, + "num_tokens": 43461699.0, + "step": 1140 + }, + { + "epoch": 0.14514692787177205, + "grad_norm": 88.5, + "learning_rate": 4.83255616786774e-07, + "loss": 0.8015, + "mean_token_accuracy": 0.780415415763855, + "num_tokens": 43494681.0, + "step": 1141 + }, + { + "epoch": 0.14527413815036255, + "grad_norm": 108.0, + "learning_rate": 4.83679525222552e-07, + "loss": 0.7718, + "mean_token_accuracy": 0.7825988531112671, + "num_tokens": 43531318.0, + "step": 1142 + }, + { + "epoch": 0.14540134842895305, + "grad_norm": 16.625, + "learning_rate": 4.841034336583298e-07, + "loss": 0.7758, + "mean_token_accuracy": 0.7824738025665283, + "num_tokens": 43569991.0, + "step": 1143 + }, + { + "epoch": 0.14552855870754358, + "grad_norm": 62.0, + "learning_rate": 4.845273420941076e-07, + "loss": 0.7575, + "mean_token_accuracy": 0.7866451740264893, + "num_tokens": 43609978.0, + "step": 1144 + }, + { + "epoch": 0.14565576898613408, + "grad_norm": 101.0, + "learning_rate": 4.849512505298855e-07, + "loss": 0.851, + "mean_token_accuracy": 0.7542632818222046, + "num_tokens": 43640419.0, + "step": 1145 + }, + { + "epoch": 0.14578297926472458, + "grad_norm": 49.25, + "learning_rate": 4.853751589656634e-07, + "loss": 0.7752, + "mean_token_accuracy": 0.7793938517570496, + "num_tokens": 43682408.0, + "step": 1146 + }, + { + "epoch": 0.1459101895433151, + "grad_norm": 27.875, + "learning_rate": 4.857990674014413e-07, + "loss": 0.817, + "mean_token_accuracy": 0.7694223523139954, + "num_tokens": 43726158.0, + "step": 1147 + }, + { + "epoch": 0.1460373998219056, + "grad_norm": 15.4375, + "learning_rate": 4.862229758372191e-07, + "loss": 0.8749, + "mean_token_accuracy": 0.7523428201675415, + "num_tokens": 43766466.0, + "step": 1148 + }, + { + "epoch": 0.1461646101004961, + "grad_norm": 29.5, + "learning_rate": 4.86646884272997e-07, + "loss": 0.8393, + "mean_token_accuracy": 0.7619951963424683, + "num_tokens": 43807255.0, + "step": 1149 + }, + { + "epoch": 0.14629182037908664, + "grad_norm": 9.0, + "learning_rate": 4.870707927087749e-07, + "loss": 0.8208, + "mean_token_accuracy": 0.7693445682525635, + "num_tokens": 43845419.0, + "step": 1150 + }, + { + "epoch": 0.14641903065767714, + "grad_norm": 19.25, + "learning_rate": 4.874947011445528e-07, + "loss": 0.8404, + "mean_token_accuracy": 0.7655578851699829, + "num_tokens": 43882942.0, + "step": 1151 + }, + { + "epoch": 0.14654624093626764, + "grad_norm": 40.5, + "learning_rate": 4.879186095803306e-07, + "loss": 0.7186, + "mean_token_accuracy": 0.794976532459259, + "num_tokens": 43923109.0, + "step": 1152 + }, + { + "epoch": 0.14667345121485817, + "grad_norm": 57.25, + "learning_rate": 4.883425180161085e-07, + "loss": 0.7724, + "mean_token_accuracy": 0.7829477190971375, + "num_tokens": 43959752.0, + "step": 1153 + }, + { + "epoch": 0.14680066149344867, + "grad_norm": 136.0, + "learning_rate": 4.887664264518864e-07, + "loss": 0.8526, + "mean_token_accuracy": 0.7658289670944214, + "num_tokens": 43996247.0, + "step": 1154 + }, + { + "epoch": 0.14692787177203917, + "grad_norm": 62.25, + "learning_rate": 4.891903348876643e-07, + "loss": 0.7929, + "mean_token_accuracy": 0.7797892689704895, + "num_tokens": 44035751.0, + "step": 1155 + }, + { + "epoch": 0.1470550820506297, + "grad_norm": 34.25, + "learning_rate": 4.896142433234421e-07, + "loss": 0.8323, + "mean_token_accuracy": 0.7700059413909912, + "num_tokens": 44073225.0, + "step": 1156 + }, + { + "epoch": 0.1471822923292202, + "grad_norm": 20.75, + "learning_rate": 4.9003815175922e-07, + "loss": 0.8596, + "mean_token_accuracy": 0.7653116583824158, + "num_tokens": 44111761.0, + "step": 1157 + }, + { + "epoch": 0.1473095026078107, + "grad_norm": 14.75, + "learning_rate": 4.904620601949979e-07, + "loss": 0.7823, + "mean_token_accuracy": 0.7771308422088623, + "num_tokens": 44151311.0, + "step": 1158 + }, + { + "epoch": 0.14743671288640123, + "grad_norm": 154.0, + "learning_rate": 4.908859686307758e-07, + "loss": 0.8759, + "mean_token_accuracy": 0.7584747076034546, + "num_tokens": 44184623.0, + "step": 1159 + }, + { + "epoch": 0.14756392316499173, + "grad_norm": 31.25, + "learning_rate": 4.913098770665536e-07, + "loss": 0.8241, + "mean_token_accuracy": 0.769841194152832, + "num_tokens": 44221753.0, + "step": 1160 + }, + { + "epoch": 0.14769113344358223, + "grad_norm": 77.0, + "learning_rate": 4.917337855023314e-07, + "loss": 0.7835, + "mean_token_accuracy": 0.7749861478805542, + "num_tokens": 44262542.0, + "step": 1161 + }, + { + "epoch": 0.14781834372217276, + "grad_norm": 55.5, + "learning_rate": 4.921576939381094e-07, + "loss": 0.8189, + "mean_token_accuracy": 0.7758574485778809, + "num_tokens": 44300294.0, + "step": 1162 + }, + { + "epoch": 0.14794555400076326, + "grad_norm": 14.0, + "learning_rate": 4.925816023738872e-07, + "loss": 0.871, + "mean_token_accuracy": 0.7572306990623474, + "num_tokens": 44338476.0, + "step": 1163 + }, + { + "epoch": 0.14807276427935376, + "grad_norm": 130.0, + "learning_rate": 4.930055108096651e-07, + "loss": 0.7807, + "mean_token_accuracy": 0.7798269987106323, + "num_tokens": 44367437.0, + "step": 1164 + }, + { + "epoch": 0.1481999745579443, + "grad_norm": 15.4375, + "learning_rate": 4.934294192454429e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7491576671600342, + "num_tokens": 44402636.0, + "step": 1165 + }, + { + "epoch": 0.1483271848365348, + "grad_norm": 39.25, + "learning_rate": 4.938533276812209e-07, + "loss": 0.8136, + "mean_token_accuracy": 0.7729188203811646, + "num_tokens": 44438187.0, + "step": 1166 + }, + { + "epoch": 0.14845439511512531, + "grad_norm": 93.0, + "learning_rate": 4.942772361169987e-07, + "loss": 0.7221, + "mean_token_accuracy": 0.7919821739196777, + "num_tokens": 44478345.0, + "step": 1167 + }, + { + "epoch": 0.14858160539371582, + "grad_norm": 79.0, + "learning_rate": 4.947011445527766e-07, + "loss": 0.8161, + "mean_token_accuracy": 0.772331953048706, + "num_tokens": 44520272.0, + "step": 1168 + }, + { + "epoch": 0.14870881567230632, + "grad_norm": 21.625, + "learning_rate": 4.951250529885544e-07, + "loss": 0.7318, + "mean_token_accuracy": 0.7945406436920166, + "num_tokens": 44561058.0, + "step": 1169 + }, + { + "epoch": 0.14883602595089684, + "grad_norm": 25.5, + "learning_rate": 4.955489614243324e-07, + "loss": 0.931, + "mean_token_accuracy": 0.7369917631149292, + "num_tokens": 44599371.0, + "step": 1170 + }, + { + "epoch": 0.14896323622948734, + "grad_norm": 7.65625, + "learning_rate": 4.959728698601102e-07, + "loss": 0.8029, + "mean_token_accuracy": 0.7647445201873779, + "num_tokens": 44640499.0, + "step": 1171 + }, + { + "epoch": 0.14909044650807785, + "grad_norm": 28.0, + "learning_rate": 4.963967782958881e-07, + "loss": 0.7817, + "mean_token_accuracy": 0.7849013805389404, + "num_tokens": 44683164.0, + "step": 1172 + }, + { + "epoch": 0.14921765678666837, + "grad_norm": 39.75, + "learning_rate": 4.968206867316659e-07, + "loss": 0.8354, + "mean_token_accuracy": 0.7654749155044556, + "num_tokens": 44721938.0, + "step": 1173 + }, + { + "epoch": 0.14934486706525887, + "grad_norm": 121.5, + "learning_rate": 4.972445951674439e-07, + "loss": 0.7863, + "mean_token_accuracy": 0.7727527618408203, + "num_tokens": 44753506.0, + "step": 1174 + }, + { + "epoch": 0.14947207734384937, + "grad_norm": 26.875, + "learning_rate": 4.976685036032216e-07, + "loss": 0.8817, + "mean_token_accuracy": 0.7544198036193848, + "num_tokens": 44792170.0, + "step": 1175 + }, + { + "epoch": 0.1495992876224399, + "grad_norm": 68.5, + "learning_rate": 4.980924120389996e-07, + "loss": 0.7215, + "mean_token_accuracy": 0.7963278889656067, + "num_tokens": 44830733.0, + "step": 1176 + }, + { + "epoch": 0.1497264979010304, + "grad_norm": 41.0, + "learning_rate": 4.985163204747774e-07, + "loss": 0.7601, + "mean_token_accuracy": 0.7886988520622253, + "num_tokens": 44869743.0, + "step": 1177 + }, + { + "epoch": 0.1498537081796209, + "grad_norm": 83.0, + "learning_rate": 4.989402289105554e-07, + "loss": 0.796, + "mean_token_accuracy": 0.7799376249313354, + "num_tokens": 44913358.0, + "step": 1178 + }, + { + "epoch": 0.14998091845821143, + "grad_norm": 12.0625, + "learning_rate": 4.993641373463331e-07, + "loss": 0.891, + "mean_token_accuracy": 0.7508729100227356, + "num_tokens": 44949963.0, + "step": 1179 + }, + { + "epoch": 0.15010812873680193, + "grad_norm": 90.5, + "learning_rate": 4.997880457821111e-07, + "loss": 0.83, + "mean_token_accuracy": 0.7650593519210815, + "num_tokens": 44988569.0, + "step": 1180 + }, + { + "epoch": 0.15023533901539243, + "grad_norm": 13.0625, + "learning_rate": 5.002119542178889e-07, + "loss": 0.7905, + "mean_token_accuracy": 0.7758833169937134, + "num_tokens": 45032634.0, + "step": 1181 + }, + { + "epoch": 0.15036254929398296, + "grad_norm": 13.3125, + "learning_rate": 5.006358626536667e-07, + "loss": 0.7551, + "mean_token_accuracy": 0.7855898141860962, + "num_tokens": 45069685.0, + "step": 1182 + }, + { + "epoch": 0.15048975957257346, + "grad_norm": 22.875, + "learning_rate": 5.010597710894446e-07, + "loss": 0.7847, + "mean_token_accuracy": 0.7787957787513733, + "num_tokens": 45107481.0, + "step": 1183 + }, + { + "epoch": 0.15061696985116396, + "grad_norm": 123.0, + "learning_rate": 5.014836795252225e-07, + "loss": 0.7615, + "mean_token_accuracy": 0.7830599546432495, + "num_tokens": 45145747.0, + "step": 1184 + }, + { + "epoch": 0.1507441801297545, + "grad_norm": 9.9375, + "learning_rate": 5.019075879610004e-07, + "loss": 0.7703, + "mean_token_accuracy": 0.7852265238761902, + "num_tokens": 45183140.0, + "step": 1185 + }, + { + "epoch": 0.150871390408345, + "grad_norm": 16.375, + "learning_rate": 5.023314963967783e-07, + "loss": 0.7919, + "mean_token_accuracy": 0.7720017433166504, + "num_tokens": 45224911.0, + "step": 1186 + }, + { + "epoch": 0.1509986006869355, + "grad_norm": 41.5, + "learning_rate": 5.027554048325562e-07, + "loss": 0.886, + "mean_token_accuracy": 0.7524060010910034, + "num_tokens": 45257388.0, + "step": 1187 + }, + { + "epoch": 0.15112581096552602, + "grad_norm": 10.875, + "learning_rate": 5.03179313268334e-07, + "loss": 0.8551, + "mean_token_accuracy": 0.7608559131622314, + "num_tokens": 45297876.0, + "step": 1188 + }, + { + "epoch": 0.15125302124411652, + "grad_norm": 116.5, + "learning_rate": 5.036032217041119e-07, + "loss": 0.8375, + "mean_token_accuracy": 0.7613928914070129, + "num_tokens": 45334040.0, + "step": 1189 + }, + { + "epoch": 0.15138023152270705, + "grad_norm": 10.0, + "learning_rate": 5.040271301398897e-07, + "loss": 0.8218, + "mean_token_accuracy": 0.7687890529632568, + "num_tokens": 45375985.0, + "step": 1190 + }, + { + "epoch": 0.15150744180129755, + "grad_norm": 26.75, + "learning_rate": 5.044510385756676e-07, + "loss": 0.772, + "mean_token_accuracy": 0.7791816592216492, + "num_tokens": 45419951.0, + "step": 1191 + }, + { + "epoch": 0.15163465207988805, + "grad_norm": 8.5, + "learning_rate": 5.048749470114455e-07, + "loss": 0.799, + "mean_token_accuracy": 0.7747311592102051, + "num_tokens": 45456988.0, + "step": 1192 + }, + { + "epoch": 0.15176186235847858, + "grad_norm": 39.25, + "learning_rate": 5.052988554472234e-07, + "loss": 0.8201, + "mean_token_accuracy": 0.7644770741462708, + "num_tokens": 45488226.0, + "step": 1193 + }, + { + "epoch": 0.15188907263706908, + "grad_norm": 21.25, + "learning_rate": 5.057227638830013e-07, + "loss": 0.8266, + "mean_token_accuracy": 0.7706223726272583, + "num_tokens": 45524661.0, + "step": 1194 + }, + { + "epoch": 0.15201628291565958, + "grad_norm": 43.75, + "learning_rate": 5.061466723187792e-07, + "loss": 0.7959, + "mean_token_accuracy": 0.7744131088256836, + "num_tokens": 45563491.0, + "step": 1195 + }, + { + "epoch": 0.1521434931942501, + "grad_norm": 17.125, + "learning_rate": 5.065705807545569e-07, + "loss": 0.7787, + "mean_token_accuracy": 0.7808395624160767, + "num_tokens": 45596771.0, + "step": 1196 + }, + { + "epoch": 0.1522707034728406, + "grad_norm": 51.25, + "learning_rate": 5.069944891903349e-07, + "loss": 0.8127, + "mean_token_accuracy": 0.770937442779541, + "num_tokens": 45640049.0, + "step": 1197 + }, + { + "epoch": 0.1523979137514311, + "grad_norm": 74.0, + "learning_rate": 5.074183976261127e-07, + "loss": 0.807, + "mean_token_accuracy": 0.7717009782791138, + "num_tokens": 45677375.0, + "step": 1198 + }, + { + "epoch": 0.15252512403002164, + "grad_norm": 79.5, + "learning_rate": 5.078423060618906e-07, + "loss": 0.792, + "mean_token_accuracy": 0.7774966955184937, + "num_tokens": 45719253.0, + "step": 1199 + }, + { + "epoch": 0.15265233430861214, + "grad_norm": 102.0, + "learning_rate": 5.082662144976685e-07, + "loss": 0.8225, + "mean_token_accuracy": 0.7629646062850952, + "num_tokens": 45756065.0, + "step": 1200 + }, + { + "epoch": 0.15277954458720264, + "grad_norm": 33.5, + "learning_rate": 5.086901229334464e-07, + "loss": 0.79, + "mean_token_accuracy": 0.7785752415657043, + "num_tokens": 45791707.0, + "step": 1201 + }, + { + "epoch": 0.15290675486579317, + "grad_norm": 147.0, + "learning_rate": 5.091140313692243e-07, + "loss": 0.8113, + "mean_token_accuracy": 0.770818829536438, + "num_tokens": 45828464.0, + "step": 1202 + }, + { + "epoch": 0.15303396514438367, + "grad_norm": 51.5, + "learning_rate": 5.095379398050022e-07, + "loss": 0.7761, + "mean_token_accuracy": 0.7771042585372925, + "num_tokens": 45867046.0, + "step": 1203 + }, + { + "epoch": 0.15316117542297417, + "grad_norm": 46.5, + "learning_rate": 5.099618482407799e-07, + "loss": 0.8193, + "mean_token_accuracy": 0.7698879241943359, + "num_tokens": 45900530.0, + "step": 1204 + }, + { + "epoch": 0.1532883857015647, + "grad_norm": 10.9375, + "learning_rate": 5.103857566765578e-07, + "loss": 0.807, + "mean_token_accuracy": 0.7758629322052002, + "num_tokens": 45940002.0, + "step": 1205 + }, + { + "epoch": 0.1534155959801552, + "grad_norm": 46.25, + "learning_rate": 5.108096651123357e-07, + "loss": 0.7613, + "mean_token_accuracy": 0.7836956977844238, + "num_tokens": 45975299.0, + "step": 1206 + }, + { + "epoch": 0.1535428062587457, + "grad_norm": 17.75, + "learning_rate": 5.112335735481135e-07, + "loss": 0.7876, + "mean_token_accuracy": 0.773439884185791, + "num_tokens": 46012693.0, + "step": 1207 + }, + { + "epoch": 0.15367001653733622, + "grad_norm": 19.0, + "learning_rate": 5.116574819838915e-07, + "loss": 0.6592, + "mean_token_accuracy": 0.8132128119468689, + "num_tokens": 46051558.0, + "step": 1208 + }, + { + "epoch": 0.15379722681592672, + "grad_norm": 47.0, + "learning_rate": 5.120813904196693e-07, + "loss": 0.7702, + "mean_token_accuracy": 0.7834689617156982, + "num_tokens": 46087173.0, + "step": 1209 + }, + { + "epoch": 0.15392443709451722, + "grad_norm": 70.5, + "learning_rate": 5.125052988554473e-07, + "loss": 0.8376, + "mean_token_accuracy": 0.7655586004257202, + "num_tokens": 46120482.0, + "step": 1210 + }, + { + "epoch": 0.15405164737310775, + "grad_norm": 26.625, + "learning_rate": 5.12929207291225e-07, + "loss": 0.7902, + "mean_token_accuracy": 0.7782875895500183, + "num_tokens": 46154478.0, + "step": 1211 + }, + { + "epoch": 0.15417885765169825, + "grad_norm": 14.375, + "learning_rate": 5.133531157270029e-07, + "loss": 0.7526, + "mean_token_accuracy": 0.786421537399292, + "num_tokens": 46195012.0, + "step": 1212 + }, + { + "epoch": 0.15430606793028875, + "grad_norm": 12.1875, + "learning_rate": 5.137770241627808e-07, + "loss": 0.7084, + "mean_token_accuracy": 0.7980983257293701, + "num_tokens": 46230641.0, + "step": 1213 + }, + { + "epoch": 0.15443327820887928, + "grad_norm": 11.3125, + "learning_rate": 5.142009325985587e-07, + "loss": 0.7611, + "mean_token_accuracy": 0.7874014377593994, + "num_tokens": 46268993.0, + "step": 1214 + }, + { + "epoch": 0.15456048848746978, + "grad_norm": 38.25, + "learning_rate": 5.146248410343365e-07, + "loss": 0.8029, + "mean_token_accuracy": 0.7748708724975586, + "num_tokens": 46309310.0, + "step": 1215 + }, + { + "epoch": 0.1546876987660603, + "grad_norm": 75.5, + "learning_rate": 5.150487494701145e-07, + "loss": 0.7195, + "mean_token_accuracy": 0.7949815988540649, + "num_tokens": 46352415.0, + "step": 1216 + }, + { + "epoch": 0.1548149090446508, + "grad_norm": 10.1875, + "learning_rate": 5.154726579058923e-07, + "loss": 0.8083, + "mean_token_accuracy": 0.7726728320121765, + "num_tokens": 46387838.0, + "step": 1217 + }, + { + "epoch": 0.1549421193232413, + "grad_norm": 31.25, + "learning_rate": 5.158965663416703e-07, + "loss": 0.7855, + "mean_token_accuracy": 0.7784601449966431, + "num_tokens": 46432669.0, + "step": 1218 + }, + { + "epoch": 0.15506932960183184, + "grad_norm": 28.5, + "learning_rate": 5.16320474777448e-07, + "loss": 0.7285, + "mean_token_accuracy": 0.7925667762756348, + "num_tokens": 46474383.0, + "step": 1219 + }, + { + "epoch": 0.15519653988042234, + "grad_norm": 15.5, + "learning_rate": 5.167443832132259e-07, + "loss": 0.794, + "mean_token_accuracy": 0.7723982334136963, + "num_tokens": 46514263.0, + "step": 1220 + }, + { + "epoch": 0.15532375015901284, + "grad_norm": 17.625, + "learning_rate": 5.171682916490038e-07, + "loss": 0.7463, + "mean_token_accuracy": 0.7866382598876953, + "num_tokens": 46549835.0, + "step": 1221 + }, + { + "epoch": 0.15545096043760337, + "grad_norm": 58.5, + "learning_rate": 5.175922000847816e-07, + "loss": 0.7503, + "mean_token_accuracy": 0.7858086824417114, + "num_tokens": 46584811.0, + "step": 1222 + }, + { + "epoch": 0.15557817071619387, + "grad_norm": 139.0, + "learning_rate": 5.180161085205595e-07, + "loss": 0.7688, + "mean_token_accuracy": 0.7834901809692383, + "num_tokens": 46621441.0, + "step": 1223 + }, + { + "epoch": 0.15570538099478437, + "grad_norm": 114.0, + "learning_rate": 5.184400169563374e-07, + "loss": 0.8368, + "mean_token_accuracy": 0.7644096612930298, + "num_tokens": 46661572.0, + "step": 1224 + }, + { + "epoch": 0.1558325912733749, + "grad_norm": 20.125, + "learning_rate": 5.188639253921153e-07, + "loss": 0.8085, + "mean_token_accuracy": 0.7709974050521851, + "num_tokens": 46702157.0, + "step": 1225 + }, + { + "epoch": 0.1559598015519654, + "grad_norm": 16.5, + "learning_rate": 5.192878338278932e-07, + "loss": 0.791, + "mean_token_accuracy": 0.7775964736938477, + "num_tokens": 46739119.0, + "step": 1226 + }, + { + "epoch": 0.1560870118305559, + "grad_norm": 19.125, + "learning_rate": 5.19711742263671e-07, + "loss": 0.8487, + "mean_token_accuracy": 0.7664370536804199, + "num_tokens": 46783990.0, + "step": 1227 + }, + { + "epoch": 0.15621422210914643, + "grad_norm": 77.5, + "learning_rate": 5.201356506994488e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.778203547000885, + "num_tokens": 46816131.0, + "step": 1228 + }, + { + "epoch": 0.15634143238773693, + "grad_norm": 40.75, + "learning_rate": 5.205595591352268e-07, + "loss": 0.8712, + "mean_token_accuracy": 0.7541468143463135, + "num_tokens": 46856453.0, + "step": 1229 + }, + { + "epoch": 0.15646864266632743, + "grad_norm": 37.0, + "learning_rate": 5.209834675710046e-07, + "loss": 0.7848, + "mean_token_accuracy": 0.7788751125335693, + "num_tokens": 46897961.0, + "step": 1230 + }, + { + "epoch": 0.15659585294491796, + "grad_norm": 15.5, + "learning_rate": 5.214073760067825e-07, + "loss": 0.8996, + "mean_token_accuracy": 0.7459200620651245, + "num_tokens": 46931961.0, + "step": 1231 + }, + { + "epoch": 0.15672306322350846, + "grad_norm": 88.5, + "learning_rate": 5.218312844425604e-07, + "loss": 0.8026, + "mean_token_accuracy": 0.7789483070373535, + "num_tokens": 46967822.0, + "step": 1232 + }, + { + "epoch": 0.15685027350209896, + "grad_norm": 19.75, + "learning_rate": 5.222551928783383e-07, + "loss": 0.7822, + "mean_token_accuracy": 0.7770800590515137, + "num_tokens": 47004511.0, + "step": 1233 + }, + { + "epoch": 0.1569774837806895, + "grad_norm": 28.25, + "learning_rate": 5.226791013141161e-07, + "loss": 0.8223, + "mean_token_accuracy": 0.7692650556564331, + "num_tokens": 47047244.0, + "step": 1234 + }, + { + "epoch": 0.15710469405928, + "grad_norm": 40.0, + "learning_rate": 5.23103009749894e-07, + "loss": 0.7649, + "mean_token_accuracy": 0.7820175290107727, + "num_tokens": 47088206.0, + "step": 1235 + }, + { + "epoch": 0.1572319043378705, + "grad_norm": 105.0, + "learning_rate": 5.235269181856718e-07, + "loss": 0.7945, + "mean_token_accuracy": 0.774429202079773, + "num_tokens": 47132733.0, + "step": 1236 + }, + { + "epoch": 0.15735911461646102, + "grad_norm": 28.25, + "learning_rate": 5.239508266214498e-07, + "loss": 0.8795, + "mean_token_accuracy": 0.7497740983963013, + "num_tokens": 47166840.0, + "step": 1237 + }, + { + "epoch": 0.15748632489505152, + "grad_norm": 47.75, + "learning_rate": 5.243747350572276e-07, + "loss": 0.8257, + "mean_token_accuracy": 0.7683917880058289, + "num_tokens": 47203444.0, + "step": 1238 + }, + { + "epoch": 0.15761353517364202, + "grad_norm": 51.75, + "learning_rate": 5.247986434930056e-07, + "loss": 0.821, + "mean_token_accuracy": 0.770139753818512, + "num_tokens": 47240380.0, + "step": 1239 + }, + { + "epoch": 0.15774074545223254, + "grad_norm": 46.75, + "learning_rate": 5.252225519287834e-07, + "loss": 0.8929, + "mean_token_accuracy": 0.7526832818984985, + "num_tokens": 47277886.0, + "step": 1240 + }, + { + "epoch": 0.15786795573082305, + "grad_norm": 12.1875, + "learning_rate": 5.256464603645613e-07, + "loss": 0.8, + "mean_token_accuracy": 0.777558445930481, + "num_tokens": 47315261.0, + "step": 1241 + }, + { + "epoch": 0.15799516600941357, + "grad_norm": 20.625, + "learning_rate": 5.260703688003391e-07, + "loss": 0.7685, + "mean_token_accuracy": 0.7822760343551636, + "num_tokens": 47357489.0, + "step": 1242 + }, + { + "epoch": 0.15812237628800407, + "grad_norm": 52.5, + "learning_rate": 5.26494277236117e-07, + "loss": 0.8363, + "mean_token_accuracy": 0.7653598189353943, + "num_tokens": 47389851.0, + "step": 1243 + }, + { + "epoch": 0.15824958656659457, + "grad_norm": 55.5, + "learning_rate": 5.269181856718948e-07, + "loss": 0.8384, + "mean_token_accuracy": 0.7612234950065613, + "num_tokens": 47427491.0, + "step": 1244 + }, + { + "epoch": 0.1583767968451851, + "grad_norm": 15.25, + "learning_rate": 5.273420941076727e-07, + "loss": 0.7692, + "mean_token_accuracy": 0.7870583534240723, + "num_tokens": 47462790.0, + "step": 1245 + }, + { + "epoch": 0.1585040071237756, + "grad_norm": 62.5, + "learning_rate": 5.277660025434506e-07, + "loss": 0.7027, + "mean_token_accuracy": 0.7999815940856934, + "num_tokens": 47502240.0, + "step": 1246 + }, + { + "epoch": 0.1586312174023661, + "grad_norm": 117.0, + "learning_rate": 5.281899109792285e-07, + "loss": 0.744, + "mean_token_accuracy": 0.788805365562439, + "num_tokens": 47542493.0, + "step": 1247 + }, + { + "epoch": 0.15875842768095663, + "grad_norm": 18.75, + "learning_rate": 5.286138194150064e-07, + "loss": 0.7834, + "mean_token_accuracy": 0.779268741607666, + "num_tokens": 47583378.0, + "step": 1248 + }, + { + "epoch": 0.15888563795954713, + "grad_norm": 48.75, + "learning_rate": 5.290377278507841e-07, + "loss": 0.82, + "mean_token_accuracy": 0.7700883150100708, + "num_tokens": 47622671.0, + "step": 1249 + }, + { + "epoch": 0.15901284823813763, + "grad_norm": 30.875, + "learning_rate": 5.294616362865621e-07, + "loss": 0.7704, + "mean_token_accuracy": 0.7864152193069458, + "num_tokens": 47663774.0, + "step": 1250 + }, + { + "epoch": 0.15914005851672816, + "grad_norm": 9.6875, + "learning_rate": 5.298855447223399e-07, + "loss": 0.8413, + "mean_token_accuracy": 0.7718521952629089, + "num_tokens": 47706143.0, + "step": 1251 + }, + { + "epoch": 0.15926726879531866, + "grad_norm": 63.5, + "learning_rate": 5.303094531581178e-07, + "loss": 0.837, + "mean_token_accuracy": 0.7629681825637817, + "num_tokens": 47741197.0, + "step": 1252 + }, + { + "epoch": 0.15939447907390916, + "grad_norm": 32.25, + "learning_rate": 5.307333615938957e-07, + "loss": 0.7331, + "mean_token_accuracy": 0.7894754409790039, + "num_tokens": 47779306.0, + "step": 1253 + }, + { + "epoch": 0.1595216893524997, + "grad_norm": 104.0, + "learning_rate": 5.311572700296736e-07, + "loss": 0.8585, + "mean_token_accuracy": 0.7583924531936646, + "num_tokens": 47811056.0, + "step": 1254 + }, + { + "epoch": 0.1596488996310902, + "grad_norm": 110.5, + "learning_rate": 5.315811784654515e-07, + "loss": 0.7289, + "mean_token_accuracy": 0.7908950448036194, + "num_tokens": 47847075.0, + "step": 1255 + }, + { + "epoch": 0.1597761099096807, + "grad_norm": 76.0, + "learning_rate": 5.320050869012294e-07, + "loss": 0.774, + "mean_token_accuracy": 0.7813632488250732, + "num_tokens": 47892639.0, + "step": 1256 + }, + { + "epoch": 0.15990332018827122, + "grad_norm": 45.5, + "learning_rate": 5.324289953370071e-07, + "loss": 0.8521, + "mean_token_accuracy": 0.7625322341918945, + "num_tokens": 47933522.0, + "step": 1257 + }, + { + "epoch": 0.16003053046686172, + "grad_norm": 7.8125, + "learning_rate": 5.328529037727851e-07, + "loss": 0.7744, + "mean_token_accuracy": 0.7815691232681274, + "num_tokens": 47977307.0, + "step": 1258 + }, + { + "epoch": 0.16015774074545222, + "grad_norm": 48.5, + "learning_rate": 5.332768122085629e-07, + "loss": 0.785, + "mean_token_accuracy": 0.7772420644760132, + "num_tokens": 48015129.0, + "step": 1259 + }, + { + "epoch": 0.16028495102404275, + "grad_norm": 156.0, + "learning_rate": 5.337007206443408e-07, + "loss": 0.8411, + "mean_token_accuracy": 0.7653079628944397, + "num_tokens": 48051904.0, + "step": 1260 + }, + { + "epoch": 0.16041216130263325, + "grad_norm": 38.0, + "learning_rate": 5.341246290801187e-07, + "loss": 0.7686, + "mean_token_accuracy": 0.7848362922668457, + "num_tokens": 48094001.0, + "step": 1261 + }, + { + "epoch": 0.16053937158122375, + "grad_norm": 11.9375, + "learning_rate": 5.345485375158966e-07, + "loss": 0.7613, + "mean_token_accuracy": 0.7841235399246216, + "num_tokens": 48134124.0, + "step": 1262 + }, + { + "epoch": 0.16066658185981428, + "grad_norm": 66.5, + "learning_rate": 5.349724459516745e-07, + "loss": 0.7639, + "mean_token_accuracy": 0.7818149328231812, + "num_tokens": 48170730.0, + "step": 1263 + }, + { + "epoch": 0.16079379213840478, + "grad_norm": 11.5625, + "learning_rate": 5.353963543874522e-07, + "loss": 0.8232, + "mean_token_accuracy": 0.7712564468383789, + "num_tokens": 48212640.0, + "step": 1264 + }, + { + "epoch": 0.1609210024169953, + "grad_norm": 35.5, + "learning_rate": 5.358202628232301e-07, + "loss": 0.8269, + "mean_token_accuracy": 0.7677581906318665, + "num_tokens": 48243136.0, + "step": 1265 + }, + { + "epoch": 0.1610482126955858, + "grad_norm": 26.125, + "learning_rate": 5.36244171259008e-07, + "loss": 0.7841, + "mean_token_accuracy": 0.7758883237838745, + "num_tokens": 48282905.0, + "step": 1266 + }, + { + "epoch": 0.1611754229741763, + "grad_norm": 25.625, + "learning_rate": 5.366680796947859e-07, + "loss": 0.8173, + "mean_token_accuracy": 0.7718914151191711, + "num_tokens": 48323325.0, + "step": 1267 + }, + { + "epoch": 0.16130263325276684, + "grad_norm": 20.125, + "learning_rate": 5.370919881305637e-07, + "loss": 0.8008, + "mean_token_accuracy": 0.7747999429702759, + "num_tokens": 48358400.0, + "step": 1268 + }, + { + "epoch": 0.16142984353135734, + "grad_norm": 14.9375, + "learning_rate": 5.375158965663417e-07, + "loss": 0.8226, + "mean_token_accuracy": 0.7727319598197937, + "num_tokens": 48393018.0, + "step": 1269 + }, + { + "epoch": 0.16155705380994784, + "grad_norm": 84.0, + "learning_rate": 5.379398050021195e-07, + "loss": 0.7874, + "mean_token_accuracy": 0.7782719135284424, + "num_tokens": 48432952.0, + "step": 1270 + }, + { + "epoch": 0.16168426408853837, + "grad_norm": 102.5, + "learning_rate": 5.383637134378975e-07, + "loss": 0.8047, + "mean_token_accuracy": 0.7735826373100281, + "num_tokens": 48466913.0, + "step": 1271 + }, + { + "epoch": 0.16181147436712887, + "grad_norm": 82.0, + "learning_rate": 5.387876218736752e-07, + "loss": 0.8662, + "mean_token_accuracy": 0.7545212507247925, + "num_tokens": 48512658.0, + "step": 1272 + }, + { + "epoch": 0.16193868464571937, + "grad_norm": 138.0, + "learning_rate": 5.392115303094531e-07, + "loss": 0.7498, + "mean_token_accuracy": 0.78708815574646, + "num_tokens": 48552283.0, + "step": 1273 + }, + { + "epoch": 0.1620658949243099, + "grad_norm": 25.0, + "learning_rate": 5.39635438745231e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.794685423374176, + "num_tokens": 48590504.0, + "step": 1274 + }, + { + "epoch": 0.1621931052029004, + "grad_norm": 58.25, + "learning_rate": 5.400593471810089e-07, + "loss": 0.8269, + "mean_token_accuracy": 0.7647992968559265, + "num_tokens": 48629993.0, + "step": 1275 + }, + { + "epoch": 0.1623203154814909, + "grad_norm": 14.1875, + "learning_rate": 5.404832556167867e-07, + "loss": 0.8435, + "mean_token_accuracy": 0.759945273399353, + "num_tokens": 48665074.0, + "step": 1276 + }, + { + "epoch": 0.16244752576008142, + "grad_norm": 26.0, + "learning_rate": 5.409071640525647e-07, + "loss": 0.7771, + "mean_token_accuracy": 0.7834875583648682, + "num_tokens": 48705524.0, + "step": 1277 + }, + { + "epoch": 0.16257473603867192, + "grad_norm": 54.75, + "learning_rate": 5.413310724883425e-07, + "loss": 0.8285, + "mean_token_accuracy": 0.7601487040519714, + "num_tokens": 48744030.0, + "step": 1278 + }, + { + "epoch": 0.16270194631726242, + "grad_norm": 12.5, + "learning_rate": 5.417549809241205e-07, + "loss": 0.8358, + "mean_token_accuracy": 0.768997311592102, + "num_tokens": 48788401.0, + "step": 1279 + }, + { + "epoch": 0.16282915659585295, + "grad_norm": 47.75, + "learning_rate": 5.421788893598982e-07, + "loss": 0.8509, + "mean_token_accuracy": 0.7561968564987183, + "num_tokens": 48823046.0, + "step": 1280 + }, + { + "epoch": 0.16295636687444345, + "grad_norm": 26.875, + "learning_rate": 5.42602797795676e-07, + "loss": 0.8322, + "mean_token_accuracy": 0.7645535469055176, + "num_tokens": 48868293.0, + "step": 1281 + }, + { + "epoch": 0.16308357715303395, + "grad_norm": 19.125, + "learning_rate": 5.43026706231454e-07, + "loss": 0.859, + "mean_token_accuracy": 0.7570993900299072, + "num_tokens": 48908775.0, + "step": 1282 + }, + { + "epoch": 0.16321078743162448, + "grad_norm": 49.0, + "learning_rate": 5.434506146672319e-07, + "loss": 0.8405, + "mean_token_accuracy": 0.7644924521446228, + "num_tokens": 48957640.0, + "step": 1283 + }, + { + "epoch": 0.16333799771021498, + "grad_norm": 9.5, + "learning_rate": 5.438745231030097e-07, + "loss": 0.8302, + "mean_token_accuracy": 0.7621530294418335, + "num_tokens": 48993601.0, + "step": 1284 + }, + { + "epoch": 0.16346520798880548, + "grad_norm": 26.875, + "learning_rate": 5.442984315387876e-07, + "loss": 0.7435, + "mean_token_accuracy": 0.786496639251709, + "num_tokens": 49028411.0, + "step": 1285 + }, + { + "epoch": 0.163592418267396, + "grad_norm": 104.0, + "learning_rate": 5.447223399745655e-07, + "loss": 0.7736, + "mean_token_accuracy": 0.7822524905204773, + "num_tokens": 49067449.0, + "step": 1286 + }, + { + "epoch": 0.1637196285459865, + "grad_norm": 111.0, + "learning_rate": 5.451462484103433e-07, + "loss": 0.7865, + "mean_token_accuracy": 0.7784304618835449, + "num_tokens": 49102650.0, + "step": 1287 + }, + { + "epoch": 0.163846838824577, + "grad_norm": 51.25, + "learning_rate": 5.455701568461212e-07, + "loss": 0.8171, + "mean_token_accuracy": 0.7683775424957275, + "num_tokens": 49137110.0, + "step": 1288 + }, + { + "epoch": 0.16397404910316754, + "grad_norm": 22.0, + "learning_rate": 5.45994065281899e-07, + "loss": 0.8788, + "mean_token_accuracy": 0.7509435415267944, + "num_tokens": 49177765.0, + "step": 1289 + }, + { + "epoch": 0.16410125938175804, + "grad_norm": 76.0, + "learning_rate": 5.46417973717677e-07, + "loss": 0.8604, + "mean_token_accuracy": 0.7579634785652161, + "num_tokens": 49207860.0, + "step": 1290 + }, + { + "epoch": 0.16422846966034857, + "grad_norm": 136.0, + "learning_rate": 5.468418821534548e-07, + "loss": 0.7479, + "mean_token_accuracy": 0.7906169295310974, + "num_tokens": 49239994.0, + "step": 1291 + }, + { + "epoch": 0.16435567993893907, + "grad_norm": 22.875, + "learning_rate": 5.472657905892327e-07, + "loss": 0.7592, + "mean_token_accuracy": 0.787712574005127, + "num_tokens": 49273464.0, + "step": 1292 + }, + { + "epoch": 0.16448289021752957, + "grad_norm": 84.5, + "learning_rate": 5.476896990250106e-07, + "loss": 0.9025, + "mean_token_accuracy": 0.7497819662094116, + "num_tokens": 49313144.0, + "step": 1293 + }, + { + "epoch": 0.1646101004961201, + "grad_norm": 13.875, + "learning_rate": 5.481136074607885e-07, + "loss": 0.754, + "mean_token_accuracy": 0.7846440076828003, + "num_tokens": 49351392.0, + "step": 1294 + }, + { + "epoch": 0.1647373107747106, + "grad_norm": 24.875, + "learning_rate": 5.485375158965663e-07, + "loss": 0.7411, + "mean_token_accuracy": 0.7875313758850098, + "num_tokens": 49396726.0, + "step": 1295 + }, + { + "epoch": 0.1648645210533011, + "grad_norm": 31.875, + "learning_rate": 5.489614243323442e-07, + "loss": 0.7922, + "mean_token_accuracy": 0.7711893320083618, + "num_tokens": 49435856.0, + "step": 1296 + }, + { + "epoch": 0.16499173133189163, + "grad_norm": 14.6875, + "learning_rate": 5.49385332768122e-07, + "loss": 0.7208, + "mean_token_accuracy": 0.7941016554832458, + "num_tokens": 49473695.0, + "step": 1297 + }, + { + "epoch": 0.16511894161048213, + "grad_norm": 55.0, + "learning_rate": 5.498092412039e-07, + "loss": 0.8901, + "mean_token_accuracy": 0.7569321393966675, + "num_tokens": 49509165.0, + "step": 1298 + }, + { + "epoch": 0.16524615188907263, + "grad_norm": 145.0, + "learning_rate": 5.502331496396778e-07, + "loss": 0.7659, + "mean_token_accuracy": 0.7832422256469727, + "num_tokens": 49553790.0, + "step": 1299 + }, + { + "epoch": 0.16537336216766316, + "grad_norm": 23.125, + "learning_rate": 5.506570580754557e-07, + "loss": 0.7899, + "mean_token_accuracy": 0.7743261456489563, + "num_tokens": 49593124.0, + "step": 1300 + }, + { + "epoch": 0.16550057244625366, + "grad_norm": 35.25, + "learning_rate": 5.510809665112336e-07, + "loss": 0.7685, + "mean_token_accuracy": 0.7837100028991699, + "num_tokens": 49632016.0, + "step": 1301 + }, + { + "epoch": 0.16562778272484416, + "grad_norm": 15.875, + "learning_rate": 5.515048749470113e-07, + "loss": 0.7824, + "mean_token_accuracy": 0.7797056436538696, + "num_tokens": 49670689.0, + "step": 1302 + }, + { + "epoch": 0.1657549930034347, + "grad_norm": 22.625, + "learning_rate": 5.519287833827893e-07, + "loss": 0.7823, + "mean_token_accuracy": 0.7804017066955566, + "num_tokens": 49709230.0, + "step": 1303 + }, + { + "epoch": 0.1658822032820252, + "grad_norm": 32.75, + "learning_rate": 5.523526918185671e-07, + "loss": 0.8462, + "mean_token_accuracy": 0.7571401000022888, + "num_tokens": 49741648.0, + "step": 1304 + }, + { + "epoch": 0.1660094135606157, + "grad_norm": 128.0, + "learning_rate": 5.52776600254345e-07, + "loss": 0.7594, + "mean_token_accuracy": 0.7789242267608643, + "num_tokens": 49773906.0, + "step": 1305 + }, + { + "epoch": 0.16613662383920622, + "grad_norm": 30.0, + "learning_rate": 5.532005086901229e-07, + "loss": 0.8028, + "mean_token_accuracy": 0.7752504348754883, + "num_tokens": 49812645.0, + "step": 1306 + }, + { + "epoch": 0.16626383411779672, + "grad_norm": 21.5, + "learning_rate": 5.536244171259008e-07, + "loss": 0.7641, + "mean_token_accuracy": 0.7822138071060181, + "num_tokens": 49854500.0, + "step": 1307 + }, + { + "epoch": 0.16639104439638722, + "grad_norm": 53.25, + "learning_rate": 5.540483255616786e-07, + "loss": 0.8924, + "mean_token_accuracy": 0.749962329864502, + "num_tokens": 49889408.0, + "step": 1308 + }, + { + "epoch": 0.16651825467497774, + "grad_norm": 124.5, + "learning_rate": 5.544722339974566e-07, + "loss": 0.8136, + "mean_token_accuracy": 0.7717872858047485, + "num_tokens": 49923581.0, + "step": 1309 + }, + { + "epoch": 0.16664546495356825, + "grad_norm": 46.0, + "learning_rate": 5.548961424332343e-07, + "loss": 0.7956, + "mean_token_accuracy": 0.7701689600944519, + "num_tokens": 49959433.0, + "step": 1310 + }, + { + "epoch": 0.16677267523215875, + "grad_norm": 32.75, + "learning_rate": 5.553200508690123e-07, + "loss": 0.7343, + "mean_token_accuracy": 0.7938594818115234, + "num_tokens": 50000307.0, + "step": 1311 + }, + { + "epoch": 0.16689988551074927, + "grad_norm": 131.0, + "learning_rate": 5.557439593047901e-07, + "loss": 0.8707, + "mean_token_accuracy": 0.7544475197792053, + "num_tokens": 50035587.0, + "step": 1312 + }, + { + "epoch": 0.16702709578933977, + "grad_norm": 19.625, + "learning_rate": 5.56167867740568e-07, + "loss": 0.7794, + "mean_token_accuracy": 0.7796443104743958, + "num_tokens": 50079425.0, + "step": 1313 + }, + { + "epoch": 0.16715430606793028, + "grad_norm": 19.0, + "learning_rate": 5.565917761763459e-07, + "loss": 0.7844, + "mean_token_accuracy": 0.7761663794517517, + "num_tokens": 50110674.0, + "step": 1314 + }, + { + "epoch": 0.1672815163465208, + "grad_norm": 14.3125, + "learning_rate": 5.570156846121238e-07, + "loss": 0.805, + "mean_token_accuracy": 0.7710082530975342, + "num_tokens": 50149033.0, + "step": 1315 + }, + { + "epoch": 0.1674087266251113, + "grad_norm": 24.875, + "learning_rate": 5.574395930479016e-07, + "loss": 0.8346, + "mean_token_accuracy": 0.7642990350723267, + "num_tokens": 50185728.0, + "step": 1316 + }, + { + "epoch": 0.16753593690370183, + "grad_norm": 10.25, + "learning_rate": 5.578635014836796e-07, + "loss": 0.7604, + "mean_token_accuracy": 0.7840211987495422, + "num_tokens": 50219671.0, + "step": 1317 + }, + { + "epoch": 0.16766314718229233, + "grad_norm": 84.5, + "learning_rate": 5.582874099194573e-07, + "loss": 0.8887, + "mean_token_accuracy": 0.7528625130653381, + "num_tokens": 50251989.0, + "step": 1318 + }, + { + "epoch": 0.16779035746088283, + "grad_norm": 19.625, + "learning_rate": 5.587113183552353e-07, + "loss": 0.8345, + "mean_token_accuracy": 0.7677448391914368, + "num_tokens": 50291079.0, + "step": 1319 + }, + { + "epoch": 0.16791756773947336, + "grad_norm": 110.0, + "learning_rate": 5.591352267910131e-07, + "loss": 0.7604, + "mean_token_accuracy": 0.784144401550293, + "num_tokens": 50332262.0, + "step": 1320 + }, + { + "epoch": 0.16804477801806386, + "grad_norm": 20.125, + "learning_rate": 5.59559135226791e-07, + "loss": 0.8043, + "mean_token_accuracy": 0.7756916880607605, + "num_tokens": 50371278.0, + "step": 1321 + }, + { + "epoch": 0.16817198829665436, + "grad_norm": 25.125, + "learning_rate": 5.599830436625689e-07, + "loss": 0.794, + "mean_token_accuracy": 0.7770780324935913, + "num_tokens": 50409463.0, + "step": 1322 + }, + { + "epoch": 0.1682991985752449, + "grad_norm": 139.0, + "learning_rate": 5.604069520983468e-07, + "loss": 0.7799, + "mean_token_accuracy": 0.7801753878593445, + "num_tokens": 50446745.0, + "step": 1323 + }, + { + "epoch": 0.1684264088538354, + "grad_norm": 143.0, + "learning_rate": 5.608308605341246e-07, + "loss": 0.777, + "mean_token_accuracy": 0.7774036526679993, + "num_tokens": 50479725.0, + "step": 1324 + }, + { + "epoch": 0.1685536191324259, + "grad_norm": 70.5, + "learning_rate": 5.612547689699024e-07, + "loss": 0.8527, + "mean_token_accuracy": 0.760860800743103, + "num_tokens": 50514734.0, + "step": 1325 + }, + { + "epoch": 0.16868082941101642, + "grad_norm": 70.0, + "learning_rate": 5.616786774056803e-07, + "loss": 0.8102, + "mean_token_accuracy": 0.7749536037445068, + "num_tokens": 50555165.0, + "step": 1326 + }, + { + "epoch": 0.16880803968960692, + "grad_norm": 16.625, + "learning_rate": 5.621025858414582e-07, + "loss": 0.8109, + "mean_token_accuracy": 0.7715675830841064, + "num_tokens": 50598251.0, + "step": 1327 + }, + { + "epoch": 0.16893524996819742, + "grad_norm": 96.0, + "learning_rate": 5.625264942772361e-07, + "loss": 0.8125, + "mean_token_accuracy": 0.7748578786849976, + "num_tokens": 50638423.0, + "step": 1328 + }, + { + "epoch": 0.16906246024678795, + "grad_norm": 25.75, + "learning_rate": 5.629504027130139e-07, + "loss": 0.8759, + "mean_token_accuracy": 0.7627424001693726, + "num_tokens": 50678226.0, + "step": 1329 + }, + { + "epoch": 0.16918967052537845, + "grad_norm": 24.125, + "learning_rate": 5.633743111487919e-07, + "loss": 0.872, + "mean_token_accuracy": 0.7538034319877625, + "num_tokens": 50717542.0, + "step": 1330 + }, + { + "epoch": 0.16931688080396895, + "grad_norm": 14.0625, + "learning_rate": 5.637982195845697e-07, + "loss": 0.7323, + "mean_token_accuracy": 0.7890002131462097, + "num_tokens": 50752914.0, + "step": 1331 + }, + { + "epoch": 0.16944409108255948, + "grad_norm": 22.0, + "learning_rate": 5.642221280203476e-07, + "loss": 0.7163, + "mean_token_accuracy": 0.7980459928512573, + "num_tokens": 50789276.0, + "step": 1332 + }, + { + "epoch": 0.16957130136114998, + "grad_norm": 56.25, + "learning_rate": 5.646460364561254e-07, + "loss": 0.774, + "mean_token_accuracy": 0.7814759016036987, + "num_tokens": 50820759.0, + "step": 1333 + }, + { + "epoch": 0.16969851163974048, + "grad_norm": 27.5, + "learning_rate": 5.650699448919033e-07, + "loss": 0.8487, + "mean_token_accuracy": 0.7604143619537354, + "num_tokens": 50863241.0, + "step": 1334 + }, + { + "epoch": 0.169825721918331, + "grad_norm": 129.0, + "learning_rate": 5.654938533276812e-07, + "loss": 0.8015, + "mean_token_accuracy": 0.772645115852356, + "num_tokens": 50904296.0, + "step": 1335 + }, + { + "epoch": 0.1699529321969215, + "grad_norm": 197.0, + "learning_rate": 5.659177617634591e-07, + "loss": 0.7805, + "mean_token_accuracy": 0.7799448370933533, + "num_tokens": 50942180.0, + "step": 1336 + }, + { + "epoch": 0.170080142475512, + "grad_norm": 156.0, + "learning_rate": 5.663416701992369e-07, + "loss": 0.808, + "mean_token_accuracy": 0.7705078721046448, + "num_tokens": 50980287.0, + "step": 1337 + }, + { + "epoch": 0.17020735275410254, + "grad_norm": 198.0, + "learning_rate": 5.667655786350149e-07, + "loss": 0.9109, + "mean_token_accuracy": 0.7346218824386597, + "num_tokens": 51014218.0, + "step": 1338 + }, + { + "epoch": 0.17033456303269304, + "grad_norm": 189.0, + "learning_rate": 5.671894870707927e-07, + "loss": 0.7787, + "mean_token_accuracy": 0.7837748527526855, + "num_tokens": 51055915.0, + "step": 1339 + }, + { + "epoch": 0.17046177331128357, + "grad_norm": 93.5, + "learning_rate": 5.676133955065705e-07, + "loss": 0.7131, + "mean_token_accuracy": 0.7923004627227783, + "num_tokens": 51090684.0, + "step": 1340 + }, + { + "epoch": 0.17058898358987407, + "grad_norm": 57.75, + "learning_rate": 5.680373039423484e-07, + "loss": 0.7736, + "mean_token_accuracy": 0.7825711369514465, + "num_tokens": 51130156.0, + "step": 1341 + }, + { + "epoch": 0.17071619386846457, + "grad_norm": 11.1875, + "learning_rate": 5.684612123781263e-07, + "loss": 0.8139, + "mean_token_accuracy": 0.7674604654312134, + "num_tokens": 51171282.0, + "step": 1342 + }, + { + "epoch": 0.1708434041470551, + "grad_norm": 118.5, + "learning_rate": 5.688851208139042e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7580453157424927, + "num_tokens": 51210616.0, + "step": 1343 + }, + { + "epoch": 0.1709706144256456, + "grad_norm": 106.0, + "learning_rate": 5.69309029249682e-07, + "loss": 0.8164, + "mean_token_accuracy": 0.7721354961395264, + "num_tokens": 51253357.0, + "step": 1344 + }, + { + "epoch": 0.1710978247042361, + "grad_norm": 49.5, + "learning_rate": 5.697329376854599e-07, + "loss": 0.7884, + "mean_token_accuracy": 0.775575578212738, + "num_tokens": 51292671.0, + "step": 1345 + }, + { + "epoch": 0.17122503498282662, + "grad_norm": 177.0, + "learning_rate": 5.701568461212378e-07, + "loss": 0.7341, + "mean_token_accuracy": 0.7935409545898438, + "num_tokens": 51329145.0, + "step": 1346 + }, + { + "epoch": 0.17135224526141712, + "grad_norm": 52.5, + "learning_rate": 5.705807545570157e-07, + "loss": 0.726, + "mean_token_accuracy": 0.7918878197669983, + "num_tokens": 51371265.0, + "step": 1347 + }, + { + "epoch": 0.17147945554000762, + "grad_norm": 31.0, + "learning_rate": 5.710046629927934e-07, + "loss": 0.7853, + "mean_token_accuracy": 0.7793645858764648, + "num_tokens": 51410262.0, + "step": 1348 + }, + { + "epoch": 0.17160666581859815, + "grad_norm": 41.0, + "learning_rate": 5.714285714285714e-07, + "loss": 0.7477, + "mean_token_accuracy": 0.7866546511650085, + "num_tokens": 51447796.0, + "step": 1349 + }, + { + "epoch": 0.17173387609718865, + "grad_norm": 60.25, + "learning_rate": 5.718524798643492e-07, + "loss": 0.8313, + "mean_token_accuracy": 0.7663658857345581, + "num_tokens": 51482677.0, + "step": 1350 + }, + { + "epoch": 0.17186108637577915, + "grad_norm": 33.75, + "learning_rate": 5.722763883001272e-07, + "loss": 0.8054, + "mean_token_accuracy": 0.7741867303848267, + "num_tokens": 51524111.0, + "step": 1351 + }, + { + "epoch": 0.17198829665436968, + "grad_norm": 19.375, + "learning_rate": 5.72700296735905e-07, + "loss": 0.7872, + "mean_token_accuracy": 0.7751563191413879, + "num_tokens": 51561008.0, + "step": 1352 + }, + { + "epoch": 0.17211550693296018, + "grad_norm": 78.0, + "learning_rate": 5.731242051716829e-07, + "loss": 0.8481, + "mean_token_accuracy": 0.7581399083137512, + "num_tokens": 51603091.0, + "step": 1353 + }, + { + "epoch": 0.17224271721155068, + "grad_norm": 64.0, + "learning_rate": 5.735481136074608e-07, + "loss": 0.7722, + "mean_token_accuracy": 0.7790061831474304, + "num_tokens": 51637789.0, + "step": 1354 + }, + { + "epoch": 0.1723699274901412, + "grad_norm": 24.625, + "learning_rate": 5.739720220432386e-07, + "loss": 0.7622, + "mean_token_accuracy": 0.7810423970222473, + "num_tokens": 51674207.0, + "step": 1355 + }, + { + "epoch": 0.1724971377687317, + "grad_norm": 40.5, + "learning_rate": 5.743959304790164e-07, + "loss": 0.8619, + "mean_token_accuracy": 0.761000394821167, + "num_tokens": 51715109.0, + "step": 1356 + }, + { + "epoch": 0.1726243480473222, + "grad_norm": 27.375, + "learning_rate": 5.748198389147944e-07, + "loss": 0.8142, + "mean_token_accuracy": 0.7694227695465088, + "num_tokens": 51753373.0, + "step": 1357 + }, + { + "epoch": 0.17275155832591274, + "grad_norm": 7.4375, + "learning_rate": 5.752437473505722e-07, + "loss": 0.8424, + "mean_token_accuracy": 0.7669243812561035, + "num_tokens": 51795100.0, + "step": 1358 + }, + { + "epoch": 0.17287876860450324, + "grad_norm": 35.0, + "learning_rate": 5.756676557863502e-07, + "loss": 0.8248, + "mean_token_accuracy": 0.7674682140350342, + "num_tokens": 51831840.0, + "step": 1359 + }, + { + "epoch": 0.17300597888309374, + "grad_norm": 17.0, + "learning_rate": 5.76091564222128e-07, + "loss": 0.7666, + "mean_token_accuracy": 0.7850795984268188, + "num_tokens": 51870227.0, + "step": 1360 + }, + { + "epoch": 0.17313318916168427, + "grad_norm": 77.5, + "learning_rate": 5.765154726579059e-07, + "loss": 0.8013, + "mean_token_accuracy": 0.7733078002929688, + "num_tokens": 51908251.0, + "step": 1361 + }, + { + "epoch": 0.17326039944027477, + "grad_norm": 64.5, + "learning_rate": 5.769393810936838e-07, + "loss": 0.7953, + "mean_token_accuracy": 0.7743208408355713, + "num_tokens": 51944531.0, + "step": 1362 + }, + { + "epoch": 0.17338760971886527, + "grad_norm": 14.625, + "learning_rate": 5.773632895294616e-07, + "loss": 0.8554, + "mean_token_accuracy": 0.7574245929718018, + "num_tokens": 51985742.0, + "step": 1363 + }, + { + "epoch": 0.1735148199974558, + "grad_norm": 175.0, + "learning_rate": 5.777871979652394e-07, + "loss": 0.7963, + "mean_token_accuracy": 0.7743295431137085, + "num_tokens": 52022036.0, + "step": 1364 + }, + { + "epoch": 0.1736420302760463, + "grad_norm": 61.75, + "learning_rate": 5.782111064010173e-07, + "loss": 0.788, + "mean_token_accuracy": 0.7747423052787781, + "num_tokens": 52066077.0, + "step": 1365 + }, + { + "epoch": 0.17376924055463683, + "grad_norm": 73.0, + "learning_rate": 5.786350148367952e-07, + "loss": 0.7811, + "mean_token_accuracy": 0.7771843671798706, + "num_tokens": 52111280.0, + "step": 1366 + }, + { + "epoch": 0.17389645083322733, + "grad_norm": 10.5625, + "learning_rate": 5.790589232725731e-07, + "loss": 0.7807, + "mean_token_accuracy": 0.776366114616394, + "num_tokens": 52153148.0, + "step": 1367 + }, + { + "epoch": 0.17402366111181783, + "grad_norm": 23.5, + "learning_rate": 5.79482831708351e-07, + "loss": 0.7663, + "mean_token_accuracy": 0.780297040939331, + "num_tokens": 52186722.0, + "step": 1368 + }, + { + "epoch": 0.17415087139040836, + "grad_norm": 21.375, + "learning_rate": 5.799067401441288e-07, + "loss": 0.7876, + "mean_token_accuracy": 0.7739752531051636, + "num_tokens": 52220837.0, + "step": 1369 + }, + { + "epoch": 0.17427808166899886, + "grad_norm": 65.5, + "learning_rate": 5.803306485799068e-07, + "loss": 0.762, + "mean_token_accuracy": 0.7879193425178528, + "num_tokens": 52257049.0, + "step": 1370 + }, + { + "epoch": 0.17440529194758936, + "grad_norm": 72.5, + "learning_rate": 5.807545570156845e-07, + "loss": 0.7361, + "mean_token_accuracy": 0.7919876575469971, + "num_tokens": 52297632.0, + "step": 1371 + }, + { + "epoch": 0.1745325022261799, + "grad_norm": 17.625, + "learning_rate": 5.811784654514624e-07, + "loss": 0.8813, + "mean_token_accuracy": 0.753835141658783, + "num_tokens": 52331008.0, + "step": 1372 + }, + { + "epoch": 0.1746597125047704, + "grad_norm": 16.625, + "learning_rate": 5.816023738872403e-07, + "loss": 0.7738, + "mean_token_accuracy": 0.7805546522140503, + "num_tokens": 52369420.0, + "step": 1373 + }, + { + "epoch": 0.1747869227833609, + "grad_norm": 132.0, + "learning_rate": 5.820262823230182e-07, + "loss": 0.814, + "mean_token_accuracy": 0.771697998046875, + "num_tokens": 52411759.0, + "step": 1374 + }, + { + "epoch": 0.17491413306195142, + "grad_norm": 29.875, + "learning_rate": 5.824501907587961e-07, + "loss": 0.7716, + "mean_token_accuracy": 0.7819116115570068, + "num_tokens": 52446459.0, + "step": 1375 + }, + { + "epoch": 0.17504134334054192, + "grad_norm": 18.25, + "learning_rate": 5.82874099194574e-07, + "loss": 0.8714, + "mean_token_accuracy": 0.7561931014060974, + "num_tokens": 52484615.0, + "step": 1376 + }, + { + "epoch": 0.17516855361913242, + "grad_norm": 21.875, + "learning_rate": 5.832980076303518e-07, + "loss": 0.8753, + "mean_token_accuracy": 0.7526453733444214, + "num_tokens": 52525195.0, + "step": 1377 + }, + { + "epoch": 0.17529576389772294, + "grad_norm": 16.125, + "learning_rate": 5.837219160661297e-07, + "loss": 0.7392, + "mean_token_accuracy": 0.7921206951141357, + "num_tokens": 52558783.0, + "step": 1378 + }, + { + "epoch": 0.17542297417631345, + "grad_norm": 183.0, + "learning_rate": 5.841458245019075e-07, + "loss": 0.7356, + "mean_token_accuracy": 0.7910772562026978, + "num_tokens": 52599176.0, + "step": 1379 + }, + { + "epoch": 0.17555018445490395, + "grad_norm": 185.0, + "learning_rate": 5.845697329376855e-07, + "loss": 0.8103, + "mean_token_accuracy": 0.771133542060852, + "num_tokens": 52630865.0, + "step": 1380 + }, + { + "epoch": 0.17567739473349447, + "grad_norm": 139.0, + "learning_rate": 5.849936413734633e-07, + "loss": 0.8651, + "mean_token_accuracy": 0.756167471408844, + "num_tokens": 52671702.0, + "step": 1381 + }, + { + "epoch": 0.17580460501208497, + "grad_norm": 112.0, + "learning_rate": 5.854175498092412e-07, + "loss": 0.7502, + "mean_token_accuracy": 0.7888933420181274, + "num_tokens": 52712288.0, + "step": 1382 + }, + { + "epoch": 0.17593181529067548, + "grad_norm": 186.0, + "learning_rate": 5.858414582450191e-07, + "loss": 0.8019, + "mean_token_accuracy": 0.7766013145446777, + "num_tokens": 52756569.0, + "step": 1383 + }, + { + "epoch": 0.176059025569266, + "grad_norm": 222.0, + "learning_rate": 5.86265366680797e-07, + "loss": 0.8434, + "mean_token_accuracy": 0.7641083002090454, + "num_tokens": 52796195.0, + "step": 1384 + }, + { + "epoch": 0.1761862358478565, + "grad_norm": 117.5, + "learning_rate": 5.866892751165748e-07, + "loss": 0.7883, + "mean_token_accuracy": 0.778772234916687, + "num_tokens": 52839687.0, + "step": 1385 + }, + { + "epoch": 0.176313446126447, + "grad_norm": 214.0, + "learning_rate": 5.871131835523526e-07, + "loss": 0.8644, + "mean_token_accuracy": 0.760186493396759, + "num_tokens": 52875660.0, + "step": 1386 + }, + { + "epoch": 0.17644065640503753, + "grad_norm": 73.5, + "learning_rate": 5.875370919881305e-07, + "loss": 0.7908, + "mean_token_accuracy": 0.7716031074523926, + "num_tokens": 52921220.0, + "step": 1387 + }, + { + "epoch": 0.17656786668362803, + "grad_norm": 39.0, + "learning_rate": 5.879610004239084e-07, + "loss": 0.8634, + "mean_token_accuracy": 0.7578738927841187, + "num_tokens": 52960042.0, + "step": 1388 + }, + { + "epoch": 0.17669507696221853, + "grad_norm": 46.5, + "learning_rate": 5.883849088596863e-07, + "loss": 0.8265, + "mean_token_accuracy": 0.7654210925102234, + "num_tokens": 53002817.0, + "step": 1389 + }, + { + "epoch": 0.17682228724080906, + "grad_norm": 49.5, + "learning_rate": 5.888088172954641e-07, + "loss": 0.8332, + "mean_token_accuracy": 0.7670255899429321, + "num_tokens": 53037751.0, + "step": 1390 + }, + { + "epoch": 0.17694949751939956, + "grad_norm": 180.0, + "learning_rate": 5.892327257312421e-07, + "loss": 0.7157, + "mean_token_accuracy": 0.7963962554931641, + "num_tokens": 53074286.0, + "step": 1391 + }, + { + "epoch": 0.1770767077979901, + "grad_norm": 38.0, + "learning_rate": 5.896566341670199e-07, + "loss": 0.7306, + "mean_token_accuracy": 0.7865879535675049, + "num_tokens": 53113525.0, + "step": 1392 + }, + { + "epoch": 0.1772039180765806, + "grad_norm": 127.5, + "learning_rate": 5.900805426027977e-07, + "loss": 0.7918, + "mean_token_accuracy": 0.7752424478530884, + "num_tokens": 53150137.0, + "step": 1393 + }, + { + "epoch": 0.1773311283551711, + "grad_norm": 36.0, + "learning_rate": 5.905044510385756e-07, + "loss": 0.7238, + "mean_token_accuracy": 0.7930609583854675, + "num_tokens": 53189519.0, + "step": 1394 + }, + { + "epoch": 0.17745833863376162, + "grad_norm": 117.0, + "learning_rate": 5.909283594743535e-07, + "loss": 0.8772, + "mean_token_accuracy": 0.7556522488594055, + "num_tokens": 53227452.0, + "step": 1395 + }, + { + "epoch": 0.17758554891235212, + "grad_norm": 26.5, + "learning_rate": 5.913522679101314e-07, + "loss": 0.8015, + "mean_token_accuracy": 0.7718983888626099, + "num_tokens": 53265875.0, + "step": 1396 + }, + { + "epoch": 0.17771275919094262, + "grad_norm": 26.25, + "learning_rate": 5.917761763459093e-07, + "loss": 0.8628, + "mean_token_accuracy": 0.7576776742935181, + "num_tokens": 53305477.0, + "step": 1397 + }, + { + "epoch": 0.17783996946953315, + "grad_norm": 105.0, + "learning_rate": 5.922000847816871e-07, + "loss": 0.7624, + "mean_token_accuracy": 0.7837153673171997, + "num_tokens": 53342800.0, + "step": 1398 + }, + { + "epoch": 0.17796717974812365, + "grad_norm": 34.5, + "learning_rate": 5.926239932174651e-07, + "loss": 0.8213, + "mean_token_accuracy": 0.7686011791229248, + "num_tokens": 53387476.0, + "step": 1399 + }, + { + "epoch": 0.17809439002671415, + "grad_norm": 15.8125, + "learning_rate": 5.930479016532429e-07, + "loss": 0.7815, + "mean_token_accuracy": 0.7746401429176331, + "num_tokens": 53424628.0, + "step": 1400 + }, + { + "epoch": 0.17822160030530468, + "grad_norm": 34.0, + "learning_rate": 5.934718100890207e-07, + "loss": 0.7722, + "mean_token_accuracy": 0.7801501750946045, + "num_tokens": 53464459.0, + "step": 1401 + }, + { + "epoch": 0.17834881058389518, + "grad_norm": 77.5, + "learning_rate": 5.938957185247986e-07, + "loss": 0.8145, + "mean_token_accuracy": 0.7665627002716064, + "num_tokens": 53501721.0, + "step": 1402 + }, + { + "epoch": 0.17847602086248568, + "grad_norm": 91.0, + "learning_rate": 5.943196269605765e-07, + "loss": 0.7796, + "mean_token_accuracy": 0.7747991681098938, + "num_tokens": 53536607.0, + "step": 1403 + }, + { + "epoch": 0.1786032311410762, + "grad_norm": 15.4375, + "learning_rate": 5.947435353963544e-07, + "loss": 0.8413, + "mean_token_accuracy": 0.7646650075912476, + "num_tokens": 53569809.0, + "step": 1404 + }, + { + "epoch": 0.1787304414196667, + "grad_norm": 19.25, + "learning_rate": 5.951674438321323e-07, + "loss": 0.8038, + "mean_token_accuracy": 0.7724686861038208, + "num_tokens": 53604423.0, + "step": 1405 + }, + { + "epoch": 0.1788576516982572, + "grad_norm": 81.5, + "learning_rate": 5.955913522679101e-07, + "loss": 0.8041, + "mean_token_accuracy": 0.7692185044288635, + "num_tokens": 53642145.0, + "step": 1406 + }, + { + "epoch": 0.17898486197684774, + "grad_norm": 79.5, + "learning_rate": 5.96015260703688e-07, + "loss": 0.8918, + "mean_token_accuracy": 0.7469756603240967, + "num_tokens": 53682998.0, + "step": 1407 + }, + { + "epoch": 0.17911207225543824, + "grad_norm": 16.125, + "learning_rate": 5.964391691394659e-07, + "loss": 0.7926, + "mean_token_accuracy": 0.7773564457893372, + "num_tokens": 53729764.0, + "step": 1408 + }, + { + "epoch": 0.17923928253402874, + "grad_norm": 49.25, + "learning_rate": 5.968630775752436e-07, + "loss": 0.8595, + "mean_token_accuracy": 0.7586232423782349, + "num_tokens": 53764522.0, + "step": 1409 + }, + { + "epoch": 0.17936649281261927, + "grad_norm": 9.25, + "learning_rate": 5.972869860110216e-07, + "loss": 0.8438, + "mean_token_accuracy": 0.758564293384552, + "num_tokens": 53799791.0, + "step": 1410 + }, + { + "epoch": 0.17949370309120977, + "grad_norm": 55.25, + "learning_rate": 5.977108944467994e-07, + "loss": 0.8626, + "mean_token_accuracy": 0.7583714127540588, + "num_tokens": 53840181.0, + "step": 1411 + }, + { + "epoch": 0.17962091336980027, + "grad_norm": 12.1875, + "learning_rate": 5.981348028825774e-07, + "loss": 0.7566, + "mean_token_accuracy": 0.7880436778068542, + "num_tokens": 53882813.0, + "step": 1412 + }, + { + "epoch": 0.1797481236483908, + "grad_norm": 31.375, + "learning_rate": 5.985587113183552e-07, + "loss": 0.817, + "mean_token_accuracy": 0.7719779014587402, + "num_tokens": 53922198.0, + "step": 1413 + }, + { + "epoch": 0.1798753339269813, + "grad_norm": 105.5, + "learning_rate": 5.989826197541331e-07, + "loss": 0.8253, + "mean_token_accuracy": 0.7724511623382568, + "num_tokens": 53959390.0, + "step": 1414 + }, + { + "epoch": 0.18000254420557182, + "grad_norm": 10.6875, + "learning_rate": 5.99406528189911e-07, + "loss": 0.8247, + "mean_token_accuracy": 0.7726479768753052, + "num_tokens": 54005707.0, + "step": 1415 + }, + { + "epoch": 0.18012975448416232, + "grad_norm": 98.5, + "learning_rate": 5.998304366256888e-07, + "loss": 0.7947, + "mean_token_accuracy": 0.7734664678573608, + "num_tokens": 54047808.0, + "step": 1416 + }, + { + "epoch": 0.18025696476275282, + "grad_norm": 16.5, + "learning_rate": 6.002543450614666e-07, + "loss": 0.7145, + "mean_token_accuracy": 0.7965540885925293, + "num_tokens": 54082929.0, + "step": 1417 + }, + { + "epoch": 0.18038417504134335, + "grad_norm": 23.25, + "learning_rate": 6.006782534972446e-07, + "loss": 0.7858, + "mean_token_accuracy": 0.776648998260498, + "num_tokens": 54120199.0, + "step": 1418 + }, + { + "epoch": 0.18051138531993385, + "grad_norm": 17.375, + "learning_rate": 6.011021619330224e-07, + "loss": 0.822, + "mean_token_accuracy": 0.7663172483444214, + "num_tokens": 54158629.0, + "step": 1419 + }, + { + "epoch": 0.18063859559852435, + "grad_norm": 64.5, + "learning_rate": 6.015260703688004e-07, + "loss": 0.8053, + "mean_token_accuracy": 0.7677383422851562, + "num_tokens": 54194772.0, + "step": 1420 + }, + { + "epoch": 0.18076580587711488, + "grad_norm": 53.5, + "learning_rate": 6.019499788045782e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.7678322792053223, + "num_tokens": 54229825.0, + "step": 1421 + }, + { + "epoch": 0.18089301615570538, + "grad_norm": 16.875, + "learning_rate": 6.023738872403561e-07, + "loss": 0.7898, + "mean_token_accuracy": 0.7780787944793701, + "num_tokens": 54260051.0, + "step": 1422 + }, + { + "epoch": 0.18102022643429588, + "grad_norm": 81.5, + "learning_rate": 6.02797795676134e-07, + "loss": 0.7896, + "mean_token_accuracy": 0.778040885925293, + "num_tokens": 54297865.0, + "step": 1423 + }, + { + "epoch": 0.1811474367128864, + "grad_norm": 110.0, + "learning_rate": 6.032217041119118e-07, + "loss": 0.822, + "mean_token_accuracy": 0.7632995247840881, + "num_tokens": 54334618.0, + "step": 1424 + }, + { + "epoch": 0.1812746469914769, + "grad_norm": 36.25, + "learning_rate": 6.036456125476896e-07, + "loss": 0.8763, + "mean_token_accuracy": 0.753918468952179, + "num_tokens": 54369761.0, + "step": 1425 + }, + { + "epoch": 0.1814018572700674, + "grad_norm": 31.5, + "learning_rate": 6.040695209834675e-07, + "loss": 0.7314, + "mean_token_accuracy": 0.7913562059402466, + "num_tokens": 54411549.0, + "step": 1426 + }, + { + "epoch": 0.18152906754865794, + "grad_norm": 58.75, + "learning_rate": 6.044934294192454e-07, + "loss": 0.8401, + "mean_token_accuracy": 0.7646322846412659, + "num_tokens": 54449137.0, + "step": 1427 + }, + { + "epoch": 0.18165627782724844, + "grad_norm": 133.0, + "learning_rate": 6.049173378550233e-07, + "loss": 0.7871, + "mean_token_accuracy": 0.7797147035598755, + "num_tokens": 54494604.0, + "step": 1428 + }, + { + "epoch": 0.18178348810583894, + "grad_norm": 57.75, + "learning_rate": 6.053412462908012e-07, + "loss": 0.7569, + "mean_token_accuracy": 0.7824462652206421, + "num_tokens": 54533533.0, + "step": 1429 + }, + { + "epoch": 0.18191069838442947, + "grad_norm": 38.5, + "learning_rate": 6.05765154726579e-07, + "loss": 0.8213, + "mean_token_accuracy": 0.7653980255126953, + "num_tokens": 54575211.0, + "step": 1430 + }, + { + "epoch": 0.18203790866301997, + "grad_norm": 139.0, + "learning_rate": 6.061890631623569e-07, + "loss": 0.7256, + "mean_token_accuracy": 0.7907825112342834, + "num_tokens": 54618308.0, + "step": 1431 + }, + { + "epoch": 0.18216511894161047, + "grad_norm": 106.5, + "learning_rate": 6.066129715981347e-07, + "loss": 0.7589, + "mean_token_accuracy": 0.783473014831543, + "num_tokens": 54660260.0, + "step": 1432 + }, + { + "epoch": 0.182292329220201, + "grad_norm": 49.75, + "learning_rate": 6.070368800339126e-07, + "loss": 0.7568, + "mean_token_accuracy": 0.7828882932662964, + "num_tokens": 54692056.0, + "step": 1433 + }, + { + "epoch": 0.1824195394987915, + "grad_norm": 130.0, + "learning_rate": 6.074607884696905e-07, + "loss": 0.7846, + "mean_token_accuracy": 0.7768372297286987, + "num_tokens": 54729252.0, + "step": 1434 + }, + { + "epoch": 0.182546749777382, + "grad_norm": 27.875, + "learning_rate": 6.078846969054684e-07, + "loss": 0.803, + "mean_token_accuracy": 0.7703762054443359, + "num_tokens": 54763422.0, + "step": 1435 + }, + { + "epoch": 0.18267396005597253, + "grad_norm": 19.75, + "learning_rate": 6.083086053412463e-07, + "loss": 0.7532, + "mean_token_accuracy": 0.7837244868278503, + "num_tokens": 54803494.0, + "step": 1436 + }, + { + "epoch": 0.18280117033456303, + "grad_norm": 93.5, + "learning_rate": 6.087325137770242e-07, + "loss": 0.7742, + "mean_token_accuracy": 0.7826562523841858, + "num_tokens": 54839349.0, + "step": 1437 + }, + { + "epoch": 0.18292838061315353, + "grad_norm": 52.25, + "learning_rate": 6.09156422212802e-07, + "loss": 0.8362, + "mean_token_accuracy": 0.7698636054992676, + "num_tokens": 54881941.0, + "step": 1438 + }, + { + "epoch": 0.18305559089174406, + "grad_norm": 83.5, + "learning_rate": 6.095803306485799e-07, + "loss": 0.7841, + "mean_token_accuracy": 0.7762624621391296, + "num_tokens": 54922927.0, + "step": 1439 + }, + { + "epoch": 0.18318280117033456, + "grad_norm": 13.5625, + "learning_rate": 6.100042390843577e-07, + "loss": 0.8002, + "mean_token_accuracy": 0.7710006833076477, + "num_tokens": 54963885.0, + "step": 1440 + }, + { + "epoch": 0.1833100114489251, + "grad_norm": 159.0, + "learning_rate": 6.104281475201356e-07, + "loss": 0.7642, + "mean_token_accuracy": 0.7817946672439575, + "num_tokens": 55001284.0, + "step": 1441 + }, + { + "epoch": 0.1834372217275156, + "grad_norm": 196.0, + "learning_rate": 6.108520559559135e-07, + "loss": 0.8582, + "mean_token_accuracy": 0.7573245167732239, + "num_tokens": 55034813.0, + "step": 1442 + }, + { + "epoch": 0.1835644320061061, + "grad_norm": 141.0, + "learning_rate": 6.112759643916914e-07, + "loss": 0.7239, + "mean_token_accuracy": 0.7942079305648804, + "num_tokens": 55070620.0, + "step": 1443 + }, + { + "epoch": 0.18369164228469662, + "grad_norm": 110.0, + "learning_rate": 6.116998728274693e-07, + "loss": 0.8176, + "mean_token_accuracy": 0.7721155881881714, + "num_tokens": 55108394.0, + "step": 1444 + }, + { + "epoch": 0.18381885256328712, + "grad_norm": 81.5, + "learning_rate": 6.121237812632472e-07, + "loss": 0.8291, + "mean_token_accuracy": 0.7661533355712891, + "num_tokens": 55148170.0, + "step": 1445 + }, + { + "epoch": 0.18394606284187762, + "grad_norm": 90.5, + "learning_rate": 6.125476896990249e-07, + "loss": 0.7352, + "mean_token_accuracy": 0.7918659448623657, + "num_tokens": 55188102.0, + "step": 1446 + }, + { + "epoch": 0.18407327312046814, + "grad_norm": 64.0, + "learning_rate": 6.129715981348028e-07, + "loss": 0.8046, + "mean_token_accuracy": 0.7748029828071594, + "num_tokens": 55225893.0, + "step": 1447 + }, + { + "epoch": 0.18420048339905865, + "grad_norm": 78.0, + "learning_rate": 6.133955065705807e-07, + "loss": 0.8314, + "mean_token_accuracy": 0.7611627578735352, + "num_tokens": 55265321.0, + "step": 1448 + }, + { + "epoch": 0.18432769367764915, + "grad_norm": 20.625, + "learning_rate": 6.138194150063585e-07, + "loss": 0.7652, + "mean_token_accuracy": 0.7841021418571472, + "num_tokens": 55304179.0, + "step": 1449 + }, + { + "epoch": 0.18445490395623967, + "grad_norm": 150.0, + "learning_rate": 6.142433234421365e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7682108879089355, + "num_tokens": 55343266.0, + "step": 1450 + }, + { + "epoch": 0.18458211423483017, + "grad_norm": 117.0, + "learning_rate": 6.146672318779143e-07, + "loss": 0.822, + "mean_token_accuracy": 0.7671886086463928, + "num_tokens": 55381147.0, + "step": 1451 + }, + { + "epoch": 0.18470932451342068, + "grad_norm": 112.5, + "learning_rate": 6.150911403136923e-07, + "loss": 0.8284, + "mean_token_accuracy": 0.7677571177482605, + "num_tokens": 55414855.0, + "step": 1452 + }, + { + "epoch": 0.1848365347920112, + "grad_norm": 85.0, + "learning_rate": 6.155150487494701e-07, + "loss": 0.7281, + "mean_token_accuracy": 0.792131781578064, + "num_tokens": 55449699.0, + "step": 1453 + }, + { + "epoch": 0.1849637450706017, + "grad_norm": 131.0, + "learning_rate": 6.159389571852479e-07, + "loss": 0.7559, + "mean_token_accuracy": 0.7877689599990845, + "num_tokens": 55493202.0, + "step": 1454 + }, + { + "epoch": 0.1850909553491922, + "grad_norm": 42.5, + "learning_rate": 6.163628656210258e-07, + "loss": 0.7791, + "mean_token_accuracy": 0.781901478767395, + "num_tokens": 55539873.0, + "step": 1455 + }, + { + "epoch": 0.18521816562778273, + "grad_norm": 175.0, + "learning_rate": 6.167867740568037e-07, + "loss": 0.876, + "mean_token_accuracy": 0.7532413005828857, + "num_tokens": 55576338.0, + "step": 1456 + }, + { + "epoch": 0.18534537590637323, + "grad_norm": 50.75, + "learning_rate": 6.172106824925815e-07, + "loss": 0.758, + "mean_token_accuracy": 0.782317042350769, + "num_tokens": 55618330.0, + "step": 1457 + }, + { + "epoch": 0.18547258618496373, + "grad_norm": 76.0, + "learning_rate": 6.176345909283595e-07, + "loss": 0.7816, + "mean_token_accuracy": 0.7762093544006348, + "num_tokens": 55657394.0, + "step": 1458 + }, + { + "epoch": 0.18559979646355426, + "grad_norm": 107.5, + "learning_rate": 6.180584993641373e-07, + "loss": 0.7832, + "mean_token_accuracy": 0.7784504890441895, + "num_tokens": 55695087.0, + "step": 1459 + }, + { + "epoch": 0.18572700674214476, + "grad_norm": 163.0, + "learning_rate": 6.184824077999153e-07, + "loss": 0.7401, + "mean_token_accuracy": 0.7873082756996155, + "num_tokens": 55730852.0, + "step": 1460 + }, + { + "epoch": 0.18585421702073526, + "grad_norm": 181.0, + "learning_rate": 6.189063162356931e-07, + "loss": 0.8792, + "mean_token_accuracy": 0.7544854879379272, + "num_tokens": 55765678.0, + "step": 1461 + }, + { + "epoch": 0.1859814272993258, + "grad_norm": 130.0, + "learning_rate": 6.193302246714709e-07, + "loss": 0.7594, + "mean_token_accuracy": 0.7836286425590515, + "num_tokens": 55804409.0, + "step": 1462 + }, + { + "epoch": 0.1861086375779163, + "grad_norm": 134.0, + "learning_rate": 6.197541331072488e-07, + "loss": 0.8139, + "mean_token_accuracy": 0.7722461819648743, + "num_tokens": 55839520.0, + "step": 1463 + }, + { + "epoch": 0.1862358478565068, + "grad_norm": 296.0, + "learning_rate": 6.201780415430267e-07, + "loss": 0.8025, + "mean_token_accuracy": 0.7730346918106079, + "num_tokens": 55879089.0, + "step": 1464 + }, + { + "epoch": 0.18636305813509732, + "grad_norm": 45.75, + "learning_rate": 6.206019499788045e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.760373592376709, + "num_tokens": 55916655.0, + "step": 1465 + }, + { + "epoch": 0.18649026841368782, + "grad_norm": 48.25, + "learning_rate": 6.210258584145825e-07, + "loss": 0.7639, + "mean_token_accuracy": 0.7791155576705933, + "num_tokens": 55953828.0, + "step": 1466 + }, + { + "epoch": 0.18661747869227835, + "grad_norm": 20.0, + "learning_rate": 6.214497668503603e-07, + "loss": 0.8042, + "mean_token_accuracy": 0.7697576284408569, + "num_tokens": 55992037.0, + "step": 1467 + }, + { + "epoch": 0.18674468897086885, + "grad_norm": 191.0, + "learning_rate": 6.218736752861383e-07, + "loss": 0.7971, + "mean_token_accuracy": 0.7749007940292358, + "num_tokens": 56020424.0, + "step": 1468 + }, + { + "epoch": 0.18687189924945935, + "grad_norm": 132.0, + "learning_rate": 6.22297583721916e-07, + "loss": 0.7721, + "mean_token_accuracy": 0.786180853843689, + "num_tokens": 56060839.0, + "step": 1469 + }, + { + "epoch": 0.18699910952804988, + "grad_norm": 178.0, + "learning_rate": 6.227214921576938e-07, + "loss": 0.8527, + "mean_token_accuracy": 0.7664976119995117, + "num_tokens": 56099095.0, + "step": 1470 + }, + { + "epoch": 0.18712631980664038, + "grad_norm": 177.0, + "learning_rate": 6.231454005934718e-07, + "loss": 0.7279, + "mean_token_accuracy": 0.7932743430137634, + "num_tokens": 56133805.0, + "step": 1471 + }, + { + "epoch": 0.18725353008523088, + "grad_norm": 222.0, + "learning_rate": 6.235693090292496e-07, + "loss": 0.8764, + "mean_token_accuracy": 0.7556459307670593, + "num_tokens": 56173064.0, + "step": 1472 + }, + { + "epoch": 0.1873807403638214, + "grad_norm": 208.0, + "learning_rate": 6.239932174650275e-07, + "loss": 0.7098, + "mean_token_accuracy": 0.8013010025024414, + "num_tokens": 56213285.0, + "step": 1473 + }, + { + "epoch": 0.1875079506424119, + "grad_norm": 166.0, + "learning_rate": 6.244171259008054e-07, + "loss": 0.824, + "mean_token_accuracy": 0.7659636735916138, + "num_tokens": 56248218.0, + "step": 1474 + }, + { + "epoch": 0.1876351609210024, + "grad_norm": 35.25, + "learning_rate": 6.248410343365833e-07, + "loss": 0.753, + "mean_token_accuracy": 0.7877026796340942, + "num_tokens": 56286098.0, + "step": 1475 + }, + { + "epoch": 0.18776237119959294, + "grad_norm": 14.875, + "learning_rate": 6.252649427723612e-07, + "loss": 0.7393, + "mean_token_accuracy": 0.7898434400558472, + "num_tokens": 56329732.0, + "step": 1476 + }, + { + "epoch": 0.18788958147818344, + "grad_norm": 20.375, + "learning_rate": 6.25688851208139e-07, + "loss": 0.7694, + "mean_token_accuracy": 0.7825940251350403, + "num_tokens": 56367838.0, + "step": 1477 + }, + { + "epoch": 0.18801679175677394, + "grad_norm": 54.5, + "learning_rate": 6.261127596439168e-07, + "loss": 0.8112, + "mean_token_accuracy": 0.7680465579032898, + "num_tokens": 56413059.0, + "step": 1478 + }, + { + "epoch": 0.18814400203536447, + "grad_norm": 256.0, + "learning_rate": 6.265366680796948e-07, + "loss": 0.7156, + "mean_token_accuracy": 0.7909085750579834, + "num_tokens": 56453806.0, + "step": 1479 + }, + { + "epoch": 0.18827121231395497, + "grad_norm": 177.0, + "learning_rate": 6.269605765154726e-07, + "loss": 0.7691, + "mean_token_accuracy": 0.7829403877258301, + "num_tokens": 56497448.0, + "step": 1480 + }, + { + "epoch": 0.18839842259254547, + "grad_norm": 158.0, + "learning_rate": 6.273844849512505e-07, + "loss": 0.7795, + "mean_token_accuracy": 0.7799586057662964, + "num_tokens": 56537731.0, + "step": 1481 + }, + { + "epoch": 0.188525632871136, + "grad_norm": 62.25, + "learning_rate": 6.278083933870284e-07, + "loss": 0.6834, + "mean_token_accuracy": 0.8042981028556824, + "num_tokens": 56580944.0, + "step": 1482 + }, + { + "epoch": 0.1886528431497265, + "grad_norm": 166.0, + "learning_rate": 6.282323018228063e-07, + "loss": 0.7493, + "mean_token_accuracy": 0.787327229976654, + "num_tokens": 56616306.0, + "step": 1483 + }, + { + "epoch": 0.188780053428317, + "grad_norm": 22.0, + "learning_rate": 6.286562102585841e-07, + "loss": 0.8095, + "mean_token_accuracy": 0.7734214067459106, + "num_tokens": 56661874.0, + "step": 1484 + }, + { + "epoch": 0.18890726370690752, + "grad_norm": 98.5, + "learning_rate": 6.29080118694362e-07, + "loss": 0.7147, + "mean_token_accuracy": 0.7934776544570923, + "num_tokens": 56696697.0, + "step": 1485 + }, + { + "epoch": 0.18903447398549802, + "grad_norm": 18.25, + "learning_rate": 6.295040271301398e-07, + "loss": 0.7189, + "mean_token_accuracy": 0.7911180257797241, + "num_tokens": 56732581.0, + "step": 1486 + }, + { + "epoch": 0.18916168426408853, + "grad_norm": 126.0, + "learning_rate": 6.299279355659178e-07, + "loss": 0.7645, + "mean_token_accuracy": 0.7833094596862793, + "num_tokens": 56777905.0, + "step": 1487 + }, + { + "epoch": 0.18928889454267905, + "grad_norm": 30.375, + "learning_rate": 6.303518440016956e-07, + "loss": 0.8086, + "mean_token_accuracy": 0.7691695690155029, + "num_tokens": 56812915.0, + "step": 1488 + }, + { + "epoch": 0.18941610482126955, + "grad_norm": 56.25, + "learning_rate": 6.307757524374735e-07, + "loss": 0.7971, + "mean_token_accuracy": 0.774745523929596, + "num_tokens": 56846907.0, + "step": 1489 + }, + { + "epoch": 0.18954331509986008, + "grad_norm": 62.5, + "learning_rate": 6.311996608732514e-07, + "loss": 0.8158, + "mean_token_accuracy": 0.7729259133338928, + "num_tokens": 56884614.0, + "step": 1490 + }, + { + "epoch": 0.18967052537845058, + "grad_norm": 27.5, + "learning_rate": 6.316235693090292e-07, + "loss": 0.8196, + "mean_token_accuracy": 0.7656347751617432, + "num_tokens": 56920508.0, + "step": 1491 + }, + { + "epoch": 0.18979773565704108, + "grad_norm": 65.0, + "learning_rate": 6.320474777448071e-07, + "loss": 0.8171, + "mean_token_accuracy": 0.7665668725967407, + "num_tokens": 56957629.0, + "step": 1492 + }, + { + "epoch": 0.1899249459356316, + "grad_norm": 40.25, + "learning_rate": 6.324713861805849e-07, + "loss": 0.7576, + "mean_token_accuracy": 0.7826743125915527, + "num_tokens": 56997265.0, + "step": 1493 + }, + { + "epoch": 0.1900521562142221, + "grad_norm": 36.75, + "learning_rate": 6.328952946163628e-07, + "loss": 0.8821, + "mean_token_accuracy": 0.7552686929702759, + "num_tokens": 57031519.0, + "step": 1494 + }, + { + "epoch": 0.1901793664928126, + "grad_norm": 16.625, + "learning_rate": 6.333192030521407e-07, + "loss": 0.7483, + "mean_token_accuracy": 0.7830939888954163, + "num_tokens": 57067264.0, + "step": 1495 + }, + { + "epoch": 0.19030657677140314, + "grad_norm": 12.8125, + "learning_rate": 6.337431114879186e-07, + "loss": 0.7952, + "mean_token_accuracy": 0.7749978303909302, + "num_tokens": 57104906.0, + "step": 1496 + }, + { + "epoch": 0.19043378704999364, + "grad_norm": 13.8125, + "learning_rate": 6.341670199236965e-07, + "loss": 0.756, + "mean_token_accuracy": 0.7830760478973389, + "num_tokens": 57142534.0, + "step": 1497 + }, + { + "epoch": 0.19056099732858414, + "grad_norm": 15.125, + "learning_rate": 6.345909283594744e-07, + "loss": 0.7188, + "mean_token_accuracy": 0.7940782308578491, + "num_tokens": 57188735.0, + "step": 1498 + }, + { + "epoch": 0.19068820760717467, + "grad_norm": 36.75, + "learning_rate": 6.350148367952522e-07, + "loss": 0.8024, + "mean_token_accuracy": 0.7688654065132141, + "num_tokens": 57224610.0, + "step": 1499 + }, + { + "epoch": 0.19081541788576517, + "grad_norm": 27.375, + "learning_rate": 6.354387452310301e-07, + "loss": 0.7575, + "mean_token_accuracy": 0.7800449132919312, + "num_tokens": 57259553.0, + "step": 1500 + }, + { + "epoch": 0.19094262816435567, + "grad_norm": 74.5, + "learning_rate": 6.358626536668079e-07, + "loss": 0.7426, + "mean_token_accuracy": 0.7886199951171875, + "num_tokens": 57295819.0, + "step": 1501 + }, + { + "epoch": 0.1910698384429462, + "grad_norm": 25.25, + "learning_rate": 6.362865621025858e-07, + "loss": 0.709, + "mean_token_accuracy": 0.795275092124939, + "num_tokens": 57339110.0, + "step": 1502 + }, + { + "epoch": 0.1911970487215367, + "grad_norm": 48.5, + "learning_rate": 6.367104705383637e-07, + "loss": 0.7695, + "mean_token_accuracy": 0.7797068357467651, + "num_tokens": 57377311.0, + "step": 1503 + }, + { + "epoch": 0.1913242590001272, + "grad_norm": 37.25, + "learning_rate": 6.371343789741416e-07, + "loss": 0.7733, + "mean_token_accuracy": 0.7769497632980347, + "num_tokens": 57419281.0, + "step": 1504 + }, + { + "epoch": 0.19145146927871773, + "grad_norm": 21.0, + "learning_rate": 6.375582874099195e-07, + "loss": 0.7551, + "mean_token_accuracy": 0.7830595374107361, + "num_tokens": 57458134.0, + "step": 1505 + }, + { + "epoch": 0.19157867955730823, + "grad_norm": 13.1875, + "learning_rate": 6.379821958456974e-07, + "loss": 0.8266, + "mean_token_accuracy": 0.763886570930481, + "num_tokens": 57492931.0, + "step": 1506 + }, + { + "epoch": 0.19170588983589873, + "grad_norm": 71.5, + "learning_rate": 6.384061042814751e-07, + "loss": 0.8036, + "mean_token_accuracy": 0.768033504486084, + "num_tokens": 57529581.0, + "step": 1507 + }, + { + "epoch": 0.19183310011448926, + "grad_norm": 73.5, + "learning_rate": 6.38830012717253e-07, + "loss": 0.7677, + "mean_token_accuracy": 0.7786853909492493, + "num_tokens": 57570302.0, + "step": 1508 + }, + { + "epoch": 0.19196031039307976, + "grad_norm": 8.375, + "learning_rate": 6.392539211530309e-07, + "loss": 0.8075, + "mean_token_accuracy": 0.769892692565918, + "num_tokens": 57615974.0, + "step": 1509 + }, + { + "epoch": 0.19208752067167026, + "grad_norm": 57.75, + "learning_rate": 6.396778295888087e-07, + "loss": 0.7614, + "mean_token_accuracy": 0.7852200269699097, + "num_tokens": 57653737.0, + "step": 1510 + }, + { + "epoch": 0.1922147309502608, + "grad_norm": 87.0, + "learning_rate": 6.401017380245867e-07, + "loss": 0.7874, + "mean_token_accuracy": 0.7752597332000732, + "num_tokens": 57693431.0, + "step": 1511 + }, + { + "epoch": 0.1923419412288513, + "grad_norm": 40.25, + "learning_rate": 6.405256464603645e-07, + "loss": 0.763, + "mean_token_accuracy": 0.7794067859649658, + "num_tokens": 57736413.0, + "step": 1512 + }, + { + "epoch": 0.1924691515074418, + "grad_norm": 25.625, + "learning_rate": 6.409495548961425e-07, + "loss": 0.7663, + "mean_token_accuracy": 0.7846698760986328, + "num_tokens": 57773293.0, + "step": 1513 + }, + { + "epoch": 0.19259636178603232, + "grad_norm": 86.0, + "learning_rate": 6.413734633319203e-07, + "loss": 0.7834, + "mean_token_accuracy": 0.7778118848800659, + "num_tokens": 57809294.0, + "step": 1514 + }, + { + "epoch": 0.19272357206462282, + "grad_norm": 56.0, + "learning_rate": 6.417973717676981e-07, + "loss": 0.7453, + "mean_token_accuracy": 0.7845760583877563, + "num_tokens": 57843610.0, + "step": 1515 + }, + { + "epoch": 0.19285078234321335, + "grad_norm": 13.125, + "learning_rate": 6.42221280203476e-07, + "loss": 0.7885, + "mean_token_accuracy": 0.773125410079956, + "num_tokens": 57893098.0, + "step": 1516 + }, + { + "epoch": 0.19297799262180385, + "grad_norm": 14.125, + "learning_rate": 6.426451886392539e-07, + "loss": 0.8389, + "mean_token_accuracy": 0.7644394636154175, + "num_tokens": 57933098.0, + "step": 1517 + }, + { + "epoch": 0.19310520290039435, + "grad_norm": 169.0, + "learning_rate": 6.430690970750317e-07, + "loss": 0.7698, + "mean_token_accuracy": 0.780837893486023, + "num_tokens": 57971502.0, + "step": 1518 + }, + { + "epoch": 0.19323241317898487, + "grad_norm": 118.5, + "learning_rate": 6.434930055108097e-07, + "loss": 0.8388, + "mean_token_accuracy": 0.7616792917251587, + "num_tokens": 58001557.0, + "step": 1519 + }, + { + "epoch": 0.19335962345757537, + "grad_norm": 35.25, + "learning_rate": 6.439169139465875e-07, + "loss": 0.8565, + "mean_token_accuracy": 0.7574433088302612, + "num_tokens": 58038058.0, + "step": 1520 + }, + { + "epoch": 0.19348683373616588, + "grad_norm": 76.5, + "learning_rate": 6.443408223823655e-07, + "loss": 0.7959, + "mean_token_accuracy": 0.7719356417655945, + "num_tokens": 58073138.0, + "step": 1521 + }, + { + "epoch": 0.1936140440147564, + "grad_norm": 140.0, + "learning_rate": 6.447647308181432e-07, + "loss": 0.7856, + "mean_token_accuracy": 0.7768352031707764, + "num_tokens": 58111793.0, + "step": 1522 + }, + { + "epoch": 0.1937412542933469, + "grad_norm": 141.0, + "learning_rate": 6.451886392539211e-07, + "loss": 0.7684, + "mean_token_accuracy": 0.7820208072662354, + "num_tokens": 58152911.0, + "step": 1523 + }, + { + "epoch": 0.1938684645719374, + "grad_norm": 79.5, + "learning_rate": 6.45612547689699e-07, + "loss": 0.796, + "mean_token_accuracy": 0.7753363847732544, + "num_tokens": 58192095.0, + "step": 1524 + }, + { + "epoch": 0.19399567485052793, + "grad_norm": 149.0, + "learning_rate": 6.460364561254769e-07, + "loss": 0.7318, + "mean_token_accuracy": 0.7911150455474854, + "num_tokens": 58232651.0, + "step": 1525 + }, + { + "epoch": 0.19412288512911843, + "grad_norm": 26.875, + "learning_rate": 6.464603645612547e-07, + "loss": 0.7358, + "mean_token_accuracy": 0.7907230257987976, + "num_tokens": 58270802.0, + "step": 1526 + }, + { + "epoch": 0.19425009540770893, + "grad_norm": 15.875, + "learning_rate": 6.468842729970327e-07, + "loss": 0.763, + "mean_token_accuracy": 0.7858390212059021, + "num_tokens": 58313179.0, + "step": 1527 + }, + { + "epoch": 0.19437730568629946, + "grad_norm": 83.0, + "learning_rate": 6.473081814328105e-07, + "loss": 0.7861, + "mean_token_accuracy": 0.779167115688324, + "num_tokens": 58354728.0, + "step": 1528 + }, + { + "epoch": 0.19450451596488996, + "grad_norm": 31.875, + "learning_rate": 6.477320898685885e-07, + "loss": 0.7989, + "mean_token_accuracy": 0.7749145030975342, + "num_tokens": 58398649.0, + "step": 1529 + }, + { + "epoch": 0.19463172624348046, + "grad_norm": 23.5, + "learning_rate": 6.481559983043662e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7725009918212891, + "num_tokens": 58438261.0, + "step": 1530 + }, + { + "epoch": 0.194758936522071, + "grad_norm": 114.5, + "learning_rate": 6.48579906740144e-07, + "loss": 0.7351, + "mean_token_accuracy": 0.7870281338691711, + "num_tokens": 58479141.0, + "step": 1531 + }, + { + "epoch": 0.1948861468006615, + "grad_norm": 13.4375, + "learning_rate": 6.49003815175922e-07, + "loss": 0.7034, + "mean_token_accuracy": 0.8001630306243896, + "num_tokens": 58515700.0, + "step": 1532 + }, + { + "epoch": 0.195013357079252, + "grad_norm": 20.625, + "learning_rate": 6.494277236116998e-07, + "loss": 0.7211, + "mean_token_accuracy": 0.7952417135238647, + "num_tokens": 58549825.0, + "step": 1533 + }, + { + "epoch": 0.19514056735784252, + "grad_norm": 21.0, + "learning_rate": 6.498516320474777e-07, + "loss": 0.8576, + "mean_token_accuracy": 0.7578556537628174, + "num_tokens": 58596110.0, + "step": 1534 + }, + { + "epoch": 0.19526777763643302, + "grad_norm": 56.25, + "learning_rate": 6.502755404832556e-07, + "loss": 0.7718, + "mean_token_accuracy": 0.7762758135795593, + "num_tokens": 58632410.0, + "step": 1535 + }, + { + "epoch": 0.19539498791502352, + "grad_norm": 34.75, + "learning_rate": 6.506994489190335e-07, + "loss": 0.8343, + "mean_token_accuracy": 0.7670558094978333, + "num_tokens": 58671339.0, + "step": 1536 + }, + { + "epoch": 0.19552219819361405, + "grad_norm": 22.875, + "learning_rate": 6.511233573548114e-07, + "loss": 0.721, + "mean_token_accuracy": 0.7913500070571899, + "num_tokens": 58711786.0, + "step": 1537 + }, + { + "epoch": 0.19564940847220455, + "grad_norm": 107.5, + "learning_rate": 6.515472657905892e-07, + "loss": 0.7949, + "mean_token_accuracy": 0.7710127234458923, + "num_tokens": 58743600.0, + "step": 1538 + }, + { + "epoch": 0.19577661875079505, + "grad_norm": 147.0, + "learning_rate": 6.51971174226367e-07, + "loss": 0.8573, + "mean_token_accuracy": 0.7571957111358643, + "num_tokens": 58780622.0, + "step": 1539 + }, + { + "epoch": 0.19590382902938558, + "grad_norm": 224.0, + "learning_rate": 6.52395082662145e-07, + "loss": 0.8489, + "mean_token_accuracy": 0.7572543621063232, + "num_tokens": 58819370.0, + "step": 1540 + }, + { + "epoch": 0.19603103930797608, + "grad_norm": 162.0, + "learning_rate": 6.528189910979228e-07, + "loss": 0.758, + "mean_token_accuracy": 0.7787019610404968, + "num_tokens": 58856270.0, + "step": 1541 + }, + { + "epoch": 0.1961582495865666, + "grad_norm": 228.0, + "learning_rate": 6.532428995337007e-07, + "loss": 0.7705, + "mean_token_accuracy": 0.7786865830421448, + "num_tokens": 58898702.0, + "step": 1542 + }, + { + "epoch": 0.1962854598651571, + "grad_norm": 153.0, + "learning_rate": 6.536668079694786e-07, + "loss": 0.8022, + "mean_token_accuracy": 0.7675952911376953, + "num_tokens": 58934602.0, + "step": 1543 + }, + { + "epoch": 0.1964126701437476, + "grad_norm": 127.5, + "learning_rate": 6.540907164052565e-07, + "loss": 0.7582, + "mean_token_accuracy": 0.7807676792144775, + "num_tokens": 58977766.0, + "step": 1544 + }, + { + "epoch": 0.19653988042233814, + "grad_norm": 108.0, + "learning_rate": 6.545146248410343e-07, + "loss": 0.7866, + "mean_token_accuracy": 0.7723968625068665, + "num_tokens": 59019649.0, + "step": 1545 + }, + { + "epoch": 0.19666709070092864, + "grad_norm": 30.875, + "learning_rate": 6.549385332768122e-07, + "loss": 0.7916, + "mean_token_accuracy": 0.7761030197143555, + "num_tokens": 59060543.0, + "step": 1546 + }, + { + "epoch": 0.19679430097951914, + "grad_norm": 16.5, + "learning_rate": 6.5536244171259e-07, + "loss": 0.7822, + "mean_token_accuracy": 0.7755714654922485, + "num_tokens": 59100049.0, + "step": 1547 + }, + { + "epoch": 0.19692151125810967, + "grad_norm": 163.0, + "learning_rate": 6.55786350148368e-07, + "loss": 0.8072, + "mean_token_accuracy": 0.7715991139411926, + "num_tokens": 59137370.0, + "step": 1548 + }, + { + "epoch": 0.19704872153670017, + "grad_norm": 236.0, + "learning_rate": 6.562102585841458e-07, + "loss": 0.7658, + "mean_token_accuracy": 0.7780276536941528, + "num_tokens": 59176196.0, + "step": 1549 + }, + { + "epoch": 0.19717593181529067, + "grad_norm": 164.0, + "learning_rate": 6.566341670199236e-07, + "loss": 0.7515, + "mean_token_accuracy": 0.7852362394332886, + "num_tokens": 59216215.0, + "step": 1550 + }, + { + "epoch": 0.1973031420938812, + "grad_norm": 244.0, + "learning_rate": 6.570580754557016e-07, + "loss": 0.7876, + "mean_token_accuracy": 0.7768376469612122, + "num_tokens": 59254173.0, + "step": 1551 + }, + { + "epoch": 0.1974303523724717, + "grad_norm": 234.0, + "learning_rate": 6.574819838914794e-07, + "loss": 0.7605, + "mean_token_accuracy": 0.7874587178230286, + "num_tokens": 59288496.0, + "step": 1552 + }, + { + "epoch": 0.1975575626510622, + "grad_norm": 50.25, + "learning_rate": 6.579058923272573e-07, + "loss": 0.8251, + "mean_token_accuracy": 0.7639642953872681, + "num_tokens": 59329043.0, + "step": 1553 + }, + { + "epoch": 0.19768477292965272, + "grad_norm": 17.125, + "learning_rate": 6.583298007630351e-07, + "loss": 0.8116, + "mean_token_accuracy": 0.7707385420799255, + "num_tokens": 59360743.0, + "step": 1554 + }, + { + "epoch": 0.19781198320824323, + "grad_norm": 62.75, + "learning_rate": 6.58753709198813e-07, + "loss": 0.7832, + "mean_token_accuracy": 0.777380108833313, + "num_tokens": 59404167.0, + "step": 1555 + }, + { + "epoch": 0.19793919348683373, + "grad_norm": 44.25, + "learning_rate": 6.591776176345909e-07, + "loss": 0.7413, + "mean_token_accuracy": 0.7860902547836304, + "num_tokens": 59447770.0, + "step": 1556 + }, + { + "epoch": 0.19806640376542425, + "grad_norm": 62.5, + "learning_rate": 6.596015260703688e-07, + "loss": 0.7546, + "mean_token_accuracy": 0.7862760424613953, + "num_tokens": 59494989.0, + "step": 1557 + }, + { + "epoch": 0.19819361404401475, + "grad_norm": 126.5, + "learning_rate": 6.600254345061466e-07, + "loss": 0.7405, + "mean_token_accuracy": 0.7893053889274597, + "num_tokens": 59534926.0, + "step": 1558 + }, + { + "epoch": 0.19832082432260525, + "grad_norm": 30.125, + "learning_rate": 6.604493429419246e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7668052911758423, + "num_tokens": 59577832.0, + "step": 1559 + }, + { + "epoch": 0.19844803460119578, + "grad_norm": 55.5, + "learning_rate": 6.608732513777023e-07, + "loss": 0.7939, + "mean_token_accuracy": 0.7810021638870239, + "num_tokens": 59624995.0, + "step": 1560 + }, + { + "epoch": 0.19857524487978628, + "grad_norm": 37.5, + "learning_rate": 6.612971598134803e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7781437635421753, + "num_tokens": 59663000.0, + "step": 1561 + }, + { + "epoch": 0.19870245515837678, + "grad_norm": 19.625, + "learning_rate": 6.617210682492581e-07, + "loss": 0.8412, + "mean_token_accuracy": 0.7599993348121643, + "num_tokens": 59700241.0, + "step": 1562 + }, + { + "epoch": 0.1988296654369673, + "grad_norm": 11.1875, + "learning_rate": 6.62144976685036e-07, + "loss": 0.7337, + "mean_token_accuracy": 0.7938799262046814, + "num_tokens": 59737244.0, + "step": 1563 + }, + { + "epoch": 0.1989568757155578, + "grad_norm": 109.5, + "learning_rate": 6.625688851208139e-07, + "loss": 0.7786, + "mean_token_accuracy": 0.7786512970924377, + "num_tokens": 59775538.0, + "step": 1564 + }, + { + "epoch": 0.19908408599414834, + "grad_norm": 36.25, + "learning_rate": 6.629927935565918e-07, + "loss": 0.781, + "mean_token_accuracy": 0.7762421369552612, + "num_tokens": 59810373.0, + "step": 1565 + }, + { + "epoch": 0.19921129627273884, + "grad_norm": 41.75, + "learning_rate": 6.634167019923696e-07, + "loss": 0.8169, + "mean_token_accuracy": 0.7726399898529053, + "num_tokens": 59849992.0, + "step": 1566 + }, + { + "epoch": 0.19933850655132934, + "grad_norm": 52.25, + "learning_rate": 6.638406104281476e-07, + "loss": 0.7961, + "mean_token_accuracy": 0.7724530100822449, + "num_tokens": 59882552.0, + "step": 1567 + }, + { + "epoch": 0.19946571682991987, + "grad_norm": 19.125, + "learning_rate": 6.642645188639253e-07, + "loss": 0.7399, + "mean_token_accuracy": 0.7900283336639404, + "num_tokens": 59925090.0, + "step": 1568 + }, + { + "epoch": 0.19959292710851037, + "grad_norm": 58.5, + "learning_rate": 6.646884272997032e-07, + "loss": 0.8102, + "mean_token_accuracy": 0.7698034048080444, + "num_tokens": 59965406.0, + "step": 1569 + }, + { + "epoch": 0.19972013738710087, + "grad_norm": 14.8125, + "learning_rate": 6.651123357354811e-07, + "loss": 0.804, + "mean_token_accuracy": 0.7718936800956726, + "num_tokens": 60001770.0, + "step": 1570 + }, + { + "epoch": 0.1998473476656914, + "grad_norm": 64.5, + "learning_rate": 6.655362441712589e-07, + "loss": 0.801, + "mean_token_accuracy": 0.7717124223709106, + "num_tokens": 60036911.0, + "step": 1571 + }, + { + "epoch": 0.1999745579442819, + "grad_norm": 21.875, + "learning_rate": 6.659601526070369e-07, + "loss": 0.8372, + "mean_token_accuracy": 0.7679131627082825, + "num_tokens": 60080210.0, + "step": 1572 + }, + { + "epoch": 0.2001017682228724, + "grad_norm": 35.25, + "learning_rate": 6.663840610428147e-07, + "loss": 0.9008, + "mean_token_accuracy": 0.7514992952346802, + "num_tokens": 60111533.0, + "step": 1573 + }, + { + "epoch": 0.20022897850146293, + "grad_norm": 46.25, + "learning_rate": 6.668079694785926e-07, + "loss": 0.7287, + "mean_token_accuracy": 0.7920616269111633, + "num_tokens": 60150733.0, + "step": 1574 + }, + { + "epoch": 0.20035618878005343, + "grad_norm": 24.125, + "learning_rate": 6.672318779143704e-07, + "loss": 0.7812, + "mean_token_accuracy": 0.7750450968742371, + "num_tokens": 60185034.0, + "step": 1575 + }, + { + "epoch": 0.20048339905864393, + "grad_norm": 32.75, + "learning_rate": 6.676557863501483e-07, + "loss": 0.813, + "mean_token_accuracy": 0.7690613269805908, + "num_tokens": 60216865.0, + "step": 1576 + }, + { + "epoch": 0.20061060933723446, + "grad_norm": 23.875, + "learning_rate": 6.680796947859262e-07, + "loss": 0.794, + "mean_token_accuracy": 0.7757829427719116, + "num_tokens": 60257685.0, + "step": 1577 + }, + { + "epoch": 0.20073781961582496, + "grad_norm": 20.125, + "learning_rate": 6.685036032217041e-07, + "loss": 0.784, + "mean_token_accuracy": 0.7781788110733032, + "num_tokens": 60294258.0, + "step": 1578 + }, + { + "epoch": 0.20086502989441546, + "grad_norm": 17.625, + "learning_rate": 6.689275116574819e-07, + "loss": 0.7713, + "mean_token_accuracy": 0.7831138372421265, + "num_tokens": 60330319.0, + "step": 1579 + }, + { + "epoch": 0.200992240173006, + "grad_norm": 22.5, + "learning_rate": 6.693514200932599e-07, + "loss": 0.7911, + "mean_token_accuracy": 0.777192234992981, + "num_tokens": 60370482.0, + "step": 1580 + }, + { + "epoch": 0.2011194504515965, + "grad_norm": 25.875, + "learning_rate": 6.697753285290377e-07, + "loss": 0.8057, + "mean_token_accuracy": 0.7728740572929382, + "num_tokens": 60412436.0, + "step": 1581 + }, + { + "epoch": 0.201246660730187, + "grad_norm": 27.125, + "learning_rate": 6.701992369648156e-07, + "loss": 0.6955, + "mean_token_accuracy": 0.7996914982795715, + "num_tokens": 60448982.0, + "step": 1582 + }, + { + "epoch": 0.20137387100877752, + "grad_norm": 13.125, + "learning_rate": 6.706231454005934e-07, + "loss": 0.7106, + "mean_token_accuracy": 0.7964116930961609, + "num_tokens": 60489878.0, + "step": 1583 + }, + { + "epoch": 0.20150108128736802, + "grad_norm": 20.25, + "learning_rate": 6.710470538363713e-07, + "loss": 0.7604, + "mean_token_accuracy": 0.7811957001686096, + "num_tokens": 60528979.0, + "step": 1584 + }, + { + "epoch": 0.20162829156595852, + "grad_norm": 25.625, + "learning_rate": 6.714709622721492e-07, + "loss": 0.7887, + "mean_token_accuracy": 0.7741953134536743, + "num_tokens": 60561670.0, + "step": 1585 + }, + { + "epoch": 0.20175550184454905, + "grad_norm": 78.0, + "learning_rate": 6.718948707079271e-07, + "loss": 0.7763, + "mean_token_accuracy": 0.7790970206260681, + "num_tokens": 60602061.0, + "step": 1586 + }, + { + "epoch": 0.20188271212313955, + "grad_norm": 14.25, + "learning_rate": 6.723187791437049e-07, + "loss": 0.7614, + "mean_token_accuracy": 0.7814781665802002, + "num_tokens": 60639805.0, + "step": 1587 + }, + { + "epoch": 0.20200992240173005, + "grad_norm": 21.5, + "learning_rate": 6.727426875794829e-07, + "loss": 0.7949, + "mean_token_accuracy": 0.7728791832923889, + "num_tokens": 60681816.0, + "step": 1588 + }, + { + "epoch": 0.20213713268032057, + "grad_norm": 140.0, + "learning_rate": 6.731665960152607e-07, + "loss": 0.789, + "mean_token_accuracy": 0.77274489402771, + "num_tokens": 60726343.0, + "step": 1589 + }, + { + "epoch": 0.20226434295891108, + "grad_norm": 26.25, + "learning_rate": 6.735905044510385e-07, + "loss": 0.834, + "mean_token_accuracy": 0.7603329420089722, + "num_tokens": 60764814.0, + "step": 1590 + }, + { + "epoch": 0.2023915532375016, + "grad_norm": 52.0, + "learning_rate": 6.740144128868164e-07, + "loss": 0.7898, + "mean_token_accuracy": 0.7741160988807678, + "num_tokens": 60802384.0, + "step": 1591 + }, + { + "epoch": 0.2025187635160921, + "grad_norm": 124.0, + "learning_rate": 6.744383213225942e-07, + "loss": 0.8362, + "mean_token_accuracy": 0.7608546018600464, + "num_tokens": 60842085.0, + "step": 1592 + }, + { + "epoch": 0.2026459737946826, + "grad_norm": 124.0, + "learning_rate": 6.748622297583722e-07, + "loss": 0.7652, + "mean_token_accuracy": 0.7774223685264587, + "num_tokens": 60874835.0, + "step": 1593 + }, + { + "epoch": 0.20277318407327313, + "grad_norm": 48.75, + "learning_rate": 6.7528613819415e-07, + "loss": 0.8108, + "mean_token_accuracy": 0.7694538831710815, + "num_tokens": 60910682.0, + "step": 1594 + }, + { + "epoch": 0.20290039435186363, + "grad_norm": 25.125, + "learning_rate": 6.757100466299279e-07, + "loss": 0.783, + "mean_token_accuracy": 0.7788625359535217, + "num_tokens": 60951352.0, + "step": 1595 + }, + { + "epoch": 0.20302760463045413, + "grad_norm": 62.0, + "learning_rate": 6.761339550657058e-07, + "loss": 0.8431, + "mean_token_accuracy": 0.7571709752082825, + "num_tokens": 60979055.0, + "step": 1596 + }, + { + "epoch": 0.20315481490904466, + "grad_norm": 23.5, + "learning_rate": 6.765578635014837e-07, + "loss": 0.7804, + "mean_token_accuracy": 0.7812941074371338, + "num_tokens": 61017437.0, + "step": 1597 + }, + { + "epoch": 0.20328202518763516, + "grad_norm": 100.0, + "learning_rate": 6.769817719372614e-07, + "loss": 0.7697, + "mean_token_accuracy": 0.7811521291732788, + "num_tokens": 61052286.0, + "step": 1598 + }, + { + "epoch": 0.20340923546622566, + "grad_norm": 186.0, + "learning_rate": 6.774056803730394e-07, + "loss": 0.8359, + "mean_token_accuracy": 0.765398383140564, + "num_tokens": 61086110.0, + "step": 1599 + }, + { + "epoch": 0.2035364457448162, + "grad_norm": 99.0, + "learning_rate": 6.778295888088172e-07, + "loss": 0.7903, + "mean_token_accuracy": 0.7776563167572021, + "num_tokens": 61124181.0, + "step": 1600 + }, + { + "epoch": 0.2036636560234067, + "grad_norm": 21.875, + "learning_rate": 6.782534972445952e-07, + "loss": 0.8103, + "mean_token_accuracy": 0.7649356126785278, + "num_tokens": 61160881.0, + "step": 1601 + }, + { + "epoch": 0.2037908663019972, + "grad_norm": 62.75, + "learning_rate": 6.78677405680373e-07, + "loss": 0.789, + "mean_token_accuracy": 0.7762120962142944, + "num_tokens": 61200755.0, + "step": 1602 + }, + { + "epoch": 0.20391807658058772, + "grad_norm": 86.0, + "learning_rate": 6.791013141161509e-07, + "loss": 0.8023, + "mean_token_accuracy": 0.7725362777709961, + "num_tokens": 61240045.0, + "step": 1603 + }, + { + "epoch": 0.20404528685917822, + "grad_norm": 15.1875, + "learning_rate": 6.795252225519288e-07, + "loss": 0.8126, + "mean_token_accuracy": 0.7677420973777771, + "num_tokens": 61282832.0, + "step": 1604 + }, + { + "epoch": 0.20417249713776872, + "grad_norm": 147.0, + "learning_rate": 6.799491309877067e-07, + "loss": 0.7821, + "mean_token_accuracy": 0.7719858884811401, + "num_tokens": 61319841.0, + "step": 1605 + }, + { + "epoch": 0.20429970741635925, + "grad_norm": 35.25, + "learning_rate": 6.803730394234844e-07, + "loss": 0.7668, + "mean_token_accuracy": 0.7814992070198059, + "num_tokens": 61352732.0, + "step": 1606 + }, + { + "epoch": 0.20442691769494975, + "grad_norm": 43.0, + "learning_rate": 6.807969478592624e-07, + "loss": 0.795, + "mean_token_accuracy": 0.778378427028656, + "num_tokens": 61388324.0, + "step": 1607 + }, + { + "epoch": 0.20455412797354025, + "grad_norm": 17.875, + "learning_rate": 6.812208562950402e-07, + "loss": 0.8135, + "mean_token_accuracy": 0.7729761004447937, + "num_tokens": 61431222.0, + "step": 1608 + }, + { + "epoch": 0.20468133825213078, + "grad_norm": 86.5, + "learning_rate": 6.816447647308182e-07, + "loss": 0.8592, + "mean_token_accuracy": 0.7585645914077759, + "num_tokens": 61475617.0, + "step": 1609 + }, + { + "epoch": 0.20480854853072128, + "grad_norm": 9.75, + "learning_rate": 6.82068673166596e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.7811699509620667, + "num_tokens": 61511338.0, + "step": 1610 + }, + { + "epoch": 0.20493575880931178, + "grad_norm": 50.25, + "learning_rate": 6.824925816023738e-07, + "loss": 0.8004, + "mean_token_accuracy": 0.7700299024581909, + "num_tokens": 61549093.0, + "step": 1611 + }, + { + "epoch": 0.2050629690879023, + "grad_norm": 53.75, + "learning_rate": 6.829164900381518e-07, + "loss": 0.7321, + "mean_token_accuracy": 0.7879981994628906, + "num_tokens": 61593843.0, + "step": 1612 + }, + { + "epoch": 0.2051901793664928, + "grad_norm": 31.625, + "learning_rate": 6.833403984739295e-07, + "loss": 0.7874, + "mean_token_accuracy": 0.7741251587867737, + "num_tokens": 61629362.0, + "step": 1613 + }, + { + "epoch": 0.2053173896450833, + "grad_norm": 118.0, + "learning_rate": 6.837643069097074e-07, + "loss": 0.8654, + "mean_token_accuracy": 0.7569445371627808, + "num_tokens": 61667037.0, + "step": 1614 + }, + { + "epoch": 0.20544459992367384, + "grad_norm": 143.0, + "learning_rate": 6.841882153454853e-07, + "loss": 0.7885, + "mean_token_accuracy": 0.7771386504173279, + "num_tokens": 61701976.0, + "step": 1615 + }, + { + "epoch": 0.20557181020226434, + "grad_norm": 72.0, + "learning_rate": 6.846121237812632e-07, + "loss": 0.735, + "mean_token_accuracy": 0.7858908772468567, + "num_tokens": 61739068.0, + "step": 1616 + }, + { + "epoch": 0.20569902048085487, + "grad_norm": 21.875, + "learning_rate": 6.850360322170411e-07, + "loss": 0.7357, + "mean_token_accuracy": 0.790404200553894, + "num_tokens": 61781515.0, + "step": 1617 + }, + { + "epoch": 0.20582623075944537, + "grad_norm": 30.5, + "learning_rate": 6.85459940652819e-07, + "loss": 0.7685, + "mean_token_accuracy": 0.7798789739608765, + "num_tokens": 61815201.0, + "step": 1618 + }, + { + "epoch": 0.20595344103803587, + "grad_norm": 26.625, + "learning_rate": 6.858838490885968e-07, + "loss": 0.8186, + "mean_token_accuracy": 0.7679142951965332, + "num_tokens": 61856639.0, + "step": 1619 + }, + { + "epoch": 0.2060806513166264, + "grad_norm": 28.375, + "learning_rate": 6.863077575243748e-07, + "loss": 0.8334, + "mean_token_accuracy": 0.7668077349662781, + "num_tokens": 61894313.0, + "step": 1620 + }, + { + "epoch": 0.2062078615952169, + "grad_norm": 41.5, + "learning_rate": 6.867316659601525e-07, + "loss": 0.7812, + "mean_token_accuracy": 0.7728838324546814, + "num_tokens": 61930748.0, + "step": 1621 + }, + { + "epoch": 0.2063350718738074, + "grad_norm": 13.3125, + "learning_rate": 6.871555743959304e-07, + "loss": 0.7736, + "mean_token_accuracy": 0.7812066674232483, + "num_tokens": 61972002.0, + "step": 1622 + }, + { + "epoch": 0.20646228215239792, + "grad_norm": 48.25, + "learning_rate": 6.875794828317083e-07, + "loss": 0.7616, + "mean_token_accuracy": 0.7825843691825867, + "num_tokens": 62010141.0, + "step": 1623 + }, + { + "epoch": 0.20658949243098843, + "grad_norm": 19.5, + "learning_rate": 6.880033912674862e-07, + "loss": 0.7311, + "mean_token_accuracy": 0.7906084060668945, + "num_tokens": 62055907.0, + "step": 1624 + }, + { + "epoch": 0.20671670270957893, + "grad_norm": 14.5625, + "learning_rate": 6.884272997032641e-07, + "loss": 0.8639, + "mean_token_accuracy": 0.7542533278465271, + "num_tokens": 62095885.0, + "step": 1625 + }, + { + "epoch": 0.20684391298816945, + "grad_norm": 37.75, + "learning_rate": 6.88851208139042e-07, + "loss": 0.7592, + "mean_token_accuracy": 0.7841033935546875, + "num_tokens": 62130497.0, + "step": 1626 + }, + { + "epoch": 0.20697112326675995, + "grad_norm": 104.0, + "learning_rate": 6.892751165748198e-07, + "loss": 0.8221, + "mean_token_accuracy": 0.7708954215049744, + "num_tokens": 62170296.0, + "step": 1627 + }, + { + "epoch": 0.20709833354535045, + "grad_norm": 59.25, + "learning_rate": 6.896990250105978e-07, + "loss": 0.7263, + "mean_token_accuracy": 0.7954622507095337, + "num_tokens": 62210168.0, + "step": 1628 + }, + { + "epoch": 0.20722554382394098, + "grad_norm": 73.0, + "learning_rate": 6.901229334463755e-07, + "loss": 0.783, + "mean_token_accuracy": 0.7756612300872803, + "num_tokens": 62245251.0, + "step": 1629 + }, + { + "epoch": 0.20735275410253148, + "grad_norm": 102.5, + "learning_rate": 6.905468418821534e-07, + "loss": 0.7549, + "mean_token_accuracy": 0.7904074788093567, + "num_tokens": 62286124.0, + "step": 1630 + }, + { + "epoch": 0.20747996438112198, + "grad_norm": 33.25, + "learning_rate": 6.909707503179313e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.7741634845733643, + "num_tokens": 62325604.0, + "step": 1631 + }, + { + "epoch": 0.2076071746597125, + "grad_norm": 20.375, + "learning_rate": 6.913946587537091e-07, + "loss": 0.7678, + "mean_token_accuracy": 0.7830315828323364, + "num_tokens": 62364467.0, + "step": 1632 + }, + { + "epoch": 0.207734384938303, + "grad_norm": 57.0, + "learning_rate": 6.918185671894871e-07, + "loss": 0.8643, + "mean_token_accuracy": 0.7538113594055176, + "num_tokens": 62402075.0, + "step": 1633 + }, + { + "epoch": 0.2078615952168935, + "grad_norm": 8.8125, + "learning_rate": 6.922424756252649e-07, + "loss": 0.7976, + "mean_token_accuracy": 0.7769125699996948, + "num_tokens": 62437802.0, + "step": 1634 + }, + { + "epoch": 0.20798880549548404, + "grad_norm": 32.5, + "learning_rate": 6.926663840610428e-07, + "loss": 0.7054, + "mean_token_accuracy": 0.7963165640830994, + "num_tokens": 62476176.0, + "step": 1635 + }, + { + "epoch": 0.20811601577407454, + "grad_norm": 47.25, + "learning_rate": 6.930902924968206e-07, + "loss": 0.7797, + "mean_token_accuracy": 0.7741709351539612, + "num_tokens": 62514378.0, + "step": 1636 + }, + { + "epoch": 0.20824322605266504, + "grad_norm": 29.625, + "learning_rate": 6.935142009325985e-07, + "loss": 0.7923, + "mean_token_accuracy": 0.7746431231498718, + "num_tokens": 62555302.0, + "step": 1637 + }, + { + "epoch": 0.20837043633125557, + "grad_norm": 80.5, + "learning_rate": 6.939381093683764e-07, + "loss": 0.8114, + "mean_token_accuracy": 0.7713855504989624, + "num_tokens": 62594356.0, + "step": 1638 + }, + { + "epoch": 0.20849764660984607, + "grad_norm": 144.0, + "learning_rate": 6.943620178041543e-07, + "loss": 0.722, + "mean_token_accuracy": 0.7920981645584106, + "num_tokens": 62632903.0, + "step": 1639 + }, + { + "epoch": 0.2086248568884366, + "grad_norm": 15.125, + "learning_rate": 6.947859262399321e-07, + "loss": 0.7851, + "mean_token_accuracy": 0.7819116115570068, + "num_tokens": 62668979.0, + "step": 1640 + }, + { + "epoch": 0.2087520671670271, + "grad_norm": 16.625, + "learning_rate": 6.952098346757101e-07, + "loss": 0.7619, + "mean_token_accuracy": 0.7796580195426941, + "num_tokens": 62705867.0, + "step": 1641 + }, + { + "epoch": 0.2088792774456176, + "grad_norm": 14.5625, + "learning_rate": 6.956337431114879e-07, + "loss": 0.7594, + "mean_token_accuracy": 0.7851452827453613, + "num_tokens": 62744559.0, + "step": 1642 + }, + { + "epoch": 0.20900648772420813, + "grad_norm": 53.0, + "learning_rate": 6.960576515472658e-07, + "loss": 0.7925, + "mean_token_accuracy": 0.776324987411499, + "num_tokens": 62785221.0, + "step": 1643 + }, + { + "epoch": 0.20913369800279863, + "grad_norm": 19.625, + "learning_rate": 6.964815599830436e-07, + "loss": 0.7884, + "mean_token_accuracy": 0.7724475860595703, + "num_tokens": 62820095.0, + "step": 1644 + }, + { + "epoch": 0.20926090828138913, + "grad_norm": 74.5, + "learning_rate": 6.969054684188215e-07, + "loss": 0.8327, + "mean_token_accuracy": 0.7614810466766357, + "num_tokens": 62863414.0, + "step": 1645 + }, + { + "epoch": 0.20938811855997966, + "grad_norm": 60.5, + "learning_rate": 6.973293768545994e-07, + "loss": 0.7405, + "mean_token_accuracy": 0.7919809818267822, + "num_tokens": 62904647.0, + "step": 1646 + }, + { + "epoch": 0.20951532883857016, + "grad_norm": 93.0, + "learning_rate": 6.977532852903773e-07, + "loss": 0.784, + "mean_token_accuracy": 0.7768278121948242, + "num_tokens": 62949476.0, + "step": 1647 + }, + { + "epoch": 0.20964253911716066, + "grad_norm": 24.125, + "learning_rate": 6.981771937261551e-07, + "loss": 0.8524, + "mean_token_accuracy": 0.7616773843765259, + "num_tokens": 62984784.0, + "step": 1648 + }, + { + "epoch": 0.2097697493957512, + "grad_norm": 34.75, + "learning_rate": 6.986011021619331e-07, + "loss": 0.7926, + "mean_token_accuracy": 0.7736479043960571, + "num_tokens": 63024395.0, + "step": 1649 + }, + { + "epoch": 0.2098969596743417, + "grad_norm": 34.25, + "learning_rate": 6.990250105977109e-07, + "loss": 0.8257, + "mean_token_accuracy": 0.7672580480575562, + "num_tokens": 63061886.0, + "step": 1650 + }, + { + "epoch": 0.2100241699529322, + "grad_norm": 23.25, + "learning_rate": 6.994489190334886e-07, + "loss": 0.7677, + "mean_token_accuracy": 0.7795875072479248, + "num_tokens": 63104150.0, + "step": 1651 + }, + { + "epoch": 0.21015138023152272, + "grad_norm": 98.0, + "learning_rate": 6.998728274692666e-07, + "loss": 0.7369, + "mean_token_accuracy": 0.7907242774963379, + "num_tokens": 63144346.0, + "step": 1652 + }, + { + "epoch": 0.21027859051011322, + "grad_norm": 31.625, + "learning_rate": 7.002967359050444e-07, + "loss": 0.7891, + "mean_token_accuracy": 0.7801510095596313, + "num_tokens": 63180042.0, + "step": 1653 + }, + { + "epoch": 0.21040580078870372, + "grad_norm": 14.3125, + "learning_rate": 7.007206443408224e-07, + "loss": 0.7077, + "mean_token_accuracy": 0.7944966554641724, + "num_tokens": 63216570.0, + "step": 1654 + }, + { + "epoch": 0.21053301106729425, + "grad_norm": 160.0, + "learning_rate": 7.011445527766002e-07, + "loss": 0.7791, + "mean_token_accuracy": 0.7747675180435181, + "num_tokens": 63253457.0, + "step": 1655 + }, + { + "epoch": 0.21066022134588475, + "grad_norm": 29.25, + "learning_rate": 7.015684612123781e-07, + "loss": 0.7592, + "mean_token_accuracy": 0.7822312712669373, + "num_tokens": 63295430.0, + "step": 1656 + }, + { + "epoch": 0.21078743162447525, + "grad_norm": 120.5, + "learning_rate": 7.01992369648156e-07, + "loss": 0.7836, + "mean_token_accuracy": 0.7794257402420044, + "num_tokens": 63340118.0, + "step": 1657 + }, + { + "epoch": 0.21091464190306577, + "grad_norm": 56.0, + "learning_rate": 7.024162780839339e-07, + "loss": 0.8394, + "mean_token_accuracy": 0.756051778793335, + "num_tokens": 63372037.0, + "step": 1658 + }, + { + "epoch": 0.21104185218165628, + "grad_norm": 29.5, + "learning_rate": 7.028401865197116e-07, + "loss": 0.8151, + "mean_token_accuracy": 0.7665196657180786, + "num_tokens": 63415508.0, + "step": 1659 + }, + { + "epoch": 0.21116906246024678, + "grad_norm": 27.0, + "learning_rate": 7.032640949554896e-07, + "loss": 0.684, + "mean_token_accuracy": 0.7978581786155701, + "num_tokens": 63459320.0, + "step": 1660 + }, + { + "epoch": 0.2112962727388373, + "grad_norm": 58.5, + "learning_rate": 7.036880033912674e-07, + "loss": 0.8176, + "mean_token_accuracy": 0.7670367360115051, + "num_tokens": 63499624.0, + "step": 1661 + }, + { + "epoch": 0.2114234830174278, + "grad_norm": 35.75, + "learning_rate": 7.041119118270454e-07, + "loss": 0.7811, + "mean_token_accuracy": 0.7757616639137268, + "num_tokens": 63536507.0, + "step": 1662 + }, + { + "epoch": 0.2115506932960183, + "grad_norm": 94.5, + "learning_rate": 7.045358202628232e-07, + "loss": 0.8039, + "mean_token_accuracy": 0.7762752175331116, + "num_tokens": 63576483.0, + "step": 1663 + }, + { + "epoch": 0.21167790357460883, + "grad_norm": 81.0, + "learning_rate": 7.049597286986011e-07, + "loss": 0.8068, + "mean_token_accuracy": 0.76641845703125, + "num_tokens": 63615568.0, + "step": 1664 + }, + { + "epoch": 0.21180511385319933, + "grad_norm": 106.5, + "learning_rate": 7.05383637134379e-07, + "loss": 0.7685, + "mean_token_accuracy": 0.782085657119751, + "num_tokens": 63650023.0, + "step": 1665 + }, + { + "epoch": 0.21193232413178986, + "grad_norm": 78.0, + "learning_rate": 7.058075455701568e-07, + "loss": 0.7911, + "mean_token_accuracy": 0.7681934237480164, + "num_tokens": 63690229.0, + "step": 1666 + }, + { + "epoch": 0.21205953441038036, + "grad_norm": 35.25, + "learning_rate": 7.062314540059346e-07, + "loss": 0.7019, + "mean_token_accuracy": 0.7970215678215027, + "num_tokens": 63728169.0, + "step": 1667 + }, + { + "epoch": 0.21218674468897086, + "grad_norm": 14.125, + "learning_rate": 7.066553624417126e-07, + "loss": 0.7682, + "mean_token_accuracy": 0.781059205532074, + "num_tokens": 63762816.0, + "step": 1668 + }, + { + "epoch": 0.2123139549675614, + "grad_norm": 17.75, + "learning_rate": 7.070792708774904e-07, + "loss": 0.7769, + "mean_token_accuracy": 0.7791011333465576, + "num_tokens": 63804288.0, + "step": 1669 + }, + { + "epoch": 0.2124411652461519, + "grad_norm": 79.5, + "learning_rate": 7.075031793132684e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.7764880657196045, + "num_tokens": 63851365.0, + "step": 1670 + }, + { + "epoch": 0.2125683755247424, + "grad_norm": 15.875, + "learning_rate": 7.079270877490462e-07, + "loss": 0.7651, + "mean_token_accuracy": 0.7858424186706543, + "num_tokens": 63886187.0, + "step": 1671 + }, + { + "epoch": 0.21269558580333292, + "grad_norm": 172.0, + "learning_rate": 7.08350996184824e-07, + "loss": 0.8457, + "mean_token_accuracy": 0.7623252868652344, + "num_tokens": 63927187.0, + "step": 1672 + }, + { + "epoch": 0.21282279608192342, + "grad_norm": 88.5, + "learning_rate": 7.08774904620602e-07, + "loss": 0.7541, + "mean_token_accuracy": 0.7908282279968262, + "num_tokens": 63964088.0, + "step": 1673 + }, + { + "epoch": 0.21295000636051392, + "grad_norm": 122.5, + "learning_rate": 7.091988130563797e-07, + "loss": 0.8117, + "mean_token_accuracy": 0.770574688911438, + "num_tokens": 64004931.0, + "step": 1674 + }, + { + "epoch": 0.21307721663910445, + "grad_norm": 61.25, + "learning_rate": 7.096227214921576e-07, + "loss": 0.8085, + "mean_token_accuracy": 0.7700371742248535, + "num_tokens": 64043966.0, + "step": 1675 + }, + { + "epoch": 0.21320442691769495, + "grad_norm": 132.0, + "learning_rate": 7.100466299279355e-07, + "loss": 0.7806, + "mean_token_accuracy": 0.7767713069915771, + "num_tokens": 64078886.0, + "step": 1676 + }, + { + "epoch": 0.21333163719628545, + "grad_norm": 130.0, + "learning_rate": 7.104705383637134e-07, + "loss": 0.8042, + "mean_token_accuracy": 0.7764964699745178, + "num_tokens": 64117291.0, + "step": 1677 + }, + { + "epoch": 0.21345884747487598, + "grad_norm": 30.0, + "learning_rate": 7.108944467994913e-07, + "loss": 0.7553, + "mean_token_accuracy": 0.7848790287971497, + "num_tokens": 64156369.0, + "step": 1678 + }, + { + "epoch": 0.21358605775346648, + "grad_norm": 29.875, + "learning_rate": 7.113183552352692e-07, + "loss": 0.8565, + "mean_token_accuracy": 0.7538017630577087, + "num_tokens": 64193646.0, + "step": 1679 + }, + { + "epoch": 0.21371326803205698, + "grad_norm": 12.5625, + "learning_rate": 7.11742263671047e-07, + "loss": 0.7197, + "mean_token_accuracy": 0.7931973338127136, + "num_tokens": 64231933.0, + "step": 1680 + }, + { + "epoch": 0.2138404783106475, + "grad_norm": 35.0, + "learning_rate": 7.12166172106825e-07, + "loss": 0.889, + "mean_token_accuracy": 0.7446358799934387, + "num_tokens": 64267418.0, + "step": 1681 + }, + { + "epoch": 0.213967688589238, + "grad_norm": 28.875, + "learning_rate": 7.125900805426027e-07, + "loss": 0.7373, + "mean_token_accuracy": 0.7875100374221802, + "num_tokens": 64302440.0, + "step": 1682 + }, + { + "epoch": 0.2140948988678285, + "grad_norm": 254.0, + "learning_rate": 7.130139889783806e-07, + "loss": 0.7982, + "mean_token_accuracy": 0.7721379995346069, + "num_tokens": 64338240.0, + "step": 1683 + }, + { + "epoch": 0.21422210914641904, + "grad_norm": 51.5, + "learning_rate": 7.134378974141585e-07, + "loss": 0.7731, + "mean_token_accuracy": 0.778895378112793, + "num_tokens": 64375734.0, + "step": 1684 + }, + { + "epoch": 0.21434931942500954, + "grad_norm": 60.5, + "learning_rate": 7.138618058499364e-07, + "loss": 0.7796, + "mean_token_accuracy": 0.7781240344047546, + "num_tokens": 64411401.0, + "step": 1685 + }, + { + "epoch": 0.21447652970360004, + "grad_norm": 12.375, + "learning_rate": 7.142857142857143e-07, + "loss": 0.7245, + "mean_token_accuracy": 0.7933320999145508, + "num_tokens": 64454719.0, + "step": 1686 + }, + { + "epoch": 0.21460373998219057, + "grad_norm": 40.5, + "learning_rate": 7.147096227214922e-07, + "loss": 0.7636, + "mean_token_accuracy": 0.7837237119674683, + "num_tokens": 64495736.0, + "step": 1687 + }, + { + "epoch": 0.21473095026078107, + "grad_norm": 12.75, + "learning_rate": 7.1513353115727e-07, + "loss": 0.7465, + "mean_token_accuracy": 0.7850382328033447, + "num_tokens": 64532843.0, + "step": 1688 + }, + { + "epoch": 0.21485816053937157, + "grad_norm": 41.75, + "learning_rate": 7.155574395930479e-07, + "loss": 0.8027, + "mean_token_accuracy": 0.7692759037017822, + "num_tokens": 64568950.0, + "step": 1689 + }, + { + "epoch": 0.2149853708179621, + "grad_norm": 75.5, + "learning_rate": 7.159813480288257e-07, + "loss": 0.7603, + "mean_token_accuracy": 0.7777426242828369, + "num_tokens": 64606878.0, + "step": 1690 + }, + { + "epoch": 0.2151125810965526, + "grad_norm": 115.0, + "learning_rate": 7.164052564646035e-07, + "loss": 0.7199, + "mean_token_accuracy": 0.7923556566238403, + "num_tokens": 64643012.0, + "step": 1691 + }, + { + "epoch": 0.21523979137514312, + "grad_norm": 133.0, + "learning_rate": 7.168291649003815e-07, + "loss": 0.7442, + "mean_token_accuracy": 0.7930741310119629, + "num_tokens": 64684444.0, + "step": 1692 + }, + { + "epoch": 0.21536700165373363, + "grad_norm": 126.0, + "learning_rate": 7.172530733361593e-07, + "loss": 0.8334, + "mean_token_accuracy": 0.7646161317825317, + "num_tokens": 64722068.0, + "step": 1693 + }, + { + "epoch": 0.21549421193232413, + "grad_norm": 121.0, + "learning_rate": 7.176769817719373e-07, + "loss": 0.7019, + "mean_token_accuracy": 0.8031504154205322, + "num_tokens": 64761829.0, + "step": 1694 + }, + { + "epoch": 0.21562142221091465, + "grad_norm": 26.0, + "learning_rate": 7.181008902077151e-07, + "loss": 0.7356, + "mean_token_accuracy": 0.7892943024635315, + "num_tokens": 64799185.0, + "step": 1695 + }, + { + "epoch": 0.21574863248950515, + "grad_norm": 86.5, + "learning_rate": 7.18524798643493e-07, + "loss": 0.782, + "mean_token_accuracy": 0.7780526280403137, + "num_tokens": 64839306.0, + "step": 1696 + }, + { + "epoch": 0.21587584276809565, + "grad_norm": 28.875, + "learning_rate": 7.189487070792708e-07, + "loss": 0.76, + "mean_token_accuracy": 0.7824044227600098, + "num_tokens": 64876011.0, + "step": 1697 + }, + { + "epoch": 0.21600305304668618, + "grad_norm": 36.25, + "learning_rate": 7.193726155150487e-07, + "loss": 0.7635, + "mean_token_accuracy": 0.7849476337432861, + "num_tokens": 64918510.0, + "step": 1698 + }, + { + "epoch": 0.21613026332527668, + "grad_norm": 72.0, + "learning_rate": 7.197965239508265e-07, + "loss": 0.7734, + "mean_token_accuracy": 0.7785279750823975, + "num_tokens": 64957156.0, + "step": 1699 + }, + { + "epoch": 0.21625747360386718, + "grad_norm": 71.0, + "learning_rate": 7.202204323866045e-07, + "loss": 0.7697, + "mean_token_accuracy": 0.7829477787017822, + "num_tokens": 64998761.0, + "step": 1700 + }, + { + "epoch": 0.2163846838824577, + "grad_norm": 55.5, + "learning_rate": 7.206443408223823e-07, + "loss": 0.7827, + "mean_token_accuracy": 0.7770757079124451, + "num_tokens": 65038849.0, + "step": 1701 + }, + { + "epoch": 0.2165118941610482, + "grad_norm": 75.5, + "learning_rate": 7.210682492581603e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.7792483568191528, + "num_tokens": 65071130.0, + "step": 1702 + }, + { + "epoch": 0.2166391044396387, + "grad_norm": 191.0, + "learning_rate": 7.214921576939381e-07, + "loss": 0.8237, + "mean_token_accuracy": 0.7595884203910828, + "num_tokens": 65111619.0, + "step": 1703 + }, + { + "epoch": 0.21676631471822924, + "grad_norm": 76.0, + "learning_rate": 7.219160661297159e-07, + "loss": 0.7427, + "mean_token_accuracy": 0.7859623432159424, + "num_tokens": 65147537.0, + "step": 1704 + }, + { + "epoch": 0.21689352499681974, + "grad_norm": 88.5, + "learning_rate": 7.223399745654938e-07, + "loss": 0.7818, + "mean_token_accuracy": 0.7766017913818359, + "num_tokens": 65187254.0, + "step": 1705 + }, + { + "epoch": 0.21702073527541024, + "grad_norm": 30.875, + "learning_rate": 7.227638830012717e-07, + "loss": 0.7991, + "mean_token_accuracy": 0.7694731950759888, + "num_tokens": 65226130.0, + "step": 1706 + }, + { + "epoch": 0.21714794555400077, + "grad_norm": 80.5, + "learning_rate": 7.231877914370495e-07, + "loss": 0.7626, + "mean_token_accuracy": 0.7837001085281372, + "num_tokens": 65261282.0, + "step": 1707 + }, + { + "epoch": 0.21727515583259127, + "grad_norm": 66.5, + "learning_rate": 7.236116998728275e-07, + "loss": 0.8258, + "mean_token_accuracy": 0.765583336353302, + "num_tokens": 65294850.0, + "step": 1708 + }, + { + "epoch": 0.21740236611118177, + "grad_norm": 48.75, + "learning_rate": 7.240356083086053e-07, + "loss": 0.6789, + "mean_token_accuracy": 0.803613007068634, + "num_tokens": 65332323.0, + "step": 1709 + }, + { + "epoch": 0.2175295763897723, + "grad_norm": 115.0, + "learning_rate": 7.244595167443833e-07, + "loss": 0.7963, + "mean_token_accuracy": 0.7731828093528748, + "num_tokens": 65376033.0, + "step": 1710 + }, + { + "epoch": 0.2176567866683628, + "grad_norm": 36.5, + "learning_rate": 7.248834251801611e-07, + "loss": 0.7427, + "mean_token_accuracy": 0.7904706597328186, + "num_tokens": 65417875.0, + "step": 1711 + }, + { + "epoch": 0.2177839969469533, + "grad_norm": 27.25, + "learning_rate": 7.253073336159388e-07, + "loss": 0.7182, + "mean_token_accuracy": 0.7925023436546326, + "num_tokens": 65456675.0, + "step": 1712 + }, + { + "epoch": 0.21791120722554383, + "grad_norm": 16.125, + "learning_rate": 7.257312420517168e-07, + "loss": 0.7283, + "mean_token_accuracy": 0.7899801731109619, + "num_tokens": 65487699.0, + "step": 1713 + }, + { + "epoch": 0.21803841750413433, + "grad_norm": 33.0, + "learning_rate": 7.261551504874946e-07, + "loss": 0.7459, + "mean_token_accuracy": 0.7812972068786621, + "num_tokens": 65526143.0, + "step": 1714 + }, + { + "epoch": 0.21816562778272486, + "grad_norm": 12.0, + "learning_rate": 7.265790589232725e-07, + "loss": 0.7532, + "mean_token_accuracy": 0.7827807664871216, + "num_tokens": 65562969.0, + "step": 1715 + }, + { + "epoch": 0.21829283806131536, + "grad_norm": 152.0, + "learning_rate": 7.270029673590504e-07, + "loss": 0.7813, + "mean_token_accuracy": 0.7768987417221069, + "num_tokens": 65605999.0, + "step": 1716 + }, + { + "epoch": 0.21842004833990586, + "grad_norm": 104.0, + "learning_rate": 7.274268757948283e-07, + "loss": 0.8394, + "mean_token_accuracy": 0.7641916275024414, + "num_tokens": 65644877.0, + "step": 1717 + }, + { + "epoch": 0.2185472586184964, + "grad_norm": 145.0, + "learning_rate": 7.278507842306062e-07, + "loss": 0.7729, + "mean_token_accuracy": 0.7816959619522095, + "num_tokens": 65678823.0, + "step": 1718 + }, + { + "epoch": 0.2186744688970869, + "grad_norm": 53.75, + "learning_rate": 7.282746926663841e-07, + "loss": 0.8423, + "mean_token_accuracy": 0.7577548027038574, + "num_tokens": 65717185.0, + "step": 1719 + }, + { + "epoch": 0.2188016791756774, + "grad_norm": 25.5, + "learning_rate": 7.286986011021618e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7886351346969604, + "num_tokens": 65754422.0, + "step": 1720 + }, + { + "epoch": 0.21892888945426792, + "grad_norm": 103.0, + "learning_rate": 7.291225095379398e-07, + "loss": 0.7682, + "mean_token_accuracy": 0.7824004888534546, + "num_tokens": 65799390.0, + "step": 1721 + }, + { + "epoch": 0.21905609973285842, + "grad_norm": 205.0, + "learning_rate": 7.295464179737176e-07, + "loss": 0.7986, + "mean_token_accuracy": 0.7776158452033997, + "num_tokens": 65837566.0, + "step": 1722 + }, + { + "epoch": 0.21918331001144892, + "grad_norm": 93.0, + "learning_rate": 7.299703264094955e-07, + "loss": 0.7849, + "mean_token_accuracy": 0.7793960571289062, + "num_tokens": 65874537.0, + "step": 1723 + }, + { + "epoch": 0.21931052029003945, + "grad_norm": 143.0, + "learning_rate": 7.303942348452734e-07, + "loss": 0.7994, + "mean_token_accuracy": 0.774758517742157, + "num_tokens": 65915616.0, + "step": 1724 + }, + { + "epoch": 0.21943773056862995, + "grad_norm": 51.0, + "learning_rate": 7.308181432810513e-07, + "loss": 0.7435, + "mean_token_accuracy": 0.7872070074081421, + "num_tokens": 65954263.0, + "step": 1725 + }, + { + "epoch": 0.21956494084722045, + "grad_norm": 241.0, + "learning_rate": 7.312420517168292e-07, + "loss": 0.7753, + "mean_token_accuracy": 0.7768526077270508, + "num_tokens": 65992880.0, + "step": 1726 + }, + { + "epoch": 0.21969215112581097, + "grad_norm": 179.0, + "learning_rate": 7.31665960152607e-07, + "loss": 0.7974, + "mean_token_accuracy": 0.7702053785324097, + "num_tokens": 66033958.0, + "step": 1727 + }, + { + "epoch": 0.21981936140440148, + "grad_norm": 144.0, + "learning_rate": 7.320898685883848e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.7821962833404541, + "num_tokens": 66072448.0, + "step": 1728 + }, + { + "epoch": 0.21994657168299198, + "grad_norm": 118.0, + "learning_rate": 7.325137770241628e-07, + "loss": 0.8905, + "mean_token_accuracy": 0.7498212456703186, + "num_tokens": 66112648.0, + "step": 1729 + }, + { + "epoch": 0.2200737819615825, + "grad_norm": 23.125, + "learning_rate": 7.329376854599406e-07, + "loss": 0.7737, + "mean_token_accuracy": 0.7803706526756287, + "num_tokens": 66154791.0, + "step": 1730 + }, + { + "epoch": 0.220200992240173, + "grad_norm": 41.5, + "learning_rate": 7.333615938957184e-07, + "loss": 0.7736, + "mean_token_accuracy": 0.779744029045105, + "num_tokens": 66192450.0, + "step": 1731 + }, + { + "epoch": 0.2203282025187635, + "grad_norm": 39.25, + "learning_rate": 7.337855023314964e-07, + "loss": 0.7671, + "mean_token_accuracy": 0.7786252498626709, + "num_tokens": 66231173.0, + "step": 1732 + }, + { + "epoch": 0.22045541279735403, + "grad_norm": 80.0, + "learning_rate": 7.342094107672742e-07, + "loss": 0.8315, + "mean_token_accuracy": 0.7634025812149048, + "num_tokens": 66272956.0, + "step": 1733 + }, + { + "epoch": 0.22058262307594453, + "grad_norm": 57.75, + "learning_rate": 7.346333192030522e-07, + "loss": 0.7752, + "mean_token_accuracy": 0.7827286720275879, + "num_tokens": 66309450.0, + "step": 1734 + }, + { + "epoch": 0.22070983335453503, + "grad_norm": 64.0, + "learning_rate": 7.350572276388299e-07, + "loss": 0.84, + "mean_token_accuracy": 0.7607845067977905, + "num_tokens": 66350947.0, + "step": 1735 + }, + { + "epoch": 0.22083704363312556, + "grad_norm": 44.75, + "learning_rate": 7.354811360746078e-07, + "loss": 0.7838, + "mean_token_accuracy": 0.7725387811660767, + "num_tokens": 66388364.0, + "step": 1736 + }, + { + "epoch": 0.22096425391171606, + "grad_norm": 114.5, + "learning_rate": 7.359050445103857e-07, + "loss": 0.8112, + "mean_token_accuracy": 0.7700737118721008, + "num_tokens": 66425029.0, + "step": 1737 + }, + { + "epoch": 0.22109146419030656, + "grad_norm": 59.75, + "learning_rate": 7.363289529461636e-07, + "loss": 0.7691, + "mean_token_accuracy": 0.7806204557418823, + "num_tokens": 66463924.0, + "step": 1738 + }, + { + "epoch": 0.2212186744688971, + "grad_norm": 21.5, + "learning_rate": 7.367528613819415e-07, + "loss": 0.7128, + "mean_token_accuracy": 0.7956492900848389, + "num_tokens": 66504824.0, + "step": 1739 + }, + { + "epoch": 0.2213458847474876, + "grad_norm": 15.875, + "learning_rate": 7.371767698177194e-07, + "loss": 0.7351, + "mean_token_accuracy": 0.7853819131851196, + "num_tokens": 66539308.0, + "step": 1740 + }, + { + "epoch": 0.22147309502607812, + "grad_norm": 97.5, + "learning_rate": 7.376006782534972e-07, + "loss": 0.7447, + "mean_token_accuracy": 0.7895525693893433, + "num_tokens": 66572785.0, + "step": 1741 + }, + { + "epoch": 0.22160030530466862, + "grad_norm": 103.0, + "learning_rate": 7.380245866892751e-07, + "loss": 0.7568, + "mean_token_accuracy": 0.7818760871887207, + "num_tokens": 66607890.0, + "step": 1742 + }, + { + "epoch": 0.22172751558325912, + "grad_norm": 26.25, + "learning_rate": 7.384484951250529e-07, + "loss": 0.7547, + "mean_token_accuracy": 0.7868102788925171, + "num_tokens": 66650553.0, + "step": 1743 + }, + { + "epoch": 0.22185472586184965, + "grad_norm": 172.0, + "learning_rate": 7.388724035608308e-07, + "loss": 0.7539, + "mean_token_accuracy": 0.7852430939674377, + "num_tokens": 66685330.0, + "step": 1744 + }, + { + "epoch": 0.22198193614044015, + "grad_norm": 133.0, + "learning_rate": 7.392963119966087e-07, + "loss": 0.7495, + "mean_token_accuracy": 0.7845652103424072, + "num_tokens": 66716816.0, + "step": 1745 + }, + { + "epoch": 0.22210914641903065, + "grad_norm": 149.0, + "learning_rate": 7.397202204323866e-07, + "loss": 0.7892, + "mean_token_accuracy": 0.7755517959594727, + "num_tokens": 66754030.0, + "step": 1746 + }, + { + "epoch": 0.22223635669762118, + "grad_norm": 52.25, + "learning_rate": 7.401441288681645e-07, + "loss": 0.7115, + "mean_token_accuracy": 0.7944361567497253, + "num_tokens": 66793650.0, + "step": 1747 + }, + { + "epoch": 0.22236356697621168, + "grad_norm": 17.125, + "learning_rate": 7.405680373039424e-07, + "loss": 0.7103, + "mean_token_accuracy": 0.7981925010681152, + "num_tokens": 66833709.0, + "step": 1748 + }, + { + "epoch": 0.22249077725480218, + "grad_norm": 24.125, + "learning_rate": 7.409919457397202e-07, + "loss": 0.8065, + "mean_token_accuracy": 0.7693134546279907, + "num_tokens": 66866456.0, + "step": 1749 + }, + { + "epoch": 0.2226179875333927, + "grad_norm": 292.0, + "learning_rate": 7.414158541754981e-07, + "loss": 0.8463, + "mean_token_accuracy": 0.7615289688110352, + "num_tokens": 66905875.0, + "step": 1750 + }, + { + "epoch": 0.2227451978119832, + "grad_norm": 228.0, + "learning_rate": 7.418397626112759e-07, + "loss": 0.7834, + "mean_token_accuracy": 0.7806596755981445, + "num_tokens": 66936009.0, + "step": 1751 + }, + { + "epoch": 0.2228724080905737, + "grad_norm": 214.0, + "learning_rate": 7.422636710470537e-07, + "loss": 0.779, + "mean_token_accuracy": 0.7794634103775024, + "num_tokens": 66978111.0, + "step": 1752 + }, + { + "epoch": 0.22299961836916424, + "grad_norm": 137.0, + "learning_rate": 7.426875794828317e-07, + "loss": 0.7925, + "mean_token_accuracy": 0.7798542976379395, + "num_tokens": 67015938.0, + "step": 1753 + }, + { + "epoch": 0.22312682864775474, + "grad_norm": 204.0, + "learning_rate": 7.431114879186095e-07, + "loss": 0.8163, + "mean_token_accuracy": 0.7710492610931396, + "num_tokens": 67055456.0, + "step": 1754 + }, + { + "epoch": 0.22325403892634524, + "grad_norm": 228.0, + "learning_rate": 7.435353963543875e-07, + "loss": 0.767, + "mean_token_accuracy": 0.7840178608894348, + "num_tokens": 67096298.0, + "step": 1755 + }, + { + "epoch": 0.22338124920493577, + "grad_norm": 180.0, + "learning_rate": 7.439593047901653e-07, + "loss": 0.7223, + "mean_token_accuracy": 0.7917342782020569, + "num_tokens": 67135280.0, + "step": 1756 + }, + { + "epoch": 0.22350845948352627, + "grad_norm": 122.0, + "learning_rate": 7.443832132259431e-07, + "loss": 0.7977, + "mean_token_accuracy": 0.7702471017837524, + "num_tokens": 67171936.0, + "step": 1757 + }, + { + "epoch": 0.22363566976211677, + "grad_norm": 60.0, + "learning_rate": 7.44807121661721e-07, + "loss": 0.7854, + "mean_token_accuracy": 0.7711290121078491, + "num_tokens": 67209854.0, + "step": 1758 + }, + { + "epoch": 0.2237628800407073, + "grad_norm": 31.25, + "learning_rate": 7.452310300974989e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7848998308181763, + "num_tokens": 67244369.0, + "step": 1759 + }, + { + "epoch": 0.2238900903192978, + "grad_norm": 294.0, + "learning_rate": 7.456549385332767e-07, + "loss": 0.8304, + "mean_token_accuracy": 0.7636831998825073, + "num_tokens": 67283994.0, + "step": 1760 + }, + { + "epoch": 0.2240173005978883, + "grad_norm": 49.25, + "learning_rate": 7.460788469690547e-07, + "loss": 0.6665, + "mean_token_accuracy": 0.8073630332946777, + "num_tokens": 67323230.0, + "step": 1761 + }, + { + "epoch": 0.22414451087647883, + "grad_norm": 219.0, + "learning_rate": 7.465027554048325e-07, + "loss": 0.8725, + "mean_token_accuracy": 0.7539811730384827, + "num_tokens": 67366198.0, + "step": 1762 + }, + { + "epoch": 0.22427172115506933, + "grad_norm": 308.0, + "learning_rate": 7.469266638406105e-07, + "loss": 0.8589, + "mean_token_accuracy": 0.7600100040435791, + "num_tokens": 67410685.0, + "step": 1763 + }, + { + "epoch": 0.22439893143365983, + "grad_norm": 212.0, + "learning_rate": 7.473505722763883e-07, + "loss": 0.7656, + "mean_token_accuracy": 0.7834604382514954, + "num_tokens": 67445818.0, + "step": 1764 + }, + { + "epoch": 0.22452614171225035, + "grad_norm": 48.5, + "learning_rate": 7.477744807121661e-07, + "loss": 0.7981, + "mean_token_accuracy": 0.7715897560119629, + "num_tokens": 67483873.0, + "step": 1765 + }, + { + "epoch": 0.22465335199084085, + "grad_norm": 26.0, + "learning_rate": 7.48198389147944e-07, + "loss": 0.7782, + "mean_token_accuracy": 0.7818137407302856, + "num_tokens": 67519524.0, + "step": 1766 + }, + { + "epoch": 0.22478056226943138, + "grad_norm": 58.5, + "learning_rate": 7.486222975837219e-07, + "loss": 0.7683, + "mean_token_accuracy": 0.7789512872695923, + "num_tokens": 67560920.0, + "step": 1767 + }, + { + "epoch": 0.22490777254802188, + "grad_norm": 51.75, + "learning_rate": 7.490462060194997e-07, + "loss": 0.7299, + "mean_token_accuracy": 0.787627100944519, + "num_tokens": 67594948.0, + "step": 1768 + }, + { + "epoch": 0.22503498282661238, + "grad_norm": 90.0, + "learning_rate": 7.494701144552777e-07, + "loss": 0.8476, + "mean_token_accuracy": 0.760198175907135, + "num_tokens": 67628867.0, + "step": 1769 + }, + { + "epoch": 0.2251621931052029, + "grad_norm": 168.0, + "learning_rate": 7.498940228910555e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.773635745048523, + "num_tokens": 67669762.0, + "step": 1770 + }, + { + "epoch": 0.2252894033837934, + "grad_norm": 198.0, + "learning_rate": 7.503179313268335e-07, + "loss": 0.7985, + "mean_token_accuracy": 0.7732564210891724, + "num_tokens": 67704750.0, + "step": 1771 + }, + { + "epoch": 0.2254166136623839, + "grad_norm": 211.0, + "learning_rate": 7.507418397626113e-07, + "loss": 0.8276, + "mean_token_accuracy": 0.7677184343338013, + "num_tokens": 67745181.0, + "step": 1772 + }, + { + "epoch": 0.22554382394097444, + "grad_norm": 175.0, + "learning_rate": 7.51165748198389e-07, + "loss": 0.7997, + "mean_token_accuracy": 0.7706862092018127, + "num_tokens": 67785929.0, + "step": 1773 + }, + { + "epoch": 0.22567103421956494, + "grad_norm": 32.5, + "learning_rate": 7.51589656634167e-07, + "loss": 0.744, + "mean_token_accuracy": 0.7831761837005615, + "num_tokens": 67821887.0, + "step": 1774 + }, + { + "epoch": 0.22579824449815544, + "grad_norm": 56.25, + "learning_rate": 7.520135650699448e-07, + "loss": 0.7872, + "mean_token_accuracy": 0.7728400826454163, + "num_tokens": 67859608.0, + "step": 1775 + }, + { + "epoch": 0.22592545477674597, + "grad_norm": 90.5, + "learning_rate": 7.524374735057227e-07, + "loss": 0.7973, + "mean_token_accuracy": 0.7751860618591309, + "num_tokens": 67895287.0, + "step": 1776 + }, + { + "epoch": 0.22605266505533647, + "grad_norm": 14.3125, + "learning_rate": 7.528613819415006e-07, + "loss": 0.7429, + "mean_token_accuracy": 0.7902317047119141, + "num_tokens": 67930358.0, + "step": 1777 + }, + { + "epoch": 0.22617987533392697, + "grad_norm": 127.5, + "learning_rate": 7.532852903772785e-07, + "loss": 0.7986, + "mean_token_accuracy": 0.7731097936630249, + "num_tokens": 67966112.0, + "step": 1778 + }, + { + "epoch": 0.2263070856125175, + "grad_norm": 139.0, + "learning_rate": 7.537091988130564e-07, + "loss": 0.8201, + "mean_token_accuracy": 0.7649648189544678, + "num_tokens": 68001297.0, + "step": 1779 + }, + { + "epoch": 0.226434295891108, + "grad_norm": 198.0, + "learning_rate": 7.541331072488342e-07, + "loss": 0.7867, + "mean_token_accuracy": 0.7757405042648315, + "num_tokens": 68037346.0, + "step": 1780 + }, + { + "epoch": 0.2265615061696985, + "grad_norm": 166.0, + "learning_rate": 7.54557015684612e-07, + "loss": 0.7685, + "mean_token_accuracy": 0.7797923684120178, + "num_tokens": 68076566.0, + "step": 1781 + }, + { + "epoch": 0.22668871644828903, + "grad_norm": 56.25, + "learning_rate": 7.5498092412039e-07, + "loss": 0.7352, + "mean_token_accuracy": 0.789191484451294, + "num_tokens": 68116834.0, + "step": 1782 + }, + { + "epoch": 0.22681592672687953, + "grad_norm": 13.8125, + "learning_rate": 7.554048325561678e-07, + "loss": 0.7123, + "mean_token_accuracy": 0.7936414480209351, + "num_tokens": 68155573.0, + "step": 1783 + }, + { + "epoch": 0.22694313700547003, + "grad_norm": 11.0, + "learning_rate": 7.558287409919457e-07, + "loss": 0.8358, + "mean_token_accuracy": 0.7682726979255676, + "num_tokens": 68193378.0, + "step": 1784 + }, + { + "epoch": 0.22707034728406056, + "grad_norm": 13.5625, + "learning_rate": 7.562526494277236e-07, + "loss": 0.8067, + "mean_token_accuracy": 0.7698858976364136, + "num_tokens": 68230183.0, + "step": 1785 + }, + { + "epoch": 0.22719755756265106, + "grad_norm": 21.75, + "learning_rate": 7.566765578635015e-07, + "loss": 0.7611, + "mean_token_accuracy": 0.7840587496757507, + "num_tokens": 68270526.0, + "step": 1786 + }, + { + "epoch": 0.22732476784124156, + "grad_norm": 115.5, + "learning_rate": 7.571004662992794e-07, + "loss": 0.7991, + "mean_token_accuracy": 0.7749987840652466, + "num_tokens": 68308690.0, + "step": 1787 + }, + { + "epoch": 0.2274519781198321, + "grad_norm": 17.625, + "learning_rate": 7.575243747350572e-07, + "loss": 0.8377, + "mean_token_accuracy": 0.7632801532745361, + "num_tokens": 68348660.0, + "step": 1788 + }, + { + "epoch": 0.2275791883984226, + "grad_norm": 70.5, + "learning_rate": 7.57948283170835e-07, + "loss": 0.7597, + "mean_token_accuracy": 0.7828409671783447, + "num_tokens": 68389129.0, + "step": 1789 + }, + { + "epoch": 0.2277063986770131, + "grad_norm": 152.0, + "learning_rate": 7.58372191606613e-07, + "loss": 0.78, + "mean_token_accuracy": 0.7798647880554199, + "num_tokens": 68423234.0, + "step": 1790 + }, + { + "epoch": 0.22783360895560362, + "grad_norm": 71.0, + "learning_rate": 7.587961000423908e-07, + "loss": 0.7725, + "mean_token_accuracy": 0.7782326340675354, + "num_tokens": 68460834.0, + "step": 1791 + }, + { + "epoch": 0.22796081923419412, + "grad_norm": 48.25, + "learning_rate": 7.592200084781686e-07, + "loss": 0.8308, + "mean_token_accuracy": 0.7654589414596558, + "num_tokens": 68495605.0, + "step": 1792 + }, + { + "epoch": 0.22808802951278465, + "grad_norm": 18.625, + "learning_rate": 7.596439169139466e-07, + "loss": 0.7713, + "mean_token_accuracy": 0.7793329954147339, + "num_tokens": 68530838.0, + "step": 1793 + }, + { + "epoch": 0.22821523979137515, + "grad_norm": 174.0, + "learning_rate": 7.600678253497244e-07, + "loss": 0.7581, + "mean_token_accuracy": 0.7837645411491394, + "num_tokens": 68571225.0, + "step": 1794 + }, + { + "epoch": 0.22834245006996565, + "grad_norm": 108.0, + "learning_rate": 7.604917337855023e-07, + "loss": 0.7408, + "mean_token_accuracy": 0.7850505709648132, + "num_tokens": 68608284.0, + "step": 1795 + }, + { + "epoch": 0.22846966034855618, + "grad_norm": 204.0, + "learning_rate": 7.609156422212801e-07, + "loss": 0.7123, + "mean_token_accuracy": 0.7967768907546997, + "num_tokens": 68645943.0, + "step": 1796 + }, + { + "epoch": 0.22859687062714668, + "grad_norm": 59.5, + "learning_rate": 7.61339550657058e-07, + "loss": 0.7373, + "mean_token_accuracy": 0.7849788665771484, + "num_tokens": 68684245.0, + "step": 1797 + }, + { + "epoch": 0.22872408090573718, + "grad_norm": 29.25, + "learning_rate": 7.617634590928359e-07, + "loss": 0.7802, + "mean_token_accuracy": 0.7781566381454468, + "num_tokens": 68717532.0, + "step": 1798 + }, + { + "epoch": 0.2288512911843277, + "grad_norm": 206.0, + "learning_rate": 7.621873675286138e-07, + "loss": 0.7857, + "mean_token_accuracy": 0.7741365432739258, + "num_tokens": 68759515.0, + "step": 1799 + }, + { + "epoch": 0.2289785014629182, + "grad_norm": 38.75, + "learning_rate": 7.626112759643916e-07, + "loss": 0.8005, + "mean_token_accuracy": 0.7698000073432922, + "num_tokens": 68796969.0, + "step": 1800 + }, + { + "epoch": 0.2291057117415087, + "grad_norm": 15.4375, + "learning_rate": 7.630351844001696e-07, + "loss": 0.7851, + "mean_token_accuracy": 0.7741513848304749, + "num_tokens": 68831138.0, + "step": 1801 + }, + { + "epoch": 0.22923292202009923, + "grad_norm": 156.0, + "learning_rate": 7.634590928359474e-07, + "loss": 0.7811, + "mean_token_accuracy": 0.778712809085846, + "num_tokens": 68868741.0, + "step": 1802 + }, + { + "epoch": 0.22936013229868973, + "grad_norm": 127.0, + "learning_rate": 7.638830012717253e-07, + "loss": 0.7875, + "mean_token_accuracy": 0.7766610383987427, + "num_tokens": 68909191.0, + "step": 1803 + }, + { + "epoch": 0.22948734257728023, + "grad_norm": 172.0, + "learning_rate": 7.643069097075031e-07, + "loss": 0.7581, + "mean_token_accuracy": 0.7819129228591919, + "num_tokens": 68949261.0, + "step": 1804 + }, + { + "epoch": 0.22961455285587076, + "grad_norm": 21.375, + "learning_rate": 7.64730818143281e-07, + "loss": 0.7021, + "mean_token_accuracy": 0.7951422929763794, + "num_tokens": 68990434.0, + "step": 1805 + }, + { + "epoch": 0.22974176313446126, + "grad_norm": 40.5, + "learning_rate": 7.651547265790589e-07, + "loss": 0.8231, + "mean_token_accuracy": 0.7674230337142944, + "num_tokens": 69029482.0, + "step": 1806 + }, + { + "epoch": 0.22986897341305176, + "grad_norm": 43.5, + "learning_rate": 7.655786350148368e-07, + "loss": 0.7883, + "mean_token_accuracy": 0.7761234045028687, + "num_tokens": 69069766.0, + "step": 1807 + }, + { + "epoch": 0.2299961836916423, + "grad_norm": 126.0, + "learning_rate": 7.660025434506146e-07, + "loss": 0.7695, + "mean_token_accuracy": 0.7816143035888672, + "num_tokens": 69114592.0, + "step": 1808 + }, + { + "epoch": 0.2301233939702328, + "grad_norm": 31.75, + "learning_rate": 7.664264518863926e-07, + "loss": 0.7655, + "mean_token_accuracy": 0.7817103266716003, + "num_tokens": 69152599.0, + "step": 1809 + }, + { + "epoch": 0.2302506042488233, + "grad_norm": 83.0, + "learning_rate": 7.668503603221704e-07, + "loss": 0.8267, + "mean_token_accuracy": 0.75914067029953, + "num_tokens": 69193157.0, + "step": 1810 + }, + { + "epoch": 0.23037781452741382, + "grad_norm": 45.25, + "learning_rate": 7.672742687579483e-07, + "loss": 0.7416, + "mean_token_accuracy": 0.783042311668396, + "num_tokens": 69230681.0, + "step": 1811 + }, + { + "epoch": 0.23050502480600432, + "grad_norm": 80.0, + "learning_rate": 7.676981771937261e-07, + "loss": 0.7153, + "mean_token_accuracy": 0.788009762763977, + "num_tokens": 69261292.0, + "step": 1812 + }, + { + "epoch": 0.23063223508459482, + "grad_norm": 16.875, + "learning_rate": 7.681220856295039e-07, + "loss": 0.7273, + "mean_token_accuracy": 0.7900620698928833, + "num_tokens": 69301795.0, + "step": 1813 + }, + { + "epoch": 0.23075944536318535, + "grad_norm": 9.9375, + "learning_rate": 7.685459940652819e-07, + "loss": 0.7743, + "mean_token_accuracy": 0.77686607837677, + "num_tokens": 69337038.0, + "step": 1814 + }, + { + "epoch": 0.23088665564177585, + "grad_norm": 95.0, + "learning_rate": 7.689699025010597e-07, + "loss": 0.7645, + "mean_token_accuracy": 0.7811826467514038, + "num_tokens": 69379871.0, + "step": 1815 + }, + { + "epoch": 0.23101386592036638, + "grad_norm": 79.0, + "learning_rate": 7.693938109368376e-07, + "loss": 0.8014, + "mean_token_accuracy": 0.7689715027809143, + "num_tokens": 69418021.0, + "step": 1816 + }, + { + "epoch": 0.23114107619895688, + "grad_norm": 154.0, + "learning_rate": 7.698177193726155e-07, + "loss": 0.7597, + "mean_token_accuracy": 0.7768421769142151, + "num_tokens": 69455574.0, + "step": 1817 + }, + { + "epoch": 0.23126828647754738, + "grad_norm": 214.0, + "learning_rate": 7.702416278083933e-07, + "loss": 0.8179, + "mean_token_accuracy": 0.7654674053192139, + "num_tokens": 69489822.0, + "step": 1818 + }, + { + "epoch": 0.2313954967561379, + "grad_norm": 96.0, + "learning_rate": 7.706655362441712e-07, + "loss": 0.7596, + "mean_token_accuracy": 0.7819409370422363, + "num_tokens": 69532821.0, + "step": 1819 + }, + { + "epoch": 0.2315227070347284, + "grad_norm": 20.25, + "learning_rate": 7.710894446799491e-07, + "loss": 0.7455, + "mean_token_accuracy": 0.7847508192062378, + "num_tokens": 69565674.0, + "step": 1820 + }, + { + "epoch": 0.2316499173133189, + "grad_norm": 150.0, + "learning_rate": 7.715133531157269e-07, + "loss": 0.728, + "mean_token_accuracy": 0.7930282354354858, + "num_tokens": 69606118.0, + "step": 1821 + }, + { + "epoch": 0.23177712759190944, + "grad_norm": 61.0, + "learning_rate": 7.719372615515049e-07, + "loss": 0.8281, + "mean_token_accuracy": 0.765401303768158, + "num_tokens": 69650102.0, + "step": 1822 + }, + { + "epoch": 0.23190433787049994, + "grad_norm": 141.0, + "learning_rate": 7.723611699872827e-07, + "loss": 0.7611, + "mean_token_accuracy": 0.7834722995758057, + "num_tokens": 69693440.0, + "step": 1823 + }, + { + "epoch": 0.23203154814909044, + "grad_norm": 71.0, + "learning_rate": 7.727850784230606e-07, + "loss": 0.787, + "mean_token_accuracy": 0.771787166595459, + "num_tokens": 69732651.0, + "step": 1824 + }, + { + "epoch": 0.23215875842768097, + "grad_norm": 93.0, + "learning_rate": 7.732089868588385e-07, + "loss": 0.7958, + "mean_token_accuracy": 0.7729930877685547, + "num_tokens": 69769915.0, + "step": 1825 + }, + { + "epoch": 0.23228596870627147, + "grad_norm": 63.25, + "learning_rate": 7.736328952946163e-07, + "loss": 0.6905, + "mean_token_accuracy": 0.8035904169082642, + "num_tokens": 69813943.0, + "step": 1826 + }, + { + "epoch": 0.23241317898486197, + "grad_norm": 100.5, + "learning_rate": 7.740568037303942e-07, + "loss": 0.7432, + "mean_token_accuracy": 0.7806854248046875, + "num_tokens": 69847039.0, + "step": 1827 + }, + { + "epoch": 0.2325403892634525, + "grad_norm": 45.25, + "learning_rate": 7.744807121661721e-07, + "loss": 0.712, + "mean_token_accuracy": 0.798412561416626, + "num_tokens": 69884588.0, + "step": 1828 + }, + { + "epoch": 0.232667599542043, + "grad_norm": 74.5, + "learning_rate": 7.749046206019499e-07, + "loss": 0.7678, + "mean_token_accuracy": 0.7827680110931396, + "num_tokens": 69925552.0, + "step": 1829 + }, + { + "epoch": 0.2327948098206335, + "grad_norm": 21.125, + "learning_rate": 7.753285290377279e-07, + "loss": 0.699, + "mean_token_accuracy": 0.797965407371521, + "num_tokens": 69961776.0, + "step": 1830 + }, + { + "epoch": 0.23292202009922403, + "grad_norm": 73.0, + "learning_rate": 7.757524374735057e-07, + "loss": 0.8103, + "mean_token_accuracy": 0.7676131725311279, + "num_tokens": 69999775.0, + "step": 1831 + }, + { + "epoch": 0.23304923037781453, + "grad_norm": 68.5, + "learning_rate": 7.761763459092836e-07, + "loss": 0.7194, + "mean_token_accuracy": 0.793714702129364, + "num_tokens": 70037294.0, + "step": 1832 + }, + { + "epoch": 0.23317644065640503, + "grad_norm": 50.25, + "learning_rate": 7.766002543450614e-07, + "loss": 0.7869, + "mean_token_accuracy": 0.7756221890449524, + "num_tokens": 70075014.0, + "step": 1833 + }, + { + "epoch": 0.23330365093499555, + "grad_norm": 21.5, + "learning_rate": 7.770241627808392e-07, + "loss": 0.6931, + "mean_token_accuracy": 0.8011385798454285, + "num_tokens": 70112186.0, + "step": 1834 + }, + { + "epoch": 0.23343086121358606, + "grad_norm": 37.5, + "learning_rate": 7.774480712166172e-07, + "loss": 0.804, + "mean_token_accuracy": 0.770804762840271, + "num_tokens": 70153285.0, + "step": 1835 + }, + { + "epoch": 0.23355807149217656, + "grad_norm": 22.875, + "learning_rate": 7.77871979652395e-07, + "loss": 0.7789, + "mean_token_accuracy": 0.777158260345459, + "num_tokens": 70193382.0, + "step": 1836 + }, + { + "epoch": 0.23368528177076708, + "grad_norm": 52.5, + "learning_rate": 7.782958880881729e-07, + "loss": 0.7933, + "mean_token_accuracy": 0.776249885559082, + "num_tokens": 70227279.0, + "step": 1837 + }, + { + "epoch": 0.23381249204935758, + "grad_norm": 19.125, + "learning_rate": 7.787197965239508e-07, + "loss": 0.7077, + "mean_token_accuracy": 0.7933639287948608, + "num_tokens": 70264395.0, + "step": 1838 + }, + { + "epoch": 0.23393970232794808, + "grad_norm": 48.75, + "learning_rate": 7.791437049597287e-07, + "loss": 0.8693, + "mean_token_accuracy": 0.7540113925933838, + "num_tokens": 70295016.0, + "step": 1839 + }, + { + "epoch": 0.2340669126065386, + "grad_norm": 55.25, + "learning_rate": 7.795676133955065e-07, + "loss": 0.8014, + "mean_token_accuracy": 0.7699359655380249, + "num_tokens": 70332774.0, + "step": 1840 + }, + { + "epoch": 0.2341941228851291, + "grad_norm": 23.25, + "learning_rate": 7.799915218312844e-07, + "loss": 0.7821, + "mean_token_accuracy": 0.7734217643737793, + "num_tokens": 70366880.0, + "step": 1841 + }, + { + "epoch": 0.23432133316371964, + "grad_norm": 26.0, + "learning_rate": 7.804154302670622e-07, + "loss": 0.7418, + "mean_token_accuracy": 0.7871636748313904, + "num_tokens": 70405235.0, + "step": 1842 + }, + { + "epoch": 0.23444854344231014, + "grad_norm": 162.0, + "learning_rate": 7.808393387028402e-07, + "loss": 0.76, + "mean_token_accuracy": 0.7807170748710632, + "num_tokens": 70446503.0, + "step": 1843 + }, + { + "epoch": 0.23457575372090064, + "grad_norm": 76.0, + "learning_rate": 7.81263247138618e-07, + "loss": 0.8306, + "mean_token_accuracy": 0.7592816948890686, + "num_tokens": 70483514.0, + "step": 1844 + }, + { + "epoch": 0.23470296399949117, + "grad_norm": 159.0, + "learning_rate": 7.816871555743959e-07, + "loss": 0.7865, + "mean_token_accuracy": 0.7740010619163513, + "num_tokens": 70520944.0, + "step": 1845 + }, + { + "epoch": 0.23483017427808167, + "grad_norm": 90.5, + "learning_rate": 7.821110640101738e-07, + "loss": 0.7989, + "mean_token_accuracy": 0.7692992687225342, + "num_tokens": 70564399.0, + "step": 1846 + }, + { + "epoch": 0.23495738455667217, + "grad_norm": 56.25, + "learning_rate": 7.825349724459517e-07, + "loss": 0.7599, + "mean_token_accuracy": 0.7839975357055664, + "num_tokens": 70604000.0, + "step": 1847 + }, + { + "epoch": 0.2350845948352627, + "grad_norm": 84.5, + "learning_rate": 7.829588808817294e-07, + "loss": 0.7568, + "mean_token_accuracy": 0.7831102013587952, + "num_tokens": 70645133.0, + "step": 1848 + }, + { + "epoch": 0.2352118051138532, + "grad_norm": 39.25, + "learning_rate": 7.833827893175074e-07, + "loss": 0.8129, + "mean_token_accuracy": 0.7629249095916748, + "num_tokens": 70675824.0, + "step": 1849 + }, + { + "epoch": 0.2353390153924437, + "grad_norm": 13.0, + "learning_rate": 7.838066977532852e-07, + "loss": 0.7715, + "mean_token_accuracy": 0.7773535847663879, + "num_tokens": 70714103.0, + "step": 1850 + }, + { + "epoch": 0.23546622567103423, + "grad_norm": 36.75, + "learning_rate": 7.842306061890632e-07, + "loss": 0.8212, + "mean_token_accuracy": 0.7679744958877563, + "num_tokens": 70754265.0, + "step": 1851 + }, + { + "epoch": 0.23559343594962473, + "grad_norm": 71.5, + "learning_rate": 7.84654514624841e-07, + "loss": 0.7626, + "mean_token_accuracy": 0.7788552641868591, + "num_tokens": 70791368.0, + "step": 1852 + }, + { + "epoch": 0.23572064622821523, + "grad_norm": 30.0, + "learning_rate": 7.850784230606188e-07, + "loss": 0.8258, + "mean_token_accuracy": 0.7660733461380005, + "num_tokens": 70827397.0, + "step": 1853 + }, + { + "epoch": 0.23584785650680576, + "grad_norm": 22.75, + "learning_rate": 7.855023314963968e-07, + "loss": 0.7927, + "mean_token_accuracy": 0.7770745158195496, + "num_tokens": 70872503.0, + "step": 1854 + }, + { + "epoch": 0.23597506678539626, + "grad_norm": 59.5, + "learning_rate": 7.859262399321746e-07, + "loss": 0.7957, + "mean_token_accuracy": 0.7688372135162354, + "num_tokens": 70914376.0, + "step": 1855 + }, + { + "epoch": 0.23610227706398676, + "grad_norm": 116.5, + "learning_rate": 7.863501483679524e-07, + "loss": 0.7456, + "mean_token_accuracy": 0.7860703468322754, + "num_tokens": 70954004.0, + "step": 1856 + }, + { + "epoch": 0.2362294873425773, + "grad_norm": 49.0, + "learning_rate": 7.867740568037303e-07, + "loss": 0.7437, + "mean_token_accuracy": 0.7835278511047363, + "num_tokens": 70991765.0, + "step": 1857 + }, + { + "epoch": 0.2363566976211678, + "grad_norm": 50.5, + "learning_rate": 7.871979652395082e-07, + "loss": 0.7873, + "mean_token_accuracy": 0.7751988172531128, + "num_tokens": 71025478.0, + "step": 1858 + }, + { + "epoch": 0.2364839078997583, + "grad_norm": 18.25, + "learning_rate": 7.876218736752861e-07, + "loss": 0.7165, + "mean_token_accuracy": 0.797536313533783, + "num_tokens": 71067287.0, + "step": 1859 + }, + { + "epoch": 0.23661111817834882, + "grad_norm": 35.5, + "learning_rate": 7.88045782111064e-07, + "loss": 0.8073, + "mean_token_accuracy": 0.7711944580078125, + "num_tokens": 71108095.0, + "step": 1860 + }, + { + "epoch": 0.23673832845693932, + "grad_norm": 72.5, + "learning_rate": 7.884696905468418e-07, + "loss": 0.7034, + "mean_token_accuracy": 0.7965154051780701, + "num_tokens": 71150297.0, + "step": 1861 + }, + { + "epoch": 0.23686553873552982, + "grad_norm": 67.0, + "learning_rate": 7.888935989826198e-07, + "loss": 0.7841, + "mean_token_accuracy": 0.7732776403427124, + "num_tokens": 71189205.0, + "step": 1862 + }, + { + "epoch": 0.23699274901412035, + "grad_norm": 20.75, + "learning_rate": 7.893175074183976e-07, + "loss": 0.7625, + "mean_token_accuracy": 0.77875816822052, + "num_tokens": 71228141.0, + "step": 1863 + }, + { + "epoch": 0.23711995929271085, + "grad_norm": 45.5, + "learning_rate": 7.897414158541754e-07, + "loss": 0.7671, + "mean_token_accuracy": 0.779188871383667, + "num_tokens": 71267854.0, + "step": 1864 + }, + { + "epoch": 0.23724716957130135, + "grad_norm": 87.5, + "learning_rate": 7.901653242899533e-07, + "loss": 0.8234, + "mean_token_accuracy": 0.768576443195343, + "num_tokens": 71305030.0, + "step": 1865 + }, + { + "epoch": 0.23737437984989188, + "grad_norm": 50.5, + "learning_rate": 7.905892327257312e-07, + "loss": 0.7653, + "mean_token_accuracy": 0.7777368426322937, + "num_tokens": 71342202.0, + "step": 1866 + }, + { + "epoch": 0.23750159012848238, + "grad_norm": 58.5, + "learning_rate": 7.910131411615091e-07, + "loss": 0.758, + "mean_token_accuracy": 0.7850555181503296, + "num_tokens": 71379702.0, + "step": 1867 + }, + { + "epoch": 0.2376288004070729, + "grad_norm": 16.125, + "learning_rate": 7.91437049597287e-07, + "loss": 0.7595, + "mean_token_accuracy": 0.7825648784637451, + "num_tokens": 71417785.0, + "step": 1868 + }, + { + "epoch": 0.2377560106856634, + "grad_norm": 28.25, + "learning_rate": 7.918609580330648e-07, + "loss": 0.7954, + "mean_token_accuracy": 0.7728165984153748, + "num_tokens": 71461358.0, + "step": 1869 + }, + { + "epoch": 0.2378832209642539, + "grad_norm": 35.5, + "learning_rate": 7.922848664688428e-07, + "loss": 0.8047, + "mean_token_accuracy": 0.7730504274368286, + "num_tokens": 71500059.0, + "step": 1870 + }, + { + "epoch": 0.23801043124284443, + "grad_norm": 40.75, + "learning_rate": 7.927087749046205e-07, + "loss": 0.7024, + "mean_token_accuracy": 0.7988390922546387, + "num_tokens": 71537043.0, + "step": 1871 + }, + { + "epoch": 0.23813764152143493, + "grad_norm": 77.0, + "learning_rate": 7.931326833403983e-07, + "loss": 0.7657, + "mean_token_accuracy": 0.7801758050918579, + "num_tokens": 71570671.0, + "step": 1872 + }, + { + "epoch": 0.23826485180002543, + "grad_norm": 24.25, + "learning_rate": 7.935565917761763e-07, + "loss": 0.7501, + "mean_token_accuracy": 0.7849211692810059, + "num_tokens": 71614813.0, + "step": 1873 + }, + { + "epoch": 0.23839206207861596, + "grad_norm": 109.0, + "learning_rate": 7.939805002119541e-07, + "loss": 0.7897, + "mean_token_accuracy": 0.772493839263916, + "num_tokens": 71650441.0, + "step": 1874 + }, + { + "epoch": 0.23851927235720646, + "grad_norm": 201.0, + "learning_rate": 7.944044086477321e-07, + "loss": 0.7856, + "mean_token_accuracy": 0.7768890857696533, + "num_tokens": 71690425.0, + "step": 1875 + }, + { + "epoch": 0.23864648263579696, + "grad_norm": 65.0, + "learning_rate": 7.948283170835099e-07, + "loss": 0.7326, + "mean_token_accuracy": 0.7876547574996948, + "num_tokens": 71727349.0, + "step": 1876 + }, + { + "epoch": 0.2387736929143875, + "grad_norm": 130.0, + "learning_rate": 7.952522255192878e-07, + "loss": 0.6746, + "mean_token_accuracy": 0.8046471476554871, + "num_tokens": 71770386.0, + "step": 1877 + }, + { + "epoch": 0.238900903192978, + "grad_norm": 12.75, + "learning_rate": 7.956761339550657e-07, + "loss": 0.7145, + "mean_token_accuracy": 0.793032705783844, + "num_tokens": 71808946.0, + "step": 1878 + }, + { + "epoch": 0.2390281134715685, + "grad_norm": 23.375, + "learning_rate": 7.961000423908435e-07, + "loss": 0.7574, + "mean_token_accuracy": 0.7886189222335815, + "num_tokens": 71848605.0, + "step": 1879 + }, + { + "epoch": 0.23915532375015902, + "grad_norm": 22.25, + "learning_rate": 7.965239508266214e-07, + "loss": 0.7381, + "mean_token_accuracy": 0.7849215865135193, + "num_tokens": 71886771.0, + "step": 1880 + }, + { + "epoch": 0.23928253402874952, + "grad_norm": 88.5, + "learning_rate": 7.969478592623993e-07, + "loss": 0.718, + "mean_token_accuracy": 0.7912648320198059, + "num_tokens": 71924905.0, + "step": 1881 + }, + { + "epoch": 0.23940974430734002, + "grad_norm": 126.0, + "learning_rate": 7.973717676981771e-07, + "loss": 0.7311, + "mean_token_accuracy": 0.7899340391159058, + "num_tokens": 71964513.0, + "step": 1882 + }, + { + "epoch": 0.23953695458593055, + "grad_norm": 216.0, + "learning_rate": 7.977956761339551e-07, + "loss": 0.7217, + "mean_token_accuracy": 0.792582631111145, + "num_tokens": 72000918.0, + "step": 1883 + }, + { + "epoch": 0.23966416486452105, + "grad_norm": 145.0, + "learning_rate": 7.982195845697329e-07, + "loss": 0.8624, + "mean_token_accuracy": 0.754334568977356, + "num_tokens": 72040840.0, + "step": 1884 + }, + { + "epoch": 0.23979137514311155, + "grad_norm": 78.0, + "learning_rate": 7.986434930055108e-07, + "loss": 0.7199, + "mean_token_accuracy": 0.7908201813697815, + "num_tokens": 72076757.0, + "step": 1885 + }, + { + "epoch": 0.23991858542170208, + "grad_norm": 39.75, + "learning_rate": 7.990674014412886e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.777633547782898, + "num_tokens": 72120760.0, + "step": 1886 + }, + { + "epoch": 0.24004579570029258, + "grad_norm": 106.5, + "learning_rate": 7.994913098770665e-07, + "loss": 0.766, + "mean_token_accuracy": 0.7768535614013672, + "num_tokens": 72159985.0, + "step": 1887 + }, + { + "epoch": 0.24017300597888308, + "grad_norm": 47.5, + "learning_rate": 7.999152183128444e-07, + "loss": 0.8112, + "mean_token_accuracy": 0.765715479850769, + "num_tokens": 72196377.0, + "step": 1888 + }, + { + "epoch": 0.2403002162574736, + "grad_norm": 17.25, + "learning_rate": 8.003391267486223e-07, + "loss": 0.7344, + "mean_token_accuracy": 0.7855030298233032, + "num_tokens": 72236585.0, + "step": 1889 + }, + { + "epoch": 0.2404274265360641, + "grad_norm": 14.6875, + "learning_rate": 8.007630351844001e-07, + "loss": 0.782, + "mean_token_accuracy": 0.7796815633773804, + "num_tokens": 72268450.0, + "step": 1890 + }, + { + "epoch": 0.24055463681465464, + "grad_norm": 208.0, + "learning_rate": 8.011869436201781e-07, + "loss": 0.768, + "mean_token_accuracy": 0.7797641754150391, + "num_tokens": 72304092.0, + "step": 1891 + }, + { + "epoch": 0.24068184709324514, + "grad_norm": 76.0, + "learning_rate": 8.016108520559559e-07, + "loss": 0.774, + "mean_token_accuracy": 0.7800313234329224, + "num_tokens": 72347312.0, + "step": 1892 + }, + { + "epoch": 0.24080905737183564, + "grad_norm": 172.0, + "learning_rate": 8.020347604917338e-07, + "loss": 0.7343, + "mean_token_accuracy": 0.788704514503479, + "num_tokens": 72382135.0, + "step": 1893 + }, + { + "epoch": 0.24093626765042617, + "grad_norm": 106.0, + "learning_rate": 8.024586689275116e-07, + "loss": 0.7764, + "mean_token_accuracy": 0.7787196636199951, + "num_tokens": 72421526.0, + "step": 1894 + }, + { + "epoch": 0.24106347792901667, + "grad_norm": 105.5, + "learning_rate": 8.028825773632894e-07, + "loss": 0.751, + "mean_token_accuracy": 0.7844237089157104, + "num_tokens": 72458592.0, + "step": 1895 + }, + { + "epoch": 0.24119068820760717, + "grad_norm": 24.25, + "learning_rate": 8.033064857990674e-07, + "loss": 0.6796, + "mean_token_accuracy": 0.7999335527420044, + "num_tokens": 72494734.0, + "step": 1896 + }, + { + "epoch": 0.2413178984861977, + "grad_norm": 64.5, + "learning_rate": 8.037303942348452e-07, + "loss": 0.7535, + "mean_token_accuracy": 0.7838910818099976, + "num_tokens": 72531560.0, + "step": 1897 + }, + { + "epoch": 0.2414451087647882, + "grad_norm": 29.625, + "learning_rate": 8.041543026706231e-07, + "loss": 0.7403, + "mean_token_accuracy": 0.7867353558540344, + "num_tokens": 72570144.0, + "step": 1898 + }, + { + "epoch": 0.2415723190433787, + "grad_norm": 137.0, + "learning_rate": 8.04578211106401e-07, + "loss": 0.7618, + "mean_token_accuracy": 0.7791595458984375, + "num_tokens": 72606738.0, + "step": 1899 + }, + { + "epoch": 0.24169952932196923, + "grad_norm": 253.0, + "learning_rate": 8.050021195421789e-07, + "loss": 0.8106, + "mean_token_accuracy": 0.7664108276367188, + "num_tokens": 72646230.0, + "step": 1900 + }, + { + "epoch": 0.24182673960055973, + "grad_norm": 342.0, + "learning_rate": 8.054260279779567e-07, + "loss": 0.8022, + "mean_token_accuracy": 0.7674619555473328, + "num_tokens": 72681811.0, + "step": 1901 + }, + { + "epoch": 0.24195394987915023, + "grad_norm": 428.0, + "learning_rate": 8.058499364137346e-07, + "loss": 0.838, + "mean_token_accuracy": 0.7609701156616211, + "num_tokens": 72721805.0, + "step": 1902 + }, + { + "epoch": 0.24208116015774075, + "grad_norm": 334.0, + "learning_rate": 8.062738448495124e-07, + "loss": 0.7195, + "mean_token_accuracy": 0.7930322885513306, + "num_tokens": 72763142.0, + "step": 1903 + }, + { + "epoch": 0.24220837043633126, + "grad_norm": 181.0, + "learning_rate": 8.066977532852904e-07, + "loss": 0.7022, + "mean_token_accuracy": 0.8012394905090332, + "num_tokens": 72802047.0, + "step": 1904 + }, + { + "epoch": 0.24233558071492176, + "grad_norm": 318.0, + "learning_rate": 8.071216617210682e-07, + "loss": 0.835, + "mean_token_accuracy": 0.7619650363922119, + "num_tokens": 72834470.0, + "step": 1905 + }, + { + "epoch": 0.24246279099351228, + "grad_norm": 146.0, + "learning_rate": 8.075455701568461e-07, + "loss": 0.7545, + "mean_token_accuracy": 0.7831929922103882, + "num_tokens": 72873101.0, + "step": 1906 + }, + { + "epoch": 0.24259000127210278, + "grad_norm": 23.75, + "learning_rate": 8.07969478592624e-07, + "loss": 0.7045, + "mean_token_accuracy": 0.800176203250885, + "num_tokens": 72915504.0, + "step": 1907 + }, + { + "epoch": 0.24271721155069328, + "grad_norm": 133.0, + "learning_rate": 8.083933870284019e-07, + "loss": 0.6883, + "mean_token_accuracy": 0.8015505075454712, + "num_tokens": 72947179.0, + "step": 1908 + }, + { + "epoch": 0.2428444218292838, + "grad_norm": 228.0, + "learning_rate": 8.088172954641796e-07, + "loss": 0.7625, + "mean_token_accuracy": 0.7803018093109131, + "num_tokens": 72983313.0, + "step": 1909 + }, + { + "epoch": 0.2429716321078743, + "grad_norm": 304.0, + "learning_rate": 8.092412038999576e-07, + "loss": 0.8662, + "mean_token_accuracy": 0.7580124735832214, + "num_tokens": 73014218.0, + "step": 1910 + }, + { + "epoch": 0.24309884238646481, + "grad_norm": 292.0, + "learning_rate": 8.096651123357354e-07, + "loss": 0.8437, + "mean_token_accuracy": 0.7608195543289185, + "num_tokens": 73048282.0, + "step": 1911 + }, + { + "epoch": 0.24322605266505534, + "grad_norm": 200.0, + "learning_rate": 8.100890207715134e-07, + "loss": 0.8042, + "mean_token_accuracy": 0.7703801393508911, + "num_tokens": 73084390.0, + "step": 1912 + }, + { + "epoch": 0.24335326294364584, + "grad_norm": 223.0, + "learning_rate": 8.105129292072912e-07, + "loss": 0.774, + "mean_token_accuracy": 0.7820001840591431, + "num_tokens": 73123296.0, + "step": 1913 + }, + { + "epoch": 0.24348047322223634, + "grad_norm": 93.0, + "learning_rate": 8.10936837643069e-07, + "loss": 0.8234, + "mean_token_accuracy": 0.7621569633483887, + "num_tokens": 73157533.0, + "step": 1914 + }, + { + "epoch": 0.24360768350082687, + "grad_norm": 56.25, + "learning_rate": 8.11360746078847e-07, + "loss": 0.7349, + "mean_token_accuracy": 0.787674069404602, + "num_tokens": 73195314.0, + "step": 1915 + }, + { + "epoch": 0.24373489377941737, + "grad_norm": 47.75, + "learning_rate": 8.117846545146248e-07, + "loss": 0.6991, + "mean_token_accuracy": 0.7944864630699158, + "num_tokens": 73236763.0, + "step": 1916 + }, + { + "epoch": 0.2438621040580079, + "grad_norm": 129.0, + "learning_rate": 8.122085629504026e-07, + "loss": 0.8213, + "mean_token_accuracy": 0.7618861198425293, + "num_tokens": 73276719.0, + "step": 1917 + }, + { + "epoch": 0.2439893143365984, + "grad_norm": 255.0, + "learning_rate": 8.126324713861805e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7796478271484375, + "num_tokens": 73318594.0, + "step": 1918 + }, + { + "epoch": 0.2441165246151889, + "grad_norm": 256.0, + "learning_rate": 8.130563798219584e-07, + "loss": 0.7414, + "mean_token_accuracy": 0.7907159924507141, + "num_tokens": 73358760.0, + "step": 1919 + }, + { + "epoch": 0.24424373489377943, + "grad_norm": 346.0, + "learning_rate": 8.134802882577363e-07, + "loss": 0.8037, + "mean_token_accuracy": 0.7743804454803467, + "num_tokens": 73398623.0, + "step": 1920 + }, + { + "epoch": 0.24437094517236993, + "grad_norm": 42.25, + "learning_rate": 8.139041966935142e-07, + "loss": 0.7471, + "mean_token_accuracy": 0.7869777083396912, + "num_tokens": 73446369.0, + "step": 1921 + }, + { + "epoch": 0.24449815545096043, + "grad_norm": 290.0, + "learning_rate": 8.14328105129292e-07, + "loss": 0.8479, + "mean_token_accuracy": 0.764772355556488, + "num_tokens": 73483152.0, + "step": 1922 + }, + { + "epoch": 0.24462536572955096, + "grad_norm": 24.5, + "learning_rate": 8.1475201356507e-07, + "loss": 0.7721, + "mean_token_accuracy": 0.7794165015220642, + "num_tokens": 73519179.0, + "step": 1923 + }, + { + "epoch": 0.24475257600814146, + "grad_norm": 26.25, + "learning_rate": 8.151759220008477e-07, + "loss": 0.7408, + "mean_token_accuracy": 0.7832821011543274, + "num_tokens": 73561947.0, + "step": 1924 + }, + { + "epoch": 0.24487978628673196, + "grad_norm": 16.5, + "learning_rate": 8.155998304366256e-07, + "loss": 0.77, + "mean_token_accuracy": 0.77982497215271, + "num_tokens": 73595841.0, + "step": 1925 + }, + { + "epoch": 0.2450069965653225, + "grad_norm": 19.5, + "learning_rate": 8.160237388724035e-07, + "loss": 0.7951, + "mean_token_accuracy": 0.7700108885765076, + "num_tokens": 73629716.0, + "step": 1926 + }, + { + "epoch": 0.245134206843913, + "grad_norm": 51.5, + "learning_rate": 8.164476473081814e-07, + "loss": 0.7829, + "mean_token_accuracy": 0.7782950401306152, + "num_tokens": 73667285.0, + "step": 1927 + }, + { + "epoch": 0.2452614171225035, + "grad_norm": 85.0, + "learning_rate": 8.168715557439593e-07, + "loss": 0.7079, + "mean_token_accuracy": 0.7955349683761597, + "num_tokens": 73708059.0, + "step": 1928 + }, + { + "epoch": 0.24538862740109402, + "grad_norm": 113.0, + "learning_rate": 8.172954641797372e-07, + "loss": 0.7119, + "mean_token_accuracy": 0.7935799360275269, + "num_tokens": 73748553.0, + "step": 1929 + }, + { + "epoch": 0.24551583767968452, + "grad_norm": 72.0, + "learning_rate": 8.17719372615515e-07, + "loss": 0.8263, + "mean_token_accuracy": 0.7653591632843018, + "num_tokens": 73796295.0, + "step": 1930 + }, + { + "epoch": 0.24564304795827502, + "grad_norm": 84.0, + "learning_rate": 8.18143281051293e-07, + "loss": 0.8243, + "mean_token_accuracy": 0.7676649689674377, + "num_tokens": 73835960.0, + "step": 1931 + }, + { + "epoch": 0.24577025823686555, + "grad_norm": 45.0, + "learning_rate": 8.185671894870707e-07, + "loss": 0.837, + "mean_token_accuracy": 0.7620968818664551, + "num_tokens": 73875302.0, + "step": 1932 + }, + { + "epoch": 0.24589746851545605, + "grad_norm": 11.0, + "learning_rate": 8.189910979228485e-07, + "loss": 0.7208, + "mean_token_accuracy": 0.7903522849082947, + "num_tokens": 73907703.0, + "step": 1933 + }, + { + "epoch": 0.24602467879404655, + "grad_norm": 63.0, + "learning_rate": 8.194150063586265e-07, + "loss": 0.8272, + "mean_token_accuracy": 0.7648314237594604, + "num_tokens": 73945449.0, + "step": 1934 + }, + { + "epoch": 0.24615188907263708, + "grad_norm": 46.75, + "learning_rate": 8.198389147944043e-07, + "loss": 0.8081, + "mean_token_accuracy": 0.7669963836669922, + "num_tokens": 73982461.0, + "step": 1935 + }, + { + "epoch": 0.24627909935122758, + "grad_norm": 94.0, + "learning_rate": 8.202628232301823e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7797219753265381, + "num_tokens": 74018252.0, + "step": 1936 + }, + { + "epoch": 0.24640630962981808, + "grad_norm": 68.0, + "learning_rate": 8.206867316659601e-07, + "loss": 0.79, + "mean_token_accuracy": 0.7720530033111572, + "num_tokens": 74065843.0, + "step": 1937 + }, + { + "epoch": 0.2465335199084086, + "grad_norm": 8.5, + "learning_rate": 8.21110640101738e-07, + "loss": 0.6659, + "mean_token_accuracy": 0.8047440648078918, + "num_tokens": 74106389.0, + "step": 1938 + }, + { + "epoch": 0.2466607301869991, + "grad_norm": 20.75, + "learning_rate": 8.215345485375159e-07, + "loss": 0.7983, + "mean_token_accuracy": 0.7705792188644409, + "num_tokens": 74150786.0, + "step": 1939 + }, + { + "epoch": 0.2467879404655896, + "grad_norm": 53.75, + "learning_rate": 8.219584569732937e-07, + "loss": 0.7683, + "mean_token_accuracy": 0.775206446647644, + "num_tokens": 74190474.0, + "step": 1940 + }, + { + "epoch": 0.24691515074418013, + "grad_norm": 35.0, + "learning_rate": 8.223823654090715e-07, + "loss": 0.7152, + "mean_token_accuracy": 0.7941348552703857, + "num_tokens": 74226437.0, + "step": 1941 + }, + { + "epoch": 0.24704236102277063, + "grad_norm": 84.5, + "learning_rate": 8.228062738448495e-07, + "loss": 0.7649, + "mean_token_accuracy": 0.7796900272369385, + "num_tokens": 74267292.0, + "step": 1942 + }, + { + "epoch": 0.24716957130136116, + "grad_norm": 67.5, + "learning_rate": 8.232301822806273e-07, + "loss": 0.7866, + "mean_token_accuracy": 0.7742376327514648, + "num_tokens": 74306738.0, + "step": 1943 + }, + { + "epoch": 0.24729678157995166, + "grad_norm": 28.625, + "learning_rate": 8.236540907164053e-07, + "loss": 0.7809, + "mean_token_accuracy": 0.778285026550293, + "num_tokens": 74343151.0, + "step": 1944 + }, + { + "epoch": 0.24742399185854216, + "grad_norm": 40.25, + "learning_rate": 8.240779991521831e-07, + "loss": 0.7417, + "mean_token_accuracy": 0.7863650918006897, + "num_tokens": 74382755.0, + "step": 1945 + }, + { + "epoch": 0.2475512021371327, + "grad_norm": 29.25, + "learning_rate": 8.24501907587961e-07, + "loss": 0.7862, + "mean_token_accuracy": 0.7735766172409058, + "num_tokens": 74426129.0, + "step": 1946 + }, + { + "epoch": 0.2476784124157232, + "grad_norm": 59.25, + "learning_rate": 8.249258160237388e-07, + "loss": 0.8238, + "mean_token_accuracy": 0.7700773477554321, + "num_tokens": 74460904.0, + "step": 1947 + }, + { + "epoch": 0.2478056226943137, + "grad_norm": 47.5, + "learning_rate": 8.253497244595167e-07, + "loss": 0.8044, + "mean_token_accuracy": 0.7714817523956299, + "num_tokens": 74497993.0, + "step": 1948 + }, + { + "epoch": 0.24793283297290422, + "grad_norm": 32.25, + "learning_rate": 8.257736328952945e-07, + "loss": 0.7862, + "mean_token_accuracy": 0.7732586860656738, + "num_tokens": 74541108.0, + "step": 1949 + }, + { + "epoch": 0.24806004325149472, + "grad_norm": 13.875, + "learning_rate": 8.261975413310725e-07, + "loss": 0.7734, + "mean_token_accuracy": 0.7814292311668396, + "num_tokens": 74579389.0, + "step": 1950 + }, + { + "epoch": 0.24818725353008522, + "grad_norm": 12.9375, + "learning_rate": 8.266214497668503e-07, + "loss": 0.7005, + "mean_token_accuracy": 0.7994669675827026, + "num_tokens": 74621546.0, + "step": 1951 + }, + { + "epoch": 0.24831446380867575, + "grad_norm": 140.0, + "learning_rate": 8.270453582026283e-07, + "loss": 0.8238, + "mean_token_accuracy": 0.7629901170730591, + "num_tokens": 74660178.0, + "step": 1952 + }, + { + "epoch": 0.24844167408726625, + "grad_norm": 72.5, + "learning_rate": 8.274692666384061e-07, + "loss": 0.762, + "mean_token_accuracy": 0.780645489692688, + "num_tokens": 74693867.0, + "step": 1953 + }, + { + "epoch": 0.24856888436585675, + "grad_norm": 21.75, + "learning_rate": 8.27893175074184e-07, + "loss": 0.7817, + "mean_token_accuracy": 0.7726646065711975, + "num_tokens": 74727561.0, + "step": 1954 + }, + { + "epoch": 0.24869609464444728, + "grad_norm": 41.0, + "learning_rate": 8.283170835099618e-07, + "loss": 0.7557, + "mean_token_accuracy": 0.7869852781295776, + "num_tokens": 74763719.0, + "step": 1955 + }, + { + "epoch": 0.24882330492303778, + "grad_norm": 35.5, + "learning_rate": 8.287409919457396e-07, + "loss": 0.7187, + "mean_token_accuracy": 0.7938677072525024, + "num_tokens": 74802666.0, + "step": 1956 + }, + { + "epoch": 0.24895051520162828, + "grad_norm": 33.0, + "learning_rate": 8.291649003815175e-07, + "loss": 0.7821, + "mean_token_accuracy": 0.777358889579773, + "num_tokens": 74840307.0, + "step": 1957 + }, + { + "epoch": 0.2490777254802188, + "grad_norm": 122.0, + "learning_rate": 8.295888088172954e-07, + "loss": 0.7822, + "mean_token_accuracy": 0.7754138708114624, + "num_tokens": 74880181.0, + "step": 1958 + }, + { + "epoch": 0.2492049357588093, + "grad_norm": 20.0, + "learning_rate": 8.300127172530733e-07, + "loss": 0.696, + "mean_token_accuracy": 0.8001834154129028, + "num_tokens": 74917148.0, + "step": 1959 + }, + { + "epoch": 0.2493321460373998, + "grad_norm": 17.75, + "learning_rate": 8.304366256888512e-07, + "loss": 0.7329, + "mean_token_accuracy": 0.7893381714820862, + "num_tokens": 74953124.0, + "step": 1960 + }, + { + "epoch": 0.24945935631599034, + "grad_norm": 29.875, + "learning_rate": 8.308605341246291e-07, + "loss": 0.7483, + "mean_token_accuracy": 0.7867541313171387, + "num_tokens": 74989726.0, + "step": 1961 + }, + { + "epoch": 0.24958656659458084, + "grad_norm": 48.75, + "learning_rate": 8.312844425604068e-07, + "loss": 0.7531, + "mean_token_accuracy": 0.7901727557182312, + "num_tokens": 75024849.0, + "step": 1962 + }, + { + "epoch": 0.24971377687317134, + "grad_norm": 37.0, + "learning_rate": 8.317083509961848e-07, + "loss": 0.7819, + "mean_token_accuracy": 0.777981162071228, + "num_tokens": 75066464.0, + "step": 1963 + }, + { + "epoch": 0.24984098715176187, + "grad_norm": 20.125, + "learning_rate": 8.321322594319626e-07, + "loss": 0.7299, + "mean_token_accuracy": 0.7866056561470032, + "num_tokens": 75106829.0, + "step": 1964 + }, + { + "epoch": 0.24996819743035237, + "grad_norm": 33.5, + "learning_rate": 8.325561678677405e-07, + "loss": 0.7867, + "mean_token_accuracy": 0.7757128477096558, + "num_tokens": 75148227.0, + "step": 1965 + }, + { + "epoch": 0.2500954077089429, + "grad_norm": 67.0, + "learning_rate": 8.329800763035184e-07, + "loss": 0.8079, + "mean_token_accuracy": 0.76847243309021, + "num_tokens": 75188719.0, + "step": 1966 + }, + { + "epoch": 0.25022261798753337, + "grad_norm": 25.375, + "learning_rate": 8.334039847392963e-07, + "loss": 0.7654, + "mean_token_accuracy": 0.7761462926864624, + "num_tokens": 75222260.0, + "step": 1967 + }, + { + "epoch": 0.2503498282661239, + "grad_norm": 97.5, + "learning_rate": 8.338278931750742e-07, + "loss": 0.7151, + "mean_token_accuracy": 0.7925851345062256, + "num_tokens": 75255489.0, + "step": 1968 + }, + { + "epoch": 0.2504770385447144, + "grad_norm": 21.25, + "learning_rate": 8.342518016108521e-07, + "loss": 0.7661, + "mean_token_accuracy": 0.7739982604980469, + "num_tokens": 75290548.0, + "step": 1969 + }, + { + "epoch": 0.2506042488233049, + "grad_norm": 22.75, + "learning_rate": 8.346757100466298e-07, + "loss": 0.7609, + "mean_token_accuracy": 0.7858622670173645, + "num_tokens": 75323912.0, + "step": 1970 + }, + { + "epoch": 0.2507314591018954, + "grad_norm": 35.0, + "learning_rate": 8.350996184824078e-07, + "loss": 0.8162, + "mean_token_accuracy": 0.7667517066001892, + "num_tokens": 75361206.0, + "step": 1971 + }, + { + "epoch": 0.25085866938048595, + "grad_norm": 30.625, + "learning_rate": 8.355235269181856e-07, + "loss": 0.7913, + "mean_token_accuracy": 0.7752000689506531, + "num_tokens": 75397292.0, + "step": 1972 + }, + { + "epoch": 0.2509858796590764, + "grad_norm": 42.5, + "learning_rate": 8.359474353539635e-07, + "loss": 0.7023, + "mean_token_accuracy": 0.7941494584083557, + "num_tokens": 75437616.0, + "step": 1973 + }, + { + "epoch": 0.25111308993766696, + "grad_norm": 93.5, + "learning_rate": 8.363713437897414e-07, + "loss": 0.8296, + "mean_token_accuracy": 0.7667273283004761, + "num_tokens": 75486172.0, + "step": 1974 + }, + { + "epoch": 0.2512403002162575, + "grad_norm": 22.625, + "learning_rate": 8.367952522255193e-07, + "loss": 0.7491, + "mean_token_accuracy": 0.7807326316833496, + "num_tokens": 75526637.0, + "step": 1975 + }, + { + "epoch": 0.25136751049484796, + "grad_norm": 50.5, + "learning_rate": 8.372191606612972e-07, + "loss": 0.762, + "mean_token_accuracy": 0.780806303024292, + "num_tokens": 75564632.0, + "step": 1976 + }, + { + "epoch": 0.2514947207734385, + "grad_norm": 20.125, + "learning_rate": 8.376430690970749e-07, + "loss": 0.7309, + "mean_token_accuracy": 0.7873607277870178, + "num_tokens": 75595553.0, + "step": 1977 + }, + { + "epoch": 0.251621931052029, + "grad_norm": 66.0, + "learning_rate": 8.380669775328528e-07, + "loss": 0.7002, + "mean_token_accuracy": 0.7985721230506897, + "num_tokens": 75635847.0, + "step": 1978 + }, + { + "epoch": 0.25174914133061954, + "grad_norm": 14.25, + "learning_rate": 8.384908859686307e-07, + "loss": 0.6999, + "mean_token_accuracy": 0.7991491556167603, + "num_tokens": 75677403.0, + "step": 1979 + }, + { + "epoch": 0.25187635160921, + "grad_norm": 104.5, + "learning_rate": 8.389147944044086e-07, + "loss": 0.7602, + "mean_token_accuracy": 0.7813237905502319, + "num_tokens": 75720703.0, + "step": 1980 + }, + { + "epoch": 0.25200356188780054, + "grad_norm": 81.5, + "learning_rate": 8.393387028401864e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7749490737915039, + "num_tokens": 75758000.0, + "step": 1981 + }, + { + "epoch": 0.25213077216639107, + "grad_norm": 126.5, + "learning_rate": 8.397626112759644e-07, + "loss": 0.8292, + "mean_token_accuracy": 0.7650099992752075, + "num_tokens": 75791113.0, + "step": 1982 + }, + { + "epoch": 0.25225798244498154, + "grad_norm": 99.5, + "learning_rate": 8.401865197117422e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.7769598364830017, + "num_tokens": 75821733.0, + "step": 1983 + }, + { + "epoch": 0.25238519272357207, + "grad_norm": 121.0, + "learning_rate": 8.406104281475202e-07, + "loss": 0.7269, + "mean_token_accuracy": 0.7871608138084412, + "num_tokens": 75861404.0, + "step": 1984 + }, + { + "epoch": 0.2525124030021626, + "grad_norm": 19.5, + "learning_rate": 8.410343365832979e-07, + "loss": 0.7194, + "mean_token_accuracy": 0.7895190715789795, + "num_tokens": 75903059.0, + "step": 1985 + }, + { + "epoch": 0.2526396132807531, + "grad_norm": 14.1875, + "learning_rate": 8.414582450190758e-07, + "loss": 0.7184, + "mean_token_accuracy": 0.7912717461585999, + "num_tokens": 75947125.0, + "step": 1986 + }, + { + "epoch": 0.2527668235593436, + "grad_norm": 15.5625, + "learning_rate": 8.418821534548537e-07, + "loss": 0.8382, + "mean_token_accuracy": 0.7627556324005127, + "num_tokens": 75987519.0, + "step": 1987 + }, + { + "epoch": 0.25289403383793413, + "grad_norm": 39.25, + "learning_rate": 8.423060618906316e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.779704213142395, + "num_tokens": 76027728.0, + "step": 1988 + }, + { + "epoch": 0.2530212441165246, + "grad_norm": 11.4375, + "learning_rate": 8.427299703264095e-07, + "loss": 0.7519, + "mean_token_accuracy": 0.7851366996765137, + "num_tokens": 76070942.0, + "step": 1989 + }, + { + "epoch": 0.25314845439511513, + "grad_norm": 16.625, + "learning_rate": 8.431538787621874e-07, + "loss": 0.7826, + "mean_token_accuracy": 0.7763780355453491, + "num_tokens": 76113330.0, + "step": 1990 + }, + { + "epoch": 0.25327566467370566, + "grad_norm": 19.0, + "learning_rate": 8.435777871979652e-07, + "loss": 0.7142, + "mean_token_accuracy": 0.7939170002937317, + "num_tokens": 76153511.0, + "step": 1991 + }, + { + "epoch": 0.25340287495229613, + "grad_norm": 84.0, + "learning_rate": 8.440016956337432e-07, + "loss": 0.7607, + "mean_token_accuracy": 0.7807692289352417, + "num_tokens": 76191505.0, + "step": 1992 + }, + { + "epoch": 0.25353008523088666, + "grad_norm": 56.25, + "learning_rate": 8.444256040695209e-07, + "loss": 0.7681, + "mean_token_accuracy": 0.7822455167770386, + "num_tokens": 76232479.0, + "step": 1993 + }, + { + "epoch": 0.2536572955094772, + "grad_norm": 37.25, + "learning_rate": 8.448495125052988e-07, + "loss": 0.782, + "mean_token_accuracy": 0.777248740196228, + "num_tokens": 76270192.0, + "step": 1994 + }, + { + "epoch": 0.25378450578806766, + "grad_norm": 22.625, + "learning_rate": 8.452734209410767e-07, + "loss": 0.7076, + "mean_token_accuracy": 0.7961335182189941, + "num_tokens": 76303981.0, + "step": 1995 + }, + { + "epoch": 0.2539117160666582, + "grad_norm": 143.0, + "learning_rate": 8.456973293768545e-07, + "loss": 0.7569, + "mean_token_accuracy": 0.7820395827293396, + "num_tokens": 76338496.0, + "step": 1996 + }, + { + "epoch": 0.2540389263452487, + "grad_norm": 30.375, + "learning_rate": 8.461212378126325e-07, + "loss": 0.7973, + "mean_token_accuracy": 0.7708348035812378, + "num_tokens": 76377904.0, + "step": 1997 + }, + { + "epoch": 0.2541661366238392, + "grad_norm": 54.5, + "learning_rate": 8.465451462484103e-07, + "loss": 0.8323, + "mean_token_accuracy": 0.7638039588928223, + "num_tokens": 76413905.0, + "step": 1998 + }, + { + "epoch": 0.2542933469024297, + "grad_norm": 75.5, + "learning_rate": 8.469690546841882e-07, + "loss": 0.7727, + "mean_token_accuracy": 0.7816044092178345, + "num_tokens": 76451591.0, + "step": 1999 + }, + { + "epoch": 0.25442055718102025, + "grad_norm": 13.0, + "learning_rate": 8.47392963119966e-07, + "loss": 0.7496, + "mean_token_accuracy": 0.782745897769928, + "num_tokens": 76486191.0, + "step": 2000 + }, + { + "epoch": 0.2545477674596107, + "grad_norm": 32.25, + "learning_rate": 8.478168715557439e-07, + "loss": 0.7509, + "mean_token_accuracy": 0.7879027128219604, + "num_tokens": 76523317.0, + "step": 2001 + }, + { + "epoch": 0.25467497773820125, + "grad_norm": 25.625, + "learning_rate": 8.482407799915217e-07, + "loss": 0.7641, + "mean_token_accuracy": 0.7828454971313477, + "num_tokens": 76562661.0, + "step": 2002 + }, + { + "epoch": 0.2548021880167918, + "grad_norm": 22.0, + "learning_rate": 8.486646884272997e-07, + "loss": 0.7632, + "mean_token_accuracy": 0.779994010925293, + "num_tokens": 76600914.0, + "step": 2003 + }, + { + "epoch": 0.25492939829538225, + "grad_norm": 14.5625, + "learning_rate": 8.490885968630775e-07, + "loss": 0.7939, + "mean_token_accuracy": 0.773053765296936, + "num_tokens": 76636310.0, + "step": 2004 + }, + { + "epoch": 0.2550566085739728, + "grad_norm": 99.0, + "learning_rate": 8.495125052988555e-07, + "loss": 0.7527, + "mean_token_accuracy": 0.789141058921814, + "num_tokens": 76680889.0, + "step": 2005 + }, + { + "epoch": 0.2551838188525633, + "grad_norm": 29.25, + "learning_rate": 8.499364137346333e-07, + "loss": 0.739, + "mean_token_accuracy": 0.7930452823638916, + "num_tokens": 76722469.0, + "step": 2006 + }, + { + "epoch": 0.2553110291311538, + "grad_norm": 34.25, + "learning_rate": 8.503603221704112e-07, + "loss": 0.8169, + "mean_token_accuracy": 0.7695317268371582, + "num_tokens": 76757668.0, + "step": 2007 + }, + { + "epoch": 0.2554382394097443, + "grad_norm": 33.5, + "learning_rate": 8.50784230606189e-07, + "loss": 0.7657, + "mean_token_accuracy": 0.7839450240135193, + "num_tokens": 76795607.0, + "step": 2008 + }, + { + "epoch": 0.25556544968833483, + "grad_norm": 29.625, + "learning_rate": 8.512081390419669e-07, + "loss": 0.7855, + "mean_token_accuracy": 0.7821503281593323, + "num_tokens": 76837161.0, + "step": 2009 + }, + { + "epoch": 0.2556926599669253, + "grad_norm": 16.625, + "learning_rate": 8.516320474777447e-07, + "loss": 0.7518, + "mean_token_accuracy": 0.7831321954727173, + "num_tokens": 76872198.0, + "step": 2010 + }, + { + "epoch": 0.25581987024551583, + "grad_norm": 68.0, + "learning_rate": 8.520559559135227e-07, + "loss": 0.7192, + "mean_token_accuracy": 0.7920563817024231, + "num_tokens": 76908976.0, + "step": 2011 + }, + { + "epoch": 0.25594708052410636, + "grad_norm": 31.875, + "learning_rate": 8.524798643493005e-07, + "loss": 0.7697, + "mean_token_accuracy": 0.7825029492378235, + "num_tokens": 76947298.0, + "step": 2012 + }, + { + "epoch": 0.25607429080269684, + "grad_norm": 162.0, + "learning_rate": 8.529037727850785e-07, + "loss": 0.73, + "mean_token_accuracy": 0.7892683744430542, + "num_tokens": 76985112.0, + "step": 2013 + }, + { + "epoch": 0.25620150108128736, + "grad_norm": 24.125, + "learning_rate": 8.533276812208563e-07, + "loss": 0.6987, + "mean_token_accuracy": 0.7992973327636719, + "num_tokens": 77020309.0, + "step": 2014 + }, + { + "epoch": 0.2563287113598779, + "grad_norm": 38.0, + "learning_rate": 8.53751589656634e-07, + "loss": 0.7483, + "mean_token_accuracy": 0.7856045365333557, + "num_tokens": 77063245.0, + "step": 2015 + }, + { + "epoch": 0.25645592163846836, + "grad_norm": 85.0, + "learning_rate": 8.54175498092412e-07, + "loss": 0.7572, + "mean_token_accuracy": 0.783464789390564, + "num_tokens": 77096791.0, + "step": 2016 + }, + { + "epoch": 0.2565831319170589, + "grad_norm": 18.625, + "learning_rate": 8.545994065281898e-07, + "loss": 0.7253, + "mean_token_accuracy": 0.7866343259811401, + "num_tokens": 77129281.0, + "step": 2017 + }, + { + "epoch": 0.2567103421956494, + "grad_norm": 109.5, + "learning_rate": 8.550233149639677e-07, + "loss": 0.8297, + "mean_token_accuracy": 0.7604614496231079, + "num_tokens": 77168785.0, + "step": 2018 + }, + { + "epoch": 0.2568375524742399, + "grad_norm": 82.5, + "learning_rate": 8.554472233997456e-07, + "loss": 0.7905, + "mean_token_accuracy": 0.7729812860488892, + "num_tokens": 77208360.0, + "step": 2019 + }, + { + "epoch": 0.2569647627528304, + "grad_norm": 66.5, + "learning_rate": 8.558711318355235e-07, + "loss": 0.8356, + "mean_token_accuracy": 0.7627792358398438, + "num_tokens": 77245539.0, + "step": 2020 + }, + { + "epoch": 0.25709197303142095, + "grad_norm": 21.375, + "learning_rate": 8.562950402713014e-07, + "loss": 0.7155, + "mean_token_accuracy": 0.7951045036315918, + "num_tokens": 77279118.0, + "step": 2021 + }, + { + "epoch": 0.2572191833100114, + "grad_norm": 53.75, + "learning_rate": 8.567189487070793e-07, + "loss": 0.816, + "mean_token_accuracy": 0.7667317390441895, + "num_tokens": 77316259.0, + "step": 2022 + }, + { + "epoch": 0.25734639358860195, + "grad_norm": 40.0, + "learning_rate": 8.57142857142857e-07, + "loss": 0.8222, + "mean_token_accuracy": 0.7671361565589905, + "num_tokens": 77359712.0, + "step": 2023 + }, + { + "epoch": 0.2574736038671925, + "grad_norm": 54.5, + "learning_rate": 8.57566765578635e-07, + "loss": 0.7043, + "mean_token_accuracy": 0.8015578985214233, + "num_tokens": 77402269.0, + "step": 2024 + }, + { + "epoch": 0.25760081414578295, + "grad_norm": 26.25, + "learning_rate": 8.579906740144128e-07, + "loss": 0.7285, + "mean_token_accuracy": 0.7894792556762695, + "num_tokens": 77445791.0, + "step": 2025 + }, + { + "epoch": 0.2577280244243735, + "grad_norm": 20.125, + "learning_rate": 8.584145824501907e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7767388820648193, + "num_tokens": 77485376.0, + "step": 2026 + }, + { + "epoch": 0.257855234702964, + "grad_norm": 20.5, + "learning_rate": 8.588384908859686e-07, + "loss": 0.7895, + "mean_token_accuracy": 0.7760486602783203, + "num_tokens": 77524344.0, + "step": 2027 + }, + { + "epoch": 0.25798244498155454, + "grad_norm": 56.75, + "learning_rate": 8.592623993217465e-07, + "loss": 0.7888, + "mean_token_accuracy": 0.7781357169151306, + "num_tokens": 77562746.0, + "step": 2028 + }, + { + "epoch": 0.258109655260145, + "grad_norm": 54.75, + "learning_rate": 8.596863077575244e-07, + "loss": 0.8022, + "mean_token_accuracy": 0.771513819694519, + "num_tokens": 77601261.0, + "step": 2029 + }, + { + "epoch": 0.25823686553873554, + "grad_norm": 60.0, + "learning_rate": 8.601102161933023e-07, + "loss": 0.7821, + "mean_token_accuracy": 0.7785452008247375, + "num_tokens": 77640054.0, + "step": 2030 + }, + { + "epoch": 0.25836407581732607, + "grad_norm": 54.5, + "learning_rate": 8.6053412462908e-07, + "loss": 0.7456, + "mean_token_accuracy": 0.7866302728652954, + "num_tokens": 77680098.0, + "step": 2031 + }, + { + "epoch": 0.25849128609591654, + "grad_norm": 163.0, + "learning_rate": 8.60958033064858e-07, + "loss": 0.7435, + "mean_token_accuracy": 0.7873238325119019, + "num_tokens": 77712517.0, + "step": 2032 + }, + { + "epoch": 0.25861849637450707, + "grad_norm": 66.5, + "learning_rate": 8.613819415006358e-07, + "loss": 0.8159, + "mean_token_accuracy": 0.769284188747406, + "num_tokens": 77751335.0, + "step": 2033 + }, + { + "epoch": 0.2587457066530976, + "grad_norm": 76.0, + "learning_rate": 8.618058499364137e-07, + "loss": 0.791, + "mean_token_accuracy": 0.7761878967285156, + "num_tokens": 77792300.0, + "step": 2034 + }, + { + "epoch": 0.25887291693168807, + "grad_norm": 88.5, + "learning_rate": 8.622297583721916e-07, + "loss": 0.7306, + "mean_token_accuracy": 0.79515141248703, + "num_tokens": 77836391.0, + "step": 2035 + }, + { + "epoch": 0.2590001272102786, + "grad_norm": 157.0, + "learning_rate": 8.626536668079695e-07, + "loss": 0.7797, + "mean_token_accuracy": 0.7767400145530701, + "num_tokens": 77871600.0, + "step": 2036 + }, + { + "epoch": 0.2591273374888691, + "grad_norm": 52.5, + "learning_rate": 8.630775752437474e-07, + "loss": 0.7692, + "mean_token_accuracy": 0.7789362668991089, + "num_tokens": 77904623.0, + "step": 2037 + }, + { + "epoch": 0.2592545477674596, + "grad_norm": 34.5, + "learning_rate": 8.635014836795251e-07, + "loss": 0.8185, + "mean_token_accuracy": 0.7657792568206787, + "num_tokens": 77941145.0, + "step": 2038 + }, + { + "epoch": 0.2593817580460501, + "grad_norm": 68.0, + "learning_rate": 8.63925392115303e-07, + "loss": 0.743, + "mean_token_accuracy": 0.7884140610694885, + "num_tokens": 77975087.0, + "step": 2039 + }, + { + "epoch": 0.25950896832464065, + "grad_norm": 64.0, + "learning_rate": 8.643493005510809e-07, + "loss": 0.7381, + "mean_token_accuracy": 0.7887741327285767, + "num_tokens": 78009526.0, + "step": 2040 + }, + { + "epoch": 0.2596361786032311, + "grad_norm": 96.0, + "learning_rate": 8.647732089868588e-07, + "loss": 0.8155, + "mean_token_accuracy": 0.7655946016311646, + "num_tokens": 78050918.0, + "step": 2041 + }, + { + "epoch": 0.25976338888182166, + "grad_norm": 70.0, + "learning_rate": 8.651971174226366e-07, + "loss": 0.7444, + "mean_token_accuracy": 0.7847690582275391, + "num_tokens": 78085547.0, + "step": 2042 + }, + { + "epoch": 0.2598905991604122, + "grad_norm": 15.8125, + "learning_rate": 8.656210258584146e-07, + "loss": 0.8184, + "mean_token_accuracy": 0.7657781839370728, + "num_tokens": 78122916.0, + "step": 2043 + }, + { + "epoch": 0.26001780943900266, + "grad_norm": 23.875, + "learning_rate": 8.660449342941924e-07, + "loss": 0.7904, + "mean_token_accuracy": 0.7710673809051514, + "num_tokens": 78161800.0, + "step": 2044 + }, + { + "epoch": 0.2601450197175932, + "grad_norm": 66.5, + "learning_rate": 8.664688427299704e-07, + "loss": 0.7871, + "mean_token_accuracy": 0.7748352885246277, + "num_tokens": 78199961.0, + "step": 2045 + }, + { + "epoch": 0.2602722299961837, + "grad_norm": 76.0, + "learning_rate": 8.668927511657481e-07, + "loss": 0.7414, + "mean_token_accuracy": 0.7823765277862549, + "num_tokens": 78231332.0, + "step": 2046 + }, + { + "epoch": 0.2603994402747742, + "grad_norm": 36.0, + "learning_rate": 8.67316659601526e-07, + "loss": 0.7866, + "mean_token_accuracy": 0.774277925491333, + "num_tokens": 78265054.0, + "step": 2047 + }, + { + "epoch": 0.2605266505533647, + "grad_norm": 63.5, + "learning_rate": 8.677405680373039e-07, + "loss": 0.7341, + "mean_token_accuracy": 0.7926850318908691, + "num_tokens": 78305445.0, + "step": 2048 + }, + { + "epoch": 0.26065386083195524, + "grad_norm": 13.625, + "learning_rate": 8.681644764730818e-07, + "loss": 0.781, + "mean_token_accuracy": 0.7745532989501953, + "num_tokens": 78339174.0, + "step": 2049 + }, + { + "epoch": 0.2607810711105457, + "grad_norm": 46.75, + "learning_rate": 8.685883849088596e-07, + "loss": 0.704, + "mean_token_accuracy": 0.7966921925544739, + "num_tokens": 78377630.0, + "step": 2050 + }, + { + "epoch": 0.26090828138913624, + "grad_norm": 49.75, + "learning_rate": 8.690122933446376e-07, + "loss": 0.752, + "mean_token_accuracy": 0.784276008605957, + "num_tokens": 78419908.0, + "step": 2051 + }, + { + "epoch": 0.26103549166772677, + "grad_norm": 19.0, + "learning_rate": 8.694362017804154e-07, + "loss": 0.7165, + "mean_token_accuracy": 0.7938785552978516, + "num_tokens": 78459561.0, + "step": 2052 + }, + { + "epoch": 0.26116270194631724, + "grad_norm": 33.5, + "learning_rate": 8.698601102161933e-07, + "loss": 0.776, + "mean_token_accuracy": 0.7782039642333984, + "num_tokens": 78502500.0, + "step": 2053 + }, + { + "epoch": 0.26128991222490777, + "grad_norm": 107.0, + "learning_rate": 8.702840186519711e-07, + "loss": 0.7179, + "mean_token_accuracy": 0.7945263385772705, + "num_tokens": 78545411.0, + "step": 2054 + }, + { + "epoch": 0.2614171225034983, + "grad_norm": 131.0, + "learning_rate": 8.70707927087749e-07, + "loss": 0.7142, + "mean_token_accuracy": 0.7942103147506714, + "num_tokens": 78585878.0, + "step": 2055 + }, + { + "epoch": 0.2615443327820888, + "grad_norm": 38.5, + "learning_rate": 8.711318355235269e-07, + "loss": 0.8232, + "mean_token_accuracy": 0.7621912360191345, + "num_tokens": 78621233.0, + "step": 2056 + }, + { + "epoch": 0.2616715430606793, + "grad_norm": 82.0, + "learning_rate": 8.715557439593047e-07, + "loss": 0.7369, + "mean_token_accuracy": 0.7874327898025513, + "num_tokens": 78654644.0, + "step": 2057 + }, + { + "epoch": 0.26179875333926983, + "grad_norm": 97.5, + "learning_rate": 8.719796523950826e-07, + "loss": 0.8426, + "mean_token_accuracy": 0.7535350322723389, + "num_tokens": 78692851.0, + "step": 2058 + }, + { + "epoch": 0.2619259636178603, + "grad_norm": 132.0, + "learning_rate": 8.724035608308605e-07, + "loss": 0.7201, + "mean_token_accuracy": 0.7923004627227783, + "num_tokens": 78725329.0, + "step": 2059 + }, + { + "epoch": 0.26205317389645083, + "grad_norm": 15.5, + "learning_rate": 8.728274692666384e-07, + "loss": 0.7643, + "mean_token_accuracy": 0.7820214629173279, + "num_tokens": 78766645.0, + "step": 2060 + }, + { + "epoch": 0.26218038417504136, + "grad_norm": 26.5, + "learning_rate": 8.732513777024162e-07, + "loss": 0.7481, + "mean_token_accuracy": 0.784429669380188, + "num_tokens": 78801576.0, + "step": 2061 + }, + { + "epoch": 0.26230759445363183, + "grad_norm": 34.0, + "learning_rate": 8.736752861381941e-07, + "loss": 0.8132, + "mean_token_accuracy": 0.7674069404602051, + "num_tokens": 78845577.0, + "step": 2062 + }, + { + "epoch": 0.26243480473222236, + "grad_norm": 77.0, + "learning_rate": 8.740991945739719e-07, + "loss": 0.7654, + "mean_token_accuracy": 0.7775576114654541, + "num_tokens": 78884229.0, + "step": 2063 + }, + { + "epoch": 0.2625620150108129, + "grad_norm": 16.75, + "learning_rate": 8.745231030097499e-07, + "loss": 0.7322, + "mean_token_accuracy": 0.7890610694885254, + "num_tokens": 78926948.0, + "step": 2064 + }, + { + "epoch": 0.26268922528940336, + "grad_norm": 32.0, + "learning_rate": 8.749470114455277e-07, + "loss": 0.7235, + "mean_token_accuracy": 0.7898101806640625, + "num_tokens": 78963582.0, + "step": 2065 + }, + { + "epoch": 0.2628164355679939, + "grad_norm": 92.0, + "learning_rate": 8.753709198813056e-07, + "loss": 0.8463, + "mean_token_accuracy": 0.761565625667572, + "num_tokens": 79007345.0, + "step": 2066 + }, + { + "epoch": 0.2629436458465844, + "grad_norm": 111.5, + "learning_rate": 8.757948283170835e-07, + "loss": 0.7142, + "mean_token_accuracy": 0.798653244972229, + "num_tokens": 79049505.0, + "step": 2067 + }, + { + "epoch": 0.2630708561251749, + "grad_norm": 19.125, + "learning_rate": 8.762187367528613e-07, + "loss": 0.7224, + "mean_token_accuracy": 0.7885693311691284, + "num_tokens": 79087419.0, + "step": 2068 + }, + { + "epoch": 0.2631980664037654, + "grad_norm": 23.75, + "learning_rate": 8.766426451886392e-07, + "loss": 0.7692, + "mean_token_accuracy": 0.7799645662307739, + "num_tokens": 79131444.0, + "step": 2069 + }, + { + "epoch": 0.26332527668235595, + "grad_norm": 41.5, + "learning_rate": 8.770665536244171e-07, + "loss": 0.7981, + "mean_token_accuracy": 0.7744371294975281, + "num_tokens": 79171976.0, + "step": 2070 + }, + { + "epoch": 0.2634524869609464, + "grad_norm": 48.75, + "learning_rate": 8.774904620601949e-07, + "loss": 0.7367, + "mean_token_accuracy": 0.7851380109786987, + "num_tokens": 79206273.0, + "step": 2071 + }, + { + "epoch": 0.26357969723953695, + "grad_norm": 85.0, + "learning_rate": 8.779143704959729e-07, + "loss": 0.7297, + "mean_token_accuracy": 0.7887148857116699, + "num_tokens": 79245564.0, + "step": 2072 + }, + { + "epoch": 0.2637069075181275, + "grad_norm": 14.1875, + "learning_rate": 8.783382789317507e-07, + "loss": 0.7355, + "mean_token_accuracy": 0.7865834832191467, + "num_tokens": 79281857.0, + "step": 2073 + }, + { + "epoch": 0.26383411779671795, + "grad_norm": 17.375, + "learning_rate": 8.787621873675286e-07, + "loss": 0.8229, + "mean_token_accuracy": 0.7624736428260803, + "num_tokens": 79312482.0, + "step": 2074 + }, + { + "epoch": 0.2639613280753085, + "grad_norm": 20.25, + "learning_rate": 8.791860958033065e-07, + "loss": 0.7417, + "mean_token_accuracy": 0.790332019329071, + "num_tokens": 79348157.0, + "step": 2075 + }, + { + "epoch": 0.264088538353899, + "grad_norm": 212.0, + "learning_rate": 8.796100042390842e-07, + "loss": 0.7114, + "mean_token_accuracy": 0.7949843406677246, + "num_tokens": 79387540.0, + "step": 2076 + }, + { + "epoch": 0.2642157486324895, + "grad_norm": 21.875, + "learning_rate": 8.800339126748622e-07, + "loss": 0.7459, + "mean_token_accuracy": 0.7833365797996521, + "num_tokens": 79421424.0, + "step": 2077 + }, + { + "epoch": 0.26434295891108, + "grad_norm": 19.875, + "learning_rate": 8.8045782111064e-07, + "loss": 0.8209, + "mean_token_accuracy": 0.7634813785552979, + "num_tokens": 79458733.0, + "step": 2078 + }, + { + "epoch": 0.26447016918967053, + "grad_norm": 173.0, + "learning_rate": 8.808817295464179e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.7794319987297058, + "num_tokens": 79495394.0, + "step": 2079 + }, + { + "epoch": 0.26459737946826106, + "grad_norm": 172.0, + "learning_rate": 8.813056379821958e-07, + "loss": 0.7448, + "mean_token_accuracy": 0.7857216596603394, + "num_tokens": 79529799.0, + "step": 2080 + }, + { + "epoch": 0.26472458974685154, + "grad_norm": 92.0, + "learning_rate": 8.817295464179737e-07, + "loss": 0.7766, + "mean_token_accuracy": 0.7770532369613647, + "num_tokens": 79564664.0, + "step": 2081 + }, + { + "epoch": 0.26485180002544206, + "grad_norm": 157.0, + "learning_rate": 8.821534548537515e-07, + "loss": 0.8618, + "mean_token_accuracy": 0.7518683075904846, + "num_tokens": 79606104.0, + "step": 2082 + }, + { + "epoch": 0.2649790103040326, + "grad_norm": 67.0, + "learning_rate": 8.825773632895295e-07, + "loss": 0.7059, + "mean_token_accuracy": 0.7982487678527832, + "num_tokens": 79647688.0, + "step": 2083 + }, + { + "epoch": 0.26510622058262306, + "grad_norm": 12.375, + "learning_rate": 8.830012717253072e-07, + "loss": 0.7465, + "mean_token_accuracy": 0.786696195602417, + "num_tokens": 79685500.0, + "step": 2084 + }, + { + "epoch": 0.2652334308612136, + "grad_norm": 52.0, + "learning_rate": 8.834251801610852e-07, + "loss": 0.7657, + "mean_token_accuracy": 0.7777054309844971, + "num_tokens": 79729912.0, + "step": 2085 + }, + { + "epoch": 0.2653606411398041, + "grad_norm": 87.5, + "learning_rate": 8.83849088596863e-07, + "loss": 0.7302, + "mean_token_accuracy": 0.7893820405006409, + "num_tokens": 79767487.0, + "step": 2086 + }, + { + "epoch": 0.2654878514183946, + "grad_norm": 239.0, + "learning_rate": 8.842729970326409e-07, + "loss": 0.6998, + "mean_token_accuracy": 0.798642635345459, + "num_tokens": 79804802.0, + "step": 2087 + }, + { + "epoch": 0.2656150616969851, + "grad_norm": 26.25, + "learning_rate": 8.846969054684188e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7654145359992981, + "num_tokens": 79845685.0, + "step": 2088 + }, + { + "epoch": 0.26574227197557565, + "grad_norm": 24.75, + "learning_rate": 8.851208139041967e-07, + "loss": 0.7137, + "mean_token_accuracy": 0.7939108610153198, + "num_tokens": 79884692.0, + "step": 2089 + }, + { + "epoch": 0.2658694822541661, + "grad_norm": 28.0, + "learning_rate": 8.855447223399745e-07, + "loss": 0.7827, + "mean_token_accuracy": 0.7702274322509766, + "num_tokens": 79919780.0, + "step": 2090 + }, + { + "epoch": 0.26599669253275665, + "grad_norm": 50.75, + "learning_rate": 8.859686307757524e-07, + "loss": 0.8169, + "mean_token_accuracy": 0.7655074000358582, + "num_tokens": 79958317.0, + "step": 2091 + }, + { + "epoch": 0.2661239028113472, + "grad_norm": 39.5, + "learning_rate": 8.863925392115302e-07, + "loss": 0.7208, + "mean_token_accuracy": 0.7889925241470337, + "num_tokens": 79994126.0, + "step": 2092 + }, + { + "epoch": 0.26625111308993765, + "grad_norm": 49.75, + "learning_rate": 8.868164476473082e-07, + "loss": 0.6943, + "mean_token_accuracy": 0.8017885088920593, + "num_tokens": 80033157.0, + "step": 2093 + }, + { + "epoch": 0.2663783233685282, + "grad_norm": 30.125, + "learning_rate": 8.87240356083086e-07, + "loss": 0.7113, + "mean_token_accuracy": 0.7945269346237183, + "num_tokens": 80073354.0, + "step": 2094 + }, + { + "epoch": 0.2665055336471187, + "grad_norm": 118.5, + "learning_rate": 8.876642645188639e-07, + "loss": 0.8054, + "mean_token_accuracy": 0.7684078216552734, + "num_tokens": 80111819.0, + "step": 2095 + }, + { + "epoch": 0.2666327439257092, + "grad_norm": 29.0, + "learning_rate": 8.880881729546418e-07, + "loss": 0.7127, + "mean_token_accuracy": 0.7950780391693115, + "num_tokens": 80147951.0, + "step": 2096 + }, + { + "epoch": 0.2667599542042997, + "grad_norm": 27.0, + "learning_rate": 8.885120813904197e-07, + "loss": 0.7117, + "mean_token_accuracy": 0.7947536110877991, + "num_tokens": 80191319.0, + "step": 2097 + }, + { + "epoch": 0.26688716448289024, + "grad_norm": 36.25, + "learning_rate": 8.889359898261976e-07, + "loss": 0.8118, + "mean_token_accuracy": 0.7675817608833313, + "num_tokens": 80226300.0, + "step": 2098 + }, + { + "epoch": 0.2670143747614807, + "grad_norm": 27.875, + "learning_rate": 8.893598982619753e-07, + "loss": 0.7288, + "mean_token_accuracy": 0.7907885313034058, + "num_tokens": 80265465.0, + "step": 2099 + }, + { + "epoch": 0.26714158504007124, + "grad_norm": 26.125, + "learning_rate": 8.897838066977532e-07, + "loss": 0.6946, + "mean_token_accuracy": 0.7962642908096313, + "num_tokens": 80312696.0, + "step": 2100 + }, + { + "epoch": 0.26726879531866177, + "grad_norm": 26.5, + "learning_rate": 8.902077151335311e-07, + "loss": 0.7627, + "mean_token_accuracy": 0.7782884836196899, + "num_tokens": 80353507.0, + "step": 2101 + }, + { + "epoch": 0.26739600559725224, + "grad_norm": 24.625, + "learning_rate": 8.90631623569309e-07, + "loss": 0.8579, + "mean_token_accuracy": 0.7512823343276978, + "num_tokens": 80390795.0, + "step": 2102 + }, + { + "epoch": 0.26752321587584277, + "grad_norm": 65.0, + "learning_rate": 8.910555320050868e-07, + "loss": 0.7485, + "mean_token_accuracy": 0.7831764221191406, + "num_tokens": 80427599.0, + "step": 2103 + }, + { + "epoch": 0.2676504261544333, + "grad_norm": 156.0, + "learning_rate": 8.914794404408648e-07, + "loss": 0.7527, + "mean_token_accuracy": 0.7839502692222595, + "num_tokens": 80468084.0, + "step": 2104 + }, + { + "epoch": 0.26777763643302377, + "grad_norm": 214.0, + "learning_rate": 8.919033488766426e-07, + "loss": 0.7044, + "mean_token_accuracy": 0.7985766530036926, + "num_tokens": 80507218.0, + "step": 2105 + }, + { + "epoch": 0.2679048467116143, + "grad_norm": 166.0, + "learning_rate": 8.923272573124204e-07, + "loss": 0.8004, + "mean_token_accuracy": 0.7670077681541443, + "num_tokens": 80541730.0, + "step": 2106 + }, + { + "epoch": 0.2680320569902048, + "grad_norm": 40.5, + "learning_rate": 8.927511657481983e-07, + "loss": 0.7995, + "mean_token_accuracy": 0.7706760168075562, + "num_tokens": 80584677.0, + "step": 2107 + }, + { + "epoch": 0.2681592672687953, + "grad_norm": 43.75, + "learning_rate": 8.931750741839762e-07, + "loss": 0.7499, + "mean_token_accuracy": 0.7850093841552734, + "num_tokens": 80616332.0, + "step": 2108 + }, + { + "epoch": 0.2682864775473858, + "grad_norm": 77.0, + "learning_rate": 8.935989826197541e-07, + "loss": 0.7295, + "mean_token_accuracy": 0.7905520796775818, + "num_tokens": 80654435.0, + "step": 2109 + }, + { + "epoch": 0.26841368782597635, + "grad_norm": 66.0, + "learning_rate": 8.94022891055532e-07, + "loss": 0.7104, + "mean_token_accuracy": 0.7970405220985413, + "num_tokens": 80691246.0, + "step": 2110 + }, + { + "epoch": 0.2685408981045668, + "grad_norm": 21.375, + "learning_rate": 8.944467994913098e-07, + "loss": 0.7742, + "mean_token_accuracy": 0.7768731713294983, + "num_tokens": 80730510.0, + "step": 2111 + }, + { + "epoch": 0.26866810838315736, + "grad_norm": 65.5, + "learning_rate": 8.948707079270878e-07, + "loss": 0.6693, + "mean_token_accuracy": 0.8054457902908325, + "num_tokens": 80769369.0, + "step": 2112 + }, + { + "epoch": 0.2687953186617479, + "grad_norm": 148.0, + "learning_rate": 8.952946163628656e-07, + "loss": 0.7571, + "mean_token_accuracy": 0.7804930806159973, + "num_tokens": 80802168.0, + "step": 2113 + }, + { + "epoch": 0.26892252894033836, + "grad_norm": 13.0, + "learning_rate": 8.957185247986434e-07, + "loss": 0.671, + "mean_token_accuracy": 0.8087128400802612, + "num_tokens": 80839389.0, + "step": 2114 + }, + { + "epoch": 0.2690497392189289, + "grad_norm": 120.5, + "learning_rate": 8.961424332344213e-07, + "loss": 0.7516, + "mean_token_accuracy": 0.7856001853942871, + "num_tokens": 80881139.0, + "step": 2115 + }, + { + "epoch": 0.2691769494975194, + "grad_norm": 19.25, + "learning_rate": 8.965663416701992e-07, + "loss": 0.7498, + "mean_token_accuracy": 0.7829147577285767, + "num_tokens": 80919809.0, + "step": 2116 + }, + { + "epoch": 0.2693041597761099, + "grad_norm": 22.375, + "learning_rate": 8.969902501059771e-07, + "loss": 0.7144, + "mean_token_accuracy": 0.7969199419021606, + "num_tokens": 80956860.0, + "step": 2117 + }, + { + "epoch": 0.2694313700547004, + "grad_norm": 178.0, + "learning_rate": 8.97414158541755e-07, + "loss": 0.6933, + "mean_token_accuracy": 0.7980151176452637, + "num_tokens": 80994322.0, + "step": 2118 + }, + { + "epoch": 0.26955858033329094, + "grad_norm": 75.5, + "learning_rate": 8.978380669775328e-07, + "loss": 0.7731, + "mean_token_accuracy": 0.7778419852256775, + "num_tokens": 81032792.0, + "step": 2119 + }, + { + "epoch": 0.2696857906118814, + "grad_norm": 66.0, + "learning_rate": 8.982619754133107e-07, + "loss": 0.7864, + "mean_token_accuracy": 0.7750107049942017, + "num_tokens": 81070011.0, + "step": 2120 + }, + { + "epoch": 0.26981300089047194, + "grad_norm": 112.5, + "learning_rate": 8.986858838490886e-07, + "loss": 0.7671, + "mean_token_accuracy": 0.7803998589515686, + "num_tokens": 81113145.0, + "step": 2121 + }, + { + "epoch": 0.26994021116906247, + "grad_norm": 101.0, + "learning_rate": 8.991097922848663e-07, + "loss": 0.7529, + "mean_token_accuracy": 0.7852916717529297, + "num_tokens": 81154175.0, + "step": 2122 + }, + { + "epoch": 0.27006742144765294, + "grad_norm": 38.5, + "learning_rate": 8.995337007206443e-07, + "loss": 0.7684, + "mean_token_accuracy": 0.7803192734718323, + "num_tokens": 81193319.0, + "step": 2123 + }, + { + "epoch": 0.2701946317262435, + "grad_norm": 31.5, + "learning_rate": 8.999576091564221e-07, + "loss": 0.7282, + "mean_token_accuracy": 0.7908725142478943, + "num_tokens": 81229057.0, + "step": 2124 + }, + { + "epoch": 0.270321842004834, + "grad_norm": 46.75, + "learning_rate": 9.003815175922001e-07, + "loss": 0.7406, + "mean_token_accuracy": 0.7871265411376953, + "num_tokens": 81262380.0, + "step": 2125 + }, + { + "epoch": 0.2704490522834245, + "grad_norm": 114.0, + "learning_rate": 9.008054260279779e-07, + "loss": 0.7349, + "mean_token_accuracy": 0.7896758317947388, + "num_tokens": 81304080.0, + "step": 2126 + }, + { + "epoch": 0.270576262562015, + "grad_norm": 58.75, + "learning_rate": 9.012293344637558e-07, + "loss": 0.7975, + "mean_token_accuracy": 0.7734614610671997, + "num_tokens": 81340226.0, + "step": 2127 + }, + { + "epoch": 0.27070347284060553, + "grad_norm": 242.0, + "learning_rate": 9.016532428995337e-07, + "loss": 0.7389, + "mean_token_accuracy": 0.7858285903930664, + "num_tokens": 81378046.0, + "step": 2128 + }, + { + "epoch": 0.27083068311919606, + "grad_norm": 140.0, + "learning_rate": 9.020771513353115e-07, + "loss": 0.7606, + "mean_token_accuracy": 0.7802364826202393, + "num_tokens": 81419486.0, + "step": 2129 + }, + { + "epoch": 0.27095789339778653, + "grad_norm": 30.875, + "learning_rate": 9.025010597710894e-07, + "loss": 0.7766, + "mean_token_accuracy": 0.7782824039459229, + "num_tokens": 81457488.0, + "step": 2130 + }, + { + "epoch": 0.27108510367637706, + "grad_norm": 26.0, + "learning_rate": 9.029249682068673e-07, + "loss": 0.7326, + "mean_token_accuracy": 0.7901873588562012, + "num_tokens": 81494812.0, + "step": 2131 + }, + { + "epoch": 0.2712123139549676, + "grad_norm": 27.875, + "learning_rate": 9.033488766426451e-07, + "loss": 0.7719, + "mean_token_accuracy": 0.7762293815612793, + "num_tokens": 81531421.0, + "step": 2132 + }, + { + "epoch": 0.27133952423355806, + "grad_norm": 49.0, + "learning_rate": 9.037727850784231e-07, + "loss": 0.7152, + "mean_token_accuracy": 0.7928662896156311, + "num_tokens": 81567221.0, + "step": 2133 + }, + { + "epoch": 0.2714667345121486, + "grad_norm": 26.625, + "learning_rate": 9.041966935142009e-07, + "loss": 0.7388, + "mean_token_accuracy": 0.786149263381958, + "num_tokens": 81605832.0, + "step": 2134 + }, + { + "epoch": 0.2715939447907391, + "grad_norm": 34.0, + "learning_rate": 9.046206019499788e-07, + "loss": 0.6813, + "mean_token_accuracy": 0.8047506809234619, + "num_tokens": 81646919.0, + "step": 2135 + }, + { + "epoch": 0.2717211550693296, + "grad_norm": 17.125, + "learning_rate": 9.050445103857567e-07, + "loss": 0.7983, + "mean_token_accuracy": 0.7779138088226318, + "num_tokens": 81687214.0, + "step": 2136 + }, + { + "epoch": 0.2718483653479201, + "grad_norm": 33.25, + "learning_rate": 9.054684188215344e-07, + "loss": 0.6796, + "mean_token_accuracy": 0.8030914068222046, + "num_tokens": 81721437.0, + "step": 2137 + }, + { + "epoch": 0.27197557562651065, + "grad_norm": 27.5, + "learning_rate": 9.058923272573124e-07, + "loss": 0.7684, + "mean_token_accuracy": 0.7838900089263916, + "num_tokens": 81756595.0, + "step": 2138 + }, + { + "epoch": 0.2721027859051011, + "grad_norm": 47.75, + "learning_rate": 9.063162356930902e-07, + "loss": 0.723, + "mean_token_accuracy": 0.7944537401199341, + "num_tokens": 81796198.0, + "step": 2139 + }, + { + "epoch": 0.27222999618369165, + "grad_norm": 132.0, + "learning_rate": 9.067401441288681e-07, + "loss": 0.769, + "mean_token_accuracy": 0.7846605181694031, + "num_tokens": 81833002.0, + "step": 2140 + }, + { + "epoch": 0.2723572064622822, + "grad_norm": 34.0, + "learning_rate": 9.07164052564646e-07, + "loss": 0.7218, + "mean_token_accuracy": 0.7838684916496277, + "num_tokens": 81872089.0, + "step": 2141 + }, + { + "epoch": 0.27248441674087265, + "grad_norm": 128.0, + "learning_rate": 9.075879610004239e-07, + "loss": 0.7742, + "mean_token_accuracy": 0.7746855020523071, + "num_tokens": 81908356.0, + "step": 2142 + }, + { + "epoch": 0.2726116270194632, + "grad_norm": 55.75, + "learning_rate": 9.080118694362017e-07, + "loss": 0.7778, + "mean_token_accuracy": 0.772913932800293, + "num_tokens": 81950251.0, + "step": 2143 + }, + { + "epoch": 0.2727388372980537, + "grad_norm": 21.75, + "learning_rate": 9.084357778719796e-07, + "loss": 0.7635, + "mean_token_accuracy": 0.7768823504447937, + "num_tokens": 81987157.0, + "step": 2144 + }, + { + "epoch": 0.2728660475766442, + "grad_norm": 40.0, + "learning_rate": 9.088596863077574e-07, + "loss": 0.7422, + "mean_token_accuracy": 0.7869187593460083, + "num_tokens": 82023633.0, + "step": 2145 + }, + { + "epoch": 0.2729932578552347, + "grad_norm": 13.5, + "learning_rate": 9.092835947435354e-07, + "loss": 0.7596, + "mean_token_accuracy": 0.780123233795166, + "num_tokens": 82061311.0, + "step": 2146 + }, + { + "epoch": 0.27312046813382523, + "grad_norm": 29.125, + "learning_rate": 9.097075031793132e-07, + "loss": 0.7858, + "mean_token_accuracy": 0.7720357179641724, + "num_tokens": 82098370.0, + "step": 2147 + }, + { + "epoch": 0.2732476784124157, + "grad_norm": 28.125, + "learning_rate": 9.101314116150911e-07, + "loss": 0.693, + "mean_token_accuracy": 0.801483154296875, + "num_tokens": 82134931.0, + "step": 2148 + }, + { + "epoch": 0.27337488869100623, + "grad_norm": 43.0, + "learning_rate": 9.10555320050869e-07, + "loss": 0.7179, + "mean_token_accuracy": 0.7913435697555542, + "num_tokens": 82171391.0, + "step": 2149 + }, + { + "epoch": 0.27350209896959676, + "grad_norm": 29.0, + "learning_rate": 9.109792284866469e-07, + "loss": 0.7072, + "mean_token_accuracy": 0.7950829267501831, + "num_tokens": 82214064.0, + "step": 2150 + }, + { + "epoch": 0.27362930924818724, + "grad_norm": 63.25, + "learning_rate": 9.114031369224247e-07, + "loss": 0.7584, + "mean_token_accuracy": 0.7815824151039124, + "num_tokens": 82252774.0, + "step": 2151 + }, + { + "epoch": 0.27375651952677776, + "grad_norm": 34.75, + "learning_rate": 9.118270453582026e-07, + "loss": 0.8073, + "mean_token_accuracy": 0.7707129716873169, + "num_tokens": 82291645.0, + "step": 2152 + }, + { + "epoch": 0.2738837298053683, + "grad_norm": 16.5, + "learning_rate": 9.122509537939804e-07, + "loss": 0.7363, + "mean_token_accuracy": 0.7834845781326294, + "num_tokens": 82333643.0, + "step": 2153 + }, + { + "epoch": 0.27401094008395877, + "grad_norm": 93.0, + "learning_rate": 9.126748622297584e-07, + "loss": 0.785, + "mean_token_accuracy": 0.7686198949813843, + "num_tokens": 82370792.0, + "step": 2154 + }, + { + "epoch": 0.2741381503625493, + "grad_norm": 17.375, + "learning_rate": 9.130987706655362e-07, + "loss": 0.7804, + "mean_token_accuracy": 0.7760292291641235, + "num_tokens": 82408626.0, + "step": 2155 + }, + { + "epoch": 0.2742653606411398, + "grad_norm": 29.0, + "learning_rate": 9.135226791013141e-07, + "loss": 0.7848, + "mean_token_accuracy": 0.7799584269523621, + "num_tokens": 82448401.0, + "step": 2156 + }, + { + "epoch": 0.2743925709197303, + "grad_norm": 19.125, + "learning_rate": 9.13946587537092e-07, + "loss": 0.7639, + "mean_token_accuracy": 0.7749298214912415, + "num_tokens": 82476644.0, + "step": 2157 + }, + { + "epoch": 0.2745197811983208, + "grad_norm": 51.5, + "learning_rate": 9.143704959728699e-07, + "loss": 0.7718, + "mean_token_accuracy": 0.7756527662277222, + "num_tokens": 82512722.0, + "step": 2158 + }, + { + "epoch": 0.27464699147691135, + "grad_norm": 100.0, + "learning_rate": 9.147944044086476e-07, + "loss": 0.8092, + "mean_token_accuracy": 0.7652664184570312, + "num_tokens": 82545393.0, + "step": 2159 + }, + { + "epoch": 0.2747742017555018, + "grad_norm": 20.625, + "learning_rate": 9.152183128444255e-07, + "loss": 0.7659, + "mean_token_accuracy": 0.7828906178474426, + "num_tokens": 82586329.0, + "step": 2160 + }, + { + "epoch": 0.27490141203409235, + "grad_norm": 16.375, + "learning_rate": 9.156422212802034e-07, + "loss": 0.8068, + "mean_token_accuracy": 0.765978217124939, + "num_tokens": 82625563.0, + "step": 2161 + }, + { + "epoch": 0.2750286223126829, + "grad_norm": 62.75, + "learning_rate": 9.160661297159813e-07, + "loss": 0.7298, + "mean_token_accuracy": 0.7916184663772583, + "num_tokens": 82660030.0, + "step": 2162 + }, + { + "epoch": 0.27515583259127335, + "grad_norm": 71.0, + "learning_rate": 9.164900381517592e-07, + "loss": 0.752, + "mean_token_accuracy": 0.7826022505760193, + "num_tokens": 82698794.0, + "step": 2163 + }, + { + "epoch": 0.2752830428698639, + "grad_norm": 44.0, + "learning_rate": 9.16913946587537e-07, + "loss": 0.7428, + "mean_token_accuracy": 0.7861111760139465, + "num_tokens": 82740015.0, + "step": 2164 + }, + { + "epoch": 0.2754102531484544, + "grad_norm": 54.75, + "learning_rate": 9.17337855023315e-07, + "loss": 0.7778, + "mean_token_accuracy": 0.7729068994522095, + "num_tokens": 82775765.0, + "step": 2165 + }, + { + "epoch": 0.2755374634270449, + "grad_norm": 33.0, + "learning_rate": 9.177617634590928e-07, + "loss": 0.6874, + "mean_token_accuracy": 0.800715982913971, + "num_tokens": 82814324.0, + "step": 2166 + }, + { + "epoch": 0.2756646737056354, + "grad_norm": 107.0, + "learning_rate": 9.181856718948706e-07, + "loss": 0.6861, + "mean_token_accuracy": 0.8008350133895874, + "num_tokens": 82851007.0, + "step": 2167 + }, + { + "epoch": 0.27579188398422594, + "grad_norm": 80.5, + "learning_rate": 9.186095803306485e-07, + "loss": 0.7351, + "mean_token_accuracy": 0.7862873673439026, + "num_tokens": 82886846.0, + "step": 2168 + }, + { + "epoch": 0.2759190942628164, + "grad_norm": 95.5, + "learning_rate": 9.190334887664264e-07, + "loss": 0.765, + "mean_token_accuracy": 0.7781859040260315, + "num_tokens": 82929607.0, + "step": 2169 + }, + { + "epoch": 0.27604630454140694, + "grad_norm": 129.0, + "learning_rate": 9.194573972022043e-07, + "loss": 0.8699, + "mean_token_accuracy": 0.7626181840896606, + "num_tokens": 82965089.0, + "step": 2170 + }, + { + "epoch": 0.27617351481999747, + "grad_norm": 18.875, + "learning_rate": 9.198813056379822e-07, + "loss": 0.8144, + "mean_token_accuracy": 0.7651336193084717, + "num_tokens": 83008345.0, + "step": 2171 + }, + { + "epoch": 0.27630072509858794, + "grad_norm": 58.5, + "learning_rate": 9.2030521407376e-07, + "loss": 0.7709, + "mean_token_accuracy": 0.7790224552154541, + "num_tokens": 83043763.0, + "step": 2172 + }, + { + "epoch": 0.27642793537717847, + "grad_norm": 125.5, + "learning_rate": 9.20729122509538e-07, + "loss": 0.7114, + "mean_token_accuracy": 0.7953469157218933, + "num_tokens": 83081389.0, + "step": 2173 + }, + { + "epoch": 0.276555145655769, + "grad_norm": 51.5, + "learning_rate": 9.211530309453158e-07, + "loss": 0.7385, + "mean_token_accuracy": 0.7836927771568298, + "num_tokens": 83120547.0, + "step": 2174 + }, + { + "epoch": 0.27668235593435947, + "grad_norm": 380.0, + "learning_rate": 9.215769393810936e-07, + "loss": 0.7994, + "mean_token_accuracy": 0.7730724811553955, + "num_tokens": 83157229.0, + "step": 2175 + }, + { + "epoch": 0.27680956621295, + "grad_norm": 75.5, + "learning_rate": 9.220008478168715e-07, + "loss": 0.841, + "mean_token_accuracy": 0.7597129940986633, + "num_tokens": 83198217.0, + "step": 2176 + }, + { + "epoch": 0.2769367764915405, + "grad_norm": 96.0, + "learning_rate": 9.224247562526494e-07, + "loss": 0.7647, + "mean_token_accuracy": 0.7793387174606323, + "num_tokens": 83238750.0, + "step": 2177 + }, + { + "epoch": 0.277063986770131, + "grad_norm": 132.0, + "learning_rate": 9.228486646884273e-07, + "loss": 0.7337, + "mean_token_accuracy": 0.7872141003608704, + "num_tokens": 83280138.0, + "step": 2178 + }, + { + "epoch": 0.2771911970487215, + "grad_norm": 30.25, + "learning_rate": 9.232725731242052e-07, + "loss": 0.824, + "mean_token_accuracy": 0.7644186019897461, + "num_tokens": 83320633.0, + "step": 2179 + }, + { + "epoch": 0.27731840732731206, + "grad_norm": 108.5, + "learning_rate": 9.23696481559983e-07, + "loss": 0.7666, + "mean_token_accuracy": 0.7757045030593872, + "num_tokens": 83358787.0, + "step": 2180 + }, + { + "epoch": 0.2774456176059026, + "grad_norm": 39.25, + "learning_rate": 9.24120389995761e-07, + "loss": 0.7104, + "mean_token_accuracy": 0.7942614555358887, + "num_tokens": 83400609.0, + "step": 2181 + }, + { + "epoch": 0.27757282788449306, + "grad_norm": 134.0, + "learning_rate": 9.245442984315387e-07, + "loss": 0.744, + "mean_token_accuracy": 0.7850695848464966, + "num_tokens": 83436058.0, + "step": 2182 + }, + { + "epoch": 0.2777000381630836, + "grad_norm": 111.0, + "learning_rate": 9.249682068673165e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7775319814682007, + "num_tokens": 83474542.0, + "step": 2183 + }, + { + "epoch": 0.2778272484416741, + "grad_norm": 46.25, + "learning_rate": 9.253921153030945e-07, + "loss": 0.8239, + "mean_token_accuracy": 0.7642793655395508, + "num_tokens": 83515551.0, + "step": 2184 + }, + { + "epoch": 0.2779544587202646, + "grad_norm": 45.5, + "learning_rate": 9.258160237388723e-07, + "loss": 0.7131, + "mean_token_accuracy": 0.7940850257873535, + "num_tokens": 83551826.0, + "step": 2185 + }, + { + "epoch": 0.2780816689988551, + "grad_norm": 32.0, + "learning_rate": 9.262399321746503e-07, + "loss": 0.7464, + "mean_token_accuracy": 0.7869664430618286, + "num_tokens": 83586598.0, + "step": 2186 + }, + { + "epoch": 0.27820887927744564, + "grad_norm": 31.125, + "learning_rate": 9.266638406104281e-07, + "loss": 0.7899, + "mean_token_accuracy": 0.7707154750823975, + "num_tokens": 83625996.0, + "step": 2187 + }, + { + "epoch": 0.2783360895560361, + "grad_norm": 24.125, + "learning_rate": 9.27087749046206e-07, + "loss": 0.7766, + "mean_token_accuracy": 0.7751765251159668, + "num_tokens": 83657322.0, + "step": 2188 + }, + { + "epoch": 0.27846329983462664, + "grad_norm": 39.75, + "learning_rate": 9.275116574819839e-07, + "loss": 0.7653, + "mean_token_accuracy": 0.7799547910690308, + "num_tokens": 83693578.0, + "step": 2189 + }, + { + "epoch": 0.27859051011321717, + "grad_norm": 91.5, + "learning_rate": 9.279355659177617e-07, + "loss": 0.7621, + "mean_token_accuracy": 0.7801242470741272, + "num_tokens": 83727672.0, + "step": 2190 + }, + { + "epoch": 0.27871772039180764, + "grad_norm": 62.0, + "learning_rate": 9.283594743535395e-07, + "loss": 0.8246, + "mean_token_accuracy": 0.7672414779663086, + "num_tokens": 83761405.0, + "step": 2191 + }, + { + "epoch": 0.2788449306703982, + "grad_norm": 20.875, + "learning_rate": 9.287833827893175e-07, + "loss": 0.7193, + "mean_token_accuracy": 0.7862248420715332, + "num_tokens": 83801564.0, + "step": 2192 + }, + { + "epoch": 0.2789721409489887, + "grad_norm": 122.5, + "learning_rate": 9.292072912250953e-07, + "loss": 0.7797, + "mean_token_accuracy": 0.7751803398132324, + "num_tokens": 83834260.0, + "step": 2193 + }, + { + "epoch": 0.2790993512275792, + "grad_norm": 25.5, + "learning_rate": 9.296311996608733e-07, + "loss": 0.6798, + "mean_token_accuracy": 0.7991218566894531, + "num_tokens": 83873518.0, + "step": 2194 + }, + { + "epoch": 0.2792265615061697, + "grad_norm": 31.5, + "learning_rate": 9.300551080966511e-07, + "loss": 0.8112, + "mean_token_accuracy": 0.7659183740615845, + "num_tokens": 83913902.0, + "step": 2195 + }, + { + "epoch": 0.27935377178476023, + "grad_norm": 22.0, + "learning_rate": 9.30479016532429e-07, + "loss": 0.7497, + "mean_token_accuracy": 0.7815330624580383, + "num_tokens": 83947952.0, + "step": 2196 + }, + { + "epoch": 0.2794809820633507, + "grad_norm": 81.0, + "learning_rate": 9.309029249682068e-07, + "loss": 0.7607, + "mean_token_accuracy": 0.7806206941604614, + "num_tokens": 83986057.0, + "step": 2197 + }, + { + "epoch": 0.27960819234194123, + "grad_norm": 140.0, + "learning_rate": 9.313268334039847e-07, + "loss": 0.7676, + "mean_token_accuracy": 0.7775278687477112, + "num_tokens": 84023881.0, + "step": 2198 + }, + { + "epoch": 0.27973540262053176, + "grad_norm": 57.25, + "learning_rate": 9.317507418397625e-07, + "loss": 0.6745, + "mean_token_accuracy": 0.8057558536529541, + "num_tokens": 84061480.0, + "step": 2199 + }, + { + "epoch": 0.27986261289912223, + "grad_norm": 89.5, + "learning_rate": 9.321746502755404e-07, + "loss": 0.8011, + "mean_token_accuracy": 0.7725868225097656, + "num_tokens": 84101782.0, + "step": 2200 + }, + { + "epoch": 0.27998982317771276, + "grad_norm": 24.875, + "learning_rate": 9.325985587113183e-07, + "loss": 0.7591, + "mean_token_accuracy": 0.7821993827819824, + "num_tokens": 84136125.0, + "step": 2201 + }, + { + "epoch": 0.2801170334563033, + "grad_norm": 29.375, + "learning_rate": 9.330224671470962e-07, + "loss": 0.7777, + "mean_token_accuracy": 0.7771921753883362, + "num_tokens": 84169748.0, + "step": 2202 + }, + { + "epoch": 0.28024424373489376, + "grad_norm": 94.0, + "learning_rate": 9.334463755828741e-07, + "loss": 0.8216, + "mean_token_accuracy": 0.7612569332122803, + "num_tokens": 84201292.0, + "step": 2203 + }, + { + "epoch": 0.2803714540134843, + "grad_norm": 66.0, + "learning_rate": 9.338702840186519e-07, + "loss": 0.8037, + "mean_token_accuracy": 0.7675267457962036, + "num_tokens": 84240652.0, + "step": 2204 + }, + { + "epoch": 0.2804986642920748, + "grad_norm": 140.0, + "learning_rate": 9.342941924544298e-07, + "loss": 0.7911, + "mean_token_accuracy": 0.7718150615692139, + "num_tokens": 84280252.0, + "step": 2205 + }, + { + "epoch": 0.2806258745706653, + "grad_norm": 163.0, + "learning_rate": 9.347181008902076e-07, + "loss": 0.7181, + "mean_token_accuracy": 0.7962092757225037, + "num_tokens": 84316505.0, + "step": 2206 + }, + { + "epoch": 0.2807530848492558, + "grad_norm": 113.0, + "learning_rate": 9.351420093259855e-07, + "loss": 0.72, + "mean_token_accuracy": 0.7928687334060669, + "num_tokens": 84349990.0, + "step": 2207 + }, + { + "epoch": 0.28088029512784635, + "grad_norm": 45.0, + "learning_rate": 9.355659177617634e-07, + "loss": 0.7189, + "mean_token_accuracy": 0.7893624901771545, + "num_tokens": 84381066.0, + "step": 2208 + }, + { + "epoch": 0.2810075054064368, + "grad_norm": 34.5, + "learning_rate": 9.359898261975413e-07, + "loss": 0.6761, + "mean_token_accuracy": 0.8022060990333557, + "num_tokens": 84422540.0, + "step": 2209 + }, + { + "epoch": 0.28113471568502735, + "grad_norm": 40.0, + "learning_rate": 9.364137346333192e-07, + "loss": 0.8267, + "mean_token_accuracy": 0.7636405229568481, + "num_tokens": 84462158.0, + "step": 2210 + }, + { + "epoch": 0.2812619259636179, + "grad_norm": 66.0, + "learning_rate": 9.368376430690971e-07, + "loss": 0.7126, + "mean_token_accuracy": 0.7924628257751465, + "num_tokens": 84495887.0, + "step": 2211 + }, + { + "epoch": 0.28138913624220835, + "grad_norm": 37.0, + "learning_rate": 9.372615515048749e-07, + "loss": 0.804, + "mean_token_accuracy": 0.76906818151474, + "num_tokens": 84530176.0, + "step": 2212 + }, + { + "epoch": 0.2815163465207989, + "grad_norm": 119.0, + "learning_rate": 9.376854599406528e-07, + "loss": 0.785, + "mean_token_accuracy": 0.7742470502853394, + "num_tokens": 84574450.0, + "step": 2213 + }, + { + "epoch": 0.2816435567993894, + "grad_norm": 39.0, + "learning_rate": 9.381093683764306e-07, + "loss": 0.7715, + "mean_token_accuracy": 0.7798346877098083, + "num_tokens": 84610661.0, + "step": 2214 + }, + { + "epoch": 0.2817707670779799, + "grad_norm": 106.0, + "learning_rate": 9.385332768122085e-07, + "loss": 0.7332, + "mean_token_accuracy": 0.7858439087867737, + "num_tokens": 84651259.0, + "step": 2215 + }, + { + "epoch": 0.2818979773565704, + "grad_norm": 81.0, + "learning_rate": 9.389571852479864e-07, + "loss": 0.7332, + "mean_token_accuracy": 0.7907202243804932, + "num_tokens": 84692389.0, + "step": 2216 + }, + { + "epoch": 0.28202518763516093, + "grad_norm": 282.0, + "learning_rate": 9.393810936837643e-07, + "loss": 0.7521, + "mean_token_accuracy": 0.7825090885162354, + "num_tokens": 84731905.0, + "step": 2217 + }, + { + "epoch": 0.2821523979137514, + "grad_norm": 95.5, + "learning_rate": 9.398050021195422e-07, + "loss": 0.769, + "mean_token_accuracy": 0.7819418907165527, + "num_tokens": 84773727.0, + "step": 2218 + }, + { + "epoch": 0.28227960819234194, + "grad_norm": 75.5, + "learning_rate": 9.402289105553201e-07, + "loss": 0.7091, + "mean_token_accuracy": 0.7925846576690674, + "num_tokens": 84816771.0, + "step": 2219 + }, + { + "epoch": 0.28240681847093246, + "grad_norm": 22.5, + "learning_rate": 9.406528189910978e-07, + "loss": 0.7309, + "mean_token_accuracy": 0.786645770072937, + "num_tokens": 84853172.0, + "step": 2220 + }, + { + "epoch": 0.28253402874952294, + "grad_norm": 33.75, + "learning_rate": 9.410767274268757e-07, + "loss": 0.7237, + "mean_token_accuracy": 0.7913668751716614, + "num_tokens": 84891508.0, + "step": 2221 + }, + { + "epoch": 0.28266123902811346, + "grad_norm": 18.75, + "learning_rate": 9.415006358626536e-07, + "loss": 0.8247, + "mean_token_accuracy": 0.7612941265106201, + "num_tokens": 84931204.0, + "step": 2222 + }, + { + "epoch": 0.282788449306704, + "grad_norm": 262.0, + "learning_rate": 9.419245442984314e-07, + "loss": 0.8321, + "mean_token_accuracy": 0.7625511884689331, + "num_tokens": 84966782.0, + "step": 2223 + }, + { + "epoch": 0.28291565958529447, + "grad_norm": 260.0, + "learning_rate": 9.423484527342094e-07, + "loss": 0.7745, + "mean_token_accuracy": 0.7767671346664429, + "num_tokens": 85001512.0, + "step": 2224 + }, + { + "epoch": 0.283042869863885, + "grad_norm": 296.0, + "learning_rate": 9.427723611699872e-07, + "loss": 0.7819, + "mean_token_accuracy": 0.7754942178726196, + "num_tokens": 85036082.0, + "step": 2225 + }, + { + "epoch": 0.2831700801424755, + "grad_norm": 302.0, + "learning_rate": 9.431962696057652e-07, + "loss": 0.7863, + "mean_token_accuracy": 0.7771806120872498, + "num_tokens": 85079433.0, + "step": 2226 + }, + { + "epoch": 0.283297290421066, + "grad_norm": 234.0, + "learning_rate": 9.43620178041543e-07, + "loss": 0.719, + "mean_token_accuracy": 0.794674277305603, + "num_tokens": 85116940.0, + "step": 2227 + }, + { + "epoch": 0.2834245006996565, + "grad_norm": 229.0, + "learning_rate": 9.440440864773208e-07, + "loss": 0.7566, + "mean_token_accuracy": 0.7835647463798523, + "num_tokens": 85153328.0, + "step": 2228 + }, + { + "epoch": 0.28355171097824705, + "grad_norm": 77.0, + "learning_rate": 9.444679949130987e-07, + "loss": 0.7711, + "mean_token_accuracy": 0.7768276929855347, + "num_tokens": 85194059.0, + "step": 2229 + }, + { + "epoch": 0.2836789212568376, + "grad_norm": 21.125, + "learning_rate": 9.448919033488766e-07, + "loss": 0.7992, + "mean_token_accuracy": 0.77072674036026, + "num_tokens": 85230820.0, + "step": 2230 + }, + { + "epoch": 0.28380613153542805, + "grad_norm": 79.0, + "learning_rate": 9.453158117846544e-07, + "loss": 0.7786, + "mean_token_accuracy": 0.7761539220809937, + "num_tokens": 85268864.0, + "step": 2231 + }, + { + "epoch": 0.2839333418140186, + "grad_norm": 151.0, + "learning_rate": 9.457397202204324e-07, + "loss": 0.8086, + "mean_token_accuracy": 0.7686559557914734, + "num_tokens": 85311377.0, + "step": 2232 + }, + { + "epoch": 0.2840605520926091, + "grad_norm": 92.0, + "learning_rate": 9.461636286562102e-07, + "loss": 0.7079, + "mean_token_accuracy": 0.797231912612915, + "num_tokens": 85347082.0, + "step": 2233 + }, + { + "epoch": 0.2841877623711996, + "grad_norm": 241.0, + "learning_rate": 9.465875370919882e-07, + "loss": 0.8162, + "mean_token_accuracy": 0.7710364460945129, + "num_tokens": 85384696.0, + "step": 2234 + }, + { + "epoch": 0.2843149726497901, + "grad_norm": 234.0, + "learning_rate": 9.470114455277659e-07, + "loss": 0.821, + "mean_token_accuracy": 0.7641196846961975, + "num_tokens": 85420192.0, + "step": 2235 + }, + { + "epoch": 0.28444218292838064, + "grad_norm": 276.0, + "learning_rate": 9.474353539635438e-07, + "loss": 0.7437, + "mean_token_accuracy": 0.7827866077423096, + "num_tokens": 85462281.0, + "step": 2236 + }, + { + "epoch": 0.2845693932069711, + "grad_norm": 149.0, + "learning_rate": 9.478592623993217e-07, + "loss": 0.7832, + "mean_token_accuracy": 0.7726786732673645, + "num_tokens": 85495191.0, + "step": 2237 + }, + { + "epoch": 0.28469660348556164, + "grad_norm": 16.375, + "learning_rate": 9.482831708350996e-07, + "loss": 0.7312, + "mean_token_accuracy": 0.7899256944656372, + "num_tokens": 85533803.0, + "step": 2238 + }, + { + "epoch": 0.28482381376415217, + "grad_norm": 80.0, + "learning_rate": 9.487070792708775e-07, + "loss": 0.7947, + "mean_token_accuracy": 0.7708649635314941, + "num_tokens": 85573424.0, + "step": 2239 + }, + { + "epoch": 0.28495102404274264, + "grad_norm": 97.0, + "learning_rate": 9.491309877066554e-07, + "loss": 0.7642, + "mean_token_accuracy": 0.7785055637359619, + "num_tokens": 85616423.0, + "step": 2240 + }, + { + "epoch": 0.28507823432133317, + "grad_norm": 102.0, + "learning_rate": 9.495548961424332e-07, + "loss": 0.8101, + "mean_token_accuracy": 0.7670317888259888, + "num_tokens": 85658410.0, + "step": 2241 + }, + { + "epoch": 0.2852054445999237, + "grad_norm": 91.0, + "learning_rate": 9.499788045782111e-07, + "loss": 0.7587, + "mean_token_accuracy": 0.7783046960830688, + "num_tokens": 85695767.0, + "step": 2242 + }, + { + "epoch": 0.28533265487851417, + "grad_norm": 59.0, + "learning_rate": 9.504027130139889e-07, + "loss": 0.7782, + "mean_token_accuracy": 0.7781447172164917, + "num_tokens": 85737771.0, + "step": 2243 + }, + { + "epoch": 0.2854598651571047, + "grad_norm": 90.0, + "learning_rate": 9.508266214497667e-07, + "loss": 0.7013, + "mean_token_accuracy": 0.794985830783844, + "num_tokens": 85774789.0, + "step": 2244 + }, + { + "epoch": 0.2855870754356952, + "grad_norm": 113.0, + "learning_rate": 9.512505298855447e-07, + "loss": 0.7552, + "mean_token_accuracy": 0.7836655378341675, + "num_tokens": 85817833.0, + "step": 2245 + }, + { + "epoch": 0.2857142857142857, + "grad_norm": 59.0, + "learning_rate": 9.516744383213225e-07, + "loss": 0.6947, + "mean_token_accuracy": 0.7962862253189087, + "num_tokens": 85854488.0, + "step": 2246 + }, + { + "epoch": 0.2858414959928762, + "grad_norm": 94.0, + "learning_rate": 9.520983467571005e-07, + "loss": 0.7815, + "mean_token_accuracy": 0.7741022706031799, + "num_tokens": 85893746.0, + "step": 2247 + }, + { + "epoch": 0.28596870627146675, + "grad_norm": 210.0, + "learning_rate": 9.525222551928783e-07, + "loss": 0.8569, + "mean_token_accuracy": 0.765052080154419, + "num_tokens": 85932798.0, + "step": 2248 + }, + { + "epoch": 0.2860959165500572, + "grad_norm": 131.0, + "learning_rate": 9.529461636286562e-07, + "loss": 0.7522, + "mean_token_accuracy": 0.7797411680221558, + "num_tokens": 85964446.0, + "step": 2249 + }, + { + "epoch": 0.28622312682864776, + "grad_norm": 201.0, + "learning_rate": 9.533700720644341e-07, + "loss": 0.7114, + "mean_token_accuracy": 0.7946450710296631, + "num_tokens": 86000587.0, + "step": 2250 + }, + { + "epoch": 0.2863503371072383, + "grad_norm": 25.0, + "learning_rate": 9.537939805002118e-07, + "loss": 0.7532, + "mean_token_accuracy": 0.7861872911453247, + "num_tokens": 86042891.0, + "step": 2251 + }, + { + "epoch": 0.28647754738582876, + "grad_norm": 25.875, + "learning_rate": 9.542178889359898e-07, + "loss": 0.7234, + "mean_token_accuracy": 0.7931737899780273, + "num_tokens": 86079973.0, + "step": 2252 + }, + { + "epoch": 0.2866047576644193, + "grad_norm": 17.875, + "learning_rate": 9.546417973717677e-07, + "loss": 0.7552, + "mean_token_accuracy": 0.7853553891181946, + "num_tokens": 86119767.0, + "step": 2253 + }, + { + "epoch": 0.2867319679430098, + "grad_norm": 41.25, + "learning_rate": 9.550657058075455e-07, + "loss": 0.7831, + "mean_token_accuracy": 0.7760775685310364, + "num_tokens": 86163176.0, + "step": 2254 + }, + { + "epoch": 0.2868591782216003, + "grad_norm": 27.875, + "learning_rate": 9.554896142433234e-07, + "loss": 0.8021, + "mean_token_accuracy": 0.7680463790893555, + "num_tokens": 86201523.0, + "step": 2255 + }, + { + "epoch": 0.2869863885001908, + "grad_norm": 105.0, + "learning_rate": 9.559135226791012e-07, + "loss": 0.8086, + "mean_token_accuracy": 0.7633643746376038, + "num_tokens": 86240277.0, + "step": 2256 + }, + { + "epoch": 0.28711359877878134, + "grad_norm": 63.5, + "learning_rate": 9.563374311148793e-07, + "loss": 0.7766, + "mean_token_accuracy": 0.7790718078613281, + "num_tokens": 86289714.0, + "step": 2257 + }, + { + "epoch": 0.2872408090573718, + "grad_norm": 139.0, + "learning_rate": 9.56761339550657e-07, + "loss": 0.755, + "mean_token_accuracy": 0.7838189005851746, + "num_tokens": 86320786.0, + "step": 2258 + }, + { + "epoch": 0.28736801933596234, + "grad_norm": 57.75, + "learning_rate": 9.57185247986435e-07, + "loss": 0.7313, + "mean_token_accuracy": 0.7917027473449707, + "num_tokens": 86358002.0, + "step": 2259 + }, + { + "epoch": 0.28749522961455287, + "grad_norm": 46.25, + "learning_rate": 9.576091564222128e-07, + "loss": 0.6842, + "mean_token_accuracy": 0.7976301908493042, + "num_tokens": 86395334.0, + "step": 2260 + }, + { + "epoch": 0.28762243989314334, + "grad_norm": 59.5, + "learning_rate": 9.580330648579906e-07, + "loss": 0.7717, + "mean_token_accuracy": 0.7756505012512207, + "num_tokens": 86432616.0, + "step": 2261 + }, + { + "epoch": 0.2877496501717339, + "grad_norm": 23.375, + "learning_rate": 9.584569732937685e-07, + "loss": 0.7665, + "mean_token_accuracy": 0.7807134389877319, + "num_tokens": 86472293.0, + "step": 2262 + }, + { + "epoch": 0.2878768604503244, + "grad_norm": 19.75, + "learning_rate": 9.588808817295463e-07, + "loss": 0.7791, + "mean_token_accuracy": 0.7721976637840271, + "num_tokens": 86506304.0, + "step": 2263 + }, + { + "epoch": 0.2880040707289149, + "grad_norm": 11.9375, + "learning_rate": 9.593047901653242e-07, + "loss": 0.7661, + "mean_token_accuracy": 0.7751096487045288, + "num_tokens": 86540520.0, + "step": 2264 + }, + { + "epoch": 0.2881312810075054, + "grad_norm": 19.25, + "learning_rate": 9.597286986011022e-07, + "loss": 0.7144, + "mean_token_accuracy": 0.7896242141723633, + "num_tokens": 86576956.0, + "step": 2265 + }, + { + "epoch": 0.28825849128609593, + "grad_norm": 21.875, + "learning_rate": 9.601526070368799e-07, + "loss": 0.8104, + "mean_token_accuracy": 0.7671276330947876, + "num_tokens": 86613799.0, + "step": 2266 + }, + { + "epoch": 0.2883857015646864, + "grad_norm": 23.625, + "learning_rate": 9.60576515472658e-07, + "loss": 0.7457, + "mean_token_accuracy": 0.7838780879974365, + "num_tokens": 86650338.0, + "step": 2267 + }, + { + "epoch": 0.28851291184327693, + "grad_norm": 22.25, + "learning_rate": 9.610004239084358e-07, + "loss": 0.7494, + "mean_token_accuracy": 0.7844067215919495, + "num_tokens": 86691605.0, + "step": 2268 + }, + { + "epoch": 0.28864012212186746, + "grad_norm": 47.25, + "learning_rate": 9.614243323442136e-07, + "loss": 0.8308, + "mean_token_accuracy": 0.7603006362915039, + "num_tokens": 86728767.0, + "step": 2269 + }, + { + "epoch": 0.28876733240045793, + "grad_norm": 45.75, + "learning_rate": 9.618482407799915e-07, + "loss": 0.7399, + "mean_token_accuracy": 0.7849304676055908, + "num_tokens": 86765240.0, + "step": 2270 + }, + { + "epoch": 0.28889454267904846, + "grad_norm": 123.5, + "learning_rate": 9.622721492157693e-07, + "loss": 0.6913, + "mean_token_accuracy": 0.8000549077987671, + "num_tokens": 86804042.0, + "step": 2271 + }, + { + "epoch": 0.289021752957639, + "grad_norm": 51.75, + "learning_rate": 9.626960576515472e-07, + "loss": 0.7597, + "mean_token_accuracy": 0.7800557613372803, + "num_tokens": 86835220.0, + "step": 2272 + }, + { + "epoch": 0.28914896323622946, + "grad_norm": 52.75, + "learning_rate": 9.63119966087325e-07, + "loss": 0.7397, + "mean_token_accuracy": 0.7850370407104492, + "num_tokens": 86873414.0, + "step": 2273 + }, + { + "epoch": 0.28927617351482, + "grad_norm": 12.75, + "learning_rate": 9.635438745231029e-07, + "loss": 0.7063, + "mean_token_accuracy": 0.7980777025222778, + "num_tokens": 86914272.0, + "step": 2274 + }, + { + "epoch": 0.2894033837934105, + "grad_norm": 40.0, + "learning_rate": 9.63967782958881e-07, + "loss": 0.7782, + "mean_token_accuracy": 0.7748173475265503, + "num_tokens": 86952425.0, + "step": 2275 + }, + { + "epoch": 0.289530594072001, + "grad_norm": 30.75, + "learning_rate": 9.643916913946588e-07, + "loss": 0.742, + "mean_token_accuracy": 0.7853078246116638, + "num_tokens": 86987842.0, + "step": 2276 + }, + { + "epoch": 0.2896578043505915, + "grad_norm": 31.0, + "learning_rate": 9.648155998304366e-07, + "loss": 0.6795, + "mean_token_accuracy": 0.8011749982833862, + "num_tokens": 87026750.0, + "step": 2277 + }, + { + "epoch": 0.28978501462918205, + "grad_norm": 40.0, + "learning_rate": 9.652395082662145e-07, + "loss": 0.7992, + "mean_token_accuracy": 0.7678777575492859, + "num_tokens": 87069685.0, + "step": 2278 + }, + { + "epoch": 0.2899122249077726, + "grad_norm": 46.0, + "learning_rate": 9.656634167019923e-07, + "loss": 0.7591, + "mean_token_accuracy": 0.7782593369483948, + "num_tokens": 87107054.0, + "step": 2279 + }, + { + "epoch": 0.29003943518636305, + "grad_norm": 37.0, + "learning_rate": 9.660873251377701e-07, + "loss": 0.7549, + "mean_token_accuracy": 0.7770693302154541, + "num_tokens": 87146254.0, + "step": 2280 + }, + { + "epoch": 0.2901666454649536, + "grad_norm": 59.5, + "learning_rate": 9.66511233573548e-07, + "loss": 0.8021, + "mean_token_accuracy": 0.7741283178329468, + "num_tokens": 87182404.0, + "step": 2281 + }, + { + "epoch": 0.2902938557435441, + "grad_norm": 22.75, + "learning_rate": 9.669351420093258e-07, + "loss": 0.72, + "mean_token_accuracy": 0.7894275188446045, + "num_tokens": 87216551.0, + "step": 2282 + }, + { + "epoch": 0.2904210660221346, + "grad_norm": 31.875, + "learning_rate": 9.67359050445104e-07, + "loss": 0.773, + "mean_token_accuracy": 0.7749820947647095, + "num_tokens": 87249402.0, + "step": 2283 + }, + { + "epoch": 0.2905482763007251, + "grad_norm": 36.0, + "learning_rate": 9.677829588808817e-07, + "loss": 0.7428, + "mean_token_accuracy": 0.7815871238708496, + "num_tokens": 87281846.0, + "step": 2284 + }, + { + "epoch": 0.29067548657931563, + "grad_norm": 78.0, + "learning_rate": 9.682068673166596e-07, + "loss": 0.8195, + "mean_token_accuracy": 0.7667784094810486, + "num_tokens": 87315765.0, + "step": 2285 + }, + { + "epoch": 0.2908026968579061, + "grad_norm": 27.875, + "learning_rate": 9.686307757524374e-07, + "loss": 0.7367, + "mean_token_accuracy": 0.7830286622047424, + "num_tokens": 87353668.0, + "step": 2286 + }, + { + "epoch": 0.29092990713649663, + "grad_norm": 62.0, + "learning_rate": 9.690546841882153e-07, + "loss": 0.6573, + "mean_token_accuracy": 0.8074421882629395, + "num_tokens": 87390747.0, + "step": 2287 + }, + { + "epoch": 0.29105711741508716, + "grad_norm": 36.25, + "learning_rate": 9.694785926239931e-07, + "loss": 0.7596, + "mean_token_accuracy": 0.782360315322876, + "num_tokens": 87430686.0, + "step": 2288 + }, + { + "epoch": 0.29118432769367764, + "grad_norm": 136.0, + "learning_rate": 9.69902501059771e-07, + "loss": 0.8246, + "mean_token_accuracy": 0.7668493986129761, + "num_tokens": 87465290.0, + "step": 2289 + }, + { + "epoch": 0.29131153797226816, + "grad_norm": 16.25, + "learning_rate": 9.703264094955488e-07, + "loss": 0.7724, + "mean_token_accuracy": 0.7763473987579346, + "num_tokens": 87504553.0, + "step": 2290 + }, + { + "epoch": 0.2914387482508587, + "grad_norm": 22.125, + "learning_rate": 9.707503179313269e-07, + "loss": 0.7295, + "mean_token_accuracy": 0.7906402349472046, + "num_tokens": 87544896.0, + "step": 2291 + }, + { + "epoch": 0.29156595852944917, + "grad_norm": 68.5, + "learning_rate": 9.711742263671047e-07, + "loss": 0.7392, + "mean_token_accuracy": 0.787523627281189, + "num_tokens": 87582075.0, + "step": 2292 + }, + { + "epoch": 0.2916931688080397, + "grad_norm": 146.0, + "learning_rate": 9.715981348028826e-07, + "loss": 0.7253, + "mean_token_accuracy": 0.7913284301757812, + "num_tokens": 87623117.0, + "step": 2293 + }, + { + "epoch": 0.2918203790866302, + "grad_norm": 178.0, + "learning_rate": 9.720220432386604e-07, + "loss": 0.7589, + "mean_token_accuracy": 0.7827026844024658, + "num_tokens": 87656263.0, + "step": 2294 + }, + { + "epoch": 0.2919475893652207, + "grad_norm": 69.0, + "learning_rate": 9.724459516744383e-07, + "loss": 0.7083, + "mean_token_accuracy": 0.7943915128707886, + "num_tokens": 87695852.0, + "step": 2295 + }, + { + "epoch": 0.2920747996438112, + "grad_norm": 12.0625, + "learning_rate": 9.728698601102161e-07, + "loss": 0.7012, + "mean_token_accuracy": 0.7977681159973145, + "num_tokens": 87734331.0, + "step": 2296 + }, + { + "epoch": 0.29220200992240175, + "grad_norm": 14.375, + "learning_rate": 9.73293768545994e-07, + "loss": 0.7562, + "mean_token_accuracy": 0.7763144969940186, + "num_tokens": 87768447.0, + "step": 2297 + }, + { + "epoch": 0.2923292202009922, + "grad_norm": 35.5, + "learning_rate": 9.737176769817718e-07, + "loss": 0.7745, + "mean_token_accuracy": 0.7808733582496643, + "num_tokens": 87813213.0, + "step": 2298 + }, + { + "epoch": 0.29245643047958275, + "grad_norm": 191.0, + "learning_rate": 9.741415854175499e-07, + "loss": 0.8407, + "mean_token_accuracy": 0.7630209922790527, + "num_tokens": 87851336.0, + "step": 2299 + }, + { + "epoch": 0.2925836407581733, + "grad_norm": 93.0, + "learning_rate": 9.745654938533277e-07, + "loss": 0.7732, + "mean_token_accuracy": 0.7762640714645386, + "num_tokens": 87888474.0, + "step": 2300 + }, + { + "epoch": 0.29271085103676375, + "grad_norm": 266.0, + "learning_rate": 9.749894022891056e-07, + "loss": 0.8158, + "mean_token_accuracy": 0.7705321907997131, + "num_tokens": 87928232.0, + "step": 2301 + }, + { + "epoch": 0.2928380613153543, + "grad_norm": 85.5, + "learning_rate": 9.754133107248834e-07, + "loss": 0.71, + "mean_token_accuracy": 0.7947935461997986, + "num_tokens": 87966982.0, + "step": 2302 + }, + { + "epoch": 0.2929652715939448, + "grad_norm": 21.5, + "learning_rate": 9.758372191606612e-07, + "loss": 0.8118, + "mean_token_accuracy": 0.761320173740387, + "num_tokens": 88003550.0, + "step": 2303 + }, + { + "epoch": 0.2930924818725353, + "grad_norm": 33.5, + "learning_rate": 9.76261127596439e-07, + "loss": 0.7025, + "mean_token_accuracy": 0.7985326051712036, + "num_tokens": 88038038.0, + "step": 2304 + }, + { + "epoch": 0.2932196921511258, + "grad_norm": 36.25, + "learning_rate": 9.76685036032217e-07, + "loss": 0.7969, + "mean_token_accuracy": 0.7684366703033447, + "num_tokens": 88074627.0, + "step": 2305 + }, + { + "epoch": 0.29334690242971634, + "grad_norm": 29.5, + "learning_rate": 9.771089444679948e-07, + "loss": 0.6992, + "mean_token_accuracy": 0.7959414720535278, + "num_tokens": 88116772.0, + "step": 2306 + }, + { + "epoch": 0.2934741127083068, + "grad_norm": 39.25, + "learning_rate": 9.775328529037728e-07, + "loss": 0.7753, + "mean_token_accuracy": 0.7770133018493652, + "num_tokens": 88152888.0, + "step": 2307 + }, + { + "epoch": 0.29360132298689734, + "grad_norm": 16.0, + "learning_rate": 9.779567613395507e-07, + "loss": 0.8036, + "mean_token_accuracy": 0.7717289328575134, + "num_tokens": 88184544.0, + "step": 2308 + }, + { + "epoch": 0.29372853326548787, + "grad_norm": 50.5, + "learning_rate": 9.783806697753285e-07, + "loss": 0.737, + "mean_token_accuracy": 0.789681077003479, + "num_tokens": 88221603.0, + "step": 2309 + }, + { + "epoch": 0.29385574354407834, + "grad_norm": 42.75, + "learning_rate": 9.788045782111064e-07, + "loss": 0.6986, + "mean_token_accuracy": 0.7972343564033508, + "num_tokens": 88265109.0, + "step": 2310 + }, + { + "epoch": 0.29398295382266887, + "grad_norm": 13.8125, + "learning_rate": 9.792284866468842e-07, + "loss": 0.6848, + "mean_token_accuracy": 0.7944858074188232, + "num_tokens": 88302756.0, + "step": 2311 + }, + { + "epoch": 0.2941101641012594, + "grad_norm": 94.5, + "learning_rate": 9.79652395082662e-07, + "loss": 0.7113, + "mean_token_accuracy": 0.7873789072036743, + "num_tokens": 88335605.0, + "step": 2312 + }, + { + "epoch": 0.29423737437984987, + "grad_norm": 41.5, + "learning_rate": 9.8007630351844e-07, + "loss": 0.7652, + "mean_token_accuracy": 0.7806216478347778, + "num_tokens": 88369447.0, + "step": 2313 + }, + { + "epoch": 0.2943645846584404, + "grad_norm": 41.5, + "learning_rate": 9.805002119542178e-07, + "loss": 0.823, + "mean_token_accuracy": 0.7665106058120728, + "num_tokens": 88404442.0, + "step": 2314 + }, + { + "epoch": 0.2944917949370309, + "grad_norm": 30.375, + "learning_rate": 9.809241203899958e-07, + "loss": 0.7425, + "mean_token_accuracy": 0.7860046029090881, + "num_tokens": 88445870.0, + "step": 2315 + }, + { + "epoch": 0.2946190052156214, + "grad_norm": 31.25, + "learning_rate": 9.813480288257737e-07, + "loss": 0.7733, + "mean_token_accuracy": 0.7730071544647217, + "num_tokens": 88480994.0, + "step": 2316 + }, + { + "epoch": 0.2947462154942119, + "grad_norm": 11.875, + "learning_rate": 9.817719372615515e-07, + "loss": 0.7035, + "mean_token_accuracy": 0.7951377034187317, + "num_tokens": 88519666.0, + "step": 2317 + }, + { + "epoch": 0.29487342577280246, + "grad_norm": 50.75, + "learning_rate": 9.821958456973294e-07, + "loss": 0.6931, + "mean_token_accuracy": 0.8010772466659546, + "num_tokens": 88553855.0, + "step": 2318 + }, + { + "epoch": 0.29500063605139293, + "grad_norm": 27.75, + "learning_rate": 9.826197541331072e-07, + "loss": 0.7086, + "mean_token_accuracy": 0.7965170741081238, + "num_tokens": 88589053.0, + "step": 2319 + }, + { + "epoch": 0.29512784632998346, + "grad_norm": 124.5, + "learning_rate": 9.83043662568885e-07, + "loss": 0.7837, + "mean_token_accuracy": 0.7758797407150269, + "num_tokens": 88627067.0, + "step": 2320 + }, + { + "epoch": 0.295255056608574, + "grad_norm": 91.0, + "learning_rate": 9.83467571004663e-07, + "loss": 0.6901, + "mean_token_accuracy": 0.7997311949729919, + "num_tokens": 88668030.0, + "step": 2321 + }, + { + "epoch": 0.29538226688716446, + "grad_norm": 36.5, + "learning_rate": 9.838914794404407e-07, + "loss": 0.697, + "mean_token_accuracy": 0.7974256277084351, + "num_tokens": 88708889.0, + "step": 2322 + }, + { + "epoch": 0.295509477165755, + "grad_norm": 73.0, + "learning_rate": 9.843153878762188e-07, + "loss": 0.7761, + "mean_token_accuracy": 0.7787987589836121, + "num_tokens": 88745689.0, + "step": 2323 + }, + { + "epoch": 0.2956366874443455, + "grad_norm": 64.5, + "learning_rate": 9.847392963119966e-07, + "loss": 0.7182, + "mean_token_accuracy": 0.7915199995040894, + "num_tokens": 88784667.0, + "step": 2324 + }, + { + "epoch": 0.295763897722936, + "grad_norm": 24.875, + "learning_rate": 9.851632047477745e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7886773347854614, + "num_tokens": 88827334.0, + "step": 2325 + }, + { + "epoch": 0.2958911080015265, + "grad_norm": 15.5625, + "learning_rate": 9.855871131835523e-07, + "loss": 0.6593, + "mean_token_accuracy": 0.804751455783844, + "num_tokens": 88863744.0, + "step": 2326 + }, + { + "epoch": 0.29601831828011704, + "grad_norm": 29.25, + "learning_rate": 9.860110216193302e-07, + "loss": 0.6883, + "mean_token_accuracy": 0.8008593320846558, + "num_tokens": 88903693.0, + "step": 2327 + }, + { + "epoch": 0.2961455285587075, + "grad_norm": 68.0, + "learning_rate": 9.86434930055108e-07, + "loss": 0.6996, + "mean_token_accuracy": 0.7954614162445068, + "num_tokens": 88936668.0, + "step": 2328 + }, + { + "epoch": 0.29627273883729804, + "grad_norm": 80.0, + "learning_rate": 9.868588384908859e-07, + "loss": 0.723, + "mean_token_accuracy": 0.7909173369407654, + "num_tokens": 88970920.0, + "step": 2329 + }, + { + "epoch": 0.2963999491158886, + "grad_norm": 132.0, + "learning_rate": 9.872827469266637e-07, + "loss": 0.7069, + "mean_token_accuracy": 0.7923727035522461, + "num_tokens": 89010856.0, + "step": 2330 + }, + { + "epoch": 0.2965271593944791, + "grad_norm": 13.1875, + "learning_rate": 9.877066553624418e-07, + "loss": 0.7155, + "mean_token_accuracy": 0.7921591401100159, + "num_tokens": 89047109.0, + "step": 2331 + }, + { + "epoch": 0.2966543696730696, + "grad_norm": 85.0, + "learning_rate": 9.881305637982196e-07, + "loss": 0.7274, + "mean_token_accuracy": 0.7897125482559204, + "num_tokens": 89087859.0, + "step": 2332 + }, + { + "epoch": 0.2967815799516601, + "grad_norm": 31.0, + "learning_rate": 9.885544722339975e-07, + "loss": 0.7226, + "mean_token_accuracy": 0.787308931350708, + "num_tokens": 89123601.0, + "step": 2333 + }, + { + "epoch": 0.29690879023025063, + "grad_norm": 58.25, + "learning_rate": 9.889783806697753e-07, + "loss": 0.6994, + "mean_token_accuracy": 0.7954375743865967, + "num_tokens": 89160133.0, + "step": 2334 + }, + { + "epoch": 0.2970360005088411, + "grad_norm": 89.5, + "learning_rate": 9.894022891055532e-07, + "loss": 0.6828, + "mean_token_accuracy": 0.8000672459602356, + "num_tokens": 89196731.0, + "step": 2335 + }, + { + "epoch": 0.29716321078743163, + "grad_norm": 76.0, + "learning_rate": 9.89826197541331e-07, + "loss": 0.6838, + "mean_token_accuracy": 0.8027614951133728, + "num_tokens": 89230159.0, + "step": 2336 + }, + { + "epoch": 0.29729042106602216, + "grad_norm": 17.5, + "learning_rate": 9.902501059771089e-07, + "loss": 0.7479, + "mean_token_accuracy": 0.7814657688140869, + "num_tokens": 89264648.0, + "step": 2337 + }, + { + "epoch": 0.29741763134461263, + "grad_norm": 20.5, + "learning_rate": 9.906740144128867e-07, + "loss": 0.7642, + "mean_token_accuracy": 0.7758075594902039, + "num_tokens": 89302505.0, + "step": 2338 + }, + { + "epoch": 0.29754484162320316, + "grad_norm": 25.5, + "learning_rate": 9.910979228486648e-07, + "loss": 0.7056, + "mean_token_accuracy": 0.7951688766479492, + "num_tokens": 89338079.0, + "step": 2339 + }, + { + "epoch": 0.2976720519017937, + "grad_norm": 15.5625, + "learning_rate": 9.915218312844426e-07, + "loss": 0.7734, + "mean_token_accuracy": 0.7824161648750305, + "num_tokens": 89376717.0, + "step": 2340 + }, + { + "epoch": 0.29779926218038416, + "grad_norm": 22.25, + "learning_rate": 9.919457397202205e-07, + "loss": 0.7549, + "mean_token_accuracy": 0.7800085544586182, + "num_tokens": 89416164.0, + "step": 2341 + }, + { + "epoch": 0.2979264724589747, + "grad_norm": 45.25, + "learning_rate": 9.923696481559983e-07, + "loss": 0.7052, + "mean_token_accuracy": 0.7970776557922363, + "num_tokens": 89446491.0, + "step": 2342 + }, + { + "epoch": 0.2980536827375652, + "grad_norm": 24.375, + "learning_rate": 9.927935565917761e-07, + "loss": 0.7649, + "mean_token_accuracy": 0.7770955562591553, + "num_tokens": 89483637.0, + "step": 2343 + }, + { + "epoch": 0.2981808930161557, + "grad_norm": 12.5625, + "learning_rate": 9.93217465027554e-07, + "loss": 0.7356, + "mean_token_accuracy": 0.7886844277381897, + "num_tokens": 89526532.0, + "step": 2344 + }, + { + "epoch": 0.2983081032947462, + "grad_norm": 42.25, + "learning_rate": 9.936413734633318e-07, + "loss": 0.8403, + "mean_token_accuracy": 0.7648181915283203, + "num_tokens": 89558353.0, + "step": 2345 + }, + { + "epoch": 0.29843531357333675, + "grad_norm": 37.25, + "learning_rate": 9.940652818991097e-07, + "loss": 0.6978, + "mean_token_accuracy": 0.7981986999511719, + "num_tokens": 89598727.0, + "step": 2346 + }, + { + "epoch": 0.2985625238519272, + "grad_norm": 53.0, + "learning_rate": 9.944891903348877e-07, + "loss": 0.7843, + "mean_token_accuracy": 0.7760242223739624, + "num_tokens": 89640093.0, + "step": 2347 + }, + { + "epoch": 0.29868973413051775, + "grad_norm": 35.0, + "learning_rate": 9.949130987706656e-07, + "loss": 0.6895, + "mean_token_accuracy": 0.799777626991272, + "num_tokens": 89681401.0, + "step": 2348 + }, + { + "epoch": 0.2988169444091083, + "grad_norm": 49.0, + "learning_rate": 9.953370072064432e-07, + "loss": 0.7158, + "mean_token_accuracy": 0.7914746403694153, + "num_tokens": 89720366.0, + "step": 2349 + }, + { + "epoch": 0.29894415468769875, + "grad_norm": 22.5, + "learning_rate": 9.957609156422213e-07, + "loss": 0.7879, + "mean_token_accuracy": 0.7715432643890381, + "num_tokens": 89762234.0, + "step": 2350 + }, + { + "epoch": 0.2990713649662893, + "grad_norm": 111.5, + "learning_rate": 9.961848240779991e-07, + "loss": 0.7547, + "mean_token_accuracy": 0.7835265398025513, + "num_tokens": 89792208.0, + "step": 2351 + }, + { + "epoch": 0.2991985752448798, + "grad_norm": 40.5, + "learning_rate": 9.96608732513777e-07, + "loss": 0.696, + "mean_token_accuracy": 0.8013895153999329, + "num_tokens": 89833246.0, + "step": 2352 + }, + { + "epoch": 0.2993257855234703, + "grad_norm": 15.1875, + "learning_rate": 9.970326409495548e-07, + "loss": 0.7886, + "mean_token_accuracy": 0.772515058517456, + "num_tokens": 89873271.0, + "step": 2353 + }, + { + "epoch": 0.2994529958020608, + "grad_norm": 19.5, + "learning_rate": 9.974565493853327e-07, + "loss": 0.8039, + "mean_token_accuracy": 0.7641130685806274, + "num_tokens": 89913069.0, + "step": 2354 + }, + { + "epoch": 0.29958020608065133, + "grad_norm": 42.0, + "learning_rate": 9.978804578211107e-07, + "loss": 0.7634, + "mean_token_accuracy": 0.7816221714019775, + "num_tokens": 89948915.0, + "step": 2355 + }, + { + "epoch": 0.2997074163592418, + "grad_norm": 141.0, + "learning_rate": 9.983043662568886e-07, + "loss": 0.7904, + "mean_token_accuracy": 0.7731141448020935, + "num_tokens": 89982123.0, + "step": 2356 + }, + { + "epoch": 0.29983462663783234, + "grad_norm": 27.875, + "learning_rate": 9.987282746926662e-07, + "loss": 0.6971, + "mean_token_accuracy": 0.7998735904693604, + "num_tokens": 90018259.0, + "step": 2357 + }, + { + "epoch": 0.29996183691642286, + "grad_norm": 35.5, + "learning_rate": 9.991521831284443e-07, + "loss": 0.7376, + "mean_token_accuracy": 0.7864787578582764, + "num_tokens": 90059041.0, + "step": 2358 + }, + { + "epoch": 0.30008904719501334, + "grad_norm": 49.5, + "learning_rate": 9.995760915642221e-07, + "loss": 0.6858, + "mean_token_accuracy": 0.7933838367462158, + "num_tokens": 90089463.0, + "step": 2359 + }, + { + "epoch": 0.30021625747360386, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7960535287857056, + "num_tokens": 90125691.0, + "step": 2360 + }, + { + "epoch": 0.3003434677521944, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7849149703979492, + "num_tokens": 90161778.0, + "step": 2361 + }, + { + "epoch": 0.30047067803078487, + "grad_norm": 63.75, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7720001339912415, + "num_tokens": 90202243.0, + "step": 2362 + }, + { + "epoch": 0.3005978883093754, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7839506268501282, + "num_tokens": 90248924.0, + "step": 2363 + }, + { + "epoch": 0.3007250985879659, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7563118934631348, + "num_tokens": 90281988.0, + "step": 2364 + }, + { + "epoch": 0.3008523088665564, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7773940563201904, + "num_tokens": 90317050.0, + "step": 2365 + }, + { + "epoch": 0.3009795191451469, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.8003835082054138, + "num_tokens": 90358064.0, + "step": 2366 + }, + { + "epoch": 0.30110672942373745, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7738383412361145, + "num_tokens": 90400692.0, + "step": 2367 + }, + { + "epoch": 0.3012339397023279, + "grad_norm": 82.0, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7785881161689758, + "num_tokens": 90432207.0, + "step": 2368 + }, + { + "epoch": 0.30136114998091845, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7776336669921875, + "num_tokens": 90466081.0, + "step": 2369 + }, + { + "epoch": 0.301488360259509, + "grad_norm": 100.0, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.79226154088974, + "num_tokens": 90501963.0, + "step": 2370 + }, + { + "epoch": 0.30161557053809945, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7984813451766968, + "num_tokens": 90540651.0, + "step": 2371 + }, + { + "epoch": 0.30174278081669, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7785680294036865, + "num_tokens": 90585105.0, + "step": 2372 + }, + { + "epoch": 0.3018699910952805, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7805287837982178, + "num_tokens": 90626773.0, + "step": 2373 + }, + { + "epoch": 0.301997201373871, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7997884750366211, + "num_tokens": 90665974.0, + "step": 2374 + }, + { + "epoch": 0.3021244116524615, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7821242809295654, + "num_tokens": 90706801.0, + "step": 2375 + }, + { + "epoch": 0.30225162193105204, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7824989557266235, + "num_tokens": 90743635.0, + "step": 2376 + }, + { + "epoch": 0.3023788322096425, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7901085615158081, + "num_tokens": 90782797.0, + "step": 2377 + }, + { + "epoch": 0.30250604248823304, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7758347988128662, + "num_tokens": 90822661.0, + "step": 2378 + }, + { + "epoch": 0.30263325276682357, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7652877569198608, + "num_tokens": 90857804.0, + "step": 2379 + }, + { + "epoch": 0.3027604630454141, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7811229825019836, + "num_tokens": 90895778.0, + "step": 2380 + }, + { + "epoch": 0.30288767332400457, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.791886568069458, + "num_tokens": 90934463.0, + "step": 2381 + }, + { + "epoch": 0.3030148836025951, + "grad_norm": 90.0, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7853102087974548, + "num_tokens": 90967888.0, + "step": 2382 + }, + { + "epoch": 0.3031420938811856, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7927504777908325, + "num_tokens": 91005200.0, + "step": 2383 + }, + { + "epoch": 0.3032693041597761, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7728800773620605, + "num_tokens": 91037840.0, + "step": 2384 + }, + { + "epoch": 0.3033965144383666, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7700024843215942, + "num_tokens": 91073336.0, + "step": 2385 + }, + { + "epoch": 0.30352372471695716, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7838106751441956, + "num_tokens": 91107001.0, + "step": 2386 + }, + { + "epoch": 0.30365093499554763, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7921839952468872, + "num_tokens": 91143937.0, + "step": 2387 + }, + { + "epoch": 0.30377814527413816, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7723255157470703, + "num_tokens": 91185391.0, + "step": 2388 + }, + { + "epoch": 0.3039053555527287, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7852879762649536, + "num_tokens": 91227653.0, + "step": 2389 + }, + { + "epoch": 0.30403256583131916, + "grad_norm": 104.0, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.8029577732086182, + "num_tokens": 91263610.0, + "step": 2390 + }, + { + "epoch": 0.3041597761099097, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7785788774490356, + "num_tokens": 91296880.0, + "step": 2391 + }, + { + "epoch": 0.3042869863885002, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7685626149177551, + "num_tokens": 91336242.0, + "step": 2392 + }, + { + "epoch": 0.3044141966670907, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7945905327796936, + "num_tokens": 91376527.0, + "step": 2393 + }, + { + "epoch": 0.3045414069456812, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7679766416549683, + "num_tokens": 91408415.0, + "step": 2394 + }, + { + "epoch": 0.30466861722427174, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7841447591781616, + "num_tokens": 91449189.0, + "step": 2395 + }, + { + "epoch": 0.3047958275028622, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7797252535820007, + "num_tokens": 91488599.0, + "step": 2396 + }, + { + "epoch": 0.30492303778145274, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7920503616333008, + "num_tokens": 91525138.0, + "step": 2397 + }, + { + "epoch": 0.30505024806004327, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7873772978782654, + "num_tokens": 91567444.0, + "step": 2398 + }, + { + "epoch": 0.30517745833863374, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7792950868606567, + "num_tokens": 91602926.0, + "step": 2399 + }, + { + "epoch": 0.3053046686172243, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7921354174613953, + "num_tokens": 91635901.0, + "step": 2400 + }, + { + "epoch": 0.3054318788958148, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7875808477401733, + "num_tokens": 91679542.0, + "step": 2401 + }, + { + "epoch": 0.3055590891744053, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7834701538085938, + "num_tokens": 91723022.0, + "step": 2402 + }, + { + "epoch": 0.3056862994529958, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7800699472427368, + "num_tokens": 91760664.0, + "step": 2403 + }, + { + "epoch": 0.30581350973158633, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7838019728660583, + "num_tokens": 91805570.0, + "step": 2404 + }, + { + "epoch": 0.3059407200101768, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.778502345085144, + "num_tokens": 91843342.0, + "step": 2405 + }, + { + "epoch": 0.30606793028876733, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7744301557540894, + "num_tokens": 91883872.0, + "step": 2406 + }, + { + "epoch": 0.30619514056735786, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7846639156341553, + "num_tokens": 91923174.0, + "step": 2407 + }, + { + "epoch": 0.30632235084594833, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7880587577819824, + "num_tokens": 91963462.0, + "step": 2408 + }, + { + "epoch": 0.30644956112453886, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7892507314682007, + "num_tokens": 92000907.0, + "step": 2409 + }, + { + "epoch": 0.3065767714031294, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7871255278587341, + "num_tokens": 92036110.0, + "step": 2410 + }, + { + "epoch": 0.30670398168171986, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7847403287887573, + "num_tokens": 92080647.0, + "step": 2411 + }, + { + "epoch": 0.3068311919603104, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7852734327316284, + "num_tokens": 92115407.0, + "step": 2412 + }, + { + "epoch": 0.3069584022389009, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.784431517124176, + "num_tokens": 92149660.0, + "step": 2413 + }, + { + "epoch": 0.3070856125174914, + "grad_norm": 88.0, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7941197156906128, + "num_tokens": 92191581.0, + "step": 2414 + }, + { + "epoch": 0.3072128227960819, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7805537581443787, + "num_tokens": 92230435.0, + "step": 2415 + }, + { + "epoch": 0.30734003307467245, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7709801197052002, + "num_tokens": 92269545.0, + "step": 2416 + }, + { + "epoch": 0.3074672433532629, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7935833930969238, + "num_tokens": 92307447.0, + "step": 2417 + }, + { + "epoch": 0.30759445363185345, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7828110456466675, + "num_tokens": 92349441.0, + "step": 2418 + }, + { + "epoch": 0.307721663910444, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7855194211006165, + "num_tokens": 92380405.0, + "step": 2419 + }, + { + "epoch": 0.30784887418903445, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7843478918075562, + "num_tokens": 92419596.0, + "step": 2420 + }, + { + "epoch": 0.307976084467625, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7741609811782837, + "num_tokens": 92454590.0, + "step": 2421 + }, + { + "epoch": 0.3081032947462155, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7750529050827026, + "num_tokens": 92491640.0, + "step": 2422 + }, + { + "epoch": 0.308230505024806, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7698401808738708, + "num_tokens": 92528682.0, + "step": 2423 + }, + { + "epoch": 0.3083577153033965, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7910218238830566, + "num_tokens": 92566368.0, + "step": 2424 + }, + { + "epoch": 0.30848492558198704, + "grad_norm": 97.5, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7620148658752441, + "num_tokens": 92601380.0, + "step": 2425 + }, + { + "epoch": 0.3086121358605775, + "grad_norm": 83.0, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.805046796798706, + "num_tokens": 92639724.0, + "step": 2426 + }, + { + "epoch": 0.30873934613916804, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7921848893165588, + "num_tokens": 92677450.0, + "step": 2427 + }, + { + "epoch": 0.30886655641775856, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.792637050151825, + "num_tokens": 92714681.0, + "step": 2428 + }, + { + "epoch": 0.3089937666963491, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7778836488723755, + "num_tokens": 92749925.0, + "step": 2429 + }, + { + "epoch": 0.30912097697493957, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7887803316116333, + "num_tokens": 92791720.0, + "step": 2430 + }, + { + "epoch": 0.3092481872535301, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.775672972202301, + "num_tokens": 92826179.0, + "step": 2431 + }, + { + "epoch": 0.3093753975321206, + "grad_norm": 77.0, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7869625687599182, + "num_tokens": 92872751.0, + "step": 2432 + }, + { + "epoch": 0.3095026078107111, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7754071950912476, + "num_tokens": 92907470.0, + "step": 2433 + }, + { + "epoch": 0.3096298180893016, + "grad_norm": 104.5, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7872029542922974, + "num_tokens": 92943321.0, + "step": 2434 + }, + { + "epoch": 0.30975702836789215, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7865146398544312, + "num_tokens": 92982136.0, + "step": 2435 + }, + { + "epoch": 0.3098842386464826, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7677542567253113, + "num_tokens": 93015974.0, + "step": 2436 + }, + { + "epoch": 0.31001144892507315, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.787556529045105, + "num_tokens": 93054010.0, + "step": 2437 + }, + { + "epoch": 0.3101386592036637, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.793491780757904, + "num_tokens": 93090226.0, + "step": 2438 + }, + { + "epoch": 0.31026586948225415, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7840828895568848, + "num_tokens": 93130242.0, + "step": 2439 + }, + { + "epoch": 0.3103930797608447, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7826462984085083, + "num_tokens": 93170990.0, + "step": 2440 + }, + { + "epoch": 0.3105202900394352, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7750554084777832, + "num_tokens": 93205478.0, + "step": 2441 + }, + { + "epoch": 0.3106475003180257, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.776327908039093, + "num_tokens": 93240952.0, + "step": 2442 + }, + { + "epoch": 0.3107747105966162, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7724524736404419, + "num_tokens": 93278882.0, + "step": 2443 + }, + { + "epoch": 0.31090192087520674, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7809798717498779, + "num_tokens": 93324573.0, + "step": 2444 + }, + { + "epoch": 0.3110291311537972, + "grad_norm": 112.5, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7988590002059937, + "num_tokens": 93361049.0, + "step": 2445 + }, + { + "epoch": 0.31115634143238774, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7843207120895386, + "num_tokens": 93401748.0, + "step": 2446 + }, + { + "epoch": 0.31128355171097827, + "grad_norm": 133.0, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7887197136878967, + "num_tokens": 93434857.0, + "step": 2447 + }, + { + "epoch": 0.31141076198956874, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7869527339935303, + "num_tokens": 93474681.0, + "step": 2448 + }, + { + "epoch": 0.31153797226815927, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7779804468154907, + "num_tokens": 93510627.0, + "step": 2449 + }, + { + "epoch": 0.3116651825467498, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.776228666305542, + "num_tokens": 93551598.0, + "step": 2450 + }, + { + "epoch": 0.31179239282534027, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7773244976997375, + "num_tokens": 93586622.0, + "step": 2451 + }, + { + "epoch": 0.3119196031039308, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7723132371902466, + "num_tokens": 93626496.0, + "step": 2452 + }, + { + "epoch": 0.3120468133825213, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7871233820915222, + "num_tokens": 93665641.0, + "step": 2453 + }, + { + "epoch": 0.3121740236611118, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7897137403488159, + "num_tokens": 93702135.0, + "step": 2454 + }, + { + "epoch": 0.3123012339397023, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7902519106864929, + "num_tokens": 93738226.0, + "step": 2455 + }, + { + "epoch": 0.31242844421829286, + "grad_norm": 177.0, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7866291999816895, + "num_tokens": 93780273.0, + "step": 2456 + }, + { + "epoch": 0.31255565449688333, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.771374523639679, + "num_tokens": 93821977.0, + "step": 2457 + }, + { + "epoch": 0.31268286477547386, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7894617319107056, + "num_tokens": 93856084.0, + "step": 2458 + }, + { + "epoch": 0.3128100750540644, + "grad_norm": 144.0, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7862323522567749, + "num_tokens": 93893491.0, + "step": 2459 + }, + { + "epoch": 0.31293728533265486, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7732688188552856, + "num_tokens": 93940240.0, + "step": 2460 + }, + { + "epoch": 0.3130644956112454, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7726666927337646, + "num_tokens": 93974504.0, + "step": 2461 + }, + { + "epoch": 0.3131917058898359, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7851340770721436, + "num_tokens": 94009208.0, + "step": 2462 + }, + { + "epoch": 0.3133189161684264, + "grad_norm": 64.5, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7862673401832581, + "num_tokens": 94049249.0, + "step": 2463 + }, + { + "epoch": 0.3134461264470169, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7844969630241394, + "num_tokens": 94085553.0, + "step": 2464 + }, + { + "epoch": 0.31357333672560744, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7659217119216919, + "num_tokens": 94127531.0, + "step": 2465 + }, + { + "epoch": 0.3137005470041979, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7944545149803162, + "num_tokens": 94171191.0, + "step": 2466 + }, + { + "epoch": 0.31382775728278844, + "grad_norm": 103.0, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7879348993301392, + "num_tokens": 94212090.0, + "step": 2467 + }, + { + "epoch": 0.313954967561379, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8051605820655823, + "num_tokens": 94246835.0, + "step": 2468 + }, + { + "epoch": 0.31408217783996945, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7779949903488159, + "num_tokens": 94284581.0, + "step": 2469 + }, + { + "epoch": 0.31420938811856, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7976375222206116, + "num_tokens": 94323050.0, + "step": 2470 + }, + { + "epoch": 0.3143365983971505, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7800304293632507, + "num_tokens": 94364702.0, + "step": 2471 + }, + { + "epoch": 0.314463808675741, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7954627871513367, + "num_tokens": 94404735.0, + "step": 2472 + }, + { + "epoch": 0.3145910189543315, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7698543071746826, + "num_tokens": 94444683.0, + "step": 2473 + }, + { + "epoch": 0.31471822923292203, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7920549511909485, + "num_tokens": 94484392.0, + "step": 2474 + }, + { + "epoch": 0.3148454395115125, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7959796786308289, + "num_tokens": 94524860.0, + "step": 2475 + }, + { + "epoch": 0.31497264979010303, + "grad_norm": 113.5, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.781341552734375, + "num_tokens": 94565076.0, + "step": 2476 + }, + { + "epoch": 0.31509986006869356, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7620185017585754, + "num_tokens": 94606191.0, + "step": 2477 + }, + { + "epoch": 0.31522707034728403, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7805225849151611, + "num_tokens": 94648021.0, + "step": 2478 + }, + { + "epoch": 0.31535428062587456, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7921677231788635, + "num_tokens": 94686305.0, + "step": 2479 + }, + { + "epoch": 0.3154814909044651, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7813258171081543, + "num_tokens": 94728606.0, + "step": 2480 + }, + { + "epoch": 0.3156087011830556, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7896691560745239, + "num_tokens": 94770162.0, + "step": 2481 + }, + { + "epoch": 0.3157359114616461, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7939804792404175, + "num_tokens": 94807735.0, + "step": 2482 + }, + { + "epoch": 0.3158631217402366, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7816259860992432, + "num_tokens": 94840366.0, + "step": 2483 + }, + { + "epoch": 0.31599033201882715, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7834718227386475, + "num_tokens": 94874293.0, + "step": 2484 + }, + { + "epoch": 0.3161175422974176, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7872614860534668, + "num_tokens": 94909347.0, + "step": 2485 + }, + { + "epoch": 0.31624475257600815, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7997651100158691, + "num_tokens": 94955120.0, + "step": 2486 + }, + { + "epoch": 0.3163719628545987, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7990987300872803, + "num_tokens": 94992319.0, + "step": 2487 + }, + { + "epoch": 0.31649917313318915, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7905269861221313, + "num_tokens": 95023464.0, + "step": 2488 + }, + { + "epoch": 0.3166263834117797, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7944093942642212, + "num_tokens": 95055963.0, + "step": 2489 + }, + { + "epoch": 0.3167535936903702, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7677261829376221, + "num_tokens": 95089628.0, + "step": 2490 + }, + { + "epoch": 0.3168808039689607, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.8015745878219604, + "num_tokens": 95128557.0, + "step": 2491 + }, + { + "epoch": 0.3170080142475512, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7737840414047241, + "num_tokens": 95168324.0, + "step": 2492 + }, + { + "epoch": 0.31713522452614173, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7687157988548279, + "num_tokens": 95208139.0, + "step": 2493 + }, + { + "epoch": 0.3172624348047322, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.778304934501648, + "num_tokens": 95247337.0, + "step": 2494 + }, + { + "epoch": 0.31738964508332274, + "grad_norm": 84.5, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7792425155639648, + "num_tokens": 95289208.0, + "step": 2495 + }, + { + "epoch": 0.31751685536191326, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7841973304748535, + "num_tokens": 95329717.0, + "step": 2496 + }, + { + "epoch": 0.31764406564050374, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.762955904006958, + "num_tokens": 95371052.0, + "step": 2497 + }, + { + "epoch": 0.31777127591909426, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.80936598777771, + "num_tokens": 95412486.0, + "step": 2498 + }, + { + "epoch": 0.3178984861976848, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8042200803756714, + "num_tokens": 95446167.0, + "step": 2499 + }, + { + "epoch": 0.31802569647627527, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7804837226867676, + "num_tokens": 95485763.0, + "step": 2500 + }, + { + "epoch": 0.3181529067548658, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7759619951248169, + "num_tokens": 95526407.0, + "step": 2501 + }, + { + "epoch": 0.3182801170334563, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7515873908996582, + "num_tokens": 95567479.0, + "step": 2502 + }, + { + "epoch": 0.3184073273120468, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7928767800331116, + "num_tokens": 95606588.0, + "step": 2503 + }, + { + "epoch": 0.3185345375906373, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7921303510665894, + "num_tokens": 95653293.0, + "step": 2504 + }, + { + "epoch": 0.31866174786922785, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7724076509475708, + "num_tokens": 95692424.0, + "step": 2505 + }, + { + "epoch": 0.3187889581478183, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7962414026260376, + "num_tokens": 95733679.0, + "step": 2506 + }, + { + "epoch": 0.31891616842640885, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7819555401802063, + "num_tokens": 95768211.0, + "step": 2507 + }, + { + "epoch": 0.3190433787049994, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7799046039581299, + "num_tokens": 95805974.0, + "step": 2508 + }, + { + "epoch": 0.31917058898358985, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7792423963546753, + "num_tokens": 95845463.0, + "step": 2509 + }, + { + "epoch": 0.3192977992621804, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7826534509658813, + "num_tokens": 95888148.0, + "step": 2510 + }, + { + "epoch": 0.3194250095407709, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.80824214220047, + "num_tokens": 95925432.0, + "step": 2511 + }, + { + "epoch": 0.3195522198193614, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7950059771537781, + "num_tokens": 95970695.0, + "step": 2512 + }, + { + "epoch": 0.3196794300979519, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7842854261398315, + "num_tokens": 96006093.0, + "step": 2513 + }, + { + "epoch": 0.31980664037654244, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7823307514190674, + "num_tokens": 96045945.0, + "step": 2514 + }, + { + "epoch": 0.3199338506551329, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.795838475227356, + "num_tokens": 96076748.0, + "step": 2515 + }, + { + "epoch": 0.32006106093372344, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7875619530677795, + "num_tokens": 96116964.0, + "step": 2516 + }, + { + "epoch": 0.32018827121231397, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7857301235198975, + "num_tokens": 96152073.0, + "step": 2517 + }, + { + "epoch": 0.32031548149090444, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7859379053115845, + "num_tokens": 96191608.0, + "step": 2518 + }, + { + "epoch": 0.32044269176949497, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.8019185066223145, + "num_tokens": 96228124.0, + "step": 2519 + }, + { + "epoch": 0.3205699020480855, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7964190244674683, + "num_tokens": 96265870.0, + "step": 2520 + }, + { + "epoch": 0.32069711232667597, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.799980103969574, + "num_tokens": 96309623.0, + "step": 2521 + }, + { + "epoch": 0.3208243226052665, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7749253511428833, + "num_tokens": 96344036.0, + "step": 2522 + }, + { + "epoch": 0.320951532883857, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.800010085105896, + "num_tokens": 96377627.0, + "step": 2523 + }, + { + "epoch": 0.3210787431624475, + "grad_norm": 88.5, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7830618619918823, + "num_tokens": 96413568.0, + "step": 2524 + }, + { + "epoch": 0.32120595344103803, + "grad_norm": 88.5, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7853963375091553, + "num_tokens": 96454682.0, + "step": 2525 + }, + { + "epoch": 0.32133316371962856, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7762909531593323, + "num_tokens": 96490456.0, + "step": 2526 + }, + { + "epoch": 0.32146037399821903, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7781918048858643, + "num_tokens": 96526891.0, + "step": 2527 + }, + { + "epoch": 0.32158758427680956, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7857190370559692, + "num_tokens": 96575226.0, + "step": 2528 + }, + { + "epoch": 0.3217147945554001, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7753236293792725, + "num_tokens": 96606383.0, + "step": 2529 + }, + { + "epoch": 0.3218420048339906, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7860186696052551, + "num_tokens": 96639826.0, + "step": 2530 + }, + { + "epoch": 0.3219692151125811, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7933581471443176, + "num_tokens": 96679151.0, + "step": 2531 + }, + { + "epoch": 0.3220964253911716, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.796226978302002, + "num_tokens": 96719365.0, + "step": 2532 + }, + { + "epoch": 0.32222363566976214, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.781785249710083, + "num_tokens": 96758318.0, + "step": 2533 + }, + { + "epoch": 0.3223508459483526, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7749233245849609, + "num_tokens": 96791159.0, + "step": 2534 + }, + { + "epoch": 0.32247805622694314, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7870782613754272, + "num_tokens": 96825352.0, + "step": 2535 + }, + { + "epoch": 0.32260526650553367, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7835841178894043, + "num_tokens": 96859700.0, + "step": 2536 + }, + { + "epoch": 0.32273247678412414, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7806534767150879, + "num_tokens": 96897077.0, + "step": 2537 + }, + { + "epoch": 0.3228596870627147, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7819434404373169, + "num_tokens": 96936936.0, + "step": 2538 + }, + { + "epoch": 0.3229868973413052, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7923187613487244, + "num_tokens": 96975702.0, + "step": 2539 + }, + { + "epoch": 0.3231141076198957, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7851859331130981, + "num_tokens": 97010585.0, + "step": 2540 + }, + { + "epoch": 0.3232413178984862, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8179594874382019, + "num_tokens": 97047210.0, + "step": 2541 + }, + { + "epoch": 0.32336852817707673, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.785673975944519, + "num_tokens": 97084835.0, + "step": 2542 + }, + { + "epoch": 0.3234957384556672, + "grad_norm": 107.0, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7851430177688599, + "num_tokens": 97125829.0, + "step": 2543 + }, + { + "epoch": 0.32362294873425773, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7793512344360352, + "num_tokens": 97162874.0, + "step": 2544 + }, + { + "epoch": 0.32375015901284826, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7828466892242432, + "num_tokens": 97197559.0, + "step": 2545 + }, + { + "epoch": 0.32387736929143873, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7704057693481445, + "num_tokens": 97227028.0, + "step": 2546 + }, + { + "epoch": 0.32400457957002926, + "grad_norm": 92.5, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7839877605438232, + "num_tokens": 97270209.0, + "step": 2547 + }, + { + "epoch": 0.3241317898486198, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7907853126525879, + "num_tokens": 97309176.0, + "step": 2548 + }, + { + "epoch": 0.32425900012721026, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7613246440887451, + "num_tokens": 97340464.0, + "step": 2549 + }, + { + "epoch": 0.3243862104058008, + "grad_norm": 170.0, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7856622934341431, + "num_tokens": 97378435.0, + "step": 2550 + }, + { + "epoch": 0.3245134206843913, + "grad_norm": 199.0, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7792339324951172, + "num_tokens": 97417886.0, + "step": 2551 + }, + { + "epoch": 0.3246406309629818, + "grad_norm": 162.0, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7845591306686401, + "num_tokens": 97450552.0, + "step": 2552 + }, + { + "epoch": 0.3247678412415723, + "grad_norm": 150.0, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7925547957420349, + "num_tokens": 97488586.0, + "step": 2553 + }, + { + "epoch": 0.32489505152016285, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7838128209114075, + "num_tokens": 97528893.0, + "step": 2554 + }, + { + "epoch": 0.3250222617987533, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.757964015007019, + "num_tokens": 97568654.0, + "step": 2555 + }, + { + "epoch": 0.32514947207734385, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7877950072288513, + "num_tokens": 97607068.0, + "step": 2556 + }, + { + "epoch": 0.3252766823559344, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8096960186958313, + "num_tokens": 97647505.0, + "step": 2557 + }, + { + "epoch": 0.32540389263452485, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7829395532608032, + "num_tokens": 97690505.0, + "step": 2558 + }, + { + "epoch": 0.3255311029131154, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.773518443107605, + "num_tokens": 97723955.0, + "step": 2559 + }, + { + "epoch": 0.3256583131917059, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7824210524559021, + "num_tokens": 97763822.0, + "step": 2560 + }, + { + "epoch": 0.3257855234702964, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7813619375228882, + "num_tokens": 97803827.0, + "step": 2561 + }, + { + "epoch": 0.3259127337488869, + "grad_norm": 115.5, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7821962237358093, + "num_tokens": 97841056.0, + "step": 2562 + }, + { + "epoch": 0.32603994402747744, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8044935464859009, + "num_tokens": 97881714.0, + "step": 2563 + }, + { + "epoch": 0.3261671543060679, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7835447192192078, + "num_tokens": 97925797.0, + "step": 2564 + }, + { + "epoch": 0.32629436458465844, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7938750982284546, + "num_tokens": 97963638.0, + "step": 2565 + }, + { + "epoch": 0.32642157486324896, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7777008414268494, + "num_tokens": 98002043.0, + "step": 2566 + }, + { + "epoch": 0.32654878514183944, + "grad_norm": 126.0, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7810108661651611, + "num_tokens": 98041031.0, + "step": 2567 + }, + { + "epoch": 0.32667599542042997, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.792230486869812, + "num_tokens": 98075510.0, + "step": 2568 + }, + { + "epoch": 0.3268032056990205, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.787447988986969, + "num_tokens": 98114162.0, + "step": 2569 + }, + { + "epoch": 0.32693041597761097, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7797673344612122, + "num_tokens": 98148722.0, + "step": 2570 + }, + { + "epoch": 0.3270576262562015, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7646000385284424, + "num_tokens": 98188255.0, + "step": 2571 + }, + { + "epoch": 0.327184836534792, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7915995121002197, + "num_tokens": 98228597.0, + "step": 2572 + }, + { + "epoch": 0.3273120468133825, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7762559652328491, + "num_tokens": 98273405.0, + "step": 2573 + }, + { + "epoch": 0.327439257091973, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7747604846954346, + "num_tokens": 98315091.0, + "step": 2574 + }, + { + "epoch": 0.32756646737056355, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7889907360076904, + "num_tokens": 98356967.0, + "step": 2575 + }, + { + "epoch": 0.327693677649154, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.772652268409729, + "num_tokens": 98394455.0, + "step": 2576 + }, + { + "epoch": 0.32782088792774455, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.796841025352478, + "num_tokens": 98430455.0, + "step": 2577 + }, + { + "epoch": 0.3279480982063351, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7749502658843994, + "num_tokens": 98472218.0, + "step": 2578 + }, + { + "epoch": 0.3280753084849256, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7906756401062012, + "num_tokens": 98509540.0, + "step": 2579 + }, + { + "epoch": 0.3282025187635161, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7897688150405884, + "num_tokens": 98544761.0, + "step": 2580 + }, + { + "epoch": 0.3283297290421066, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.8007001280784607, + "num_tokens": 98580522.0, + "step": 2581 + }, + { + "epoch": 0.32845693932069714, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7918884754180908, + "num_tokens": 98612988.0, + "step": 2582 + }, + { + "epoch": 0.3285841495992876, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7736616730690002, + "num_tokens": 98654213.0, + "step": 2583 + }, + { + "epoch": 0.32871135987787814, + "grad_norm": 100.0, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7916757464408875, + "num_tokens": 98697168.0, + "step": 2584 + }, + { + "epoch": 0.32883857015646867, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7899760603904724, + "num_tokens": 98735664.0, + "step": 2585 + }, + { + "epoch": 0.32896578043505914, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7804534435272217, + "num_tokens": 98777257.0, + "step": 2586 + }, + { + "epoch": 0.32909299071364967, + "grad_norm": 132.0, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7848471403121948, + "num_tokens": 98812443.0, + "step": 2587 + }, + { + "epoch": 0.3292202009922402, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7910153865814209, + "num_tokens": 98848231.0, + "step": 2588 + }, + { + "epoch": 0.32934741127083067, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7891762256622314, + "num_tokens": 98886487.0, + "step": 2589 + }, + { + "epoch": 0.3294746215494212, + "grad_norm": 177.0, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7737815380096436, + "num_tokens": 98924750.0, + "step": 2590 + }, + { + "epoch": 0.3296018318280117, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7607886791229248, + "num_tokens": 98964151.0, + "step": 2591 + }, + { + "epoch": 0.3297290421066022, + "grad_norm": 134.0, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7986541986465454, + "num_tokens": 99000277.0, + "step": 2592 + }, + { + "epoch": 0.3298562523851927, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7786219120025635, + "num_tokens": 99042335.0, + "step": 2593 + }, + { + "epoch": 0.32998346266378326, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7843077182769775, + "num_tokens": 99082964.0, + "step": 2594 + }, + { + "epoch": 0.33011067294237373, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8152175545692444, + "num_tokens": 99123440.0, + "step": 2595 + }, + { + "epoch": 0.33023788322096426, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7863132953643799, + "num_tokens": 99167047.0, + "step": 2596 + }, + { + "epoch": 0.3303650934995548, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7826721668243408, + "num_tokens": 99202594.0, + "step": 2597 + }, + { + "epoch": 0.33049230377814526, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7959929704666138, + "num_tokens": 99244732.0, + "step": 2598 + }, + { + "epoch": 0.3306195140567358, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.8035435676574707, + "num_tokens": 99287161.0, + "step": 2599 + }, + { + "epoch": 0.3307467243353263, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7768347263336182, + "num_tokens": 99322650.0, + "step": 2600 + }, + { + "epoch": 0.3308739346139168, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7872428297996521, + "num_tokens": 99358529.0, + "step": 2601 + }, + { + "epoch": 0.3310011448925073, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7973504066467285, + "num_tokens": 99394010.0, + "step": 2602 + }, + { + "epoch": 0.33112835517109784, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.787940263748169, + "num_tokens": 99434764.0, + "step": 2603 + }, + { + "epoch": 0.3312555654496883, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.791102945804596, + "num_tokens": 99481322.0, + "step": 2604 + }, + { + "epoch": 0.33138277572827884, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7739830017089844, + "num_tokens": 99517719.0, + "step": 2605 + }, + { + "epoch": 0.3315099860068694, + "grad_norm": 145.0, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7892751097679138, + "num_tokens": 99553662.0, + "step": 2606 + }, + { + "epoch": 0.33163719628545985, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7857910990715027, + "num_tokens": 99589195.0, + "step": 2607 + }, + { + "epoch": 0.3317644065640504, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7867770195007324, + "num_tokens": 99626092.0, + "step": 2608 + }, + { + "epoch": 0.3318916168426409, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7603384852409363, + "num_tokens": 99667087.0, + "step": 2609 + }, + { + "epoch": 0.3320188271212314, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7797625660896301, + "num_tokens": 99705929.0, + "step": 2610 + }, + { + "epoch": 0.3321460373998219, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7847287058830261, + "num_tokens": 99742044.0, + "step": 2611 + }, + { + "epoch": 0.33227324767841243, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7813915610313416, + "num_tokens": 99772621.0, + "step": 2612 + }, + { + "epoch": 0.3324004579570029, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7948242425918579, + "num_tokens": 99814405.0, + "step": 2613 + }, + { + "epoch": 0.33252766823559343, + "grad_norm": 109.0, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.8031137585639954, + "num_tokens": 99852197.0, + "step": 2614 + }, + { + "epoch": 0.33265487851418396, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7866196036338806, + "num_tokens": 99894254.0, + "step": 2615 + }, + { + "epoch": 0.33278208879277443, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7813464403152466, + "num_tokens": 99936230.0, + "step": 2616 + }, + { + "epoch": 0.33290929907136496, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.78331458568573, + "num_tokens": 99974964.0, + "step": 2617 + }, + { + "epoch": 0.3330365093499555, + "grad_norm": 136.0, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7757762670516968, + "num_tokens": 100018750.0, + "step": 2618 + }, + { + "epoch": 0.33316371962854596, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7813701629638672, + "num_tokens": 100054064.0, + "step": 2619 + }, + { + "epoch": 0.3332909299071365, + "grad_norm": 116.0, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7873785495758057, + "num_tokens": 100092549.0, + "step": 2620 + }, + { + "epoch": 0.333418140185727, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7559027671813965, + "num_tokens": 100123973.0, + "step": 2621 + }, + { + "epoch": 0.3335453504643175, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7840726375579834, + "num_tokens": 100162142.0, + "step": 2622 + }, + { + "epoch": 0.333672560742908, + "grad_norm": 159.0, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7832437753677368, + "num_tokens": 100199352.0, + "step": 2623 + }, + { + "epoch": 0.33379977102149855, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7702255845069885, + "num_tokens": 100237446.0, + "step": 2624 + }, + { + "epoch": 0.333926981300089, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7930421829223633, + "num_tokens": 100272986.0, + "step": 2625 + }, + { + "epoch": 0.33405419157867955, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7733174562454224, + "num_tokens": 100304998.0, + "step": 2626 + }, + { + "epoch": 0.3341814018572701, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.778400182723999, + "num_tokens": 100338473.0, + "step": 2627 + }, + { + "epoch": 0.33430861213586055, + "grad_norm": 117.0, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7668219804763794, + "num_tokens": 100372644.0, + "step": 2628 + }, + { + "epoch": 0.3344358224144511, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.761807918548584, + "num_tokens": 100413248.0, + "step": 2629 + }, + { + "epoch": 0.3345630326930416, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7766478657722473, + "num_tokens": 100448022.0, + "step": 2630 + }, + { + "epoch": 0.33469024297163213, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7762651443481445, + "num_tokens": 100493222.0, + "step": 2631 + }, + { + "epoch": 0.3348174532502226, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7981241941452026, + "num_tokens": 100534736.0, + "step": 2632 + }, + { + "epoch": 0.33494466352881314, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7967157363891602, + "num_tokens": 100576754.0, + "step": 2633 + }, + { + "epoch": 0.33507187380740366, + "grad_norm": 156.0, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.772053599357605, + "num_tokens": 100611407.0, + "step": 2634 + }, + { + "epoch": 0.33519908408599414, + "grad_norm": 216.0, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7844517230987549, + "num_tokens": 100643531.0, + "step": 2635 + }, + { + "epoch": 0.33532629436458466, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7881230711936951, + "num_tokens": 100678612.0, + "step": 2636 + }, + { + "epoch": 0.3354535046431752, + "grad_norm": 132.0, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7923647165298462, + "num_tokens": 100718873.0, + "step": 2637 + }, + { + "epoch": 0.33558071492176567, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7895740270614624, + "num_tokens": 100759883.0, + "step": 2638 + }, + { + "epoch": 0.3357079252003562, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7645496129989624, + "num_tokens": 100800022.0, + "step": 2639 + }, + { + "epoch": 0.3358351354789467, + "grad_norm": 92.5, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7646172046661377, + "num_tokens": 100837299.0, + "step": 2640 + }, + { + "epoch": 0.3359623457575372, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7769012451171875, + "num_tokens": 100868433.0, + "step": 2641 + }, + { + "epoch": 0.3360895560361277, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.773284912109375, + "num_tokens": 100907768.0, + "step": 2642 + }, + { + "epoch": 0.33621676631471825, + "grad_norm": 186.0, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7578641772270203, + "num_tokens": 100942723.0, + "step": 2643 + }, + { + "epoch": 0.3363439765933087, + "grad_norm": 131.0, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7954851388931274, + "num_tokens": 100984997.0, + "step": 2644 + }, + { + "epoch": 0.33647118687189925, + "grad_norm": 144.0, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.8020977973937988, + "num_tokens": 101026290.0, + "step": 2645 + }, + { + "epoch": 0.3365983971504898, + "grad_norm": 314.0, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7731566429138184, + "num_tokens": 101066766.0, + "step": 2646 + }, + { + "epoch": 0.33672560742908025, + "grad_norm": 292.0, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7854726910591125, + "num_tokens": 101100833.0, + "step": 2647 + }, + { + "epoch": 0.3368528177076708, + "grad_norm": 133.0, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.798625648021698, + "num_tokens": 101145552.0, + "step": 2648 + }, + { + "epoch": 0.3369800279862613, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7884703874588013, + "num_tokens": 101186644.0, + "step": 2649 + }, + { + "epoch": 0.3371072382648518, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7831023931503296, + "num_tokens": 101223356.0, + "step": 2650 + }, + { + "epoch": 0.3372344485434423, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.799927294254303, + "num_tokens": 101260527.0, + "step": 2651 + }, + { + "epoch": 0.33736165882203284, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7873702049255371, + "num_tokens": 101305403.0, + "step": 2652 + }, + { + "epoch": 0.3374888691006233, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.772441029548645, + "num_tokens": 101345858.0, + "step": 2653 + }, + { + "epoch": 0.33761607937921384, + "grad_norm": 99.5, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7578755617141724, + "num_tokens": 101380321.0, + "step": 2654 + }, + { + "epoch": 0.33774328965780437, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7961346507072449, + "num_tokens": 101419120.0, + "step": 2655 + }, + { + "epoch": 0.33787049993639484, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7858009338378906, + "num_tokens": 101453272.0, + "step": 2656 + }, + { + "epoch": 0.33799771021498537, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7723811864852905, + "num_tokens": 101486879.0, + "step": 2657 + }, + { + "epoch": 0.3381249204935759, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7778923511505127, + "num_tokens": 101523691.0, + "step": 2658 + }, + { + "epoch": 0.33825213077216637, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7865683436393738, + "num_tokens": 101555279.0, + "step": 2659 + }, + { + "epoch": 0.3383793410507569, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8060009479522705, + "num_tokens": 101591780.0, + "step": 2660 + }, + { + "epoch": 0.3385065513293474, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8079884052276611, + "num_tokens": 101631041.0, + "step": 2661 + }, + { + "epoch": 0.3386337616079379, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7747974395751953, + "num_tokens": 101666315.0, + "step": 2662 + }, + { + "epoch": 0.33876097188652843, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.8043028116226196, + "num_tokens": 101704046.0, + "step": 2663 + }, + { + "epoch": 0.33888818216511896, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7963137030601501, + "num_tokens": 101743076.0, + "step": 2664 + }, + { + "epoch": 0.33901539244370943, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8011794686317444, + "num_tokens": 101784006.0, + "step": 2665 + }, + { + "epoch": 0.33914260272229996, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7918868064880371, + "num_tokens": 101821845.0, + "step": 2666 + }, + { + "epoch": 0.3392698130008905, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7909467220306396, + "num_tokens": 101861121.0, + "step": 2667 + }, + { + "epoch": 0.33939702327948096, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7815904021263123, + "num_tokens": 101894330.0, + "step": 2668 + }, + { + "epoch": 0.3395242335580715, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7788856625556946, + "num_tokens": 101934835.0, + "step": 2669 + }, + { + "epoch": 0.339651443836662, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7728589177131653, + "num_tokens": 101970943.0, + "step": 2670 + }, + { + "epoch": 0.3397786541152525, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.771998405456543, + "num_tokens": 102001211.0, + "step": 2671 + }, + { + "epoch": 0.339905864393843, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7903056144714355, + "num_tokens": 102039399.0, + "step": 2672 + }, + { + "epoch": 0.34003307467243354, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7704512476921082, + "num_tokens": 102070635.0, + "step": 2673 + }, + { + "epoch": 0.340160284951024, + "grad_norm": 64.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7937513589859009, + "num_tokens": 102104590.0, + "step": 2674 + }, + { + "epoch": 0.34028749522961454, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7979395389556885, + "num_tokens": 102149267.0, + "step": 2675 + }, + { + "epoch": 0.3404147055082051, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7692860960960388, + "num_tokens": 102188089.0, + "step": 2676 + }, + { + "epoch": 0.34054191578679555, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7968748211860657, + "num_tokens": 102224335.0, + "step": 2677 + }, + { + "epoch": 0.3406691260653861, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.787919282913208, + "num_tokens": 102263421.0, + "step": 2678 + }, + { + "epoch": 0.3407963363439766, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7701592445373535, + "num_tokens": 102301133.0, + "step": 2679 + }, + { + "epoch": 0.34092354662256713, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7901575565338135, + "num_tokens": 102338516.0, + "step": 2680 + }, + { + "epoch": 0.3410507569011576, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7777067422866821, + "num_tokens": 102376010.0, + "step": 2681 + }, + { + "epoch": 0.34117796717974813, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.81138014793396, + "num_tokens": 102417080.0, + "step": 2682 + }, + { + "epoch": 0.34130517745833866, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7920296788215637, + "num_tokens": 102458042.0, + "step": 2683 + }, + { + "epoch": 0.34143238773692913, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7932906746864319, + "num_tokens": 102497152.0, + "step": 2684 + }, + { + "epoch": 0.34155959801551966, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7863858938217163, + "num_tokens": 102531486.0, + "step": 2685 + }, + { + "epoch": 0.3416868082941102, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7864916324615479, + "num_tokens": 102571552.0, + "step": 2686 + }, + { + "epoch": 0.34181401857270066, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7903296947479248, + "num_tokens": 102607338.0, + "step": 2687 + }, + { + "epoch": 0.3419412288512912, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7877933979034424, + "num_tokens": 102654445.0, + "step": 2688 + }, + { + "epoch": 0.3420684391298817, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7948253154754639, + "num_tokens": 102695265.0, + "step": 2689 + }, + { + "epoch": 0.3421956494084722, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7885891199111938, + "num_tokens": 102728621.0, + "step": 2690 + }, + { + "epoch": 0.3423228596870627, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7802392244338989, + "num_tokens": 102766378.0, + "step": 2691 + }, + { + "epoch": 0.34245006996565325, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7857784032821655, + "num_tokens": 102806331.0, + "step": 2692 + }, + { + "epoch": 0.3425772802442437, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7826754450798035, + "num_tokens": 102845314.0, + "step": 2693 + }, + { + "epoch": 0.34270449052283425, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7887882590293884, + "num_tokens": 102888614.0, + "step": 2694 + }, + { + "epoch": 0.3428317008014248, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7814801931381226, + "num_tokens": 102925988.0, + "step": 2695 + }, + { + "epoch": 0.34295891108001525, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7952673435211182, + "num_tokens": 102963150.0, + "step": 2696 + }, + { + "epoch": 0.3430861213586058, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7891038060188293, + "num_tokens": 102999164.0, + "step": 2697 + }, + { + "epoch": 0.3432133316371963, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7913209795951843, + "num_tokens": 103034337.0, + "step": 2698 + }, + { + "epoch": 0.3433405419157868, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7945880889892578, + "num_tokens": 103075062.0, + "step": 2699 + }, + { + "epoch": 0.3434677521943773, + "grad_norm": 119.0, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7771666049957275, + "num_tokens": 103113209.0, + "step": 2700 + }, + { + "epoch": 0.34359496247296784, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7917884588241577, + "num_tokens": 103153117.0, + "step": 2701 + }, + { + "epoch": 0.3437221727515583, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7874912023544312, + "num_tokens": 103192234.0, + "step": 2702 + }, + { + "epoch": 0.34384938303014884, + "grad_norm": 63.75, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7818540930747986, + "num_tokens": 103224958.0, + "step": 2703 + }, + { + "epoch": 0.34397659330873936, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7886568307876587, + "num_tokens": 103259679.0, + "step": 2704 + }, + { + "epoch": 0.34410380358732984, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7795238494873047, + "num_tokens": 103299987.0, + "step": 2705 + }, + { + "epoch": 0.34423101386592037, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.777478814125061, + "num_tokens": 103344457.0, + "step": 2706 + }, + { + "epoch": 0.3443582241445109, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7905159592628479, + "num_tokens": 103385173.0, + "step": 2707 + }, + { + "epoch": 0.34448543442310137, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7693212032318115, + "num_tokens": 103425972.0, + "step": 2708 + }, + { + "epoch": 0.3446126447016919, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7850041389465332, + "num_tokens": 103466819.0, + "step": 2709 + }, + { + "epoch": 0.3447398549802824, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7716277837753296, + "num_tokens": 103509660.0, + "step": 2710 + }, + { + "epoch": 0.3448670652588729, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7789299488067627, + "num_tokens": 103544557.0, + "step": 2711 + }, + { + "epoch": 0.3449942755374634, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.8011497259140015, + "num_tokens": 103585132.0, + "step": 2712 + }, + { + "epoch": 0.34512148581605395, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7942978143692017, + "num_tokens": 103626353.0, + "step": 2713 + }, + { + "epoch": 0.3452486960946444, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7882125377655029, + "num_tokens": 103659035.0, + "step": 2714 + }, + { + "epoch": 0.34537590637323495, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7982243299484253, + "num_tokens": 103699497.0, + "step": 2715 + }, + { + "epoch": 0.3455031166518255, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7692979574203491, + "num_tokens": 103733685.0, + "step": 2716 + }, + { + "epoch": 0.34563032693041595, + "grad_norm": 62.5, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7683778405189514, + "num_tokens": 103769399.0, + "step": 2717 + }, + { + "epoch": 0.3457575372090065, + "grad_norm": 103.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.794869601726532, + "num_tokens": 103809078.0, + "step": 2718 + }, + { + "epoch": 0.345884747487597, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8100003004074097, + "num_tokens": 103846764.0, + "step": 2719 + }, + { + "epoch": 0.3460119577661875, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7977493405342102, + "num_tokens": 103883089.0, + "step": 2720 + }, + { + "epoch": 0.346139168044778, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.8012809157371521, + "num_tokens": 103915775.0, + "step": 2721 + }, + { + "epoch": 0.34626637832336854, + "grad_norm": 103.5, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7852157950401306, + "num_tokens": 103953703.0, + "step": 2722 + }, + { + "epoch": 0.346393588601959, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7753151655197144, + "num_tokens": 103990775.0, + "step": 2723 + }, + { + "epoch": 0.34652079888054954, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7898919582366943, + "num_tokens": 104025927.0, + "step": 2724 + }, + { + "epoch": 0.34664800915914007, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7808908224105835, + "num_tokens": 104065860.0, + "step": 2725 + }, + { + "epoch": 0.34677521943773054, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7958542108535767, + "num_tokens": 104102117.0, + "step": 2726 + }, + { + "epoch": 0.34690242971632107, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7817168235778809, + "num_tokens": 104140896.0, + "step": 2727 + }, + { + "epoch": 0.3470296399949116, + "grad_norm": 95.5, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7885826230049133, + "num_tokens": 104181355.0, + "step": 2728 + }, + { + "epoch": 0.3471568502735021, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.781954824924469, + "num_tokens": 104218564.0, + "step": 2729 + }, + { + "epoch": 0.3472840605520926, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7926393747329712, + "num_tokens": 104256553.0, + "step": 2730 + }, + { + "epoch": 0.3474112708306831, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7995592951774597, + "num_tokens": 104291150.0, + "step": 2731 + }, + { + "epoch": 0.34753848110927366, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7760254144668579, + "num_tokens": 104327977.0, + "step": 2732 + }, + { + "epoch": 0.34766569138786413, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7801622152328491, + "num_tokens": 104365167.0, + "step": 2733 + }, + { + "epoch": 0.34779290166645466, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7924696803092957, + "num_tokens": 104400304.0, + "step": 2734 + }, + { + "epoch": 0.3479201119450452, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7713540196418762, + "num_tokens": 104437974.0, + "step": 2735 + }, + { + "epoch": 0.34804732222363566, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7854868769645691, + "num_tokens": 104474474.0, + "step": 2736 + }, + { + "epoch": 0.3481745325022262, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7739691734313965, + "num_tokens": 104511004.0, + "step": 2737 + }, + { + "epoch": 0.3483017427808167, + "grad_norm": 89.0, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7792893052101135, + "num_tokens": 104551889.0, + "step": 2738 + }, + { + "epoch": 0.3484289530594072, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7701199054718018, + "num_tokens": 104591962.0, + "step": 2739 + }, + { + "epoch": 0.3485561633379977, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.784681499004364, + "num_tokens": 104636201.0, + "step": 2740 + }, + { + "epoch": 0.34868337361658824, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7846651077270508, + "num_tokens": 104673448.0, + "step": 2741 + }, + { + "epoch": 0.3488105838951787, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7938237190246582, + "num_tokens": 104708879.0, + "step": 2742 + }, + { + "epoch": 0.34893779417376924, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7987885475158691, + "num_tokens": 104747073.0, + "step": 2743 + }, + { + "epoch": 0.3490650044523598, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7834475636482239, + "num_tokens": 104786950.0, + "step": 2744 + }, + { + "epoch": 0.34919221473095025, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.790856122970581, + "num_tokens": 104828306.0, + "step": 2745 + }, + { + "epoch": 0.3493194250095408, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7786434888839722, + "num_tokens": 104867030.0, + "step": 2746 + }, + { + "epoch": 0.3494466352881313, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7895141839981079, + "num_tokens": 104909182.0, + "step": 2747 + }, + { + "epoch": 0.3495738455667218, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7836015224456787, + "num_tokens": 104950648.0, + "step": 2748 + }, + { + "epoch": 0.3497010558453123, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7994444966316223, + "num_tokens": 104994780.0, + "step": 2749 + }, + { + "epoch": 0.34982826612390283, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7984066009521484, + "num_tokens": 105029469.0, + "step": 2750 + }, + { + "epoch": 0.3499554764024933, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7800838947296143, + "num_tokens": 105066797.0, + "step": 2751 + }, + { + "epoch": 0.35008268668108383, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8037720918655396, + "num_tokens": 105100869.0, + "step": 2752 + }, + { + "epoch": 0.35020989695967436, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8002947568893433, + "num_tokens": 105135409.0, + "step": 2753 + }, + { + "epoch": 0.35033710723826483, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7943553924560547, + "num_tokens": 105172839.0, + "step": 2754 + }, + { + "epoch": 0.35046431751685536, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7946414351463318, + "num_tokens": 105216656.0, + "step": 2755 + }, + { + "epoch": 0.3505915277954459, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7809284925460815, + "num_tokens": 105253740.0, + "step": 2756 + }, + { + "epoch": 0.35071873807403636, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7891609072685242, + "num_tokens": 105290073.0, + "step": 2757 + }, + { + "epoch": 0.3508459483526269, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7901421189308167, + "num_tokens": 105325933.0, + "step": 2758 + }, + { + "epoch": 0.3509731586312174, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.795835554599762, + "num_tokens": 105362962.0, + "step": 2759 + }, + { + "epoch": 0.3511003689098079, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.799015462398529, + "num_tokens": 105410074.0, + "step": 2760 + }, + { + "epoch": 0.3512275791883984, + "grad_norm": 86.5, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7730164527893066, + "num_tokens": 105450264.0, + "step": 2761 + }, + { + "epoch": 0.35135478946698895, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7778559923171997, + "num_tokens": 105486566.0, + "step": 2762 + }, + { + "epoch": 0.3514819997455794, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7928131222724915, + "num_tokens": 105527185.0, + "step": 2763 + }, + { + "epoch": 0.35160921002416995, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7831953167915344, + "num_tokens": 105557401.0, + "step": 2764 + }, + { + "epoch": 0.3517364203027605, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7916309237480164, + "num_tokens": 105586877.0, + "step": 2765 + }, + { + "epoch": 0.35186363058135095, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7965723276138306, + "num_tokens": 105626988.0, + "step": 2766 + }, + { + "epoch": 0.3519908408599415, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7813162803649902, + "num_tokens": 105665796.0, + "step": 2767 + }, + { + "epoch": 0.352118051138532, + "grad_norm": 107.5, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7781850099563599, + "num_tokens": 105708892.0, + "step": 2768 + }, + { + "epoch": 0.3522452614171225, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.781624436378479, + "num_tokens": 105748221.0, + "step": 2769 + }, + { + "epoch": 0.352372471695713, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.787592887878418, + "num_tokens": 105784568.0, + "step": 2770 + }, + { + "epoch": 0.35249968197430354, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7750416994094849, + "num_tokens": 105827223.0, + "step": 2771 + }, + { + "epoch": 0.352626892252894, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.810406506061554, + "num_tokens": 105871014.0, + "step": 2772 + }, + { + "epoch": 0.35275410253148454, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7824877500534058, + "num_tokens": 105908995.0, + "step": 2773 + }, + { + "epoch": 0.35288131281007507, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7798596620559692, + "num_tokens": 105948243.0, + "step": 2774 + }, + { + "epoch": 0.35300852308866554, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7930409908294678, + "num_tokens": 105983997.0, + "step": 2775 + }, + { + "epoch": 0.35313573336725607, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7735191583633423, + "num_tokens": 106020336.0, + "step": 2776 + }, + { + "epoch": 0.3532629436458466, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7843164205551147, + "num_tokens": 106053300.0, + "step": 2777 + }, + { + "epoch": 0.35339015392443707, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.797399640083313, + "num_tokens": 106091790.0, + "step": 2778 + }, + { + "epoch": 0.3535173642030276, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7866146564483643, + "num_tokens": 106123768.0, + "step": 2779 + }, + { + "epoch": 0.3536445744816181, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7962362766265869, + "num_tokens": 106158219.0, + "step": 2780 + }, + { + "epoch": 0.35377178476020865, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7806721329689026, + "num_tokens": 106195795.0, + "step": 2781 + }, + { + "epoch": 0.3538989950387991, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7913020849227905, + "num_tokens": 106234973.0, + "step": 2782 + }, + { + "epoch": 0.35402620531738965, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7680241465568542, + "num_tokens": 106267297.0, + "step": 2783 + }, + { + "epoch": 0.3541534155959802, + "grad_norm": 105.0, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7981656789779663, + "num_tokens": 106309070.0, + "step": 2784 + }, + { + "epoch": 0.35428062587457065, + "grad_norm": 135.0, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7860805988311768, + "num_tokens": 106345784.0, + "step": 2785 + }, + { + "epoch": 0.3544078361531612, + "grad_norm": 216.0, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7891852855682373, + "num_tokens": 106383172.0, + "step": 2786 + }, + { + "epoch": 0.3545350464317517, + "grad_norm": 105.0, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7889460325241089, + "num_tokens": 106416916.0, + "step": 2787 + }, + { + "epoch": 0.3546622567103422, + "grad_norm": 119.0, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.771805465221405, + "num_tokens": 106456969.0, + "step": 2788 + }, + { + "epoch": 0.3547894669889327, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7823002338409424, + "num_tokens": 106490476.0, + "step": 2789 + }, + { + "epoch": 0.35491667726752324, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7886431217193604, + "num_tokens": 106525990.0, + "step": 2790 + }, + { + "epoch": 0.3550438875461137, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7575668692588806, + "num_tokens": 106562953.0, + "step": 2791 + }, + { + "epoch": 0.35517109782470424, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.788215160369873, + "num_tokens": 106599142.0, + "step": 2792 + }, + { + "epoch": 0.35529830810329477, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.803870439529419, + "num_tokens": 106636743.0, + "step": 2793 + }, + { + "epoch": 0.35542551838188524, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7937780022621155, + "num_tokens": 106678992.0, + "step": 2794 + }, + { + "epoch": 0.35555272866047577, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7773406505584717, + "num_tokens": 106719884.0, + "step": 2795 + }, + { + "epoch": 0.3556799389390663, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7857666015625, + "num_tokens": 106755224.0, + "step": 2796 + }, + { + "epoch": 0.35580714921765677, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.8037106394767761, + "num_tokens": 106789366.0, + "step": 2797 + }, + { + "epoch": 0.3559343594962473, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7758342623710632, + "num_tokens": 106826382.0, + "step": 2798 + }, + { + "epoch": 0.3560615697748378, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7736291289329529, + "num_tokens": 106860027.0, + "step": 2799 + }, + { + "epoch": 0.3561887800534283, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.794127881526947, + "num_tokens": 106898684.0, + "step": 2800 + }, + { + "epoch": 0.35631599033201883, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7922749519348145, + "num_tokens": 106934496.0, + "step": 2801 + }, + { + "epoch": 0.35644320061060936, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8105005621910095, + "num_tokens": 106974335.0, + "step": 2802 + }, + { + "epoch": 0.35657041088919983, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7583280801773071, + "num_tokens": 107013914.0, + "step": 2803 + }, + { + "epoch": 0.35669762116779036, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7963547110557556, + "num_tokens": 107053747.0, + "step": 2804 + }, + { + "epoch": 0.3568248314463809, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7781780958175659, + "num_tokens": 107095245.0, + "step": 2805 + }, + { + "epoch": 0.35695204172497136, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7898483276367188, + "num_tokens": 107133627.0, + "step": 2806 + }, + { + "epoch": 0.3570792520035619, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7639791965484619, + "num_tokens": 107170128.0, + "step": 2807 + }, + { + "epoch": 0.3572064622821524, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8032338619232178, + "num_tokens": 107209137.0, + "step": 2808 + }, + { + "epoch": 0.3573336725607429, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7852780818939209, + "num_tokens": 107244866.0, + "step": 2809 + }, + { + "epoch": 0.3574608828393334, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.789900541305542, + "num_tokens": 107287399.0, + "step": 2810 + }, + { + "epoch": 0.35758809311792394, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.776378870010376, + "num_tokens": 107325627.0, + "step": 2811 + }, + { + "epoch": 0.3577153033965144, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8036535978317261, + "num_tokens": 107359375.0, + "step": 2812 + }, + { + "epoch": 0.35784251367510495, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7773464918136597, + "num_tokens": 107400286.0, + "step": 2813 + }, + { + "epoch": 0.3579697239536955, + "grad_norm": 111.5, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.774247944355011, + "num_tokens": 107431768.0, + "step": 2814 + }, + { + "epoch": 0.35809693423228595, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.800445556640625, + "num_tokens": 107470875.0, + "step": 2815 + }, + { + "epoch": 0.3582241445108765, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7762234210968018, + "num_tokens": 107510988.0, + "step": 2816 + }, + { + "epoch": 0.358351354789467, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7967676520347595, + "num_tokens": 107552423.0, + "step": 2817 + }, + { + "epoch": 0.3584785650680575, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7815858721733093, + "num_tokens": 107591693.0, + "step": 2818 + }, + { + "epoch": 0.358605775346648, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7930215001106262, + "num_tokens": 107626017.0, + "step": 2819 + }, + { + "epoch": 0.35873298562523853, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8090858459472656, + "num_tokens": 107664927.0, + "step": 2820 + }, + { + "epoch": 0.358860195903829, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7869260907173157, + "num_tokens": 107702487.0, + "step": 2821 + }, + { + "epoch": 0.35898740618241953, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8046339750289917, + "num_tokens": 107741263.0, + "step": 2822 + }, + { + "epoch": 0.35911461646101006, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7925233840942383, + "num_tokens": 107781079.0, + "step": 2823 + }, + { + "epoch": 0.35924182673960053, + "grad_norm": 156.0, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8092244863510132, + "num_tokens": 107819309.0, + "step": 2824 + }, + { + "epoch": 0.35936903701819106, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7886414527893066, + "num_tokens": 107853251.0, + "step": 2825 + }, + { + "epoch": 0.3594962472967816, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8090405464172363, + "num_tokens": 107893476.0, + "step": 2826 + }, + { + "epoch": 0.35962345757537206, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7961817979812622, + "num_tokens": 107931204.0, + "step": 2827 + }, + { + "epoch": 0.3597506678539626, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7682408690452576, + "num_tokens": 107966471.0, + "step": 2828 + }, + { + "epoch": 0.3598778781325531, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7894534468650818, + "num_tokens": 108005328.0, + "step": 2829 + }, + { + "epoch": 0.36000508841114365, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.778494656085968, + "num_tokens": 108046295.0, + "step": 2830 + }, + { + "epoch": 0.3601322986897341, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7903780341148376, + "num_tokens": 108078164.0, + "step": 2831 + }, + { + "epoch": 0.36025950896832465, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7847228050231934, + "num_tokens": 108117105.0, + "step": 2832 + }, + { + "epoch": 0.3603867192469152, + "grad_norm": 89.0, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7847137451171875, + "num_tokens": 108155301.0, + "step": 2833 + }, + { + "epoch": 0.36051392952550565, + "grad_norm": 109.0, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.8011786937713623, + "num_tokens": 108193091.0, + "step": 2834 + }, + { + "epoch": 0.3606411398040962, + "grad_norm": 76.5, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7867248058319092, + "num_tokens": 108233719.0, + "step": 2835 + }, + { + "epoch": 0.3607683500826867, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7914397716522217, + "num_tokens": 108265775.0, + "step": 2836 + }, + { + "epoch": 0.3608955603612772, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.8004817962646484, + "num_tokens": 108307953.0, + "step": 2837 + }, + { + "epoch": 0.3610227706398677, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7883223295211792, + "num_tokens": 108351998.0, + "step": 2838 + }, + { + "epoch": 0.36114998091845824, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7787088751792908, + "num_tokens": 108395968.0, + "step": 2839 + }, + { + "epoch": 0.3612771911970487, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7693487405776978, + "num_tokens": 108433045.0, + "step": 2840 + }, + { + "epoch": 0.36140440147563924, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7828064560890198, + "num_tokens": 108470964.0, + "step": 2841 + }, + { + "epoch": 0.36153161175422976, + "grad_norm": 108.0, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.786385178565979, + "num_tokens": 108512572.0, + "step": 2842 + }, + { + "epoch": 0.36165882203282024, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7704018354415894, + "num_tokens": 108548526.0, + "step": 2843 + }, + { + "epoch": 0.36178603231141077, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7840200662612915, + "num_tokens": 108588447.0, + "step": 2844 + }, + { + "epoch": 0.3619132425900013, + "grad_norm": 104.0, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7751814126968384, + "num_tokens": 108628426.0, + "step": 2845 + }, + { + "epoch": 0.36204045286859177, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7778550982475281, + "num_tokens": 108661355.0, + "step": 2846 + }, + { + "epoch": 0.3621676631471823, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8056992292404175, + "num_tokens": 108698505.0, + "step": 2847 + }, + { + "epoch": 0.3622948734257728, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7959625720977783, + "num_tokens": 108737409.0, + "step": 2848 + }, + { + "epoch": 0.3624220837043633, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7885692119598389, + "num_tokens": 108772969.0, + "step": 2849 + }, + { + "epoch": 0.3625492939829538, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7930793166160583, + "num_tokens": 108805129.0, + "step": 2850 + }, + { + "epoch": 0.36267650426154435, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7710196375846863, + "num_tokens": 108841197.0, + "step": 2851 + }, + { + "epoch": 0.3628037145401348, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8054889440536499, + "num_tokens": 108876442.0, + "step": 2852 + }, + { + "epoch": 0.36293092481872535, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7672094106674194, + "num_tokens": 108911432.0, + "step": 2853 + }, + { + "epoch": 0.3630581350973159, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7734649777412415, + "num_tokens": 108952977.0, + "step": 2854 + }, + { + "epoch": 0.36318534537590635, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.764409065246582, + "num_tokens": 108995602.0, + "step": 2855 + }, + { + "epoch": 0.3633125556544969, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.794898271560669, + "num_tokens": 109040226.0, + "step": 2856 + }, + { + "epoch": 0.3634397659330874, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7939014434814453, + "num_tokens": 109082898.0, + "step": 2857 + }, + { + "epoch": 0.3635669762116779, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7812472581863403, + "num_tokens": 109122460.0, + "step": 2858 + }, + { + "epoch": 0.3636941864902684, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7887486219406128, + "num_tokens": 109161680.0, + "step": 2859 + }, + { + "epoch": 0.36382139676885894, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8063497543334961, + "num_tokens": 109198943.0, + "step": 2860 + }, + { + "epoch": 0.3639486070474494, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7960076332092285, + "num_tokens": 109235719.0, + "step": 2861 + }, + { + "epoch": 0.36407581732603994, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7776885628700256, + "num_tokens": 109272681.0, + "step": 2862 + }, + { + "epoch": 0.36420302760463047, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7704550623893738, + "num_tokens": 109306114.0, + "step": 2863 + }, + { + "epoch": 0.36433023788322094, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7727890014648438, + "num_tokens": 109337728.0, + "step": 2864 + }, + { + "epoch": 0.36445744816181147, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7827757596969604, + "num_tokens": 109377809.0, + "step": 2865 + }, + { + "epoch": 0.364584658440402, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8045480251312256, + "num_tokens": 109413499.0, + "step": 2866 + }, + { + "epoch": 0.36471186871899247, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7838563323020935, + "num_tokens": 109460961.0, + "step": 2867 + }, + { + "epoch": 0.364839078997583, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7983849048614502, + "num_tokens": 109500615.0, + "step": 2868 + }, + { + "epoch": 0.36496628927617353, + "grad_norm": 131.0, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7884424924850464, + "num_tokens": 109533921.0, + "step": 2869 + }, + { + "epoch": 0.365093499554764, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7814691066741943, + "num_tokens": 109567069.0, + "step": 2870 + }, + { + "epoch": 0.36522070983335453, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.8025221228599548, + "num_tokens": 109603696.0, + "step": 2871 + }, + { + "epoch": 0.36534792011194506, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8065743446350098, + "num_tokens": 109644590.0, + "step": 2872 + }, + { + "epoch": 0.36547513039053553, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7850416898727417, + "num_tokens": 109678869.0, + "step": 2873 + }, + { + "epoch": 0.36560234066912606, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7830835580825806, + "num_tokens": 109722517.0, + "step": 2874 + }, + { + "epoch": 0.3657295509477166, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7844141125679016, + "num_tokens": 109757869.0, + "step": 2875 + }, + { + "epoch": 0.36585676122630706, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7813662886619568, + "num_tokens": 109793162.0, + "step": 2876 + }, + { + "epoch": 0.3659839715048976, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7976288795471191, + "num_tokens": 109832942.0, + "step": 2877 + }, + { + "epoch": 0.3661111817834881, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7906373739242554, + "num_tokens": 109876230.0, + "step": 2878 + }, + { + "epoch": 0.3662383920620786, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7708652019500732, + "num_tokens": 109911844.0, + "step": 2879 + }, + { + "epoch": 0.3663656023406691, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7797337174415588, + "num_tokens": 109957619.0, + "step": 2880 + }, + { + "epoch": 0.36649281261925964, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7754085659980774, + "num_tokens": 109999117.0, + "step": 2881 + }, + { + "epoch": 0.3666200228978502, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7926868200302124, + "num_tokens": 110039557.0, + "step": 2882 + }, + { + "epoch": 0.36674723317644065, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7775564789772034, + "num_tokens": 110074418.0, + "step": 2883 + }, + { + "epoch": 0.3668744434550312, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7932054996490479, + "num_tokens": 110109255.0, + "step": 2884 + }, + { + "epoch": 0.3670016537336217, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7962497472763062, + "num_tokens": 110150180.0, + "step": 2885 + }, + { + "epoch": 0.3671288640122122, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7740085124969482, + "num_tokens": 110193653.0, + "step": 2886 + }, + { + "epoch": 0.3672560742908027, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7950377464294434, + "num_tokens": 110231020.0, + "step": 2887 + }, + { + "epoch": 0.36738328456939323, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8068243265151978, + "num_tokens": 110269397.0, + "step": 2888 + }, + { + "epoch": 0.3675104948479837, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8060206174850464, + "num_tokens": 110311188.0, + "step": 2889 + }, + { + "epoch": 0.36763770512657423, + "grad_norm": 117.0, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7920137643814087, + "num_tokens": 110343801.0, + "step": 2890 + }, + { + "epoch": 0.36776491540516476, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7716254591941833, + "num_tokens": 110385830.0, + "step": 2891 + }, + { + "epoch": 0.36789212568375523, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7832440137863159, + "num_tokens": 110421494.0, + "step": 2892 + }, + { + "epoch": 0.36801933596234576, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7896385192871094, + "num_tokens": 110463132.0, + "step": 2893 + }, + { + "epoch": 0.3681465462409363, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7925456762313843, + "num_tokens": 110498330.0, + "step": 2894 + }, + { + "epoch": 0.36827375651952676, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7921188473701477, + "num_tokens": 110534438.0, + "step": 2895 + }, + { + "epoch": 0.3684009667981173, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7838705778121948, + "num_tokens": 110572059.0, + "step": 2896 + }, + { + "epoch": 0.3685281770767078, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7887672185897827, + "num_tokens": 110611012.0, + "step": 2897 + }, + { + "epoch": 0.3686553873552983, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7987111210823059, + "num_tokens": 110647429.0, + "step": 2898 + }, + { + "epoch": 0.3687825976338888, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7828830480575562, + "num_tokens": 110685052.0, + "step": 2899 + }, + { + "epoch": 0.36890980791247935, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7653619050979614, + "num_tokens": 110725222.0, + "step": 2900 + }, + { + "epoch": 0.3690370181910698, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7814199924468994, + "num_tokens": 110764314.0, + "step": 2901 + }, + { + "epoch": 0.36916422846966035, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.798039972782135, + "num_tokens": 110803585.0, + "step": 2902 + }, + { + "epoch": 0.3692914387482509, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7853385210037231, + "num_tokens": 110846532.0, + "step": 2903 + }, + { + "epoch": 0.36941864902684135, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.784955620765686, + "num_tokens": 110889807.0, + "step": 2904 + }, + { + "epoch": 0.3695458593054319, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7670270204544067, + "num_tokens": 110930567.0, + "step": 2905 + }, + { + "epoch": 0.3696730695840224, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7979820966720581, + "num_tokens": 110968725.0, + "step": 2906 + }, + { + "epoch": 0.3698002798626129, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8137560486793518, + "num_tokens": 111002724.0, + "step": 2907 + }, + { + "epoch": 0.3699274901412034, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.784553050994873, + "num_tokens": 111036668.0, + "step": 2908 + }, + { + "epoch": 0.37005470041979394, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7767195701599121, + "num_tokens": 111077712.0, + "step": 2909 + }, + { + "epoch": 0.3701819106983844, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7777389287948608, + "num_tokens": 111118856.0, + "step": 2910 + }, + { + "epoch": 0.37030912097697494, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.783431887626648, + "num_tokens": 111156712.0, + "step": 2911 + }, + { + "epoch": 0.37043633125556547, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.790664792060852, + "num_tokens": 111195542.0, + "step": 2912 + }, + { + "epoch": 0.37056354153415594, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.770987331867218, + "num_tokens": 111237827.0, + "step": 2913 + }, + { + "epoch": 0.37069075181274647, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7971162796020508, + "num_tokens": 111278802.0, + "step": 2914 + }, + { + "epoch": 0.370817962091337, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7836389541625977, + "num_tokens": 111313795.0, + "step": 2915 + }, + { + "epoch": 0.37094517236992747, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7920329570770264, + "num_tokens": 111353798.0, + "step": 2916 + }, + { + "epoch": 0.371072382648518, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7923198938369751, + "num_tokens": 111390613.0, + "step": 2917 + }, + { + "epoch": 0.3711995929271085, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.792319655418396, + "num_tokens": 111432043.0, + "step": 2918 + }, + { + "epoch": 0.371326803205699, + "grad_norm": 77.0, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7991353273391724, + "num_tokens": 111476575.0, + "step": 2919 + }, + { + "epoch": 0.3714540134842895, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7861482501029968, + "num_tokens": 111515304.0, + "step": 2920 + }, + { + "epoch": 0.37158122376288005, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.8002477884292603, + "num_tokens": 111550246.0, + "step": 2921 + }, + { + "epoch": 0.3717084340414705, + "grad_norm": 89.0, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7929859161376953, + "num_tokens": 111592189.0, + "step": 2922 + }, + { + "epoch": 0.37183564432006105, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7991385459899902, + "num_tokens": 111638177.0, + "step": 2923 + }, + { + "epoch": 0.3719628545986516, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7887827157974243, + "num_tokens": 111674365.0, + "step": 2924 + }, + { + "epoch": 0.37209006487724205, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.784650981426239, + "num_tokens": 111714571.0, + "step": 2925 + }, + { + "epoch": 0.3722172751558326, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7859271764755249, + "num_tokens": 111755214.0, + "step": 2926 + }, + { + "epoch": 0.3723444854344231, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.762211799621582, + "num_tokens": 111796516.0, + "step": 2927 + }, + { + "epoch": 0.3724716957130136, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7672191858291626, + "num_tokens": 111832615.0, + "step": 2928 + }, + { + "epoch": 0.3725989059916041, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7804069519042969, + "num_tokens": 111874682.0, + "step": 2929 + }, + { + "epoch": 0.37272611627019464, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7705014944076538, + "num_tokens": 111916369.0, + "step": 2930 + }, + { + "epoch": 0.37285332654878517, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7814604043960571, + "num_tokens": 111950671.0, + "step": 2931 + }, + { + "epoch": 0.37298053682737564, + "grad_norm": 92.5, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.766778826713562, + "num_tokens": 111988846.0, + "step": 2932 + }, + { + "epoch": 0.37310774710596617, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7930756211280823, + "num_tokens": 112022853.0, + "step": 2933 + }, + { + "epoch": 0.3732349573845567, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7841700315475464, + "num_tokens": 112065631.0, + "step": 2934 + }, + { + "epoch": 0.37336216766314717, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.779404878616333, + "num_tokens": 112107236.0, + "step": 2935 + }, + { + "epoch": 0.3734893779417377, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7664300203323364, + "num_tokens": 112153914.0, + "step": 2936 + }, + { + "epoch": 0.3736165882203282, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7884308099746704, + "num_tokens": 112187596.0, + "step": 2937 + }, + { + "epoch": 0.3737437984989187, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7818828225135803, + "num_tokens": 112222409.0, + "step": 2938 + }, + { + "epoch": 0.37387100877750923, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7755575776100159, + "num_tokens": 112257699.0, + "step": 2939 + }, + { + "epoch": 0.37399821905609976, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7981608510017395, + "num_tokens": 112292431.0, + "step": 2940 + }, + { + "epoch": 0.37412542933469023, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7895233631134033, + "num_tokens": 112331241.0, + "step": 2941 + }, + { + "epoch": 0.37425263961328076, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7837525606155396, + "num_tokens": 112368486.0, + "step": 2942 + }, + { + "epoch": 0.3743798498918713, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7770974636077881, + "num_tokens": 112409638.0, + "step": 2943 + }, + { + "epoch": 0.37450706017046176, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7851840257644653, + "num_tokens": 112448257.0, + "step": 2944 + }, + { + "epoch": 0.3746342704490523, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.80271977186203, + "num_tokens": 112488739.0, + "step": 2945 + }, + { + "epoch": 0.3747614807276428, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.773330569267273, + "num_tokens": 112528621.0, + "step": 2946 + }, + { + "epoch": 0.3748886910062333, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7804956436157227, + "num_tokens": 112564615.0, + "step": 2947 + }, + { + "epoch": 0.3750159012848238, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7728313207626343, + "num_tokens": 112598190.0, + "step": 2948 + }, + { + "epoch": 0.37514311156341434, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7864524126052856, + "num_tokens": 112632078.0, + "step": 2949 + }, + { + "epoch": 0.3752703218420048, + "grad_norm": 115.5, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7637441754341125, + "num_tokens": 112667848.0, + "step": 2950 + }, + { + "epoch": 0.37539753212059535, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7824482321739197, + "num_tokens": 112707037.0, + "step": 2951 + }, + { + "epoch": 0.3755247423991859, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7982741594314575, + "num_tokens": 112751921.0, + "step": 2952 + }, + { + "epoch": 0.37565195267777635, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7968756556510925, + "num_tokens": 112793061.0, + "step": 2953 + }, + { + "epoch": 0.3757791629563669, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8043205738067627, + "num_tokens": 112833852.0, + "step": 2954 + }, + { + "epoch": 0.3759063732349574, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7812469005584717, + "num_tokens": 112866718.0, + "step": 2955 + }, + { + "epoch": 0.3760335835135479, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7907898426055908, + "num_tokens": 112909420.0, + "step": 2956 + }, + { + "epoch": 0.3761607937921384, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7860094308853149, + "num_tokens": 112944570.0, + "step": 2957 + }, + { + "epoch": 0.37628800407072893, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8043937683105469, + "num_tokens": 112984107.0, + "step": 2958 + }, + { + "epoch": 0.3764152143493194, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8138371706008911, + "num_tokens": 113023066.0, + "step": 2959 + }, + { + "epoch": 0.37654242462790993, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7806954383850098, + "num_tokens": 113060093.0, + "step": 2960 + }, + { + "epoch": 0.37666963490650046, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7779711484909058, + "num_tokens": 113095940.0, + "step": 2961 + }, + { + "epoch": 0.37679684518509093, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7964491844177246, + "num_tokens": 113130699.0, + "step": 2962 + }, + { + "epoch": 0.37692405546368146, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7888137102127075, + "num_tokens": 113169145.0, + "step": 2963 + }, + { + "epoch": 0.377051265742272, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8000514507293701, + "num_tokens": 113201987.0, + "step": 2964 + }, + { + "epoch": 0.37717847602086246, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7944972515106201, + "num_tokens": 113239588.0, + "step": 2965 + }, + { + "epoch": 0.377305686299453, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.787947416305542, + "num_tokens": 113277075.0, + "step": 2966 + }, + { + "epoch": 0.3774328965780435, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.8228566646575928, + "num_tokens": 113313708.0, + "step": 2967 + }, + { + "epoch": 0.377560106856634, + "grad_norm": 88.0, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7647650837898254, + "num_tokens": 113344027.0, + "step": 2968 + }, + { + "epoch": 0.3776873171352245, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.794682502746582, + "num_tokens": 113388444.0, + "step": 2969 + }, + { + "epoch": 0.37781452741381505, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7806978821754456, + "num_tokens": 113428660.0, + "step": 2970 + }, + { + "epoch": 0.3779417376924055, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7636110186576843, + "num_tokens": 113469160.0, + "step": 2971 + }, + { + "epoch": 0.37806894797099605, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7596322298049927, + "num_tokens": 113504291.0, + "step": 2972 + }, + { + "epoch": 0.3781961582495866, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7897837162017822, + "num_tokens": 113548085.0, + "step": 2973 + }, + { + "epoch": 0.37832336852817705, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7943840622901917, + "num_tokens": 113581664.0, + "step": 2974 + }, + { + "epoch": 0.3784505788067676, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7988908886909485, + "num_tokens": 113618552.0, + "step": 2975 + }, + { + "epoch": 0.3785777890853581, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7613829374313354, + "num_tokens": 113653406.0, + "step": 2976 + }, + { + "epoch": 0.3787049993639486, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7977064251899719, + "num_tokens": 113693803.0, + "step": 2977 + }, + { + "epoch": 0.3788322096425391, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.8048367500305176, + "num_tokens": 113734637.0, + "step": 2978 + }, + { + "epoch": 0.37895941992112964, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7764381170272827, + "num_tokens": 113776579.0, + "step": 2979 + }, + { + "epoch": 0.37908663019972016, + "grad_norm": 122.5, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7690235376358032, + "num_tokens": 113818857.0, + "step": 2980 + }, + { + "epoch": 0.37921384047831064, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7810513973236084, + "num_tokens": 113853248.0, + "step": 2981 + }, + { + "epoch": 0.37934105075690117, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7938805222511292, + "num_tokens": 113894206.0, + "step": 2982 + }, + { + "epoch": 0.3794682610354917, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7971401810646057, + "num_tokens": 113931309.0, + "step": 2983 + }, + { + "epoch": 0.37959547131408217, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7790846228599548, + "num_tokens": 113976559.0, + "step": 2984 + }, + { + "epoch": 0.3797226815926727, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7865278720855713, + "num_tokens": 114010814.0, + "step": 2985 + }, + { + "epoch": 0.3798498918712632, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8042111992835999, + "num_tokens": 114049389.0, + "step": 2986 + }, + { + "epoch": 0.3799771021498537, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8038747310638428, + "num_tokens": 114081463.0, + "step": 2987 + }, + { + "epoch": 0.3801043124284442, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7997356057167053, + "num_tokens": 114117175.0, + "step": 2988 + }, + { + "epoch": 0.38023152270703475, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7773443460464478, + "num_tokens": 114153764.0, + "step": 2989 + }, + { + "epoch": 0.3803587329856252, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7866668105125427, + "num_tokens": 114191487.0, + "step": 2990 + }, + { + "epoch": 0.38048594326421575, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7650407552719116, + "num_tokens": 114233880.0, + "step": 2991 + }, + { + "epoch": 0.3806131535428063, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7855837941169739, + "num_tokens": 114269238.0, + "step": 2992 + }, + { + "epoch": 0.38074036382139675, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.8029780387878418, + "num_tokens": 114307645.0, + "step": 2993 + }, + { + "epoch": 0.3808675740999873, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.79168701171875, + "num_tokens": 114341800.0, + "step": 2994 + }, + { + "epoch": 0.3809947843785778, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7981323003768921, + "num_tokens": 114378363.0, + "step": 2995 + }, + { + "epoch": 0.3811219946571683, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7761603593826294, + "num_tokens": 114414489.0, + "step": 2996 + }, + { + "epoch": 0.3812492049357588, + "grad_norm": 106.5, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.780460000038147, + "num_tokens": 114449197.0, + "step": 2997 + }, + { + "epoch": 0.38137641521434934, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7848951816558838, + "num_tokens": 114478654.0, + "step": 2998 + }, + { + "epoch": 0.3815036254929398, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.8014122247695923, + "num_tokens": 114521807.0, + "step": 2999 + }, + { + "epoch": 0.38163083577153034, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.805858850479126, + "num_tokens": 114558297.0, + "step": 3000 + }, + { + "epoch": 0.38175804605012087, + "grad_norm": 96.5, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7974488735198975, + "num_tokens": 114599630.0, + "step": 3001 + }, + { + "epoch": 0.38188525632871134, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7868831157684326, + "num_tokens": 114638697.0, + "step": 3002 + }, + { + "epoch": 0.38201246660730187, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7861741781234741, + "num_tokens": 114676106.0, + "step": 3003 + }, + { + "epoch": 0.3821396768858924, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7824755907058716, + "num_tokens": 114712026.0, + "step": 3004 + }, + { + "epoch": 0.38226688716448287, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7654540538787842, + "num_tokens": 114752348.0, + "step": 3005 + }, + { + "epoch": 0.3823940974430734, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7942680716514587, + "num_tokens": 114789922.0, + "step": 3006 + }, + { + "epoch": 0.38252130772166393, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.782781720161438, + "num_tokens": 114838834.0, + "step": 3007 + }, + { + "epoch": 0.3826485180002544, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7668099999427795, + "num_tokens": 114873320.0, + "step": 3008 + }, + { + "epoch": 0.38277572827884493, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7975077629089355, + "num_tokens": 114908293.0, + "step": 3009 + }, + { + "epoch": 0.38290293855743546, + "grad_norm": 88.0, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7777817845344543, + "num_tokens": 114950083.0, + "step": 3010 + }, + { + "epoch": 0.38303014883602593, + "grad_norm": 77.0, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7887580394744873, + "num_tokens": 114982353.0, + "step": 3011 + }, + { + "epoch": 0.38315735911461646, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7952730655670166, + "num_tokens": 115017668.0, + "step": 3012 + }, + { + "epoch": 0.383284569393207, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7981442213058472, + "num_tokens": 115055266.0, + "step": 3013 + }, + { + "epoch": 0.38341177967179746, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7803084254264832, + "num_tokens": 115089728.0, + "step": 3014 + }, + { + "epoch": 0.383538989950388, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7780223488807678, + "num_tokens": 115135192.0, + "step": 3015 + }, + { + "epoch": 0.3836662002289785, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7674330472946167, + "num_tokens": 115170350.0, + "step": 3016 + }, + { + "epoch": 0.383793410507569, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7979892492294312, + "num_tokens": 115202724.0, + "step": 3017 + }, + { + "epoch": 0.3839206207861595, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7854416966438293, + "num_tokens": 115239274.0, + "step": 3018 + }, + { + "epoch": 0.38404783106475004, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.8017697334289551, + "num_tokens": 115278330.0, + "step": 3019 + }, + { + "epoch": 0.3841750413433405, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.784166693687439, + "num_tokens": 115321258.0, + "step": 3020 + }, + { + "epoch": 0.38430225162193105, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7886165380477905, + "num_tokens": 115362609.0, + "step": 3021 + }, + { + "epoch": 0.3844294619005216, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7651782035827637, + "num_tokens": 115405814.0, + "step": 3022 + }, + { + "epoch": 0.38455667217911205, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7846417427062988, + "num_tokens": 115440856.0, + "step": 3023 + }, + { + "epoch": 0.3846838824577026, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7964059114456177, + "num_tokens": 115482582.0, + "step": 3024 + }, + { + "epoch": 0.3848110927362931, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7815595865249634, + "num_tokens": 115527268.0, + "step": 3025 + }, + { + "epoch": 0.3849383030148836, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7840231657028198, + "num_tokens": 115563757.0, + "step": 3026 + }, + { + "epoch": 0.3850655132934741, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7926191687583923, + "num_tokens": 115599629.0, + "step": 3027 + }, + { + "epoch": 0.38519272357206463, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7887967824935913, + "num_tokens": 115634261.0, + "step": 3028 + }, + { + "epoch": 0.3853199338506551, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8036356568336487, + "num_tokens": 115671150.0, + "step": 3029 + }, + { + "epoch": 0.38544714412924563, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.804663360118866, + "num_tokens": 115714328.0, + "step": 3030 + }, + { + "epoch": 0.38557435440783616, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8063430786132812, + "num_tokens": 115747118.0, + "step": 3031 + }, + { + "epoch": 0.3857015646864267, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7735748291015625, + "num_tokens": 115790309.0, + "step": 3032 + }, + { + "epoch": 0.38582877496501716, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8016791343688965, + "num_tokens": 115825630.0, + "step": 3033 + }, + { + "epoch": 0.3859559852436077, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7933052778244019, + "num_tokens": 115858848.0, + "step": 3034 + }, + { + "epoch": 0.3860831955221982, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7996501326560974, + "num_tokens": 115900024.0, + "step": 3035 + }, + { + "epoch": 0.3862104058007887, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7647671103477478, + "num_tokens": 115938899.0, + "step": 3036 + }, + { + "epoch": 0.3863376160793792, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.792170524597168, + "num_tokens": 115972199.0, + "step": 3037 + }, + { + "epoch": 0.38646482635796975, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7967431545257568, + "num_tokens": 116011506.0, + "step": 3038 + }, + { + "epoch": 0.3865920366365602, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7713420391082764, + "num_tokens": 116051428.0, + "step": 3039 + }, + { + "epoch": 0.38671924691515075, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.797039270401001, + "num_tokens": 116090307.0, + "step": 3040 + }, + { + "epoch": 0.3868464571937413, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7761155962944031, + "num_tokens": 116123381.0, + "step": 3041 + }, + { + "epoch": 0.38697366747233175, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7919526100158691, + "num_tokens": 116157755.0, + "step": 3042 + }, + { + "epoch": 0.3871008777509223, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7927442789077759, + "num_tokens": 116197371.0, + "step": 3043 + }, + { + "epoch": 0.3872280880295128, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7736446857452393, + "num_tokens": 116240544.0, + "step": 3044 + }, + { + "epoch": 0.3873552983081033, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7824792265892029, + "num_tokens": 116272079.0, + "step": 3045 + }, + { + "epoch": 0.3874825085866938, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7855046987533569, + "num_tokens": 116309641.0, + "step": 3046 + }, + { + "epoch": 0.38760971886528434, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7792922854423523, + "num_tokens": 116344470.0, + "step": 3047 + }, + { + "epoch": 0.3877369291438748, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7898913621902466, + "num_tokens": 116380177.0, + "step": 3048 + }, + { + "epoch": 0.38786413942246534, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.8000938296318054, + "num_tokens": 116416482.0, + "step": 3049 + }, + { + "epoch": 0.38799134970105587, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7741144895553589, + "num_tokens": 116456103.0, + "step": 3050 + }, + { + "epoch": 0.38811855997964634, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.8015466928482056, + "num_tokens": 116496810.0, + "step": 3051 + }, + { + "epoch": 0.38824577025823687, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7840194702148438, + "num_tokens": 116532999.0, + "step": 3052 + }, + { + "epoch": 0.3883729805368274, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7969616651535034, + "num_tokens": 116565700.0, + "step": 3053 + }, + { + "epoch": 0.38850019081541787, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7941307425498962, + "num_tokens": 116602798.0, + "step": 3054 + }, + { + "epoch": 0.3886274010940084, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.796124279499054, + "num_tokens": 116638743.0, + "step": 3055 + }, + { + "epoch": 0.3887546113725989, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7988862991333008, + "num_tokens": 116672290.0, + "step": 3056 + }, + { + "epoch": 0.3888818216511894, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7904660105705261, + "num_tokens": 116712593.0, + "step": 3057 + }, + { + "epoch": 0.3890090319297799, + "grad_norm": 153.0, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.780631422996521, + "num_tokens": 116748663.0, + "step": 3058 + }, + { + "epoch": 0.38913624220837045, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7871902585029602, + "num_tokens": 116796018.0, + "step": 3059 + }, + { + "epoch": 0.3892634524869609, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7894811630249023, + "num_tokens": 116837287.0, + "step": 3060 + }, + { + "epoch": 0.38939066276555145, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7679541110992432, + "num_tokens": 116872163.0, + "step": 3061 + }, + { + "epoch": 0.389517873044142, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7989945411682129, + "num_tokens": 116913445.0, + "step": 3062 + }, + { + "epoch": 0.38964508332273246, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7842115163803101, + "num_tokens": 116954469.0, + "step": 3063 + }, + { + "epoch": 0.389772293601323, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7896472215652466, + "num_tokens": 116994339.0, + "step": 3064 + }, + { + "epoch": 0.3898995038799135, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.8004539012908936, + "num_tokens": 117030867.0, + "step": 3065 + }, + { + "epoch": 0.390026714158504, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.8016225695610046, + "num_tokens": 117068309.0, + "step": 3066 + }, + { + "epoch": 0.3901539244370945, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7827693819999695, + "num_tokens": 117105278.0, + "step": 3067 + }, + { + "epoch": 0.39028113471568504, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7796565890312195, + "num_tokens": 117143096.0, + "step": 3068 + }, + { + "epoch": 0.3904083449942755, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7888645529747009, + "num_tokens": 117179638.0, + "step": 3069 + }, + { + "epoch": 0.39053555527286604, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7968018054962158, + "num_tokens": 117211560.0, + "step": 3070 + }, + { + "epoch": 0.39066276555145657, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7795901894569397, + "num_tokens": 117250622.0, + "step": 3071 + }, + { + "epoch": 0.39078997583004704, + "grad_norm": 83.0, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8079532384872437, + "num_tokens": 117287600.0, + "step": 3072 + }, + { + "epoch": 0.39091718610863757, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7761461734771729, + "num_tokens": 117328188.0, + "step": 3073 + }, + { + "epoch": 0.3910443963872281, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7972943782806396, + "num_tokens": 117367282.0, + "step": 3074 + }, + { + "epoch": 0.39117160666581857, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.8000949025154114, + "num_tokens": 117403376.0, + "step": 3075 + }, + { + "epoch": 0.3912988169444091, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7746246457099915, + "num_tokens": 117438661.0, + "step": 3076 + }, + { + "epoch": 0.39142602722299963, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.794243574142456, + "num_tokens": 117477351.0, + "step": 3077 + }, + { + "epoch": 0.3915532375015901, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7830433249473572, + "num_tokens": 117512773.0, + "step": 3078 + }, + { + "epoch": 0.39168044778018063, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7793252468109131, + "num_tokens": 117555370.0, + "step": 3079 + }, + { + "epoch": 0.39180765805877116, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7971851229667664, + "num_tokens": 117594208.0, + "step": 3080 + }, + { + "epoch": 0.3919348683373617, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7927250862121582, + "num_tokens": 117632715.0, + "step": 3081 + }, + { + "epoch": 0.39206207861595216, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7819185853004456, + "num_tokens": 117669221.0, + "step": 3082 + }, + { + "epoch": 0.3921892888945427, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7833588123321533, + "num_tokens": 117709559.0, + "step": 3083 + }, + { + "epoch": 0.3923164991731332, + "grad_norm": 111.5, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7925010919570923, + "num_tokens": 117745240.0, + "step": 3084 + }, + { + "epoch": 0.3924437094517237, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7931073307991028, + "num_tokens": 117787897.0, + "step": 3085 + }, + { + "epoch": 0.3925709197303142, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7830616235733032, + "num_tokens": 117826862.0, + "step": 3086 + }, + { + "epoch": 0.39269813000890474, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7863174080848694, + "num_tokens": 117866932.0, + "step": 3087 + }, + { + "epoch": 0.3928253402874952, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7974992990493774, + "num_tokens": 117905538.0, + "step": 3088 + }, + { + "epoch": 0.39295255056608575, + "grad_norm": 180.0, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7753181457519531, + "num_tokens": 117945123.0, + "step": 3089 + }, + { + "epoch": 0.3930797608446763, + "grad_norm": 188.0, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7921634912490845, + "num_tokens": 117988743.0, + "step": 3090 + }, + { + "epoch": 0.39320697112326675, + "grad_norm": 184.0, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8036273717880249, + "num_tokens": 118026078.0, + "step": 3091 + }, + { + "epoch": 0.3933341814018573, + "grad_norm": 113.5, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7985358238220215, + "num_tokens": 118060105.0, + "step": 3092 + }, + { + "epoch": 0.3934613916804478, + "grad_norm": 97.5, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8132103681564331, + "num_tokens": 118089677.0, + "step": 3093 + }, + { + "epoch": 0.3935886019590383, + "grad_norm": 113.0, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8031812310218811, + "num_tokens": 118128449.0, + "step": 3094 + }, + { + "epoch": 0.3937158122376288, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.8040963411331177, + "num_tokens": 118168411.0, + "step": 3095 + }, + { + "epoch": 0.39384302251621933, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7998048067092896, + "num_tokens": 118206097.0, + "step": 3096 + }, + { + "epoch": 0.3939702327948098, + "grad_norm": 117.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8042581081390381, + "num_tokens": 118244429.0, + "step": 3097 + }, + { + "epoch": 0.39409744307340033, + "grad_norm": 218.0, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7898300886154175, + "num_tokens": 118279125.0, + "step": 3098 + }, + { + "epoch": 0.39422465335199086, + "grad_norm": 231.0, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.782525897026062, + "num_tokens": 118313156.0, + "step": 3099 + }, + { + "epoch": 0.39435186363058133, + "grad_norm": 253.0, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7714072465896606, + "num_tokens": 118349868.0, + "step": 3100 + }, + { + "epoch": 0.39447907390917186, + "grad_norm": 262.0, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7751457691192627, + "num_tokens": 118394998.0, + "step": 3101 + }, + { + "epoch": 0.3946062841877624, + "grad_norm": 122.5, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8031624555587769, + "num_tokens": 118436002.0, + "step": 3102 + }, + { + "epoch": 0.39473349446635286, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8007093667984009, + "num_tokens": 118475094.0, + "step": 3103 + }, + { + "epoch": 0.3948607047449434, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7890978455543518, + "num_tokens": 118514164.0, + "step": 3104 + }, + { + "epoch": 0.3949879150235339, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7845684289932251, + "num_tokens": 118554271.0, + "step": 3105 + }, + { + "epoch": 0.3951151253021244, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7906419038772583, + "num_tokens": 118597123.0, + "step": 3106 + }, + { + "epoch": 0.3952423355807149, + "grad_norm": 157.0, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7849738001823425, + "num_tokens": 118636683.0, + "step": 3107 + }, + { + "epoch": 0.39536954585930545, + "grad_norm": 158.0, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7816780805587769, + "num_tokens": 118665770.0, + "step": 3108 + }, + { + "epoch": 0.3954967561378959, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7960358262062073, + "num_tokens": 118701652.0, + "step": 3109 + }, + { + "epoch": 0.39562396641648645, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7969353795051575, + "num_tokens": 118741756.0, + "step": 3110 + }, + { + "epoch": 0.395751176695077, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7868139743804932, + "num_tokens": 118784320.0, + "step": 3111 + }, + { + "epoch": 0.39587838697366745, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7838380336761475, + "num_tokens": 118821456.0, + "step": 3112 + }, + { + "epoch": 0.396005597252258, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7890944480895996, + "num_tokens": 118861746.0, + "step": 3113 + }, + { + "epoch": 0.3961328075308485, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7930737137794495, + "num_tokens": 118900346.0, + "step": 3114 + }, + { + "epoch": 0.396260017809439, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7915695905685425, + "num_tokens": 118939707.0, + "step": 3115 + }, + { + "epoch": 0.3963872280880295, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7875658273696899, + "num_tokens": 118971154.0, + "step": 3116 + }, + { + "epoch": 0.39651443836662004, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7746834754943848, + "num_tokens": 119012344.0, + "step": 3117 + }, + { + "epoch": 0.3966416486452105, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7907745838165283, + "num_tokens": 119047677.0, + "step": 3118 + }, + { + "epoch": 0.39676885892380104, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8112229704856873, + "num_tokens": 119086357.0, + "step": 3119 + }, + { + "epoch": 0.39689606920239157, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7965413331985474, + "num_tokens": 119122854.0, + "step": 3120 + }, + { + "epoch": 0.39702327948098204, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.8007866144180298, + "num_tokens": 119155449.0, + "step": 3121 + }, + { + "epoch": 0.39715048975957257, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.786137580871582, + "num_tokens": 119197287.0, + "step": 3122 + }, + { + "epoch": 0.3972777000381631, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7692432403564453, + "num_tokens": 119235749.0, + "step": 3123 + }, + { + "epoch": 0.39740491031675357, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.782323956489563, + "num_tokens": 119267282.0, + "step": 3124 + }, + { + "epoch": 0.3975321205953441, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7844707369804382, + "num_tokens": 119304387.0, + "step": 3125 + }, + { + "epoch": 0.3976593308739346, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7797516584396362, + "num_tokens": 119341470.0, + "step": 3126 + }, + { + "epoch": 0.3977865411525251, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.776934027671814, + "num_tokens": 119381067.0, + "step": 3127 + }, + { + "epoch": 0.3979137514311156, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7883301377296448, + "num_tokens": 119415680.0, + "step": 3128 + }, + { + "epoch": 0.39804096170970615, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7803608775138855, + "num_tokens": 119454300.0, + "step": 3129 + }, + { + "epoch": 0.3981681719882967, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.803070068359375, + "num_tokens": 119485599.0, + "step": 3130 + }, + { + "epoch": 0.39829538226688715, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7761628031730652, + "num_tokens": 119525820.0, + "step": 3131 + }, + { + "epoch": 0.3984225925454777, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7986577153205872, + "num_tokens": 119568216.0, + "step": 3132 + }, + { + "epoch": 0.3985498028240682, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7771030068397522, + "num_tokens": 119602914.0, + "step": 3133 + }, + { + "epoch": 0.3986770131026587, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7847704887390137, + "num_tokens": 119637024.0, + "step": 3134 + }, + { + "epoch": 0.3988042233812492, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7812396287918091, + "num_tokens": 119679265.0, + "step": 3135 + }, + { + "epoch": 0.39893143365983974, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7815006971359253, + "num_tokens": 119716254.0, + "step": 3136 + }, + { + "epoch": 0.3990586439384302, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7850664854049683, + "num_tokens": 119761325.0, + "step": 3137 + }, + { + "epoch": 0.39918585421702074, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8156285285949707, + "num_tokens": 119795468.0, + "step": 3138 + }, + { + "epoch": 0.39931306449561127, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7813761234283447, + "num_tokens": 119833804.0, + "step": 3139 + }, + { + "epoch": 0.39944027477420174, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7867312431335449, + "num_tokens": 119870852.0, + "step": 3140 + }, + { + "epoch": 0.39956748505279227, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7902132272720337, + "num_tokens": 119909258.0, + "step": 3141 + }, + { + "epoch": 0.3996946953313828, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.778403639793396, + "num_tokens": 119940701.0, + "step": 3142 + }, + { + "epoch": 0.39982190560997327, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7846938371658325, + "num_tokens": 119976780.0, + "step": 3143 + }, + { + "epoch": 0.3999491158885638, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7856107950210571, + "num_tokens": 120017543.0, + "step": 3144 + }, + { + "epoch": 0.40007632616715433, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7822927236557007, + "num_tokens": 120050036.0, + "step": 3145 + }, + { + "epoch": 0.4002035364457448, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.8032840490341187, + "num_tokens": 120084582.0, + "step": 3146 + }, + { + "epoch": 0.40033074672433533, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7855051159858704, + "num_tokens": 120120218.0, + "step": 3147 + }, + { + "epoch": 0.40045795700292586, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8132385611534119, + "num_tokens": 120161650.0, + "step": 3148 + }, + { + "epoch": 0.40058516728151633, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7997938990592957, + "num_tokens": 120197409.0, + "step": 3149 + }, + { + "epoch": 0.40071237756010686, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7822808623313904, + "num_tokens": 120240700.0, + "step": 3150 + }, + { + "epoch": 0.4008395878386974, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7813857793807983, + "num_tokens": 120286741.0, + "step": 3151 + }, + { + "epoch": 0.40096679811728786, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7928789854049683, + "num_tokens": 120323236.0, + "step": 3152 + }, + { + "epoch": 0.4010940083958784, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7901204824447632, + "num_tokens": 120362185.0, + "step": 3153 + }, + { + "epoch": 0.4012212186744689, + "grad_norm": 101.5, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7784807682037354, + "num_tokens": 120397524.0, + "step": 3154 + }, + { + "epoch": 0.4013484289530594, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7940233945846558, + "num_tokens": 120439723.0, + "step": 3155 + }, + { + "epoch": 0.4014756392316499, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7920321226119995, + "num_tokens": 120483693.0, + "step": 3156 + }, + { + "epoch": 0.40160284951024044, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7978722453117371, + "num_tokens": 120520558.0, + "step": 3157 + }, + { + "epoch": 0.4017300597888309, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.8029342293739319, + "num_tokens": 120557901.0, + "step": 3158 + }, + { + "epoch": 0.40185727006742145, + "grad_norm": 92.0, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7998716831207275, + "num_tokens": 120593851.0, + "step": 3159 + }, + { + "epoch": 0.401984480346012, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7791825532913208, + "num_tokens": 120624533.0, + "step": 3160 + }, + { + "epoch": 0.40211169062460245, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7794806957244873, + "num_tokens": 120660820.0, + "step": 3161 + }, + { + "epoch": 0.402238900903193, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8048669099807739, + "num_tokens": 120701485.0, + "step": 3162 + }, + { + "epoch": 0.4023661111817835, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.8020550012588501, + "num_tokens": 120736677.0, + "step": 3163 + }, + { + "epoch": 0.402493321460374, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7854516506195068, + "num_tokens": 120773933.0, + "step": 3164 + }, + { + "epoch": 0.4026205317389645, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7904312610626221, + "num_tokens": 120804671.0, + "step": 3165 + }, + { + "epoch": 0.40274774201755503, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7923758029937744, + "num_tokens": 120837042.0, + "step": 3166 + }, + { + "epoch": 0.4028749522961455, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7863831520080566, + "num_tokens": 120875127.0, + "step": 3167 + }, + { + "epoch": 0.40300216257473603, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.790899932384491, + "num_tokens": 120915937.0, + "step": 3168 + }, + { + "epoch": 0.40312937285332656, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7747958302497864, + "num_tokens": 120953334.0, + "step": 3169 + }, + { + "epoch": 0.40325658313191703, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7910250425338745, + "num_tokens": 120986674.0, + "step": 3170 + }, + { + "epoch": 0.40338379341050756, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7973544001579285, + "num_tokens": 121026305.0, + "step": 3171 + }, + { + "epoch": 0.4035110036890981, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.8030310869216919, + "num_tokens": 121061644.0, + "step": 3172 + }, + { + "epoch": 0.40363821396768856, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.785963773727417, + "num_tokens": 121102860.0, + "step": 3173 + }, + { + "epoch": 0.4037654242462791, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8020306825637817, + "num_tokens": 121136566.0, + "step": 3174 + }, + { + "epoch": 0.4038926345248696, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8003935217857361, + "num_tokens": 121173759.0, + "step": 3175 + }, + { + "epoch": 0.4040198448034601, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7971005439758301, + "num_tokens": 121219881.0, + "step": 3176 + }, + { + "epoch": 0.4041470550820506, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7883601784706116, + "num_tokens": 121258719.0, + "step": 3177 + }, + { + "epoch": 0.40427426536064115, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.766211748123169, + "num_tokens": 121298013.0, + "step": 3178 + }, + { + "epoch": 0.4044014756392316, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8031743764877319, + "num_tokens": 121329615.0, + "step": 3179 + }, + { + "epoch": 0.40452868591782215, + "grad_norm": 88.0, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.8005536794662476, + "num_tokens": 121365199.0, + "step": 3180 + }, + { + "epoch": 0.4046558961964127, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7837169170379639, + "num_tokens": 121402393.0, + "step": 3181 + }, + { + "epoch": 0.4047831064750032, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7895453572273254, + "num_tokens": 121439678.0, + "step": 3182 + }, + { + "epoch": 0.4049103167535937, + "grad_norm": 90.5, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7835838794708252, + "num_tokens": 121480781.0, + "step": 3183 + }, + { + "epoch": 0.4050375270321842, + "grad_norm": 78.0, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7999624609947205, + "num_tokens": 121517614.0, + "step": 3184 + }, + { + "epoch": 0.40516473731077474, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7870550155639648, + "num_tokens": 121552225.0, + "step": 3185 + }, + { + "epoch": 0.4052919475893652, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7821093797683716, + "num_tokens": 121587177.0, + "step": 3186 + }, + { + "epoch": 0.40541915786795574, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7691856622695923, + "num_tokens": 121619547.0, + "step": 3187 + }, + { + "epoch": 0.40554636814654627, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7965155839920044, + "num_tokens": 121657396.0, + "step": 3188 + }, + { + "epoch": 0.40567357842513674, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.768704891204834, + "num_tokens": 121691867.0, + "step": 3189 + }, + { + "epoch": 0.40580078870372727, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7780214548110962, + "num_tokens": 121723546.0, + "step": 3190 + }, + { + "epoch": 0.4059279989823178, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7930960655212402, + "num_tokens": 121764589.0, + "step": 3191 + }, + { + "epoch": 0.40605520926090827, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7760639190673828, + "num_tokens": 121803281.0, + "step": 3192 + }, + { + "epoch": 0.4061824195394988, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7776329517364502, + "num_tokens": 121841710.0, + "step": 3193 + }, + { + "epoch": 0.4063096298180893, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7874784469604492, + "num_tokens": 121881979.0, + "step": 3194 + }, + { + "epoch": 0.4064368400966798, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.794508159160614, + "num_tokens": 121921599.0, + "step": 3195 + }, + { + "epoch": 0.4065640503752703, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7977545261383057, + "num_tokens": 121960545.0, + "step": 3196 + }, + { + "epoch": 0.40669126065386085, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7711424827575684, + "num_tokens": 121999259.0, + "step": 3197 + }, + { + "epoch": 0.4068184709324513, + "grad_norm": 103.0, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.806810200214386, + "num_tokens": 122039066.0, + "step": 3198 + }, + { + "epoch": 0.40694568121104185, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8067935705184937, + "num_tokens": 122085035.0, + "step": 3199 + }, + { + "epoch": 0.4070728914896324, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8061461448669434, + "num_tokens": 122128007.0, + "step": 3200 + }, + { + "epoch": 0.40720010176822286, + "grad_norm": 90.0, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7872344255447388, + "num_tokens": 122169623.0, + "step": 3201 + }, + { + "epoch": 0.4073273120468134, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7854787707328796, + "num_tokens": 122201755.0, + "step": 3202 + }, + { + "epoch": 0.4074545223254039, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7974919080734253, + "num_tokens": 122233680.0, + "step": 3203 + }, + { + "epoch": 0.4075817326039944, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7822871208190918, + "num_tokens": 122265107.0, + "step": 3204 + }, + { + "epoch": 0.4077089428825849, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8030869960784912, + "num_tokens": 122302750.0, + "step": 3205 + }, + { + "epoch": 0.40783615316117544, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7836999893188477, + "num_tokens": 122346820.0, + "step": 3206 + }, + { + "epoch": 0.4079633634397659, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7904661893844604, + "num_tokens": 122383254.0, + "step": 3207 + }, + { + "epoch": 0.40809057371835644, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.8007309436798096, + "num_tokens": 122416693.0, + "step": 3208 + }, + { + "epoch": 0.40821778399694697, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7945408821105957, + "num_tokens": 122452403.0, + "step": 3209 + }, + { + "epoch": 0.40834499427553744, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7771684527397156, + "num_tokens": 122489685.0, + "step": 3210 + }, + { + "epoch": 0.40847220455412797, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7704957723617554, + "num_tokens": 122529079.0, + "step": 3211 + }, + { + "epoch": 0.4085994148327185, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7805672883987427, + "num_tokens": 122569933.0, + "step": 3212 + }, + { + "epoch": 0.40872662511130897, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7715722918510437, + "num_tokens": 122607884.0, + "step": 3213 + }, + { + "epoch": 0.4088538353898995, + "grad_norm": 131.0, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7904877662658691, + "num_tokens": 122651968.0, + "step": 3214 + }, + { + "epoch": 0.40898104566849003, + "grad_norm": 126.0, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.794266939163208, + "num_tokens": 122688680.0, + "step": 3215 + }, + { + "epoch": 0.4091082559470805, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7959817051887512, + "num_tokens": 122728271.0, + "step": 3216 + }, + { + "epoch": 0.40923546622567103, + "grad_norm": 212.0, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7870316505432129, + "num_tokens": 122759691.0, + "step": 3217 + }, + { + "epoch": 0.40936267650426156, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7856738567352295, + "num_tokens": 122800185.0, + "step": 3218 + }, + { + "epoch": 0.40948988678285203, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7999613285064697, + "num_tokens": 122839548.0, + "step": 3219 + }, + { + "epoch": 0.40961709706144256, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7740712761878967, + "num_tokens": 122879621.0, + "step": 3220 + }, + { + "epoch": 0.4097443073400331, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7820439338684082, + "num_tokens": 122921495.0, + "step": 3221 + }, + { + "epoch": 0.40987151761862356, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7919206619262695, + "num_tokens": 122957197.0, + "step": 3222 + }, + { + "epoch": 0.4099987278972141, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7685883045196533, + "num_tokens": 122999999.0, + "step": 3223 + }, + { + "epoch": 0.4101259381758046, + "grad_norm": 158.0, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7949930429458618, + "num_tokens": 123040402.0, + "step": 3224 + }, + { + "epoch": 0.4102531484543951, + "grad_norm": 128.0, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7930870056152344, + "num_tokens": 123080086.0, + "step": 3225 + }, + { + "epoch": 0.4103803587329856, + "grad_norm": 98.5, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7933473587036133, + "num_tokens": 123120058.0, + "step": 3226 + }, + { + "epoch": 0.41050756901157615, + "grad_norm": 110.5, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7893496751785278, + "num_tokens": 123162003.0, + "step": 3227 + }, + { + "epoch": 0.4106347792901666, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7600705027580261, + "num_tokens": 123202249.0, + "step": 3228 + }, + { + "epoch": 0.41076198956875715, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7793044447898865, + "num_tokens": 123237401.0, + "step": 3229 + }, + { + "epoch": 0.4108891998473477, + "grad_norm": 193.0, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.8057039976119995, + "num_tokens": 123269364.0, + "step": 3230 + }, + { + "epoch": 0.4110164101259382, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7942606806755066, + "num_tokens": 123307743.0, + "step": 3231 + }, + { + "epoch": 0.4111436204045287, + "grad_norm": 118.0, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7935967445373535, + "num_tokens": 123340102.0, + "step": 3232 + }, + { + "epoch": 0.4112708306831192, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7825790047645569, + "num_tokens": 123374956.0, + "step": 3233 + }, + { + "epoch": 0.41139804096170973, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.8022997379302979, + "num_tokens": 123418070.0, + "step": 3234 + }, + { + "epoch": 0.4115252512403002, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.785887598991394, + "num_tokens": 123453183.0, + "step": 3235 + }, + { + "epoch": 0.41165246151889073, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7882630825042725, + "num_tokens": 123491526.0, + "step": 3236 + }, + { + "epoch": 0.41177967179748126, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7924674153327942, + "num_tokens": 123529743.0, + "step": 3237 + }, + { + "epoch": 0.41190688207607173, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7930798530578613, + "num_tokens": 123573402.0, + "step": 3238 + }, + { + "epoch": 0.41203409235466226, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7811905145645142, + "num_tokens": 123604167.0, + "step": 3239 + }, + { + "epoch": 0.4121613026332528, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7915149927139282, + "num_tokens": 123648104.0, + "step": 3240 + }, + { + "epoch": 0.41228851291184326, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7793704867362976, + "num_tokens": 123689678.0, + "step": 3241 + }, + { + "epoch": 0.4124157231904338, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.8004764318466187, + "num_tokens": 123729257.0, + "step": 3242 + }, + { + "epoch": 0.4125429334690243, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7934310436248779, + "num_tokens": 123772251.0, + "step": 3243 + }, + { + "epoch": 0.4126701437476148, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8041579723358154, + "num_tokens": 123810345.0, + "step": 3244 + }, + { + "epoch": 0.4127973540262053, + "grad_norm": 124.0, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7942443490028381, + "num_tokens": 123855276.0, + "step": 3245 + }, + { + "epoch": 0.41292456430479585, + "grad_norm": 233.0, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7923170924186707, + "num_tokens": 123896613.0, + "step": 3246 + }, + { + "epoch": 0.4130517745833863, + "grad_norm": 203.0, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7716165781021118, + "num_tokens": 123940713.0, + "step": 3247 + }, + { + "epoch": 0.41317898486197685, + "grad_norm": 132.0, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7793999910354614, + "num_tokens": 123981485.0, + "step": 3248 + }, + { + "epoch": 0.4133061951405674, + "grad_norm": 274.0, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7799553275108337, + "num_tokens": 124018159.0, + "step": 3249 + }, + { + "epoch": 0.41343340541915785, + "grad_norm": 182.0, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.79647296667099, + "num_tokens": 124049911.0, + "step": 3250 + }, + { + "epoch": 0.4135606156977484, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.806678295135498, + "num_tokens": 124084260.0, + "step": 3251 + }, + { + "epoch": 0.4136878259763389, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7925227880477905, + "num_tokens": 124122741.0, + "step": 3252 + }, + { + "epoch": 0.4138150362549294, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7793607711791992, + "num_tokens": 124162793.0, + "step": 3253 + }, + { + "epoch": 0.4139422465335199, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7856712341308594, + "num_tokens": 124206889.0, + "step": 3254 + }, + { + "epoch": 0.41406945681211044, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7998952269554138, + "num_tokens": 124244551.0, + "step": 3255 + }, + { + "epoch": 0.4141966670907009, + "grad_norm": 95.5, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7962161302566528, + "num_tokens": 124290485.0, + "step": 3256 + }, + { + "epoch": 0.41432387736929144, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7915222644805908, + "num_tokens": 124323763.0, + "step": 3257 + }, + { + "epoch": 0.41445108764788197, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7942291498184204, + "num_tokens": 124363318.0, + "step": 3258 + }, + { + "epoch": 0.41457829792647244, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8157055974006653, + "num_tokens": 124400845.0, + "step": 3259 + }, + { + "epoch": 0.41470550820506297, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7971335053443909, + "num_tokens": 124442546.0, + "step": 3260 + }, + { + "epoch": 0.4148327184836535, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7850740551948547, + "num_tokens": 124477737.0, + "step": 3261 + }, + { + "epoch": 0.41495992876224397, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7768257260322571, + "num_tokens": 124518081.0, + "step": 3262 + }, + { + "epoch": 0.4150871390408345, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7818183302879333, + "num_tokens": 124556361.0, + "step": 3263 + }, + { + "epoch": 0.415214349319425, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7838228344917297, + "num_tokens": 124588490.0, + "step": 3264 + }, + { + "epoch": 0.4153415595980155, + "grad_norm": 110.0, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7837643623352051, + "num_tokens": 124624656.0, + "step": 3265 + }, + { + "epoch": 0.415468769876606, + "grad_norm": 165.0, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7709113359451294, + "num_tokens": 124660681.0, + "step": 3266 + }, + { + "epoch": 0.41559598015519655, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7915987968444824, + "num_tokens": 124703448.0, + "step": 3267 + }, + { + "epoch": 0.415723190433787, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7928125262260437, + "num_tokens": 124742078.0, + "step": 3268 + }, + { + "epoch": 0.41585040071237755, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.77727210521698, + "num_tokens": 124779894.0, + "step": 3269 + }, + { + "epoch": 0.4159776109909681, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7926934957504272, + "num_tokens": 124817918.0, + "step": 3270 + }, + { + "epoch": 0.41610482126955856, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8076503276824951, + "num_tokens": 124854839.0, + "step": 3271 + }, + { + "epoch": 0.4162320315481491, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7713432312011719, + "num_tokens": 124886779.0, + "step": 3272 + }, + { + "epoch": 0.4163592418267396, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.8003815412521362, + "num_tokens": 124927919.0, + "step": 3273 + }, + { + "epoch": 0.4164864521053301, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7899537086486816, + "num_tokens": 124966475.0, + "step": 3274 + }, + { + "epoch": 0.4166136623839206, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7657718658447266, + "num_tokens": 125004110.0, + "step": 3275 + }, + { + "epoch": 0.41674087266251114, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8096168041229248, + "num_tokens": 125043164.0, + "step": 3276 + }, + { + "epoch": 0.4168680829411016, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7931304574012756, + "num_tokens": 125084255.0, + "step": 3277 + }, + { + "epoch": 0.41699529321969214, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8035836219787598, + "num_tokens": 125124971.0, + "step": 3278 + }, + { + "epoch": 0.41712250349828267, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7834782600402832, + "num_tokens": 125161470.0, + "step": 3279 + }, + { + "epoch": 0.4172497137768732, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7774465680122375, + "num_tokens": 125199117.0, + "step": 3280 + }, + { + "epoch": 0.41737692405546367, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7746207118034363, + "num_tokens": 125238136.0, + "step": 3281 + }, + { + "epoch": 0.4175041343340542, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7908539175987244, + "num_tokens": 125281041.0, + "step": 3282 + }, + { + "epoch": 0.41763134461264473, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7900843024253845, + "num_tokens": 125314739.0, + "step": 3283 + }, + { + "epoch": 0.4177585548912352, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7811876535415649, + "num_tokens": 125353129.0, + "step": 3284 + }, + { + "epoch": 0.41788576516982573, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7906612753868103, + "num_tokens": 125394092.0, + "step": 3285 + }, + { + "epoch": 0.41801297544841626, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7829221487045288, + "num_tokens": 125438262.0, + "step": 3286 + }, + { + "epoch": 0.41814018572700673, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7856602668762207, + "num_tokens": 125477842.0, + "step": 3287 + }, + { + "epoch": 0.41826739600559726, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8026759624481201, + "num_tokens": 125517236.0, + "step": 3288 + }, + { + "epoch": 0.4183946062841878, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7885145545005798, + "num_tokens": 125561598.0, + "step": 3289 + }, + { + "epoch": 0.41852181656277826, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7859117388725281, + "num_tokens": 125599104.0, + "step": 3290 + }, + { + "epoch": 0.4186490268413688, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7849603891372681, + "num_tokens": 125635286.0, + "step": 3291 + }, + { + "epoch": 0.4187762371199593, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7930588722229004, + "num_tokens": 125669534.0, + "step": 3292 + }, + { + "epoch": 0.4189034473985498, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7904993295669556, + "num_tokens": 125707966.0, + "step": 3293 + }, + { + "epoch": 0.4190306576771403, + "grad_norm": 153.0, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7750605344772339, + "num_tokens": 125741448.0, + "step": 3294 + }, + { + "epoch": 0.41915786795573085, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7819064259529114, + "num_tokens": 125776887.0, + "step": 3295 + }, + { + "epoch": 0.4192850782343213, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7918010354042053, + "num_tokens": 125807899.0, + "step": 3296 + }, + { + "epoch": 0.41941228851291185, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.8004089593887329, + "num_tokens": 125847711.0, + "step": 3297 + }, + { + "epoch": 0.4195394987915024, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8043025135993958, + "num_tokens": 125883248.0, + "step": 3298 + }, + { + "epoch": 0.41966670907009285, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7813810706138611, + "num_tokens": 125925771.0, + "step": 3299 + }, + { + "epoch": 0.4197939193486834, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7756738066673279, + "num_tokens": 125968815.0, + "step": 3300 + }, + { + "epoch": 0.4199211296272739, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7849217653274536, + "num_tokens": 126007761.0, + "step": 3301 + }, + { + "epoch": 0.4200483399058644, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7721570134162903, + "num_tokens": 126047872.0, + "step": 3302 + }, + { + "epoch": 0.4201755501844549, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7919138669967651, + "num_tokens": 126081273.0, + "step": 3303 + }, + { + "epoch": 0.42030276046304543, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7748632431030273, + "num_tokens": 126121590.0, + "step": 3304 + }, + { + "epoch": 0.4204299707416359, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7739707231521606, + "num_tokens": 126168396.0, + "step": 3305 + }, + { + "epoch": 0.42055718102022643, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7864944934844971, + "num_tokens": 126205220.0, + "step": 3306 + }, + { + "epoch": 0.42068439129881696, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7786831855773926, + "num_tokens": 126239131.0, + "step": 3307 + }, + { + "epoch": 0.42081160157740743, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7860284447669983, + "num_tokens": 126271931.0, + "step": 3308 + }, + { + "epoch": 0.42093881185599796, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.8007174134254456, + "num_tokens": 126307730.0, + "step": 3309 + }, + { + "epoch": 0.4210660221345885, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7927265167236328, + "num_tokens": 126344528.0, + "step": 3310 + }, + { + "epoch": 0.42119323241317896, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7937573194503784, + "num_tokens": 126377260.0, + "step": 3311 + }, + { + "epoch": 0.4213204426917695, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7827959656715393, + "num_tokens": 126409723.0, + "step": 3312 + }, + { + "epoch": 0.42144765297036, + "grad_norm": 224.0, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7893490791320801, + "num_tokens": 126450929.0, + "step": 3313 + }, + { + "epoch": 0.4215748632489505, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7983072400093079, + "num_tokens": 126486878.0, + "step": 3314 + }, + { + "epoch": 0.421702073527541, + "grad_norm": 133.0, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7771259546279907, + "num_tokens": 126522884.0, + "step": 3315 + }, + { + "epoch": 0.42182928380613155, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7922496795654297, + "num_tokens": 126560848.0, + "step": 3316 + }, + { + "epoch": 0.421956494084722, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7871072292327881, + "num_tokens": 126592947.0, + "step": 3317 + }, + { + "epoch": 0.42208370436331255, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8040961027145386, + "num_tokens": 126630631.0, + "step": 3318 + }, + { + "epoch": 0.4222109146419031, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7884044647216797, + "num_tokens": 126673395.0, + "step": 3319 + }, + { + "epoch": 0.42233812492049355, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8132466077804565, + "num_tokens": 126711504.0, + "step": 3320 + }, + { + "epoch": 0.4224653351990841, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8039427995681763, + "num_tokens": 126753217.0, + "step": 3321 + }, + { + "epoch": 0.4225925454776746, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7992243766784668, + "num_tokens": 126792982.0, + "step": 3322 + }, + { + "epoch": 0.4227197557562651, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7802867889404297, + "num_tokens": 126823771.0, + "step": 3323 + }, + { + "epoch": 0.4228469660348556, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7972307205200195, + "num_tokens": 126867137.0, + "step": 3324 + }, + { + "epoch": 0.42297417631344614, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.797187089920044, + "num_tokens": 126902680.0, + "step": 3325 + }, + { + "epoch": 0.4231013865920366, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8076978921890259, + "num_tokens": 126941037.0, + "step": 3326 + }, + { + "epoch": 0.42322859687062714, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7912821769714355, + "num_tokens": 126980009.0, + "step": 3327 + }, + { + "epoch": 0.42335580714921767, + "grad_norm": 98.5, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8085907101631165, + "num_tokens": 127020022.0, + "step": 3328 + }, + { + "epoch": 0.42348301742780814, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7690408825874329, + "num_tokens": 127060500.0, + "step": 3329 + }, + { + "epoch": 0.42361022770639867, + "grad_norm": 84.0, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7836247086524963, + "num_tokens": 127095886.0, + "step": 3330 + }, + { + "epoch": 0.4237374379849892, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7786206007003784, + "num_tokens": 127134334.0, + "step": 3331 + }, + { + "epoch": 0.4238646482635797, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7818085551261902, + "num_tokens": 127174288.0, + "step": 3332 + }, + { + "epoch": 0.4239918585421702, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7870643138885498, + "num_tokens": 127210311.0, + "step": 3333 + }, + { + "epoch": 0.4241190688207607, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8148213624954224, + "num_tokens": 127252813.0, + "step": 3334 + }, + { + "epoch": 0.42424627909935125, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.776432991027832, + "num_tokens": 127285847.0, + "step": 3335 + }, + { + "epoch": 0.4243734893779417, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8060430288314819, + "num_tokens": 127325360.0, + "step": 3336 + }, + { + "epoch": 0.42450069965653225, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7959661483764648, + "num_tokens": 127366446.0, + "step": 3337 + }, + { + "epoch": 0.4246279099351228, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7998331785202026, + "num_tokens": 127402608.0, + "step": 3338 + }, + { + "epoch": 0.42475512021371326, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7999039888381958, + "num_tokens": 127445355.0, + "step": 3339 + }, + { + "epoch": 0.4248823304923038, + "grad_norm": 147.0, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7835235595703125, + "num_tokens": 127480523.0, + "step": 3340 + }, + { + "epoch": 0.4250095407708943, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8181522488594055, + "num_tokens": 127520787.0, + "step": 3341 + }, + { + "epoch": 0.4251367510494848, + "grad_norm": 110.0, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7865715026855469, + "num_tokens": 127565576.0, + "step": 3342 + }, + { + "epoch": 0.4252639613280753, + "grad_norm": 119.5, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7942029237747192, + "num_tokens": 127599144.0, + "step": 3343 + }, + { + "epoch": 0.42539117160666584, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.81015944480896, + "num_tokens": 127637361.0, + "step": 3344 + }, + { + "epoch": 0.4255183818852563, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7781466245651245, + "num_tokens": 127675139.0, + "step": 3345 + }, + { + "epoch": 0.42564559216384684, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.8011469841003418, + "num_tokens": 127718822.0, + "step": 3346 + }, + { + "epoch": 0.42577280244243737, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.796898603439331, + "num_tokens": 127763332.0, + "step": 3347 + }, + { + "epoch": 0.42590001272102784, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8033241033554077, + "num_tokens": 127800346.0, + "step": 3348 + }, + { + "epoch": 0.42602722299961837, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7911185026168823, + "num_tokens": 127845142.0, + "step": 3349 + }, + { + "epoch": 0.4261544332782089, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7766735553741455, + "num_tokens": 127881131.0, + "step": 3350 + }, + { + "epoch": 0.4262816435567994, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8016271591186523, + "num_tokens": 127917941.0, + "step": 3351 + }, + { + "epoch": 0.4264088538353899, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7785207033157349, + "num_tokens": 127963345.0, + "step": 3352 + }, + { + "epoch": 0.42653606411398043, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.799685001373291, + "num_tokens": 128004839.0, + "step": 3353 + }, + { + "epoch": 0.4266632743925709, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7849308252334595, + "num_tokens": 128042709.0, + "step": 3354 + }, + { + "epoch": 0.42679048467116143, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7741828560829163, + "num_tokens": 128085069.0, + "step": 3355 + }, + { + "epoch": 0.42691769494975196, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7890173196792603, + "num_tokens": 128124300.0, + "step": 3356 + }, + { + "epoch": 0.42704490522834243, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.793455183506012, + "num_tokens": 128164410.0, + "step": 3357 + }, + { + "epoch": 0.42717211550693296, + "grad_norm": 119.5, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.8037028312683105, + "num_tokens": 128206524.0, + "step": 3358 + }, + { + "epoch": 0.4272993257855235, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7807478904724121, + "num_tokens": 128246387.0, + "step": 3359 + }, + { + "epoch": 0.42742653606411396, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8066891431808472, + "num_tokens": 128287794.0, + "step": 3360 + }, + { + "epoch": 0.4275537463427045, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7907378077507019, + "num_tokens": 128328467.0, + "step": 3361 + }, + { + "epoch": 0.427680956621295, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7983666658401489, + "num_tokens": 128364319.0, + "step": 3362 + }, + { + "epoch": 0.4278081668998855, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7843154072761536, + "num_tokens": 128403892.0, + "step": 3363 + }, + { + "epoch": 0.427935377178476, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7701634168624878, + "num_tokens": 128446760.0, + "step": 3364 + }, + { + "epoch": 0.42806258745706655, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7638249397277832, + "num_tokens": 128485498.0, + "step": 3365 + }, + { + "epoch": 0.428189797735657, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7859214544296265, + "num_tokens": 128527871.0, + "step": 3366 + }, + { + "epoch": 0.42831700801424755, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7803696393966675, + "num_tokens": 128567792.0, + "step": 3367 + }, + { + "epoch": 0.4284442182928381, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.782405436038971, + "num_tokens": 128607217.0, + "step": 3368 + }, + { + "epoch": 0.42857142857142855, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7924306392669678, + "num_tokens": 128642787.0, + "step": 3369 + }, + { + "epoch": 0.4286986388500191, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8035264611244202, + "num_tokens": 128680462.0, + "step": 3370 + }, + { + "epoch": 0.4288258491286096, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7854031324386597, + "num_tokens": 128715101.0, + "step": 3371 + }, + { + "epoch": 0.4289530594072001, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7802280187606812, + "num_tokens": 128749601.0, + "step": 3372 + }, + { + "epoch": 0.4290802696857906, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7902661561965942, + "num_tokens": 128788744.0, + "step": 3373 + }, + { + "epoch": 0.42920747996438113, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7907504439353943, + "num_tokens": 128828102.0, + "step": 3374 + }, + { + "epoch": 0.4293346902429716, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7861785888671875, + "num_tokens": 128863610.0, + "step": 3375 + }, + { + "epoch": 0.42946190052156213, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8070162534713745, + "num_tokens": 128898836.0, + "step": 3376 + }, + { + "epoch": 0.42958911080015266, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7913150787353516, + "num_tokens": 128934821.0, + "step": 3377 + }, + { + "epoch": 0.42971632107874314, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7823284864425659, + "num_tokens": 128966914.0, + "step": 3378 + }, + { + "epoch": 0.42984353135733366, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7971968650817871, + "num_tokens": 129008305.0, + "step": 3379 + }, + { + "epoch": 0.4299707416359242, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8135688304901123, + "num_tokens": 129043331.0, + "step": 3380 + }, + { + "epoch": 0.4300979519145147, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7975943684577942, + "num_tokens": 129083513.0, + "step": 3381 + }, + { + "epoch": 0.4302251621931052, + "grad_norm": 78.5, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7838809490203857, + "num_tokens": 129113463.0, + "step": 3382 + }, + { + "epoch": 0.4303523724716957, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7626596689224243, + "num_tokens": 129153080.0, + "step": 3383 + }, + { + "epoch": 0.43047958275028625, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.789935290813446, + "num_tokens": 129193338.0, + "step": 3384 + }, + { + "epoch": 0.4306067930288767, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.78168123960495, + "num_tokens": 129234816.0, + "step": 3385 + }, + { + "epoch": 0.43073400330746725, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7985148429870605, + "num_tokens": 129275228.0, + "step": 3386 + }, + { + "epoch": 0.4308612135860578, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7724206447601318, + "num_tokens": 129309747.0, + "step": 3387 + }, + { + "epoch": 0.43098842386464825, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.79367995262146, + "num_tokens": 129342580.0, + "step": 3388 + }, + { + "epoch": 0.4311156341432388, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7873635292053223, + "num_tokens": 129376599.0, + "step": 3389 + }, + { + "epoch": 0.4312428444218293, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7880998849868774, + "num_tokens": 129414783.0, + "step": 3390 + }, + { + "epoch": 0.4313700547004198, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7870064377784729, + "num_tokens": 129450287.0, + "step": 3391 + }, + { + "epoch": 0.4314972649790103, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7740549445152283, + "num_tokens": 129486733.0, + "step": 3392 + }, + { + "epoch": 0.43162447525760084, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7924196720123291, + "num_tokens": 129524392.0, + "step": 3393 + }, + { + "epoch": 0.4317516855361913, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7699548006057739, + "num_tokens": 129558452.0, + "step": 3394 + }, + { + "epoch": 0.43187889581478184, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7875810265541077, + "num_tokens": 129597076.0, + "step": 3395 + }, + { + "epoch": 0.43200610609337237, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7791842222213745, + "num_tokens": 129635861.0, + "step": 3396 + }, + { + "epoch": 0.43213331637196284, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7889301180839539, + "num_tokens": 129673760.0, + "step": 3397 + }, + { + "epoch": 0.43226052665055337, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7926629781723022, + "num_tokens": 129710730.0, + "step": 3398 + }, + { + "epoch": 0.4323877369291439, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7829735279083252, + "num_tokens": 129748724.0, + "step": 3399 + }, + { + "epoch": 0.43251494720773437, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8029599785804749, + "num_tokens": 129787478.0, + "step": 3400 + }, + { + "epoch": 0.4326421574863249, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.8005102872848511, + "num_tokens": 129829852.0, + "step": 3401 + }, + { + "epoch": 0.4327693677649154, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.790742814540863, + "num_tokens": 129869279.0, + "step": 3402 + }, + { + "epoch": 0.4328965780435059, + "grad_norm": 118.0, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7824910879135132, + "num_tokens": 129903347.0, + "step": 3403 + }, + { + "epoch": 0.4330237883220964, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.8003942370414734, + "num_tokens": 129940028.0, + "step": 3404 + }, + { + "epoch": 0.43315099860068695, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7939203977584839, + "num_tokens": 129976438.0, + "step": 3405 + }, + { + "epoch": 0.4332782088792774, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7914717197418213, + "num_tokens": 130015698.0, + "step": 3406 + }, + { + "epoch": 0.43340541915786795, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8056332468986511, + "num_tokens": 130056000.0, + "step": 3407 + }, + { + "epoch": 0.4335326294364585, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7851417064666748, + "num_tokens": 130096990.0, + "step": 3408 + }, + { + "epoch": 0.43365983971504896, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7974361777305603, + "num_tokens": 130128556.0, + "step": 3409 + }, + { + "epoch": 0.4337870499936395, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8016019463539124, + "num_tokens": 130170607.0, + "step": 3410 + }, + { + "epoch": 0.43391426027223, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7881797552108765, + "num_tokens": 130212511.0, + "step": 3411 + }, + { + "epoch": 0.4340414705508205, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7868647575378418, + "num_tokens": 130249246.0, + "step": 3412 + }, + { + "epoch": 0.434168680829411, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8136347532272339, + "num_tokens": 130291250.0, + "step": 3413 + }, + { + "epoch": 0.43429589110800154, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7678719758987427, + "num_tokens": 130327094.0, + "step": 3414 + }, + { + "epoch": 0.434423101386592, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7802309989929199, + "num_tokens": 130366482.0, + "step": 3415 + }, + { + "epoch": 0.43455031166518254, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7929430603981018, + "num_tokens": 130407416.0, + "step": 3416 + }, + { + "epoch": 0.43467752194377307, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7959014177322388, + "num_tokens": 130447125.0, + "step": 3417 + }, + { + "epoch": 0.43480473222236354, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7974874973297119, + "num_tokens": 130480955.0, + "step": 3418 + }, + { + "epoch": 0.43493194250095407, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8033693432807922, + "num_tokens": 130511771.0, + "step": 3419 + }, + { + "epoch": 0.4350591527795446, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7900379300117493, + "num_tokens": 130550080.0, + "step": 3420 + }, + { + "epoch": 0.4351863630581351, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7740051746368408, + "num_tokens": 130588109.0, + "step": 3421 + }, + { + "epoch": 0.4353135733367256, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7748485803604126, + "num_tokens": 130626832.0, + "step": 3422 + }, + { + "epoch": 0.43544078361531613, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7927449345588684, + "num_tokens": 130662202.0, + "step": 3423 + }, + { + "epoch": 0.4355679938939066, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7906582355499268, + "num_tokens": 130700747.0, + "step": 3424 + }, + { + "epoch": 0.43569520417249713, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7912272214889526, + "num_tokens": 130742849.0, + "step": 3425 + }, + { + "epoch": 0.43582241445108766, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8072806000709534, + "num_tokens": 130780156.0, + "step": 3426 + }, + { + "epoch": 0.43594962472967813, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.792463481426239, + "num_tokens": 130825824.0, + "step": 3427 + }, + { + "epoch": 0.43607683500826866, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.798616886138916, + "num_tokens": 130863603.0, + "step": 3428 + }, + { + "epoch": 0.4362040452868592, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7938515543937683, + "num_tokens": 130906689.0, + "step": 3429 + }, + { + "epoch": 0.4363312555654497, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7926090955734253, + "num_tokens": 130947785.0, + "step": 3430 + }, + { + "epoch": 0.4364584658440402, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.79897141456604, + "num_tokens": 130985489.0, + "step": 3431 + }, + { + "epoch": 0.4365856761226307, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7991193532943726, + "num_tokens": 131025274.0, + "step": 3432 + }, + { + "epoch": 0.43671288640122125, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7891495823860168, + "num_tokens": 131060132.0, + "step": 3433 + }, + { + "epoch": 0.4368400966798117, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7828025221824646, + "num_tokens": 131105109.0, + "step": 3434 + }, + { + "epoch": 0.43696730695840225, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.8062256574630737, + "num_tokens": 131142094.0, + "step": 3435 + }, + { + "epoch": 0.4370945172369928, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7901265025138855, + "num_tokens": 131174648.0, + "step": 3436 + }, + { + "epoch": 0.43722172751558325, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8086241483688354, + "num_tokens": 131213447.0, + "step": 3437 + }, + { + "epoch": 0.4373489377941738, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7882809042930603, + "num_tokens": 131249622.0, + "step": 3438 + }, + { + "epoch": 0.4374761480727643, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8088167309761047, + "num_tokens": 131285883.0, + "step": 3439 + }, + { + "epoch": 0.4376033583513548, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7943748831748962, + "num_tokens": 131322774.0, + "step": 3440 + }, + { + "epoch": 0.4377305686299453, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7988075017929077, + "num_tokens": 131364513.0, + "step": 3441 + }, + { + "epoch": 0.43785777890853583, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7864394187927246, + "num_tokens": 131402299.0, + "step": 3442 + }, + { + "epoch": 0.4379849891871263, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7970622181892395, + "num_tokens": 131447830.0, + "step": 3443 + }, + { + "epoch": 0.43811219946571683, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7815634608268738, + "num_tokens": 131481058.0, + "step": 3444 + }, + { + "epoch": 0.43823940974430736, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7839250564575195, + "num_tokens": 131518376.0, + "step": 3445 + }, + { + "epoch": 0.43836662002289783, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7736843824386597, + "num_tokens": 131553702.0, + "step": 3446 + }, + { + "epoch": 0.43849383030148836, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7922765016555786, + "num_tokens": 131587026.0, + "step": 3447 + }, + { + "epoch": 0.4386210405800789, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7993956804275513, + "num_tokens": 131621145.0, + "step": 3448 + }, + { + "epoch": 0.43874825085866936, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7967668771743774, + "num_tokens": 131661618.0, + "step": 3449 + }, + { + "epoch": 0.4388754611372599, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7832828760147095, + "num_tokens": 131693840.0, + "step": 3450 + }, + { + "epoch": 0.4390026714158504, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7936379909515381, + "num_tokens": 131730856.0, + "step": 3451 + }, + { + "epoch": 0.4391298816944409, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7718539237976074, + "num_tokens": 131769172.0, + "step": 3452 + }, + { + "epoch": 0.4392570919730314, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7950905561447144, + "num_tokens": 131810481.0, + "step": 3453 + }, + { + "epoch": 0.43938430225162195, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7936182022094727, + "num_tokens": 131850742.0, + "step": 3454 + }, + { + "epoch": 0.4395115125302124, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7886383533477783, + "num_tokens": 131886675.0, + "step": 3455 + }, + { + "epoch": 0.43963872280880295, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8042221069335938, + "num_tokens": 131926831.0, + "step": 3456 + }, + { + "epoch": 0.4397659330873935, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7925516963005066, + "num_tokens": 131962265.0, + "step": 3457 + }, + { + "epoch": 0.43989314336598395, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7893237471580505, + "num_tokens": 131999906.0, + "step": 3458 + }, + { + "epoch": 0.4400203536445745, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8045833110809326, + "num_tokens": 132031330.0, + "step": 3459 + }, + { + "epoch": 0.440147563923165, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7872250080108643, + "num_tokens": 132065049.0, + "step": 3460 + }, + { + "epoch": 0.4402747742017555, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7951617240905762, + "num_tokens": 132100993.0, + "step": 3461 + }, + { + "epoch": 0.440401984480346, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7901013493537903, + "num_tokens": 132138169.0, + "step": 3462 + }, + { + "epoch": 0.44052919475893654, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7954682111740112, + "num_tokens": 132170498.0, + "step": 3463 + }, + { + "epoch": 0.440656405037527, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.776519775390625, + "num_tokens": 132213583.0, + "step": 3464 + }, + { + "epoch": 0.44078361531611754, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7905203104019165, + "num_tokens": 132255427.0, + "step": 3465 + }, + { + "epoch": 0.44091082559470807, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7881728410720825, + "num_tokens": 132295795.0, + "step": 3466 + }, + { + "epoch": 0.44103803587329854, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7756052613258362, + "num_tokens": 132336881.0, + "step": 3467 + }, + { + "epoch": 0.44116524615188907, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7876583933830261, + "num_tokens": 132375136.0, + "step": 3468 + }, + { + "epoch": 0.4412924564304796, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7900568842887878, + "num_tokens": 132412893.0, + "step": 3469 + }, + { + "epoch": 0.44141966670907007, + "grad_norm": 160.0, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7744195461273193, + "num_tokens": 132444360.0, + "step": 3470 + }, + { + "epoch": 0.4415468769876606, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7618093490600586, + "num_tokens": 132485022.0, + "step": 3471 + }, + { + "epoch": 0.4416740872662511, + "grad_norm": 198.0, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7696568965911865, + "num_tokens": 132522800.0, + "step": 3472 + }, + { + "epoch": 0.4418012975448416, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7935059070587158, + "num_tokens": 132560527.0, + "step": 3473 + }, + { + "epoch": 0.4419285078234321, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7871055603027344, + "num_tokens": 132601075.0, + "step": 3474 + }, + { + "epoch": 0.44205571810202265, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7872806787490845, + "num_tokens": 132639851.0, + "step": 3475 + }, + { + "epoch": 0.4421829283806131, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7940990328788757, + "num_tokens": 132679849.0, + "step": 3476 + }, + { + "epoch": 0.44231013865920366, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.794504702091217, + "num_tokens": 132721020.0, + "step": 3477 + }, + { + "epoch": 0.4424373489377942, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7779122591018677, + "num_tokens": 132760933.0, + "step": 3478 + }, + { + "epoch": 0.44256455921638466, + "grad_norm": 87.0, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.8029611706733704, + "num_tokens": 132800627.0, + "step": 3479 + }, + { + "epoch": 0.4426917694949752, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7887991070747375, + "num_tokens": 132833232.0, + "step": 3480 + }, + { + "epoch": 0.4428189797735657, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.792427659034729, + "num_tokens": 132874670.0, + "step": 3481 + }, + { + "epoch": 0.44294619005215624, + "grad_norm": 86.5, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7946242690086365, + "num_tokens": 132916187.0, + "step": 3482 + }, + { + "epoch": 0.4430734003307467, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7927600145339966, + "num_tokens": 132956948.0, + "step": 3483 + }, + { + "epoch": 0.44320061060933724, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7974858283996582, + "num_tokens": 132995213.0, + "step": 3484 + }, + { + "epoch": 0.44332782088792777, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7914748787879944, + "num_tokens": 133038850.0, + "step": 3485 + }, + { + "epoch": 0.44345503116651824, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7922313213348389, + "num_tokens": 133080766.0, + "step": 3486 + }, + { + "epoch": 0.44358224144510877, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7890677452087402, + "num_tokens": 133120845.0, + "step": 3487 + }, + { + "epoch": 0.4437094517236993, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7943246960639954, + "num_tokens": 133166011.0, + "step": 3488 + }, + { + "epoch": 0.4438366620022898, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7894808053970337, + "num_tokens": 133203507.0, + "step": 3489 + }, + { + "epoch": 0.4439638722808803, + "grad_norm": 124.0, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7671023011207581, + "num_tokens": 133240554.0, + "step": 3490 + }, + { + "epoch": 0.44409108255947083, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7749706506729126, + "num_tokens": 133274987.0, + "step": 3491 + }, + { + "epoch": 0.4442182928380613, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7651286125183105, + "num_tokens": 133319287.0, + "step": 3492 + }, + { + "epoch": 0.44434550311665183, + "grad_norm": 158.0, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7874364852905273, + "num_tokens": 133353670.0, + "step": 3493 + }, + { + "epoch": 0.44447271339524236, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7942072153091431, + "num_tokens": 133391950.0, + "step": 3494 + }, + { + "epoch": 0.44459992367383283, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7941595911979675, + "num_tokens": 133424758.0, + "step": 3495 + }, + { + "epoch": 0.44472713395242336, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8199961185455322, + "num_tokens": 133459869.0, + "step": 3496 + }, + { + "epoch": 0.4448543442310139, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7858791351318359, + "num_tokens": 133494598.0, + "step": 3497 + }, + { + "epoch": 0.44498155450960436, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.8016676902770996, + "num_tokens": 133536403.0, + "step": 3498 + }, + { + "epoch": 0.4451087647881949, + "grad_norm": 106.0, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7609529495239258, + "num_tokens": 133567319.0, + "step": 3499 + }, + { + "epoch": 0.4452359750667854, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7824558019638062, + "num_tokens": 133604829.0, + "step": 3500 + }, + { + "epoch": 0.4453631853453759, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7911282777786255, + "num_tokens": 133638373.0, + "step": 3501 + }, + { + "epoch": 0.4454903956239664, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8028555512428284, + "num_tokens": 133674110.0, + "step": 3502 + }, + { + "epoch": 0.44561760590255695, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7928993701934814, + "num_tokens": 133711005.0, + "step": 3503 + }, + { + "epoch": 0.4457448161811474, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8022878766059875, + "num_tokens": 133746258.0, + "step": 3504 + }, + { + "epoch": 0.44587202645973795, + "grad_norm": 78.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7987572550773621, + "num_tokens": 133790695.0, + "step": 3505 + }, + { + "epoch": 0.4459992367383285, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7884070873260498, + "num_tokens": 133830181.0, + "step": 3506 + }, + { + "epoch": 0.44612644701691895, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7960964441299438, + "num_tokens": 133866849.0, + "step": 3507 + }, + { + "epoch": 0.4462536572955095, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7935487031936646, + "num_tokens": 133906812.0, + "step": 3508 + }, + { + "epoch": 0.4463808675741, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7867260575294495, + "num_tokens": 133948956.0, + "step": 3509 + }, + { + "epoch": 0.4465080778526905, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7753366231918335, + "num_tokens": 133988764.0, + "step": 3510 + }, + { + "epoch": 0.446635288131281, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7978546619415283, + "num_tokens": 134028148.0, + "step": 3511 + }, + { + "epoch": 0.44676249840987153, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7660487294197083, + "num_tokens": 134065909.0, + "step": 3512 + }, + { + "epoch": 0.446889708688462, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7995011210441589, + "num_tokens": 134108342.0, + "step": 3513 + }, + { + "epoch": 0.44701691896705253, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7915812730789185, + "num_tokens": 134145928.0, + "step": 3514 + }, + { + "epoch": 0.44714412924564306, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7875088453292847, + "num_tokens": 134185955.0, + "step": 3515 + }, + { + "epoch": 0.44727133952423354, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7980333566665649, + "num_tokens": 134225829.0, + "step": 3516 + }, + { + "epoch": 0.44739854980282406, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7756708860397339, + "num_tokens": 134261381.0, + "step": 3517 + }, + { + "epoch": 0.4475257600814146, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7930743098258972, + "num_tokens": 134294697.0, + "step": 3518 + }, + { + "epoch": 0.44765297036000506, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8131903409957886, + "num_tokens": 134334349.0, + "step": 3519 + }, + { + "epoch": 0.4477801806385956, + "grad_norm": 78.5, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.779901385307312, + "num_tokens": 134373381.0, + "step": 3520 + }, + { + "epoch": 0.4479073909171861, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7937555909156799, + "num_tokens": 134407952.0, + "step": 3521 + }, + { + "epoch": 0.4480346011957766, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7860689163208008, + "num_tokens": 134458862.0, + "step": 3522 + }, + { + "epoch": 0.4481618114743671, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7837915420532227, + "num_tokens": 134494007.0, + "step": 3523 + }, + { + "epoch": 0.44828902175295765, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7773851156234741, + "num_tokens": 134534195.0, + "step": 3524 + }, + { + "epoch": 0.4484162320315481, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.795164942741394, + "num_tokens": 134570241.0, + "step": 3525 + }, + { + "epoch": 0.44854344231013865, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7902150750160217, + "num_tokens": 134611904.0, + "step": 3526 + }, + { + "epoch": 0.4486706525887292, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7777143716812134, + "num_tokens": 134644426.0, + "step": 3527 + }, + { + "epoch": 0.44879786286731965, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7998038530349731, + "num_tokens": 134678979.0, + "step": 3528 + }, + { + "epoch": 0.4489250731459102, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7931535840034485, + "num_tokens": 134717270.0, + "step": 3529 + }, + { + "epoch": 0.4490522834245007, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7900136709213257, + "num_tokens": 134755464.0, + "step": 3530 + }, + { + "epoch": 0.44917949370309124, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7847166061401367, + "num_tokens": 134793737.0, + "step": 3531 + }, + { + "epoch": 0.4493067039816817, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7906352281570435, + "num_tokens": 134834332.0, + "step": 3532 + }, + { + "epoch": 0.44943391426027224, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8010501265525818, + "num_tokens": 134871027.0, + "step": 3533 + }, + { + "epoch": 0.44956112453886277, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7912697792053223, + "num_tokens": 134908529.0, + "step": 3534 + }, + { + "epoch": 0.44968833481745324, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.801102876663208, + "num_tokens": 134944177.0, + "step": 3535 + }, + { + "epoch": 0.44981554509604377, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7844402194023132, + "num_tokens": 134977928.0, + "step": 3536 + }, + { + "epoch": 0.4499427553746343, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7901760935783386, + "num_tokens": 135017554.0, + "step": 3537 + }, + { + "epoch": 0.45006996565322477, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7981147766113281, + "num_tokens": 135058694.0, + "step": 3538 + }, + { + "epoch": 0.4501971759318153, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.794593334197998, + "num_tokens": 135094037.0, + "step": 3539 + }, + { + "epoch": 0.4503243862104058, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7868764400482178, + "num_tokens": 135132758.0, + "step": 3540 + }, + { + "epoch": 0.4504515964889963, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8019993305206299, + "num_tokens": 135169398.0, + "step": 3541 + }, + { + "epoch": 0.4505788067675868, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7726240754127502, + "num_tokens": 135204312.0, + "step": 3542 + }, + { + "epoch": 0.45070601704617735, + "grad_norm": 84.0, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7893942594528198, + "num_tokens": 135246287.0, + "step": 3543 + }, + { + "epoch": 0.4508332273247678, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7948357462882996, + "num_tokens": 135280970.0, + "step": 3544 + }, + { + "epoch": 0.45096043760335836, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8159715533256531, + "num_tokens": 135317873.0, + "step": 3545 + }, + { + "epoch": 0.4510876478819489, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.8019238114356995, + "num_tokens": 135354994.0, + "step": 3546 + }, + { + "epoch": 0.45121485816053936, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7854495644569397, + "num_tokens": 135393050.0, + "step": 3547 + }, + { + "epoch": 0.4513420684391299, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7962300777435303, + "num_tokens": 135431704.0, + "step": 3548 + }, + { + "epoch": 0.4514692787177204, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7667646408081055, + "num_tokens": 135471306.0, + "step": 3549 + }, + { + "epoch": 0.4515964889963109, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7794556617736816, + "num_tokens": 135509647.0, + "step": 3550 + }, + { + "epoch": 0.4517236992749014, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7998119592666626, + "num_tokens": 135550469.0, + "step": 3551 + }, + { + "epoch": 0.45185090955349194, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.791732668876648, + "num_tokens": 135589290.0, + "step": 3552 + }, + { + "epoch": 0.4519781198320824, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7843267917633057, + "num_tokens": 135624416.0, + "step": 3553 + }, + { + "epoch": 0.45210533011067294, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7957533597946167, + "num_tokens": 135662316.0, + "step": 3554 + }, + { + "epoch": 0.45223254038926347, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7935824394226074, + "num_tokens": 135701739.0, + "step": 3555 + }, + { + "epoch": 0.45235975066785394, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7923598885536194, + "num_tokens": 135741372.0, + "step": 3556 + }, + { + "epoch": 0.45248696094644447, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.8170039653778076, + "num_tokens": 135779945.0, + "step": 3557 + }, + { + "epoch": 0.452614171225035, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7801094055175781, + "num_tokens": 135813179.0, + "step": 3558 + }, + { + "epoch": 0.4527413815036255, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7948426008224487, + "num_tokens": 135855147.0, + "step": 3559 + }, + { + "epoch": 0.452868591782216, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.8011395335197449, + "num_tokens": 135892990.0, + "step": 3560 + }, + { + "epoch": 0.45299580206080653, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8124284148216248, + "num_tokens": 135926911.0, + "step": 3561 + }, + { + "epoch": 0.453123012339397, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.803175687789917, + "num_tokens": 135963530.0, + "step": 3562 + }, + { + "epoch": 0.45325022261798753, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7772563099861145, + "num_tokens": 136000677.0, + "step": 3563 + }, + { + "epoch": 0.45337743289657806, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7778573036193848, + "num_tokens": 136037283.0, + "step": 3564 + }, + { + "epoch": 0.45350464317516853, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7687324285507202, + "num_tokens": 136071454.0, + "step": 3565 + }, + { + "epoch": 0.45363185345375906, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7793674468994141, + "num_tokens": 136105638.0, + "step": 3566 + }, + { + "epoch": 0.4537590637323496, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.797898530960083, + "num_tokens": 136147968.0, + "step": 3567 + }, + { + "epoch": 0.45388627401094006, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7930872440338135, + "num_tokens": 136190468.0, + "step": 3568 + }, + { + "epoch": 0.4540134842895306, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.797888994216919, + "num_tokens": 136229363.0, + "step": 3569 + }, + { + "epoch": 0.4541406945681211, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7846727967262268, + "num_tokens": 136268834.0, + "step": 3570 + }, + { + "epoch": 0.4542679048467116, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7650585174560547, + "num_tokens": 136306761.0, + "step": 3571 + }, + { + "epoch": 0.4543951151253021, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7916775941848755, + "num_tokens": 136347736.0, + "step": 3572 + }, + { + "epoch": 0.45452232540389265, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7935212850570679, + "num_tokens": 136381009.0, + "step": 3573 + }, + { + "epoch": 0.4546495356824831, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.803174614906311, + "num_tokens": 136418050.0, + "step": 3574 + }, + { + "epoch": 0.45477674596107365, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7765548229217529, + "num_tokens": 136459109.0, + "step": 3575 + }, + { + "epoch": 0.4549039562396642, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7911807894706726, + "num_tokens": 136497993.0, + "step": 3576 + }, + { + "epoch": 0.45503116651825465, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7938370704650879, + "num_tokens": 136534446.0, + "step": 3577 + }, + { + "epoch": 0.4551583767968452, + "grad_norm": 77.0, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7774938344955444, + "num_tokens": 136571117.0, + "step": 3578 + }, + { + "epoch": 0.4552855870754357, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7982102036476135, + "num_tokens": 136601706.0, + "step": 3579 + }, + { + "epoch": 0.4554127973540262, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.8004478812217712, + "num_tokens": 136639735.0, + "step": 3580 + }, + { + "epoch": 0.4555400076326167, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7808034420013428, + "num_tokens": 136675647.0, + "step": 3581 + }, + { + "epoch": 0.45566721791120723, + "grad_norm": 157.0, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7943726778030396, + "num_tokens": 136713361.0, + "step": 3582 + }, + { + "epoch": 0.45579442818979776, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7854008078575134, + "num_tokens": 136751440.0, + "step": 3583 + }, + { + "epoch": 0.45592163846838824, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.790581464767456, + "num_tokens": 136786078.0, + "step": 3584 + }, + { + "epoch": 0.45604884874697876, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.8020468950271606, + "num_tokens": 136828313.0, + "step": 3585 + }, + { + "epoch": 0.4561760590255693, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7897016406059265, + "num_tokens": 136862549.0, + "step": 3586 + }, + { + "epoch": 0.45630326930415976, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7939594984054565, + "num_tokens": 136906423.0, + "step": 3587 + }, + { + "epoch": 0.4564304795827503, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.78215491771698, + "num_tokens": 136941550.0, + "step": 3588 + }, + { + "epoch": 0.4565576898613408, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7946251034736633, + "num_tokens": 136980234.0, + "step": 3589 + }, + { + "epoch": 0.4566849001399313, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.796603798866272, + "num_tokens": 137023991.0, + "step": 3590 + }, + { + "epoch": 0.4568121104185218, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.777227520942688, + "num_tokens": 137063261.0, + "step": 3591 + }, + { + "epoch": 0.45693932069711235, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.8010345697402954, + "num_tokens": 137103935.0, + "step": 3592 + }, + { + "epoch": 0.4570665309757028, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7915030121803284, + "num_tokens": 137140194.0, + "step": 3593 + }, + { + "epoch": 0.45719374125429335, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7831472754478455, + "num_tokens": 137179086.0, + "step": 3594 + }, + { + "epoch": 0.4573209515328839, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7941573858261108, + "num_tokens": 137216887.0, + "step": 3595 + }, + { + "epoch": 0.45744816181147435, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8081684708595276, + "num_tokens": 137255029.0, + "step": 3596 + }, + { + "epoch": 0.4575753720900649, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7951445579528809, + "num_tokens": 137291399.0, + "step": 3597 + }, + { + "epoch": 0.4577025823686554, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7992397546768188, + "num_tokens": 137326569.0, + "step": 3598 + }, + { + "epoch": 0.4578297926472459, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7928391695022583, + "num_tokens": 137370035.0, + "step": 3599 + }, + { + "epoch": 0.4579570029258364, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.790717601776123, + "num_tokens": 137407969.0, + "step": 3600 + }, + { + "epoch": 0.45808421320442694, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7884640097618103, + "num_tokens": 137447256.0, + "step": 3601 + }, + { + "epoch": 0.4582114234830174, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7783791422843933, + "num_tokens": 137479762.0, + "step": 3602 + }, + { + "epoch": 0.45833863376160794, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7871711850166321, + "num_tokens": 137521823.0, + "step": 3603 + }, + { + "epoch": 0.45846584404019847, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.794379472732544, + "num_tokens": 137558174.0, + "step": 3604 + }, + { + "epoch": 0.45859305431878894, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7680410146713257, + "num_tokens": 137597265.0, + "step": 3605 + }, + { + "epoch": 0.45872026459737947, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8059271574020386, + "num_tokens": 137635954.0, + "step": 3606 + }, + { + "epoch": 0.45884747487597, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7919049263000488, + "num_tokens": 137676137.0, + "step": 3607 + }, + { + "epoch": 0.45897468515456047, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7973347902297974, + "num_tokens": 137716723.0, + "step": 3608 + }, + { + "epoch": 0.459101895433151, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.77325040102005, + "num_tokens": 137752435.0, + "step": 3609 + }, + { + "epoch": 0.4592291057117415, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7795542478561401, + "num_tokens": 137786684.0, + "step": 3610 + }, + { + "epoch": 0.459356315990332, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7877910137176514, + "num_tokens": 137830668.0, + "step": 3611 + }, + { + "epoch": 0.4594835262689225, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7955931425094604, + "num_tokens": 137865262.0, + "step": 3612 + }, + { + "epoch": 0.45961073654751305, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7934578657150269, + "num_tokens": 137898867.0, + "step": 3613 + }, + { + "epoch": 0.4597379468261035, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7900075912475586, + "num_tokens": 137935108.0, + "step": 3614 + }, + { + "epoch": 0.45986515710469406, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8077623248100281, + "num_tokens": 137977714.0, + "step": 3615 + }, + { + "epoch": 0.4599923673832846, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8030656576156616, + "num_tokens": 138006345.0, + "step": 3616 + }, + { + "epoch": 0.46011957766187506, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.796345591545105, + "num_tokens": 138045106.0, + "step": 3617 + }, + { + "epoch": 0.4602467879404656, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7826650142669678, + "num_tokens": 138079993.0, + "step": 3618 + }, + { + "epoch": 0.4603739982190561, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7947050929069519, + "num_tokens": 138118555.0, + "step": 3619 + }, + { + "epoch": 0.4605012084976466, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.8005180358886719, + "num_tokens": 138155096.0, + "step": 3620 + }, + { + "epoch": 0.4606284187762371, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7963733673095703, + "num_tokens": 138194529.0, + "step": 3621 + }, + { + "epoch": 0.46075562905482764, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.765764057636261, + "num_tokens": 138235520.0, + "step": 3622 + }, + { + "epoch": 0.4608828393334181, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.802055299282074, + "num_tokens": 138271391.0, + "step": 3623 + }, + { + "epoch": 0.46101004961200864, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7904982566833496, + "num_tokens": 138313923.0, + "step": 3624 + }, + { + "epoch": 0.46113725989059917, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7836709022521973, + "num_tokens": 138355992.0, + "step": 3625 + }, + { + "epoch": 0.46126447016918964, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8154737949371338, + "num_tokens": 138393074.0, + "step": 3626 + }, + { + "epoch": 0.4613916804477802, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7995648980140686, + "num_tokens": 138427854.0, + "step": 3627 + }, + { + "epoch": 0.4615188907263707, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.8043969869613647, + "num_tokens": 138467359.0, + "step": 3628 + }, + { + "epoch": 0.4616461010049612, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7765651941299438, + "num_tokens": 138505355.0, + "step": 3629 + }, + { + "epoch": 0.4617733112835517, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.796744704246521, + "num_tokens": 138543703.0, + "step": 3630 + }, + { + "epoch": 0.46190052156214223, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.783995509147644, + "num_tokens": 138579860.0, + "step": 3631 + }, + { + "epoch": 0.46202773184073276, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.8029166460037231, + "num_tokens": 138615755.0, + "step": 3632 + }, + { + "epoch": 0.46215494211932323, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.785344123840332, + "num_tokens": 138647607.0, + "step": 3633 + }, + { + "epoch": 0.46228215239791376, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7950170040130615, + "num_tokens": 138679284.0, + "step": 3634 + }, + { + "epoch": 0.4624093626765043, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8014975786209106, + "num_tokens": 138719690.0, + "step": 3635 + }, + { + "epoch": 0.46253657295509476, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7753605842590332, + "num_tokens": 138758031.0, + "step": 3636 + }, + { + "epoch": 0.4626637832336853, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7950993776321411, + "num_tokens": 138797283.0, + "step": 3637 + }, + { + "epoch": 0.4627909935122758, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7889032363891602, + "num_tokens": 138834672.0, + "step": 3638 + }, + { + "epoch": 0.4629182037908663, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8031215667724609, + "num_tokens": 138871105.0, + "step": 3639 + }, + { + "epoch": 0.4630454140694568, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7895596027374268, + "num_tokens": 138903516.0, + "step": 3640 + }, + { + "epoch": 0.46317262434804735, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7965361475944519, + "num_tokens": 138942021.0, + "step": 3641 + }, + { + "epoch": 0.4632998346266378, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.803038477897644, + "num_tokens": 138978210.0, + "step": 3642 + }, + { + "epoch": 0.46342704490522835, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.800663948059082, + "num_tokens": 139019122.0, + "step": 3643 + }, + { + "epoch": 0.4635542551838189, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8043030500411987, + "num_tokens": 139062187.0, + "step": 3644 + }, + { + "epoch": 0.46368146546240935, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8107700943946838, + "num_tokens": 139102960.0, + "step": 3645 + }, + { + "epoch": 0.4638086757409999, + "grad_norm": 94.5, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7907385230064392, + "num_tokens": 139137302.0, + "step": 3646 + }, + { + "epoch": 0.4639358860195904, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.770837664604187, + "num_tokens": 139177195.0, + "step": 3647 + }, + { + "epoch": 0.4640630962981809, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7901331186294556, + "num_tokens": 139214722.0, + "step": 3648 + }, + { + "epoch": 0.4641903065767714, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7879214882850647, + "num_tokens": 139248569.0, + "step": 3649 + }, + { + "epoch": 0.46431751685536193, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7973048686981201, + "num_tokens": 139289141.0, + "step": 3650 + }, + { + "epoch": 0.4644447271339524, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8088785409927368, + "num_tokens": 139330791.0, + "step": 3651 + }, + { + "epoch": 0.46457193741254293, + "grad_norm": 93.0, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7602977752685547, + "num_tokens": 139373009.0, + "step": 3652 + }, + { + "epoch": 0.46469914769113346, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7903775572776794, + "num_tokens": 139408894.0, + "step": 3653 + }, + { + "epoch": 0.46482635796972394, + "grad_norm": 93.5, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7915087342262268, + "num_tokens": 139444040.0, + "step": 3654 + }, + { + "epoch": 0.46495356824831446, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7974650263786316, + "num_tokens": 139479675.0, + "step": 3655 + }, + { + "epoch": 0.465080778526905, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7992316484451294, + "num_tokens": 139521792.0, + "step": 3656 + }, + { + "epoch": 0.46520798880549546, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8036506772041321, + "num_tokens": 139563471.0, + "step": 3657 + }, + { + "epoch": 0.465335199084086, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7968957424163818, + "num_tokens": 139604702.0, + "step": 3658 + }, + { + "epoch": 0.4654624093626765, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7897851467132568, + "num_tokens": 139644573.0, + "step": 3659 + }, + { + "epoch": 0.465589619641267, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7732235193252563, + "num_tokens": 139680924.0, + "step": 3660 + }, + { + "epoch": 0.4657168299198575, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7848243713378906, + "num_tokens": 139724369.0, + "step": 3661 + }, + { + "epoch": 0.46584404019844805, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7896323204040527, + "num_tokens": 139769412.0, + "step": 3662 + }, + { + "epoch": 0.4659712504770385, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7670705914497375, + "num_tokens": 139808066.0, + "step": 3663 + }, + { + "epoch": 0.46609846075562905, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.8001046180725098, + "num_tokens": 139851169.0, + "step": 3664 + }, + { + "epoch": 0.4662256710342196, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8030990958213806, + "num_tokens": 139890550.0, + "step": 3665 + }, + { + "epoch": 0.46635288131281005, + "grad_norm": 111.0, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.767426609992981, + "num_tokens": 139931496.0, + "step": 3666 + }, + { + "epoch": 0.4664800915914006, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.784557580947876, + "num_tokens": 139965872.0, + "step": 3667 + }, + { + "epoch": 0.4666073018699911, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7779189944267273, + "num_tokens": 140000394.0, + "step": 3668 + }, + { + "epoch": 0.4667345121485816, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7768582105636597, + "num_tokens": 140039967.0, + "step": 3669 + }, + { + "epoch": 0.4668617224271721, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7931931614875793, + "num_tokens": 140075739.0, + "step": 3670 + }, + { + "epoch": 0.46698893270576264, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7990345358848572, + "num_tokens": 140110016.0, + "step": 3671 + }, + { + "epoch": 0.4671161429843531, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7814593315124512, + "num_tokens": 140147905.0, + "step": 3672 + }, + { + "epoch": 0.46724335326294364, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7782493233680725, + "num_tokens": 140182284.0, + "step": 3673 + }, + { + "epoch": 0.46737056354153417, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7849499583244324, + "num_tokens": 140220669.0, + "step": 3674 + }, + { + "epoch": 0.46749777382012464, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7791246175765991, + "num_tokens": 140258399.0, + "step": 3675 + }, + { + "epoch": 0.46762498409871517, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8055597543716431, + "num_tokens": 140296015.0, + "step": 3676 + }, + { + "epoch": 0.4677521943773057, + "grad_norm": 130.0, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7786828279495239, + "num_tokens": 140338618.0, + "step": 3677 + }, + { + "epoch": 0.46787940465589617, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7950336933135986, + "num_tokens": 140379151.0, + "step": 3678 + }, + { + "epoch": 0.4680066149344867, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7801492214202881, + "num_tokens": 140417740.0, + "step": 3679 + }, + { + "epoch": 0.4681338252130772, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.789181649684906, + "num_tokens": 140452454.0, + "step": 3680 + }, + { + "epoch": 0.46826103549166775, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.766872763633728, + "num_tokens": 140492981.0, + "step": 3681 + }, + { + "epoch": 0.4683882457702582, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.779341459274292, + "num_tokens": 140534131.0, + "step": 3682 + }, + { + "epoch": 0.46851545604884876, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7842693328857422, + "num_tokens": 140576761.0, + "step": 3683 + }, + { + "epoch": 0.4686426663274393, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7699851393699646, + "num_tokens": 140609538.0, + "step": 3684 + }, + { + "epoch": 0.46876987660602976, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7893154621124268, + "num_tokens": 140641802.0, + "step": 3685 + }, + { + "epoch": 0.4688970868846203, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7835273146629333, + "num_tokens": 140685832.0, + "step": 3686 + }, + { + "epoch": 0.4690242971632108, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7796503305435181, + "num_tokens": 140725333.0, + "step": 3687 + }, + { + "epoch": 0.4691515074418013, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7847731113433838, + "num_tokens": 140760838.0, + "step": 3688 + }, + { + "epoch": 0.4692787177203918, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7746615409851074, + "num_tokens": 140797367.0, + "step": 3689 + }, + { + "epoch": 0.46940592799898234, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7882478833198547, + "num_tokens": 140839778.0, + "step": 3690 + }, + { + "epoch": 0.4695331382775728, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7840976715087891, + "num_tokens": 140880340.0, + "step": 3691 + }, + { + "epoch": 0.46966034855616334, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7844640016555786, + "num_tokens": 140915451.0, + "step": 3692 + }, + { + "epoch": 0.46978755883475387, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7926369309425354, + "num_tokens": 140956967.0, + "step": 3693 + }, + { + "epoch": 0.46991476911334434, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8078900575637817, + "num_tokens": 140996640.0, + "step": 3694 + }, + { + "epoch": 0.47004197939193487, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7925962209701538, + "num_tokens": 141036094.0, + "step": 3695 + }, + { + "epoch": 0.4701691896705254, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7786380052566528, + "num_tokens": 141069033.0, + "step": 3696 + }, + { + "epoch": 0.4702963999491159, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8102818131446838, + "num_tokens": 141111075.0, + "step": 3697 + }, + { + "epoch": 0.4704236102277064, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8040712475776672, + "num_tokens": 141145270.0, + "step": 3698 + }, + { + "epoch": 0.47055082050629693, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8057410717010498, + "num_tokens": 141182132.0, + "step": 3699 + }, + { + "epoch": 0.4706780307848874, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.785988450050354, + "num_tokens": 141217455.0, + "step": 3700 + }, + { + "epoch": 0.47080524106347793, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7867803573608398, + "num_tokens": 141249680.0, + "step": 3701 + }, + { + "epoch": 0.47093245134206846, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7947707176208496, + "num_tokens": 141292085.0, + "step": 3702 + }, + { + "epoch": 0.47105966162065893, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7990329265594482, + "num_tokens": 141331455.0, + "step": 3703 + }, + { + "epoch": 0.47118687189924946, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7883299589157104, + "num_tokens": 141365623.0, + "step": 3704 + }, + { + "epoch": 0.47131408217784, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7757755517959595, + "num_tokens": 141399033.0, + "step": 3705 + }, + { + "epoch": 0.47144129245643046, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7960578203201294, + "num_tokens": 141430436.0, + "step": 3706 + }, + { + "epoch": 0.471568502735021, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8047827482223511, + "num_tokens": 141463169.0, + "step": 3707 + }, + { + "epoch": 0.4716957130136115, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8092527389526367, + "num_tokens": 141495524.0, + "step": 3708 + }, + { + "epoch": 0.471822923292202, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8029125928878784, + "num_tokens": 141533099.0, + "step": 3709 + }, + { + "epoch": 0.4719501335707925, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7991378307342529, + "num_tokens": 141576389.0, + "step": 3710 + }, + { + "epoch": 0.47207734384938305, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7936396598815918, + "num_tokens": 141611870.0, + "step": 3711 + }, + { + "epoch": 0.4722045541279735, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.8002010583877563, + "num_tokens": 141648539.0, + "step": 3712 + }, + { + "epoch": 0.47233176440656405, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7978631258010864, + "num_tokens": 141688549.0, + "step": 3713 + }, + { + "epoch": 0.4724589746851546, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7984645366668701, + "num_tokens": 141725294.0, + "step": 3714 + }, + { + "epoch": 0.47258618496374505, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7713298797607422, + "num_tokens": 141756761.0, + "step": 3715 + }, + { + "epoch": 0.4727133952423356, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7697083353996277, + "num_tokens": 141790417.0, + "step": 3716 + }, + { + "epoch": 0.4728406055209261, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8065370321273804, + "num_tokens": 141830749.0, + "step": 3717 + }, + { + "epoch": 0.4729678157995166, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7980363965034485, + "num_tokens": 141868433.0, + "step": 3718 + }, + { + "epoch": 0.4730950260781071, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8136553764343262, + "num_tokens": 141907140.0, + "step": 3719 + }, + { + "epoch": 0.47322223635669763, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8072119355201721, + "num_tokens": 141947436.0, + "step": 3720 + }, + { + "epoch": 0.4733494466352881, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7839890718460083, + "num_tokens": 141986876.0, + "step": 3721 + }, + { + "epoch": 0.47347665691387864, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7911633253097534, + "num_tokens": 142025806.0, + "step": 3722 + }, + { + "epoch": 0.47360386719246916, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7872869968414307, + "num_tokens": 142057932.0, + "step": 3723 + }, + { + "epoch": 0.47373107747105964, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7931699752807617, + "num_tokens": 142088880.0, + "step": 3724 + }, + { + "epoch": 0.47385828774965016, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8078835606575012, + "num_tokens": 142124959.0, + "step": 3725 + }, + { + "epoch": 0.4739854980282407, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7746415138244629, + "num_tokens": 142160667.0, + "step": 3726 + }, + { + "epoch": 0.47411270830683117, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7943533658981323, + "num_tokens": 142205017.0, + "step": 3727 + }, + { + "epoch": 0.4742399185854217, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7947670221328735, + "num_tokens": 142241479.0, + "step": 3728 + }, + { + "epoch": 0.4743671288640122, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7979825735092163, + "num_tokens": 142280881.0, + "step": 3729 + }, + { + "epoch": 0.4744943391426027, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8065245747566223, + "num_tokens": 142328248.0, + "step": 3730 + }, + { + "epoch": 0.4746215494211932, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7724368572235107, + "num_tokens": 142364872.0, + "step": 3731 + }, + { + "epoch": 0.47474875969978375, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7734384536743164, + "num_tokens": 142400494.0, + "step": 3732 + }, + { + "epoch": 0.4748759699783743, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.79036545753479, + "num_tokens": 142433739.0, + "step": 3733 + }, + { + "epoch": 0.47500318025696475, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7987097501754761, + "num_tokens": 142469777.0, + "step": 3734 + }, + { + "epoch": 0.4751303905355553, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7841261625289917, + "num_tokens": 142510052.0, + "step": 3735 + }, + { + "epoch": 0.4752576008141458, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7789016366004944, + "num_tokens": 142545689.0, + "step": 3736 + }, + { + "epoch": 0.4753848110927363, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7944839000701904, + "num_tokens": 142585916.0, + "step": 3737 + }, + { + "epoch": 0.4755120213713268, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7990095615386963, + "num_tokens": 142623019.0, + "step": 3738 + }, + { + "epoch": 0.47563923164991734, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7882680296897888, + "num_tokens": 142660986.0, + "step": 3739 + }, + { + "epoch": 0.4757664419285078, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7934785485267639, + "num_tokens": 142702876.0, + "step": 3740 + }, + { + "epoch": 0.47589365220709834, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7811384797096252, + "num_tokens": 142744082.0, + "step": 3741 + }, + { + "epoch": 0.47602086248568887, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7705832123756409, + "num_tokens": 142777511.0, + "step": 3742 + }, + { + "epoch": 0.47614807276427934, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7886734008789062, + "num_tokens": 142818882.0, + "step": 3743 + }, + { + "epoch": 0.47627528304286987, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7961417436599731, + "num_tokens": 142852771.0, + "step": 3744 + }, + { + "epoch": 0.4764024933214604, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7905207276344299, + "num_tokens": 142888398.0, + "step": 3745 + }, + { + "epoch": 0.47652970360005087, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8060041666030884, + "num_tokens": 142924789.0, + "step": 3746 + }, + { + "epoch": 0.4766569138786414, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7996612787246704, + "num_tokens": 142958843.0, + "step": 3747 + }, + { + "epoch": 0.4767841241572319, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8066046237945557, + "num_tokens": 142996408.0, + "step": 3748 + }, + { + "epoch": 0.4769113344358224, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7855938673019409, + "num_tokens": 143040031.0, + "step": 3749 + }, + { + "epoch": 0.4770385447144129, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7772781848907471, + "num_tokens": 143081648.0, + "step": 3750 + }, + { + "epoch": 0.47716575499300345, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7679629325866699, + "num_tokens": 143118913.0, + "step": 3751 + }, + { + "epoch": 0.4772929652715939, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.803925633430481, + "num_tokens": 143153858.0, + "step": 3752 + }, + { + "epoch": 0.47742017555018446, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7997118234634399, + "num_tokens": 143190977.0, + "step": 3753 + }, + { + "epoch": 0.477547385828775, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7877428531646729, + "num_tokens": 143228009.0, + "step": 3754 + }, + { + "epoch": 0.47767459610736546, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7815605401992798, + "num_tokens": 143269820.0, + "step": 3755 + }, + { + "epoch": 0.477801806385956, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7970679998397827, + "num_tokens": 143303441.0, + "step": 3756 + }, + { + "epoch": 0.4779290166645465, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7842892408370972, + "num_tokens": 143338748.0, + "step": 3757 + }, + { + "epoch": 0.478056226943137, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7878555655479431, + "num_tokens": 143373900.0, + "step": 3758 + }, + { + "epoch": 0.4781834372217275, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7875192165374756, + "num_tokens": 143411190.0, + "step": 3759 + }, + { + "epoch": 0.47831064750031804, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7861864566802979, + "num_tokens": 143449470.0, + "step": 3760 + }, + { + "epoch": 0.4784378577789085, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7996932864189148, + "num_tokens": 143485868.0, + "step": 3761 + }, + { + "epoch": 0.47856506805749904, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7950567007064819, + "num_tokens": 143520334.0, + "step": 3762 + }, + { + "epoch": 0.47869227833608957, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7781995534896851, + "num_tokens": 143557752.0, + "step": 3763 + }, + { + "epoch": 0.47881948861468004, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7922079563140869, + "num_tokens": 143593385.0, + "step": 3764 + }, + { + "epoch": 0.4789466988932706, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7691217064857483, + "num_tokens": 143631975.0, + "step": 3765 + }, + { + "epoch": 0.4790739091718611, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7725834846496582, + "num_tokens": 143668382.0, + "step": 3766 + }, + { + "epoch": 0.4792011194504516, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7789411544799805, + "num_tokens": 143706623.0, + "step": 3767 + }, + { + "epoch": 0.4793283297290421, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7969983816146851, + "num_tokens": 143745664.0, + "step": 3768 + }, + { + "epoch": 0.47945554000763263, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7828043699264526, + "num_tokens": 143780761.0, + "step": 3769 + }, + { + "epoch": 0.4795827502862231, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7947022914886475, + "num_tokens": 143825298.0, + "step": 3770 + }, + { + "epoch": 0.47970996056481363, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8093940615653992, + "num_tokens": 143859855.0, + "step": 3771 + }, + { + "epoch": 0.47983717084340416, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.807380199432373, + "num_tokens": 143898553.0, + "step": 3772 + }, + { + "epoch": 0.47996438112199463, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7925278544425964, + "num_tokens": 143935362.0, + "step": 3773 + }, + { + "epoch": 0.48009159140058516, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7974576950073242, + "num_tokens": 143978692.0, + "step": 3774 + }, + { + "epoch": 0.4802188016791757, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7969458103179932, + "num_tokens": 144020951.0, + "step": 3775 + }, + { + "epoch": 0.48034601195776616, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.801658570766449, + "num_tokens": 144058341.0, + "step": 3776 + }, + { + "epoch": 0.4804732222363567, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.787825882434845, + "num_tokens": 144098232.0, + "step": 3777 + }, + { + "epoch": 0.4806004325149472, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7856550216674805, + "num_tokens": 144137537.0, + "step": 3778 + }, + { + "epoch": 0.4807276427935377, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7984676361083984, + "num_tokens": 144178459.0, + "step": 3779 + }, + { + "epoch": 0.4808548530721282, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7885231971740723, + "num_tokens": 144219257.0, + "step": 3780 + }, + { + "epoch": 0.48098206335071875, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7974708080291748, + "num_tokens": 144262232.0, + "step": 3781 + }, + { + "epoch": 0.4811092736293093, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7686446905136108, + "num_tokens": 144297144.0, + "step": 3782 + }, + { + "epoch": 0.48123648390789975, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7953286170959473, + "num_tokens": 144336078.0, + "step": 3783 + }, + { + "epoch": 0.4813636941864903, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7899256944656372, + "num_tokens": 144380162.0, + "step": 3784 + }, + { + "epoch": 0.4814909044650808, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7746765613555908, + "num_tokens": 144416735.0, + "step": 3785 + }, + { + "epoch": 0.4816181147436713, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7807680368423462, + "num_tokens": 144452780.0, + "step": 3786 + }, + { + "epoch": 0.4817453250222618, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8065088987350464, + "num_tokens": 144494155.0, + "step": 3787 + }, + { + "epoch": 0.48187253530085233, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8066045641899109, + "num_tokens": 144533989.0, + "step": 3788 + }, + { + "epoch": 0.4819997455794428, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7896561622619629, + "num_tokens": 144577409.0, + "step": 3789 + }, + { + "epoch": 0.48212695585803333, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7882010340690613, + "num_tokens": 144618501.0, + "step": 3790 + }, + { + "epoch": 0.48225416613662386, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.8001961708068848, + "num_tokens": 144659454.0, + "step": 3791 + }, + { + "epoch": 0.48238137641521434, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7958922982215881, + "num_tokens": 144695580.0, + "step": 3792 + }, + { + "epoch": 0.48250858669380486, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7748368978500366, + "num_tokens": 144735586.0, + "step": 3793 + }, + { + "epoch": 0.4826357969723954, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7953041791915894, + "num_tokens": 144776371.0, + "step": 3794 + }, + { + "epoch": 0.48276300725098586, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7875359654426575, + "num_tokens": 144816724.0, + "step": 3795 + }, + { + "epoch": 0.4828902175295764, + "grad_norm": 128.0, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7957957983016968, + "num_tokens": 144854591.0, + "step": 3796 + }, + { + "epoch": 0.4830174278081669, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7956207990646362, + "num_tokens": 144893694.0, + "step": 3797 + }, + { + "epoch": 0.4831446380867574, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7884426116943359, + "num_tokens": 144932985.0, + "step": 3798 + }, + { + "epoch": 0.4832718483653479, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7881655693054199, + "num_tokens": 144967823.0, + "step": 3799 + }, + { + "epoch": 0.48339905864393845, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8200737237930298, + "num_tokens": 145007750.0, + "step": 3800 + }, + { + "epoch": 0.4835262689225289, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7869247198104858, + "num_tokens": 145048596.0, + "step": 3801 + }, + { + "epoch": 0.48365347920111945, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7781950235366821, + "num_tokens": 145080868.0, + "step": 3802 + }, + { + "epoch": 0.48378068947971, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.801061749458313, + "num_tokens": 145115865.0, + "step": 3803 + }, + { + "epoch": 0.48390789975830045, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8087121248245239, + "num_tokens": 145151748.0, + "step": 3804 + }, + { + "epoch": 0.484035110036891, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7934805154800415, + "num_tokens": 145187074.0, + "step": 3805 + }, + { + "epoch": 0.4841623203154815, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8124102354049683, + "num_tokens": 145224855.0, + "step": 3806 + }, + { + "epoch": 0.484289530594072, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8014184236526489, + "num_tokens": 145266538.0, + "step": 3807 + }, + { + "epoch": 0.4844167408726625, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7916759848594666, + "num_tokens": 145303560.0, + "step": 3808 + }, + { + "epoch": 0.48454395115125304, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7893093824386597, + "num_tokens": 145345985.0, + "step": 3809 + }, + { + "epoch": 0.4846711614298435, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7853551506996155, + "num_tokens": 145382132.0, + "step": 3810 + }, + { + "epoch": 0.48479837170843404, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7798532247543335, + "num_tokens": 145418992.0, + "step": 3811 + }, + { + "epoch": 0.48492558198702457, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7893057465553284, + "num_tokens": 145455892.0, + "step": 3812 + }, + { + "epoch": 0.48505279226561504, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7937475442886353, + "num_tokens": 145497217.0, + "step": 3813 + }, + { + "epoch": 0.48518000254420557, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.8001391887664795, + "num_tokens": 145533822.0, + "step": 3814 + }, + { + "epoch": 0.4853072128227961, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8131650686264038, + "num_tokens": 145572967.0, + "step": 3815 + }, + { + "epoch": 0.48543442310138657, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.81301349401474, + "num_tokens": 145609436.0, + "step": 3816 + }, + { + "epoch": 0.4855616333799771, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.802259087562561, + "num_tokens": 145645083.0, + "step": 3817 + }, + { + "epoch": 0.4856888436585676, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7713832855224609, + "num_tokens": 145680249.0, + "step": 3818 + }, + { + "epoch": 0.4858160539371581, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7979563474655151, + "num_tokens": 145718569.0, + "step": 3819 + }, + { + "epoch": 0.4859432642157486, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7783049941062927, + "num_tokens": 145751868.0, + "step": 3820 + }, + { + "epoch": 0.48607047449433916, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7910325527191162, + "num_tokens": 145787881.0, + "step": 3821 + }, + { + "epoch": 0.48619768477292963, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7946740984916687, + "num_tokens": 145824010.0, + "step": 3822 + }, + { + "epoch": 0.48632489505152016, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.80267333984375, + "num_tokens": 145860981.0, + "step": 3823 + }, + { + "epoch": 0.4864521053301107, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7912195324897766, + "num_tokens": 145900687.0, + "step": 3824 + }, + { + "epoch": 0.48657931560870116, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7787191271781921, + "num_tokens": 145938888.0, + "step": 3825 + }, + { + "epoch": 0.4867065258872917, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7898373007774353, + "num_tokens": 145987060.0, + "step": 3826 + }, + { + "epoch": 0.4868337361658822, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7895440459251404, + "num_tokens": 146026447.0, + "step": 3827 + }, + { + "epoch": 0.4869609464444727, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7967614531517029, + "num_tokens": 146067099.0, + "step": 3828 + }, + { + "epoch": 0.4870881567230632, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7810651659965515, + "num_tokens": 146110767.0, + "step": 3829 + }, + { + "epoch": 0.48721536700165374, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.782442569732666, + "num_tokens": 146147603.0, + "step": 3830 + }, + { + "epoch": 0.48734257728024427, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.806300938129425, + "num_tokens": 146188821.0, + "step": 3831 + }, + { + "epoch": 0.48746978755883474, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7927905321121216, + "num_tokens": 146231886.0, + "step": 3832 + }, + { + "epoch": 0.48759699783742527, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.76716148853302, + "num_tokens": 146271965.0, + "step": 3833 + }, + { + "epoch": 0.4877242081160158, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7798779606819153, + "num_tokens": 146316686.0, + "step": 3834 + }, + { + "epoch": 0.4878514183946063, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7863736152648926, + "num_tokens": 146355098.0, + "step": 3835 + }, + { + "epoch": 0.4879786286731968, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7961150407791138, + "num_tokens": 146390666.0, + "step": 3836 + }, + { + "epoch": 0.48810583895178733, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.799903154373169, + "num_tokens": 146426740.0, + "step": 3837 + }, + { + "epoch": 0.4882330492303778, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7843335866928101, + "num_tokens": 146462668.0, + "step": 3838 + }, + { + "epoch": 0.48836025950896833, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7832121253013611, + "num_tokens": 146500160.0, + "step": 3839 + }, + { + "epoch": 0.48848746978755886, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7869735956192017, + "num_tokens": 146542963.0, + "step": 3840 + }, + { + "epoch": 0.48861468006614933, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8033050298690796, + "num_tokens": 146580649.0, + "step": 3841 + }, + { + "epoch": 0.48874189034473986, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8043081164360046, + "num_tokens": 146620823.0, + "step": 3842 + }, + { + "epoch": 0.4888691006233304, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8041884899139404, + "num_tokens": 146663700.0, + "step": 3843 + }, + { + "epoch": 0.48899631090192086, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7855491638183594, + "num_tokens": 146698211.0, + "step": 3844 + }, + { + "epoch": 0.4891235211805114, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.795588493347168, + "num_tokens": 146738380.0, + "step": 3845 + }, + { + "epoch": 0.4892507314591019, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7904435396194458, + "num_tokens": 146775413.0, + "step": 3846 + }, + { + "epoch": 0.4893779417376924, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.794825553894043, + "num_tokens": 146814774.0, + "step": 3847 + }, + { + "epoch": 0.4895051520162829, + "grad_norm": 131.0, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8030865788459778, + "num_tokens": 146853957.0, + "step": 3848 + }, + { + "epoch": 0.48963236229487345, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7835231423377991, + "num_tokens": 146889060.0, + "step": 3849 + }, + { + "epoch": 0.4897595725734639, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8064560294151306, + "num_tokens": 146929849.0, + "step": 3850 + }, + { + "epoch": 0.48988678285205445, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.794469952583313, + "num_tokens": 146968376.0, + "step": 3851 + }, + { + "epoch": 0.490013993130645, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7722402215003967, + "num_tokens": 147002871.0, + "step": 3852 + }, + { + "epoch": 0.49014120340923545, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7871124744415283, + "num_tokens": 147039412.0, + "step": 3853 + }, + { + "epoch": 0.490268413687826, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7954554557800293, + "num_tokens": 147080307.0, + "step": 3854 + }, + { + "epoch": 0.4903956239664165, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.799892783164978, + "num_tokens": 147120131.0, + "step": 3855 + }, + { + "epoch": 0.490522834245007, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7841075658798218, + "num_tokens": 147155760.0, + "step": 3856 + }, + { + "epoch": 0.4906500445235975, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7965866327285767, + "num_tokens": 147195406.0, + "step": 3857 + }, + { + "epoch": 0.49077725480218803, + "grad_norm": 104.0, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8036272525787354, + "num_tokens": 147235688.0, + "step": 3858 + }, + { + "epoch": 0.4909044650807785, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8004180192947388, + "num_tokens": 147271575.0, + "step": 3859 + }, + { + "epoch": 0.49103167535936904, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7951261401176453, + "num_tokens": 147315473.0, + "step": 3860 + }, + { + "epoch": 0.49115888563795956, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8047897815704346, + "num_tokens": 147359058.0, + "step": 3861 + }, + { + "epoch": 0.49128609591655004, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7942004203796387, + "num_tokens": 147394443.0, + "step": 3862 + }, + { + "epoch": 0.49141330619514056, + "grad_norm": 82.0, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7858998775482178, + "num_tokens": 147436225.0, + "step": 3863 + }, + { + "epoch": 0.4915405164737311, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.803471565246582, + "num_tokens": 147473449.0, + "step": 3864 + }, + { + "epoch": 0.49166772675232157, + "grad_norm": 111.5, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7868142127990723, + "num_tokens": 147510330.0, + "step": 3865 + }, + { + "epoch": 0.4917949370309121, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8044946193695068, + "num_tokens": 147544683.0, + "step": 3866 + }, + { + "epoch": 0.4919221473095026, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8167881965637207, + "num_tokens": 147589330.0, + "step": 3867 + }, + { + "epoch": 0.4920493575880931, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7593985199928284, + "num_tokens": 147626864.0, + "step": 3868 + }, + { + "epoch": 0.4921765678666836, + "grad_norm": 103.0, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7783898711204529, + "num_tokens": 147672006.0, + "step": 3869 + }, + { + "epoch": 0.49230377814527415, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7889525890350342, + "num_tokens": 147717388.0, + "step": 3870 + }, + { + "epoch": 0.4924309884238646, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7799496650695801, + "num_tokens": 147759827.0, + "step": 3871 + }, + { + "epoch": 0.49255819870245515, + "grad_norm": 91.0, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7665222883224487, + "num_tokens": 147793742.0, + "step": 3872 + }, + { + "epoch": 0.4926854089810457, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8081949353218079, + "num_tokens": 147829881.0, + "step": 3873 + }, + { + "epoch": 0.49281261925963615, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7797457575798035, + "num_tokens": 147875051.0, + "step": 3874 + }, + { + "epoch": 0.4929398295382267, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7867501974105835, + "num_tokens": 147912786.0, + "step": 3875 + }, + { + "epoch": 0.4930670398168172, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7772622108459473, + "num_tokens": 147954981.0, + "step": 3876 + }, + { + "epoch": 0.4931942500954077, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7855277061462402, + "num_tokens": 147992179.0, + "step": 3877 + }, + { + "epoch": 0.4933214603739982, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7962906360626221, + "num_tokens": 148029866.0, + "step": 3878 + }, + { + "epoch": 0.49344867065258874, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7981018424034119, + "num_tokens": 148070057.0, + "step": 3879 + }, + { + "epoch": 0.4935758809311792, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7987675666809082, + "num_tokens": 148108461.0, + "step": 3880 + }, + { + "epoch": 0.49370309120976974, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7858554124832153, + "num_tokens": 148147609.0, + "step": 3881 + }, + { + "epoch": 0.49383030148836027, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.79039067029953, + "num_tokens": 148190475.0, + "step": 3882 + }, + { + "epoch": 0.4939575117669508, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7991514205932617, + "num_tokens": 148221187.0, + "step": 3883 + }, + { + "epoch": 0.49408472204554127, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.801417350769043, + "num_tokens": 148260916.0, + "step": 3884 + }, + { + "epoch": 0.4942119323241318, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7964516878128052, + "num_tokens": 148295591.0, + "step": 3885 + }, + { + "epoch": 0.4943391426027223, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7877129912376404, + "num_tokens": 148333445.0, + "step": 3886 + }, + { + "epoch": 0.4944663528813128, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7988849878311157, + "num_tokens": 148367951.0, + "step": 3887 + }, + { + "epoch": 0.4945935631599033, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7868672609329224, + "num_tokens": 148408491.0, + "step": 3888 + }, + { + "epoch": 0.49472077343849385, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7937412858009338, + "num_tokens": 148443883.0, + "step": 3889 + }, + { + "epoch": 0.4948479837170843, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.795732855796814, + "num_tokens": 148480511.0, + "step": 3890 + }, + { + "epoch": 0.49497519399567486, + "grad_norm": 95.5, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7774837613105774, + "num_tokens": 148514759.0, + "step": 3891 + }, + { + "epoch": 0.4951024042742654, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7972632050514221, + "num_tokens": 148553209.0, + "step": 3892 + }, + { + "epoch": 0.49522961455285586, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7820936441421509, + "num_tokens": 148590387.0, + "step": 3893 + }, + { + "epoch": 0.4953568248314464, + "grad_norm": 90.5, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7878928780555725, + "num_tokens": 148625201.0, + "step": 3894 + }, + { + "epoch": 0.4954840351100369, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7977790832519531, + "num_tokens": 148665989.0, + "step": 3895 + }, + { + "epoch": 0.4956112453886274, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7913762927055359, + "num_tokens": 148704839.0, + "step": 3896 + }, + { + "epoch": 0.4957384556672179, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7998803853988647, + "num_tokens": 148740523.0, + "step": 3897 + }, + { + "epoch": 0.49586566594580844, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8126755356788635, + "num_tokens": 148779103.0, + "step": 3898 + }, + { + "epoch": 0.4959928762243989, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8002438545227051, + "num_tokens": 148814659.0, + "step": 3899 + }, + { + "epoch": 0.49612008650298944, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8150383830070496, + "num_tokens": 148856864.0, + "step": 3900 + }, + { + "epoch": 0.49624729678157997, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7968869209289551, + "num_tokens": 148890309.0, + "step": 3901 + }, + { + "epoch": 0.49637450706017044, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7747423648834229, + "num_tokens": 148926662.0, + "step": 3902 + }, + { + "epoch": 0.496501717338761, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7799893021583557, + "num_tokens": 148963032.0, + "step": 3903 + }, + { + "epoch": 0.4966289276173515, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7961395382881165, + "num_tokens": 148996684.0, + "step": 3904 + }, + { + "epoch": 0.496756137895942, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7903045415878296, + "num_tokens": 149032735.0, + "step": 3905 + }, + { + "epoch": 0.4968833481745325, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7942727208137512, + "num_tokens": 149069415.0, + "step": 3906 + }, + { + "epoch": 0.49701055845312303, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.78272545337677, + "num_tokens": 149115042.0, + "step": 3907 + }, + { + "epoch": 0.4971377687317135, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8026742339134216, + "num_tokens": 149149767.0, + "step": 3908 + }, + { + "epoch": 0.49726497901030403, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7871078252792358, + "num_tokens": 149193000.0, + "step": 3909 + }, + { + "epoch": 0.49739218928889456, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7888375520706177, + "num_tokens": 149226963.0, + "step": 3910 + }, + { + "epoch": 0.49751939956748503, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7901632785797119, + "num_tokens": 149268139.0, + "step": 3911 + }, + { + "epoch": 0.49764660984607556, + "grad_norm": 115.0, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7977352142333984, + "num_tokens": 149305882.0, + "step": 3912 + }, + { + "epoch": 0.4977738201246661, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7933595776557922, + "num_tokens": 149344119.0, + "step": 3913 + }, + { + "epoch": 0.49790103040325656, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7836920022964478, + "num_tokens": 149384102.0, + "step": 3914 + }, + { + "epoch": 0.4980282406818471, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.790760338306427, + "num_tokens": 149425677.0, + "step": 3915 + }, + { + "epoch": 0.4981554509604376, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7893295288085938, + "num_tokens": 149463275.0, + "step": 3916 + }, + { + "epoch": 0.4982826612390281, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8065274953842163, + "num_tokens": 149503950.0, + "step": 3917 + }, + { + "epoch": 0.4984098715176186, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7829936146736145, + "num_tokens": 149541192.0, + "step": 3918 + }, + { + "epoch": 0.49853708179620915, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7893966436386108, + "num_tokens": 149581784.0, + "step": 3919 + }, + { + "epoch": 0.4986642920747996, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7956960797309875, + "num_tokens": 149619312.0, + "step": 3920 + }, + { + "epoch": 0.49879150235339015, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7889339923858643, + "num_tokens": 149657443.0, + "step": 3921 + }, + { + "epoch": 0.4989187126319807, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8003886938095093, + "num_tokens": 149694992.0, + "step": 3922 + }, + { + "epoch": 0.49904592291057115, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7884269952774048, + "num_tokens": 149729554.0, + "step": 3923 + }, + { + "epoch": 0.4991731331891617, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8021512627601624, + "num_tokens": 149770051.0, + "step": 3924 + }, + { + "epoch": 0.4993003434677522, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7582566738128662, + "num_tokens": 149806724.0, + "step": 3925 + }, + { + "epoch": 0.4994275537463427, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7841017246246338, + "num_tokens": 149849874.0, + "step": 3926 + }, + { + "epoch": 0.4995547640249332, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7934168577194214, + "num_tokens": 149888076.0, + "step": 3927 + }, + { + "epoch": 0.49968197430352373, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7894048094749451, + "num_tokens": 149923750.0, + "step": 3928 + }, + { + "epoch": 0.4998091845821142, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7736138105392456, + "num_tokens": 149960657.0, + "step": 3929 + }, + { + "epoch": 0.49993639486070474, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8045686483383179, + "num_tokens": 149995983.0, + "step": 3930 + }, + { + "epoch": 0.5000636051392953, + "grad_norm": 84.5, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7703673839569092, + "num_tokens": 150037508.0, + "step": 3931 + }, + { + "epoch": 0.5001908154178858, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7889317870140076, + "num_tokens": 150077806.0, + "step": 3932 + }, + { + "epoch": 0.5003180256964763, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7942033410072327, + "num_tokens": 150117589.0, + "step": 3933 + }, + { + "epoch": 0.5004452359750667, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.792424201965332, + "num_tokens": 150161579.0, + "step": 3934 + }, + { + "epoch": 0.5005724462536573, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7795686721801758, + "num_tokens": 150208018.0, + "step": 3935 + }, + { + "epoch": 0.5006996565322478, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7915071249008179, + "num_tokens": 150250164.0, + "step": 3936 + }, + { + "epoch": 0.5008268668108383, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8158639669418335, + "num_tokens": 150289972.0, + "step": 3937 + }, + { + "epoch": 0.5009540770894289, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7947462201118469, + "num_tokens": 150333656.0, + "step": 3938 + }, + { + "epoch": 0.5010812873680194, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8001132011413574, + "num_tokens": 150374086.0, + "step": 3939 + }, + { + "epoch": 0.5012084976466098, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.794429361820221, + "num_tokens": 150407472.0, + "step": 3940 + }, + { + "epoch": 0.5013357079252003, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.8007315993309021, + "num_tokens": 150444560.0, + "step": 3941 + }, + { + "epoch": 0.5014629182037909, + "grad_norm": 77.0, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7987103462219238, + "num_tokens": 150483621.0, + "step": 3942 + }, + { + "epoch": 0.5015901284823814, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7973295450210571, + "num_tokens": 150520414.0, + "step": 3943 + }, + { + "epoch": 0.5017173387609719, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7854886054992676, + "num_tokens": 150556589.0, + "step": 3944 + }, + { + "epoch": 0.5018445490395624, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7891652584075928, + "num_tokens": 150592393.0, + "step": 3945 + }, + { + "epoch": 0.5019717593181529, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7800086736679077, + "num_tokens": 150630457.0, + "step": 3946 + }, + { + "epoch": 0.5020989695967434, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7991645932197571, + "num_tokens": 150659482.0, + "step": 3947 + }, + { + "epoch": 0.5022261798753339, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.795549213886261, + "num_tokens": 150700920.0, + "step": 3948 + }, + { + "epoch": 0.5023533901539244, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8021012544631958, + "num_tokens": 150736393.0, + "step": 3949 + }, + { + "epoch": 0.502480600432515, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7927030920982361, + "num_tokens": 150783848.0, + "step": 3950 + }, + { + "epoch": 0.5026078107111055, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7895840406417847, + "num_tokens": 150824179.0, + "step": 3951 + }, + { + "epoch": 0.5027350209896959, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7886353731155396, + "num_tokens": 150862775.0, + "step": 3952 + }, + { + "epoch": 0.5028622312682864, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7912545204162598, + "num_tokens": 150900005.0, + "step": 3953 + }, + { + "epoch": 0.502989441546877, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7873599529266357, + "num_tokens": 150936148.0, + "step": 3954 + }, + { + "epoch": 0.5031166518254675, + "grad_norm": 91.5, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7895538210868835, + "num_tokens": 150971661.0, + "step": 3955 + }, + { + "epoch": 0.503243862104058, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8016278743743896, + "num_tokens": 151009521.0, + "step": 3956 + }, + { + "epoch": 0.5033710723826486, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7882487773895264, + "num_tokens": 151053641.0, + "step": 3957 + }, + { + "epoch": 0.5034982826612391, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7943034172058105, + "num_tokens": 151094476.0, + "step": 3958 + }, + { + "epoch": 0.5036254929398295, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7871515154838562, + "num_tokens": 151126381.0, + "step": 3959 + }, + { + "epoch": 0.50375270321842, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7942546010017395, + "num_tokens": 151156875.0, + "step": 3960 + }, + { + "epoch": 0.5038799134970106, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7891194224357605, + "num_tokens": 151194394.0, + "step": 3961 + }, + { + "epoch": 0.5040071237756011, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.775290846824646, + "num_tokens": 151234472.0, + "step": 3962 + }, + { + "epoch": 0.5041343340541916, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7865076661109924, + "num_tokens": 151270426.0, + "step": 3963 + }, + { + "epoch": 0.5042615443327821, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7886407971382141, + "num_tokens": 151305793.0, + "step": 3964 + }, + { + "epoch": 0.5043887546113726, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7798731923103333, + "num_tokens": 151345116.0, + "step": 3965 + }, + { + "epoch": 0.5045159648899631, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7987338304519653, + "num_tokens": 151381901.0, + "step": 3966 + }, + { + "epoch": 0.5046431751685536, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8012372851371765, + "num_tokens": 151426884.0, + "step": 3967 + }, + { + "epoch": 0.5047703854471441, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7853593826293945, + "num_tokens": 151467596.0, + "step": 3968 + }, + { + "epoch": 0.5048975957257347, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8101024627685547, + "num_tokens": 151502131.0, + "step": 3969 + }, + { + "epoch": 0.5050248060043252, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7844448685646057, + "num_tokens": 151538932.0, + "step": 3970 + }, + { + "epoch": 0.5051520162829156, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7926591634750366, + "num_tokens": 151577839.0, + "step": 3971 + }, + { + "epoch": 0.5052792265615061, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7900041341781616, + "num_tokens": 151618984.0, + "step": 3972 + }, + { + "epoch": 0.5054064368400967, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7936345338821411, + "num_tokens": 151658392.0, + "step": 3973 + }, + { + "epoch": 0.5055336471186872, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.8022486567497253, + "num_tokens": 151689781.0, + "step": 3974 + }, + { + "epoch": 0.5056608573972777, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7774441838264465, + "num_tokens": 151729612.0, + "step": 3975 + }, + { + "epoch": 0.5057880676758683, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.782534658908844, + "num_tokens": 151769270.0, + "step": 3976 + }, + { + "epoch": 0.5059152779544587, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7708770036697388, + "num_tokens": 151803936.0, + "step": 3977 + }, + { + "epoch": 0.5060424882330492, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7835179567337036, + "num_tokens": 151840429.0, + "step": 3978 + }, + { + "epoch": 0.5061696985116397, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.797681987285614, + "num_tokens": 151883296.0, + "step": 3979 + }, + { + "epoch": 0.5062969087902303, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8094609975814819, + "num_tokens": 151917658.0, + "step": 3980 + }, + { + "epoch": 0.5064241190688208, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7922526001930237, + "num_tokens": 151958822.0, + "step": 3981 + }, + { + "epoch": 0.5065513293474113, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7875579595565796, + "num_tokens": 151994964.0, + "step": 3982 + }, + { + "epoch": 0.5066785396260017, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7869434952735901, + "num_tokens": 152037013.0, + "step": 3983 + }, + { + "epoch": 0.5068057499045923, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7894378900527954, + "num_tokens": 152074821.0, + "step": 3984 + }, + { + "epoch": 0.5069329601831828, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7886893153190613, + "num_tokens": 152116667.0, + "step": 3985 + }, + { + "epoch": 0.5070601704617733, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7946768999099731, + "num_tokens": 152152714.0, + "step": 3986 + }, + { + "epoch": 0.5071873807403638, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7765217423439026, + "num_tokens": 152190230.0, + "step": 3987 + }, + { + "epoch": 0.5073145910189544, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7918145656585693, + "num_tokens": 152228887.0, + "step": 3988 + }, + { + "epoch": 0.5074418012975448, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8042707443237305, + "num_tokens": 152271863.0, + "step": 3989 + }, + { + "epoch": 0.5075690115761353, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7803128361701965, + "num_tokens": 152310765.0, + "step": 3990 + }, + { + "epoch": 0.5076962218547258, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8030411005020142, + "num_tokens": 152342545.0, + "step": 3991 + }, + { + "epoch": 0.5078234321333164, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.778020977973938, + "num_tokens": 152386223.0, + "step": 3992 + }, + { + "epoch": 0.5079506424119069, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.8004240989685059, + "num_tokens": 152421811.0, + "step": 3993 + }, + { + "epoch": 0.5080778526904974, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7712678909301758, + "num_tokens": 152454581.0, + "step": 3994 + }, + { + "epoch": 0.5082050629690879, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7795884013175964, + "num_tokens": 152490591.0, + "step": 3995 + }, + { + "epoch": 0.5083322732476784, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.818117618560791, + "num_tokens": 152532290.0, + "step": 3996 + }, + { + "epoch": 0.5084594835262689, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7979733943939209, + "num_tokens": 152571984.0, + "step": 3997 + }, + { + "epoch": 0.5085866938048594, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7901821732521057, + "num_tokens": 152611082.0, + "step": 3998 + }, + { + "epoch": 0.50871390408345, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8024295568466187, + "num_tokens": 152655303.0, + "step": 3999 + }, + { + "epoch": 0.5088411143620405, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7886459827423096, + "num_tokens": 152688432.0, + "step": 4000 + }, + { + "epoch": 0.5089683246406309, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7859463095664978, + "num_tokens": 152725637.0, + "step": 4001 + }, + { + "epoch": 0.5090955349192214, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7902777791023254, + "num_tokens": 152763108.0, + "step": 4002 + }, + { + "epoch": 0.509222745197812, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7995038628578186, + "num_tokens": 152800659.0, + "step": 4003 + }, + { + "epoch": 0.5093499554764025, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7858078479766846, + "num_tokens": 152842797.0, + "step": 4004 + }, + { + "epoch": 0.509477165754993, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7842458486557007, + "num_tokens": 152888257.0, + "step": 4005 + }, + { + "epoch": 0.5096043760335836, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7959838509559631, + "num_tokens": 152929089.0, + "step": 4006 + }, + { + "epoch": 0.5097315863121741, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7851302623748779, + "num_tokens": 152972710.0, + "step": 4007 + }, + { + "epoch": 0.5098587965907645, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7934550046920776, + "num_tokens": 153008867.0, + "step": 4008 + }, + { + "epoch": 0.509986006869355, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7870643734931946, + "num_tokens": 153051497.0, + "step": 4009 + }, + { + "epoch": 0.5101132171479456, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8089612126350403, + "num_tokens": 153089295.0, + "step": 4010 + }, + { + "epoch": 0.5102404274265361, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.8002682328224182, + "num_tokens": 153123957.0, + "step": 4011 + }, + { + "epoch": 0.5103676377051266, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7875767350196838, + "num_tokens": 153165248.0, + "step": 4012 + }, + { + "epoch": 0.5104948479837171, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8067737817764282, + "num_tokens": 153206163.0, + "step": 4013 + }, + { + "epoch": 0.5106220582623076, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7852935791015625, + "num_tokens": 153248666.0, + "step": 4014 + }, + { + "epoch": 0.5107492685408981, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8146556615829468, + "num_tokens": 153289009.0, + "step": 4015 + }, + { + "epoch": 0.5108764788194886, + "grad_norm": 111.5, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7940650582313538, + "num_tokens": 153321354.0, + "step": 4016 + }, + { + "epoch": 0.5110036890980791, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7953250408172607, + "num_tokens": 153361061.0, + "step": 4017 + }, + { + "epoch": 0.5111308993766697, + "grad_norm": 97.0, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7843615412712097, + "num_tokens": 153397002.0, + "step": 4018 + }, + { + "epoch": 0.5112581096552602, + "grad_norm": 154.0, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7746559381484985, + "num_tokens": 153435264.0, + "step": 4019 + }, + { + "epoch": 0.5113853199338506, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.791205883026123, + "num_tokens": 153474751.0, + "step": 4020 + }, + { + "epoch": 0.5115125302124411, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.799917995929718, + "num_tokens": 153510084.0, + "step": 4021 + }, + { + "epoch": 0.5116397404910317, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7772022485733032, + "num_tokens": 153550387.0, + "step": 4022 + }, + { + "epoch": 0.5117669507696222, + "grad_norm": 132.0, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.8163755536079407, + "num_tokens": 153592441.0, + "step": 4023 + }, + { + "epoch": 0.5118941610482127, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8020816445350647, + "num_tokens": 153629202.0, + "step": 4024 + }, + { + "epoch": 0.5120213713268033, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7896648049354553, + "num_tokens": 153669243.0, + "step": 4025 + }, + { + "epoch": 0.5121485816053937, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7873259782791138, + "num_tokens": 153709206.0, + "step": 4026 + }, + { + "epoch": 0.5122757918839842, + "grad_norm": 175.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7990434169769287, + "num_tokens": 153741251.0, + "step": 4027 + }, + { + "epoch": 0.5124030021625747, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.779790997505188, + "num_tokens": 153775559.0, + "step": 4028 + }, + { + "epoch": 0.5125302124411653, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7984870076179504, + "num_tokens": 153819158.0, + "step": 4029 + }, + { + "epoch": 0.5126574227197558, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7835251092910767, + "num_tokens": 153856231.0, + "step": 4030 + }, + { + "epoch": 0.5127846329983463, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8039554357528687, + "num_tokens": 153897125.0, + "step": 4031 + }, + { + "epoch": 0.5129118432769367, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7961562871932983, + "num_tokens": 153933552.0, + "step": 4032 + }, + { + "epoch": 0.5130390535555273, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7775475978851318, + "num_tokens": 153974337.0, + "step": 4033 + }, + { + "epoch": 0.5131662638341178, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7869446873664856, + "num_tokens": 154018911.0, + "step": 4034 + }, + { + "epoch": 0.5132934741127083, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8069190979003906, + "num_tokens": 154056441.0, + "step": 4035 + }, + { + "epoch": 0.5134206843912988, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7882537841796875, + "num_tokens": 154096326.0, + "step": 4036 + }, + { + "epoch": 0.5135478946698894, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7955489158630371, + "num_tokens": 154137223.0, + "step": 4037 + }, + { + "epoch": 0.5136751049484798, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7880721092224121, + "num_tokens": 154168312.0, + "step": 4038 + }, + { + "epoch": 0.5138023152270703, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8045584559440613, + "num_tokens": 154203125.0, + "step": 4039 + }, + { + "epoch": 0.5139295255056608, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7857101559638977, + "num_tokens": 154248437.0, + "step": 4040 + }, + { + "epoch": 0.5140567357842514, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.796485185623169, + "num_tokens": 154286303.0, + "step": 4041 + }, + { + "epoch": 0.5141839460628419, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7817096710205078, + "num_tokens": 154328638.0, + "step": 4042 + }, + { + "epoch": 0.5143111563414324, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7886159420013428, + "num_tokens": 154362953.0, + "step": 4043 + }, + { + "epoch": 0.5144383666200228, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7758033275604248, + "num_tokens": 154401289.0, + "step": 4044 + }, + { + "epoch": 0.5145655768986134, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7970175743103027, + "num_tokens": 154438779.0, + "step": 4045 + }, + { + "epoch": 0.5146927871772039, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7830454707145691, + "num_tokens": 154467032.0, + "step": 4046 + }, + { + "epoch": 0.5148199974557944, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8036168217658997, + "num_tokens": 154505222.0, + "step": 4047 + }, + { + "epoch": 0.514947207734385, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7783914804458618, + "num_tokens": 154544953.0, + "step": 4048 + }, + { + "epoch": 0.5150744180129755, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8083157539367676, + "num_tokens": 154586193.0, + "step": 4049 + }, + { + "epoch": 0.5152016282915659, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.8001195192337036, + "num_tokens": 154629531.0, + "step": 4050 + }, + { + "epoch": 0.5153288385701564, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8041324615478516, + "num_tokens": 154659482.0, + "step": 4051 + }, + { + "epoch": 0.515456048848747, + "grad_norm": 84.0, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7585408687591553, + "num_tokens": 154702645.0, + "step": 4052 + }, + { + "epoch": 0.5155832591273375, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8022291660308838, + "num_tokens": 154743329.0, + "step": 4053 + }, + { + "epoch": 0.515710469405928, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7901084423065186, + "num_tokens": 154783460.0, + "step": 4054 + }, + { + "epoch": 0.5158376796845185, + "grad_norm": 117.0, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.792908251285553, + "num_tokens": 154827209.0, + "step": 4055 + }, + { + "epoch": 0.5159648899631091, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7682613730430603, + "num_tokens": 154866577.0, + "step": 4056 + }, + { + "epoch": 0.5160921002416995, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7963575124740601, + "num_tokens": 154912191.0, + "step": 4057 + }, + { + "epoch": 0.51621931052029, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7799760103225708, + "num_tokens": 154953412.0, + "step": 4058 + }, + { + "epoch": 0.5163465207988805, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8008015751838684, + "num_tokens": 154989913.0, + "step": 4059 + }, + { + "epoch": 0.5164737310774711, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7829023599624634, + "num_tokens": 155038094.0, + "step": 4060 + }, + { + "epoch": 0.5166009413560616, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7844437956809998, + "num_tokens": 155075900.0, + "step": 4061 + }, + { + "epoch": 0.5167281516346521, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.789177417755127, + "num_tokens": 155117038.0, + "step": 4062 + }, + { + "epoch": 0.5168553619132426, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8026281595230103, + "num_tokens": 155151442.0, + "step": 4063 + }, + { + "epoch": 0.5169825721918331, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.801386833190918, + "num_tokens": 155192288.0, + "step": 4064 + }, + { + "epoch": 0.5171097824704236, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7997286319732666, + "num_tokens": 155238931.0, + "step": 4065 + }, + { + "epoch": 0.5172369927490141, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8071249723434448, + "num_tokens": 155276644.0, + "step": 4066 + }, + { + "epoch": 0.5173642030276047, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8045192360877991, + "num_tokens": 155307638.0, + "step": 4067 + }, + { + "epoch": 0.5174914133061952, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7802777290344238, + "num_tokens": 155343255.0, + "step": 4068 + }, + { + "epoch": 0.5176186235847856, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7928245663642883, + "num_tokens": 155382283.0, + "step": 4069 + }, + { + "epoch": 0.5177458338633761, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7724912166595459, + "num_tokens": 155427096.0, + "step": 4070 + }, + { + "epoch": 0.5178730441419667, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7799836993217468, + "num_tokens": 155467472.0, + "step": 4071 + }, + { + "epoch": 0.5180002544205572, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7930921316146851, + "num_tokens": 155508350.0, + "step": 4072 + }, + { + "epoch": 0.5181274646991477, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8173836469650269, + "num_tokens": 155547215.0, + "step": 4073 + }, + { + "epoch": 0.5182546749777382, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7949783802032471, + "num_tokens": 155586658.0, + "step": 4074 + }, + { + "epoch": 0.5183818852563287, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7921181321144104, + "num_tokens": 155620944.0, + "step": 4075 + }, + { + "epoch": 0.5185090955349192, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7937296628952026, + "num_tokens": 155662699.0, + "step": 4076 + }, + { + "epoch": 0.5186363058135097, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7825449705123901, + "num_tokens": 155707783.0, + "step": 4077 + }, + { + "epoch": 0.5187635160921003, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7912175059318542, + "num_tokens": 155745227.0, + "step": 4078 + }, + { + "epoch": 0.5188907263706908, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7968547344207764, + "num_tokens": 155786466.0, + "step": 4079 + }, + { + "epoch": 0.5190179366492813, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.794837474822998, + "num_tokens": 155833145.0, + "step": 4080 + }, + { + "epoch": 0.5191451469278717, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7841755747795105, + "num_tokens": 155869533.0, + "step": 4081 + }, + { + "epoch": 0.5192723572064623, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7900634407997131, + "num_tokens": 155909329.0, + "step": 4082 + }, + { + "epoch": 0.5193995674850528, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8055645823478699, + "num_tokens": 155947848.0, + "step": 4083 + }, + { + "epoch": 0.5195267777636433, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7922091484069824, + "num_tokens": 155982693.0, + "step": 4084 + }, + { + "epoch": 0.5196539880422338, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7942187786102295, + "num_tokens": 156023414.0, + "step": 4085 + }, + { + "epoch": 0.5197811983208244, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7930728793144226, + "num_tokens": 156063488.0, + "step": 4086 + }, + { + "epoch": 0.5199084085994148, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7893767952919006, + "num_tokens": 156103081.0, + "step": 4087 + }, + { + "epoch": 0.5200356188780053, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7802078127861023, + "num_tokens": 156139453.0, + "step": 4088 + }, + { + "epoch": 0.5201628291565958, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7880704402923584, + "num_tokens": 156176564.0, + "step": 4089 + }, + { + "epoch": 0.5202900394351864, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7971565127372742, + "num_tokens": 156221278.0, + "step": 4090 + }, + { + "epoch": 0.5204172497137769, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.801977276802063, + "num_tokens": 156259124.0, + "step": 4091 + }, + { + "epoch": 0.5205444599923674, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7794793844223022, + "num_tokens": 156297832.0, + "step": 4092 + }, + { + "epoch": 0.5206716702709578, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7946452498435974, + "num_tokens": 156329745.0, + "step": 4093 + }, + { + "epoch": 0.5207988805495484, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7856485247612, + "num_tokens": 156360154.0, + "step": 4094 + }, + { + "epoch": 0.5209260908281389, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7951618432998657, + "num_tokens": 156399360.0, + "step": 4095 + }, + { + "epoch": 0.5210533011067294, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7922106981277466, + "num_tokens": 156438782.0, + "step": 4096 + }, + { + "epoch": 0.52118051138532, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7927002310752869, + "num_tokens": 156470261.0, + "step": 4097 + }, + { + "epoch": 0.5213077216639105, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7727979421615601, + "num_tokens": 156510493.0, + "step": 4098 + }, + { + "epoch": 0.5214349319425009, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7864367961883545, + "num_tokens": 156545040.0, + "step": 4099 + }, + { + "epoch": 0.5215621422210914, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7943634986877441, + "num_tokens": 156583474.0, + "step": 4100 + }, + { + "epoch": 0.521689352499682, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7983136773109436, + "num_tokens": 156627137.0, + "step": 4101 + }, + { + "epoch": 0.5218165627782725, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7810118198394775, + "num_tokens": 156664726.0, + "step": 4102 + }, + { + "epoch": 0.521943773056863, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8123880624771118, + "num_tokens": 156707014.0, + "step": 4103 + }, + { + "epoch": 0.5220709833354535, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7877901792526245, + "num_tokens": 156747784.0, + "step": 4104 + }, + { + "epoch": 0.522198193614044, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7913335561752319, + "num_tokens": 156784760.0, + "step": 4105 + }, + { + "epoch": 0.5223254038926345, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7902024984359741, + "num_tokens": 156822899.0, + "step": 4106 + }, + { + "epoch": 0.522452614171225, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7961193323135376, + "num_tokens": 156861308.0, + "step": 4107 + }, + { + "epoch": 0.5225798244498155, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7984755039215088, + "num_tokens": 156901981.0, + "step": 4108 + }, + { + "epoch": 0.5227070347284061, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7960861921310425, + "num_tokens": 156942215.0, + "step": 4109 + }, + { + "epoch": 0.5228342450069966, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.798341691493988, + "num_tokens": 156983820.0, + "step": 4110 + }, + { + "epoch": 0.5229614552855871, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8058538436889648, + "num_tokens": 157016215.0, + "step": 4111 + }, + { + "epoch": 0.5230886655641775, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7891668081283569, + "num_tokens": 157052071.0, + "step": 4112 + }, + { + "epoch": 0.5232158758427681, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.770352840423584, + "num_tokens": 157097711.0, + "step": 4113 + }, + { + "epoch": 0.5233430861213586, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7948377132415771, + "num_tokens": 157134801.0, + "step": 4114 + }, + { + "epoch": 0.5234702963999491, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7876015901565552, + "num_tokens": 157171885.0, + "step": 4115 + }, + { + "epoch": 0.5235975066785397, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7891412973403931, + "num_tokens": 157211915.0, + "step": 4116 + }, + { + "epoch": 0.5237247169571302, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7866179347038269, + "num_tokens": 157260564.0, + "step": 4117 + }, + { + "epoch": 0.5238519272357206, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7933180332183838, + "num_tokens": 157294676.0, + "step": 4118 + }, + { + "epoch": 0.5239791375143111, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8019070029258728, + "num_tokens": 157334607.0, + "step": 4119 + }, + { + "epoch": 0.5241063477929017, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.791847825050354, + "num_tokens": 157372021.0, + "step": 4120 + }, + { + "epoch": 0.5242335580714922, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7953832149505615, + "num_tokens": 157410092.0, + "step": 4121 + }, + { + "epoch": 0.5243607683500827, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7839519381523132, + "num_tokens": 157447979.0, + "step": 4122 + }, + { + "epoch": 0.5244879786286732, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7837362289428711, + "num_tokens": 157481376.0, + "step": 4123 + }, + { + "epoch": 0.5246151889072637, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7994692921638489, + "num_tokens": 157519555.0, + "step": 4124 + }, + { + "epoch": 0.5247423991858542, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7853797674179077, + "num_tokens": 157551849.0, + "step": 4125 + }, + { + "epoch": 0.5248696094644447, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7928964495658875, + "num_tokens": 157590823.0, + "step": 4126 + }, + { + "epoch": 0.5249968197430352, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8044342398643494, + "num_tokens": 157633771.0, + "step": 4127 + }, + { + "epoch": 0.5251240300216258, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.808700680732727, + "num_tokens": 157667796.0, + "step": 4128 + }, + { + "epoch": 0.5252512403002163, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8035277128219604, + "num_tokens": 157709452.0, + "step": 4129 + }, + { + "epoch": 0.5253784505788067, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.777917742729187, + "num_tokens": 157741407.0, + "step": 4130 + }, + { + "epoch": 0.5255056608573972, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8027207255363464, + "num_tokens": 157777693.0, + "step": 4131 + }, + { + "epoch": 0.5256328711359878, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7873610258102417, + "num_tokens": 157814729.0, + "step": 4132 + }, + { + "epoch": 0.5257600814145783, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.8042689561843872, + "num_tokens": 157853698.0, + "step": 4133 + }, + { + "epoch": 0.5258872916931688, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8022735118865967, + "num_tokens": 157888580.0, + "step": 4134 + }, + { + "epoch": 0.5260145019717594, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7878082394599915, + "num_tokens": 157928434.0, + "step": 4135 + }, + { + "epoch": 0.5261417122503498, + "grad_norm": 117.0, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7865312099456787, + "num_tokens": 157968936.0, + "step": 4136 + }, + { + "epoch": 0.5262689225289403, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7835569977760315, + "num_tokens": 158005496.0, + "step": 4137 + }, + { + "epoch": 0.5263961328075308, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8037286996841431, + "num_tokens": 158047569.0, + "step": 4138 + }, + { + "epoch": 0.5265233430861214, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7916163802146912, + "num_tokens": 158085733.0, + "step": 4139 + }, + { + "epoch": 0.5266505533647119, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.782455563545227, + "num_tokens": 158123906.0, + "step": 4140 + }, + { + "epoch": 0.5267777636433024, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8010098934173584, + "num_tokens": 158163663.0, + "step": 4141 + }, + { + "epoch": 0.5269049739218928, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7811145782470703, + "num_tokens": 158205878.0, + "step": 4142 + }, + { + "epoch": 0.5270321842004834, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7977054715156555, + "num_tokens": 158241608.0, + "step": 4143 + }, + { + "epoch": 0.5271593944790739, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8084524869918823, + "num_tokens": 158274259.0, + "step": 4144 + }, + { + "epoch": 0.5272866047576644, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7997794151306152, + "num_tokens": 158304793.0, + "step": 4145 + }, + { + "epoch": 0.527413815036255, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8082084655761719, + "num_tokens": 158349420.0, + "step": 4146 + }, + { + "epoch": 0.5275410253148455, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7925922274589539, + "num_tokens": 158395834.0, + "step": 4147 + }, + { + "epoch": 0.5276682355934359, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7985687255859375, + "num_tokens": 158435568.0, + "step": 4148 + }, + { + "epoch": 0.5277954458720264, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8033249378204346, + "num_tokens": 158473584.0, + "step": 4149 + }, + { + "epoch": 0.527922656150617, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7941297888755798, + "num_tokens": 158513435.0, + "step": 4150 + }, + { + "epoch": 0.5280498664292075, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7924925684928894, + "num_tokens": 158549478.0, + "step": 4151 + }, + { + "epoch": 0.528177076707798, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8050788640975952, + "num_tokens": 158598742.0, + "step": 4152 + }, + { + "epoch": 0.5283042869863885, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7815014719963074, + "num_tokens": 158636806.0, + "step": 4153 + }, + { + "epoch": 0.528431497264979, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7849986553192139, + "num_tokens": 158676043.0, + "step": 4154 + }, + { + "epoch": 0.5285587075435695, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7764754891395569, + "num_tokens": 158711772.0, + "step": 4155 + }, + { + "epoch": 0.52868591782216, + "grad_norm": 91.5, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7873533964157104, + "num_tokens": 158749608.0, + "step": 4156 + }, + { + "epoch": 0.5288131281007505, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7951236963272095, + "num_tokens": 158790502.0, + "step": 4157 + }, + { + "epoch": 0.5289403383793411, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7711194753646851, + "num_tokens": 158828415.0, + "step": 4158 + }, + { + "epoch": 0.5290675486579316, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7857149243354797, + "num_tokens": 158866972.0, + "step": 4159 + }, + { + "epoch": 0.5291947589365221, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7879747152328491, + "num_tokens": 158902991.0, + "step": 4160 + }, + { + "epoch": 0.5293219692151125, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8065915107727051, + "num_tokens": 158945963.0, + "step": 4161 + }, + { + "epoch": 0.5294491794937031, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7917829751968384, + "num_tokens": 158985407.0, + "step": 4162 + }, + { + "epoch": 0.5295763897722936, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7862139940261841, + "num_tokens": 159024387.0, + "step": 4163 + }, + { + "epoch": 0.5297036000508841, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.782315731048584, + "num_tokens": 159061430.0, + "step": 4164 + }, + { + "epoch": 0.5298308103294747, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7959593534469604, + "num_tokens": 159099350.0, + "step": 4165 + }, + { + "epoch": 0.5299580206080652, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7660994529724121, + "num_tokens": 159136276.0, + "step": 4166 + }, + { + "epoch": 0.5300852308866556, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8186063766479492, + "num_tokens": 159177102.0, + "step": 4167 + }, + { + "epoch": 0.5302124411652461, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8049983978271484, + "num_tokens": 159213582.0, + "step": 4168 + }, + { + "epoch": 0.5303396514438367, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7949017286300659, + "num_tokens": 159254283.0, + "step": 4169 + }, + { + "epoch": 0.5304668617224272, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7968342900276184, + "num_tokens": 159288557.0, + "step": 4170 + }, + { + "epoch": 0.5305940720010177, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8055932521820068, + "num_tokens": 159325047.0, + "step": 4171 + }, + { + "epoch": 0.5307212822796082, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7975971102714539, + "num_tokens": 159366291.0, + "step": 4172 + }, + { + "epoch": 0.5308484925581987, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7883015871047974, + "num_tokens": 159401634.0, + "step": 4173 + }, + { + "epoch": 0.5309757028367892, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7850997447967529, + "num_tokens": 159435316.0, + "step": 4174 + }, + { + "epoch": 0.5311029131153797, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7903363704681396, + "num_tokens": 159472192.0, + "step": 4175 + }, + { + "epoch": 0.5312301233939702, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7998445630073547, + "num_tokens": 159515342.0, + "step": 4176 + }, + { + "epoch": 0.5313573336725608, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7972978949546814, + "num_tokens": 159553932.0, + "step": 4177 + }, + { + "epoch": 0.5314845439511513, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7914438247680664, + "num_tokens": 159584711.0, + "step": 4178 + }, + { + "epoch": 0.5316117542297417, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.77860027551651, + "num_tokens": 159618784.0, + "step": 4179 + }, + { + "epoch": 0.5317389645083322, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.789588451385498, + "num_tokens": 159661288.0, + "step": 4180 + }, + { + "epoch": 0.5318661747869228, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8031842708587646, + "num_tokens": 159701286.0, + "step": 4181 + }, + { + "epoch": 0.5319933850655133, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7819961309432983, + "num_tokens": 159746631.0, + "step": 4182 + }, + { + "epoch": 0.5321205953441038, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7980020046234131, + "num_tokens": 159787753.0, + "step": 4183 + }, + { + "epoch": 0.5322478056226944, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7872458696365356, + "num_tokens": 159823092.0, + "step": 4184 + }, + { + "epoch": 0.5323750159012848, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7810367345809937, + "num_tokens": 159858480.0, + "step": 4185 + }, + { + "epoch": 0.5325022261798753, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7829563617706299, + "num_tokens": 159896220.0, + "step": 4186 + }, + { + "epoch": 0.5326294364584658, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7888128757476807, + "num_tokens": 159930995.0, + "step": 4187 + }, + { + "epoch": 0.5327566467370564, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7901080250740051, + "num_tokens": 159968013.0, + "step": 4188 + }, + { + "epoch": 0.5328838570156469, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7844394445419312, + "num_tokens": 160004085.0, + "step": 4189 + }, + { + "epoch": 0.5330110672942374, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8048017621040344, + "num_tokens": 160040606.0, + "step": 4190 + }, + { + "epoch": 0.5331382775728278, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8040651082992554, + "num_tokens": 160080792.0, + "step": 4191 + }, + { + "epoch": 0.5332654878514184, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7822604775428772, + "num_tokens": 160117055.0, + "step": 4192 + }, + { + "epoch": 0.5333926981300089, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7967828512191772, + "num_tokens": 160153223.0, + "step": 4193 + }, + { + "epoch": 0.5335199084085994, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7780062556266785, + "num_tokens": 160190920.0, + "step": 4194 + }, + { + "epoch": 0.53364711868719, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.786855161190033, + "num_tokens": 160221898.0, + "step": 4195 + }, + { + "epoch": 0.5337743289657805, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.8007774353027344, + "num_tokens": 160267102.0, + "step": 4196 + }, + { + "epoch": 0.5339015392443709, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7940587997436523, + "num_tokens": 160304422.0, + "step": 4197 + }, + { + "epoch": 0.5340287495229614, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7897043228149414, + "num_tokens": 160339006.0, + "step": 4198 + }, + { + "epoch": 0.534155959801552, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8129004240036011, + "num_tokens": 160377504.0, + "step": 4199 + }, + { + "epoch": 0.5342831700801425, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7715839147567749, + "num_tokens": 160413315.0, + "step": 4200 + }, + { + "epoch": 0.534410380358733, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.80913245677948, + "num_tokens": 160455034.0, + "step": 4201 + }, + { + "epoch": 0.5345375906373235, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7796417474746704, + "num_tokens": 160490505.0, + "step": 4202 + }, + { + "epoch": 0.534664800915914, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8060746192932129, + "num_tokens": 160528712.0, + "step": 4203 + }, + { + "epoch": 0.5347920111945045, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7864225506782532, + "num_tokens": 160567945.0, + "step": 4204 + }, + { + "epoch": 0.534919221473095, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7913976311683655, + "num_tokens": 160596767.0, + "step": 4205 + }, + { + "epoch": 0.5350464317516855, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7874009609222412, + "num_tokens": 160636912.0, + "step": 4206 + }, + { + "epoch": 0.5351736420302761, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7875161170959473, + "num_tokens": 160674588.0, + "step": 4207 + }, + { + "epoch": 0.5353008523088666, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.783600926399231, + "num_tokens": 160713441.0, + "step": 4208 + }, + { + "epoch": 0.5354280625874571, + "grad_norm": 101.0, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.8005704283714294, + "num_tokens": 160752716.0, + "step": 4209 + }, + { + "epoch": 0.5355552728660475, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8068767786026001, + "num_tokens": 160798771.0, + "step": 4210 + }, + { + "epoch": 0.5356824831446381, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8135710954666138, + "num_tokens": 160843085.0, + "step": 4211 + }, + { + "epoch": 0.5358096934232286, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7866222858428955, + "num_tokens": 160883073.0, + "step": 4212 + }, + { + "epoch": 0.5359369037018191, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7997976541519165, + "num_tokens": 160923734.0, + "step": 4213 + }, + { + "epoch": 0.5360641139804097, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7948466539382935, + "num_tokens": 160963460.0, + "step": 4214 + }, + { + "epoch": 0.5361913242590002, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7907079458236694, + "num_tokens": 161003691.0, + "step": 4215 + }, + { + "epoch": 0.5363185345375906, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7929520606994629, + "num_tokens": 161042056.0, + "step": 4216 + }, + { + "epoch": 0.5364457448161811, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8196747303009033, + "num_tokens": 161078670.0, + "step": 4217 + }, + { + "epoch": 0.5365729550947717, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8044910430908203, + "num_tokens": 161113992.0, + "step": 4218 + }, + { + "epoch": 0.5367001653733622, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7994036674499512, + "num_tokens": 161153763.0, + "step": 4219 + }, + { + "epoch": 0.5368273756519527, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7767882347106934, + "num_tokens": 161184186.0, + "step": 4220 + }, + { + "epoch": 0.5369545859305432, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7905601263046265, + "num_tokens": 161219298.0, + "step": 4221 + }, + { + "epoch": 0.5370817962091337, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8075417876243591, + "num_tokens": 161255252.0, + "step": 4222 + }, + { + "epoch": 0.5372090064877242, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7989346981048584, + "num_tokens": 161295890.0, + "step": 4223 + }, + { + "epoch": 0.5373362167663147, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7910545468330383, + "num_tokens": 161330808.0, + "step": 4224 + }, + { + "epoch": 0.5374634270449052, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8010814785957336, + "num_tokens": 161370515.0, + "step": 4225 + }, + { + "epoch": 0.5375906373234958, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8049993515014648, + "num_tokens": 161405257.0, + "step": 4226 + }, + { + "epoch": 0.5377178476020863, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.793439507484436, + "num_tokens": 161443526.0, + "step": 4227 + }, + { + "epoch": 0.5378450578806767, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7882476449012756, + "num_tokens": 161481262.0, + "step": 4228 + }, + { + "epoch": 0.5379722681592672, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7799350023269653, + "num_tokens": 161517756.0, + "step": 4229 + }, + { + "epoch": 0.5380994784378578, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8008260726928711, + "num_tokens": 161552680.0, + "step": 4230 + }, + { + "epoch": 0.5382266887164483, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7856075167655945, + "num_tokens": 161588607.0, + "step": 4231 + }, + { + "epoch": 0.5383538989950388, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7693725824356079, + "num_tokens": 161626427.0, + "step": 4232 + }, + { + "epoch": 0.5384811092736294, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7986562252044678, + "num_tokens": 161665182.0, + "step": 4233 + }, + { + "epoch": 0.5386083195522198, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7950664162635803, + "num_tokens": 161704685.0, + "step": 4234 + }, + { + "epoch": 0.5387355298308103, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7952117919921875, + "num_tokens": 161742960.0, + "step": 4235 + }, + { + "epoch": 0.5388627401094008, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7908915281295776, + "num_tokens": 161778199.0, + "step": 4236 + }, + { + "epoch": 0.5389899503879914, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7882604598999023, + "num_tokens": 161811235.0, + "step": 4237 + }, + { + "epoch": 0.5391171606665819, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.78629469871521, + "num_tokens": 161847669.0, + "step": 4238 + }, + { + "epoch": 0.5392443709451724, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7765753269195557, + "num_tokens": 161888470.0, + "step": 4239 + }, + { + "epoch": 0.5393715812237628, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7694286108016968, + "num_tokens": 161924563.0, + "step": 4240 + }, + { + "epoch": 0.5394987915023534, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7760481238365173, + "num_tokens": 161959062.0, + "step": 4241 + }, + { + "epoch": 0.5396260017809439, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.797853946685791, + "num_tokens": 161994755.0, + "step": 4242 + }, + { + "epoch": 0.5397532120595344, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8021409511566162, + "num_tokens": 162032812.0, + "step": 4243 + }, + { + "epoch": 0.5398804223381249, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7961639165878296, + "num_tokens": 162067130.0, + "step": 4244 + }, + { + "epoch": 0.5400076326167155, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.794622540473938, + "num_tokens": 162107452.0, + "step": 4245 + }, + { + "epoch": 0.5401348428953059, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7974541783332825, + "num_tokens": 162144220.0, + "step": 4246 + }, + { + "epoch": 0.5402620531738964, + "grad_norm": 104.0, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.8020450472831726, + "num_tokens": 162181798.0, + "step": 4247 + }, + { + "epoch": 0.540389263452487, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7812503576278687, + "num_tokens": 162222501.0, + "step": 4248 + }, + { + "epoch": 0.5405164737310775, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7904000282287598, + "num_tokens": 162265533.0, + "step": 4249 + }, + { + "epoch": 0.540643684009668, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8158501386642456, + "num_tokens": 162304984.0, + "step": 4250 + }, + { + "epoch": 0.5407708942882585, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7810766696929932, + "num_tokens": 162342358.0, + "step": 4251 + }, + { + "epoch": 0.540898104566849, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7955460548400879, + "num_tokens": 162377343.0, + "step": 4252 + }, + { + "epoch": 0.5410253148454395, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7942138910293579, + "num_tokens": 162416416.0, + "step": 4253 + }, + { + "epoch": 0.54115252512403, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.8230330348014832, + "num_tokens": 162447108.0, + "step": 4254 + }, + { + "epoch": 0.5412797354026205, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7968226075172424, + "num_tokens": 162489659.0, + "step": 4255 + }, + { + "epoch": 0.5414069456812111, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7809344530105591, + "num_tokens": 162525816.0, + "step": 4256 + }, + { + "epoch": 0.5415341559598016, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7943969368934631, + "num_tokens": 162564154.0, + "step": 4257 + }, + { + "epoch": 0.5416613662383921, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7896310091018677, + "num_tokens": 162602728.0, + "step": 4258 + }, + { + "epoch": 0.5417885765169825, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.8007632493972778, + "num_tokens": 162636659.0, + "step": 4259 + }, + { + "epoch": 0.5419157867955731, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8103269338607788, + "num_tokens": 162676639.0, + "step": 4260 + }, + { + "epoch": 0.5420429970741636, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7741541862487793, + "num_tokens": 162716763.0, + "step": 4261 + }, + { + "epoch": 0.5421702073527541, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.780194878578186, + "num_tokens": 162753943.0, + "step": 4262 + }, + { + "epoch": 0.5422974176313446, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7935456037521362, + "num_tokens": 162792938.0, + "step": 4263 + }, + { + "epoch": 0.5424246279099352, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7937774658203125, + "num_tokens": 162838938.0, + "step": 4264 + }, + { + "epoch": 0.5425518381885256, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7943947911262512, + "num_tokens": 162875026.0, + "step": 4265 + }, + { + "epoch": 0.5426790484671161, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7985074520111084, + "num_tokens": 162910724.0, + "step": 4266 + }, + { + "epoch": 0.5428062587457066, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7898276448249817, + "num_tokens": 162941602.0, + "step": 4267 + }, + { + "epoch": 0.5429334690242972, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7794094085693359, + "num_tokens": 162980760.0, + "step": 4268 + }, + { + "epoch": 0.5430606793028877, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.807563304901123, + "num_tokens": 163016155.0, + "step": 4269 + }, + { + "epoch": 0.5431878895814782, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7882276177406311, + "num_tokens": 163052519.0, + "step": 4270 + }, + { + "epoch": 0.5433150998600687, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.808712363243103, + "num_tokens": 163091806.0, + "step": 4271 + }, + { + "epoch": 0.5434423101386592, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8146939873695374, + "num_tokens": 163128830.0, + "step": 4272 + }, + { + "epoch": 0.5435695204172497, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7926107048988342, + "num_tokens": 163169567.0, + "step": 4273 + }, + { + "epoch": 0.5436967306958402, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7810643911361694, + "num_tokens": 163202913.0, + "step": 4274 + }, + { + "epoch": 0.5438239409744308, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.8001020550727844, + "num_tokens": 163236741.0, + "step": 4275 + }, + { + "epoch": 0.5439511512530213, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7919516563415527, + "num_tokens": 163281402.0, + "step": 4276 + }, + { + "epoch": 0.5440783615316117, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.807404637336731, + "num_tokens": 163325288.0, + "step": 4277 + }, + { + "epoch": 0.5442055718102022, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7871090173721313, + "num_tokens": 163367471.0, + "step": 4278 + }, + { + "epoch": 0.5443327820887928, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7913416624069214, + "num_tokens": 163400077.0, + "step": 4279 + }, + { + "epoch": 0.5444599923673833, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7986833453178406, + "num_tokens": 163433739.0, + "step": 4280 + }, + { + "epoch": 0.5445872026459738, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7940589189529419, + "num_tokens": 163467447.0, + "step": 4281 + }, + { + "epoch": 0.5447144129245644, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.788759708404541, + "num_tokens": 163506633.0, + "step": 4282 + }, + { + "epoch": 0.5448416232031548, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7754522562026978, + "num_tokens": 163542773.0, + "step": 4283 + }, + { + "epoch": 0.5449688334817453, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7590832710266113, + "num_tokens": 163585247.0, + "step": 4284 + }, + { + "epoch": 0.5450960437603358, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7773780822753906, + "num_tokens": 163622802.0, + "step": 4285 + }, + { + "epoch": 0.5452232540389264, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8207306861877441, + "num_tokens": 163655980.0, + "step": 4286 + }, + { + "epoch": 0.5453504643175169, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8022094964981079, + "num_tokens": 163691494.0, + "step": 4287 + }, + { + "epoch": 0.5454776745961074, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8043380975723267, + "num_tokens": 163725169.0, + "step": 4288 + }, + { + "epoch": 0.5456048848746978, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7976003289222717, + "num_tokens": 163766050.0, + "step": 4289 + }, + { + "epoch": 0.5457320951532884, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7878854274749756, + "num_tokens": 163799559.0, + "step": 4290 + }, + { + "epoch": 0.5458593054318789, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7897821068763733, + "num_tokens": 163837052.0, + "step": 4291 + }, + { + "epoch": 0.5459865157104694, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.788821816444397, + "num_tokens": 163872494.0, + "step": 4292 + }, + { + "epoch": 0.5461137259890599, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7873640060424805, + "num_tokens": 163913499.0, + "step": 4293 + }, + { + "epoch": 0.5462409362676505, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7759931087493896, + "num_tokens": 163957965.0, + "step": 4294 + }, + { + "epoch": 0.5463681465462409, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7549165487289429, + "num_tokens": 164005661.0, + "step": 4295 + }, + { + "epoch": 0.5464953568248314, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8148285746574402, + "num_tokens": 164044717.0, + "step": 4296 + }, + { + "epoch": 0.5466225671034219, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.790650486946106, + "num_tokens": 164080022.0, + "step": 4297 + }, + { + "epoch": 0.5467497773820125, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.8014526963233948, + "num_tokens": 164112749.0, + "step": 4298 + }, + { + "epoch": 0.546876987660603, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7878443002700806, + "num_tokens": 164152517.0, + "step": 4299 + }, + { + "epoch": 0.5470041979391935, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7977702617645264, + "num_tokens": 164195245.0, + "step": 4300 + }, + { + "epoch": 0.5471314082177839, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7878658175468445, + "num_tokens": 164230643.0, + "step": 4301 + }, + { + "epoch": 0.5472586184963745, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7911960482597351, + "num_tokens": 164271089.0, + "step": 4302 + }, + { + "epoch": 0.547385828774965, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7890695333480835, + "num_tokens": 164310618.0, + "step": 4303 + }, + { + "epoch": 0.5475130390535555, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7803671360015869, + "num_tokens": 164348630.0, + "step": 4304 + }, + { + "epoch": 0.5476402493321461, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7958147525787354, + "num_tokens": 164383735.0, + "step": 4305 + }, + { + "epoch": 0.5477674596107366, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7961061000823975, + "num_tokens": 164426115.0, + "step": 4306 + }, + { + "epoch": 0.5478946698893271, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7842617630958557, + "num_tokens": 164462119.0, + "step": 4307 + }, + { + "epoch": 0.5480218801679175, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.78996342420578, + "num_tokens": 164503030.0, + "step": 4308 + }, + { + "epoch": 0.5481490904465081, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7962157726287842, + "num_tokens": 164541091.0, + "step": 4309 + }, + { + "epoch": 0.5482763007250986, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7874013185501099, + "num_tokens": 164577678.0, + "step": 4310 + }, + { + "epoch": 0.5484035110036891, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7960142493247986, + "num_tokens": 164612779.0, + "step": 4311 + }, + { + "epoch": 0.5485307212822796, + "grad_norm": 62.5, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7980136871337891, + "num_tokens": 164649849.0, + "step": 4312 + }, + { + "epoch": 0.5486579315608702, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7955030202865601, + "num_tokens": 164690538.0, + "step": 4313 + }, + { + "epoch": 0.5487851418394606, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7937415838241577, + "num_tokens": 164729338.0, + "step": 4314 + }, + { + "epoch": 0.5489123521180511, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8096414804458618, + "num_tokens": 164766648.0, + "step": 4315 + }, + { + "epoch": 0.5490395623966416, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7915395498275757, + "num_tokens": 164804254.0, + "step": 4316 + }, + { + "epoch": 0.5491667726752322, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8093159198760986, + "num_tokens": 164839943.0, + "step": 4317 + }, + { + "epoch": 0.5492939829538227, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7847005128860474, + "num_tokens": 164887057.0, + "step": 4318 + }, + { + "epoch": 0.5494211932324132, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7990440726280212, + "num_tokens": 164928515.0, + "step": 4319 + }, + { + "epoch": 0.5495484035110036, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7930763959884644, + "num_tokens": 164963818.0, + "step": 4320 + }, + { + "epoch": 0.5496756137895942, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.809717059135437, + "num_tokens": 165001091.0, + "step": 4321 + }, + { + "epoch": 0.5498028240681847, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7768774032592773, + "num_tokens": 165046009.0, + "step": 4322 + }, + { + "epoch": 0.5499300343467752, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7904202938079834, + "num_tokens": 165090729.0, + "step": 4323 + }, + { + "epoch": 0.5500572446253658, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.813145637512207, + "num_tokens": 165129771.0, + "step": 4324 + }, + { + "epoch": 0.5501844549039563, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8147465586662292, + "num_tokens": 165164310.0, + "step": 4325 + }, + { + "epoch": 0.5503116651825467, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7908463478088379, + "num_tokens": 165202842.0, + "step": 4326 + }, + { + "epoch": 0.5504388754611372, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7898602485656738, + "num_tokens": 165235998.0, + "step": 4327 + }, + { + "epoch": 0.5505660857397278, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7932289838790894, + "num_tokens": 165271779.0, + "step": 4328 + }, + { + "epoch": 0.5506932960183183, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7871382236480713, + "num_tokens": 165306136.0, + "step": 4329 + }, + { + "epoch": 0.5508205062969088, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8088271617889404, + "num_tokens": 165342401.0, + "step": 4330 + }, + { + "epoch": 0.5509477165754993, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7998359203338623, + "num_tokens": 165380120.0, + "step": 4331 + }, + { + "epoch": 0.5510749268540898, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.781096339225769, + "num_tokens": 165418210.0, + "step": 4332 + }, + { + "epoch": 0.5512021371326803, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7757039666175842, + "num_tokens": 165460952.0, + "step": 4333 + }, + { + "epoch": 0.5513293474112708, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8083053827285767, + "num_tokens": 165500370.0, + "step": 4334 + }, + { + "epoch": 0.5514565576898613, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7923966646194458, + "num_tokens": 165534089.0, + "step": 4335 + }, + { + "epoch": 0.5515837679684519, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7887985706329346, + "num_tokens": 165573111.0, + "step": 4336 + }, + { + "epoch": 0.5517109782470424, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7817321419715881, + "num_tokens": 165611942.0, + "step": 4337 + }, + { + "epoch": 0.5518381885256328, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8026028871536255, + "num_tokens": 165651703.0, + "step": 4338 + }, + { + "epoch": 0.5519653988042234, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7911428809165955, + "num_tokens": 165691257.0, + "step": 4339 + }, + { + "epoch": 0.5520926090828139, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8110926747322083, + "num_tokens": 165727947.0, + "step": 4340 + }, + { + "epoch": 0.5522198193614044, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8058578968048096, + "num_tokens": 165763400.0, + "step": 4341 + }, + { + "epoch": 0.5523470296399949, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7935450077056885, + "num_tokens": 165809586.0, + "step": 4342 + }, + { + "epoch": 0.5524742399185855, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7932857275009155, + "num_tokens": 165839565.0, + "step": 4343 + }, + { + "epoch": 0.5526014501971759, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7918550968170166, + "num_tokens": 165873539.0, + "step": 4344 + }, + { + "epoch": 0.5527286604757664, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8082964420318604, + "num_tokens": 165920987.0, + "step": 4345 + }, + { + "epoch": 0.5528558707543569, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7853651642799377, + "num_tokens": 165965661.0, + "step": 4346 + }, + { + "epoch": 0.5529830810329475, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7806192636489868, + "num_tokens": 165998239.0, + "step": 4347 + }, + { + "epoch": 0.553110291311538, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7924337983131409, + "num_tokens": 166036489.0, + "step": 4348 + }, + { + "epoch": 0.5532375015901285, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7965829968452454, + "num_tokens": 166078774.0, + "step": 4349 + }, + { + "epoch": 0.5533647118687189, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7965679168701172, + "num_tokens": 166119555.0, + "step": 4350 + }, + { + "epoch": 0.5534919221473095, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7935115098953247, + "num_tokens": 166159788.0, + "step": 4351 + }, + { + "epoch": 0.5536191324259, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7880344986915588, + "num_tokens": 166200230.0, + "step": 4352 + }, + { + "epoch": 0.5537463427044905, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7982584834098816, + "num_tokens": 166237864.0, + "step": 4353 + }, + { + "epoch": 0.553873552983081, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7988104224205017, + "num_tokens": 166278347.0, + "step": 4354 + }, + { + "epoch": 0.5540007632616716, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7868969440460205, + "num_tokens": 166315744.0, + "step": 4355 + }, + { + "epoch": 0.554127973540262, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7917052507400513, + "num_tokens": 166352705.0, + "step": 4356 + }, + { + "epoch": 0.5542551838188525, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8063803315162659, + "num_tokens": 166393956.0, + "step": 4357 + }, + { + "epoch": 0.554382394097443, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7711499929428101, + "num_tokens": 166430172.0, + "step": 4358 + }, + { + "epoch": 0.5545096043760336, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7961299419403076, + "num_tokens": 166468732.0, + "step": 4359 + }, + { + "epoch": 0.5546368146546241, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7952736020088196, + "num_tokens": 166511383.0, + "step": 4360 + }, + { + "epoch": 0.5547640249332146, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7943947911262512, + "num_tokens": 166547674.0, + "step": 4361 + }, + { + "epoch": 0.5548912352118052, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7842831611633301, + "num_tokens": 166581763.0, + "step": 4362 + }, + { + "epoch": 0.5550184454903956, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7854639291763306, + "num_tokens": 166618604.0, + "step": 4363 + }, + { + "epoch": 0.5551456557689861, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8030037879943848, + "num_tokens": 166654812.0, + "step": 4364 + }, + { + "epoch": 0.5552728660475766, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7793475985527039, + "num_tokens": 166693206.0, + "step": 4365 + }, + { + "epoch": 0.5554000763261672, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7893240451812744, + "num_tokens": 166734531.0, + "step": 4366 + }, + { + "epoch": 0.5555272866047577, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7884637117385864, + "num_tokens": 166771362.0, + "step": 4367 + }, + { + "epoch": 0.5556544968833482, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7821530103683472, + "num_tokens": 166802784.0, + "step": 4368 + }, + { + "epoch": 0.5557817071619386, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7879235744476318, + "num_tokens": 166835590.0, + "step": 4369 + }, + { + "epoch": 0.5559089174405292, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7819579839706421, + "num_tokens": 166872145.0, + "step": 4370 + }, + { + "epoch": 0.5560361277191197, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.78758704662323, + "num_tokens": 166912855.0, + "step": 4371 + }, + { + "epoch": 0.5561633379977102, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7745147347450256, + "num_tokens": 166949482.0, + "step": 4372 + }, + { + "epoch": 0.5562905482763008, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7941607236862183, + "num_tokens": 166986727.0, + "step": 4373 + }, + { + "epoch": 0.5564177585548913, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7841775417327881, + "num_tokens": 167025869.0, + "step": 4374 + }, + { + "epoch": 0.5565449688334817, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8062208294868469, + "num_tokens": 167059825.0, + "step": 4375 + }, + { + "epoch": 0.5566721791120722, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.8020932674407959, + "num_tokens": 167097375.0, + "step": 4376 + }, + { + "epoch": 0.5567993893906628, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7874420881271362, + "num_tokens": 167136555.0, + "step": 4377 + }, + { + "epoch": 0.5569265996692533, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.79123854637146, + "num_tokens": 167179870.0, + "step": 4378 + }, + { + "epoch": 0.5570538099478438, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7933500409126282, + "num_tokens": 167215242.0, + "step": 4379 + }, + { + "epoch": 0.5571810202264343, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7973048686981201, + "num_tokens": 167250690.0, + "step": 4380 + }, + { + "epoch": 0.5573082305050248, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7774227857589722, + "num_tokens": 167283677.0, + "step": 4381 + }, + { + "epoch": 0.5574354407836153, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7921364307403564, + "num_tokens": 167321355.0, + "step": 4382 + }, + { + "epoch": 0.5575626510622058, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7906230688095093, + "num_tokens": 167363103.0, + "step": 4383 + }, + { + "epoch": 0.5576898613407963, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.808513879776001, + "num_tokens": 167399732.0, + "step": 4384 + }, + { + "epoch": 0.5578170716193869, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7985743284225464, + "num_tokens": 167434938.0, + "step": 4385 + }, + { + "epoch": 0.5579442818979774, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7930734753608704, + "num_tokens": 167472293.0, + "step": 4386 + }, + { + "epoch": 0.5580714921765678, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7984170913696289, + "num_tokens": 167511015.0, + "step": 4387 + }, + { + "epoch": 0.5581987024551583, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7953094244003296, + "num_tokens": 167549777.0, + "step": 4388 + }, + { + "epoch": 0.5583259127337489, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7988330125808716, + "num_tokens": 167592342.0, + "step": 4389 + }, + { + "epoch": 0.5584531230123394, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.8006237745285034, + "num_tokens": 167630829.0, + "step": 4390 + }, + { + "epoch": 0.5585803332909299, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7847303748130798, + "num_tokens": 167675938.0, + "step": 4391 + }, + { + "epoch": 0.5587075435695205, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.8007445931434631, + "num_tokens": 167716550.0, + "step": 4392 + }, + { + "epoch": 0.5588347538481109, + "grad_norm": 100.0, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7762311697006226, + "num_tokens": 167754401.0, + "step": 4393 + }, + { + "epoch": 0.5589619641267014, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7993758916854858, + "num_tokens": 167792124.0, + "step": 4394 + }, + { + "epoch": 0.5590891744052919, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8167753219604492, + "num_tokens": 167829226.0, + "step": 4395 + }, + { + "epoch": 0.5592163846838825, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8030980825424194, + "num_tokens": 167863439.0, + "step": 4396 + }, + { + "epoch": 0.559343594962473, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.798922061920166, + "num_tokens": 167907019.0, + "step": 4397 + }, + { + "epoch": 0.5594708052410635, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7879105806350708, + "num_tokens": 167942539.0, + "step": 4398 + }, + { + "epoch": 0.5595980155196539, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7958672046661377, + "num_tokens": 167982848.0, + "step": 4399 + }, + { + "epoch": 0.5597252257982445, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8027629852294922, + "num_tokens": 168017500.0, + "step": 4400 + }, + { + "epoch": 0.559852436076835, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7749776840209961, + "num_tokens": 168052790.0, + "step": 4401 + }, + { + "epoch": 0.5599796463554255, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7947118282318115, + "num_tokens": 168082541.0, + "step": 4402 + }, + { + "epoch": 0.560106856634016, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8030511736869812, + "num_tokens": 168128218.0, + "step": 4403 + }, + { + "epoch": 0.5602340669126066, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7961462140083313, + "num_tokens": 168165496.0, + "step": 4404 + }, + { + "epoch": 0.560361277191197, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7794308066368103, + "num_tokens": 168204675.0, + "step": 4405 + }, + { + "epoch": 0.5604884874697875, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7896120548248291, + "num_tokens": 168244832.0, + "step": 4406 + }, + { + "epoch": 0.560615697748378, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7927548885345459, + "num_tokens": 168282440.0, + "step": 4407 + }, + { + "epoch": 0.5607429080269686, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8024005889892578, + "num_tokens": 168325303.0, + "step": 4408 + }, + { + "epoch": 0.5608701183055591, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7985216975212097, + "num_tokens": 168360628.0, + "step": 4409 + }, + { + "epoch": 0.5609973285841496, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7760981321334839, + "num_tokens": 168398601.0, + "step": 4410 + }, + { + "epoch": 0.5611245388627402, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7790476083755493, + "num_tokens": 168436893.0, + "step": 4411 + }, + { + "epoch": 0.5612517491413306, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7913655042648315, + "num_tokens": 168478794.0, + "step": 4412 + }, + { + "epoch": 0.5613789594199211, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7980794906616211, + "num_tokens": 168520184.0, + "step": 4413 + }, + { + "epoch": 0.5615061696985116, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7888299822807312, + "num_tokens": 168554305.0, + "step": 4414 + }, + { + "epoch": 0.5616333799771022, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7760801911354065, + "num_tokens": 168597752.0, + "step": 4415 + }, + { + "epoch": 0.5617605902556927, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8002327084541321, + "num_tokens": 168637764.0, + "step": 4416 + }, + { + "epoch": 0.5618878005342832, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7703784108161926, + "num_tokens": 168672260.0, + "step": 4417 + }, + { + "epoch": 0.5620150108128736, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8063986301422119, + "num_tokens": 168711478.0, + "step": 4418 + }, + { + "epoch": 0.5621422210914642, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7985638380050659, + "num_tokens": 168751773.0, + "step": 4419 + }, + { + "epoch": 0.5622694313700547, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8062918782234192, + "num_tokens": 168786364.0, + "step": 4420 + }, + { + "epoch": 0.5623966416486452, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7941280603408813, + "num_tokens": 168826819.0, + "step": 4421 + }, + { + "epoch": 0.5625238519272358, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7925171852111816, + "num_tokens": 168869405.0, + "step": 4422 + }, + { + "epoch": 0.5626510622058263, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.8017996549606323, + "num_tokens": 168904323.0, + "step": 4423 + }, + { + "epoch": 0.5627782724844167, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7853003740310669, + "num_tokens": 168948654.0, + "step": 4424 + }, + { + "epoch": 0.5629054827630072, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7985478639602661, + "num_tokens": 168984333.0, + "step": 4425 + }, + { + "epoch": 0.5630326930415978, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7898719310760498, + "num_tokens": 169019843.0, + "step": 4426 + }, + { + "epoch": 0.5631599033201883, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.800171971321106, + "num_tokens": 169058177.0, + "step": 4427 + }, + { + "epoch": 0.5632871135987788, + "grad_norm": 103.0, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7991276979446411, + "num_tokens": 169102268.0, + "step": 4428 + }, + { + "epoch": 0.5634143238773693, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7925022840499878, + "num_tokens": 169141709.0, + "step": 4429 + }, + { + "epoch": 0.5635415341559598, + "grad_norm": 96.0, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7894917726516724, + "num_tokens": 169177107.0, + "step": 4430 + }, + { + "epoch": 0.5636687444345503, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7928119897842407, + "num_tokens": 169211531.0, + "step": 4431 + }, + { + "epoch": 0.5637959547131408, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.78754061460495, + "num_tokens": 169247507.0, + "step": 4432 + }, + { + "epoch": 0.5639231649917313, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8071110248565674, + "num_tokens": 169281007.0, + "step": 4433 + }, + { + "epoch": 0.5640503752703219, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7994124889373779, + "num_tokens": 169322295.0, + "step": 4434 + }, + { + "epoch": 0.5641775855489124, + "grad_norm": 107.0, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7898205518722534, + "num_tokens": 169358184.0, + "step": 4435 + }, + { + "epoch": 0.5643047958275028, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7917912006378174, + "num_tokens": 169391518.0, + "step": 4436 + }, + { + "epoch": 0.5644320061060933, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7945008277893066, + "num_tokens": 169426743.0, + "step": 4437 + }, + { + "epoch": 0.5645592163846839, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8184875249862671, + "num_tokens": 169466452.0, + "step": 4438 + }, + { + "epoch": 0.5646864266632744, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.801419734954834, + "num_tokens": 169500295.0, + "step": 4439 + }, + { + "epoch": 0.5648136369418649, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7964274883270264, + "num_tokens": 169541829.0, + "step": 4440 + }, + { + "epoch": 0.5649408472204555, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7774604558944702, + "num_tokens": 169581547.0, + "step": 4441 + }, + { + "epoch": 0.5650680574990459, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.803330659866333, + "num_tokens": 169622107.0, + "step": 4442 + }, + { + "epoch": 0.5651952677776364, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7919641733169556, + "num_tokens": 169657258.0, + "step": 4443 + }, + { + "epoch": 0.5653224780562269, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7810177803039551, + "num_tokens": 169691568.0, + "step": 4444 + }, + { + "epoch": 0.5654496883348175, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8026630282402039, + "num_tokens": 169732429.0, + "step": 4445 + }, + { + "epoch": 0.565576898613408, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7816308736801147, + "num_tokens": 169769224.0, + "step": 4446 + }, + { + "epoch": 0.5657041088919985, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7970702052116394, + "num_tokens": 169812830.0, + "step": 4447 + }, + { + "epoch": 0.5658313191705889, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7880107164382935, + "num_tokens": 169843823.0, + "step": 4448 + }, + { + "epoch": 0.5659585294491795, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8013366460800171, + "num_tokens": 169882847.0, + "step": 4449 + }, + { + "epoch": 0.56608573972777, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7758992910385132, + "num_tokens": 169921264.0, + "step": 4450 + }, + { + "epoch": 0.5662129500063605, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7930810451507568, + "num_tokens": 169949681.0, + "step": 4451 + }, + { + "epoch": 0.566340160284951, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8095330595970154, + "num_tokens": 169986254.0, + "step": 4452 + }, + { + "epoch": 0.5664673705635416, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7864125370979309, + "num_tokens": 170022745.0, + "step": 4453 + }, + { + "epoch": 0.566594580842132, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8032636642456055, + "num_tokens": 170057038.0, + "step": 4454 + }, + { + "epoch": 0.5667217911207225, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8129719495773315, + "num_tokens": 170098460.0, + "step": 4455 + }, + { + "epoch": 0.566849001399313, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7818349599838257, + "num_tokens": 170137689.0, + "step": 4456 + }, + { + "epoch": 0.5669762116779036, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7983066439628601, + "num_tokens": 170176716.0, + "step": 4457 + }, + { + "epoch": 0.5671034219564941, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7891888618469238, + "num_tokens": 170212621.0, + "step": 4458 + }, + { + "epoch": 0.5672306322350846, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7893624305725098, + "num_tokens": 170254493.0, + "step": 4459 + }, + { + "epoch": 0.5673578425136752, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8141130805015564, + "num_tokens": 170290559.0, + "step": 4460 + }, + { + "epoch": 0.5674850527922656, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7907435894012451, + "num_tokens": 170335079.0, + "step": 4461 + }, + { + "epoch": 0.5676122630708561, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7970486283302307, + "num_tokens": 170370665.0, + "step": 4462 + }, + { + "epoch": 0.5677394733494466, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7884500622749329, + "num_tokens": 170412585.0, + "step": 4463 + }, + { + "epoch": 0.5678666836280372, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.800041913986206, + "num_tokens": 170450806.0, + "step": 4464 + }, + { + "epoch": 0.5679938939066277, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7900046706199646, + "num_tokens": 170483265.0, + "step": 4465 + }, + { + "epoch": 0.5681211041852182, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.812598705291748, + "num_tokens": 170520910.0, + "step": 4466 + }, + { + "epoch": 0.5682483144638086, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7726045250892639, + "num_tokens": 170556516.0, + "step": 4467 + }, + { + "epoch": 0.5683755247423992, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.796114444732666, + "num_tokens": 170595571.0, + "step": 4468 + }, + { + "epoch": 0.5685027350209897, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8166388869285583, + "num_tokens": 170634117.0, + "step": 4469 + }, + { + "epoch": 0.5686299452995802, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7895790338516235, + "num_tokens": 170677624.0, + "step": 4470 + }, + { + "epoch": 0.5687571555781707, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7817466259002686, + "num_tokens": 170716732.0, + "step": 4471 + }, + { + "epoch": 0.5688843658567613, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7877019643783569, + "num_tokens": 170758521.0, + "step": 4472 + }, + { + "epoch": 0.5690115761353517, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7910926938056946, + "num_tokens": 170796799.0, + "step": 4473 + }, + { + "epoch": 0.5691387864139422, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.778344452381134, + "num_tokens": 170838569.0, + "step": 4474 + }, + { + "epoch": 0.5692659966925327, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8020422458648682, + "num_tokens": 170878206.0, + "step": 4475 + }, + { + "epoch": 0.5693932069711233, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.776570737361908, + "num_tokens": 170915506.0, + "step": 4476 + }, + { + "epoch": 0.5695204172497138, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7910735011100769, + "num_tokens": 170952303.0, + "step": 4477 + }, + { + "epoch": 0.5696476275283043, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7851623892784119, + "num_tokens": 170993757.0, + "step": 4478 + }, + { + "epoch": 0.5697748378068948, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7942380905151367, + "num_tokens": 171030240.0, + "step": 4479 + }, + { + "epoch": 0.5699020480854853, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7703405618667603, + "num_tokens": 171069654.0, + "step": 4480 + }, + { + "epoch": 0.5700292583640758, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7758014798164368, + "num_tokens": 171107341.0, + "step": 4481 + }, + { + "epoch": 0.5701564686426663, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7858473658561707, + "num_tokens": 171144063.0, + "step": 4482 + }, + { + "epoch": 0.5702836789212569, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7810567021369934, + "num_tokens": 171180448.0, + "step": 4483 + }, + { + "epoch": 0.5704108891998474, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7997013926506042, + "num_tokens": 171214220.0, + "step": 4484 + }, + { + "epoch": 0.5705380994784378, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.801800012588501, + "num_tokens": 171252798.0, + "step": 4485 + }, + { + "epoch": 0.5706653097570283, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8034549355506897, + "num_tokens": 171295434.0, + "step": 4486 + }, + { + "epoch": 0.5707925200356189, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7996987104415894, + "num_tokens": 171334129.0, + "step": 4487 + }, + { + "epoch": 0.5709197303142094, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7914329767227173, + "num_tokens": 171370239.0, + "step": 4488 + }, + { + "epoch": 0.5710469405927999, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.803073525428772, + "num_tokens": 171402411.0, + "step": 4489 + }, + { + "epoch": 0.5711741508713905, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8028066158294678, + "num_tokens": 171438736.0, + "step": 4490 + }, + { + "epoch": 0.5713013611499809, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7974953055381775, + "num_tokens": 171477949.0, + "step": 4491 + }, + { + "epoch": 0.5714285714285714, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7918981313705444, + "num_tokens": 171520953.0, + "step": 4492 + }, + { + "epoch": 0.5715557817071619, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.800490140914917, + "num_tokens": 171559415.0, + "step": 4493 + }, + { + "epoch": 0.5716829919857525, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7995721101760864, + "num_tokens": 171593118.0, + "step": 4494 + }, + { + "epoch": 0.571810202264343, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7793652415275574, + "num_tokens": 171634514.0, + "step": 4495 + }, + { + "epoch": 0.5719374125429335, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.799117922782898, + "num_tokens": 171675239.0, + "step": 4496 + }, + { + "epoch": 0.5720646228215239, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8032583594322205, + "num_tokens": 171713154.0, + "step": 4497 + }, + { + "epoch": 0.5721918331001145, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7870416641235352, + "num_tokens": 171750046.0, + "step": 4498 + }, + { + "epoch": 0.572319043378705, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7815117835998535, + "num_tokens": 171782708.0, + "step": 4499 + }, + { + "epoch": 0.5724462536572955, + "grad_norm": 93.5, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7893027663230896, + "num_tokens": 171819421.0, + "step": 4500 + }, + { + "epoch": 0.572573463935886, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7923369407653809, + "num_tokens": 171860993.0, + "step": 4501 + }, + { + "epoch": 0.5727006742144766, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7897934317588806, + "num_tokens": 171896894.0, + "step": 4502 + }, + { + "epoch": 0.572827884493067, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7928797602653503, + "num_tokens": 171938694.0, + "step": 4503 + }, + { + "epoch": 0.5729550947716575, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.821853756904602, + "num_tokens": 171972606.0, + "step": 4504 + }, + { + "epoch": 0.573082305050248, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7967140078544617, + "num_tokens": 172010861.0, + "step": 4505 + }, + { + "epoch": 0.5732095153288386, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7759536504745483, + "num_tokens": 172049867.0, + "step": 4506 + }, + { + "epoch": 0.5733367256074291, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7978067994117737, + "num_tokens": 172092176.0, + "step": 4507 + }, + { + "epoch": 0.5734639358860196, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7948592305183411, + "num_tokens": 172132789.0, + "step": 4508 + }, + { + "epoch": 0.5735911461646102, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8130748271942139, + "num_tokens": 172170536.0, + "step": 4509 + }, + { + "epoch": 0.5737183564432006, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7953174710273743, + "num_tokens": 172206828.0, + "step": 4510 + }, + { + "epoch": 0.5738455667217911, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7838019132614136, + "num_tokens": 172242311.0, + "step": 4511 + }, + { + "epoch": 0.5739727770003816, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7881485223770142, + "num_tokens": 172275135.0, + "step": 4512 + }, + { + "epoch": 0.5740999872789722, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8023392558097839, + "num_tokens": 172308682.0, + "step": 4513 + }, + { + "epoch": 0.5742271975575627, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7863504886627197, + "num_tokens": 172345834.0, + "step": 4514 + }, + { + "epoch": 0.5743544078361532, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7685158252716064, + "num_tokens": 172387976.0, + "step": 4515 + }, + { + "epoch": 0.5744816181147436, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7940798401832581, + "num_tokens": 172427447.0, + "step": 4516 + }, + { + "epoch": 0.5746088283933342, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7824485898017883, + "num_tokens": 172468686.0, + "step": 4517 + }, + { + "epoch": 0.5747360386719247, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7870148420333862, + "num_tokens": 172508190.0, + "step": 4518 + }, + { + "epoch": 0.5748632489505152, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8119802474975586, + "num_tokens": 172552295.0, + "step": 4519 + }, + { + "epoch": 0.5749904592291057, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7995308637619019, + "num_tokens": 172591065.0, + "step": 4520 + }, + { + "epoch": 0.5751176695076963, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7991682291030884, + "num_tokens": 172633741.0, + "step": 4521 + }, + { + "epoch": 0.5752448797862867, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8044247031211853, + "num_tokens": 172669081.0, + "step": 4522 + }, + { + "epoch": 0.5753720900648772, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7651697397232056, + "num_tokens": 172704956.0, + "step": 4523 + }, + { + "epoch": 0.5754993003434677, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7991719245910645, + "num_tokens": 172742270.0, + "step": 4524 + }, + { + "epoch": 0.5756265106220583, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7769383192062378, + "num_tokens": 172778065.0, + "step": 4525 + }, + { + "epoch": 0.5757537209006488, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.810810387134552, + "num_tokens": 172815825.0, + "step": 4526 + }, + { + "epoch": 0.5758809311792393, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7930077910423279, + "num_tokens": 172849560.0, + "step": 4527 + }, + { + "epoch": 0.5760081414578297, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7926602363586426, + "num_tokens": 172885203.0, + "step": 4528 + }, + { + "epoch": 0.5761353517364203, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7916387319564819, + "num_tokens": 172922341.0, + "step": 4529 + }, + { + "epoch": 0.5762625620150108, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7941606044769287, + "num_tokens": 172953363.0, + "step": 4530 + }, + { + "epoch": 0.5763897722936013, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8010649681091309, + "num_tokens": 172986253.0, + "step": 4531 + }, + { + "epoch": 0.5765169825721919, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8049930334091187, + "num_tokens": 173023466.0, + "step": 4532 + }, + { + "epoch": 0.5766441928507824, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7871541976928711, + "num_tokens": 173063329.0, + "step": 4533 + }, + { + "epoch": 0.5767714031293728, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8051915168762207, + "num_tokens": 173100676.0, + "step": 4534 + }, + { + "epoch": 0.5768986134079633, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.803612470626831, + "num_tokens": 173135088.0, + "step": 4535 + }, + { + "epoch": 0.5770258236865539, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7878639698028564, + "num_tokens": 173175810.0, + "step": 4536 + }, + { + "epoch": 0.5771530339651444, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7974514961242676, + "num_tokens": 173220826.0, + "step": 4537 + }, + { + "epoch": 0.5772802442437349, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8048316240310669, + "num_tokens": 173258411.0, + "step": 4538 + }, + { + "epoch": 0.5774074545223254, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7754856944084167, + "num_tokens": 173297298.0, + "step": 4539 + }, + { + "epoch": 0.5775346648009159, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.772955060005188, + "num_tokens": 173330839.0, + "step": 4540 + }, + { + "epoch": 0.5776618750795064, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7867132425308228, + "num_tokens": 173370219.0, + "step": 4541 + }, + { + "epoch": 0.5777890853580969, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7976467609405518, + "num_tokens": 173410425.0, + "step": 4542 + }, + { + "epoch": 0.5779162956366874, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7902513146400452, + "num_tokens": 173447109.0, + "step": 4543 + }, + { + "epoch": 0.578043505915278, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.803012490272522, + "num_tokens": 173490064.0, + "step": 4544 + }, + { + "epoch": 0.5781707161938685, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.791999876499176, + "num_tokens": 173523791.0, + "step": 4545 + }, + { + "epoch": 0.5782979264724589, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7996216416358948, + "num_tokens": 173564069.0, + "step": 4546 + }, + { + "epoch": 0.5784251367510495, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7889560461044312, + "num_tokens": 173601893.0, + "step": 4547 + }, + { + "epoch": 0.57855234702964, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8094313144683838, + "num_tokens": 173642299.0, + "step": 4548 + }, + { + "epoch": 0.5786795573082305, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7938821315765381, + "num_tokens": 173675824.0, + "step": 4549 + }, + { + "epoch": 0.578806767586821, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7946130037307739, + "num_tokens": 173712192.0, + "step": 4550 + }, + { + "epoch": 0.5789339778654116, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7995871901512146, + "num_tokens": 173746949.0, + "step": 4551 + }, + { + "epoch": 0.579061188144002, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7934672832489014, + "num_tokens": 173784527.0, + "step": 4552 + }, + { + "epoch": 0.5791883984225925, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8025723695755005, + "num_tokens": 173820648.0, + "step": 4553 + }, + { + "epoch": 0.579315608701183, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7861809730529785, + "num_tokens": 173860427.0, + "step": 4554 + }, + { + "epoch": 0.5794428189797736, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7951439619064331, + "num_tokens": 173899876.0, + "step": 4555 + }, + { + "epoch": 0.5795700292583641, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8057048916816711, + "num_tokens": 173937354.0, + "step": 4556 + }, + { + "epoch": 0.5796972395369546, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7811826467514038, + "num_tokens": 173976039.0, + "step": 4557 + }, + { + "epoch": 0.5798244498155452, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7686012983322144, + "num_tokens": 174006359.0, + "step": 4558 + }, + { + "epoch": 0.5799516600941356, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7859789133071899, + "num_tokens": 174044662.0, + "step": 4559 + }, + { + "epoch": 0.5800788703727261, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7845621109008789, + "num_tokens": 174077133.0, + "step": 4560 + }, + { + "epoch": 0.5802060806513166, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7899621725082397, + "num_tokens": 174121873.0, + "step": 4561 + }, + { + "epoch": 0.5803332909299072, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7963579893112183, + "num_tokens": 174158873.0, + "step": 4562 + }, + { + "epoch": 0.5804605012084977, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.8031962513923645, + "num_tokens": 174189222.0, + "step": 4563 + }, + { + "epoch": 0.5805877114870882, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7862445712089539, + "num_tokens": 174229018.0, + "step": 4564 + }, + { + "epoch": 0.5807149217656786, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7949556708335876, + "num_tokens": 174269215.0, + "step": 4565 + }, + { + "epoch": 0.5808421320442692, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8018804788589478, + "num_tokens": 174303926.0, + "step": 4566 + }, + { + "epoch": 0.5809693423228597, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7968539595603943, + "num_tokens": 174347491.0, + "step": 4567 + }, + { + "epoch": 0.5810965526014502, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7948936223983765, + "num_tokens": 174384246.0, + "step": 4568 + }, + { + "epoch": 0.5812237628800407, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7944512367248535, + "num_tokens": 174417081.0, + "step": 4569 + }, + { + "epoch": 0.5813509731586313, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7815538644790649, + "num_tokens": 174457364.0, + "step": 4570 + }, + { + "epoch": 0.5814781834372217, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7926862835884094, + "num_tokens": 174494304.0, + "step": 4571 + }, + { + "epoch": 0.5816053937158122, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7967219352722168, + "num_tokens": 174528047.0, + "step": 4572 + }, + { + "epoch": 0.5817326039944027, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7949610948562622, + "num_tokens": 174565990.0, + "step": 4573 + }, + { + "epoch": 0.5818598142729933, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7791122794151306, + "num_tokens": 174599039.0, + "step": 4574 + }, + { + "epoch": 0.5819870245515838, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7854203581809998, + "num_tokens": 174634786.0, + "step": 4575 + }, + { + "epoch": 0.5821142348301743, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7841842174530029, + "num_tokens": 174677875.0, + "step": 4576 + }, + { + "epoch": 0.5822414451087647, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7889851331710815, + "num_tokens": 174715421.0, + "step": 4577 + }, + { + "epoch": 0.5823686553873553, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7849907875061035, + "num_tokens": 174746555.0, + "step": 4578 + }, + { + "epoch": 0.5824958656659458, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7839576601982117, + "num_tokens": 174786470.0, + "step": 4579 + }, + { + "epoch": 0.5826230759445363, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7936230301856995, + "num_tokens": 174827131.0, + "step": 4580 + }, + { + "epoch": 0.5827502862231269, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7851996421813965, + "num_tokens": 174861391.0, + "step": 4581 + }, + { + "epoch": 0.5828774965017174, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7960458993911743, + "num_tokens": 174897437.0, + "step": 4582 + }, + { + "epoch": 0.5830047067803078, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7807326912879944, + "num_tokens": 174932504.0, + "step": 4583 + }, + { + "epoch": 0.5831319170588983, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7858555912971497, + "num_tokens": 174974746.0, + "step": 4584 + }, + { + "epoch": 0.5832591273374889, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7880290746688843, + "num_tokens": 175016155.0, + "step": 4585 + }, + { + "epoch": 0.5833863376160794, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7839956283569336, + "num_tokens": 175060595.0, + "step": 4586 + }, + { + "epoch": 0.5835135478946699, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7937993407249451, + "num_tokens": 175101278.0, + "step": 4587 + }, + { + "epoch": 0.5836407581732604, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7864172458648682, + "num_tokens": 175139479.0, + "step": 4588 + }, + { + "epoch": 0.5837679684518509, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7773880958557129, + "num_tokens": 175182376.0, + "step": 4589 + }, + { + "epoch": 0.5838951787304414, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7894281148910522, + "num_tokens": 175225449.0, + "step": 4590 + }, + { + "epoch": 0.5840223890090319, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.788108229637146, + "num_tokens": 175255959.0, + "step": 4591 + }, + { + "epoch": 0.5841495992876224, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7796167731285095, + "num_tokens": 175292142.0, + "step": 4592 + }, + { + "epoch": 0.584276809566213, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7896493077278137, + "num_tokens": 175336630.0, + "step": 4593 + }, + { + "epoch": 0.5844040198448035, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.8048195838928223, + "num_tokens": 175372697.0, + "step": 4594 + }, + { + "epoch": 0.5845312301233939, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7776387929916382, + "num_tokens": 175412395.0, + "step": 4595 + }, + { + "epoch": 0.5846584404019844, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7924487590789795, + "num_tokens": 175447295.0, + "step": 4596 + }, + { + "epoch": 0.584785650680575, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8017095327377319, + "num_tokens": 175480229.0, + "step": 4597 + }, + { + "epoch": 0.5849128609591655, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7907787561416626, + "num_tokens": 175523081.0, + "step": 4598 + }, + { + "epoch": 0.585040071237756, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.797419548034668, + "num_tokens": 175565036.0, + "step": 4599 + }, + { + "epoch": 0.5851672815163466, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7980575561523438, + "num_tokens": 175606733.0, + "step": 4600 + }, + { + "epoch": 0.585294491794937, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8050155639648438, + "num_tokens": 175637663.0, + "step": 4601 + }, + { + "epoch": 0.5854217020735275, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7979320287704468, + "num_tokens": 175677381.0, + "step": 4602 + }, + { + "epoch": 0.585548912352118, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8021117448806763, + "num_tokens": 175713049.0, + "step": 4603 + }, + { + "epoch": 0.5856761226307086, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7840479612350464, + "num_tokens": 175753681.0, + "step": 4604 + }, + { + "epoch": 0.5858033329092991, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7799554467201233, + "num_tokens": 175789725.0, + "step": 4605 + }, + { + "epoch": 0.5859305431878896, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.788483202457428, + "num_tokens": 175823203.0, + "step": 4606 + }, + { + "epoch": 0.5860577534664801, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7866612076759338, + "num_tokens": 175860506.0, + "step": 4607 + }, + { + "epoch": 0.5861849637450706, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8063893914222717, + "num_tokens": 175895080.0, + "step": 4608 + }, + { + "epoch": 0.5863121740236611, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.787619948387146, + "num_tokens": 175937113.0, + "step": 4609 + }, + { + "epoch": 0.5864393843022516, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.802544355392456, + "num_tokens": 175966193.0, + "step": 4610 + }, + { + "epoch": 0.5865665945808421, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7890613079071045, + "num_tokens": 176003771.0, + "step": 4611 + }, + { + "epoch": 0.5866938048594327, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7991891503334045, + "num_tokens": 176050069.0, + "step": 4612 + }, + { + "epoch": 0.5868210151380232, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8124061822891235, + "num_tokens": 176088763.0, + "step": 4613 + }, + { + "epoch": 0.5869482254166136, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7974408268928528, + "num_tokens": 176127853.0, + "step": 4614 + }, + { + "epoch": 0.5870754356952042, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7811592817306519, + "num_tokens": 176173775.0, + "step": 4615 + }, + { + "epoch": 0.5872026459737947, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7932379245758057, + "num_tokens": 176210878.0, + "step": 4616 + }, + { + "epoch": 0.5873298562523852, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7859376668930054, + "num_tokens": 176255094.0, + "step": 4617 + }, + { + "epoch": 0.5874570665309757, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7881171703338623, + "num_tokens": 176293586.0, + "step": 4618 + }, + { + "epoch": 0.5875842768095663, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.796454906463623, + "num_tokens": 176329203.0, + "step": 4619 + }, + { + "epoch": 0.5877114870881567, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8130215406417847, + "num_tokens": 176367984.0, + "step": 4620 + }, + { + "epoch": 0.5878386973667472, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8088167905807495, + "num_tokens": 176408949.0, + "step": 4621 + }, + { + "epoch": 0.5879659076453377, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8122109770774841, + "num_tokens": 176451073.0, + "step": 4622 + }, + { + "epoch": 0.5880931179239283, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.803476095199585, + "num_tokens": 176490623.0, + "step": 4623 + }, + { + "epoch": 0.5882203282025188, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7808040380477905, + "num_tokens": 176525771.0, + "step": 4624 + }, + { + "epoch": 0.5883475384811093, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8141895532608032, + "num_tokens": 176565417.0, + "step": 4625 + }, + { + "epoch": 0.5884747487596997, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7864502668380737, + "num_tokens": 176599005.0, + "step": 4626 + }, + { + "epoch": 0.5886019590382903, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.797137975692749, + "num_tokens": 176637947.0, + "step": 4627 + }, + { + "epoch": 0.5887291693168808, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7860283851623535, + "num_tokens": 176675241.0, + "step": 4628 + }, + { + "epoch": 0.5888563795954713, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7886989712715149, + "num_tokens": 176715375.0, + "step": 4629 + }, + { + "epoch": 0.5889835898740619, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.784968912601471, + "num_tokens": 176750977.0, + "step": 4630 + }, + { + "epoch": 0.5891108001526524, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7956010103225708, + "num_tokens": 176788968.0, + "step": 4631 + }, + { + "epoch": 0.5892380104312428, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7940704226493835, + "num_tokens": 176824399.0, + "step": 4632 + }, + { + "epoch": 0.5893652207098333, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7863864898681641, + "num_tokens": 176862053.0, + "step": 4633 + }, + { + "epoch": 0.5894924309884239, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.808074414730072, + "num_tokens": 176896943.0, + "step": 4634 + }, + { + "epoch": 0.5896196412670144, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7916203141212463, + "num_tokens": 176936263.0, + "step": 4635 + }, + { + "epoch": 0.5897468515456049, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7852063179016113, + "num_tokens": 176976632.0, + "step": 4636 + }, + { + "epoch": 0.5898740618241954, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7961825728416443, + "num_tokens": 177007036.0, + "step": 4637 + }, + { + "epoch": 0.5900012721027859, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7843139171600342, + "num_tokens": 177042189.0, + "step": 4638 + }, + { + "epoch": 0.5901284823813764, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7863389253616333, + "num_tokens": 177082428.0, + "step": 4639 + }, + { + "epoch": 0.5902556926599669, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7946359515190125, + "num_tokens": 177124447.0, + "step": 4640 + }, + { + "epoch": 0.5903829029385574, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.810012936592102, + "num_tokens": 177160516.0, + "step": 4641 + }, + { + "epoch": 0.590510113217148, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8028079867362976, + "num_tokens": 177198100.0, + "step": 4642 + }, + { + "epoch": 0.5906373234957385, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7881964445114136, + "num_tokens": 177243170.0, + "step": 4643 + }, + { + "epoch": 0.5907645337743289, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7885712385177612, + "num_tokens": 177282976.0, + "step": 4644 + }, + { + "epoch": 0.5908917440529194, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7801807522773743, + "num_tokens": 177323725.0, + "step": 4645 + }, + { + "epoch": 0.59101895433151, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7930001020431519, + "num_tokens": 177352178.0, + "step": 4646 + }, + { + "epoch": 0.5911461646101005, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7936862111091614, + "num_tokens": 177392322.0, + "step": 4647 + }, + { + "epoch": 0.591273374888691, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7924976348876953, + "num_tokens": 177432013.0, + "step": 4648 + }, + { + "epoch": 0.5914005851672816, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8042340278625488, + "num_tokens": 177473542.0, + "step": 4649 + }, + { + "epoch": 0.591527795445872, + "grad_norm": 84.0, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.790542721748352, + "num_tokens": 177516590.0, + "step": 4650 + }, + { + "epoch": 0.5916550057244625, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8051995635032654, + "num_tokens": 177553750.0, + "step": 4651 + }, + { + "epoch": 0.591782216003053, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7829304933547974, + "num_tokens": 177586093.0, + "step": 4652 + }, + { + "epoch": 0.5919094262816436, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7877402305603027, + "num_tokens": 177626800.0, + "step": 4653 + }, + { + "epoch": 0.5920366365602341, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7897084951400757, + "num_tokens": 177662336.0, + "step": 4654 + }, + { + "epoch": 0.5921638468388246, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8061918020248413, + "num_tokens": 177706922.0, + "step": 4655 + }, + { + "epoch": 0.592291057117415, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7953378558158875, + "num_tokens": 177743922.0, + "step": 4656 + }, + { + "epoch": 0.5924182673960056, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7960554957389832, + "num_tokens": 177777069.0, + "step": 4657 + }, + { + "epoch": 0.5925454776745961, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7980155944824219, + "num_tokens": 177810959.0, + "step": 4658 + }, + { + "epoch": 0.5926726879531866, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7814493179321289, + "num_tokens": 177846366.0, + "step": 4659 + }, + { + "epoch": 0.5927998982317771, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.808199405670166, + "num_tokens": 177881450.0, + "step": 4660 + }, + { + "epoch": 0.5929271085103677, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7955519556999207, + "num_tokens": 177922931.0, + "step": 4661 + }, + { + "epoch": 0.5930543187889582, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7728278040885925, + "num_tokens": 177962722.0, + "step": 4662 + }, + { + "epoch": 0.5931815290675486, + "grad_norm": 153.0, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7855603694915771, + "num_tokens": 177999956.0, + "step": 4663 + }, + { + "epoch": 0.5933087393461391, + "grad_norm": 110.5, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7804572582244873, + "num_tokens": 178032450.0, + "step": 4664 + }, + { + "epoch": 0.5934359496247297, + "grad_norm": 91.0, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.8002520203590393, + "num_tokens": 178075028.0, + "step": 4665 + }, + { + "epoch": 0.5935631599033202, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7895668745040894, + "num_tokens": 178108267.0, + "step": 4666 + }, + { + "epoch": 0.5936903701819107, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7989558577537537, + "num_tokens": 178148091.0, + "step": 4667 + }, + { + "epoch": 0.5938175804605013, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8078210353851318, + "num_tokens": 178185626.0, + "step": 4668 + }, + { + "epoch": 0.5939447907390917, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.804047703742981, + "num_tokens": 178224165.0, + "step": 4669 + }, + { + "epoch": 0.5940720010176822, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.793079137802124, + "num_tokens": 178264118.0, + "step": 4670 + }, + { + "epoch": 0.5941992112962727, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7974739670753479, + "num_tokens": 178295634.0, + "step": 4671 + }, + { + "epoch": 0.5943264215748633, + "grad_norm": 180.0, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7896056175231934, + "num_tokens": 178327076.0, + "step": 4672 + }, + { + "epoch": 0.5944536318534538, + "grad_norm": 125.0, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.779866099357605, + "num_tokens": 178357469.0, + "step": 4673 + }, + { + "epoch": 0.5945808421320443, + "grad_norm": 109.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7996004819869995, + "num_tokens": 178393830.0, + "step": 4674 + }, + { + "epoch": 0.5947080524106347, + "grad_norm": 96.0, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7897950410842896, + "num_tokens": 178427560.0, + "step": 4675 + }, + { + "epoch": 0.5948352626892253, + "grad_norm": 125.0, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7886945605278015, + "num_tokens": 178464666.0, + "step": 4676 + }, + { + "epoch": 0.5949624729678158, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7831408381462097, + "num_tokens": 178508771.0, + "step": 4677 + }, + { + "epoch": 0.5950896832464063, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7952084541320801, + "num_tokens": 178548516.0, + "step": 4678 + }, + { + "epoch": 0.5952168935249968, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.755989670753479, + "num_tokens": 178591276.0, + "step": 4679 + }, + { + "epoch": 0.5953441038035874, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7892280220985413, + "num_tokens": 178632580.0, + "step": 4680 + }, + { + "epoch": 0.5954713140821778, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7894527912139893, + "num_tokens": 178672858.0, + "step": 4681 + }, + { + "epoch": 0.5955985243607683, + "grad_norm": 80.0, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7919427752494812, + "num_tokens": 178706435.0, + "step": 4682 + }, + { + "epoch": 0.5957257346393589, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7912874221801758, + "num_tokens": 178746309.0, + "step": 4683 + }, + { + "epoch": 0.5958529449179494, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7783850431442261, + "num_tokens": 178787639.0, + "step": 4684 + }, + { + "epoch": 0.5959801551965399, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7973894476890564, + "num_tokens": 178828271.0, + "step": 4685 + }, + { + "epoch": 0.5961073654751304, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8156353235244751, + "num_tokens": 178862862.0, + "step": 4686 + }, + { + "epoch": 0.5962345757537209, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8051410913467407, + "num_tokens": 178897969.0, + "step": 4687 + }, + { + "epoch": 0.5963617860323114, + "grad_norm": 107.5, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.794698178768158, + "num_tokens": 178942002.0, + "step": 4688 + }, + { + "epoch": 0.5964889963109019, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8035409450531006, + "num_tokens": 178983473.0, + "step": 4689 + }, + { + "epoch": 0.5966162065894924, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7790669202804565, + "num_tokens": 179022370.0, + "step": 4690 + }, + { + "epoch": 0.596743416868083, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.8124265670776367, + "num_tokens": 179064841.0, + "step": 4691 + }, + { + "epoch": 0.5968706271466735, + "grad_norm": 107.5, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7599316835403442, + "num_tokens": 179099399.0, + "step": 4692 + }, + { + "epoch": 0.5969978374252639, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7802108526229858, + "num_tokens": 179133969.0, + "step": 4693 + }, + { + "epoch": 0.5971250477038544, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.78314208984375, + "num_tokens": 179165456.0, + "step": 4694 + }, + { + "epoch": 0.597252257982445, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7886312007904053, + "num_tokens": 179198209.0, + "step": 4695 + }, + { + "epoch": 0.5973794682610355, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7798526287078857, + "num_tokens": 179233872.0, + "step": 4696 + }, + { + "epoch": 0.597506678539626, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.816271185874939, + "num_tokens": 179272721.0, + "step": 4697 + }, + { + "epoch": 0.5976338888182166, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7955107688903809, + "num_tokens": 179305781.0, + "step": 4698 + }, + { + "epoch": 0.597761099096807, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7981257438659668, + "num_tokens": 179350049.0, + "step": 4699 + }, + { + "epoch": 0.5978883093753975, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7932584881782532, + "num_tokens": 179388063.0, + "step": 4700 + }, + { + "epoch": 0.598015519653988, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8017712831497192, + "num_tokens": 179427607.0, + "step": 4701 + }, + { + "epoch": 0.5981427299325786, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7773890495300293, + "num_tokens": 179471456.0, + "step": 4702 + }, + { + "epoch": 0.5982699402111691, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8109586834907532, + "num_tokens": 179514318.0, + "step": 4703 + }, + { + "epoch": 0.5983971504897596, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8115406036376953, + "num_tokens": 179553307.0, + "step": 4704 + }, + { + "epoch": 0.59852436076835, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7877749800682068, + "num_tokens": 179597288.0, + "step": 4705 + }, + { + "epoch": 0.5986515710469406, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7925382852554321, + "num_tokens": 179629583.0, + "step": 4706 + }, + { + "epoch": 0.5987787813255311, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7866051197052002, + "num_tokens": 179672104.0, + "step": 4707 + }, + { + "epoch": 0.5989059916041216, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7815150022506714, + "num_tokens": 179712951.0, + "step": 4708 + }, + { + "epoch": 0.5990332018827121, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.792044997215271, + "num_tokens": 179754626.0, + "step": 4709 + }, + { + "epoch": 0.5991604121613027, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7997187376022339, + "num_tokens": 179790068.0, + "step": 4710 + }, + { + "epoch": 0.5992876224398932, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7738699913024902, + "num_tokens": 179823936.0, + "step": 4711 + }, + { + "epoch": 0.5994148327184836, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7961774468421936, + "num_tokens": 179860920.0, + "step": 4712 + }, + { + "epoch": 0.5995420429970741, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.783950686454773, + "num_tokens": 179899848.0, + "step": 4713 + }, + { + "epoch": 0.5996692532756647, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8115867376327515, + "num_tokens": 179934462.0, + "step": 4714 + }, + { + "epoch": 0.5997964635542552, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7929757237434387, + "num_tokens": 179973955.0, + "step": 4715 + }, + { + "epoch": 0.5999236738328457, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.8011822700500488, + "num_tokens": 180015256.0, + "step": 4716 + }, + { + "epoch": 0.6000508841114363, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.8052822351455688, + "num_tokens": 180054145.0, + "step": 4717 + }, + { + "epoch": 0.6001780943900267, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8022331595420837, + "num_tokens": 180093411.0, + "step": 4718 + }, + { + "epoch": 0.6003053046686172, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8034343719482422, + "num_tokens": 180135799.0, + "step": 4719 + }, + { + "epoch": 0.6004325149472077, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7838530540466309, + "num_tokens": 180172623.0, + "step": 4720 + }, + { + "epoch": 0.6005597252257983, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7919355630874634, + "num_tokens": 180213833.0, + "step": 4721 + }, + { + "epoch": 0.6006869355043888, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7969472408294678, + "num_tokens": 180254982.0, + "step": 4722 + }, + { + "epoch": 0.6008141457829793, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.798959493637085, + "num_tokens": 180288544.0, + "step": 4723 + }, + { + "epoch": 0.6009413560615697, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7896391153335571, + "num_tokens": 180330233.0, + "step": 4724 + }, + { + "epoch": 0.6010685663401603, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7987256050109863, + "num_tokens": 180361516.0, + "step": 4725 + }, + { + "epoch": 0.6011957766187508, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7957619428634644, + "num_tokens": 180404380.0, + "step": 4726 + }, + { + "epoch": 0.6013229868973413, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7871112823486328, + "num_tokens": 180436916.0, + "step": 4727 + }, + { + "epoch": 0.6014501971759318, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7796993255615234, + "num_tokens": 180476895.0, + "step": 4728 + }, + { + "epoch": 0.6015774074545224, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7937602400779724, + "num_tokens": 180515248.0, + "step": 4729 + }, + { + "epoch": 0.6017046177331128, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.8022469282150269, + "num_tokens": 180554575.0, + "step": 4730 + }, + { + "epoch": 0.6018318280117033, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7825761437416077, + "num_tokens": 180594485.0, + "step": 4731 + }, + { + "epoch": 0.6019590382902938, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7759692668914795, + "num_tokens": 180634746.0, + "step": 4732 + }, + { + "epoch": 0.6020862485688844, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7847076058387756, + "num_tokens": 180674946.0, + "step": 4733 + }, + { + "epoch": 0.6022134588474749, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7951493859291077, + "num_tokens": 180715538.0, + "step": 4734 + }, + { + "epoch": 0.6023406691260654, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7947021722793579, + "num_tokens": 180755298.0, + "step": 4735 + }, + { + "epoch": 0.6024678794046558, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.789708137512207, + "num_tokens": 180797104.0, + "step": 4736 + }, + { + "epoch": 0.6025950896832464, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8096187114715576, + "num_tokens": 180839845.0, + "step": 4737 + }, + { + "epoch": 0.6027222999618369, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8069097995758057, + "num_tokens": 180878849.0, + "step": 4738 + }, + { + "epoch": 0.6028495102404274, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7928099632263184, + "num_tokens": 180918427.0, + "step": 4739 + }, + { + "epoch": 0.602976720519018, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7824974060058594, + "num_tokens": 180956206.0, + "step": 4740 + }, + { + "epoch": 0.6031039307976085, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7996475696563721, + "num_tokens": 180996945.0, + "step": 4741 + }, + { + "epoch": 0.6032311410761989, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7973217964172363, + "num_tokens": 181031763.0, + "step": 4742 + }, + { + "epoch": 0.6033583513547894, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8022686243057251, + "num_tokens": 181070661.0, + "step": 4743 + }, + { + "epoch": 0.60348556163338, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7808343172073364, + "num_tokens": 181112769.0, + "step": 4744 + }, + { + "epoch": 0.6036127719119705, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7958852648735046, + "num_tokens": 181151155.0, + "step": 4745 + }, + { + "epoch": 0.603739982190561, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7943292856216431, + "num_tokens": 181188689.0, + "step": 4746 + }, + { + "epoch": 0.6038671924691515, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7962452173233032, + "num_tokens": 181230031.0, + "step": 4747 + }, + { + "epoch": 0.603994402747742, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7966318726539612, + "num_tokens": 181271464.0, + "step": 4748 + }, + { + "epoch": 0.6041216130263325, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7833361625671387, + "num_tokens": 181312607.0, + "step": 4749 + }, + { + "epoch": 0.604248823304923, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8017981648445129, + "num_tokens": 181348628.0, + "step": 4750 + }, + { + "epoch": 0.6043760335835135, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7863259315490723, + "num_tokens": 181387975.0, + "step": 4751 + }, + { + "epoch": 0.6045032438621041, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8093973398208618, + "num_tokens": 181424163.0, + "step": 4752 + }, + { + "epoch": 0.6046304541406946, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7840311527252197, + "num_tokens": 181457002.0, + "step": 4753 + }, + { + "epoch": 0.604757664419285, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7873642444610596, + "num_tokens": 181494410.0, + "step": 4754 + }, + { + "epoch": 0.6048848746978756, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.790061354637146, + "num_tokens": 181534595.0, + "step": 4755 + }, + { + "epoch": 0.6050120849764661, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7991098761558533, + "num_tokens": 181570654.0, + "step": 4756 + }, + { + "epoch": 0.6051392952550566, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7894972562789917, + "num_tokens": 181608366.0, + "step": 4757 + }, + { + "epoch": 0.6052665055336471, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7753483653068542, + "num_tokens": 181647168.0, + "step": 4758 + }, + { + "epoch": 0.6053937158122377, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7997365593910217, + "num_tokens": 181689177.0, + "step": 4759 + }, + { + "epoch": 0.6055209260908282, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7998386025428772, + "num_tokens": 181723742.0, + "step": 4760 + }, + { + "epoch": 0.6056481363694186, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8022530674934387, + "num_tokens": 181761539.0, + "step": 4761 + }, + { + "epoch": 0.6057753466480091, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7927782535552979, + "num_tokens": 181799674.0, + "step": 4762 + }, + { + "epoch": 0.6059025569265997, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7790517807006836, + "num_tokens": 181838784.0, + "step": 4763 + }, + { + "epoch": 0.6060297672051902, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7850614786148071, + "num_tokens": 181875669.0, + "step": 4764 + }, + { + "epoch": 0.6061569774837807, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7758347392082214, + "num_tokens": 181912450.0, + "step": 4765 + }, + { + "epoch": 0.6062841877623713, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7941924929618835, + "num_tokens": 181947691.0, + "step": 4766 + }, + { + "epoch": 0.6064113980409617, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8010014295578003, + "num_tokens": 181982364.0, + "step": 4767 + }, + { + "epoch": 0.6065386083195522, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7929971814155579, + "num_tokens": 182017071.0, + "step": 4768 + }, + { + "epoch": 0.6066658185981427, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.795944094657898, + "num_tokens": 182057234.0, + "step": 4769 + }, + { + "epoch": 0.6067930288767333, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8045865893363953, + "num_tokens": 182092450.0, + "step": 4770 + }, + { + "epoch": 0.6069202391553238, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7855886220932007, + "num_tokens": 182126398.0, + "step": 4771 + }, + { + "epoch": 0.6070474494339143, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8038181066513062, + "num_tokens": 182160848.0, + "step": 4772 + }, + { + "epoch": 0.6071746597125047, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8039852380752563, + "num_tokens": 182202596.0, + "step": 4773 + }, + { + "epoch": 0.6073018699910953, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7877224087715149, + "num_tokens": 182239235.0, + "step": 4774 + }, + { + "epoch": 0.6074290802696858, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.807349681854248, + "num_tokens": 182273662.0, + "step": 4775 + }, + { + "epoch": 0.6075562905482763, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7802024483680725, + "num_tokens": 182311991.0, + "step": 4776 + }, + { + "epoch": 0.6076835008268668, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7960983514785767, + "num_tokens": 182350722.0, + "step": 4777 + }, + { + "epoch": 0.6078107111054574, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7996232509613037, + "num_tokens": 182386755.0, + "step": 4778 + }, + { + "epoch": 0.6079379213840478, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8061954975128174, + "num_tokens": 182423128.0, + "step": 4779 + }, + { + "epoch": 0.6080651316626383, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7762056589126587, + "num_tokens": 182461518.0, + "step": 4780 + }, + { + "epoch": 0.6081923419412288, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7924680709838867, + "num_tokens": 182495707.0, + "step": 4781 + }, + { + "epoch": 0.6083195522198194, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7871455550193787, + "num_tokens": 182539058.0, + "step": 4782 + }, + { + "epoch": 0.6084467624984099, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7848191857337952, + "num_tokens": 182572482.0, + "step": 4783 + }, + { + "epoch": 0.6085739727770004, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7974805235862732, + "num_tokens": 182607665.0, + "step": 4784 + }, + { + "epoch": 0.6087011830555908, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8013390302658081, + "num_tokens": 182650503.0, + "step": 4785 + }, + { + "epoch": 0.6088283933341814, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8019661903381348, + "num_tokens": 182681662.0, + "step": 4786 + }, + { + "epoch": 0.6089556036127719, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7783815264701843, + "num_tokens": 182714912.0, + "step": 4787 + }, + { + "epoch": 0.6090828138913624, + "grad_norm": 78.0, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7903754711151123, + "num_tokens": 182748813.0, + "step": 4788 + }, + { + "epoch": 0.609210024169953, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7745189070701599, + "num_tokens": 182786123.0, + "step": 4789 + }, + { + "epoch": 0.6093372344485435, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.80648273229599, + "num_tokens": 182820550.0, + "step": 4790 + }, + { + "epoch": 0.6094644447271339, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.798508882522583, + "num_tokens": 182863135.0, + "step": 4791 + }, + { + "epoch": 0.6095916550057244, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.798440158367157, + "num_tokens": 182905193.0, + "step": 4792 + }, + { + "epoch": 0.609718865284315, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7961246967315674, + "num_tokens": 182944934.0, + "step": 4793 + }, + { + "epoch": 0.6098460755629055, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7840896844863892, + "num_tokens": 182984604.0, + "step": 4794 + }, + { + "epoch": 0.609973285841496, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7910647392272949, + "num_tokens": 183020421.0, + "step": 4795 + }, + { + "epoch": 0.6101004961200865, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7829105257987976, + "num_tokens": 183057635.0, + "step": 4796 + }, + { + "epoch": 0.610227706398677, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7782819271087646, + "num_tokens": 183094689.0, + "step": 4797 + }, + { + "epoch": 0.6103549166772675, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8033140301704407, + "num_tokens": 183129118.0, + "step": 4798 + }, + { + "epoch": 0.610482126955858, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7945889234542847, + "num_tokens": 183163365.0, + "step": 4799 + }, + { + "epoch": 0.6106093372344485, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.8012282848358154, + "num_tokens": 183199271.0, + "step": 4800 + }, + { + "epoch": 0.6107365475130391, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.811513364315033, + "num_tokens": 183234523.0, + "step": 4801 + }, + { + "epoch": 0.6108637577916296, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7940237522125244, + "num_tokens": 183267098.0, + "step": 4802 + }, + { + "epoch": 0.61099096807022, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7925817966461182, + "num_tokens": 183306825.0, + "step": 4803 + }, + { + "epoch": 0.6111181783488105, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7972978353500366, + "num_tokens": 183348712.0, + "step": 4804 + }, + { + "epoch": 0.6112453886274011, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8071086406707764, + "num_tokens": 183383797.0, + "step": 4805 + }, + { + "epoch": 0.6113725989059916, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8070203065872192, + "num_tokens": 183417484.0, + "step": 4806 + }, + { + "epoch": 0.6114998091845821, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8030543327331543, + "num_tokens": 183453808.0, + "step": 4807 + }, + { + "epoch": 0.6116270194631727, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8092724680900574, + "num_tokens": 183496693.0, + "step": 4808 + }, + { + "epoch": 0.6117542297417632, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7989128828048706, + "num_tokens": 183536727.0, + "step": 4809 + }, + { + "epoch": 0.6118814400203536, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7951496243476868, + "num_tokens": 183571809.0, + "step": 4810 + }, + { + "epoch": 0.6120086502989441, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7963744401931763, + "num_tokens": 183610761.0, + "step": 4811 + }, + { + "epoch": 0.6121358605775347, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7920920848846436, + "num_tokens": 183648304.0, + "step": 4812 + }, + { + "epoch": 0.6122630708561252, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7829104661941528, + "num_tokens": 183684201.0, + "step": 4813 + }, + { + "epoch": 0.6123902811347157, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7897031307220459, + "num_tokens": 183726595.0, + "step": 4814 + }, + { + "epoch": 0.6125174914133062, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8060749769210815, + "num_tokens": 183762222.0, + "step": 4815 + }, + { + "epoch": 0.6126447016918967, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8099747896194458, + "num_tokens": 183797167.0, + "step": 4816 + }, + { + "epoch": 0.6127719119704872, + "grad_norm": 62.5, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7971503734588623, + "num_tokens": 183839591.0, + "step": 4817 + }, + { + "epoch": 0.6128991222490777, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7944244146347046, + "num_tokens": 183877862.0, + "step": 4818 + }, + { + "epoch": 0.6130263325276682, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7997069358825684, + "num_tokens": 183915964.0, + "step": 4819 + }, + { + "epoch": 0.6131535428062588, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7882930040359497, + "num_tokens": 183952823.0, + "step": 4820 + }, + { + "epoch": 0.6132807530848493, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7908481359481812, + "num_tokens": 183991650.0, + "step": 4821 + }, + { + "epoch": 0.6134079633634397, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7979814410209656, + "num_tokens": 184027768.0, + "step": 4822 + }, + { + "epoch": 0.6135351736420303, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8020131587982178, + "num_tokens": 184063754.0, + "step": 4823 + }, + { + "epoch": 0.6136623839206208, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7890477776527405, + "num_tokens": 184105286.0, + "step": 4824 + }, + { + "epoch": 0.6137895941992113, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7944782376289368, + "num_tokens": 184139312.0, + "step": 4825 + }, + { + "epoch": 0.6139168044778018, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.797065794467926, + "num_tokens": 184181541.0, + "step": 4826 + }, + { + "epoch": 0.6140440147563924, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.791576087474823, + "num_tokens": 184216857.0, + "step": 4827 + }, + { + "epoch": 0.6141712250349828, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7988077998161316, + "num_tokens": 184253318.0, + "step": 4828 + }, + { + "epoch": 0.6142984353135733, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8115084767341614, + "num_tokens": 184298625.0, + "step": 4829 + }, + { + "epoch": 0.6144256455921638, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7910923957824707, + "num_tokens": 184334569.0, + "step": 4830 + }, + { + "epoch": 0.6145528558707544, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7808641791343689, + "num_tokens": 184377477.0, + "step": 4831 + }, + { + "epoch": 0.6146800661493449, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7812730669975281, + "num_tokens": 184416671.0, + "step": 4832 + }, + { + "epoch": 0.6148072764279354, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.78229820728302, + "num_tokens": 184453174.0, + "step": 4833 + }, + { + "epoch": 0.6149344867065258, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.8014907836914062, + "num_tokens": 184488313.0, + "step": 4834 + }, + { + "epoch": 0.6150616969851164, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7733496427536011, + "num_tokens": 184523733.0, + "step": 4835 + }, + { + "epoch": 0.6151889072637069, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7966965436935425, + "num_tokens": 184568399.0, + "step": 4836 + }, + { + "epoch": 0.6153161175422974, + "grad_norm": 95.0, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7958965301513672, + "num_tokens": 184603302.0, + "step": 4837 + }, + { + "epoch": 0.615443327820888, + "grad_norm": 120.5, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8091307878494263, + "num_tokens": 184637346.0, + "step": 4838 + }, + { + "epoch": 0.6155705380994785, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8009679913520813, + "num_tokens": 184671687.0, + "step": 4839 + }, + { + "epoch": 0.6156977483780689, + "grad_norm": 97.0, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8063489198684692, + "num_tokens": 184706867.0, + "step": 4840 + }, + { + "epoch": 0.6158249586566594, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7906637191772461, + "num_tokens": 184743916.0, + "step": 4841 + }, + { + "epoch": 0.61595216893525, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8059811592102051, + "num_tokens": 184783247.0, + "step": 4842 + }, + { + "epoch": 0.6160793792138405, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7962021231651306, + "num_tokens": 184826501.0, + "step": 4843 + }, + { + "epoch": 0.616206589492431, + "grad_norm": 134.0, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7916398048400879, + "num_tokens": 184868743.0, + "step": 4844 + }, + { + "epoch": 0.6163337997710215, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7940350770950317, + "num_tokens": 184908992.0, + "step": 4845 + }, + { + "epoch": 0.616461010049612, + "grad_norm": 133.0, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7803762555122375, + "num_tokens": 184942456.0, + "step": 4846 + }, + { + "epoch": 0.6165882203282025, + "grad_norm": 162.0, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7802841663360596, + "num_tokens": 184975834.0, + "step": 4847 + }, + { + "epoch": 0.616715430606793, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8198715448379517, + "num_tokens": 185012295.0, + "step": 4848 + }, + { + "epoch": 0.6168426408853835, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8017131686210632, + "num_tokens": 185049272.0, + "step": 4849 + }, + { + "epoch": 0.6169698511639741, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.802353024482727, + "num_tokens": 185081054.0, + "step": 4850 + }, + { + "epoch": 0.6170970614425646, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7881867289543152, + "num_tokens": 185119809.0, + "step": 4851 + }, + { + "epoch": 0.617224271721155, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8056120276451111, + "num_tokens": 185161462.0, + "step": 4852 + }, + { + "epoch": 0.6173514819997455, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7882946729660034, + "num_tokens": 185196238.0, + "step": 4853 + }, + { + "epoch": 0.6174786922783361, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7993628978729248, + "num_tokens": 185239317.0, + "step": 4854 + }, + { + "epoch": 0.6176059025569266, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8061733245849609, + "num_tokens": 185273555.0, + "step": 4855 + }, + { + "epoch": 0.6177331128355171, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8040223121643066, + "num_tokens": 185309721.0, + "step": 4856 + }, + { + "epoch": 0.6178603231141077, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7915798425674438, + "num_tokens": 185350846.0, + "step": 4857 + }, + { + "epoch": 0.6179875333926982, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7886813879013062, + "num_tokens": 185392472.0, + "step": 4858 + }, + { + "epoch": 0.6181147436712886, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.794213056564331, + "num_tokens": 185438250.0, + "step": 4859 + }, + { + "epoch": 0.6182419539498791, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7925118803977966, + "num_tokens": 185479455.0, + "step": 4860 + }, + { + "epoch": 0.6183691642284697, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7960577011108398, + "num_tokens": 185518719.0, + "step": 4861 + }, + { + "epoch": 0.6184963745070602, + "grad_norm": 101.5, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7778430581092834, + "num_tokens": 185556671.0, + "step": 4862 + }, + { + "epoch": 0.6186235847856507, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7876728773117065, + "num_tokens": 185595063.0, + "step": 4863 + }, + { + "epoch": 0.6187507950642412, + "grad_norm": 90.5, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7824077606201172, + "num_tokens": 185635626.0, + "step": 4864 + }, + { + "epoch": 0.6188780053428317, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7809858918190002, + "num_tokens": 185671687.0, + "step": 4865 + }, + { + "epoch": 0.6190052156214222, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7959045171737671, + "num_tokens": 185706459.0, + "step": 4866 + }, + { + "epoch": 0.6191324259000127, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8129910826683044, + "num_tokens": 185743162.0, + "step": 4867 + }, + { + "epoch": 0.6192596361786032, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7788545489311218, + "num_tokens": 185784036.0, + "step": 4868 + }, + { + "epoch": 0.6193868464571938, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.789980411529541, + "num_tokens": 185821407.0, + "step": 4869 + }, + { + "epoch": 0.6195140567357843, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.801483690738678, + "num_tokens": 185857789.0, + "step": 4870 + }, + { + "epoch": 0.6196412670143747, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8062288165092468, + "num_tokens": 185896989.0, + "step": 4871 + }, + { + "epoch": 0.6197684772929652, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8050453662872314, + "num_tokens": 185938123.0, + "step": 4872 + }, + { + "epoch": 0.6198956875715558, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7933405637741089, + "num_tokens": 185973440.0, + "step": 4873 + }, + { + "epoch": 0.6200228978501463, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.799582839012146, + "num_tokens": 186010069.0, + "step": 4874 + }, + { + "epoch": 0.6201501081287368, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7884783744812012, + "num_tokens": 186048365.0, + "step": 4875 + }, + { + "epoch": 0.6202773184073274, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7991342544555664, + "num_tokens": 186085358.0, + "step": 4876 + }, + { + "epoch": 0.6204045286859178, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7954167127609253, + "num_tokens": 186118476.0, + "step": 4877 + }, + { + "epoch": 0.6205317389645083, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7898186445236206, + "num_tokens": 186155487.0, + "step": 4878 + }, + { + "epoch": 0.6206589492430988, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8048398494720459, + "num_tokens": 186192941.0, + "step": 4879 + }, + { + "epoch": 0.6207861595216894, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.793282151222229, + "num_tokens": 186233860.0, + "step": 4880 + }, + { + "epoch": 0.6209133698002799, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7785722017288208, + "num_tokens": 186273081.0, + "step": 4881 + }, + { + "epoch": 0.6210405800788704, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7992566823959351, + "num_tokens": 186310596.0, + "step": 4882 + }, + { + "epoch": 0.6211677903574608, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7935101985931396, + "num_tokens": 186346262.0, + "step": 4883 + }, + { + "epoch": 0.6212950006360514, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.797038197517395, + "num_tokens": 186386508.0, + "step": 4884 + }, + { + "epoch": 0.6214222109146419, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7917253971099854, + "num_tokens": 186425184.0, + "step": 4885 + }, + { + "epoch": 0.6215494211932324, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7928507924079895, + "num_tokens": 186463072.0, + "step": 4886 + }, + { + "epoch": 0.621676631471823, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8115929365158081, + "num_tokens": 186504062.0, + "step": 4887 + }, + { + "epoch": 0.6218038417504135, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7874242067337036, + "num_tokens": 186544564.0, + "step": 4888 + }, + { + "epoch": 0.6219310520290039, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7779534459114075, + "num_tokens": 186587196.0, + "step": 4889 + }, + { + "epoch": 0.6220582623075944, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.794737696647644, + "num_tokens": 186620673.0, + "step": 4890 + }, + { + "epoch": 0.622185472586185, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7981372475624084, + "num_tokens": 186654028.0, + "step": 4891 + }, + { + "epoch": 0.6223126828647755, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7895214557647705, + "num_tokens": 186693395.0, + "step": 4892 + }, + { + "epoch": 0.622439893143366, + "grad_norm": 99.0, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7753174304962158, + "num_tokens": 186734440.0, + "step": 4893 + }, + { + "epoch": 0.6225671034219565, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7976205945014954, + "num_tokens": 186769637.0, + "step": 4894 + }, + { + "epoch": 0.622694313700547, + "grad_norm": 97.5, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7918627262115479, + "num_tokens": 186810188.0, + "step": 4895 + }, + { + "epoch": 0.6228215239791375, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7801566123962402, + "num_tokens": 186847512.0, + "step": 4896 + }, + { + "epoch": 0.622948734257728, + "grad_norm": 91.5, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8037671446800232, + "num_tokens": 186878721.0, + "step": 4897 + }, + { + "epoch": 0.6230759445363185, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8067516684532166, + "num_tokens": 186920239.0, + "step": 4898 + }, + { + "epoch": 0.6232031548149091, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7850267887115479, + "num_tokens": 186964158.0, + "step": 4899 + }, + { + "epoch": 0.6233303650934996, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8139068484306335, + "num_tokens": 187003176.0, + "step": 4900 + }, + { + "epoch": 0.62345757537209, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.794238269329071, + "num_tokens": 187046152.0, + "step": 4901 + }, + { + "epoch": 0.6235847856506805, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8079427480697632, + "num_tokens": 187086629.0, + "step": 4902 + }, + { + "epoch": 0.6237119959292711, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7868228554725647, + "num_tokens": 187122856.0, + "step": 4903 + }, + { + "epoch": 0.6238392062078616, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8121768832206726, + "num_tokens": 187161568.0, + "step": 4904 + }, + { + "epoch": 0.6239664164864521, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7970438599586487, + "num_tokens": 187202831.0, + "step": 4905 + }, + { + "epoch": 0.6240936267650427, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8019101619720459, + "num_tokens": 187239059.0, + "step": 4906 + }, + { + "epoch": 0.6242208370436332, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7922830581665039, + "num_tokens": 187272940.0, + "step": 4907 + }, + { + "epoch": 0.6243480473222236, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8098745346069336, + "num_tokens": 187310963.0, + "step": 4908 + }, + { + "epoch": 0.6244752576008141, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8177490234375, + "num_tokens": 187345698.0, + "step": 4909 + }, + { + "epoch": 0.6246024678794047, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7942492365837097, + "num_tokens": 187384762.0, + "step": 4910 + }, + { + "epoch": 0.6247296781579952, + "grad_norm": 151.0, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7788947224617004, + "num_tokens": 187424288.0, + "step": 4911 + }, + { + "epoch": 0.6248568884365857, + "grad_norm": 61.5, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7985025644302368, + "num_tokens": 187461481.0, + "step": 4912 + }, + { + "epoch": 0.6249840987151762, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8123254776000977, + "num_tokens": 187497690.0, + "step": 4913 + }, + { + "epoch": 0.6251113089937667, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7867152094841003, + "num_tokens": 187537825.0, + "step": 4914 + }, + { + "epoch": 0.6252385192723572, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8086317777633667, + "num_tokens": 187583998.0, + "step": 4915 + }, + { + "epoch": 0.6253657295509477, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7897566556930542, + "num_tokens": 187621962.0, + "step": 4916 + }, + { + "epoch": 0.6254929398295382, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8006480932235718, + "num_tokens": 187655913.0, + "step": 4917 + }, + { + "epoch": 0.6256201501081288, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7864762544631958, + "num_tokens": 187692433.0, + "step": 4918 + }, + { + "epoch": 0.6257473603867193, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8048782348632812, + "num_tokens": 187724311.0, + "step": 4919 + }, + { + "epoch": 0.6258745706653097, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8076560497283936, + "num_tokens": 187766614.0, + "step": 4920 + }, + { + "epoch": 0.6260017809439002, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8061928153038025, + "num_tokens": 187804847.0, + "step": 4921 + }, + { + "epoch": 0.6261289912224908, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.794536292552948, + "num_tokens": 187842791.0, + "step": 4922 + }, + { + "epoch": 0.6262562015010813, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7860867977142334, + "num_tokens": 187878304.0, + "step": 4923 + }, + { + "epoch": 0.6263834117796718, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7977372407913208, + "num_tokens": 187915022.0, + "step": 4924 + }, + { + "epoch": 0.6265106220582624, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7943975925445557, + "num_tokens": 187950533.0, + "step": 4925 + }, + { + "epoch": 0.6266378323368528, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7906427383422852, + "num_tokens": 187989893.0, + "step": 4926 + }, + { + "epoch": 0.6267650426154433, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7891457080841064, + "num_tokens": 188025031.0, + "step": 4927 + }, + { + "epoch": 0.6268922528940338, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7860552072525024, + "num_tokens": 188060797.0, + "step": 4928 + }, + { + "epoch": 0.6270194631726244, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7868480682373047, + "num_tokens": 188096271.0, + "step": 4929 + }, + { + "epoch": 0.6271466734512149, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7928045988082886, + "num_tokens": 188136697.0, + "step": 4930 + }, + { + "epoch": 0.6272738837298054, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7969936728477478, + "num_tokens": 188174715.0, + "step": 4931 + }, + { + "epoch": 0.6274010940083958, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7721114158630371, + "num_tokens": 188216379.0, + "step": 4932 + }, + { + "epoch": 0.6275283042869864, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8032300472259521, + "num_tokens": 188252183.0, + "step": 4933 + }, + { + "epoch": 0.6276555145655769, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7982399463653564, + "num_tokens": 188291128.0, + "step": 4934 + }, + { + "epoch": 0.6277827248441674, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.778344452381134, + "num_tokens": 188334256.0, + "step": 4935 + }, + { + "epoch": 0.627909935122758, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7873312830924988, + "num_tokens": 188373207.0, + "step": 4936 + }, + { + "epoch": 0.6280371454013485, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7923523187637329, + "num_tokens": 188414937.0, + "step": 4937 + }, + { + "epoch": 0.6281643556799389, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7925413846969604, + "num_tokens": 188455398.0, + "step": 4938 + }, + { + "epoch": 0.6282915659585294, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7894415855407715, + "num_tokens": 188496561.0, + "step": 4939 + }, + { + "epoch": 0.62841877623712, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7958371043205261, + "num_tokens": 188531318.0, + "step": 4940 + }, + { + "epoch": 0.6285459865157105, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7850498557090759, + "num_tokens": 188566807.0, + "step": 4941 + }, + { + "epoch": 0.628673196794301, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7953976988792419, + "num_tokens": 188608062.0, + "step": 4942 + }, + { + "epoch": 0.6288004070728915, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.8012694120407104, + "num_tokens": 188644953.0, + "step": 4943 + }, + { + "epoch": 0.628927617351482, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8174660205841064, + "num_tokens": 188687720.0, + "step": 4944 + }, + { + "epoch": 0.6290548276300725, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.790812075138092, + "num_tokens": 188730077.0, + "step": 4945 + }, + { + "epoch": 0.629182037908663, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8010410666465759, + "num_tokens": 188768306.0, + "step": 4946 + }, + { + "epoch": 0.6293092481872535, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7776639461517334, + "num_tokens": 188807480.0, + "step": 4947 + }, + { + "epoch": 0.6294364584658441, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7754859924316406, + "num_tokens": 188836383.0, + "step": 4948 + }, + { + "epoch": 0.6295636687444346, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7970468401908875, + "num_tokens": 188876079.0, + "step": 4949 + }, + { + "epoch": 0.629690879023025, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7897595763206482, + "num_tokens": 188912936.0, + "step": 4950 + }, + { + "epoch": 0.6298180893016155, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7932488322257996, + "num_tokens": 188951475.0, + "step": 4951 + }, + { + "epoch": 0.6299452995802061, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7790058851242065, + "num_tokens": 188996972.0, + "step": 4952 + }, + { + "epoch": 0.6300725098587966, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.8194842338562012, + "num_tokens": 189030706.0, + "step": 4953 + }, + { + "epoch": 0.6301997201373871, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8202255964279175, + "num_tokens": 189065638.0, + "step": 4954 + }, + { + "epoch": 0.6303269304159776, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8113448619842529, + "num_tokens": 189101747.0, + "step": 4955 + }, + { + "epoch": 0.6304541406945681, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7927367687225342, + "num_tokens": 189142965.0, + "step": 4956 + }, + { + "epoch": 0.6305813509731586, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7941096425056458, + "num_tokens": 189177822.0, + "step": 4957 + }, + { + "epoch": 0.6307085612517491, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.786332368850708, + "num_tokens": 189215710.0, + "step": 4958 + }, + { + "epoch": 0.6308357715303397, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.763221025466919, + "num_tokens": 189250788.0, + "step": 4959 + }, + { + "epoch": 0.6309629818089302, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7864366769790649, + "num_tokens": 189278451.0, + "step": 4960 + }, + { + "epoch": 0.6310901920875207, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8026536107063293, + "num_tokens": 189316857.0, + "step": 4961 + }, + { + "epoch": 0.6312174023661112, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8079921007156372, + "num_tokens": 189356600.0, + "step": 4962 + }, + { + "epoch": 0.6313446126447017, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7909535765647888, + "num_tokens": 189393134.0, + "step": 4963 + }, + { + "epoch": 0.6314718229232922, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7795253396034241, + "num_tokens": 189433431.0, + "step": 4964 + }, + { + "epoch": 0.6315990332018827, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7767212390899658, + "num_tokens": 189473738.0, + "step": 4965 + }, + { + "epoch": 0.6317262434804732, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8074831962585449, + "num_tokens": 189509930.0, + "step": 4966 + }, + { + "epoch": 0.6318534537590638, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8049085140228271, + "num_tokens": 189547801.0, + "step": 4967 + }, + { + "epoch": 0.6319806640376543, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8064709901809692, + "num_tokens": 189582226.0, + "step": 4968 + }, + { + "epoch": 0.6321078743162447, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7944004535675049, + "num_tokens": 189622915.0, + "step": 4969 + }, + { + "epoch": 0.6322350845948352, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8014020323753357, + "num_tokens": 189658182.0, + "step": 4970 + }, + { + "epoch": 0.6323622948734258, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7928432822227478, + "num_tokens": 189696627.0, + "step": 4971 + }, + { + "epoch": 0.6324895051520163, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7912750244140625, + "num_tokens": 189729826.0, + "step": 4972 + }, + { + "epoch": 0.6326167154306068, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8002787232398987, + "num_tokens": 189771504.0, + "step": 4973 + }, + { + "epoch": 0.6327439257091974, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8103460073471069, + "num_tokens": 189809220.0, + "step": 4974 + }, + { + "epoch": 0.6328711359877878, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8017787933349609, + "num_tokens": 189849586.0, + "step": 4975 + }, + { + "epoch": 0.6329983462663783, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7961233854293823, + "num_tokens": 189885054.0, + "step": 4976 + }, + { + "epoch": 0.6331255565449688, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7823700308799744, + "num_tokens": 189918286.0, + "step": 4977 + }, + { + "epoch": 0.6332527668235594, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7771816849708557, + "num_tokens": 189956845.0, + "step": 4978 + }, + { + "epoch": 0.6333799771021499, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8138271570205688, + "num_tokens": 189994827.0, + "step": 4979 + }, + { + "epoch": 0.6335071873807404, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7937026619911194, + "num_tokens": 190036066.0, + "step": 4980 + }, + { + "epoch": 0.6336343976593308, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.807054877281189, + "num_tokens": 190073326.0, + "step": 4981 + }, + { + "epoch": 0.6337616079379214, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7916207313537598, + "num_tokens": 190109125.0, + "step": 4982 + }, + { + "epoch": 0.6338888182165119, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7976008653640747, + "num_tokens": 190148520.0, + "step": 4983 + }, + { + "epoch": 0.6340160284951024, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7940707206726074, + "num_tokens": 190184272.0, + "step": 4984 + }, + { + "epoch": 0.6341432387736929, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7982767224311829, + "num_tokens": 190221483.0, + "step": 4985 + }, + { + "epoch": 0.6342704490522835, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8082280158996582, + "num_tokens": 190266430.0, + "step": 4986 + }, + { + "epoch": 0.6343976593308739, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8115667700767517, + "num_tokens": 190308575.0, + "step": 4987 + }, + { + "epoch": 0.6345248696094644, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8070236444473267, + "num_tokens": 190344584.0, + "step": 4988 + }, + { + "epoch": 0.6346520798880549, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7989353537559509, + "num_tokens": 190376497.0, + "step": 4989 + }, + { + "epoch": 0.6347792901666455, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.783625602722168, + "num_tokens": 190416879.0, + "step": 4990 + }, + { + "epoch": 0.634906500445236, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8181841373443604, + "num_tokens": 190454684.0, + "step": 4991 + }, + { + "epoch": 0.6350337107238265, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7892338633537292, + "num_tokens": 190490524.0, + "step": 4992 + }, + { + "epoch": 0.635160921002417, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.8017940521240234, + "num_tokens": 190526558.0, + "step": 4993 + }, + { + "epoch": 0.6352881312810075, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8022391200065613, + "num_tokens": 190571206.0, + "step": 4994 + }, + { + "epoch": 0.635415341559598, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7875230312347412, + "num_tokens": 190611480.0, + "step": 4995 + }, + { + "epoch": 0.6355425518381885, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7654595375061035, + "num_tokens": 190647699.0, + "step": 4996 + }, + { + "epoch": 0.6356697621167791, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7957134246826172, + "num_tokens": 190684863.0, + "step": 4997 + }, + { + "epoch": 0.6357969723953696, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7835999727249146, + "num_tokens": 190725006.0, + "step": 4998 + }, + { + "epoch": 0.63592418267396, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7941883206367493, + "num_tokens": 190757431.0, + "step": 4999 + }, + { + "epoch": 0.6360513929525505, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7975946664810181, + "num_tokens": 190793118.0, + "step": 5000 + }, + { + "epoch": 0.6361786032311411, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8081604242324829, + "num_tokens": 190837105.0, + "step": 5001 + }, + { + "epoch": 0.6363058135097316, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7912192344665527, + "num_tokens": 190875372.0, + "step": 5002 + }, + { + "epoch": 0.6364330237883221, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8079843521118164, + "num_tokens": 190921307.0, + "step": 5003 + }, + { + "epoch": 0.6365602340669126, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7963934540748596, + "num_tokens": 190967919.0, + "step": 5004 + }, + { + "epoch": 0.6366874443455031, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7930465340614319, + "num_tokens": 191005668.0, + "step": 5005 + }, + { + "epoch": 0.6368146546240936, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7869857549667358, + "num_tokens": 191044455.0, + "step": 5006 + }, + { + "epoch": 0.6369418649026841, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8005617260932922, + "num_tokens": 191086369.0, + "step": 5007 + }, + { + "epoch": 0.6370690751812746, + "grad_norm": 63.75, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7925331592559814, + "num_tokens": 191127070.0, + "step": 5008 + }, + { + "epoch": 0.6371962854598652, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8125772476196289, + "num_tokens": 191163623.0, + "step": 5009 + }, + { + "epoch": 0.6373234957384557, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.789153516292572, + "num_tokens": 191196790.0, + "step": 5010 + }, + { + "epoch": 0.6374507060170462, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7928240299224854, + "num_tokens": 191230265.0, + "step": 5011 + }, + { + "epoch": 0.6375779162956366, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7851516604423523, + "num_tokens": 191266781.0, + "step": 5012 + }, + { + "epoch": 0.6377051265742272, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7848379015922546, + "num_tokens": 191305869.0, + "step": 5013 + }, + { + "epoch": 0.6378323368528177, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.784514307975769, + "num_tokens": 191343498.0, + "step": 5014 + }, + { + "epoch": 0.6379595471314082, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7806597352027893, + "num_tokens": 191383260.0, + "step": 5015 + }, + { + "epoch": 0.6380867574099988, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7849295139312744, + "num_tokens": 191417350.0, + "step": 5016 + }, + { + "epoch": 0.6382139676885893, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7963453531265259, + "num_tokens": 191454336.0, + "step": 5017 + }, + { + "epoch": 0.6383411779671797, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.799491286277771, + "num_tokens": 191489977.0, + "step": 5018 + }, + { + "epoch": 0.6384683882457702, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7841601371765137, + "num_tokens": 191524622.0, + "step": 5019 + }, + { + "epoch": 0.6385955985243608, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7866665124893188, + "num_tokens": 191565389.0, + "step": 5020 + }, + { + "epoch": 0.6387228088029513, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6097, + "mean_token_accuracy": 0.8138166666030884, + "num_tokens": 191604640.0, + "step": 5021 + }, + { + "epoch": 0.6388500190815418, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7966581583023071, + "num_tokens": 191651029.0, + "step": 5022 + }, + { + "epoch": 0.6389772293601323, + "grad_norm": 118.5, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7952739596366882, + "num_tokens": 191690524.0, + "step": 5023 + }, + { + "epoch": 0.6391044396387228, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7938882112503052, + "num_tokens": 191733992.0, + "step": 5024 + }, + { + "epoch": 0.6392316499173133, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8094613552093506, + "num_tokens": 191769717.0, + "step": 5025 + }, + { + "epoch": 0.6393588601959038, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7829194664955139, + "num_tokens": 191813334.0, + "step": 5026 + }, + { + "epoch": 0.6394860704744944, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7960342764854431, + "num_tokens": 191848295.0, + "step": 5027 + }, + { + "epoch": 0.6396132807530849, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7852762937545776, + "num_tokens": 191885329.0, + "step": 5028 + }, + { + "epoch": 0.6397404910316754, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8054277896881104, + "num_tokens": 191918070.0, + "step": 5029 + }, + { + "epoch": 0.6398677013102658, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.803633987903595, + "num_tokens": 191966084.0, + "step": 5030 + }, + { + "epoch": 0.6399949115888564, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7922762632369995, + "num_tokens": 192004907.0, + "step": 5031 + }, + { + "epoch": 0.6401221218674469, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7760748267173767, + "num_tokens": 192042942.0, + "step": 5032 + }, + { + "epoch": 0.6402493321460374, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7953065037727356, + "num_tokens": 192081133.0, + "step": 5033 + }, + { + "epoch": 0.6403765424246279, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7927915453910828, + "num_tokens": 192120171.0, + "step": 5034 + }, + { + "epoch": 0.6405037527032185, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7928440570831299, + "num_tokens": 192159583.0, + "step": 5035 + }, + { + "epoch": 0.6406309629818089, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.789764404296875, + "num_tokens": 192199513.0, + "step": 5036 + }, + { + "epoch": 0.6407581732603994, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7914244532585144, + "num_tokens": 192228352.0, + "step": 5037 + }, + { + "epoch": 0.6408853835389899, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7894703149795532, + "num_tokens": 192264480.0, + "step": 5038 + }, + { + "epoch": 0.6410125938175805, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7976366281509399, + "num_tokens": 192303307.0, + "step": 5039 + }, + { + "epoch": 0.641139804096171, + "grad_norm": 98.5, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7955482006072998, + "num_tokens": 192342064.0, + "step": 5040 + }, + { + "epoch": 0.6412670143747615, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.802249014377594, + "num_tokens": 192379297.0, + "step": 5041 + }, + { + "epoch": 0.6413942246533519, + "grad_norm": 114.5, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7987275719642639, + "num_tokens": 192416087.0, + "step": 5042 + }, + { + "epoch": 0.6415214349319425, + "grad_norm": 146.0, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8034697771072388, + "num_tokens": 192454421.0, + "step": 5043 + }, + { + "epoch": 0.641648645210533, + "grad_norm": 150.0, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7928998470306396, + "num_tokens": 192493860.0, + "step": 5044 + }, + { + "epoch": 0.6417758554891235, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7840548753738403, + "num_tokens": 192533672.0, + "step": 5045 + }, + { + "epoch": 0.641903065767714, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7819363474845886, + "num_tokens": 192569569.0, + "step": 5046 + }, + { + "epoch": 0.6420302760463046, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.805777370929718, + "num_tokens": 192610716.0, + "step": 5047 + }, + { + "epoch": 0.642157486324895, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7848401665687561, + "num_tokens": 192648040.0, + "step": 5048 + }, + { + "epoch": 0.6422846966034855, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.79655522108078, + "num_tokens": 192690063.0, + "step": 5049 + }, + { + "epoch": 0.6424119068820761, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7949839234352112, + "num_tokens": 192732067.0, + "step": 5050 + }, + { + "epoch": 0.6425391171606666, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8073602318763733, + "num_tokens": 192768598.0, + "step": 5051 + }, + { + "epoch": 0.6426663274392571, + "grad_norm": 128.0, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7600991725921631, + "num_tokens": 192805420.0, + "step": 5052 + }, + { + "epoch": 0.6427935377178476, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7931863069534302, + "num_tokens": 192852082.0, + "step": 5053 + }, + { + "epoch": 0.6429207479964381, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7964313626289368, + "num_tokens": 192892846.0, + "step": 5054 + }, + { + "epoch": 0.6430479582750286, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8132932782173157, + "num_tokens": 192932091.0, + "step": 5055 + }, + { + "epoch": 0.6431751685536191, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7858092188835144, + "num_tokens": 192960628.0, + "step": 5056 + }, + { + "epoch": 0.6433023788322096, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7962343692779541, + "num_tokens": 193002075.0, + "step": 5057 + }, + { + "epoch": 0.6434295891108002, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7870578169822693, + "num_tokens": 193039720.0, + "step": 5058 + }, + { + "epoch": 0.6435567993893907, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7901514768600464, + "num_tokens": 193072806.0, + "step": 5059 + }, + { + "epoch": 0.6436840096679812, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8033579587936401, + "num_tokens": 193108478.0, + "step": 5060 + }, + { + "epoch": 0.6438112199465716, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7983471155166626, + "num_tokens": 193150663.0, + "step": 5061 + }, + { + "epoch": 0.6439384302251622, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7968515157699585, + "num_tokens": 193191499.0, + "step": 5062 + }, + { + "epoch": 0.6440656405037527, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7889079451560974, + "num_tokens": 193236103.0, + "step": 5063 + }, + { + "epoch": 0.6441928507823432, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7720423936843872, + "num_tokens": 193274095.0, + "step": 5064 + }, + { + "epoch": 0.6443200610609338, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8053789734840393, + "num_tokens": 193316988.0, + "step": 5065 + }, + { + "epoch": 0.6444472713395243, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7943565249443054, + "num_tokens": 193357430.0, + "step": 5066 + }, + { + "epoch": 0.6445744816181147, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8021808862686157, + "num_tokens": 193390392.0, + "step": 5067 + }, + { + "epoch": 0.6447016918967052, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7950739860534668, + "num_tokens": 193430961.0, + "step": 5068 + }, + { + "epoch": 0.6448289021752958, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7916862368583679, + "num_tokens": 193473767.0, + "step": 5069 + }, + { + "epoch": 0.6449561124538863, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7826278209686279, + "num_tokens": 193511375.0, + "step": 5070 + }, + { + "epoch": 0.6450833227324768, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8000000715255737, + "num_tokens": 193551296.0, + "step": 5071 + }, + { + "epoch": 0.6452105330110673, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.8014390468597412, + "num_tokens": 193596484.0, + "step": 5072 + }, + { + "epoch": 0.6453377432896578, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.8034830093383789, + "num_tokens": 193637513.0, + "step": 5073 + }, + { + "epoch": 0.6454649535682483, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7883279323577881, + "num_tokens": 193672808.0, + "step": 5074 + }, + { + "epoch": 0.6455921638468388, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8142164945602417, + "num_tokens": 193713018.0, + "step": 5075 + }, + { + "epoch": 0.6457193741254293, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.8002029061317444, + "num_tokens": 193757649.0, + "step": 5076 + }, + { + "epoch": 0.6458465844040199, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7929221391677856, + "num_tokens": 193792390.0, + "step": 5077 + }, + { + "epoch": 0.6459737946826104, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8014060258865356, + "num_tokens": 193827721.0, + "step": 5078 + }, + { + "epoch": 0.6461010049612008, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7823677659034729, + "num_tokens": 193867217.0, + "step": 5079 + }, + { + "epoch": 0.6462282152397913, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7986067533493042, + "num_tokens": 193900128.0, + "step": 5080 + }, + { + "epoch": 0.6463554255183819, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8093442320823669, + "num_tokens": 193935744.0, + "step": 5081 + }, + { + "epoch": 0.6464826357969724, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7924289703369141, + "num_tokens": 193977344.0, + "step": 5082 + }, + { + "epoch": 0.6466098460755629, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7613409161567688, + "num_tokens": 194014295.0, + "step": 5083 + }, + { + "epoch": 0.6467370563541535, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.77686607837677, + "num_tokens": 194055245.0, + "step": 5084 + }, + { + "epoch": 0.6468642666327439, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.783437967300415, + "num_tokens": 194089313.0, + "step": 5085 + }, + { + "epoch": 0.6469914769113344, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8059744834899902, + "num_tokens": 194128293.0, + "step": 5086 + }, + { + "epoch": 0.6471186871899249, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7865403890609741, + "num_tokens": 194166632.0, + "step": 5087 + }, + { + "epoch": 0.6472458974685155, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7800144553184509, + "num_tokens": 194209176.0, + "step": 5088 + }, + { + "epoch": 0.647373107747106, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.780900239944458, + "num_tokens": 194245617.0, + "step": 5089 + }, + { + "epoch": 0.6475003180256965, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8018466234207153, + "num_tokens": 194285983.0, + "step": 5090 + }, + { + "epoch": 0.6476275283042869, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7895516753196716, + "num_tokens": 194317088.0, + "step": 5091 + }, + { + "epoch": 0.6477547385828775, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8000767230987549, + "num_tokens": 194354025.0, + "step": 5092 + }, + { + "epoch": 0.647881948861468, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7932670712471008, + "num_tokens": 194388539.0, + "step": 5093 + }, + { + "epoch": 0.6480091591400585, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7928997278213501, + "num_tokens": 194424518.0, + "step": 5094 + }, + { + "epoch": 0.648136369418649, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7930400371551514, + "num_tokens": 194463310.0, + "step": 5095 + }, + { + "epoch": 0.6482635796972396, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7930859923362732, + "num_tokens": 194499911.0, + "step": 5096 + }, + { + "epoch": 0.64839078997583, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7879021763801575, + "num_tokens": 194536731.0, + "step": 5097 + }, + { + "epoch": 0.6485180002544205, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7843329310417175, + "num_tokens": 194574014.0, + "step": 5098 + }, + { + "epoch": 0.648645210533011, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7845076322555542, + "num_tokens": 194611944.0, + "step": 5099 + }, + { + "epoch": 0.6487724208116016, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8050278425216675, + "num_tokens": 194649734.0, + "step": 5100 + }, + { + "epoch": 0.6488996310901921, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.8044141530990601, + "num_tokens": 194688344.0, + "step": 5101 + }, + { + "epoch": 0.6490268413687826, + "grad_norm": 99.5, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7779477834701538, + "num_tokens": 194728582.0, + "step": 5102 + }, + { + "epoch": 0.649154051647373, + "grad_norm": 93.0, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7885566353797913, + "num_tokens": 194762753.0, + "step": 5103 + }, + { + "epoch": 0.6492812619259636, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8083003163337708, + "num_tokens": 194802325.0, + "step": 5104 + }, + { + "epoch": 0.6494084722045541, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7987509369850159, + "num_tokens": 194846349.0, + "step": 5105 + }, + { + "epoch": 0.6495356824831446, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7950554490089417, + "num_tokens": 194880334.0, + "step": 5106 + }, + { + "epoch": 0.6496628927617352, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7883448600769043, + "num_tokens": 194919343.0, + "step": 5107 + }, + { + "epoch": 0.6497901030403257, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7978640198707581, + "num_tokens": 194958800.0, + "step": 5108 + }, + { + "epoch": 0.6499173133189162, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7679616212844849, + "num_tokens": 194997249.0, + "step": 5109 + }, + { + "epoch": 0.6500445235975066, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7975509166717529, + "num_tokens": 195035879.0, + "step": 5110 + }, + { + "epoch": 0.6501717338760972, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7936666011810303, + "num_tokens": 195075925.0, + "step": 5111 + }, + { + "epoch": 0.6502989441546877, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7998871207237244, + "num_tokens": 195110185.0, + "step": 5112 + }, + { + "epoch": 0.6504261544332782, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.789332389831543, + "num_tokens": 195154666.0, + "step": 5113 + }, + { + "epoch": 0.6505533647118688, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8057102560997009, + "num_tokens": 195188998.0, + "step": 5114 + }, + { + "epoch": 0.6506805749904593, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7996585965156555, + "num_tokens": 195224124.0, + "step": 5115 + }, + { + "epoch": 0.6508077852690497, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7917357683181763, + "num_tokens": 195253601.0, + "step": 5116 + }, + { + "epoch": 0.6509349955476402, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7790898084640503, + "num_tokens": 195289045.0, + "step": 5117 + }, + { + "epoch": 0.6510622058262308, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7997155785560608, + "num_tokens": 195326983.0, + "step": 5118 + }, + { + "epoch": 0.6511894161048213, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8041722178459167, + "num_tokens": 195360425.0, + "step": 5119 + }, + { + "epoch": 0.6513166263834118, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7822635173797607, + "num_tokens": 195402648.0, + "step": 5120 + }, + { + "epoch": 0.6514438366620023, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.811585009098053, + "num_tokens": 195440945.0, + "step": 5121 + }, + { + "epoch": 0.6515710469405928, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.798233687877655, + "num_tokens": 195481872.0, + "step": 5122 + }, + { + "epoch": 0.6516982572191833, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8026224374771118, + "num_tokens": 195525460.0, + "step": 5123 + }, + { + "epoch": 0.6518254674977738, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7935819625854492, + "num_tokens": 195563522.0, + "step": 5124 + }, + { + "epoch": 0.6519526777763643, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8033092617988586, + "num_tokens": 195598114.0, + "step": 5125 + }, + { + "epoch": 0.6520798880549549, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7839972376823425, + "num_tokens": 195639826.0, + "step": 5126 + }, + { + "epoch": 0.6522070983335454, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8088158369064331, + "num_tokens": 195682047.0, + "step": 5127 + }, + { + "epoch": 0.6523343086121358, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.796751856803894, + "num_tokens": 195722453.0, + "step": 5128 + }, + { + "epoch": 0.6524615188907263, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.786495566368103, + "num_tokens": 195759182.0, + "step": 5129 + }, + { + "epoch": 0.6525887291693169, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7981979250907898, + "num_tokens": 195795641.0, + "step": 5130 + }, + { + "epoch": 0.6527159394479074, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8077991008758545, + "num_tokens": 195831200.0, + "step": 5131 + }, + { + "epoch": 0.6528431497264979, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8019198775291443, + "num_tokens": 195875110.0, + "step": 5132 + }, + { + "epoch": 0.6529703600050885, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.799152672290802, + "num_tokens": 195916724.0, + "step": 5133 + }, + { + "epoch": 0.6530975702836789, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7889472246170044, + "num_tokens": 195956219.0, + "step": 5134 + }, + { + "epoch": 0.6532247805622694, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7951539754867554, + "num_tokens": 195991563.0, + "step": 5135 + }, + { + "epoch": 0.6533519908408599, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8022447824478149, + "num_tokens": 196031864.0, + "step": 5136 + }, + { + "epoch": 0.6534792011194505, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7881982326507568, + "num_tokens": 196067208.0, + "step": 5137 + }, + { + "epoch": 0.653606411398041, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8023951053619385, + "num_tokens": 196103934.0, + "step": 5138 + }, + { + "epoch": 0.6537336216766315, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7878482341766357, + "num_tokens": 196140886.0, + "step": 5139 + }, + { + "epoch": 0.6538608319552219, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8043897151947021, + "num_tokens": 196178940.0, + "step": 5140 + }, + { + "epoch": 0.6539880422338125, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8019795417785645, + "num_tokens": 196229461.0, + "step": 5141 + }, + { + "epoch": 0.654115252512403, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.809057891368866, + "num_tokens": 196265793.0, + "step": 5142 + }, + { + "epoch": 0.6542424627909935, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8039612174034119, + "num_tokens": 196304919.0, + "step": 5143 + }, + { + "epoch": 0.654369673069584, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8122906684875488, + "num_tokens": 196344761.0, + "step": 5144 + }, + { + "epoch": 0.6544968833481746, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8112161159515381, + "num_tokens": 196382625.0, + "step": 5145 + }, + { + "epoch": 0.654624093626765, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8110367059707642, + "num_tokens": 196422279.0, + "step": 5146 + }, + { + "epoch": 0.6547513039053555, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7913419008255005, + "num_tokens": 196463684.0, + "step": 5147 + }, + { + "epoch": 0.654878514183946, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8015743494033813, + "num_tokens": 196506319.0, + "step": 5148 + }, + { + "epoch": 0.6550057244625366, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7866660356521606, + "num_tokens": 196546367.0, + "step": 5149 + }, + { + "epoch": 0.6551329347411271, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7766058444976807, + "num_tokens": 196576598.0, + "step": 5150 + }, + { + "epoch": 0.6552601450197176, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7876386046409607, + "num_tokens": 196610633.0, + "step": 5151 + }, + { + "epoch": 0.655387355298308, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7891873717308044, + "num_tokens": 196646183.0, + "step": 5152 + }, + { + "epoch": 0.6555145655768986, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8066253662109375, + "num_tokens": 196689454.0, + "step": 5153 + }, + { + "epoch": 0.6556417758554891, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7797961235046387, + "num_tokens": 196731820.0, + "step": 5154 + }, + { + "epoch": 0.6557689861340796, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7869142293930054, + "num_tokens": 196768542.0, + "step": 5155 + }, + { + "epoch": 0.6558961964126702, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7838667035102844, + "num_tokens": 196810300.0, + "step": 5156 + }, + { + "epoch": 0.6560234066912607, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7772335410118103, + "num_tokens": 196847271.0, + "step": 5157 + }, + { + "epoch": 0.6561506169698512, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8068932294845581, + "num_tokens": 196880439.0, + "step": 5158 + }, + { + "epoch": 0.6562778272484416, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7877536416053772, + "num_tokens": 196914911.0, + "step": 5159 + }, + { + "epoch": 0.6564050375270322, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7965620160102844, + "num_tokens": 196949419.0, + "step": 5160 + }, + { + "epoch": 0.6565322478056227, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7887162566184998, + "num_tokens": 196989401.0, + "step": 5161 + }, + { + "epoch": 0.6566594580842132, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7867847681045532, + "num_tokens": 197028377.0, + "step": 5162 + }, + { + "epoch": 0.6567866683628037, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7730894088745117, + "num_tokens": 197068334.0, + "step": 5163 + }, + { + "epoch": 0.6569138786413943, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7980483770370483, + "num_tokens": 197105978.0, + "step": 5164 + }, + { + "epoch": 0.6570410889199847, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7809798121452332, + "num_tokens": 197148641.0, + "step": 5165 + }, + { + "epoch": 0.6571682991985752, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7718410491943359, + "num_tokens": 197188820.0, + "step": 5166 + }, + { + "epoch": 0.6572955094771658, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7824596166610718, + "num_tokens": 197227109.0, + "step": 5167 + }, + { + "epoch": 0.6574227197557563, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.8005501627922058, + "num_tokens": 197264083.0, + "step": 5168 + }, + { + "epoch": 0.6575499300343468, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7910748720169067, + "num_tokens": 197301349.0, + "step": 5169 + }, + { + "epoch": 0.6576771403129373, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7853602170944214, + "num_tokens": 197348176.0, + "step": 5170 + }, + { + "epoch": 0.6578043505915278, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7867720127105713, + "num_tokens": 197380490.0, + "step": 5171 + }, + { + "epoch": 0.6579315608701183, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7814463376998901, + "num_tokens": 197423075.0, + "step": 5172 + }, + { + "epoch": 0.6580587711487088, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7985080480575562, + "num_tokens": 197461730.0, + "step": 5173 + }, + { + "epoch": 0.6581859814272993, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7834702134132385, + "num_tokens": 197505801.0, + "step": 5174 + }, + { + "epoch": 0.6583131917058899, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8041330575942993, + "num_tokens": 197534643.0, + "step": 5175 + }, + { + "epoch": 0.6584404019844804, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8003683090209961, + "num_tokens": 197576490.0, + "step": 5176 + }, + { + "epoch": 0.6585676122630708, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.805269718170166, + "num_tokens": 197612745.0, + "step": 5177 + }, + { + "epoch": 0.6586948225416613, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7911589741706848, + "num_tokens": 197653756.0, + "step": 5178 + }, + { + "epoch": 0.6588220328202519, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8223811388015747, + "num_tokens": 197694417.0, + "step": 5179 + }, + { + "epoch": 0.6589492430988424, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7900546789169312, + "num_tokens": 197732738.0, + "step": 5180 + }, + { + "epoch": 0.6590764533774329, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7839654088020325, + "num_tokens": 197759408.0, + "step": 5181 + }, + { + "epoch": 0.6592036636560235, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7936682105064392, + "num_tokens": 197794334.0, + "step": 5182 + }, + { + "epoch": 0.6593308739346139, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.804172158241272, + "num_tokens": 197832497.0, + "step": 5183 + }, + { + "epoch": 0.6594580842132044, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7899107933044434, + "num_tokens": 197869021.0, + "step": 5184 + }, + { + "epoch": 0.6595852944917949, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7755887508392334, + "num_tokens": 197906365.0, + "step": 5185 + }, + { + "epoch": 0.6597125047703855, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.8006371259689331, + "num_tokens": 197950725.0, + "step": 5186 + }, + { + "epoch": 0.659839715048976, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8066226243972778, + "num_tokens": 197991178.0, + "step": 5187 + }, + { + "epoch": 0.6599669253275665, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7839217185974121, + "num_tokens": 198030176.0, + "step": 5188 + }, + { + "epoch": 0.6600941356061569, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7864408493041992, + "num_tokens": 198067169.0, + "step": 5189 + }, + { + "epoch": 0.6602213458847475, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7851426005363464, + "num_tokens": 198104994.0, + "step": 5190 + }, + { + "epoch": 0.660348556163338, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8051406145095825, + "num_tokens": 198142791.0, + "step": 5191 + }, + { + "epoch": 0.6604757664419285, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7837355136871338, + "num_tokens": 198179300.0, + "step": 5192 + }, + { + "epoch": 0.660602976720519, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8117692470550537, + "num_tokens": 198220230.0, + "step": 5193 + }, + { + "epoch": 0.6607301869991096, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7829565405845642, + "num_tokens": 198251067.0, + "step": 5194 + }, + { + "epoch": 0.6608573972777, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8023321628570557, + "num_tokens": 198282130.0, + "step": 5195 + }, + { + "epoch": 0.6609846075562905, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8057041764259338, + "num_tokens": 198325609.0, + "step": 5196 + }, + { + "epoch": 0.661111817834881, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7974218130111694, + "num_tokens": 198365116.0, + "step": 5197 + }, + { + "epoch": 0.6612390281134716, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.8010928630828857, + "num_tokens": 198403467.0, + "step": 5198 + }, + { + "epoch": 0.6613662383920621, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8036938905715942, + "num_tokens": 198440676.0, + "step": 5199 + }, + { + "epoch": 0.6614934486706526, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7998190522193909, + "num_tokens": 198479082.0, + "step": 5200 + }, + { + "epoch": 0.661620658949243, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7965015769004822, + "num_tokens": 198521741.0, + "step": 5201 + }, + { + "epoch": 0.6617478692278336, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7927871346473694, + "num_tokens": 198562177.0, + "step": 5202 + }, + { + "epoch": 0.6618750795064241, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.796612560749054, + "num_tokens": 198600726.0, + "step": 5203 + }, + { + "epoch": 0.6620022897850146, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7636128664016724, + "num_tokens": 198638156.0, + "step": 5204 + }, + { + "epoch": 0.6621295000636052, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7757413983345032, + "num_tokens": 198680312.0, + "step": 5205 + }, + { + "epoch": 0.6622567103421957, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8155739307403564, + "num_tokens": 198723886.0, + "step": 5206 + }, + { + "epoch": 0.6623839206207861, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7965157628059387, + "num_tokens": 198765331.0, + "step": 5207 + }, + { + "epoch": 0.6625111308993766, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7980138659477234, + "num_tokens": 198802421.0, + "step": 5208 + }, + { + "epoch": 0.6626383411779672, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8070858716964722, + "num_tokens": 198843335.0, + "step": 5209 + }, + { + "epoch": 0.6627655514565577, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8016721606254578, + "num_tokens": 198880925.0, + "step": 5210 + }, + { + "epoch": 0.6628927617351482, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7923460006713867, + "num_tokens": 198915448.0, + "step": 5211 + }, + { + "epoch": 0.6630199720137387, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8044370412826538, + "num_tokens": 198952254.0, + "step": 5212 + }, + { + "epoch": 0.6631471822923293, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8013385534286499, + "num_tokens": 198990841.0, + "step": 5213 + }, + { + "epoch": 0.6632743925709197, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7972467541694641, + "num_tokens": 199029383.0, + "step": 5214 + }, + { + "epoch": 0.6634016028495102, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7925510406494141, + "num_tokens": 199066731.0, + "step": 5215 + }, + { + "epoch": 0.6635288131281007, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8121219873428345, + "num_tokens": 199108752.0, + "step": 5216 + }, + { + "epoch": 0.6636560234066913, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8025882840156555, + "num_tokens": 199146438.0, + "step": 5217 + }, + { + "epoch": 0.6637832336852818, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7980908155441284, + "num_tokens": 199186367.0, + "step": 5218 + }, + { + "epoch": 0.6639104439638723, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8119726181030273, + "num_tokens": 199225785.0, + "step": 5219 + }, + { + "epoch": 0.6640376542424627, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7941089868545532, + "num_tokens": 199262574.0, + "step": 5220 + }, + { + "epoch": 0.6641648645210533, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7986083030700684, + "num_tokens": 199299093.0, + "step": 5221 + }, + { + "epoch": 0.6642920747996438, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7993261218070984, + "num_tokens": 199337446.0, + "step": 5222 + }, + { + "epoch": 0.6644192850782343, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7989187836647034, + "num_tokens": 199377321.0, + "step": 5223 + }, + { + "epoch": 0.6645464953568249, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8102052807807922, + "num_tokens": 199413057.0, + "step": 5224 + }, + { + "epoch": 0.6646737056354154, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.8019296526908875, + "num_tokens": 199452730.0, + "step": 5225 + }, + { + "epoch": 0.6648009159140058, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8014664053916931, + "num_tokens": 199487860.0, + "step": 5226 + }, + { + "epoch": 0.6649281261925963, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8061118125915527, + "num_tokens": 199532790.0, + "step": 5227 + }, + { + "epoch": 0.6650553364711869, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8155620098114014, + "num_tokens": 199561072.0, + "step": 5228 + }, + { + "epoch": 0.6651825467497774, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8057477474212646, + "num_tokens": 199595820.0, + "step": 5229 + }, + { + "epoch": 0.6653097570283679, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7828530073165894, + "num_tokens": 199640101.0, + "step": 5230 + }, + { + "epoch": 0.6654369673069584, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7965447902679443, + "num_tokens": 199680287.0, + "step": 5231 + }, + { + "epoch": 0.6655641775855489, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.8008037209510803, + "num_tokens": 199721958.0, + "step": 5232 + }, + { + "epoch": 0.6656913878641394, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7994629144668579, + "num_tokens": 199756021.0, + "step": 5233 + }, + { + "epoch": 0.6658185981427299, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8036342263221741, + "num_tokens": 199793586.0, + "step": 5234 + }, + { + "epoch": 0.6659458084213205, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7865544557571411, + "num_tokens": 199835728.0, + "step": 5235 + }, + { + "epoch": 0.666073018699911, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8055741786956787, + "num_tokens": 199873348.0, + "step": 5236 + }, + { + "epoch": 0.6662002289785015, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7983099818229675, + "num_tokens": 199908223.0, + "step": 5237 + }, + { + "epoch": 0.6663274392570919, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.794933557510376, + "num_tokens": 199947017.0, + "step": 5238 + }, + { + "epoch": 0.6664546495356825, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8061969876289368, + "num_tokens": 199984806.0, + "step": 5239 + }, + { + "epoch": 0.666581859814273, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7889885902404785, + "num_tokens": 200022600.0, + "step": 5240 + }, + { + "epoch": 0.6667090700928635, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8056192994117737, + "num_tokens": 200058407.0, + "step": 5241 + }, + { + "epoch": 0.666836280371454, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7665790319442749, + "num_tokens": 200101871.0, + "step": 5242 + }, + { + "epoch": 0.6669634906500446, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.797936737537384, + "num_tokens": 200138501.0, + "step": 5243 + }, + { + "epoch": 0.667090700928635, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7959856986999512, + "num_tokens": 200176639.0, + "step": 5244 + }, + { + "epoch": 0.6672179112072255, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.790481448173523, + "num_tokens": 200211873.0, + "step": 5245 + }, + { + "epoch": 0.667345121485816, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7901921272277832, + "num_tokens": 200246596.0, + "step": 5246 + }, + { + "epoch": 0.6674723317644066, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8143476247787476, + "num_tokens": 200282761.0, + "step": 5247 + }, + { + "epoch": 0.6675995420429971, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7783738374710083, + "num_tokens": 200321821.0, + "step": 5248 + }, + { + "epoch": 0.6677267523215876, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7813434600830078, + "num_tokens": 200369066.0, + "step": 5249 + }, + { + "epoch": 0.667853962600178, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.789707601070404, + "num_tokens": 200408067.0, + "step": 5250 + }, + { + "epoch": 0.6679811728787686, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8163470029830933, + "num_tokens": 200446693.0, + "step": 5251 + }, + { + "epoch": 0.6681083831573591, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7761662602424622, + "num_tokens": 200485340.0, + "step": 5252 + }, + { + "epoch": 0.6682355934359496, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7930513620376587, + "num_tokens": 200521978.0, + "step": 5253 + }, + { + "epoch": 0.6683628037145402, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7969048619270325, + "num_tokens": 200560604.0, + "step": 5254 + }, + { + "epoch": 0.6684900139931307, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7924309372901917, + "num_tokens": 200601718.0, + "step": 5255 + }, + { + "epoch": 0.6686172242717211, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7765030860900879, + "num_tokens": 200646186.0, + "step": 5256 + }, + { + "epoch": 0.6687444345503116, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8049073219299316, + "num_tokens": 200684849.0, + "step": 5257 + }, + { + "epoch": 0.6688716448289022, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7986085414886475, + "num_tokens": 200719542.0, + "step": 5258 + }, + { + "epoch": 0.6689988551074927, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7985302209854126, + "num_tokens": 200756212.0, + "step": 5259 + }, + { + "epoch": 0.6691260653860832, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7996034622192383, + "num_tokens": 200791184.0, + "step": 5260 + }, + { + "epoch": 0.6692532756646737, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7717239260673523, + "num_tokens": 200823966.0, + "step": 5261 + }, + { + "epoch": 0.6693804859432643, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8076232671737671, + "num_tokens": 200862718.0, + "step": 5262 + }, + { + "epoch": 0.6695076962218547, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7915642261505127, + "num_tokens": 200901755.0, + "step": 5263 + }, + { + "epoch": 0.6696349065004452, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.8087356090545654, + "num_tokens": 200935788.0, + "step": 5264 + }, + { + "epoch": 0.6697621167790357, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8150080442428589, + "num_tokens": 200975159.0, + "step": 5265 + }, + { + "epoch": 0.6698893270576263, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8118585348129272, + "num_tokens": 201010813.0, + "step": 5266 + }, + { + "epoch": 0.6700165373362168, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7991445064544678, + "num_tokens": 201051937.0, + "step": 5267 + }, + { + "epoch": 0.6701437476148073, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8002271056175232, + "num_tokens": 201090926.0, + "step": 5268 + }, + { + "epoch": 0.6702709578933977, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7724277973175049, + "num_tokens": 201126086.0, + "step": 5269 + }, + { + "epoch": 0.6703981681719883, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7848634719848633, + "num_tokens": 201165933.0, + "step": 5270 + }, + { + "epoch": 0.6705253784505788, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7978173494338989, + "num_tokens": 201198794.0, + "step": 5271 + }, + { + "epoch": 0.6706525887291693, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7888131141662598, + "num_tokens": 201234081.0, + "step": 5272 + }, + { + "epoch": 0.6707797990077599, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.791861891746521, + "num_tokens": 201271796.0, + "step": 5273 + }, + { + "epoch": 0.6709070092863504, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7803622484207153, + "num_tokens": 201309526.0, + "step": 5274 + }, + { + "epoch": 0.6710342195649408, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8006752729415894, + "num_tokens": 201344015.0, + "step": 5275 + }, + { + "epoch": 0.6711614298435313, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7961009740829468, + "num_tokens": 201385201.0, + "step": 5276 + }, + { + "epoch": 0.6712886401221219, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.8006937503814697, + "num_tokens": 201422370.0, + "step": 5277 + }, + { + "epoch": 0.6714158504007124, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7844306230545044, + "num_tokens": 201467085.0, + "step": 5278 + }, + { + "epoch": 0.6715430606793029, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8106332421302795, + "num_tokens": 201503432.0, + "step": 5279 + }, + { + "epoch": 0.6716702709578934, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8076746463775635, + "num_tokens": 201541358.0, + "step": 5280 + }, + { + "epoch": 0.6717974812364839, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7914787530899048, + "num_tokens": 201580235.0, + "step": 5281 + }, + { + "epoch": 0.6719246915150744, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7985250353813171, + "num_tokens": 201627703.0, + "step": 5282 + }, + { + "epoch": 0.6720519017936649, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7980149984359741, + "num_tokens": 201667182.0, + "step": 5283 + }, + { + "epoch": 0.6721791120722554, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.8015624284744263, + "num_tokens": 201699876.0, + "step": 5284 + }, + { + "epoch": 0.672306322350846, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7837883234024048, + "num_tokens": 201732770.0, + "step": 5285 + }, + { + "epoch": 0.6724335326294365, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8044969439506531, + "num_tokens": 201771388.0, + "step": 5286 + }, + { + "epoch": 0.6725607429080269, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.786949872970581, + "num_tokens": 201809283.0, + "step": 5287 + }, + { + "epoch": 0.6726879531866174, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.8000837564468384, + "num_tokens": 201854950.0, + "step": 5288 + }, + { + "epoch": 0.672815163465208, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.8004859685897827, + "num_tokens": 201895638.0, + "step": 5289 + }, + { + "epoch": 0.6729423737437985, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7795077562332153, + "num_tokens": 201928243.0, + "step": 5290 + }, + { + "epoch": 0.673069584022389, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7916867733001709, + "num_tokens": 201964687.0, + "step": 5291 + }, + { + "epoch": 0.6731967943009796, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8101202249526978, + "num_tokens": 201997166.0, + "step": 5292 + }, + { + "epoch": 0.67332400457957, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7978521585464478, + "num_tokens": 202030771.0, + "step": 5293 + }, + { + "epoch": 0.6734512148581605, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7902088165283203, + "num_tokens": 202077280.0, + "step": 5294 + }, + { + "epoch": 0.673578425136751, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.788610577583313, + "num_tokens": 202115273.0, + "step": 5295 + }, + { + "epoch": 0.6737056354153416, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7989716529846191, + "num_tokens": 202145347.0, + "step": 5296 + }, + { + "epoch": 0.6738328456939321, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8022557497024536, + "num_tokens": 202184621.0, + "step": 5297 + }, + { + "epoch": 0.6739600559725226, + "grad_norm": 92.5, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.8025262951850891, + "num_tokens": 202230099.0, + "step": 5298 + }, + { + "epoch": 0.674087266251113, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7936725616455078, + "num_tokens": 202267872.0, + "step": 5299 + }, + { + "epoch": 0.6742144765297036, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7961479425430298, + "num_tokens": 202305517.0, + "step": 5300 + }, + { + "epoch": 0.6743416868082941, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7765274047851562, + "num_tokens": 202340671.0, + "step": 5301 + }, + { + "epoch": 0.6744688970868846, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8064411878585815, + "num_tokens": 202379552.0, + "step": 5302 + }, + { + "epoch": 0.6745961073654752, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7930058240890503, + "num_tokens": 202421404.0, + "step": 5303 + }, + { + "epoch": 0.6747233176440657, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7814356684684753, + "num_tokens": 202463765.0, + "step": 5304 + }, + { + "epoch": 0.6748505279226561, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7899534702301025, + "num_tokens": 202505872.0, + "step": 5305 + }, + { + "epoch": 0.6749777382012466, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7956875562667847, + "num_tokens": 202544085.0, + "step": 5306 + }, + { + "epoch": 0.6751049484798372, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7961386442184448, + "num_tokens": 202588237.0, + "step": 5307 + }, + { + "epoch": 0.6752321587584277, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7918387651443481, + "num_tokens": 202628462.0, + "step": 5308 + }, + { + "epoch": 0.6753593690370182, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7991886138916016, + "num_tokens": 202668149.0, + "step": 5309 + }, + { + "epoch": 0.6754865793156087, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8141864538192749, + "num_tokens": 202708651.0, + "step": 5310 + }, + { + "epoch": 0.6756137895941993, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7972143888473511, + "num_tokens": 202738613.0, + "step": 5311 + }, + { + "epoch": 0.6757409998727897, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7955257892608643, + "num_tokens": 202779279.0, + "step": 5312 + }, + { + "epoch": 0.6758682101513802, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7949801683425903, + "num_tokens": 202812095.0, + "step": 5313 + }, + { + "epoch": 0.6759954204299707, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7933357954025269, + "num_tokens": 202855705.0, + "step": 5314 + }, + { + "epoch": 0.6761226307085613, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7783041000366211, + "num_tokens": 202894180.0, + "step": 5315 + }, + { + "epoch": 0.6762498409871518, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8027347326278687, + "num_tokens": 202925842.0, + "step": 5316 + }, + { + "epoch": 0.6763770512657423, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8068559169769287, + "num_tokens": 202959103.0, + "step": 5317 + }, + { + "epoch": 0.6765042615443327, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7970798015594482, + "num_tokens": 203000393.0, + "step": 5318 + }, + { + "epoch": 0.6766314718229233, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7978808879852295, + "num_tokens": 203041613.0, + "step": 5319 + }, + { + "epoch": 0.6767586821015138, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8104304075241089, + "num_tokens": 203073744.0, + "step": 5320 + }, + { + "epoch": 0.6768858923801043, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.776807427406311, + "num_tokens": 203108350.0, + "step": 5321 + }, + { + "epoch": 0.6770131026586949, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7935174107551575, + "num_tokens": 203139149.0, + "step": 5322 + }, + { + "epoch": 0.6771403129372854, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.774904191493988, + "num_tokens": 203173953.0, + "step": 5323 + }, + { + "epoch": 0.6772675232158758, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8011900186538696, + "num_tokens": 203208915.0, + "step": 5324 + }, + { + "epoch": 0.6773947334944663, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7853991985321045, + "num_tokens": 203251906.0, + "step": 5325 + }, + { + "epoch": 0.6775219437730569, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7998985052108765, + "num_tokens": 203289460.0, + "step": 5326 + }, + { + "epoch": 0.6776491540516474, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.8078218102455139, + "num_tokens": 203325647.0, + "step": 5327 + }, + { + "epoch": 0.6777763643302379, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7959831357002258, + "num_tokens": 203364972.0, + "step": 5328 + }, + { + "epoch": 0.6779035746088284, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8004598617553711, + "num_tokens": 203407778.0, + "step": 5329 + }, + { + "epoch": 0.6780307848874189, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7973837852478027, + "num_tokens": 203443820.0, + "step": 5330 + }, + { + "epoch": 0.6781579951660094, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.793806791305542, + "num_tokens": 203475742.0, + "step": 5331 + }, + { + "epoch": 0.6782852054445999, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7731869220733643, + "num_tokens": 203512617.0, + "step": 5332 + }, + { + "epoch": 0.6784124157231904, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7778880596160889, + "num_tokens": 203546778.0, + "step": 5333 + }, + { + "epoch": 0.678539626001781, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7981062531471252, + "num_tokens": 203585222.0, + "step": 5334 + }, + { + "epoch": 0.6786668362803715, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7884790897369385, + "num_tokens": 203627014.0, + "step": 5335 + }, + { + "epoch": 0.6787940465589619, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7974529266357422, + "num_tokens": 203659189.0, + "step": 5336 + }, + { + "epoch": 0.6789212568375524, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8025325536727905, + "num_tokens": 203695355.0, + "step": 5337 + }, + { + "epoch": 0.679048467116143, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8112924098968506, + "num_tokens": 203733446.0, + "step": 5338 + }, + { + "epoch": 0.6791756773947335, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7788254618644714, + "num_tokens": 203775387.0, + "step": 5339 + }, + { + "epoch": 0.679302887673324, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.790471076965332, + "num_tokens": 203810713.0, + "step": 5340 + }, + { + "epoch": 0.6794300979519146, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8098629713058472, + "num_tokens": 203847144.0, + "step": 5341 + }, + { + "epoch": 0.679557308230505, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8035147190093994, + "num_tokens": 203889716.0, + "step": 5342 + }, + { + "epoch": 0.6796845185090955, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7847999334335327, + "num_tokens": 203927692.0, + "step": 5343 + }, + { + "epoch": 0.679811728787686, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7775949835777283, + "num_tokens": 203960180.0, + "step": 5344 + }, + { + "epoch": 0.6799389390662766, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8112062215805054, + "num_tokens": 203997375.0, + "step": 5345 + }, + { + "epoch": 0.6800661493448671, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7919328808784485, + "num_tokens": 204037588.0, + "step": 5346 + }, + { + "epoch": 0.6801933596234576, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7918045520782471, + "num_tokens": 204080993.0, + "step": 5347 + }, + { + "epoch": 0.680320569902048, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7993974685668945, + "num_tokens": 204116863.0, + "step": 5348 + }, + { + "epoch": 0.6804477801806386, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.790860116481781, + "num_tokens": 204156551.0, + "step": 5349 + }, + { + "epoch": 0.6805749904592291, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7905638217926025, + "num_tokens": 204196210.0, + "step": 5350 + }, + { + "epoch": 0.6807022007378196, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7920277118682861, + "num_tokens": 204234791.0, + "step": 5351 + }, + { + "epoch": 0.6808294110164101, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7979733943939209, + "num_tokens": 204271631.0, + "step": 5352 + }, + { + "epoch": 0.6809566212950007, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.810047447681427, + "num_tokens": 204311879.0, + "step": 5353 + }, + { + "epoch": 0.6810838315735911, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.8006371855735779, + "num_tokens": 204357123.0, + "step": 5354 + }, + { + "epoch": 0.6812110418521816, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7957121133804321, + "num_tokens": 204397196.0, + "step": 5355 + }, + { + "epoch": 0.6813382521307721, + "grad_norm": 104.5, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7835569977760315, + "num_tokens": 204434654.0, + "step": 5356 + }, + { + "epoch": 0.6814654624093627, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.8152838349342346, + "num_tokens": 204466581.0, + "step": 5357 + }, + { + "epoch": 0.6815926726879532, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8079895377159119, + "num_tokens": 204503890.0, + "step": 5358 + }, + { + "epoch": 0.6817198829665437, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7916418313980103, + "num_tokens": 204537160.0, + "step": 5359 + }, + { + "epoch": 0.6818470932451343, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8045251369476318, + "num_tokens": 204570077.0, + "step": 5360 + }, + { + "epoch": 0.6819743035237247, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8127403259277344, + "num_tokens": 204603638.0, + "step": 5361 + }, + { + "epoch": 0.6821015138023152, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7954568862915039, + "num_tokens": 204644584.0, + "step": 5362 + }, + { + "epoch": 0.6822287240809057, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.778272271156311, + "num_tokens": 204678459.0, + "step": 5363 + }, + { + "epoch": 0.6823559343594963, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7859752774238586, + "num_tokens": 204712923.0, + "step": 5364 + }, + { + "epoch": 0.6824831446380868, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7751146554946899, + "num_tokens": 204751303.0, + "step": 5365 + }, + { + "epoch": 0.6826103549166773, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8015505075454712, + "num_tokens": 204791685.0, + "step": 5366 + }, + { + "epoch": 0.6827375651952677, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7925087213516235, + "num_tokens": 204827748.0, + "step": 5367 + }, + { + "epoch": 0.6828647754738583, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.8168967962265015, + "num_tokens": 204866568.0, + "step": 5368 + }, + { + "epoch": 0.6829919857524488, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7938246130943298, + "num_tokens": 204908334.0, + "step": 5369 + }, + { + "epoch": 0.6831191960310393, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.802351176738739, + "num_tokens": 204949929.0, + "step": 5370 + }, + { + "epoch": 0.6832464063096299, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7848261594772339, + "num_tokens": 204991435.0, + "step": 5371 + }, + { + "epoch": 0.6833736165882204, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8051080107688904, + "num_tokens": 205028434.0, + "step": 5372 + }, + { + "epoch": 0.6835008268668108, + "grad_norm": 140.0, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7876906394958496, + "num_tokens": 205064138.0, + "step": 5373 + }, + { + "epoch": 0.6836280371454013, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7876418828964233, + "num_tokens": 205104058.0, + "step": 5374 + }, + { + "epoch": 0.6837552474239919, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7942339181900024, + "num_tokens": 205142609.0, + "step": 5375 + }, + { + "epoch": 0.6838824577025824, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8053734302520752, + "num_tokens": 205180407.0, + "step": 5376 + }, + { + "epoch": 0.6840096679811729, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8219935894012451, + "num_tokens": 205217340.0, + "step": 5377 + }, + { + "epoch": 0.6841368782597634, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7856260538101196, + "num_tokens": 205263671.0, + "step": 5378 + }, + { + "epoch": 0.6842640885383539, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7976065874099731, + "num_tokens": 205302970.0, + "step": 5379 + }, + { + "epoch": 0.6843912988169444, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.801681399345398, + "num_tokens": 205342247.0, + "step": 5380 + }, + { + "epoch": 0.6845185090955349, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7775756120681763, + "num_tokens": 205374897.0, + "step": 5381 + }, + { + "epoch": 0.6846457193741254, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7907663583755493, + "num_tokens": 205411645.0, + "step": 5382 + }, + { + "epoch": 0.684772929652716, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7940726280212402, + "num_tokens": 205450972.0, + "step": 5383 + }, + { + "epoch": 0.6849001399313065, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.8001033663749695, + "num_tokens": 205494314.0, + "step": 5384 + }, + { + "epoch": 0.6850273502098969, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7856855988502502, + "num_tokens": 205534696.0, + "step": 5385 + }, + { + "epoch": 0.6851545604884874, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.80790776014328, + "num_tokens": 205575579.0, + "step": 5386 + }, + { + "epoch": 0.685281770767078, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7951943874359131, + "num_tokens": 205609249.0, + "step": 5387 + }, + { + "epoch": 0.6854089810456685, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7914308309555054, + "num_tokens": 205644135.0, + "step": 5388 + }, + { + "epoch": 0.685536191324259, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7853140830993652, + "num_tokens": 205685312.0, + "step": 5389 + }, + { + "epoch": 0.6856634016028496, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7876200079917908, + "num_tokens": 205720525.0, + "step": 5390 + }, + { + "epoch": 0.68579061188144, + "grad_norm": 88.0, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7742447853088379, + "num_tokens": 205763358.0, + "step": 5391 + }, + { + "epoch": 0.6859178221600305, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7967197299003601, + "num_tokens": 205801413.0, + "step": 5392 + }, + { + "epoch": 0.686045032438621, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7979359030723572, + "num_tokens": 205840442.0, + "step": 5393 + }, + { + "epoch": 0.6861722427172116, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8089644312858582, + "num_tokens": 205877319.0, + "step": 5394 + }, + { + "epoch": 0.6862994529958021, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7936838865280151, + "num_tokens": 205916186.0, + "step": 5395 + }, + { + "epoch": 0.6864266632743926, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7826369404792786, + "num_tokens": 205956336.0, + "step": 5396 + }, + { + "epoch": 0.686553873552983, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7945750951766968, + "num_tokens": 205989570.0, + "step": 5397 + }, + { + "epoch": 0.6866810838315736, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7854301929473877, + "num_tokens": 206026139.0, + "step": 5398 + }, + { + "epoch": 0.6868082941101641, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7910593152046204, + "num_tokens": 206070059.0, + "step": 5399 + }, + { + "epoch": 0.6869355043887546, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7971755266189575, + "num_tokens": 206104645.0, + "step": 5400 + }, + { + "epoch": 0.6870627146673451, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7912836074829102, + "num_tokens": 206141220.0, + "step": 5401 + }, + { + "epoch": 0.6871899249459357, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7948786020278931, + "num_tokens": 206177917.0, + "step": 5402 + }, + { + "epoch": 0.6873171352245261, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7930988073348999, + "num_tokens": 206213419.0, + "step": 5403 + }, + { + "epoch": 0.6874443455031166, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.780127763748169, + "num_tokens": 206252421.0, + "step": 5404 + }, + { + "epoch": 0.6875715557817071, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8053728342056274, + "num_tokens": 206287195.0, + "step": 5405 + }, + { + "epoch": 0.6876987660602977, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7790518403053284, + "num_tokens": 206328339.0, + "step": 5406 + }, + { + "epoch": 0.6878259763388882, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7816517949104309, + "num_tokens": 206369523.0, + "step": 5407 + }, + { + "epoch": 0.6879531866174787, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6008, + "mean_token_accuracy": 0.8190790414810181, + "num_tokens": 206402864.0, + "step": 5408 + }, + { + "epoch": 0.6880803968960693, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.779386579990387, + "num_tokens": 206439004.0, + "step": 5409 + }, + { + "epoch": 0.6882076071746597, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7940181493759155, + "num_tokens": 206472320.0, + "step": 5410 + }, + { + "epoch": 0.6883348174532502, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7877481579780579, + "num_tokens": 206518649.0, + "step": 5411 + }, + { + "epoch": 0.6884620277318407, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8142199516296387, + "num_tokens": 206551318.0, + "step": 5412 + }, + { + "epoch": 0.6885892380104313, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7801788449287415, + "num_tokens": 206588173.0, + "step": 5413 + }, + { + "epoch": 0.6887164482890218, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7938026189804077, + "num_tokens": 206624013.0, + "step": 5414 + }, + { + "epoch": 0.6888436585676123, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8014163970947266, + "num_tokens": 206668180.0, + "step": 5415 + }, + { + "epoch": 0.6889708688462027, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7895866632461548, + "num_tokens": 206710619.0, + "step": 5416 + }, + { + "epoch": 0.6890980791247933, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8047722578048706, + "num_tokens": 206750121.0, + "step": 5417 + }, + { + "epoch": 0.6892252894033838, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8082634806632996, + "num_tokens": 206789538.0, + "step": 5418 + }, + { + "epoch": 0.6893524996819743, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7935869097709656, + "num_tokens": 206827513.0, + "step": 5419 + }, + { + "epoch": 0.6894797099605648, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8078917264938354, + "num_tokens": 206866563.0, + "step": 5420 + }, + { + "epoch": 0.6896069202391554, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8110673427581787, + "num_tokens": 206906281.0, + "step": 5421 + }, + { + "epoch": 0.6897341305177458, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7807241678237915, + "num_tokens": 206944055.0, + "step": 5422 + }, + { + "epoch": 0.6898613407963363, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7934361696243286, + "num_tokens": 206984421.0, + "step": 5423 + }, + { + "epoch": 0.6899885510749268, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7831177115440369, + "num_tokens": 207020319.0, + "step": 5424 + }, + { + "epoch": 0.6901157613535174, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8031613230705261, + "num_tokens": 207061487.0, + "step": 5425 + }, + { + "epoch": 0.6902429716321079, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8015578985214233, + "num_tokens": 207098343.0, + "step": 5426 + }, + { + "epoch": 0.6903701819106984, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7964891195297241, + "num_tokens": 207140381.0, + "step": 5427 + }, + { + "epoch": 0.6904973921892888, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8074809908866882, + "num_tokens": 207173923.0, + "step": 5428 + }, + { + "epoch": 0.6906246024678794, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8034787178039551, + "num_tokens": 207214840.0, + "step": 5429 + }, + { + "epoch": 0.6907518127464699, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7911874055862427, + "num_tokens": 207247850.0, + "step": 5430 + }, + { + "epoch": 0.6908790230250604, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7923340201377869, + "num_tokens": 207286698.0, + "step": 5431 + }, + { + "epoch": 0.691006233303651, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.792558491230011, + "num_tokens": 207322470.0, + "step": 5432 + }, + { + "epoch": 0.6911334435822415, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7858009934425354, + "num_tokens": 207361517.0, + "step": 5433 + }, + { + "epoch": 0.6912606538608319, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8051769137382507, + "num_tokens": 207394504.0, + "step": 5434 + }, + { + "epoch": 0.6913878641394224, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7752096056938171, + "num_tokens": 207430148.0, + "step": 5435 + }, + { + "epoch": 0.691515074418013, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7849910855293274, + "num_tokens": 207461094.0, + "step": 5436 + }, + { + "epoch": 0.6916422846966035, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8037940859794617, + "num_tokens": 207500572.0, + "step": 5437 + }, + { + "epoch": 0.691769494975194, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7869904041290283, + "num_tokens": 207534977.0, + "step": 5438 + }, + { + "epoch": 0.6918967052537845, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8029779195785522, + "num_tokens": 207569809.0, + "step": 5439 + }, + { + "epoch": 0.692023915532375, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7985339760780334, + "num_tokens": 207606785.0, + "step": 5440 + }, + { + "epoch": 0.6921511258109655, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7929832339286804, + "num_tokens": 207648344.0, + "step": 5441 + }, + { + "epoch": 0.692278336089556, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8007868528366089, + "num_tokens": 207680962.0, + "step": 5442 + }, + { + "epoch": 0.6924055463681466, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.5803, + "mean_token_accuracy": 0.8223980069160461, + "num_tokens": 207709162.0, + "step": 5443 + }, + { + "epoch": 0.6925327566467371, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.790704607963562, + "num_tokens": 207746089.0, + "step": 5444 + }, + { + "epoch": 0.6926599669253276, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.80816650390625, + "num_tokens": 207791235.0, + "step": 5445 + }, + { + "epoch": 0.692787177203918, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7851712107658386, + "num_tokens": 207824527.0, + "step": 5446 + }, + { + "epoch": 0.6929143874825086, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8094118237495422, + "num_tokens": 207863535.0, + "step": 5447 + }, + { + "epoch": 0.6930415977610991, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7877485752105713, + "num_tokens": 207899694.0, + "step": 5448 + }, + { + "epoch": 0.6931688080396896, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7853348255157471, + "num_tokens": 207931440.0, + "step": 5449 + }, + { + "epoch": 0.6932960183182801, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7940298318862915, + "num_tokens": 207963758.0, + "step": 5450 + }, + { + "epoch": 0.6934232285968707, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7983189821243286, + "num_tokens": 208002383.0, + "step": 5451 + }, + { + "epoch": 0.6935504388754611, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8013407588005066, + "num_tokens": 208038723.0, + "step": 5452 + }, + { + "epoch": 0.6936776491540516, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7940530776977539, + "num_tokens": 208071851.0, + "step": 5453 + }, + { + "epoch": 0.6938048594326421, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7775484323501587, + "num_tokens": 208106916.0, + "step": 5454 + }, + { + "epoch": 0.6939320697112327, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8024592995643616, + "num_tokens": 208146883.0, + "step": 5455 + }, + { + "epoch": 0.6940592799898232, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8020055890083313, + "num_tokens": 208187364.0, + "step": 5456 + }, + { + "epoch": 0.6941864902684137, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8019605875015259, + "num_tokens": 208226087.0, + "step": 5457 + }, + { + "epoch": 0.6943137005470043, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7867030501365662, + "num_tokens": 208265193.0, + "step": 5458 + }, + { + "epoch": 0.6944409108255947, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8128564953804016, + "num_tokens": 208305163.0, + "step": 5459 + }, + { + "epoch": 0.6945681211041852, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.777343213558197, + "num_tokens": 208342229.0, + "step": 5460 + }, + { + "epoch": 0.6946953313827757, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7832098007202148, + "num_tokens": 208384606.0, + "step": 5461 + }, + { + "epoch": 0.6948225416613663, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8122062087059021, + "num_tokens": 208416777.0, + "step": 5462 + }, + { + "epoch": 0.6949497519399568, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7788596153259277, + "num_tokens": 208456101.0, + "step": 5463 + }, + { + "epoch": 0.6950769622185473, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8012382984161377, + "num_tokens": 208489201.0, + "step": 5464 + }, + { + "epoch": 0.6952041724971377, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8084180951118469, + "num_tokens": 208525006.0, + "step": 5465 + }, + { + "epoch": 0.6953313827757283, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7937357425689697, + "num_tokens": 208568997.0, + "step": 5466 + }, + { + "epoch": 0.6954585930543188, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7848652601242065, + "num_tokens": 208605925.0, + "step": 5467 + }, + { + "epoch": 0.6955858033329093, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7738577723503113, + "num_tokens": 208646290.0, + "step": 5468 + }, + { + "epoch": 0.6957130136114998, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7917119264602661, + "num_tokens": 208686424.0, + "step": 5469 + }, + { + "epoch": 0.6958402238900904, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7942071557044983, + "num_tokens": 208719643.0, + "step": 5470 + }, + { + "epoch": 0.6959674341686808, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7812156081199646, + "num_tokens": 208760804.0, + "step": 5471 + }, + { + "epoch": 0.6960946444472713, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7917104959487915, + "num_tokens": 208803022.0, + "step": 5472 + }, + { + "epoch": 0.6962218547258618, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7981610298156738, + "num_tokens": 208844205.0, + "step": 5473 + }, + { + "epoch": 0.6963490650044524, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8109921216964722, + "num_tokens": 208886882.0, + "step": 5474 + }, + { + "epoch": 0.6964762752830429, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7962628602981567, + "num_tokens": 208926649.0, + "step": 5475 + }, + { + "epoch": 0.6966034855616334, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7855636477470398, + "num_tokens": 208963403.0, + "step": 5476 + }, + { + "epoch": 0.6967306958402238, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7891198992729187, + "num_tokens": 209002475.0, + "step": 5477 + }, + { + "epoch": 0.6968579061188144, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7965493202209473, + "num_tokens": 209044216.0, + "step": 5478 + }, + { + "epoch": 0.6969851163974049, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7837875485420227, + "num_tokens": 209080774.0, + "step": 5479 + }, + { + "epoch": 0.6971123266759954, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7982338070869446, + "num_tokens": 209117167.0, + "step": 5480 + }, + { + "epoch": 0.697239536954586, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7863587141036987, + "num_tokens": 209148872.0, + "step": 5481 + }, + { + "epoch": 0.6973667472331765, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8001021146774292, + "num_tokens": 209187912.0, + "step": 5482 + }, + { + "epoch": 0.6974939575117669, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8046116232872009, + "num_tokens": 209227055.0, + "step": 5483 + }, + { + "epoch": 0.6976211677903574, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8014750480651855, + "num_tokens": 209255343.0, + "step": 5484 + }, + { + "epoch": 0.697748378068948, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7910811901092529, + "num_tokens": 209294621.0, + "step": 5485 + }, + { + "epoch": 0.6978755883475385, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8138813376426697, + "num_tokens": 209336519.0, + "step": 5486 + }, + { + "epoch": 0.698002798626129, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.795440673828125, + "num_tokens": 209372913.0, + "step": 5487 + }, + { + "epoch": 0.6981300089047195, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7903937697410583, + "num_tokens": 209417768.0, + "step": 5488 + }, + { + "epoch": 0.69825721918331, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7961738705635071, + "num_tokens": 209451221.0, + "step": 5489 + }, + { + "epoch": 0.6983844294619005, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7963650226593018, + "num_tokens": 209488710.0, + "step": 5490 + }, + { + "epoch": 0.698511639740491, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7885235548019409, + "num_tokens": 209525166.0, + "step": 5491 + }, + { + "epoch": 0.6986388500190815, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.802377462387085, + "num_tokens": 209564804.0, + "step": 5492 + }, + { + "epoch": 0.6987660602976721, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8016639947891235, + "num_tokens": 209607155.0, + "step": 5493 + }, + { + "epoch": 0.6988932705762626, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.782930850982666, + "num_tokens": 209641226.0, + "step": 5494 + }, + { + "epoch": 0.699020480854853, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7864334583282471, + "num_tokens": 209679180.0, + "step": 5495 + }, + { + "epoch": 0.6991476911334435, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7913012504577637, + "num_tokens": 209716412.0, + "step": 5496 + }, + { + "epoch": 0.6992749014120341, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7881243228912354, + "num_tokens": 209755697.0, + "step": 5497 + }, + { + "epoch": 0.6994021116906246, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8001619577407837, + "num_tokens": 209793654.0, + "step": 5498 + }, + { + "epoch": 0.6995293219692151, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7933441996574402, + "num_tokens": 209830703.0, + "step": 5499 + }, + { + "epoch": 0.6996565322478057, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7972978949546814, + "num_tokens": 209868994.0, + "step": 5500 + }, + { + "epoch": 0.6997837425263961, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8075937032699585, + "num_tokens": 209909605.0, + "step": 5501 + }, + { + "epoch": 0.6999109528049866, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7974995374679565, + "num_tokens": 209945196.0, + "step": 5502 + }, + { + "epoch": 0.7000381630835771, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7976659536361694, + "num_tokens": 209988585.0, + "step": 5503 + }, + { + "epoch": 0.7001653733621677, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7993808388710022, + "num_tokens": 210032084.0, + "step": 5504 + }, + { + "epoch": 0.7002925836407582, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7864084243774414, + "num_tokens": 210070639.0, + "step": 5505 + }, + { + "epoch": 0.7004197939193487, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.808969259262085, + "num_tokens": 210113793.0, + "step": 5506 + }, + { + "epoch": 0.7005470041979391, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7895665168762207, + "num_tokens": 210153517.0, + "step": 5507 + }, + { + "epoch": 0.7006742144765297, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7845199108123779, + "num_tokens": 210186818.0, + "step": 5508 + }, + { + "epoch": 0.7008014247551202, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7872837781906128, + "num_tokens": 210224283.0, + "step": 5509 + }, + { + "epoch": 0.7009286350337107, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7910130620002747, + "num_tokens": 210263439.0, + "step": 5510 + }, + { + "epoch": 0.7010558453123013, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8025619983673096, + "num_tokens": 210305887.0, + "step": 5511 + }, + { + "epoch": 0.7011830555908918, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7880696058273315, + "num_tokens": 210340118.0, + "step": 5512 + }, + { + "epoch": 0.7013102658694823, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7983748912811279, + "num_tokens": 210376098.0, + "step": 5513 + }, + { + "epoch": 0.7014374761480727, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7984501123428345, + "num_tokens": 210418801.0, + "step": 5514 + }, + { + "epoch": 0.7015646864266633, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7835522890090942, + "num_tokens": 210452392.0, + "step": 5515 + }, + { + "epoch": 0.7016918967052538, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.8009561896324158, + "num_tokens": 210489697.0, + "step": 5516 + }, + { + "epoch": 0.7018191069838443, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7764078974723816, + "num_tokens": 210530212.0, + "step": 5517 + }, + { + "epoch": 0.7019463172624348, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8037662506103516, + "num_tokens": 210564974.0, + "step": 5518 + }, + { + "epoch": 0.7020735275410254, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7964274287223816, + "num_tokens": 210601272.0, + "step": 5519 + }, + { + "epoch": 0.7022007378196158, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7994491457939148, + "num_tokens": 210636908.0, + "step": 5520 + }, + { + "epoch": 0.7023279480982063, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7909089922904968, + "num_tokens": 210672629.0, + "step": 5521 + }, + { + "epoch": 0.7024551583767968, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8058421611785889, + "num_tokens": 210703531.0, + "step": 5522 + }, + { + "epoch": 0.7025823686553874, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8120280504226685, + "num_tokens": 210742946.0, + "step": 5523 + }, + { + "epoch": 0.7027095789339779, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8088862895965576, + "num_tokens": 210778345.0, + "step": 5524 + }, + { + "epoch": 0.7028367892125684, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8058800101280212, + "num_tokens": 210821544.0, + "step": 5525 + }, + { + "epoch": 0.7029639994911588, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.8030779361724854, + "num_tokens": 210858060.0, + "step": 5526 + }, + { + "epoch": 0.7030912097697494, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7954400181770325, + "num_tokens": 210900365.0, + "step": 5527 + }, + { + "epoch": 0.7032184200483399, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7907288074493408, + "num_tokens": 210941118.0, + "step": 5528 + }, + { + "epoch": 0.7033456303269304, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8030945658683777, + "num_tokens": 210979865.0, + "step": 5529 + }, + { + "epoch": 0.703472840605521, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7947748899459839, + "num_tokens": 211016385.0, + "step": 5530 + }, + { + "epoch": 0.7036000508841115, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7832581996917725, + "num_tokens": 211059071.0, + "step": 5531 + }, + { + "epoch": 0.7037272611627019, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7959520220756531, + "num_tokens": 211094189.0, + "step": 5532 + }, + { + "epoch": 0.7038544714412924, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8050238490104675, + "num_tokens": 211130524.0, + "step": 5533 + }, + { + "epoch": 0.703981681719883, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7928841710090637, + "num_tokens": 211172335.0, + "step": 5534 + }, + { + "epoch": 0.7041088919984735, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7964679598808289, + "num_tokens": 211205406.0, + "step": 5535 + }, + { + "epoch": 0.704236102277064, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7946162819862366, + "num_tokens": 211246443.0, + "step": 5536 + }, + { + "epoch": 0.7043633125556545, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8147679567337036, + "num_tokens": 211286707.0, + "step": 5537 + }, + { + "epoch": 0.704490522834245, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8034622669219971, + "num_tokens": 211319448.0, + "step": 5538 + }, + { + "epoch": 0.7046177331128355, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7946640849113464, + "num_tokens": 211363747.0, + "step": 5539 + }, + { + "epoch": 0.704744943391426, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8032344579696655, + "num_tokens": 211408080.0, + "step": 5540 + }, + { + "epoch": 0.7048721536700165, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7798382639884949, + "num_tokens": 211445656.0, + "step": 5541 + }, + { + "epoch": 0.7049993639486071, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8029646277427673, + "num_tokens": 211484342.0, + "step": 5542 + }, + { + "epoch": 0.7051265742271976, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7901748418807983, + "num_tokens": 211518535.0, + "step": 5543 + }, + { + "epoch": 0.705253784505788, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7873572707176208, + "num_tokens": 211557236.0, + "step": 5544 + }, + { + "epoch": 0.7053809947843785, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8104501366615295, + "num_tokens": 211599216.0, + "step": 5545 + }, + { + "epoch": 0.7055082050629691, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.776500940322876, + "num_tokens": 211638930.0, + "step": 5546 + }, + { + "epoch": 0.7056354153415596, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7890055775642395, + "num_tokens": 211680765.0, + "step": 5547 + }, + { + "epoch": 0.7057626256201501, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8136075735092163, + "num_tokens": 211719037.0, + "step": 5548 + }, + { + "epoch": 0.7058898358987407, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7945595979690552, + "num_tokens": 211759655.0, + "step": 5549 + }, + { + "epoch": 0.7060170461773311, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7814604640007019, + "num_tokens": 211800405.0, + "step": 5550 + }, + { + "epoch": 0.7061442564559216, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7786647081375122, + "num_tokens": 211846057.0, + "step": 5551 + }, + { + "epoch": 0.7062714667345121, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7908844351768494, + "num_tokens": 211885944.0, + "step": 5552 + }, + { + "epoch": 0.7063986770131027, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7987586855888367, + "num_tokens": 211928727.0, + "step": 5553 + }, + { + "epoch": 0.7065258872916932, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7818363904953003, + "num_tokens": 211967593.0, + "step": 5554 + }, + { + "epoch": 0.7066530975702837, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7895718812942505, + "num_tokens": 212001023.0, + "step": 5555 + }, + { + "epoch": 0.7067803078488741, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7902700901031494, + "num_tokens": 212045611.0, + "step": 5556 + }, + { + "epoch": 0.7069075181274647, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7886418104171753, + "num_tokens": 212087167.0, + "step": 5557 + }, + { + "epoch": 0.7070347284060552, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7945238351821899, + "num_tokens": 212124254.0, + "step": 5558 + }, + { + "epoch": 0.7071619386846457, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.789368748664856, + "num_tokens": 212162706.0, + "step": 5559 + }, + { + "epoch": 0.7072891489632362, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7942771911621094, + "num_tokens": 212197359.0, + "step": 5560 + }, + { + "epoch": 0.7074163592418268, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8061727285385132, + "num_tokens": 212237300.0, + "step": 5561 + }, + { + "epoch": 0.7075435695204173, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7805097103118896, + "num_tokens": 212283472.0, + "step": 5562 + }, + { + "epoch": 0.7076707797990077, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.791240930557251, + "num_tokens": 212321433.0, + "step": 5563 + }, + { + "epoch": 0.7077979900775982, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7948306202888489, + "num_tokens": 212362234.0, + "step": 5564 + }, + { + "epoch": 0.7079252003561888, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7826720476150513, + "num_tokens": 212399259.0, + "step": 5565 + }, + { + "epoch": 0.7080524106347793, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7923502326011658, + "num_tokens": 212438489.0, + "step": 5566 + }, + { + "epoch": 0.7081796209133698, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7872961163520813, + "num_tokens": 212482283.0, + "step": 5567 + }, + { + "epoch": 0.7083068311919604, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7902565598487854, + "num_tokens": 212522749.0, + "step": 5568 + }, + { + "epoch": 0.7084340414705508, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7865523099899292, + "num_tokens": 212559937.0, + "step": 5569 + }, + { + "epoch": 0.7085612517491413, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8034390211105347, + "num_tokens": 212595578.0, + "step": 5570 + }, + { + "epoch": 0.7086884620277318, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8064146041870117, + "num_tokens": 212634085.0, + "step": 5571 + }, + { + "epoch": 0.7088156723063224, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7842804193496704, + "num_tokens": 212673557.0, + "step": 5572 + }, + { + "epoch": 0.7089428825849129, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8083248734474182, + "num_tokens": 212711142.0, + "step": 5573 + }, + { + "epoch": 0.7090700928635034, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7873093485832214, + "num_tokens": 212745570.0, + "step": 5574 + }, + { + "epoch": 0.7091973031420938, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7979626655578613, + "num_tokens": 212783048.0, + "step": 5575 + }, + { + "epoch": 0.7093245134206844, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.814736008644104, + "num_tokens": 212821231.0, + "step": 5576 + }, + { + "epoch": 0.7094517236992749, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7931758165359497, + "num_tokens": 212861715.0, + "step": 5577 + }, + { + "epoch": 0.7095789339778654, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7837343215942383, + "num_tokens": 212900655.0, + "step": 5578 + }, + { + "epoch": 0.709706144256456, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7902254462242126, + "num_tokens": 212939308.0, + "step": 5579 + }, + { + "epoch": 0.7098333545350465, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8019818067550659, + "num_tokens": 212981174.0, + "step": 5580 + }, + { + "epoch": 0.7099605648136369, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8074355125427246, + "num_tokens": 213016401.0, + "step": 5581 + }, + { + "epoch": 0.7100877750922274, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7849063277244568, + "num_tokens": 213058480.0, + "step": 5582 + }, + { + "epoch": 0.710214985370818, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8068238496780396, + "num_tokens": 213093527.0, + "step": 5583 + }, + { + "epoch": 0.7103421956494085, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.794424831867218, + "num_tokens": 213126778.0, + "step": 5584 + }, + { + "epoch": 0.710469405927999, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8012661933898926, + "num_tokens": 213170863.0, + "step": 5585 + }, + { + "epoch": 0.7105966162065895, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7716915607452393, + "num_tokens": 213211336.0, + "step": 5586 + }, + { + "epoch": 0.71072382648518, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7983146905899048, + "num_tokens": 213245262.0, + "step": 5587 + }, + { + "epoch": 0.7108510367637705, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7889835834503174, + "num_tokens": 213282150.0, + "step": 5588 + }, + { + "epoch": 0.710978247042361, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8014323711395264, + "num_tokens": 213325573.0, + "step": 5589 + }, + { + "epoch": 0.7111054573209515, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8057256937026978, + "num_tokens": 213363738.0, + "step": 5590 + }, + { + "epoch": 0.7112326675995421, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7993682622909546, + "num_tokens": 213404796.0, + "step": 5591 + }, + { + "epoch": 0.7113598778781326, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7859023809432983, + "num_tokens": 213436711.0, + "step": 5592 + }, + { + "epoch": 0.711487088156723, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7956107258796692, + "num_tokens": 213473458.0, + "step": 5593 + }, + { + "epoch": 0.7116142984353135, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8085643649101257, + "num_tokens": 213507355.0, + "step": 5594 + }, + { + "epoch": 0.7117415087139041, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7739457488059998, + "num_tokens": 213543771.0, + "step": 5595 + }, + { + "epoch": 0.7118687189924946, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.793458104133606, + "num_tokens": 213578317.0, + "step": 5596 + }, + { + "epoch": 0.7119959292710851, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7915933132171631, + "num_tokens": 213615795.0, + "step": 5597 + }, + { + "epoch": 0.7121231395496757, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.8071128129959106, + "num_tokens": 213655452.0, + "step": 5598 + }, + { + "epoch": 0.7122503498282661, + "grad_norm": 78.5, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7835397720336914, + "num_tokens": 213688079.0, + "step": 5599 + }, + { + "epoch": 0.7123775601068566, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.8124809861183167, + "num_tokens": 213725310.0, + "step": 5600 + }, + { + "epoch": 0.7125047703854471, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7925452589988708, + "num_tokens": 213757950.0, + "step": 5601 + }, + { + "epoch": 0.7126319806640377, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7812972068786621, + "num_tokens": 213794146.0, + "step": 5602 + }, + { + "epoch": 0.7127591909426282, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7976861596107483, + "num_tokens": 213834922.0, + "step": 5603 + }, + { + "epoch": 0.7128864012212187, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8189699649810791, + "num_tokens": 213879651.0, + "step": 5604 + }, + { + "epoch": 0.7130136114998091, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7964357733726501, + "num_tokens": 213921833.0, + "step": 5605 + }, + { + "epoch": 0.7131408217783997, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8053779602050781, + "num_tokens": 213962909.0, + "step": 5606 + }, + { + "epoch": 0.7132680320569902, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7985538840293884, + "num_tokens": 214002030.0, + "step": 5607 + }, + { + "epoch": 0.7133952423355807, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7884931564331055, + "num_tokens": 214037094.0, + "step": 5608 + }, + { + "epoch": 0.7135224526141712, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.805830717086792, + "num_tokens": 214076649.0, + "step": 5609 + }, + { + "epoch": 0.7136496628927618, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.791074812412262, + "num_tokens": 214114194.0, + "step": 5610 + }, + { + "epoch": 0.7137768731713523, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7951650619506836, + "num_tokens": 214149652.0, + "step": 5611 + }, + { + "epoch": 0.7139040834499427, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8052565455436707, + "num_tokens": 214189068.0, + "step": 5612 + }, + { + "epoch": 0.7140312937285332, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7774223685264587, + "num_tokens": 214220251.0, + "step": 5613 + }, + { + "epoch": 0.7141585040071238, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7896212339401245, + "num_tokens": 214256705.0, + "step": 5614 + }, + { + "epoch": 0.7142857142857143, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8118951320648193, + "num_tokens": 214292563.0, + "step": 5615 + }, + { + "epoch": 0.7144129245643048, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7911554574966431, + "num_tokens": 214333843.0, + "step": 5616 + }, + { + "epoch": 0.7145401348428954, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8045551776885986, + "num_tokens": 214372272.0, + "step": 5617 + }, + { + "epoch": 0.7146673451214858, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8020138740539551, + "num_tokens": 214416364.0, + "step": 5618 + }, + { + "epoch": 0.7147945554000763, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.794496476650238, + "num_tokens": 214453102.0, + "step": 5619 + }, + { + "epoch": 0.7149217656786668, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.798662543296814, + "num_tokens": 214492374.0, + "step": 5620 + }, + { + "epoch": 0.7150489759572574, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7951974272727966, + "num_tokens": 214530918.0, + "step": 5621 + }, + { + "epoch": 0.7151761862358479, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8114187121391296, + "num_tokens": 214570227.0, + "step": 5622 + }, + { + "epoch": 0.7153033965144384, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7998175621032715, + "num_tokens": 214609519.0, + "step": 5623 + }, + { + "epoch": 0.7154306067930288, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7873921990394592, + "num_tokens": 214643564.0, + "step": 5624 + }, + { + "epoch": 0.7155578170716194, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8042612075805664, + "num_tokens": 214677632.0, + "step": 5625 + }, + { + "epoch": 0.7156850273502099, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.798440158367157, + "num_tokens": 214710363.0, + "step": 5626 + }, + { + "epoch": 0.7158122376288004, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.801719069480896, + "num_tokens": 214746982.0, + "step": 5627 + }, + { + "epoch": 0.715939447907391, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7882784008979797, + "num_tokens": 214786447.0, + "step": 5628 + }, + { + "epoch": 0.7160666581859815, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8033576011657715, + "num_tokens": 214824627.0, + "step": 5629 + }, + { + "epoch": 0.7161938684645719, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7738754749298096, + "num_tokens": 214863051.0, + "step": 5630 + }, + { + "epoch": 0.7163210787431624, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7843672633171082, + "num_tokens": 214900524.0, + "step": 5631 + }, + { + "epoch": 0.716448289021753, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8065448999404907, + "num_tokens": 214943949.0, + "step": 5632 + }, + { + "epoch": 0.7165754993003435, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.8008564114570618, + "num_tokens": 214979126.0, + "step": 5633 + }, + { + "epoch": 0.716702709578934, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7990705966949463, + "num_tokens": 215017740.0, + "step": 5634 + }, + { + "epoch": 0.7168299198575245, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8033783435821533, + "num_tokens": 215054472.0, + "step": 5635 + }, + { + "epoch": 0.716957130136115, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7965173125267029, + "num_tokens": 215086894.0, + "step": 5636 + }, + { + "epoch": 0.7170843404147055, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8053765296936035, + "num_tokens": 215128157.0, + "step": 5637 + }, + { + "epoch": 0.717211550693296, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8001316785812378, + "num_tokens": 215161008.0, + "step": 5638 + }, + { + "epoch": 0.7173387609718865, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7788756489753723, + "num_tokens": 215197157.0, + "step": 5639 + }, + { + "epoch": 0.7174659712504771, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.806160032749176, + "num_tokens": 215240270.0, + "step": 5640 + }, + { + "epoch": 0.7175931815290676, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8010498285293579, + "num_tokens": 215279828.0, + "step": 5641 + }, + { + "epoch": 0.717720391807658, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7889742851257324, + "num_tokens": 215315271.0, + "step": 5642 + }, + { + "epoch": 0.7178476020862485, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.795285701751709, + "num_tokens": 215352621.0, + "step": 5643 + }, + { + "epoch": 0.7179748123648391, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.795494556427002, + "num_tokens": 215390231.0, + "step": 5644 + }, + { + "epoch": 0.7181020226434296, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.8153712749481201, + "num_tokens": 215426812.0, + "step": 5645 + }, + { + "epoch": 0.7182292329220201, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7803983688354492, + "num_tokens": 215467551.0, + "step": 5646 + }, + { + "epoch": 0.7183564432006107, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8130662441253662, + "num_tokens": 215508559.0, + "step": 5647 + }, + { + "epoch": 0.7184836534792011, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7874508500099182, + "num_tokens": 215548709.0, + "step": 5648 + }, + { + "epoch": 0.7186108637577916, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7995665073394775, + "num_tokens": 215593844.0, + "step": 5649 + }, + { + "epoch": 0.7187380740363821, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8067358136177063, + "num_tokens": 215633853.0, + "step": 5650 + }, + { + "epoch": 0.7188652843149727, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7951960563659668, + "num_tokens": 215667272.0, + "step": 5651 + }, + { + "epoch": 0.7189924945935632, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7718368768692017, + "num_tokens": 215705889.0, + "step": 5652 + }, + { + "epoch": 0.7191197048721537, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.806740403175354, + "num_tokens": 215749143.0, + "step": 5653 + }, + { + "epoch": 0.7192469151507441, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7895640730857849, + "num_tokens": 215787583.0, + "step": 5654 + }, + { + "epoch": 0.7193741254293347, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7962276935577393, + "num_tokens": 215825552.0, + "step": 5655 + }, + { + "epoch": 0.7195013357079252, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8038643598556519, + "num_tokens": 215858310.0, + "step": 5656 + }, + { + "epoch": 0.7196285459865157, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7839330434799194, + "num_tokens": 215898403.0, + "step": 5657 + }, + { + "epoch": 0.7197557562651062, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7720867395401001, + "num_tokens": 215935930.0, + "step": 5658 + }, + { + "epoch": 0.7198829665436968, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7909890413284302, + "num_tokens": 215971509.0, + "step": 5659 + }, + { + "epoch": 0.7200101768222873, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8039081692695618, + "num_tokens": 216008696.0, + "step": 5660 + }, + { + "epoch": 0.7201373871008777, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7989562153816223, + "num_tokens": 216045861.0, + "step": 5661 + }, + { + "epoch": 0.7202645973794682, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.8079221248626709, + "num_tokens": 216081307.0, + "step": 5662 + }, + { + "epoch": 0.7203918076580588, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8080936074256897, + "num_tokens": 216118506.0, + "step": 5663 + }, + { + "epoch": 0.7205190179366493, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7898515462875366, + "num_tokens": 216159000.0, + "step": 5664 + }, + { + "epoch": 0.7206462282152398, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7924402952194214, + "num_tokens": 216204712.0, + "step": 5665 + }, + { + "epoch": 0.7207734384938304, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7834349870681763, + "num_tokens": 216235001.0, + "step": 5666 + }, + { + "epoch": 0.7209006487724208, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7877836227416992, + "num_tokens": 216271383.0, + "step": 5667 + }, + { + "epoch": 0.7210278590510113, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7961294651031494, + "num_tokens": 216307923.0, + "step": 5668 + }, + { + "epoch": 0.7211550693296018, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8173331022262573, + "num_tokens": 216345147.0, + "step": 5669 + }, + { + "epoch": 0.7212822796081924, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7856099009513855, + "num_tokens": 216386991.0, + "step": 5670 + }, + { + "epoch": 0.7214094898867829, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7970679998397827, + "num_tokens": 216425534.0, + "step": 5671 + }, + { + "epoch": 0.7215367001653734, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7980086803436279, + "num_tokens": 216466686.0, + "step": 5672 + }, + { + "epoch": 0.7216639104439638, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8118675947189331, + "num_tokens": 216505245.0, + "step": 5673 + }, + { + "epoch": 0.7217911207225544, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7939820289611816, + "num_tokens": 216538024.0, + "step": 5674 + }, + { + "epoch": 0.7219183310011449, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7945659160614014, + "num_tokens": 216573722.0, + "step": 5675 + }, + { + "epoch": 0.7220455412797354, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8009375333786011, + "num_tokens": 216610362.0, + "step": 5676 + }, + { + "epoch": 0.7221727515583259, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7877861261367798, + "num_tokens": 216648732.0, + "step": 5677 + }, + { + "epoch": 0.7222999618369165, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8082638382911682, + "num_tokens": 216680324.0, + "step": 5678 + }, + { + "epoch": 0.7224271721155069, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8004804849624634, + "num_tokens": 216712773.0, + "step": 5679 + }, + { + "epoch": 0.7225543823940974, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7918009161949158, + "num_tokens": 216752538.0, + "step": 5680 + }, + { + "epoch": 0.722681592672688, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8019604682922363, + "num_tokens": 216791795.0, + "step": 5681 + }, + { + "epoch": 0.7228088029512785, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.791904091835022, + "num_tokens": 216831729.0, + "step": 5682 + }, + { + "epoch": 0.722936013229869, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7967555522918701, + "num_tokens": 216869494.0, + "step": 5683 + }, + { + "epoch": 0.7230632235084595, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7781267762184143, + "num_tokens": 216906672.0, + "step": 5684 + }, + { + "epoch": 0.72319043378705, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8002961277961731, + "num_tokens": 216941587.0, + "step": 5685 + }, + { + "epoch": 0.7233176440656405, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7923722267150879, + "num_tokens": 216981812.0, + "step": 5686 + }, + { + "epoch": 0.723444854344231, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8071441650390625, + "num_tokens": 217024137.0, + "step": 5687 + }, + { + "epoch": 0.7235720646228215, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.8147172331809998, + "num_tokens": 217063753.0, + "step": 5688 + }, + { + "epoch": 0.7236992749014121, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7984007596969604, + "num_tokens": 217105611.0, + "step": 5689 + }, + { + "epoch": 0.7238264851800026, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7799862623214722, + "num_tokens": 217145989.0, + "step": 5690 + }, + { + "epoch": 0.723953695458593, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.81883305311203, + "num_tokens": 217183726.0, + "step": 5691 + }, + { + "epoch": 0.7240809057371835, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7970629930496216, + "num_tokens": 217226644.0, + "step": 5692 + }, + { + "epoch": 0.7242081160157741, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7827851176261902, + "num_tokens": 217263327.0, + "step": 5693 + }, + { + "epoch": 0.7243353262943646, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8036680817604065, + "num_tokens": 217300595.0, + "step": 5694 + }, + { + "epoch": 0.7244625365729551, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7841718196868896, + "num_tokens": 217336385.0, + "step": 5695 + }, + { + "epoch": 0.7245897468515456, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8004925847053528, + "num_tokens": 217374309.0, + "step": 5696 + }, + { + "epoch": 0.7247169571301361, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7873384356498718, + "num_tokens": 217408386.0, + "step": 5697 + }, + { + "epoch": 0.7248441674087266, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7771012783050537, + "num_tokens": 217442855.0, + "step": 5698 + }, + { + "epoch": 0.7249713776873171, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8012686371803284, + "num_tokens": 217476892.0, + "step": 5699 + }, + { + "epoch": 0.7250985879659076, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7887757420539856, + "num_tokens": 217513569.0, + "step": 5700 + }, + { + "epoch": 0.7252257982444982, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7849390506744385, + "num_tokens": 217545585.0, + "step": 5701 + }, + { + "epoch": 0.7253530085230887, + "grad_norm": 89.0, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7956684827804565, + "num_tokens": 217583156.0, + "step": 5702 + }, + { + "epoch": 0.7254802188016791, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7994034290313721, + "num_tokens": 217619630.0, + "step": 5703 + }, + { + "epoch": 0.7256074290802697, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8075052499771118, + "num_tokens": 217653616.0, + "step": 5704 + }, + { + "epoch": 0.7257346393588602, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7964652180671692, + "num_tokens": 217692758.0, + "step": 5705 + }, + { + "epoch": 0.7258618496374507, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7812654376029968, + "num_tokens": 217734175.0, + "step": 5706 + }, + { + "epoch": 0.7259890599160412, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7753542065620422, + "num_tokens": 217773372.0, + "step": 5707 + }, + { + "epoch": 0.7261162701946318, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8073214292526245, + "num_tokens": 217809174.0, + "step": 5708 + }, + { + "epoch": 0.7262434804732223, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7883957028388977, + "num_tokens": 217845882.0, + "step": 5709 + }, + { + "epoch": 0.7263706907518127, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7956851720809937, + "num_tokens": 217885798.0, + "step": 5710 + }, + { + "epoch": 0.7264979010304032, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7882295250892639, + "num_tokens": 217924967.0, + "step": 5711 + }, + { + "epoch": 0.7266251113089938, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8060342073440552, + "num_tokens": 217967381.0, + "step": 5712 + }, + { + "epoch": 0.7267523215875843, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8000904321670532, + "num_tokens": 218001097.0, + "step": 5713 + }, + { + "epoch": 0.7268795318661748, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.797162652015686, + "num_tokens": 218040705.0, + "step": 5714 + }, + { + "epoch": 0.7270067421447653, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7948834896087646, + "num_tokens": 218077549.0, + "step": 5715 + }, + { + "epoch": 0.7271339524233558, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7930178642272949, + "num_tokens": 218116251.0, + "step": 5716 + }, + { + "epoch": 0.7272611627019463, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7725597620010376, + "num_tokens": 218155798.0, + "step": 5717 + }, + { + "epoch": 0.7273883729805368, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8023099899291992, + "num_tokens": 218191277.0, + "step": 5718 + }, + { + "epoch": 0.7275155832591274, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8080686330795288, + "num_tokens": 218233165.0, + "step": 5719 + }, + { + "epoch": 0.7276427935377179, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7745329737663269, + "num_tokens": 218271263.0, + "step": 5720 + }, + { + "epoch": 0.7277700038163084, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7950742840766907, + "num_tokens": 218309081.0, + "step": 5721 + }, + { + "epoch": 0.7278972140948988, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7875310182571411, + "num_tokens": 218348314.0, + "step": 5722 + }, + { + "epoch": 0.7280244243734894, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7957225441932678, + "num_tokens": 218387556.0, + "step": 5723 + }, + { + "epoch": 0.7281516346520799, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7887508869171143, + "num_tokens": 218423130.0, + "step": 5724 + }, + { + "epoch": 0.7282788449306704, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7762356400489807, + "num_tokens": 218464079.0, + "step": 5725 + }, + { + "epoch": 0.7284060552092609, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8018902540206909, + "num_tokens": 218501307.0, + "step": 5726 + }, + { + "epoch": 0.7285332654878515, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8061630725860596, + "num_tokens": 218537710.0, + "step": 5727 + }, + { + "epoch": 0.7286604757664419, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7897080183029175, + "num_tokens": 218576026.0, + "step": 5728 + }, + { + "epoch": 0.7287876860450324, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7915596961975098, + "num_tokens": 218612491.0, + "step": 5729 + }, + { + "epoch": 0.7289148963236229, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8041333556175232, + "num_tokens": 218642541.0, + "step": 5730 + }, + { + "epoch": 0.7290421066022135, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7767698764801025, + "num_tokens": 218678514.0, + "step": 5731 + }, + { + "epoch": 0.729169316880804, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8004343509674072, + "num_tokens": 218715415.0, + "step": 5732 + }, + { + "epoch": 0.7292965271593945, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8024083375930786, + "num_tokens": 218754422.0, + "step": 5733 + }, + { + "epoch": 0.7294237374379849, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7811441421508789, + "num_tokens": 218787141.0, + "step": 5734 + }, + { + "epoch": 0.7295509477165755, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8108102083206177, + "num_tokens": 218824764.0, + "step": 5735 + }, + { + "epoch": 0.729678157995166, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7972202301025391, + "num_tokens": 218866142.0, + "step": 5736 + }, + { + "epoch": 0.7298053682737565, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7939020395278931, + "num_tokens": 218909208.0, + "step": 5737 + }, + { + "epoch": 0.7299325785523471, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8061213493347168, + "num_tokens": 218942469.0, + "step": 5738 + }, + { + "epoch": 0.7300597888309376, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7863941192626953, + "num_tokens": 218982934.0, + "step": 5739 + }, + { + "epoch": 0.730186999109528, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7933924198150635, + "num_tokens": 219027055.0, + "step": 5740 + }, + { + "epoch": 0.7303142093881185, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8006075024604797, + "num_tokens": 219069639.0, + "step": 5741 + }, + { + "epoch": 0.7304414196667091, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7905462384223938, + "num_tokens": 219111774.0, + "step": 5742 + }, + { + "epoch": 0.7305686299452996, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7832083702087402, + "num_tokens": 219156962.0, + "step": 5743 + }, + { + "epoch": 0.7306958402238901, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8053910732269287, + "num_tokens": 219193352.0, + "step": 5744 + }, + { + "epoch": 0.7308230505024806, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7810152769088745, + "num_tokens": 219230041.0, + "step": 5745 + }, + { + "epoch": 0.7309502607810711, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7734245657920837, + "num_tokens": 219262748.0, + "step": 5746 + }, + { + "epoch": 0.7310774710596616, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.811741828918457, + "num_tokens": 219296556.0, + "step": 5747 + }, + { + "epoch": 0.7312046813382521, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.785345196723938, + "num_tokens": 219330801.0, + "step": 5748 + }, + { + "epoch": 0.7313318916168426, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7947908639907837, + "num_tokens": 219367470.0, + "step": 5749 + }, + { + "epoch": 0.7314591018954332, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7769085168838501, + "num_tokens": 219405291.0, + "step": 5750 + }, + { + "epoch": 0.7315863121740237, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7990335822105408, + "num_tokens": 219443365.0, + "step": 5751 + }, + { + "epoch": 0.7317135224526141, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7980289459228516, + "num_tokens": 219477524.0, + "step": 5752 + }, + { + "epoch": 0.7318407327312046, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7838276028633118, + "num_tokens": 219514576.0, + "step": 5753 + }, + { + "epoch": 0.7319679430097952, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7956562042236328, + "num_tokens": 219558253.0, + "step": 5754 + }, + { + "epoch": 0.7320951532883857, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7925134897232056, + "num_tokens": 219601156.0, + "step": 5755 + }, + { + "epoch": 0.7322223635669762, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7881439328193665, + "num_tokens": 219638241.0, + "step": 5756 + }, + { + "epoch": 0.7323495738455668, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7985304594039917, + "num_tokens": 219678168.0, + "step": 5757 + }, + { + "epoch": 0.7324767841241572, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.794215977191925, + "num_tokens": 219718115.0, + "step": 5758 + }, + { + "epoch": 0.7326039944027477, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7956076860427856, + "num_tokens": 219754221.0, + "step": 5759 + }, + { + "epoch": 0.7327312046813382, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7962735891342163, + "num_tokens": 219786880.0, + "step": 5760 + }, + { + "epoch": 0.7328584149599288, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8036765456199646, + "num_tokens": 219830585.0, + "step": 5761 + }, + { + "epoch": 0.7329856252385193, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7967404127120972, + "num_tokens": 219867249.0, + "step": 5762 + }, + { + "epoch": 0.7331128355171098, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7883313894271851, + "num_tokens": 219906046.0, + "step": 5763 + }, + { + "epoch": 0.7332400457957003, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.78738933801651, + "num_tokens": 219947466.0, + "step": 5764 + }, + { + "epoch": 0.7333672560742908, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7946348786354065, + "num_tokens": 219985698.0, + "step": 5765 + }, + { + "epoch": 0.7334944663528813, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7970495223999023, + "num_tokens": 220026158.0, + "step": 5766 + }, + { + "epoch": 0.7336216766314718, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8009361028671265, + "num_tokens": 220071849.0, + "step": 5767 + }, + { + "epoch": 0.7337488869100623, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7990785241127014, + "num_tokens": 220110255.0, + "step": 5768 + }, + { + "epoch": 0.7338760971886529, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7784830927848816, + "num_tokens": 220144962.0, + "step": 5769 + }, + { + "epoch": 0.7340033074672434, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7887014746665955, + "num_tokens": 220182118.0, + "step": 5770 + }, + { + "epoch": 0.7341305177458338, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.789982259273529, + "num_tokens": 220218954.0, + "step": 5771 + }, + { + "epoch": 0.7342577280244243, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7961521148681641, + "num_tokens": 220259116.0, + "step": 5772 + }, + { + "epoch": 0.7343849383030149, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7921921014785767, + "num_tokens": 220301742.0, + "step": 5773 + }, + { + "epoch": 0.7345121485816054, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7993240356445312, + "num_tokens": 220342572.0, + "step": 5774 + }, + { + "epoch": 0.7346393588601959, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7990680932998657, + "num_tokens": 220376373.0, + "step": 5775 + }, + { + "epoch": 0.7347665691387865, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8153730034828186, + "num_tokens": 220413290.0, + "step": 5776 + }, + { + "epoch": 0.7348937794173769, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7946512699127197, + "num_tokens": 220453219.0, + "step": 5777 + }, + { + "epoch": 0.7350209896959674, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.793781042098999, + "num_tokens": 220488369.0, + "step": 5778 + }, + { + "epoch": 0.7351481999745579, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8023884296417236, + "num_tokens": 220527798.0, + "step": 5779 + }, + { + "epoch": 0.7352754102531485, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8196709752082825, + "num_tokens": 220565398.0, + "step": 5780 + }, + { + "epoch": 0.735402620531739, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8114038705825806, + "num_tokens": 220603903.0, + "step": 5781 + }, + { + "epoch": 0.7355298308103295, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7808904647827148, + "num_tokens": 220644168.0, + "step": 5782 + }, + { + "epoch": 0.7356570410889199, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8043622970581055, + "num_tokens": 220679709.0, + "step": 5783 + }, + { + "epoch": 0.7357842513675105, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.794845461845398, + "num_tokens": 220721184.0, + "step": 5784 + }, + { + "epoch": 0.735911461646101, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8027583360671997, + "num_tokens": 220755027.0, + "step": 5785 + }, + { + "epoch": 0.7360386719246915, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7707056999206543, + "num_tokens": 220793597.0, + "step": 5786 + }, + { + "epoch": 0.736165882203282, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7984523177146912, + "num_tokens": 220833549.0, + "step": 5787 + }, + { + "epoch": 0.7362930924818726, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8008942604064941, + "num_tokens": 220865551.0, + "step": 5788 + }, + { + "epoch": 0.736420302760463, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8088574409484863, + "num_tokens": 220911626.0, + "step": 5789 + }, + { + "epoch": 0.7365475130390535, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.796934962272644, + "num_tokens": 220949839.0, + "step": 5790 + }, + { + "epoch": 0.736674723317644, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7708749771118164, + "num_tokens": 220984069.0, + "step": 5791 + }, + { + "epoch": 0.7368019335962346, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8171483278274536, + "num_tokens": 221020999.0, + "step": 5792 + }, + { + "epoch": 0.7369291438748251, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7800444960594177, + "num_tokens": 221051082.0, + "step": 5793 + }, + { + "epoch": 0.7370563541534156, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7781906723976135, + "num_tokens": 221095444.0, + "step": 5794 + }, + { + "epoch": 0.737183564432006, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8012828826904297, + "num_tokens": 221136433.0, + "step": 5795 + }, + { + "epoch": 0.7373107747105966, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7935057282447815, + "num_tokens": 221171278.0, + "step": 5796 + }, + { + "epoch": 0.7374379849891871, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7923897504806519, + "num_tokens": 221209380.0, + "step": 5797 + }, + { + "epoch": 0.7375651952677776, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7954405546188354, + "num_tokens": 221253076.0, + "step": 5798 + }, + { + "epoch": 0.7376924055463682, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7920795679092407, + "num_tokens": 221290348.0, + "step": 5799 + }, + { + "epoch": 0.7378196158249587, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8045447468757629, + "num_tokens": 221329635.0, + "step": 5800 + }, + { + "epoch": 0.7379468261035491, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8044703006744385, + "num_tokens": 221368521.0, + "step": 5801 + }, + { + "epoch": 0.7380740363821396, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8129644393920898, + "num_tokens": 221407090.0, + "step": 5802 + }, + { + "epoch": 0.7382012466607302, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8104440569877625, + "num_tokens": 221450340.0, + "step": 5803 + }, + { + "epoch": 0.7383284569393207, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.8018444180488586, + "num_tokens": 221491996.0, + "step": 5804 + }, + { + "epoch": 0.7384556672179112, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8000519871711731, + "num_tokens": 221531466.0, + "step": 5805 + }, + { + "epoch": 0.7385828774965018, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7775604724884033, + "num_tokens": 221567003.0, + "step": 5806 + }, + { + "epoch": 0.7387100877750922, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.805326521396637, + "num_tokens": 221607396.0, + "step": 5807 + }, + { + "epoch": 0.7388372980536827, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7915246486663818, + "num_tokens": 221644776.0, + "step": 5808 + }, + { + "epoch": 0.7389645083322732, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8115454316139221, + "num_tokens": 221685643.0, + "step": 5809 + }, + { + "epoch": 0.7390917186108638, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7894865274429321, + "num_tokens": 221724673.0, + "step": 5810 + }, + { + "epoch": 0.7392189288894543, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7685443162918091, + "num_tokens": 221766287.0, + "step": 5811 + }, + { + "epoch": 0.7393461391680448, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7966057062149048, + "num_tokens": 221805357.0, + "step": 5812 + }, + { + "epoch": 0.7394733494466353, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7939897775650024, + "num_tokens": 221838513.0, + "step": 5813 + }, + { + "epoch": 0.7396005597252258, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.786314070224762, + "num_tokens": 221882808.0, + "step": 5814 + }, + { + "epoch": 0.7397277700038163, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8070005178451538, + "num_tokens": 221918505.0, + "step": 5815 + }, + { + "epoch": 0.7398549802824068, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7933832406997681, + "num_tokens": 221955983.0, + "step": 5816 + }, + { + "epoch": 0.7399821905609973, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7915288209915161, + "num_tokens": 221997363.0, + "step": 5817 + }, + { + "epoch": 0.7401094008395879, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7795605659484863, + "num_tokens": 222039302.0, + "step": 5818 + }, + { + "epoch": 0.7402366111181784, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7929033041000366, + "num_tokens": 222075779.0, + "step": 5819 + }, + { + "epoch": 0.7403638213967688, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8146102428436279, + "num_tokens": 222115120.0, + "step": 5820 + }, + { + "epoch": 0.7404910316753593, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7894772291183472, + "num_tokens": 222156570.0, + "step": 5821 + }, + { + "epoch": 0.7406182419539499, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7989879846572876, + "num_tokens": 222189487.0, + "step": 5822 + }, + { + "epoch": 0.7407454522325404, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7905669808387756, + "num_tokens": 222223163.0, + "step": 5823 + }, + { + "epoch": 0.7408726625111309, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.806001603603363, + "num_tokens": 222259042.0, + "step": 5824 + }, + { + "epoch": 0.7409998727897215, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7786251306533813, + "num_tokens": 222300497.0, + "step": 5825 + }, + { + "epoch": 0.7411270830683119, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7875687479972839, + "num_tokens": 222333630.0, + "step": 5826 + }, + { + "epoch": 0.7412542933469024, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7913298606872559, + "num_tokens": 222371715.0, + "step": 5827 + }, + { + "epoch": 0.7413815036254929, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7940288782119751, + "num_tokens": 222412044.0, + "step": 5828 + }, + { + "epoch": 0.7415087139040835, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7999066710472107, + "num_tokens": 222452668.0, + "step": 5829 + }, + { + "epoch": 0.741635924182674, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7930914163589478, + "num_tokens": 222485631.0, + "step": 5830 + }, + { + "epoch": 0.7417631344612645, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8105220794677734, + "num_tokens": 222522967.0, + "step": 5831 + }, + { + "epoch": 0.7418903447398549, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7861753106117249, + "num_tokens": 222563258.0, + "step": 5832 + }, + { + "epoch": 0.7420175550184455, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7871601581573486, + "num_tokens": 222599248.0, + "step": 5833 + }, + { + "epoch": 0.742144765297036, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7993543148040771, + "num_tokens": 222642575.0, + "step": 5834 + }, + { + "epoch": 0.7422719755756265, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7706064581871033, + "num_tokens": 222687330.0, + "step": 5835 + }, + { + "epoch": 0.742399185854217, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7645909786224365, + "num_tokens": 222720020.0, + "step": 5836 + }, + { + "epoch": 0.7425263961328076, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8080017566680908, + "num_tokens": 222758634.0, + "step": 5837 + }, + { + "epoch": 0.742653606411398, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8017681241035461, + "num_tokens": 222794823.0, + "step": 5838 + }, + { + "epoch": 0.7427808166899885, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7979087829589844, + "num_tokens": 222833836.0, + "step": 5839 + }, + { + "epoch": 0.742908026968579, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.801301121711731, + "num_tokens": 222875712.0, + "step": 5840 + }, + { + "epoch": 0.7430352372471696, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7870534658432007, + "num_tokens": 222915494.0, + "step": 5841 + }, + { + "epoch": 0.7431624475257601, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.8143337965011597, + "num_tokens": 222951675.0, + "step": 5842 + }, + { + "epoch": 0.7432896578043506, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7969825863838196, + "num_tokens": 222993245.0, + "step": 5843 + }, + { + "epoch": 0.743416868082941, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8019029498100281, + "num_tokens": 223029876.0, + "step": 5844 + }, + { + "epoch": 0.7435440783615316, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8029255270957947, + "num_tokens": 223072550.0, + "step": 5845 + }, + { + "epoch": 0.7436712886401221, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7919319868087769, + "num_tokens": 223112892.0, + "step": 5846 + }, + { + "epoch": 0.7437984989187126, + "grad_norm": 9.8125, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8022252321243286, + "num_tokens": 223157093.0, + "step": 5847 + }, + { + "epoch": 0.7439257091973032, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7812613844871521, + "num_tokens": 223190756.0, + "step": 5848 + }, + { + "epoch": 0.7440529194758937, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7847486734390259, + "num_tokens": 223226013.0, + "step": 5849 + }, + { + "epoch": 0.7441801297544841, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8076820373535156, + "num_tokens": 223262858.0, + "step": 5850 + }, + { + "epoch": 0.7443073400330746, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8031031489372253, + "num_tokens": 223310528.0, + "step": 5851 + }, + { + "epoch": 0.7444345503116652, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7958360910415649, + "num_tokens": 223345021.0, + "step": 5852 + }, + { + "epoch": 0.7445617605902557, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8080061674118042, + "num_tokens": 223382055.0, + "step": 5853 + }, + { + "epoch": 0.7446889708688462, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7870743870735168, + "num_tokens": 223418441.0, + "step": 5854 + }, + { + "epoch": 0.7448161811474368, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.796628475189209, + "num_tokens": 223458150.0, + "step": 5855 + }, + { + "epoch": 0.7449433914260272, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7915033102035522, + "num_tokens": 223496916.0, + "step": 5856 + }, + { + "epoch": 0.7450706017046177, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8056105971336365, + "num_tokens": 223534808.0, + "step": 5857 + }, + { + "epoch": 0.7451978119832082, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8049677610397339, + "num_tokens": 223571255.0, + "step": 5858 + }, + { + "epoch": 0.7453250222617988, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8041293621063232, + "num_tokens": 223616608.0, + "step": 5859 + }, + { + "epoch": 0.7454522325403893, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.798890233039856, + "num_tokens": 223656188.0, + "step": 5860 + }, + { + "epoch": 0.7455794428189798, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7906022667884827, + "num_tokens": 223693787.0, + "step": 5861 + }, + { + "epoch": 0.7457066530975703, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8068879842758179, + "num_tokens": 223729350.0, + "step": 5862 + }, + { + "epoch": 0.7458338633761608, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7870644330978394, + "num_tokens": 223770846.0, + "step": 5863 + }, + { + "epoch": 0.7459610736547513, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7943955659866333, + "num_tokens": 223814194.0, + "step": 5864 + }, + { + "epoch": 0.7460882839333418, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.788555920124054, + "num_tokens": 223849141.0, + "step": 5865 + }, + { + "epoch": 0.7462154942119323, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8195446133613586, + "num_tokens": 223888654.0, + "step": 5866 + }, + { + "epoch": 0.7463427044905229, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7991523146629333, + "num_tokens": 223924509.0, + "step": 5867 + }, + { + "epoch": 0.7464699147691134, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7818430066108704, + "num_tokens": 223966191.0, + "step": 5868 + }, + { + "epoch": 0.7465971250477038, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7968674302101135, + "num_tokens": 224001415.0, + "step": 5869 + }, + { + "epoch": 0.7467243353262943, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8087790012359619, + "num_tokens": 224042820.0, + "step": 5870 + }, + { + "epoch": 0.7468515456048849, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8013504147529602, + "num_tokens": 224078411.0, + "step": 5871 + }, + { + "epoch": 0.7469787558834754, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8045281171798706, + "num_tokens": 224117136.0, + "step": 5872 + }, + { + "epoch": 0.7471059661620659, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7840984463691711, + "num_tokens": 224155879.0, + "step": 5873 + }, + { + "epoch": 0.7472331764406565, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8104584217071533, + "num_tokens": 224195186.0, + "step": 5874 + }, + { + "epoch": 0.7473603867192469, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7904670238494873, + "num_tokens": 224235722.0, + "step": 5875 + }, + { + "epoch": 0.7474875969978374, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8062748908996582, + "num_tokens": 224268461.0, + "step": 5876 + }, + { + "epoch": 0.7476148072764279, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.805695652961731, + "num_tokens": 224302921.0, + "step": 5877 + }, + { + "epoch": 0.7477420175550185, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8007006645202637, + "num_tokens": 224344081.0, + "step": 5878 + }, + { + "epoch": 0.747869227833609, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7814410924911499, + "num_tokens": 224384464.0, + "step": 5879 + }, + { + "epoch": 0.7479964381121995, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8084885478019714, + "num_tokens": 224425542.0, + "step": 5880 + }, + { + "epoch": 0.7481236483907899, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8154434561729431, + "num_tokens": 224465342.0, + "step": 5881 + }, + { + "epoch": 0.7482508586693805, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.796716570854187, + "num_tokens": 224504787.0, + "step": 5882 + }, + { + "epoch": 0.748378068947971, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7969576120376587, + "num_tokens": 224548534.0, + "step": 5883 + }, + { + "epoch": 0.7485052792265615, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7914663553237915, + "num_tokens": 224590251.0, + "step": 5884 + }, + { + "epoch": 0.748632489505152, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8091277480125427, + "num_tokens": 224633745.0, + "step": 5885 + }, + { + "epoch": 0.7487596997837426, + "grad_norm": 60.5, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7736002206802368, + "num_tokens": 224671319.0, + "step": 5886 + }, + { + "epoch": 0.748886910062333, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7974954843521118, + "num_tokens": 224712670.0, + "step": 5887 + }, + { + "epoch": 0.7490141203409235, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7912764549255371, + "num_tokens": 224747840.0, + "step": 5888 + }, + { + "epoch": 0.749141330619514, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8075125217437744, + "num_tokens": 224783274.0, + "step": 5889 + }, + { + "epoch": 0.7492685408981046, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7981166839599609, + "num_tokens": 224820456.0, + "step": 5890 + }, + { + "epoch": 0.7493957511766951, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7959417104721069, + "num_tokens": 224858569.0, + "step": 5891 + }, + { + "epoch": 0.7495229614552856, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7963501214981079, + "num_tokens": 224901033.0, + "step": 5892 + }, + { + "epoch": 0.749650171733876, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8118756413459778, + "num_tokens": 224934345.0, + "step": 5893 + }, + { + "epoch": 0.7497773820124666, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7902044057846069, + "num_tokens": 224972479.0, + "step": 5894 + }, + { + "epoch": 0.7499045922910571, + "grad_norm": 120.5, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8129551410675049, + "num_tokens": 225014213.0, + "step": 5895 + }, + { + "epoch": 0.7500318025696476, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8056038022041321, + "num_tokens": 225049929.0, + "step": 5896 + }, + { + "epoch": 0.7501590128482382, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7789883613586426, + "num_tokens": 225086409.0, + "step": 5897 + }, + { + "epoch": 0.7502862231268287, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7783268094062805, + "num_tokens": 225124443.0, + "step": 5898 + }, + { + "epoch": 0.7504134334054191, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7858264446258545, + "num_tokens": 225166665.0, + "step": 5899 + }, + { + "epoch": 0.7505406436840096, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7698222398757935, + "num_tokens": 225203841.0, + "step": 5900 + }, + { + "epoch": 0.7506678539626002, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7727710604667664, + "num_tokens": 225234844.0, + "step": 5901 + }, + { + "epoch": 0.7507950642411907, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7731294631958008, + "num_tokens": 225269828.0, + "step": 5902 + }, + { + "epoch": 0.7509222745197812, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8016940951347351, + "num_tokens": 225312361.0, + "step": 5903 + }, + { + "epoch": 0.7510494847983717, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7822266221046448, + "num_tokens": 225353708.0, + "step": 5904 + }, + { + "epoch": 0.7511766950769622, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.8190286159515381, + "num_tokens": 225399730.0, + "step": 5905 + }, + { + "epoch": 0.7513039053555527, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7881947755813599, + "num_tokens": 225441836.0, + "step": 5906 + }, + { + "epoch": 0.7514311156341432, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7971883416175842, + "num_tokens": 225471004.0, + "step": 5907 + }, + { + "epoch": 0.7515583259127337, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7742594480514526, + "num_tokens": 225505789.0, + "step": 5908 + }, + { + "epoch": 0.7516855361913243, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7994123697280884, + "num_tokens": 225541138.0, + "step": 5909 + }, + { + "epoch": 0.7518127464699148, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.798089325428009, + "num_tokens": 225581695.0, + "step": 5910 + }, + { + "epoch": 0.7519399567485053, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8043949604034424, + "num_tokens": 225619532.0, + "step": 5911 + }, + { + "epoch": 0.7520671670270958, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7876144647598267, + "num_tokens": 225657403.0, + "step": 5912 + }, + { + "epoch": 0.7521943773056863, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7919194102287292, + "num_tokens": 225693402.0, + "step": 5913 + }, + { + "epoch": 0.7523215875842768, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7918981313705444, + "num_tokens": 225727208.0, + "step": 5914 + }, + { + "epoch": 0.7524487978628673, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7984322905540466, + "num_tokens": 225762117.0, + "step": 5915 + }, + { + "epoch": 0.7525760081414579, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7726723551750183, + "num_tokens": 225794737.0, + "step": 5916 + }, + { + "epoch": 0.7527032184200484, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7908917665481567, + "num_tokens": 225829074.0, + "step": 5917 + }, + { + "epoch": 0.7528304286986388, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8110416531562805, + "num_tokens": 225871344.0, + "step": 5918 + }, + { + "epoch": 0.7529576389772293, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7930612564086914, + "num_tokens": 225911321.0, + "step": 5919 + }, + { + "epoch": 0.7530848492558199, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7992744445800781, + "num_tokens": 225944613.0, + "step": 5920 + }, + { + "epoch": 0.7532120595344104, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.8027902841567993, + "num_tokens": 225989501.0, + "step": 5921 + }, + { + "epoch": 0.7533392698130009, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7948081493377686, + "num_tokens": 226024984.0, + "step": 5922 + }, + { + "epoch": 0.7534664800915915, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8115946054458618, + "num_tokens": 226063255.0, + "step": 5923 + }, + { + "epoch": 0.7535936903701819, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7828834652900696, + "num_tokens": 226105538.0, + "step": 5924 + }, + { + "epoch": 0.7537209006487724, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7953248023986816, + "num_tokens": 226137449.0, + "step": 5925 + }, + { + "epoch": 0.7538481109273629, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7875796556472778, + "num_tokens": 226171078.0, + "step": 5926 + }, + { + "epoch": 0.7539753212059535, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8026149868965149, + "num_tokens": 226208060.0, + "step": 5927 + }, + { + "epoch": 0.754102531484544, + "grad_norm": 10.5, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8045291304588318, + "num_tokens": 226245330.0, + "step": 5928 + }, + { + "epoch": 0.7542297417631345, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8184038400650024, + "num_tokens": 226281974.0, + "step": 5929 + }, + { + "epoch": 0.7543569520417249, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7893047332763672, + "num_tokens": 226323302.0, + "step": 5930 + }, + { + "epoch": 0.7544841623203155, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7815360426902771, + "num_tokens": 226361346.0, + "step": 5931 + }, + { + "epoch": 0.754611372598906, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7883890867233276, + "num_tokens": 226397744.0, + "step": 5932 + }, + { + "epoch": 0.7547385828774965, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8067370057106018, + "num_tokens": 226436745.0, + "step": 5933 + }, + { + "epoch": 0.754865793156087, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8035550117492676, + "num_tokens": 226470302.0, + "step": 5934 + }, + { + "epoch": 0.7549930034346776, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.792187511920929, + "num_tokens": 226500090.0, + "step": 5935 + }, + { + "epoch": 0.755120213713268, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7983541488647461, + "num_tokens": 226533359.0, + "step": 5936 + }, + { + "epoch": 0.7552474239918585, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7896252274513245, + "num_tokens": 226567060.0, + "step": 5937 + }, + { + "epoch": 0.755374634270449, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7930734753608704, + "num_tokens": 226604270.0, + "step": 5938 + }, + { + "epoch": 0.7555018445490396, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.792941689491272, + "num_tokens": 226641310.0, + "step": 5939 + }, + { + "epoch": 0.7556290548276301, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7856103181838989, + "num_tokens": 226678305.0, + "step": 5940 + }, + { + "epoch": 0.7557562651062206, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8028866052627563, + "num_tokens": 226715628.0, + "step": 5941 + }, + { + "epoch": 0.755883475384811, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8066281676292419, + "num_tokens": 226755346.0, + "step": 5942 + }, + { + "epoch": 0.7560106856634016, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7826049327850342, + "num_tokens": 226792243.0, + "step": 5943 + }, + { + "epoch": 0.7561378959419921, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8082528114318848, + "num_tokens": 226829033.0, + "step": 5944 + }, + { + "epoch": 0.7562651062205826, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8094367980957031, + "num_tokens": 226870645.0, + "step": 5945 + }, + { + "epoch": 0.7563923164991732, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7993517518043518, + "num_tokens": 226902395.0, + "step": 5946 + }, + { + "epoch": 0.7565195267777637, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8034926652908325, + "num_tokens": 226942620.0, + "step": 5947 + }, + { + "epoch": 0.7566467370563541, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8041433095932007, + "num_tokens": 226976472.0, + "step": 5948 + }, + { + "epoch": 0.7567739473349446, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7883764505386353, + "num_tokens": 227013626.0, + "step": 5949 + }, + { + "epoch": 0.7569011576135352, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7798149585723877, + "num_tokens": 227049336.0, + "step": 5950 + }, + { + "epoch": 0.7570283678921257, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8186384439468384, + "num_tokens": 227087371.0, + "step": 5951 + }, + { + "epoch": 0.7571555781707162, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8014782071113586, + "num_tokens": 227129188.0, + "step": 5952 + }, + { + "epoch": 0.7572827884493067, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7841647863388062, + "num_tokens": 227161807.0, + "step": 5953 + }, + { + "epoch": 0.7574099987278972, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7864905595779419, + "num_tokens": 227198431.0, + "step": 5954 + }, + { + "epoch": 0.7575372090064877, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8182348012924194, + "num_tokens": 227236046.0, + "step": 5955 + }, + { + "epoch": 0.7576644192850782, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8191291093826294, + "num_tokens": 227272925.0, + "step": 5956 + }, + { + "epoch": 0.7577916295636687, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8027859926223755, + "num_tokens": 227312311.0, + "step": 5957 + }, + { + "epoch": 0.7579188398422593, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8189090490341187, + "num_tokens": 227349632.0, + "step": 5958 + }, + { + "epoch": 0.7580460501208498, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7961491346359253, + "num_tokens": 227392982.0, + "step": 5959 + }, + { + "epoch": 0.7581732603994403, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7949433326721191, + "num_tokens": 227431878.0, + "step": 5960 + }, + { + "epoch": 0.7583004706780307, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7861236333847046, + "num_tokens": 227470401.0, + "step": 5961 + }, + { + "epoch": 0.7584276809566213, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8029527068138123, + "num_tokens": 227507189.0, + "step": 5962 + }, + { + "epoch": 0.7585548912352118, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.802200436592102, + "num_tokens": 227548070.0, + "step": 5963 + }, + { + "epoch": 0.7586821015138023, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8016571998596191, + "num_tokens": 227584360.0, + "step": 5964 + }, + { + "epoch": 0.7588093117923929, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.8026451468467712, + "num_tokens": 227618961.0, + "step": 5965 + }, + { + "epoch": 0.7589365220709834, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7961945533752441, + "num_tokens": 227658304.0, + "step": 5966 + }, + { + "epoch": 0.7590637323495738, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.794082522392273, + "num_tokens": 227701526.0, + "step": 5967 + }, + { + "epoch": 0.7591909426281643, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.805619478225708, + "num_tokens": 227744930.0, + "step": 5968 + }, + { + "epoch": 0.7593181529067549, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8076144456863403, + "num_tokens": 227780607.0, + "step": 5969 + }, + { + "epoch": 0.7594453631853454, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7952929139137268, + "num_tokens": 227825742.0, + "step": 5970 + }, + { + "epoch": 0.7595725734639359, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7741649150848389, + "num_tokens": 227865278.0, + "step": 5971 + }, + { + "epoch": 0.7596997837425264, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7996331453323364, + "num_tokens": 227906912.0, + "step": 5972 + }, + { + "epoch": 0.7598269940211169, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7995597720146179, + "num_tokens": 227947067.0, + "step": 5973 + }, + { + "epoch": 0.7599542042997074, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7786705493927002, + "num_tokens": 227988563.0, + "step": 5974 + }, + { + "epoch": 0.7600814145782979, + "grad_norm": 86.0, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7893324494361877, + "num_tokens": 228028556.0, + "step": 5975 + }, + { + "epoch": 0.7602086248568884, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8144159317016602, + "num_tokens": 228066265.0, + "step": 5976 + }, + { + "epoch": 0.760335835135479, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7950960397720337, + "num_tokens": 228103732.0, + "step": 5977 + }, + { + "epoch": 0.7604630454140695, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7980924844741821, + "num_tokens": 228149609.0, + "step": 5978 + }, + { + "epoch": 0.7605902556926599, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7972724437713623, + "num_tokens": 228188462.0, + "step": 5979 + }, + { + "epoch": 0.7607174659712505, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8088463544845581, + "num_tokens": 228229252.0, + "step": 5980 + }, + { + "epoch": 0.760844676249841, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7936840653419495, + "num_tokens": 228267397.0, + "step": 5981 + }, + { + "epoch": 0.7609718865284315, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8069231510162354, + "num_tokens": 228307561.0, + "step": 5982 + }, + { + "epoch": 0.761099096807022, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8024661540985107, + "num_tokens": 228347667.0, + "step": 5983 + }, + { + "epoch": 0.7612263070856126, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.793344259262085, + "num_tokens": 228386820.0, + "step": 5984 + }, + { + "epoch": 0.761353517364203, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8060325384140015, + "num_tokens": 228427304.0, + "step": 5985 + }, + { + "epoch": 0.7614807276427935, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7887403964996338, + "num_tokens": 228470779.0, + "step": 5986 + }, + { + "epoch": 0.761607937921384, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.814418613910675, + "num_tokens": 228500454.0, + "step": 5987 + }, + { + "epoch": 0.7617351481999746, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8075968027114868, + "num_tokens": 228537374.0, + "step": 5988 + }, + { + "epoch": 0.7618623584785651, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7979110479354858, + "num_tokens": 228577215.0, + "step": 5989 + }, + { + "epoch": 0.7619895687571556, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8000969886779785, + "num_tokens": 228611693.0, + "step": 5990 + }, + { + "epoch": 0.762116779035746, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7924127578735352, + "num_tokens": 228644231.0, + "step": 5991 + }, + { + "epoch": 0.7622439893143366, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.80776047706604, + "num_tokens": 228684776.0, + "step": 5992 + }, + { + "epoch": 0.7623711995929271, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8040612936019897, + "num_tokens": 228724214.0, + "step": 5993 + }, + { + "epoch": 0.7624984098715176, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7947123050689697, + "num_tokens": 228765098.0, + "step": 5994 + }, + { + "epoch": 0.7626256201501082, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7974166870117188, + "num_tokens": 228802423.0, + "step": 5995 + }, + { + "epoch": 0.7627528304286987, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7855519652366638, + "num_tokens": 228840218.0, + "step": 5996 + }, + { + "epoch": 0.7628800407072891, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8072803020477295, + "num_tokens": 228882810.0, + "step": 5997 + }, + { + "epoch": 0.7630072509858796, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.8002698421478271, + "num_tokens": 228917664.0, + "step": 5998 + }, + { + "epoch": 0.7631344612644702, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8041689395904541, + "num_tokens": 228951017.0, + "step": 5999 + }, + { + "epoch": 0.7632616715430607, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8003764152526855, + "num_tokens": 228990949.0, + "step": 6000 + }, + { + "epoch": 0.7633888818216512, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8025614619255066, + "num_tokens": 229026604.0, + "step": 6001 + }, + { + "epoch": 0.7635160921002417, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7930405735969543, + "num_tokens": 229061810.0, + "step": 6002 + }, + { + "epoch": 0.7636433023788322, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7745708227157593, + "num_tokens": 229107765.0, + "step": 6003 + }, + { + "epoch": 0.7637705126574227, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7879597544670105, + "num_tokens": 229147822.0, + "step": 6004 + }, + { + "epoch": 0.7638977229360132, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8074027895927429, + "num_tokens": 229184421.0, + "step": 6005 + }, + { + "epoch": 0.7640249332146037, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.796244740486145, + "num_tokens": 229221737.0, + "step": 6006 + }, + { + "epoch": 0.7641521434931943, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.789260983467102, + "num_tokens": 229254727.0, + "step": 6007 + }, + { + "epoch": 0.7642793537717848, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.807839035987854, + "num_tokens": 229292648.0, + "step": 6008 + }, + { + "epoch": 0.7644065640503753, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8025779724121094, + "num_tokens": 229335708.0, + "step": 6009 + }, + { + "epoch": 0.7645337743289657, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8068809509277344, + "num_tokens": 229373791.0, + "step": 6010 + }, + { + "epoch": 0.7646609846075563, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8071799278259277, + "num_tokens": 229409891.0, + "step": 6011 + }, + { + "epoch": 0.7647881948861468, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8006112575531006, + "num_tokens": 229449859.0, + "step": 6012 + }, + { + "epoch": 0.7649154051647373, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8033353090286255, + "num_tokens": 229484224.0, + "step": 6013 + }, + { + "epoch": 0.7650426154433279, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7913863658905029, + "num_tokens": 229521104.0, + "step": 6014 + }, + { + "epoch": 0.7651698257219184, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7813002467155457, + "num_tokens": 229558918.0, + "step": 6015 + }, + { + "epoch": 0.7652970360005088, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7825953960418701, + "num_tokens": 229596931.0, + "step": 6016 + }, + { + "epoch": 0.7654242462790993, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7841607332229614, + "num_tokens": 229633903.0, + "step": 6017 + }, + { + "epoch": 0.7655514565576899, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7921315431594849, + "num_tokens": 229674224.0, + "step": 6018 + }, + { + "epoch": 0.7656786668362804, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7941123247146606, + "num_tokens": 229710547.0, + "step": 6019 + }, + { + "epoch": 0.7658058771148709, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8165513873100281, + "num_tokens": 229752020.0, + "step": 6020 + }, + { + "epoch": 0.7659330873934614, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7983219623565674, + "num_tokens": 229785710.0, + "step": 6021 + }, + { + "epoch": 0.7660602976720519, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8048274517059326, + "num_tokens": 229824118.0, + "step": 6022 + }, + { + "epoch": 0.7661875079506424, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8015457391738892, + "num_tokens": 229854231.0, + "step": 6023 + }, + { + "epoch": 0.7663147182292329, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7888169884681702, + "num_tokens": 229888102.0, + "step": 6024 + }, + { + "epoch": 0.7664419285078234, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.8015342950820923, + "num_tokens": 229927634.0, + "step": 6025 + }, + { + "epoch": 0.766569138786414, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8044575452804565, + "num_tokens": 229966526.0, + "step": 6026 + }, + { + "epoch": 0.7666963490650045, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8107829689979553, + "num_tokens": 230002761.0, + "step": 6027 + }, + { + "epoch": 0.7668235593435949, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8055797815322876, + "num_tokens": 230041303.0, + "step": 6028 + }, + { + "epoch": 0.7669507696221854, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7993208765983582, + "num_tokens": 230081117.0, + "step": 6029 + }, + { + "epoch": 0.767077979900776, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7931594848632812, + "num_tokens": 230119411.0, + "step": 6030 + }, + { + "epoch": 0.7672051901793665, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8070551156997681, + "num_tokens": 230161111.0, + "step": 6031 + }, + { + "epoch": 0.767332400457957, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7916674613952637, + "num_tokens": 230196999.0, + "step": 6032 + }, + { + "epoch": 0.7674596107365476, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8021644353866577, + "num_tokens": 230231495.0, + "step": 6033 + }, + { + "epoch": 0.767586821015138, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7921832203865051, + "num_tokens": 230267732.0, + "step": 6034 + }, + { + "epoch": 0.7677140312937285, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7964836359024048, + "num_tokens": 230297559.0, + "step": 6035 + }, + { + "epoch": 0.767841241572319, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7809180021286011, + "num_tokens": 230336396.0, + "step": 6036 + }, + { + "epoch": 0.7679684518509096, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7919352650642395, + "num_tokens": 230368152.0, + "step": 6037 + }, + { + "epoch": 0.7680956621295001, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7989676594734192, + "num_tokens": 230408081.0, + "step": 6038 + }, + { + "epoch": 0.7682228724080906, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7887247800827026, + "num_tokens": 230441691.0, + "step": 6039 + }, + { + "epoch": 0.768350082686681, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7968207597732544, + "num_tokens": 230480089.0, + "step": 6040 + }, + { + "epoch": 0.7684772929652716, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7910513877868652, + "num_tokens": 230518655.0, + "step": 6041 + }, + { + "epoch": 0.7686045032438621, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7993699312210083, + "num_tokens": 230555717.0, + "step": 6042 + }, + { + "epoch": 0.7687317135224526, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8075112104415894, + "num_tokens": 230596699.0, + "step": 6043 + }, + { + "epoch": 0.7688589238010431, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8007574081420898, + "num_tokens": 230633284.0, + "step": 6044 + }, + { + "epoch": 0.7689861340796337, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8002680540084839, + "num_tokens": 230673133.0, + "step": 6045 + }, + { + "epoch": 0.7691133443582241, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8031702041625977, + "num_tokens": 230719003.0, + "step": 6046 + }, + { + "epoch": 0.7692405546368146, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8104828596115112, + "num_tokens": 230754805.0, + "step": 6047 + }, + { + "epoch": 0.7693677649154052, + "grad_norm": 130.0, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.785416305065155, + "num_tokens": 230793249.0, + "step": 6048 + }, + { + "epoch": 0.7694949751939957, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8075960874557495, + "num_tokens": 230826025.0, + "step": 6049 + }, + { + "epoch": 0.7696221854725862, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8137508034706116, + "num_tokens": 230866615.0, + "step": 6050 + }, + { + "epoch": 0.7697493957511767, + "grad_norm": 106.0, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7829372882843018, + "num_tokens": 230901587.0, + "step": 6051 + }, + { + "epoch": 0.7698766060297672, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8111958503723145, + "num_tokens": 230944267.0, + "step": 6052 + }, + { + "epoch": 0.7700038163083577, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7925675511360168, + "num_tokens": 230980022.0, + "step": 6053 + }, + { + "epoch": 0.7701310265869482, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.800460696220398, + "num_tokens": 231018110.0, + "step": 6054 + }, + { + "epoch": 0.7702582368655387, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8018541932106018, + "num_tokens": 231056531.0, + "step": 6055 + }, + { + "epoch": 0.7703854471441293, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8003556132316589, + "num_tokens": 231096270.0, + "step": 6056 + }, + { + "epoch": 0.7705126574227198, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7915880084037781, + "num_tokens": 231136300.0, + "step": 6057 + }, + { + "epoch": 0.7706398677013102, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7841657400131226, + "num_tokens": 231169005.0, + "step": 6058 + }, + { + "epoch": 0.7707670779799007, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.800060510635376, + "num_tokens": 231205274.0, + "step": 6059 + }, + { + "epoch": 0.7708942882584913, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7970373630523682, + "num_tokens": 231239022.0, + "step": 6060 + }, + { + "epoch": 0.7710214985370818, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7803888320922852, + "num_tokens": 231282952.0, + "step": 6061 + }, + { + "epoch": 0.7711487088156723, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7976171374320984, + "num_tokens": 231322980.0, + "step": 6062 + }, + { + "epoch": 0.7712759190942629, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7882444858551025, + "num_tokens": 231358382.0, + "step": 6063 + }, + { + "epoch": 0.7714031293728534, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8062291145324707, + "num_tokens": 231401915.0, + "step": 6064 + }, + { + "epoch": 0.7715303396514438, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7717313766479492, + "num_tokens": 231433241.0, + "step": 6065 + }, + { + "epoch": 0.7716575499300343, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7940251231193542, + "num_tokens": 231478318.0, + "step": 6066 + }, + { + "epoch": 0.7717847602086249, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7982311248779297, + "num_tokens": 231518584.0, + "step": 6067 + }, + { + "epoch": 0.7719119704872154, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7993865013122559, + "num_tokens": 231556689.0, + "step": 6068 + }, + { + "epoch": 0.7720391807658059, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7971140146255493, + "num_tokens": 231593936.0, + "step": 6069 + }, + { + "epoch": 0.7721663910443964, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8201149702072144, + "num_tokens": 231634785.0, + "step": 6070 + }, + { + "epoch": 0.7722936013229869, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7929689884185791, + "num_tokens": 231675798.0, + "step": 6071 + }, + { + "epoch": 0.7724208116015774, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7926546335220337, + "num_tokens": 231711611.0, + "step": 6072 + }, + { + "epoch": 0.7725480218801679, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7851738929748535, + "num_tokens": 231746997.0, + "step": 6073 + }, + { + "epoch": 0.7726752321587584, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7805681824684143, + "num_tokens": 231787696.0, + "step": 6074 + }, + { + "epoch": 0.772802442437349, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7924216985702515, + "num_tokens": 231826127.0, + "step": 6075 + }, + { + "epoch": 0.7729296527159395, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.783662736415863, + "num_tokens": 231857538.0, + "step": 6076 + }, + { + "epoch": 0.7730568629945299, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7984124422073364, + "num_tokens": 231893913.0, + "step": 6077 + }, + { + "epoch": 0.7731840732731204, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8000320196151733, + "num_tokens": 231934274.0, + "step": 6078 + }, + { + "epoch": 0.773311283551711, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7940264344215393, + "num_tokens": 231983197.0, + "step": 6079 + }, + { + "epoch": 0.7734384938303015, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7961364388465881, + "num_tokens": 232021265.0, + "step": 6080 + }, + { + "epoch": 0.773565704108892, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7962023019790649, + "num_tokens": 232060992.0, + "step": 6081 + }, + { + "epoch": 0.7736929143874826, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7955889701843262, + "num_tokens": 232096758.0, + "step": 6082 + }, + { + "epoch": 0.773820124666073, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7933446168899536, + "num_tokens": 232133773.0, + "step": 6083 + }, + { + "epoch": 0.7739473349446635, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7967538237571716, + "num_tokens": 232173182.0, + "step": 6084 + }, + { + "epoch": 0.774074545223254, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8015917539596558, + "num_tokens": 232206478.0, + "step": 6085 + }, + { + "epoch": 0.7742017555018446, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7883550524711609, + "num_tokens": 232242231.0, + "step": 6086 + }, + { + "epoch": 0.7743289657804351, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7941794991493225, + "num_tokens": 232270636.0, + "step": 6087 + }, + { + "epoch": 0.7744561760590256, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.804456889629364, + "num_tokens": 232309974.0, + "step": 6088 + }, + { + "epoch": 0.774583386337616, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8032335042953491, + "num_tokens": 232344118.0, + "step": 6089 + }, + { + "epoch": 0.7747105966162066, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7859744429588318, + "num_tokens": 232385183.0, + "step": 6090 + }, + { + "epoch": 0.7748378068947971, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7971622943878174, + "num_tokens": 232421670.0, + "step": 6091 + }, + { + "epoch": 0.7749650171733876, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7807009220123291, + "num_tokens": 232459563.0, + "step": 6092 + }, + { + "epoch": 0.7750922274519781, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.796353816986084, + "num_tokens": 232496357.0, + "step": 6093 + }, + { + "epoch": 0.7752194377305687, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7962364554405212, + "num_tokens": 232530436.0, + "step": 6094 + }, + { + "epoch": 0.7753466480091591, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7911127805709839, + "num_tokens": 232568585.0, + "step": 6095 + }, + { + "epoch": 0.7754738582877496, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8022812604904175, + "num_tokens": 232602506.0, + "step": 6096 + }, + { + "epoch": 0.7756010685663401, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8040320873260498, + "num_tokens": 232642819.0, + "step": 6097 + }, + { + "epoch": 0.7757282788449307, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7877119779586792, + "num_tokens": 232679505.0, + "step": 6098 + }, + { + "epoch": 0.7758554891235212, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.8017817735671997, + "num_tokens": 232715701.0, + "step": 6099 + }, + { + "epoch": 0.7759826994021117, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8021829724311829, + "num_tokens": 232750786.0, + "step": 6100 + }, + { + "epoch": 0.7761099096807021, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7846974730491638, + "num_tokens": 232791370.0, + "step": 6101 + }, + { + "epoch": 0.7762371199592927, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.799501895904541, + "num_tokens": 232830895.0, + "step": 6102 + }, + { + "epoch": 0.7763643302378832, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7945219874382019, + "num_tokens": 232869940.0, + "step": 6103 + }, + { + "epoch": 0.7764915405164737, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7951359748840332, + "num_tokens": 232910743.0, + "step": 6104 + }, + { + "epoch": 0.7766187507950643, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7957519888877869, + "num_tokens": 232948428.0, + "step": 6105 + }, + { + "epoch": 0.7767459610736548, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8018558025360107, + "num_tokens": 232985142.0, + "step": 6106 + }, + { + "epoch": 0.7768731713522452, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8024693727493286, + "num_tokens": 233020810.0, + "step": 6107 + }, + { + "epoch": 0.7770003816308357, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7880650758743286, + "num_tokens": 233055674.0, + "step": 6108 + }, + { + "epoch": 0.7771275919094263, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7893317937850952, + "num_tokens": 233084506.0, + "step": 6109 + }, + { + "epoch": 0.7772548021880168, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8141738176345825, + "num_tokens": 233120172.0, + "step": 6110 + }, + { + "epoch": 0.7773820124666073, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7833387851715088, + "num_tokens": 233160075.0, + "step": 6111 + }, + { + "epoch": 0.7775092227451978, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7947936654090881, + "num_tokens": 233201714.0, + "step": 6112 + }, + { + "epoch": 0.7776364330237884, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8046659231185913, + "num_tokens": 233237642.0, + "step": 6113 + }, + { + "epoch": 0.7777636433023788, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.806023895740509, + "num_tokens": 233279712.0, + "step": 6114 + }, + { + "epoch": 0.7778908535809693, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8040599822998047, + "num_tokens": 233322441.0, + "step": 6115 + }, + { + "epoch": 0.7780180638595598, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7929495573043823, + "num_tokens": 233371053.0, + "step": 6116 + }, + { + "epoch": 0.7781452741381504, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8003696203231812, + "num_tokens": 233405372.0, + "step": 6117 + }, + { + "epoch": 0.7782724844167409, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8076078295707703, + "num_tokens": 233447209.0, + "step": 6118 + }, + { + "epoch": 0.7783996946953314, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7990598678588867, + "num_tokens": 233484621.0, + "step": 6119 + }, + { + "epoch": 0.7785269049739219, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7785919904708862, + "num_tokens": 233520756.0, + "step": 6120 + }, + { + "epoch": 0.7786541152525124, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7949410080909729, + "num_tokens": 233559120.0, + "step": 6121 + }, + { + "epoch": 0.7787813255311029, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8003231287002563, + "num_tokens": 233590513.0, + "step": 6122 + }, + { + "epoch": 0.7789085358096934, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8096266388893127, + "num_tokens": 233625986.0, + "step": 6123 + }, + { + "epoch": 0.779035746088284, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8052870035171509, + "num_tokens": 233655097.0, + "step": 6124 + }, + { + "epoch": 0.7791629563668745, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7901476621627808, + "num_tokens": 233687815.0, + "step": 6125 + }, + { + "epoch": 0.7792901666454649, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8165085315704346, + "num_tokens": 233722611.0, + "step": 6126 + }, + { + "epoch": 0.7794173769240554, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7869666218757629, + "num_tokens": 233765345.0, + "step": 6127 + }, + { + "epoch": 0.779544587202646, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7993623614311218, + "num_tokens": 233804358.0, + "step": 6128 + }, + { + "epoch": 0.7796717974812365, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7859235405921936, + "num_tokens": 233845117.0, + "step": 6129 + }, + { + "epoch": 0.779799007759827, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7895472049713135, + "num_tokens": 233884091.0, + "step": 6130 + }, + { + "epoch": 0.7799262180384176, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.792801022529602, + "num_tokens": 233922590.0, + "step": 6131 + }, + { + "epoch": 0.780053428317008, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7956904172897339, + "num_tokens": 233954138.0, + "step": 6132 + }, + { + "epoch": 0.7801806385955985, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8081988096237183, + "num_tokens": 233988035.0, + "step": 6133 + }, + { + "epoch": 0.780307848874189, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.788743257522583, + "num_tokens": 234023747.0, + "step": 6134 + }, + { + "epoch": 0.7804350591527796, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8094983696937561, + "num_tokens": 234064182.0, + "step": 6135 + }, + { + "epoch": 0.7805622694313701, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.780992865562439, + "num_tokens": 234106533.0, + "step": 6136 + }, + { + "epoch": 0.7806894797099606, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8112665414810181, + "num_tokens": 234145492.0, + "step": 6137 + }, + { + "epoch": 0.780816689988551, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8036357164382935, + "num_tokens": 234181709.0, + "step": 6138 + }, + { + "epoch": 0.7809439002671416, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7978047132492065, + "num_tokens": 234217854.0, + "step": 6139 + }, + { + "epoch": 0.7810711105457321, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7973374128341675, + "num_tokens": 234255769.0, + "step": 6140 + }, + { + "epoch": 0.7811983208243226, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7794812917709351, + "num_tokens": 234290828.0, + "step": 6141 + }, + { + "epoch": 0.7813255311029131, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.795974612236023, + "num_tokens": 234326336.0, + "step": 6142 + }, + { + "epoch": 0.7814527413815037, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.788179874420166, + "num_tokens": 234366210.0, + "step": 6143 + }, + { + "epoch": 0.7815799516600941, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7870932817459106, + "num_tokens": 234398975.0, + "step": 6144 + }, + { + "epoch": 0.7817071619386846, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7909299731254578, + "num_tokens": 234437746.0, + "step": 6145 + }, + { + "epoch": 0.7818343722172751, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7994852066040039, + "num_tokens": 234478402.0, + "step": 6146 + }, + { + "epoch": 0.7819615824958657, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7827304601669312, + "num_tokens": 234509699.0, + "step": 6147 + }, + { + "epoch": 0.7820887927744562, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7947264909744263, + "num_tokens": 234551515.0, + "step": 6148 + }, + { + "epoch": 0.7822160030530467, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7999864816665649, + "num_tokens": 234582811.0, + "step": 6149 + }, + { + "epoch": 0.7823432133316371, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7874687314033508, + "num_tokens": 234625299.0, + "step": 6150 + }, + { + "epoch": 0.7824704236102277, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8054704070091248, + "num_tokens": 234665641.0, + "step": 6151 + }, + { + "epoch": 0.7825976338888182, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7928920984268188, + "num_tokens": 234704668.0, + "step": 6152 + }, + { + "epoch": 0.7827248441674087, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8015812635421753, + "num_tokens": 234738080.0, + "step": 6153 + }, + { + "epoch": 0.7828520544459993, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.792578935623169, + "num_tokens": 234781077.0, + "step": 6154 + }, + { + "epoch": 0.7829792647245898, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7765669822692871, + "num_tokens": 234809374.0, + "step": 6155 + }, + { + "epoch": 0.7831064750031802, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7825096845626831, + "num_tokens": 234845585.0, + "step": 6156 + }, + { + "epoch": 0.7832336852817707, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7983265519142151, + "num_tokens": 234883891.0, + "step": 6157 + }, + { + "epoch": 0.7833608955603613, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7917215824127197, + "num_tokens": 234919445.0, + "step": 6158 + }, + { + "epoch": 0.7834881058389518, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7970402240753174, + "num_tokens": 234951465.0, + "step": 6159 + }, + { + "epoch": 0.7836153161175423, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7874488830566406, + "num_tokens": 234989970.0, + "step": 6160 + }, + { + "epoch": 0.7837425263961328, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7981051206588745, + "num_tokens": 235029673.0, + "step": 6161 + }, + { + "epoch": 0.7838697366747234, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7979809045791626, + "num_tokens": 235065389.0, + "step": 6162 + }, + { + "epoch": 0.7839969469533138, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8056939840316772, + "num_tokens": 235099795.0, + "step": 6163 + }, + { + "epoch": 0.7841241572319043, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7993831038475037, + "num_tokens": 235137475.0, + "step": 6164 + }, + { + "epoch": 0.7842513675104948, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7940764427185059, + "num_tokens": 235178454.0, + "step": 6165 + }, + { + "epoch": 0.7843785777890854, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8203647136688232, + "num_tokens": 235216139.0, + "step": 6166 + }, + { + "epoch": 0.7845057880676759, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8125731348991394, + "num_tokens": 235253571.0, + "step": 6167 + }, + { + "epoch": 0.7846329983462664, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7865800857543945, + "num_tokens": 235293950.0, + "step": 6168 + }, + { + "epoch": 0.7847602086248568, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8022171854972839, + "num_tokens": 235336419.0, + "step": 6169 + }, + { + "epoch": 0.7848874189034474, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8023985028266907, + "num_tokens": 235376039.0, + "step": 6170 + }, + { + "epoch": 0.7850146291820379, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7839429974555969, + "num_tokens": 235412736.0, + "step": 6171 + }, + { + "epoch": 0.7851418394606284, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8022570610046387, + "num_tokens": 235457318.0, + "step": 6172 + }, + { + "epoch": 0.785269049739219, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.786695122718811, + "num_tokens": 235491471.0, + "step": 6173 + }, + { + "epoch": 0.7853962600178095, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8074203729629517, + "num_tokens": 235531324.0, + "step": 6174 + }, + { + "epoch": 0.7855234702963999, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7857860326766968, + "num_tokens": 235573233.0, + "step": 6175 + }, + { + "epoch": 0.7856506805749904, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8019081354141235, + "num_tokens": 235614400.0, + "step": 6176 + }, + { + "epoch": 0.785777890853581, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8071260452270508, + "num_tokens": 235652247.0, + "step": 6177 + }, + { + "epoch": 0.7859051011321715, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7921576499938965, + "num_tokens": 235681810.0, + "step": 6178 + }, + { + "epoch": 0.786032311410762, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8068740367889404, + "num_tokens": 235718422.0, + "step": 6179 + }, + { + "epoch": 0.7861595216893525, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7901352643966675, + "num_tokens": 235758886.0, + "step": 6180 + }, + { + "epoch": 0.786286731967943, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.797988772392273, + "num_tokens": 235792270.0, + "step": 6181 + }, + { + "epoch": 0.7864139422465335, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7884188294410706, + "num_tokens": 235826678.0, + "step": 6182 + }, + { + "epoch": 0.786541152525124, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8092248439788818, + "num_tokens": 235860437.0, + "step": 6183 + }, + { + "epoch": 0.7866683628037145, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7994762063026428, + "num_tokens": 235900865.0, + "step": 6184 + }, + { + "epoch": 0.7867955730823051, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7824074029922485, + "num_tokens": 235937419.0, + "step": 6185 + }, + { + "epoch": 0.7869227833608956, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7885088920593262, + "num_tokens": 235976425.0, + "step": 6186 + }, + { + "epoch": 0.787049993639486, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7880168557167053, + "num_tokens": 236012968.0, + "step": 6187 + }, + { + "epoch": 0.7871772039180766, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.799302339553833, + "num_tokens": 236052462.0, + "step": 6188 + }, + { + "epoch": 0.7873044141966671, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.809748113155365, + "num_tokens": 236087604.0, + "step": 6189 + }, + { + "epoch": 0.7874316244752576, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.795369029045105, + "num_tokens": 236126344.0, + "step": 6190 + }, + { + "epoch": 0.7875588347538481, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8040460348129272, + "num_tokens": 236161913.0, + "step": 6191 + }, + { + "epoch": 0.7876860450324387, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7996257543563843, + "num_tokens": 236200612.0, + "step": 6192 + }, + { + "epoch": 0.7878132553110291, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8040798306465149, + "num_tokens": 236238136.0, + "step": 6193 + }, + { + "epoch": 0.7879404655896196, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7969919443130493, + "num_tokens": 236281303.0, + "step": 6194 + }, + { + "epoch": 0.7880676758682101, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7944988012313843, + "num_tokens": 236322355.0, + "step": 6195 + }, + { + "epoch": 0.7881948861468007, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7945570945739746, + "num_tokens": 236354263.0, + "step": 6196 + }, + { + "epoch": 0.7883220964253912, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7942303419113159, + "num_tokens": 236391581.0, + "step": 6197 + }, + { + "epoch": 0.7884493067039817, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8040724992752075, + "num_tokens": 236426003.0, + "step": 6198 + }, + { + "epoch": 0.7885765169825721, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8083536624908447, + "num_tokens": 236464874.0, + "step": 6199 + }, + { + "epoch": 0.7887037272611627, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8067523837089539, + "num_tokens": 236499826.0, + "step": 6200 + }, + { + "epoch": 0.7888309375397532, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8105833530426025, + "num_tokens": 236538615.0, + "step": 6201 + }, + { + "epoch": 0.7889581478183437, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7986422777175903, + "num_tokens": 236577860.0, + "step": 6202 + }, + { + "epoch": 0.7890853580969343, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8150883316993713, + "num_tokens": 236619041.0, + "step": 6203 + }, + { + "epoch": 0.7892125683755248, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8011389970779419, + "num_tokens": 236655788.0, + "step": 6204 + }, + { + "epoch": 0.7893397786541152, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8135746717453003, + "num_tokens": 236691532.0, + "step": 6205 + }, + { + "epoch": 0.7894669889327057, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8100180625915527, + "num_tokens": 236733792.0, + "step": 6206 + }, + { + "epoch": 0.7895941992112963, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8134645819664001, + "num_tokens": 236775454.0, + "step": 6207 + }, + { + "epoch": 0.7897214094898868, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8043017387390137, + "num_tokens": 236811572.0, + "step": 6208 + }, + { + "epoch": 0.7898486197684773, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.787020742893219, + "num_tokens": 236851018.0, + "step": 6209 + }, + { + "epoch": 0.7899758300470678, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8066849112510681, + "num_tokens": 236891871.0, + "step": 6210 + }, + { + "epoch": 0.7901030403256584, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7974308729171753, + "num_tokens": 236926771.0, + "step": 6211 + }, + { + "epoch": 0.7902302506042488, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7818673849105835, + "num_tokens": 236964768.0, + "step": 6212 + }, + { + "epoch": 0.7903574608828393, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7827738523483276, + "num_tokens": 237005075.0, + "step": 6213 + }, + { + "epoch": 0.7904846711614298, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7956963777542114, + "num_tokens": 237035517.0, + "step": 6214 + }, + { + "epoch": 0.7906118814400204, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8144536018371582, + "num_tokens": 237076319.0, + "step": 6215 + }, + { + "epoch": 0.7907390917186109, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7869492173194885, + "num_tokens": 237117935.0, + "step": 6216 + }, + { + "epoch": 0.7908663019972014, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8063360452651978, + "num_tokens": 237155789.0, + "step": 6217 + }, + { + "epoch": 0.7909935122757918, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8062089681625366, + "num_tokens": 237198870.0, + "step": 6218 + }, + { + "epoch": 0.7911207225543824, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8054810166358948, + "num_tokens": 237243084.0, + "step": 6219 + }, + { + "epoch": 0.7912479328329729, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7842309474945068, + "num_tokens": 237277245.0, + "step": 6220 + }, + { + "epoch": 0.7913751431115634, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8030728101730347, + "num_tokens": 237317841.0, + "step": 6221 + }, + { + "epoch": 0.791502353390154, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7812249660491943, + "num_tokens": 237349327.0, + "step": 6222 + }, + { + "epoch": 0.7916295636687445, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7878188490867615, + "num_tokens": 237396179.0, + "step": 6223 + }, + { + "epoch": 0.7917567739473349, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7895458936691284, + "num_tokens": 237437766.0, + "step": 6224 + }, + { + "epoch": 0.7918839842259254, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8078404664993286, + "num_tokens": 237479239.0, + "step": 6225 + }, + { + "epoch": 0.792011194504516, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8184836506843567, + "num_tokens": 237514641.0, + "step": 6226 + }, + { + "epoch": 0.7921384047831065, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8091766834259033, + "num_tokens": 237555319.0, + "step": 6227 + }, + { + "epoch": 0.792265615061697, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7999080419540405, + "num_tokens": 237597786.0, + "step": 6228 + }, + { + "epoch": 0.7923928253402875, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8113834857940674, + "num_tokens": 237636584.0, + "step": 6229 + }, + { + "epoch": 0.792520035618878, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.790991485118866, + "num_tokens": 237676000.0, + "step": 6230 + }, + { + "epoch": 0.7926472458974685, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5808, + "mean_token_accuracy": 0.8295733332633972, + "num_tokens": 237713003.0, + "step": 6231 + }, + { + "epoch": 0.792774456176059, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7984614372253418, + "num_tokens": 237742054.0, + "step": 6232 + }, + { + "epoch": 0.7929016664546495, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7867529392242432, + "num_tokens": 237783147.0, + "step": 6233 + }, + { + "epoch": 0.7930288767332401, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7907875776290894, + "num_tokens": 237814322.0, + "step": 6234 + }, + { + "epoch": 0.7931560870118306, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8008370399475098, + "num_tokens": 237849278.0, + "step": 6235 + }, + { + "epoch": 0.793283297290421, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.808619499206543, + "num_tokens": 237893434.0, + "step": 6236 + }, + { + "epoch": 0.7934105075690115, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8073713779449463, + "num_tokens": 237932135.0, + "step": 6237 + }, + { + "epoch": 0.7935377178476021, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7932960391044617, + "num_tokens": 237966320.0, + "step": 6238 + }, + { + "epoch": 0.7936649281261926, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8029191493988037, + "num_tokens": 238006365.0, + "step": 6239 + }, + { + "epoch": 0.7937921384047831, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7857447266578674, + "num_tokens": 238039307.0, + "step": 6240 + }, + { + "epoch": 0.7939193486833737, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7865437865257263, + "num_tokens": 238078015.0, + "step": 6241 + }, + { + "epoch": 0.7940465589619641, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7956613302230835, + "num_tokens": 238116790.0, + "step": 6242 + }, + { + "epoch": 0.7941737692405546, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8070212602615356, + "num_tokens": 238155330.0, + "step": 6243 + }, + { + "epoch": 0.7943009795191451, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7822524905204773, + "num_tokens": 238194492.0, + "step": 6244 + }, + { + "epoch": 0.7944281897977357, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7927576899528503, + "num_tokens": 238230451.0, + "step": 6245 + }, + { + "epoch": 0.7945554000763262, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8064919710159302, + "num_tokens": 238266331.0, + "step": 6246 + }, + { + "epoch": 0.7946826103549167, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7860817909240723, + "num_tokens": 238307753.0, + "step": 6247 + }, + { + "epoch": 0.7948098206335071, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.801774799823761, + "num_tokens": 238349644.0, + "step": 6248 + }, + { + "epoch": 0.7949370309120977, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7803550958633423, + "num_tokens": 238386308.0, + "step": 6249 + }, + { + "epoch": 0.7950642411906882, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7885974645614624, + "num_tokens": 238427935.0, + "step": 6250 + }, + { + "epoch": 0.7951914514692787, + "grad_norm": 60.5, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8038398027420044, + "num_tokens": 238463639.0, + "step": 6251 + }, + { + "epoch": 0.7953186617478692, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8088802695274353, + "num_tokens": 238498295.0, + "step": 6252 + }, + { + "epoch": 0.7954458720264598, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7935405969619751, + "num_tokens": 238538109.0, + "step": 6253 + }, + { + "epoch": 0.7955730823050502, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7829681634902954, + "num_tokens": 238574415.0, + "step": 6254 + }, + { + "epoch": 0.7957002925836407, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.800955593585968, + "num_tokens": 238616368.0, + "step": 6255 + }, + { + "epoch": 0.7958275028622313, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.8021141886711121, + "num_tokens": 238650634.0, + "step": 6256 + }, + { + "epoch": 0.7959547131408218, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7925674915313721, + "num_tokens": 238684383.0, + "step": 6257 + }, + { + "epoch": 0.7960819234194123, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7923684120178223, + "num_tokens": 238722863.0, + "step": 6258 + }, + { + "epoch": 0.7962091336980028, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8016449213027954, + "num_tokens": 238765027.0, + "step": 6259 + }, + { + "epoch": 0.7963363439765934, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8063029646873474, + "num_tokens": 238801950.0, + "step": 6260 + }, + { + "epoch": 0.7964635542551838, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7904472351074219, + "num_tokens": 238840963.0, + "step": 6261 + }, + { + "epoch": 0.7965907645337743, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.781375527381897, + "num_tokens": 238876167.0, + "step": 6262 + }, + { + "epoch": 0.7967179748123648, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8047562837600708, + "num_tokens": 238910293.0, + "step": 6263 + }, + { + "epoch": 0.7968451850909554, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7916402816772461, + "num_tokens": 238947279.0, + "step": 6264 + }, + { + "epoch": 0.7969723953695459, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7904386520385742, + "num_tokens": 238984148.0, + "step": 6265 + }, + { + "epoch": 0.7970996056481364, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7815005779266357, + "num_tokens": 239022370.0, + "step": 6266 + }, + { + "epoch": 0.7972268159267268, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.8017934560775757, + "num_tokens": 239064542.0, + "step": 6267 + }, + { + "epoch": 0.7973540262053174, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7950171828269958, + "num_tokens": 239103193.0, + "step": 6268 + }, + { + "epoch": 0.7974812364839079, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8048216104507446, + "num_tokens": 239137952.0, + "step": 6269 + }, + { + "epoch": 0.7976084467624984, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7812414169311523, + "num_tokens": 239174600.0, + "step": 6270 + }, + { + "epoch": 0.797735657041089, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8045245409011841, + "num_tokens": 239206157.0, + "step": 6271 + }, + { + "epoch": 0.7978628673196795, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8089909553527832, + "num_tokens": 239243094.0, + "step": 6272 + }, + { + "epoch": 0.7979900775982699, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.796751856803894, + "num_tokens": 239281208.0, + "step": 6273 + }, + { + "epoch": 0.7981172878768604, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7932787537574768, + "num_tokens": 239320159.0, + "step": 6274 + }, + { + "epoch": 0.798244498155451, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7846531867980957, + "num_tokens": 239360160.0, + "step": 6275 + }, + { + "epoch": 0.7983717084340415, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8029600381851196, + "num_tokens": 239398101.0, + "step": 6276 + }, + { + "epoch": 0.798498918712632, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8150712251663208, + "num_tokens": 239438560.0, + "step": 6277 + }, + { + "epoch": 0.7986261289912225, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7885751724243164, + "num_tokens": 239475851.0, + "step": 6278 + }, + { + "epoch": 0.798753339269813, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8075320720672607, + "num_tokens": 239519165.0, + "step": 6279 + }, + { + "epoch": 0.7988805495484035, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7874659299850464, + "num_tokens": 239560602.0, + "step": 6280 + }, + { + "epoch": 0.799007759826994, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7916055917739868, + "num_tokens": 239591933.0, + "step": 6281 + }, + { + "epoch": 0.7991349701055845, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7943023443222046, + "num_tokens": 239625495.0, + "step": 6282 + }, + { + "epoch": 0.7992621803841751, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8157123327255249, + "num_tokens": 239658249.0, + "step": 6283 + }, + { + "epoch": 0.7993893906627656, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7908310294151306, + "num_tokens": 239690194.0, + "step": 6284 + }, + { + "epoch": 0.799516600941356, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7941646575927734, + "num_tokens": 239728315.0, + "step": 6285 + }, + { + "epoch": 0.7996438112199465, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8020463585853577, + "num_tokens": 239764505.0, + "step": 6286 + }, + { + "epoch": 0.7997710214985371, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7805742025375366, + "num_tokens": 239809461.0, + "step": 6287 + }, + { + "epoch": 0.7998982317771276, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7913047075271606, + "num_tokens": 239845232.0, + "step": 6288 + }, + { + "epoch": 0.8000254420557181, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.8000808954238892, + "num_tokens": 239885489.0, + "step": 6289 + }, + { + "epoch": 0.8001526523343087, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7819550037384033, + "num_tokens": 239921159.0, + "step": 6290 + }, + { + "epoch": 0.8002798626128991, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8018488883972168, + "num_tokens": 239960879.0, + "step": 6291 + }, + { + "epoch": 0.8004070728914896, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7891839742660522, + "num_tokens": 239994302.0, + "step": 6292 + }, + { + "epoch": 0.8005342831700801, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8017485737800598, + "num_tokens": 240027759.0, + "step": 6293 + }, + { + "epoch": 0.8006614934486707, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.796170175075531, + "num_tokens": 240065049.0, + "step": 6294 + }, + { + "epoch": 0.8007887037272612, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7829208374023438, + "num_tokens": 240101453.0, + "step": 6295 + }, + { + "epoch": 0.8009159140058517, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7747138738632202, + "num_tokens": 240136831.0, + "step": 6296 + }, + { + "epoch": 0.8010431242844421, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7992967963218689, + "num_tokens": 240173364.0, + "step": 6297 + }, + { + "epoch": 0.8011703345630327, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7759058475494385, + "num_tokens": 240206718.0, + "step": 6298 + }, + { + "epoch": 0.8012975448416232, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7845221161842346, + "num_tokens": 240245633.0, + "step": 6299 + }, + { + "epoch": 0.8014247551202137, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8112720847129822, + "num_tokens": 240283887.0, + "step": 6300 + }, + { + "epoch": 0.8015519653988042, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7905133962631226, + "num_tokens": 240322208.0, + "step": 6301 + }, + { + "epoch": 0.8016791756773948, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7946878671646118, + "num_tokens": 240356154.0, + "step": 6302 + }, + { + "epoch": 0.8018063859559852, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7913363575935364, + "num_tokens": 240389103.0, + "step": 6303 + }, + { + "epoch": 0.8019335962345757, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8168136477470398, + "num_tokens": 240429545.0, + "step": 6304 + }, + { + "epoch": 0.8020608065131662, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8148030042648315, + "num_tokens": 240468556.0, + "step": 6305 + }, + { + "epoch": 0.8021880167917568, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7955607175827026, + "num_tokens": 240508534.0, + "step": 6306 + }, + { + "epoch": 0.8023152270703473, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.794535756111145, + "num_tokens": 240540833.0, + "step": 6307 + }, + { + "epoch": 0.8024424373489378, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7859101295471191, + "num_tokens": 240577768.0, + "step": 6308 + }, + { + "epoch": 0.8025696476275284, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7920740246772766, + "num_tokens": 240610607.0, + "step": 6309 + }, + { + "epoch": 0.8026968579061188, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7915445566177368, + "num_tokens": 240646056.0, + "step": 6310 + }, + { + "epoch": 0.8028240681847093, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8044163584709167, + "num_tokens": 240682513.0, + "step": 6311 + }, + { + "epoch": 0.8029512784632998, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.802557110786438, + "num_tokens": 240716311.0, + "step": 6312 + }, + { + "epoch": 0.8030784887418904, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7866967916488647, + "num_tokens": 240758873.0, + "step": 6313 + }, + { + "epoch": 0.8032056990204809, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7987614870071411, + "num_tokens": 240797163.0, + "step": 6314 + }, + { + "epoch": 0.8033329092990714, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8138096928596497, + "num_tokens": 240835719.0, + "step": 6315 + }, + { + "epoch": 0.8034601195776618, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7900471687316895, + "num_tokens": 240874359.0, + "step": 6316 + }, + { + "epoch": 0.8035873298562524, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8026145696640015, + "num_tokens": 240909125.0, + "step": 6317 + }, + { + "epoch": 0.8037145401348429, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.801703929901123, + "num_tokens": 240949391.0, + "step": 6318 + }, + { + "epoch": 0.8038417504134334, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7895667552947998, + "num_tokens": 240987390.0, + "step": 6319 + }, + { + "epoch": 0.803968960692024, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7895045280456543, + "num_tokens": 241025359.0, + "step": 6320 + }, + { + "epoch": 0.8040961709706145, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7885289788246155, + "num_tokens": 241060532.0, + "step": 6321 + }, + { + "epoch": 0.8042233812492049, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8064593076705933, + "num_tokens": 241098547.0, + "step": 6322 + }, + { + "epoch": 0.8043505915277954, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7924597263336182, + "num_tokens": 241131091.0, + "step": 6323 + }, + { + "epoch": 0.804477801806386, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7874877452850342, + "num_tokens": 241164495.0, + "step": 6324 + }, + { + "epoch": 0.8046050120849765, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7930664420127869, + "num_tokens": 241203777.0, + "step": 6325 + }, + { + "epoch": 0.804732222363567, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7920662760734558, + "num_tokens": 241243219.0, + "step": 6326 + }, + { + "epoch": 0.8048594326421575, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.779807448387146, + "num_tokens": 241284547.0, + "step": 6327 + }, + { + "epoch": 0.804986642920748, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7764977216720581, + "num_tokens": 241331110.0, + "step": 6328 + }, + { + "epoch": 0.8051138531993385, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7925732135772705, + "num_tokens": 241372720.0, + "step": 6329 + }, + { + "epoch": 0.805241063477929, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7984965443611145, + "num_tokens": 241408173.0, + "step": 6330 + }, + { + "epoch": 0.8053682737565195, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8078866600990295, + "num_tokens": 241445021.0, + "step": 6331 + }, + { + "epoch": 0.8054954840351101, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8026244640350342, + "num_tokens": 241481874.0, + "step": 6332 + }, + { + "epoch": 0.8056226943137006, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7834736108779907, + "num_tokens": 241523140.0, + "step": 6333 + }, + { + "epoch": 0.805749904592291, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.773322343826294, + "num_tokens": 241563424.0, + "step": 6334 + }, + { + "epoch": 0.8058771148708815, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8087489604949951, + "num_tokens": 241600004.0, + "step": 6335 + }, + { + "epoch": 0.8060043251494721, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7956752181053162, + "num_tokens": 241642704.0, + "step": 6336 + }, + { + "epoch": 0.8061315354280626, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8064625263214111, + "num_tokens": 241679176.0, + "step": 6337 + }, + { + "epoch": 0.8062587457066531, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.802871823310852, + "num_tokens": 241721168.0, + "step": 6338 + }, + { + "epoch": 0.8063859559852437, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8099198341369629, + "num_tokens": 241759735.0, + "step": 6339 + }, + { + "epoch": 0.8065131662638341, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7993206977844238, + "num_tokens": 241799445.0, + "step": 6340 + }, + { + "epoch": 0.8066403765424246, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.8040599226951599, + "num_tokens": 241839333.0, + "step": 6341 + }, + { + "epoch": 0.8067675868210151, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7825167179107666, + "num_tokens": 241880248.0, + "step": 6342 + }, + { + "epoch": 0.8068947970996057, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8118671774864197, + "num_tokens": 241926717.0, + "step": 6343 + }, + { + "epoch": 0.8070220073781962, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7869392037391663, + "num_tokens": 241966080.0, + "step": 6344 + }, + { + "epoch": 0.8071492176567867, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.792904257774353, + "num_tokens": 242011975.0, + "step": 6345 + }, + { + "epoch": 0.8072764279353771, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8099571466445923, + "num_tokens": 242046320.0, + "step": 6346 + }, + { + "epoch": 0.8074036382139677, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.8005415201187134, + "num_tokens": 242085157.0, + "step": 6347 + }, + { + "epoch": 0.8075308484925582, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7874042391777039, + "num_tokens": 242116548.0, + "step": 6348 + }, + { + "epoch": 0.8076580587711487, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8074642419815063, + "num_tokens": 242151883.0, + "step": 6349 + }, + { + "epoch": 0.8077852690497392, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7969329357147217, + "num_tokens": 242190486.0, + "step": 6350 + }, + { + "epoch": 0.8079124793283298, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8023149371147156, + "num_tokens": 242232720.0, + "step": 6351 + }, + { + "epoch": 0.8080396896069202, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7964979410171509, + "num_tokens": 242273926.0, + "step": 6352 + }, + { + "epoch": 0.8081668998855107, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7861976623535156, + "num_tokens": 242313484.0, + "step": 6353 + }, + { + "epoch": 0.8082941101641012, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.792629599571228, + "num_tokens": 242352101.0, + "step": 6354 + }, + { + "epoch": 0.8084213204426918, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8205461502075195, + "num_tokens": 242392958.0, + "step": 6355 + }, + { + "epoch": 0.8085485307212823, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8023370504379272, + "num_tokens": 242432327.0, + "step": 6356 + }, + { + "epoch": 0.8086757409998728, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7730816602706909, + "num_tokens": 242473714.0, + "step": 6357 + }, + { + "epoch": 0.8088029512784632, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8079952597618103, + "num_tokens": 242513179.0, + "step": 6358 + }, + { + "epoch": 0.8089301615570538, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7974728345870972, + "num_tokens": 242549398.0, + "step": 6359 + }, + { + "epoch": 0.8090573718356443, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7871819734573364, + "num_tokens": 242583705.0, + "step": 6360 + }, + { + "epoch": 0.8091845821142348, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8055496215820312, + "num_tokens": 242625900.0, + "step": 6361 + }, + { + "epoch": 0.8093117923928254, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7856920957565308, + "num_tokens": 242665766.0, + "step": 6362 + }, + { + "epoch": 0.8094390026714159, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8018443584442139, + "num_tokens": 242707634.0, + "step": 6363 + }, + { + "epoch": 0.8095662129500064, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7958396673202515, + "num_tokens": 242746791.0, + "step": 6364 + }, + { + "epoch": 0.8096934232285968, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8117862939834595, + "num_tokens": 242784861.0, + "step": 6365 + }, + { + "epoch": 0.8098206335071874, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.807152271270752, + "num_tokens": 242817612.0, + "step": 6366 + }, + { + "epoch": 0.8099478437857779, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8039528131484985, + "num_tokens": 242854951.0, + "step": 6367 + }, + { + "epoch": 0.8100750540643684, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8165997266769409, + "num_tokens": 242894166.0, + "step": 6368 + }, + { + "epoch": 0.8102022643429589, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8043094873428345, + "num_tokens": 242936230.0, + "step": 6369 + }, + { + "epoch": 0.8103294746215495, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8097853064537048, + "num_tokens": 242973148.0, + "step": 6370 + }, + { + "epoch": 0.8104566849001399, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7990961074829102, + "num_tokens": 243013029.0, + "step": 6371 + }, + { + "epoch": 0.8105838951787304, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7939420938491821, + "num_tokens": 243054847.0, + "step": 6372 + }, + { + "epoch": 0.810711105457321, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7941412925720215, + "num_tokens": 243096524.0, + "step": 6373 + }, + { + "epoch": 0.8108383157359115, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7889773845672607, + "num_tokens": 243132397.0, + "step": 6374 + }, + { + "epoch": 0.810965526014502, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7815209627151489, + "num_tokens": 243171790.0, + "step": 6375 + }, + { + "epoch": 0.8110927362930925, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7980032563209534, + "num_tokens": 243208217.0, + "step": 6376 + }, + { + "epoch": 0.811219946571683, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8013981580734253, + "num_tokens": 243245260.0, + "step": 6377 + }, + { + "epoch": 0.8113471568502735, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.8035089373588562, + "num_tokens": 243283924.0, + "step": 6378 + }, + { + "epoch": 0.811474367128864, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8043326735496521, + "num_tokens": 243321977.0, + "step": 6379 + }, + { + "epoch": 0.8116015774074545, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.793063759803772, + "num_tokens": 243359349.0, + "step": 6380 + }, + { + "epoch": 0.8117287876860451, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8016330003738403, + "num_tokens": 243395722.0, + "step": 6381 + }, + { + "epoch": 0.8118559979646356, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7822388410568237, + "num_tokens": 243429022.0, + "step": 6382 + }, + { + "epoch": 0.811983208243226, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.779329776763916, + "num_tokens": 243468474.0, + "step": 6383 + }, + { + "epoch": 0.8121104185218165, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8088862895965576, + "num_tokens": 243503045.0, + "step": 6384 + }, + { + "epoch": 0.8122376288004071, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7914770245552063, + "num_tokens": 243543865.0, + "step": 6385 + }, + { + "epoch": 0.8123648390789976, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.791000485420227, + "num_tokens": 243581258.0, + "step": 6386 + }, + { + "epoch": 0.8124920493575881, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7921371459960938, + "num_tokens": 243622777.0, + "step": 6387 + }, + { + "epoch": 0.8126192596361786, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7858577370643616, + "num_tokens": 243664549.0, + "step": 6388 + }, + { + "epoch": 0.8127464699147691, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7962316870689392, + "num_tokens": 243710408.0, + "step": 6389 + }, + { + "epoch": 0.8128736801933596, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.8034977912902832, + "num_tokens": 243752598.0, + "step": 6390 + }, + { + "epoch": 0.8130008904719501, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7969680428504944, + "num_tokens": 243789805.0, + "step": 6391 + }, + { + "epoch": 0.8131281007505406, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8071075081825256, + "num_tokens": 243824124.0, + "step": 6392 + }, + { + "epoch": 0.8132553110291312, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7982860207557678, + "num_tokens": 243863448.0, + "step": 6393 + }, + { + "epoch": 0.8133825213077217, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7995672821998596, + "num_tokens": 243902085.0, + "step": 6394 + }, + { + "epoch": 0.8135097315863121, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7967873811721802, + "num_tokens": 243945349.0, + "step": 6395 + }, + { + "epoch": 0.8136369418649027, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7960010170936584, + "num_tokens": 243982221.0, + "step": 6396 + }, + { + "epoch": 0.8137641521434932, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7818015217781067, + "num_tokens": 244012576.0, + "step": 6397 + }, + { + "epoch": 0.8138913624220837, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8094677329063416, + "num_tokens": 244052089.0, + "step": 6398 + }, + { + "epoch": 0.8140185727006742, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7984001636505127, + "num_tokens": 244088712.0, + "step": 6399 + }, + { + "epoch": 0.8141457829792648, + "grad_norm": 64.5, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8011602163314819, + "num_tokens": 244127655.0, + "step": 6400 + }, + { + "epoch": 0.8142729932578552, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7972991466522217, + "num_tokens": 244158588.0, + "step": 6401 + }, + { + "epoch": 0.8144002035364457, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6032, + "mean_token_accuracy": 0.8143275380134583, + "num_tokens": 244193872.0, + "step": 6402 + }, + { + "epoch": 0.8145274138150362, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7805818319320679, + "num_tokens": 244234055.0, + "step": 6403 + }, + { + "epoch": 0.8146546240936268, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.797444224357605, + "num_tokens": 244275163.0, + "step": 6404 + }, + { + "epoch": 0.8147818343722173, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7936331629753113, + "num_tokens": 244315241.0, + "step": 6405 + }, + { + "epoch": 0.8149090446508078, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8091342449188232, + "num_tokens": 244351947.0, + "step": 6406 + }, + { + "epoch": 0.8150362549293982, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7966586947441101, + "num_tokens": 244387706.0, + "step": 6407 + }, + { + "epoch": 0.8151634652079888, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8051198720932007, + "num_tokens": 244428447.0, + "step": 6408 + }, + { + "epoch": 0.8152906754865793, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7833040952682495, + "num_tokens": 244471191.0, + "step": 6409 + }, + { + "epoch": 0.8154178857651698, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7765483856201172, + "num_tokens": 244509198.0, + "step": 6410 + }, + { + "epoch": 0.8155450960437604, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.787218451499939, + "num_tokens": 244549677.0, + "step": 6411 + }, + { + "epoch": 0.8156723063223509, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7777906656265259, + "num_tokens": 244583586.0, + "step": 6412 + }, + { + "epoch": 0.8157995166009414, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.806442141532898, + "num_tokens": 244621220.0, + "step": 6413 + }, + { + "epoch": 0.8159267268795318, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8054725527763367, + "num_tokens": 244655941.0, + "step": 6414 + }, + { + "epoch": 0.8160539371581224, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8091703057289124, + "num_tokens": 244685807.0, + "step": 6415 + }, + { + "epoch": 0.8161811474367129, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7955648899078369, + "num_tokens": 244726866.0, + "step": 6416 + }, + { + "epoch": 0.8163083577153034, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7874433994293213, + "num_tokens": 244767273.0, + "step": 6417 + }, + { + "epoch": 0.8164355679938939, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7875443696975708, + "num_tokens": 244806111.0, + "step": 6418 + }, + { + "epoch": 0.8165627782724845, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.804773211479187, + "num_tokens": 244841706.0, + "step": 6419 + }, + { + "epoch": 0.8166899885510749, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7986178994178772, + "num_tokens": 244883908.0, + "step": 6420 + }, + { + "epoch": 0.8168171988296654, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.785102128982544, + "num_tokens": 244919090.0, + "step": 6421 + }, + { + "epoch": 0.8169444091082559, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8142259120941162, + "num_tokens": 244961141.0, + "step": 6422 + }, + { + "epoch": 0.8170716193868465, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7803884148597717, + "num_tokens": 245004503.0, + "step": 6423 + }, + { + "epoch": 0.817198829665437, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7950115203857422, + "num_tokens": 245045042.0, + "step": 6424 + }, + { + "epoch": 0.8173260399440275, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7931420803070068, + "num_tokens": 245085416.0, + "step": 6425 + }, + { + "epoch": 0.8174532502226179, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7836010456085205, + "num_tokens": 245116137.0, + "step": 6426 + }, + { + "epoch": 0.8175804605012085, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8098733425140381, + "num_tokens": 245159181.0, + "step": 6427 + }, + { + "epoch": 0.817707670779799, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.791681170463562, + "num_tokens": 245205199.0, + "step": 6428 + }, + { + "epoch": 0.8178348810583895, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8019590377807617, + "num_tokens": 245242455.0, + "step": 6429 + }, + { + "epoch": 0.8179620913369801, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8139039278030396, + "num_tokens": 245280675.0, + "step": 6430 + }, + { + "epoch": 0.8180893016155706, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8125244379043579, + "num_tokens": 245316947.0, + "step": 6431 + }, + { + "epoch": 0.818216511894161, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7996333837509155, + "num_tokens": 245355145.0, + "step": 6432 + }, + { + "epoch": 0.8183437221727515, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8067957162857056, + "num_tokens": 245390282.0, + "step": 6433 + }, + { + "epoch": 0.8184709324513421, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8092356324195862, + "num_tokens": 245430262.0, + "step": 6434 + }, + { + "epoch": 0.8185981427299326, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7920383810997009, + "num_tokens": 245470874.0, + "step": 6435 + }, + { + "epoch": 0.8187253530085231, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8047153949737549, + "num_tokens": 245506626.0, + "step": 6436 + }, + { + "epoch": 0.8188525632871136, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7791576981544495, + "num_tokens": 245546794.0, + "step": 6437 + }, + { + "epoch": 0.8189797735657041, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7984520196914673, + "num_tokens": 245589544.0, + "step": 6438 + }, + { + "epoch": 0.8191069838442946, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8094735145568848, + "num_tokens": 245628889.0, + "step": 6439 + }, + { + "epoch": 0.8192341941228851, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7952533960342407, + "num_tokens": 245674615.0, + "step": 6440 + }, + { + "epoch": 0.8193614044014756, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7915810346603394, + "num_tokens": 245713966.0, + "step": 6441 + }, + { + "epoch": 0.8194886146800662, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8017820715904236, + "num_tokens": 245750035.0, + "step": 6442 + }, + { + "epoch": 0.8196158249586567, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8129421472549438, + "num_tokens": 245788506.0, + "step": 6443 + }, + { + "epoch": 0.8197430352372471, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.802337646484375, + "num_tokens": 245826547.0, + "step": 6444 + }, + { + "epoch": 0.8198702455158376, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7927086353302002, + "num_tokens": 245874363.0, + "step": 6445 + }, + { + "epoch": 0.8199974557944282, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7773054242134094, + "num_tokens": 245917813.0, + "step": 6446 + }, + { + "epoch": 0.8201246660730187, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7924008369445801, + "num_tokens": 245952177.0, + "step": 6447 + }, + { + "epoch": 0.8202518763516092, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.8000770807266235, + "num_tokens": 245992897.0, + "step": 6448 + }, + { + "epoch": 0.8203790866301998, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8088687062263489, + "num_tokens": 246029657.0, + "step": 6449 + }, + { + "epoch": 0.8205062969087902, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7888048887252808, + "num_tokens": 246072425.0, + "step": 6450 + }, + { + "epoch": 0.8206335071873807, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8163946866989136, + "num_tokens": 246113131.0, + "step": 6451 + }, + { + "epoch": 0.8207607174659712, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8005194664001465, + "num_tokens": 246154548.0, + "step": 6452 + }, + { + "epoch": 0.8208879277445618, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7939013838768005, + "num_tokens": 246189518.0, + "step": 6453 + }, + { + "epoch": 0.8210151380231523, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7928929328918457, + "num_tokens": 246235248.0, + "step": 6454 + }, + { + "epoch": 0.8211423483017428, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7895363569259644, + "num_tokens": 246277016.0, + "step": 6455 + }, + { + "epoch": 0.8212695585803332, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.79057776927948, + "num_tokens": 246312352.0, + "step": 6456 + }, + { + "epoch": 0.8213967688589238, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.782947301864624, + "num_tokens": 246355079.0, + "step": 6457 + }, + { + "epoch": 0.8215239791375143, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7804854512214661, + "num_tokens": 246391073.0, + "step": 6458 + }, + { + "epoch": 0.8216511894161048, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.803127646446228, + "num_tokens": 246435407.0, + "step": 6459 + }, + { + "epoch": 0.8217783996946953, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7854846119880676, + "num_tokens": 246475303.0, + "step": 6460 + }, + { + "epoch": 0.8219056099732859, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7937319278717041, + "num_tokens": 246515237.0, + "step": 6461 + }, + { + "epoch": 0.8220328202518764, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7914678454399109, + "num_tokens": 246554958.0, + "step": 6462 + }, + { + "epoch": 0.8221600305304668, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.799053966999054, + "num_tokens": 246592327.0, + "step": 6463 + }, + { + "epoch": 0.8222872408090574, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8040065169334412, + "num_tokens": 246626965.0, + "step": 6464 + }, + { + "epoch": 0.8224144510876479, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7942785024642944, + "num_tokens": 246662049.0, + "step": 6465 + }, + { + "epoch": 0.8225416613662384, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7944208979606628, + "num_tokens": 246696989.0, + "step": 6466 + }, + { + "epoch": 0.8226688716448289, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.8017756938934326, + "num_tokens": 246738418.0, + "step": 6467 + }, + { + "epoch": 0.8227960819234195, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7960383296012878, + "num_tokens": 246776779.0, + "step": 6468 + }, + { + "epoch": 0.8229232922020099, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8035421371459961, + "num_tokens": 246814029.0, + "step": 6469 + }, + { + "epoch": 0.8230505024806004, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7905469536781311, + "num_tokens": 246856390.0, + "step": 6470 + }, + { + "epoch": 0.8231777127591909, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7976382970809937, + "num_tokens": 246895563.0, + "step": 6471 + }, + { + "epoch": 0.8233049230377815, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7941001057624817, + "num_tokens": 246935956.0, + "step": 6472 + }, + { + "epoch": 0.823432133316372, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7962292432785034, + "num_tokens": 246984765.0, + "step": 6473 + }, + { + "epoch": 0.8235593435949625, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7951732873916626, + "num_tokens": 247017086.0, + "step": 6474 + }, + { + "epoch": 0.8236865538735529, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8033638000488281, + "num_tokens": 247056195.0, + "step": 6475 + }, + { + "epoch": 0.8238137641521435, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8104602098464966, + "num_tokens": 247094063.0, + "step": 6476 + }, + { + "epoch": 0.823940974430734, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7922828197479248, + "num_tokens": 247131743.0, + "step": 6477 + }, + { + "epoch": 0.8240681847093245, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7963651418685913, + "num_tokens": 247169477.0, + "step": 6478 + }, + { + "epoch": 0.824195394987915, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8042893409729004, + "num_tokens": 247209159.0, + "step": 6479 + }, + { + "epoch": 0.8243226052665056, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8039740920066833, + "num_tokens": 247247603.0, + "step": 6480 + }, + { + "epoch": 0.824449815545096, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8082640767097473, + "num_tokens": 247290184.0, + "step": 6481 + }, + { + "epoch": 0.8245770258236865, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.78294438123703, + "num_tokens": 247331862.0, + "step": 6482 + }, + { + "epoch": 0.824704236102277, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7950683832168579, + "num_tokens": 247368271.0, + "step": 6483 + }, + { + "epoch": 0.8248314463808676, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7962639331817627, + "num_tokens": 247408077.0, + "step": 6484 + }, + { + "epoch": 0.8249586566594581, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8076062798500061, + "num_tokens": 247447408.0, + "step": 6485 + }, + { + "epoch": 0.8250858669380486, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8122365474700928, + "num_tokens": 247489863.0, + "step": 6486 + }, + { + "epoch": 0.8252130772166391, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7963595390319824, + "num_tokens": 247532233.0, + "step": 6487 + }, + { + "epoch": 0.8253402874952296, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7677094340324402, + "num_tokens": 247570719.0, + "step": 6488 + }, + { + "epoch": 0.8254674977738201, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.798628568649292, + "num_tokens": 247603302.0, + "step": 6489 + }, + { + "epoch": 0.8255947080524106, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8004763126373291, + "num_tokens": 247642837.0, + "step": 6490 + }, + { + "epoch": 0.8257219183310012, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7964515089988708, + "num_tokens": 247685246.0, + "step": 6491 + }, + { + "epoch": 0.8258491286095917, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.788829505443573, + "num_tokens": 247720871.0, + "step": 6492 + }, + { + "epoch": 0.8259763388881821, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8016068935394287, + "num_tokens": 247759796.0, + "step": 6493 + }, + { + "epoch": 0.8261035491667726, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7773503065109253, + "num_tokens": 247806486.0, + "step": 6494 + }, + { + "epoch": 0.8262307594453632, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.808081865310669, + "num_tokens": 247842031.0, + "step": 6495 + }, + { + "epoch": 0.8263579697239537, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7987203598022461, + "num_tokens": 247883438.0, + "step": 6496 + }, + { + "epoch": 0.8264851800025442, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7994338274002075, + "num_tokens": 247921491.0, + "step": 6497 + }, + { + "epoch": 0.8266123902811348, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8077688217163086, + "num_tokens": 247957454.0, + "step": 6498 + }, + { + "epoch": 0.8267396005597252, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7795787453651428, + "num_tokens": 247997705.0, + "step": 6499 + }, + { + "epoch": 0.8268668108383157, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7898589372634888, + "num_tokens": 248034536.0, + "step": 6500 + }, + { + "epoch": 0.8269940211169062, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7948508858680725, + "num_tokens": 248080916.0, + "step": 6501 + }, + { + "epoch": 0.8271212313954968, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7942930459976196, + "num_tokens": 248119719.0, + "step": 6502 + }, + { + "epoch": 0.8272484416740873, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7956691980361938, + "num_tokens": 248156221.0, + "step": 6503 + }, + { + "epoch": 0.8273756519526778, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7984251379966736, + "num_tokens": 248191780.0, + "step": 6504 + }, + { + "epoch": 0.8275028622312682, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7925100922584534, + "num_tokens": 248228594.0, + "step": 6505 + }, + { + "epoch": 0.8276300725098588, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8023233413696289, + "num_tokens": 248260372.0, + "step": 6506 + }, + { + "epoch": 0.8277572827884493, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7879143953323364, + "num_tokens": 248299361.0, + "step": 6507 + }, + { + "epoch": 0.8278844930670398, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7867894172668457, + "num_tokens": 248337370.0, + "step": 6508 + }, + { + "epoch": 0.8280117033456303, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7985175251960754, + "num_tokens": 248377541.0, + "step": 6509 + }, + { + "epoch": 0.8281389136242209, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7790147066116333, + "num_tokens": 248413928.0, + "step": 6510 + }, + { + "epoch": 0.8282661239028114, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7987903952598572, + "num_tokens": 248453283.0, + "step": 6511 + }, + { + "epoch": 0.8283933341814018, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8069037199020386, + "num_tokens": 248490286.0, + "step": 6512 + }, + { + "epoch": 0.8285205444599923, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7814135551452637, + "num_tokens": 248526349.0, + "step": 6513 + }, + { + "epoch": 0.8286477547385829, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.782684862613678, + "num_tokens": 248561751.0, + "step": 6514 + }, + { + "epoch": 0.8287749650171734, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8054834604263306, + "num_tokens": 248601921.0, + "step": 6515 + }, + { + "epoch": 0.8289021752957639, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8088697195053101, + "num_tokens": 248641218.0, + "step": 6516 + }, + { + "epoch": 0.8290293855743545, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.794083833694458, + "num_tokens": 248672616.0, + "step": 6517 + }, + { + "epoch": 0.8291565958529449, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7981964945793152, + "num_tokens": 248708212.0, + "step": 6518 + }, + { + "epoch": 0.8292838061315354, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8016564846038818, + "num_tokens": 248746575.0, + "step": 6519 + }, + { + "epoch": 0.8294110164101259, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7924736738204956, + "num_tokens": 248789423.0, + "step": 6520 + }, + { + "epoch": 0.8295382266887165, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8034893870353699, + "num_tokens": 248824044.0, + "step": 6521 + }, + { + "epoch": 0.829665436967307, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.8006930351257324, + "num_tokens": 248860805.0, + "step": 6522 + }, + { + "epoch": 0.8297926472458975, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8077130317687988, + "num_tokens": 248898958.0, + "step": 6523 + }, + { + "epoch": 0.8299198575244879, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7977811694145203, + "num_tokens": 248938847.0, + "step": 6524 + }, + { + "epoch": 0.8300470678030785, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7980026006698608, + "num_tokens": 248977944.0, + "step": 6525 + }, + { + "epoch": 0.830174278081669, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7849644422531128, + "num_tokens": 249014610.0, + "step": 6526 + }, + { + "epoch": 0.8303014883602595, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.762649655342102, + "num_tokens": 249053869.0, + "step": 6527 + }, + { + "epoch": 0.83042869863885, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7925043702125549, + "num_tokens": 249096874.0, + "step": 6528 + }, + { + "epoch": 0.8305559089174406, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7919406890869141, + "num_tokens": 249133141.0, + "step": 6529 + }, + { + "epoch": 0.830683119196031, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8009852170944214, + "num_tokens": 249163813.0, + "step": 6530 + }, + { + "epoch": 0.8308103294746215, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7924555540084839, + "num_tokens": 249202935.0, + "step": 6531 + }, + { + "epoch": 0.830937539753212, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8040701150894165, + "num_tokens": 249241061.0, + "step": 6532 + }, + { + "epoch": 0.8310647500318026, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7826245427131653, + "num_tokens": 249277863.0, + "step": 6533 + }, + { + "epoch": 0.8311919603103931, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7942441701889038, + "num_tokens": 249318047.0, + "step": 6534 + }, + { + "epoch": 0.8313191705889836, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7955989837646484, + "num_tokens": 249362095.0, + "step": 6535 + }, + { + "epoch": 0.831446380867574, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8131242990493774, + "num_tokens": 249403362.0, + "step": 6536 + }, + { + "epoch": 0.8315735911461646, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8004037141799927, + "num_tokens": 249438051.0, + "step": 6537 + }, + { + "epoch": 0.8317008014247551, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8025590181350708, + "num_tokens": 249477185.0, + "step": 6538 + }, + { + "epoch": 0.8318280117033456, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7884186506271362, + "num_tokens": 249516880.0, + "step": 6539 + }, + { + "epoch": 0.8319552219819362, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8086094856262207, + "num_tokens": 249555864.0, + "step": 6540 + }, + { + "epoch": 0.8320824322605267, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7946945428848267, + "num_tokens": 249590566.0, + "step": 6541 + }, + { + "epoch": 0.8322096425391171, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8128608465194702, + "num_tokens": 249631523.0, + "step": 6542 + }, + { + "epoch": 0.8323368528177076, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8080062866210938, + "num_tokens": 249674113.0, + "step": 6543 + }, + { + "epoch": 0.8324640630962982, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8100380897521973, + "num_tokens": 249712934.0, + "step": 6544 + }, + { + "epoch": 0.8325912733748887, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8010623455047607, + "num_tokens": 249750295.0, + "step": 6545 + }, + { + "epoch": 0.8327184836534792, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8035774230957031, + "num_tokens": 249790289.0, + "step": 6546 + }, + { + "epoch": 0.8328456939320698, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8084455728530884, + "num_tokens": 249822873.0, + "step": 6547 + }, + { + "epoch": 0.8329729042106602, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.8006306886672974, + "num_tokens": 249860564.0, + "step": 6548 + }, + { + "epoch": 0.8331001144892507, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.8001016974449158, + "num_tokens": 249901448.0, + "step": 6549 + }, + { + "epoch": 0.8332273247678412, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7836716175079346, + "num_tokens": 249933524.0, + "step": 6550 + }, + { + "epoch": 0.8333545350464318, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7931047677993774, + "num_tokens": 249971856.0, + "step": 6551 + }, + { + "epoch": 0.8334817453250223, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7917556762695312, + "num_tokens": 250013940.0, + "step": 6552 + }, + { + "epoch": 0.8336089556036128, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7801342010498047, + "num_tokens": 250044826.0, + "step": 6553 + }, + { + "epoch": 0.8337361658822032, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7779006361961365, + "num_tokens": 250087409.0, + "step": 6554 + }, + { + "epoch": 0.8338633761607938, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8031488656997681, + "num_tokens": 250126236.0, + "step": 6555 + }, + { + "epoch": 0.8339905864393843, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8081209659576416, + "num_tokens": 250163559.0, + "step": 6556 + }, + { + "epoch": 0.8341177967179748, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7946786880493164, + "num_tokens": 250203536.0, + "step": 6557 + }, + { + "epoch": 0.8342450069965653, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.798259437084198, + "num_tokens": 250237455.0, + "step": 6558 + }, + { + "epoch": 0.8343722172751559, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8070195913314819, + "num_tokens": 250274559.0, + "step": 6559 + }, + { + "epoch": 0.8344994275537464, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7959581613540649, + "num_tokens": 250309392.0, + "step": 6560 + }, + { + "epoch": 0.8346266378323368, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8049686551094055, + "num_tokens": 250352010.0, + "step": 6561 + }, + { + "epoch": 0.8347538481109273, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7798487544059753, + "num_tokens": 250385745.0, + "step": 6562 + }, + { + "epoch": 0.8348810583895179, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.798648476600647, + "num_tokens": 250426055.0, + "step": 6563 + }, + { + "epoch": 0.8350082686681084, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7940266132354736, + "num_tokens": 250468795.0, + "step": 6564 + }, + { + "epoch": 0.8351354789466989, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7949509620666504, + "num_tokens": 250509115.0, + "step": 6565 + }, + { + "epoch": 0.8352626892252895, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7949575185775757, + "num_tokens": 250547770.0, + "step": 6566 + }, + { + "epoch": 0.8353898995038799, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7930601835250854, + "num_tokens": 250589737.0, + "step": 6567 + }, + { + "epoch": 0.8355171097824704, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8009248971939087, + "num_tokens": 250623058.0, + "step": 6568 + }, + { + "epoch": 0.8356443200610609, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7966187596321106, + "num_tokens": 250660526.0, + "step": 6569 + }, + { + "epoch": 0.8357715303396515, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7816755771636963, + "num_tokens": 250698218.0, + "step": 6570 + }, + { + "epoch": 0.835898740618242, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7913419008255005, + "num_tokens": 250743410.0, + "step": 6571 + }, + { + "epoch": 0.8360259508968325, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8054366111755371, + "num_tokens": 250781919.0, + "step": 6572 + }, + { + "epoch": 0.8361531611754229, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.79966801404953, + "num_tokens": 250817051.0, + "step": 6573 + }, + { + "epoch": 0.8362803714540135, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7955154180526733, + "num_tokens": 250857241.0, + "step": 6574 + }, + { + "epoch": 0.836407581732604, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.8007245659828186, + "num_tokens": 250903591.0, + "step": 6575 + }, + { + "epoch": 0.8365347920111945, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7862597703933716, + "num_tokens": 250945788.0, + "step": 6576 + }, + { + "epoch": 0.836662002289785, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7953054904937744, + "num_tokens": 250984884.0, + "step": 6577 + }, + { + "epoch": 0.8367892125683756, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8125448822975159, + "num_tokens": 251024347.0, + "step": 6578 + }, + { + "epoch": 0.836916422846966, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8240383863449097, + "num_tokens": 251068527.0, + "step": 6579 + }, + { + "epoch": 0.8370436331255565, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8030415773391724, + "num_tokens": 251109914.0, + "step": 6580 + }, + { + "epoch": 0.837170843404147, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7872387170791626, + "num_tokens": 251144760.0, + "step": 6581 + }, + { + "epoch": 0.8372980536827376, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7959154844284058, + "num_tokens": 251184456.0, + "step": 6582 + }, + { + "epoch": 0.8374252639613281, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7899277210235596, + "num_tokens": 251229493.0, + "step": 6583 + }, + { + "epoch": 0.8375524742399186, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8008301258087158, + "num_tokens": 251270886.0, + "step": 6584 + }, + { + "epoch": 0.837679684518509, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8073229789733887, + "num_tokens": 251304833.0, + "step": 6585 + }, + { + "epoch": 0.8378068947970996, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8048402667045593, + "num_tokens": 251334196.0, + "step": 6586 + }, + { + "epoch": 0.8379341050756901, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.784919261932373, + "num_tokens": 251380278.0, + "step": 6587 + }, + { + "epoch": 0.8380613153542806, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7883346080780029, + "num_tokens": 251418432.0, + "step": 6588 + }, + { + "epoch": 0.8381885256328712, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7919825315475464, + "num_tokens": 251453747.0, + "step": 6589 + }, + { + "epoch": 0.8383157359114617, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7946720123291016, + "num_tokens": 251492439.0, + "step": 6590 + }, + { + "epoch": 0.8384429461900521, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.794215738773346, + "num_tokens": 251527945.0, + "step": 6591 + }, + { + "epoch": 0.8385701564686426, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8052202463150024, + "num_tokens": 251567161.0, + "step": 6592 + }, + { + "epoch": 0.8386973667472332, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7957556843757629, + "num_tokens": 251601189.0, + "step": 6593 + }, + { + "epoch": 0.8388245770258237, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7823962569236755, + "num_tokens": 251630846.0, + "step": 6594 + }, + { + "epoch": 0.8389517873044142, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8089491724967957, + "num_tokens": 251665701.0, + "step": 6595 + }, + { + "epoch": 0.8390789975830047, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7883012294769287, + "num_tokens": 251699909.0, + "step": 6596 + }, + { + "epoch": 0.8392062078615952, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7940106987953186, + "num_tokens": 251734052.0, + "step": 6597 + }, + { + "epoch": 0.8393334181401857, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7828978300094604, + "num_tokens": 251772518.0, + "step": 6598 + }, + { + "epoch": 0.8394606284187762, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7946159839630127, + "num_tokens": 251816378.0, + "step": 6599 + }, + { + "epoch": 0.8395878386973668, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7955647706985474, + "num_tokens": 251851264.0, + "step": 6600 + }, + { + "epoch": 0.8397150489759573, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8007709980010986, + "num_tokens": 251888906.0, + "step": 6601 + }, + { + "epoch": 0.8398422592545478, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8054440021514893, + "num_tokens": 251930641.0, + "step": 6602 + }, + { + "epoch": 0.8399694695331382, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7789990305900574, + "num_tokens": 251968929.0, + "step": 6603 + }, + { + "epoch": 0.8400966798117288, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7892752885818481, + "num_tokens": 252006751.0, + "step": 6604 + }, + { + "epoch": 0.8402238900903193, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7919038534164429, + "num_tokens": 252045861.0, + "step": 6605 + }, + { + "epoch": 0.8403511003689098, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7873637676239014, + "num_tokens": 252084412.0, + "step": 6606 + }, + { + "epoch": 0.8404783106475003, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8034459352493286, + "num_tokens": 252120576.0, + "step": 6607 + }, + { + "epoch": 0.8406055209260909, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7975836992263794, + "num_tokens": 252161636.0, + "step": 6608 + }, + { + "epoch": 0.8407327312046813, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.796938955783844, + "num_tokens": 252199669.0, + "step": 6609 + }, + { + "epoch": 0.8408599414832718, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8081420660018921, + "num_tokens": 252236832.0, + "step": 6610 + }, + { + "epoch": 0.8409871517618623, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7954167127609253, + "num_tokens": 252282518.0, + "step": 6611 + }, + { + "epoch": 0.8411143620404529, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7903507947921753, + "num_tokens": 252316479.0, + "step": 6612 + }, + { + "epoch": 0.8412415723190434, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7865061163902283, + "num_tokens": 252353789.0, + "step": 6613 + }, + { + "epoch": 0.8413687825976339, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.801434338092804, + "num_tokens": 252391609.0, + "step": 6614 + }, + { + "epoch": 0.8414959928762245, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8128687143325806, + "num_tokens": 252433514.0, + "step": 6615 + }, + { + "epoch": 0.8416232031548149, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7994793653488159, + "num_tokens": 252469700.0, + "step": 6616 + }, + { + "epoch": 0.8417504134334054, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7955554723739624, + "num_tokens": 252501658.0, + "step": 6617 + }, + { + "epoch": 0.8418776237119959, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7996057868003845, + "num_tokens": 252534546.0, + "step": 6618 + }, + { + "epoch": 0.8420048339905865, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8063753843307495, + "num_tokens": 252573172.0, + "step": 6619 + }, + { + "epoch": 0.842132044269177, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8099088668823242, + "num_tokens": 252607258.0, + "step": 6620 + }, + { + "epoch": 0.8422592545477675, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8140991926193237, + "num_tokens": 252651451.0, + "step": 6621 + }, + { + "epoch": 0.8423864648263579, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8118449449539185, + "num_tokens": 252686348.0, + "step": 6622 + }, + { + "epoch": 0.8425136751049485, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7919650673866272, + "num_tokens": 252724316.0, + "step": 6623 + }, + { + "epoch": 0.842640885383539, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7864881157875061, + "num_tokens": 252764612.0, + "step": 6624 + }, + { + "epoch": 0.8427680956621295, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7793376445770264, + "num_tokens": 252809510.0, + "step": 6625 + }, + { + "epoch": 0.84289530594072, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7907103300094604, + "num_tokens": 252847598.0, + "step": 6626 + }, + { + "epoch": 0.8430225162193106, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7776817679405212, + "num_tokens": 252884401.0, + "step": 6627 + }, + { + "epoch": 0.843149726497901, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7778971195220947, + "num_tokens": 252928348.0, + "step": 6628 + }, + { + "epoch": 0.8432769367764915, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7950575947761536, + "num_tokens": 252965084.0, + "step": 6629 + }, + { + "epoch": 0.843404147055082, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8062132596969604, + "num_tokens": 253003697.0, + "step": 6630 + }, + { + "epoch": 0.8435313573336726, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7791745066642761, + "num_tokens": 253037239.0, + "step": 6631 + }, + { + "epoch": 0.8436585676122631, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7902498245239258, + "num_tokens": 253078163.0, + "step": 6632 + }, + { + "epoch": 0.8437857778908536, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7936679720878601, + "num_tokens": 253118907.0, + "step": 6633 + }, + { + "epoch": 0.843912988169444, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8150218725204468, + "num_tokens": 253153132.0, + "step": 6634 + }, + { + "epoch": 0.8440401984480346, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7939984798431396, + "num_tokens": 253194446.0, + "step": 6635 + }, + { + "epoch": 0.8441674087266251, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7982040643692017, + "num_tokens": 253230798.0, + "step": 6636 + }, + { + "epoch": 0.8442946190052156, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7864042520523071, + "num_tokens": 253269029.0, + "step": 6637 + }, + { + "epoch": 0.8444218292838062, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7977516651153564, + "num_tokens": 253310656.0, + "step": 6638 + }, + { + "epoch": 0.8445490395623967, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.818315863609314, + "num_tokens": 253346882.0, + "step": 6639 + }, + { + "epoch": 0.8446762498409871, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7866907119750977, + "num_tokens": 253389601.0, + "step": 6640 + }, + { + "epoch": 0.8448034601195776, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7952255606651306, + "num_tokens": 253428725.0, + "step": 6641 + }, + { + "epoch": 0.8449306703981682, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8099420666694641, + "num_tokens": 253461404.0, + "step": 6642 + }, + { + "epoch": 0.8450578806767587, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7997456789016724, + "num_tokens": 253500324.0, + "step": 6643 + }, + { + "epoch": 0.8451850909553492, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7823845744132996, + "num_tokens": 253542500.0, + "step": 6644 + }, + { + "epoch": 0.8453123012339397, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7778826951980591, + "num_tokens": 253581887.0, + "step": 6645 + }, + { + "epoch": 0.8454395115125302, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8061936497688293, + "num_tokens": 253616942.0, + "step": 6646 + }, + { + "epoch": 0.8455667217911207, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7877181768417358, + "num_tokens": 253652610.0, + "step": 6647 + }, + { + "epoch": 0.8456939320697112, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7907373905181885, + "num_tokens": 253686269.0, + "step": 6648 + }, + { + "epoch": 0.8458211423483017, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7919261455535889, + "num_tokens": 253723632.0, + "step": 6649 + }, + { + "epoch": 0.8459483526268923, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7888921499252319, + "num_tokens": 253755990.0, + "step": 6650 + }, + { + "epoch": 0.8460755629054828, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7886592745780945, + "num_tokens": 253794200.0, + "step": 6651 + }, + { + "epoch": 0.8462027731840732, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8139025568962097, + "num_tokens": 253840554.0, + "step": 6652 + }, + { + "epoch": 0.8463299834626637, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8080073595046997, + "num_tokens": 253873642.0, + "step": 6653 + }, + { + "epoch": 0.8464571937412543, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7997387647628784, + "num_tokens": 253911839.0, + "step": 6654 + }, + { + "epoch": 0.8465844040198448, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7841761112213135, + "num_tokens": 253944649.0, + "step": 6655 + }, + { + "epoch": 0.8467116142984353, + "grad_norm": 176.0, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7976044416427612, + "num_tokens": 253977739.0, + "step": 6656 + }, + { + "epoch": 0.8468388245770259, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7801960706710815, + "num_tokens": 254021216.0, + "step": 6657 + }, + { + "epoch": 0.8469660348556163, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7947699427604675, + "num_tokens": 254063209.0, + "step": 6658 + }, + { + "epoch": 0.8470932451342068, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7924888134002686, + "num_tokens": 254102325.0, + "step": 6659 + }, + { + "epoch": 0.8472204554127973, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8016029596328735, + "num_tokens": 254136991.0, + "step": 6660 + }, + { + "epoch": 0.8473476656913879, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7989050149917603, + "num_tokens": 254179520.0, + "step": 6661 + }, + { + "epoch": 0.8474748759699784, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7965815663337708, + "num_tokens": 254220661.0, + "step": 6662 + }, + { + "epoch": 0.8476020862485689, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8058872818946838, + "num_tokens": 254261853.0, + "step": 6663 + }, + { + "epoch": 0.8477292965271594, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.798688530921936, + "num_tokens": 254301577.0, + "step": 6664 + }, + { + "epoch": 0.8478565068057499, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8150104284286499, + "num_tokens": 254341792.0, + "step": 6665 + }, + { + "epoch": 0.8479837170843404, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7826207280158997, + "num_tokens": 254376755.0, + "step": 6666 + }, + { + "epoch": 0.8481109273629309, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8086560964584351, + "num_tokens": 254413565.0, + "step": 6667 + }, + { + "epoch": 0.8482381376415215, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7847885489463806, + "num_tokens": 254449889.0, + "step": 6668 + }, + { + "epoch": 0.848365347920112, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8082143068313599, + "num_tokens": 254488129.0, + "step": 6669 + }, + { + "epoch": 0.8484925581987025, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8055506348609924, + "num_tokens": 254522031.0, + "step": 6670 + }, + { + "epoch": 0.8486197684772929, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8086485862731934, + "num_tokens": 254559249.0, + "step": 6671 + }, + { + "epoch": 0.8487469787558835, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7935884594917297, + "num_tokens": 254603968.0, + "step": 6672 + }, + { + "epoch": 0.848874189034474, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.808806836605072, + "num_tokens": 254642178.0, + "step": 6673 + }, + { + "epoch": 0.8490013993130645, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7870514392852783, + "num_tokens": 254679129.0, + "step": 6674 + }, + { + "epoch": 0.849128609591655, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7671826481819153, + "num_tokens": 254714952.0, + "step": 6675 + }, + { + "epoch": 0.8492558198702456, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7856878042221069, + "num_tokens": 254743323.0, + "step": 6676 + }, + { + "epoch": 0.849383030148836, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8082365989685059, + "num_tokens": 254784749.0, + "step": 6677 + }, + { + "epoch": 0.8495102404274265, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8028801083564758, + "num_tokens": 254820102.0, + "step": 6678 + }, + { + "epoch": 0.849637450706017, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.794120192527771, + "num_tokens": 254859988.0, + "step": 6679 + }, + { + "epoch": 0.8497646609846076, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7950422167778015, + "num_tokens": 254906792.0, + "step": 6680 + }, + { + "epoch": 0.8498918712631981, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7996104955673218, + "num_tokens": 254941552.0, + "step": 6681 + }, + { + "epoch": 0.8500190815417886, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7838209271430969, + "num_tokens": 254974653.0, + "step": 6682 + }, + { + "epoch": 0.850146291820379, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7828721404075623, + "num_tokens": 255018418.0, + "step": 6683 + }, + { + "epoch": 0.8502735020989696, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7957818508148193, + "num_tokens": 255049626.0, + "step": 6684 + }, + { + "epoch": 0.8504007123775601, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7892201542854309, + "num_tokens": 255086323.0, + "step": 6685 + }, + { + "epoch": 0.8505279226561506, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8023054003715515, + "num_tokens": 255124102.0, + "step": 6686 + }, + { + "epoch": 0.8506551329347412, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.8028599619865417, + "num_tokens": 255162176.0, + "step": 6687 + }, + { + "epoch": 0.8507823432133317, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7961881160736084, + "num_tokens": 255196256.0, + "step": 6688 + }, + { + "epoch": 0.8509095534919221, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7895379066467285, + "num_tokens": 255236752.0, + "step": 6689 + }, + { + "epoch": 0.8510367637705126, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7971683740615845, + "num_tokens": 255268694.0, + "step": 6690 + }, + { + "epoch": 0.8511639740491032, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8026416897773743, + "num_tokens": 255309445.0, + "step": 6691 + }, + { + "epoch": 0.8512911843276937, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7936587333679199, + "num_tokens": 255341126.0, + "step": 6692 + }, + { + "epoch": 0.8514183946062842, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8056691884994507, + "num_tokens": 255379553.0, + "step": 6693 + }, + { + "epoch": 0.8515456048848747, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8001232147216797, + "num_tokens": 255416588.0, + "step": 6694 + }, + { + "epoch": 0.8516728151634652, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7811300754547119, + "num_tokens": 255449574.0, + "step": 6695 + }, + { + "epoch": 0.8518000254420557, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7965513467788696, + "num_tokens": 255494088.0, + "step": 6696 + }, + { + "epoch": 0.8519272357206462, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8042306303977966, + "num_tokens": 255529785.0, + "step": 6697 + }, + { + "epoch": 0.8520544459992367, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7874412536621094, + "num_tokens": 255566675.0, + "step": 6698 + }, + { + "epoch": 0.8521816562778273, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8023260831832886, + "num_tokens": 255606502.0, + "step": 6699 + }, + { + "epoch": 0.8523088665564178, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7889844179153442, + "num_tokens": 255649179.0, + "step": 6700 + }, + { + "epoch": 0.8524360768350082, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7981629967689514, + "num_tokens": 255688190.0, + "step": 6701 + }, + { + "epoch": 0.8525632871135987, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.798527717590332, + "num_tokens": 255730401.0, + "step": 6702 + }, + { + "epoch": 0.8526904973921893, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7916548252105713, + "num_tokens": 255767881.0, + "step": 6703 + }, + { + "epoch": 0.8528177076707798, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.787611722946167, + "num_tokens": 255808347.0, + "step": 6704 + }, + { + "epoch": 0.8529449179493703, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7810897827148438, + "num_tokens": 255849538.0, + "step": 6705 + }, + { + "epoch": 0.8530721282279609, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8082712888717651, + "num_tokens": 255890228.0, + "step": 6706 + }, + { + "epoch": 0.8531993385065513, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7986622452735901, + "num_tokens": 255930189.0, + "step": 6707 + }, + { + "epoch": 0.8533265487851418, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8004467487335205, + "num_tokens": 255968310.0, + "step": 6708 + }, + { + "epoch": 0.8534537590637323, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8112221956253052, + "num_tokens": 256005872.0, + "step": 6709 + }, + { + "epoch": 0.8535809693423229, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8033450245857239, + "num_tokens": 256045374.0, + "step": 6710 + }, + { + "epoch": 0.8537081796209134, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.8011107444763184, + "num_tokens": 256084874.0, + "step": 6711 + }, + { + "epoch": 0.8538353898995039, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7875986099243164, + "num_tokens": 256122251.0, + "step": 6712 + }, + { + "epoch": 0.8539626001780944, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7977741956710815, + "num_tokens": 256165025.0, + "step": 6713 + }, + { + "epoch": 0.8540898104566849, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.806553840637207, + "num_tokens": 256202220.0, + "step": 6714 + }, + { + "epoch": 0.8542170207352754, + "grad_norm": 88.5, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.795556902885437, + "num_tokens": 256234036.0, + "step": 6715 + }, + { + "epoch": 0.8543442310138659, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7982825040817261, + "num_tokens": 256275253.0, + "step": 6716 + }, + { + "epoch": 0.8544714412924564, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7874182462692261, + "num_tokens": 256311296.0, + "step": 6717 + }, + { + "epoch": 0.854598651571047, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8019711971282959, + "num_tokens": 256348149.0, + "step": 6718 + }, + { + "epoch": 0.8547258618496375, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8127464652061462, + "num_tokens": 256387861.0, + "step": 6719 + }, + { + "epoch": 0.8548530721282279, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8046086430549622, + "num_tokens": 256427131.0, + "step": 6720 + }, + { + "epoch": 0.8549802824068184, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8006902933120728, + "num_tokens": 256465041.0, + "step": 6721 + }, + { + "epoch": 0.855107492685409, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7861172556877136, + "num_tokens": 256502576.0, + "step": 6722 + }, + { + "epoch": 0.8552347029639995, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.800588071346283, + "num_tokens": 256540493.0, + "step": 6723 + }, + { + "epoch": 0.85536191324259, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7796115875244141, + "num_tokens": 256578115.0, + "step": 6724 + }, + { + "epoch": 0.8554891235211806, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8056764006614685, + "num_tokens": 256615957.0, + "step": 6725 + }, + { + "epoch": 0.855616333799771, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7956501245498657, + "num_tokens": 256649972.0, + "step": 6726 + }, + { + "epoch": 0.8557435440783615, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7993354797363281, + "num_tokens": 256686683.0, + "step": 6727 + }, + { + "epoch": 0.855870754356952, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7914458513259888, + "num_tokens": 256723587.0, + "step": 6728 + }, + { + "epoch": 0.8559979646355426, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7971776723861694, + "num_tokens": 256765245.0, + "step": 6729 + }, + { + "epoch": 0.8561251749141331, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.5626, + "mean_token_accuracy": 0.8274540305137634, + "num_tokens": 256806930.0, + "step": 6730 + }, + { + "epoch": 0.8562523851927236, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7966036796569824, + "num_tokens": 256844568.0, + "step": 6731 + }, + { + "epoch": 0.856379595471314, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7882808446884155, + "num_tokens": 256877483.0, + "step": 6732 + }, + { + "epoch": 0.8565068057499046, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8057547211647034, + "num_tokens": 256921139.0, + "step": 6733 + }, + { + "epoch": 0.8566340160284951, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7986788749694824, + "num_tokens": 256957861.0, + "step": 6734 + }, + { + "epoch": 0.8567612263070856, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.800990879535675, + "num_tokens": 256998582.0, + "step": 6735 + }, + { + "epoch": 0.8568884365856761, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7942475080490112, + "num_tokens": 257037638.0, + "step": 6736 + }, + { + "epoch": 0.8570156468642667, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8004044890403748, + "num_tokens": 257074427.0, + "step": 6737 + }, + { + "epoch": 0.8571428571428571, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8019299507141113, + "num_tokens": 257113826.0, + "step": 6738 + }, + { + "epoch": 0.8572700674214476, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7815593481063843, + "num_tokens": 257152387.0, + "step": 6739 + }, + { + "epoch": 0.8573972777000382, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7999135255813599, + "num_tokens": 257192083.0, + "step": 6740 + }, + { + "epoch": 0.8575244879786287, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7891286611557007, + "num_tokens": 257235425.0, + "step": 6741 + }, + { + "epoch": 0.8576516982572192, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7891431450843811, + "num_tokens": 257268621.0, + "step": 6742 + }, + { + "epoch": 0.8577789085358097, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7997167706489563, + "num_tokens": 257306909.0, + "step": 6743 + }, + { + "epoch": 0.8579061188144002, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8156837224960327, + "num_tokens": 257343388.0, + "step": 6744 + }, + { + "epoch": 0.8580333290929907, + "grad_norm": 80.0, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8036044836044312, + "num_tokens": 257377172.0, + "step": 6745 + }, + { + "epoch": 0.8581605393715812, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7906319499015808, + "num_tokens": 257410312.0, + "step": 6746 + }, + { + "epoch": 0.8582877496501717, + "grad_norm": 80.0, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8018962144851685, + "num_tokens": 257449919.0, + "step": 6747 + }, + { + "epoch": 0.8584149599287623, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.80931556224823, + "num_tokens": 257488296.0, + "step": 6748 + }, + { + "epoch": 0.8585421702073528, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8081142902374268, + "num_tokens": 257524678.0, + "step": 6749 + }, + { + "epoch": 0.8586693804859432, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8011186122894287, + "num_tokens": 257559060.0, + "step": 6750 + }, + { + "epoch": 0.8587965907645337, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.80831378698349, + "num_tokens": 257594340.0, + "step": 6751 + }, + { + "epoch": 0.8589238010431243, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8063462376594543, + "num_tokens": 257627936.0, + "step": 6752 + }, + { + "epoch": 0.8590510113217148, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7914984822273254, + "num_tokens": 257660311.0, + "step": 6753 + }, + { + "epoch": 0.8591782216003053, + "grad_norm": 86.5, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7892992496490479, + "num_tokens": 257697496.0, + "step": 6754 + }, + { + "epoch": 0.8593054318788959, + "grad_norm": 85.0, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7746419906616211, + "num_tokens": 257737242.0, + "step": 6755 + }, + { + "epoch": 0.8594326421574863, + "grad_norm": 140.0, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7960413694381714, + "num_tokens": 257772946.0, + "step": 6756 + }, + { + "epoch": 0.8595598524360768, + "grad_norm": 208.0, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7888975739479065, + "num_tokens": 257807198.0, + "step": 6757 + }, + { + "epoch": 0.8596870627146673, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.804240345954895, + "num_tokens": 257848618.0, + "step": 6758 + }, + { + "epoch": 0.8598142729932579, + "grad_norm": 160.0, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8091869354248047, + "num_tokens": 257887805.0, + "step": 6759 + }, + { + "epoch": 0.8599414832718484, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8077999353408813, + "num_tokens": 257929276.0, + "step": 6760 + }, + { + "epoch": 0.8600686935504389, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.781904935836792, + "num_tokens": 257958629.0, + "step": 6761 + }, + { + "epoch": 0.8601959038290294, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7919703722000122, + "num_tokens": 257990303.0, + "step": 6762 + }, + { + "epoch": 0.8603231141076199, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7805296182632446, + "num_tokens": 258028704.0, + "step": 6763 + }, + { + "epoch": 0.8604503243862104, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8053001761436462, + "num_tokens": 258062757.0, + "step": 6764 + }, + { + "epoch": 0.8605775346648009, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7958072423934937, + "num_tokens": 258098574.0, + "step": 6765 + }, + { + "epoch": 0.8607047449433914, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8059826493263245, + "num_tokens": 258130937.0, + "step": 6766 + }, + { + "epoch": 0.860831955221982, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.8001404404640198, + "num_tokens": 258165142.0, + "step": 6767 + }, + { + "epoch": 0.8609591655005725, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.8220565915107727, + "num_tokens": 258203986.0, + "step": 6768 + }, + { + "epoch": 0.8610863757791629, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7968654632568359, + "num_tokens": 258238692.0, + "step": 6769 + }, + { + "epoch": 0.8612135860577534, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7781018614768982, + "num_tokens": 258280942.0, + "step": 6770 + }, + { + "epoch": 0.861340796336344, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8026343584060669, + "num_tokens": 258318835.0, + "step": 6771 + }, + { + "epoch": 0.8614680066149345, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7865299582481384, + "num_tokens": 258359461.0, + "step": 6772 + }, + { + "epoch": 0.861595216893525, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7945599555969238, + "num_tokens": 258394582.0, + "step": 6773 + }, + { + "epoch": 0.8617224271721156, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.8003512620925903, + "num_tokens": 258438526.0, + "step": 6774 + }, + { + "epoch": 0.861849637450706, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.780968964099884, + "num_tokens": 258479332.0, + "step": 6775 + }, + { + "epoch": 0.8619768477292965, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7882099747657776, + "num_tokens": 258513202.0, + "step": 6776 + }, + { + "epoch": 0.862104058007887, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7998629808425903, + "num_tokens": 258549145.0, + "step": 6777 + }, + { + "epoch": 0.8622312682864776, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7914486527442932, + "num_tokens": 258586421.0, + "step": 6778 + }, + { + "epoch": 0.8623584785650681, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7933031916618347, + "num_tokens": 258620014.0, + "step": 6779 + }, + { + "epoch": 0.8624856888436586, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8039355278015137, + "num_tokens": 258656124.0, + "step": 6780 + }, + { + "epoch": 0.862612899122249, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7836439609527588, + "num_tokens": 258692537.0, + "step": 6781 + }, + { + "epoch": 0.8627401094008396, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.803691565990448, + "num_tokens": 258732737.0, + "step": 6782 + }, + { + "epoch": 0.8628673196794301, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7873610258102417, + "num_tokens": 258770634.0, + "step": 6783 + }, + { + "epoch": 0.8629945299580206, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.8022074103355408, + "num_tokens": 258809643.0, + "step": 6784 + }, + { + "epoch": 0.8631217402366111, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8026130795478821, + "num_tokens": 258845336.0, + "step": 6785 + }, + { + "epoch": 0.8632489505152017, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7999837398529053, + "num_tokens": 258882206.0, + "step": 6786 + }, + { + "epoch": 0.8633761607937921, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8015437126159668, + "num_tokens": 258918661.0, + "step": 6787 + }, + { + "epoch": 0.8635033710723826, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7938320636749268, + "num_tokens": 258957452.0, + "step": 6788 + }, + { + "epoch": 0.8636305813509731, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7888845801353455, + "num_tokens": 259003507.0, + "step": 6789 + }, + { + "epoch": 0.8637577916295637, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7909139394760132, + "num_tokens": 259042284.0, + "step": 6790 + }, + { + "epoch": 0.8638850019081542, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.81302809715271, + "num_tokens": 259082929.0, + "step": 6791 + }, + { + "epoch": 0.8640122121867447, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8135570287704468, + "num_tokens": 259125920.0, + "step": 6792 + }, + { + "epoch": 0.8641394224653351, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8008972406387329, + "num_tokens": 259163140.0, + "step": 6793 + }, + { + "epoch": 0.8642666327439257, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7959715723991394, + "num_tokens": 259199038.0, + "step": 6794 + }, + { + "epoch": 0.8643938430225162, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7929326295852661, + "num_tokens": 259235713.0, + "step": 6795 + }, + { + "epoch": 0.8645210533011067, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7910414338111877, + "num_tokens": 259278314.0, + "step": 6796 + }, + { + "epoch": 0.8646482635796973, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8007739782333374, + "num_tokens": 259313039.0, + "step": 6797 + }, + { + "epoch": 0.8647754738582878, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8082139492034912, + "num_tokens": 259348845.0, + "step": 6798 + }, + { + "epoch": 0.8649026841368782, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7933465242385864, + "num_tokens": 259387345.0, + "step": 6799 + }, + { + "epoch": 0.8650298944154687, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7965009212493896, + "num_tokens": 259427835.0, + "step": 6800 + }, + { + "epoch": 0.8651571046940593, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8030085563659668, + "num_tokens": 259460453.0, + "step": 6801 + }, + { + "epoch": 0.8652843149726498, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.802074670791626, + "num_tokens": 259499296.0, + "step": 6802 + }, + { + "epoch": 0.8654115252512403, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7887442111968994, + "num_tokens": 259537649.0, + "step": 6803 + }, + { + "epoch": 0.8655387355298308, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7945528030395508, + "num_tokens": 259583136.0, + "step": 6804 + }, + { + "epoch": 0.8656659458084213, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8062877655029297, + "num_tokens": 259620422.0, + "step": 6805 + }, + { + "epoch": 0.8657931560870118, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7871400713920593, + "num_tokens": 259663948.0, + "step": 6806 + }, + { + "epoch": 0.8659203663656023, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8005926609039307, + "num_tokens": 259701986.0, + "step": 6807 + }, + { + "epoch": 0.8660475766441929, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8151648044586182, + "num_tokens": 259740684.0, + "step": 6808 + }, + { + "epoch": 0.8661747869227834, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7851364612579346, + "num_tokens": 259781659.0, + "step": 6809 + }, + { + "epoch": 0.8663019972013739, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7935175895690918, + "num_tokens": 259818554.0, + "step": 6810 + }, + { + "epoch": 0.8664292074799644, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8040451407432556, + "num_tokens": 259856652.0, + "step": 6811 + }, + { + "epoch": 0.8665564177585549, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8179234862327576, + "num_tokens": 259898255.0, + "step": 6812 + }, + { + "epoch": 0.8666836280371454, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8015454411506653, + "num_tokens": 259935305.0, + "step": 6813 + }, + { + "epoch": 0.8668108383157359, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8023389577865601, + "num_tokens": 259975245.0, + "step": 6814 + }, + { + "epoch": 0.8669380485943264, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8062782287597656, + "num_tokens": 260014150.0, + "step": 6815 + }, + { + "epoch": 0.867065258872917, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7900264263153076, + "num_tokens": 260054655.0, + "step": 6816 + }, + { + "epoch": 0.8671924691515075, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.8024502992630005, + "num_tokens": 260091528.0, + "step": 6817 + }, + { + "epoch": 0.8673196794300979, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8089471459388733, + "num_tokens": 260123069.0, + "step": 6818 + }, + { + "epoch": 0.8674468897086884, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7783368229866028, + "num_tokens": 260154482.0, + "step": 6819 + }, + { + "epoch": 0.867574099987279, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8016654253005981, + "num_tokens": 260193983.0, + "step": 6820 + }, + { + "epoch": 0.8677013102658695, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8004066944122314, + "num_tokens": 260228705.0, + "step": 6821 + }, + { + "epoch": 0.86782852054446, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7928789258003235, + "num_tokens": 260262821.0, + "step": 6822 + }, + { + "epoch": 0.8679557308230506, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7967967391014099, + "num_tokens": 260298949.0, + "step": 6823 + }, + { + "epoch": 0.868082941101641, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8084907531738281, + "num_tokens": 260340028.0, + "step": 6824 + }, + { + "epoch": 0.8682101513802315, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.8002678751945496, + "num_tokens": 260381328.0, + "step": 6825 + }, + { + "epoch": 0.868337361658822, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.800628662109375, + "num_tokens": 260423421.0, + "step": 6826 + }, + { + "epoch": 0.8684645719374126, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7963002920150757, + "num_tokens": 260461985.0, + "step": 6827 + }, + { + "epoch": 0.8685917822160031, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7864813208580017, + "num_tokens": 260503041.0, + "step": 6828 + }, + { + "epoch": 0.8687189924945936, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8074647188186646, + "num_tokens": 260545553.0, + "step": 6829 + }, + { + "epoch": 0.868846202773184, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8003483414649963, + "num_tokens": 260577160.0, + "step": 6830 + }, + { + "epoch": 0.8689734130517746, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8128091096878052, + "num_tokens": 260614748.0, + "step": 6831 + }, + { + "epoch": 0.8691006233303651, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8011468648910522, + "num_tokens": 260651762.0, + "step": 6832 + }, + { + "epoch": 0.8692278336089556, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.786439061164856, + "num_tokens": 260694438.0, + "step": 6833 + }, + { + "epoch": 0.8693550438875461, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8063037395477295, + "num_tokens": 260730256.0, + "step": 6834 + }, + { + "epoch": 0.8694822541661367, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7746384143829346, + "num_tokens": 260766190.0, + "step": 6835 + }, + { + "epoch": 0.8696094644447271, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8109644055366516, + "num_tokens": 260801733.0, + "step": 6836 + }, + { + "epoch": 0.8697366747233176, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8149604797363281, + "num_tokens": 260843461.0, + "step": 6837 + }, + { + "epoch": 0.8698638850019081, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7951114177703857, + "num_tokens": 260875006.0, + "step": 6838 + }, + { + "epoch": 0.8699910952804987, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7910776138305664, + "num_tokens": 260917553.0, + "step": 6839 + }, + { + "epoch": 0.8701183055590892, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7735679149627686, + "num_tokens": 260955293.0, + "step": 6840 + }, + { + "epoch": 0.8702455158376797, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7955898642539978, + "num_tokens": 260983966.0, + "step": 6841 + }, + { + "epoch": 0.8703727261162701, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8060042858123779, + "num_tokens": 261019434.0, + "step": 6842 + }, + { + "epoch": 0.8704999363948607, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7856552600860596, + "num_tokens": 261053761.0, + "step": 6843 + }, + { + "epoch": 0.8706271466734512, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7870105504989624, + "num_tokens": 261093490.0, + "step": 6844 + }, + { + "epoch": 0.8707543569520417, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7970269918441772, + "num_tokens": 261129518.0, + "step": 6845 + }, + { + "epoch": 0.8708815672306323, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7814061641693115, + "num_tokens": 261164282.0, + "step": 6846 + }, + { + "epoch": 0.8710087775092228, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7784600257873535, + "num_tokens": 261205055.0, + "step": 6847 + }, + { + "epoch": 0.8711359877878132, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7899484634399414, + "num_tokens": 261242329.0, + "step": 6848 + }, + { + "epoch": 0.8712631980664037, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7726075053215027, + "num_tokens": 261280330.0, + "step": 6849 + }, + { + "epoch": 0.8713904083449943, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8076132535934448, + "num_tokens": 261318232.0, + "step": 6850 + }, + { + "epoch": 0.8715176186235848, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8097666501998901, + "num_tokens": 261351483.0, + "step": 6851 + }, + { + "epoch": 0.8716448289021753, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7990016937255859, + "num_tokens": 261387058.0, + "step": 6852 + }, + { + "epoch": 0.8717720391807658, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8197813034057617, + "num_tokens": 261430110.0, + "step": 6853 + }, + { + "epoch": 0.8718992494593563, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8046982288360596, + "num_tokens": 261470707.0, + "step": 6854 + }, + { + "epoch": 0.8720264597379468, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8063243627548218, + "num_tokens": 261507146.0, + "step": 6855 + }, + { + "epoch": 0.8721536700165373, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7817996144294739, + "num_tokens": 261549723.0, + "step": 6856 + }, + { + "epoch": 0.8722808802951278, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8076521754264832, + "num_tokens": 261591030.0, + "step": 6857 + }, + { + "epoch": 0.8724080905737184, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7997480630874634, + "num_tokens": 261623919.0, + "step": 6858 + }, + { + "epoch": 0.8725353008523089, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8074076771736145, + "num_tokens": 261655277.0, + "step": 6859 + }, + { + "epoch": 0.8726625111308994, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7869046926498413, + "num_tokens": 261697462.0, + "step": 6860 + }, + { + "epoch": 0.8727897214094898, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8010584115982056, + "num_tokens": 261739134.0, + "step": 6861 + }, + { + "epoch": 0.8729169316880804, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7994574308395386, + "num_tokens": 261776295.0, + "step": 6862 + }, + { + "epoch": 0.8730441419666709, + "grad_norm": 73.0, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7973199486732483, + "num_tokens": 261812388.0, + "step": 6863 + }, + { + "epoch": 0.8731713522452614, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8043266534805298, + "num_tokens": 261855537.0, + "step": 6864 + }, + { + "epoch": 0.873298562523852, + "grad_norm": 91.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.809748649597168, + "num_tokens": 261899922.0, + "step": 6865 + }, + { + "epoch": 0.8734257728024425, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8099536895751953, + "num_tokens": 261934475.0, + "step": 6866 + }, + { + "epoch": 0.8735529830810329, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7991575002670288, + "num_tokens": 261973395.0, + "step": 6867 + }, + { + "epoch": 0.8736801933596234, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7804527282714844, + "num_tokens": 262011644.0, + "step": 6868 + }, + { + "epoch": 0.873807403638214, + "grad_norm": 93.5, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7850145697593689, + "num_tokens": 262044038.0, + "step": 6869 + }, + { + "epoch": 0.8739346139168045, + "grad_norm": 126.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7962182760238647, + "num_tokens": 262081023.0, + "step": 6870 + }, + { + "epoch": 0.874061824195395, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7913261651992798, + "num_tokens": 262123981.0, + "step": 6871 + }, + { + "epoch": 0.8741890344739855, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7864007949829102, + "num_tokens": 262161542.0, + "step": 6872 + }, + { + "epoch": 0.874316244752576, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7871177196502686, + "num_tokens": 262202482.0, + "step": 6873 + }, + { + "epoch": 0.8744434550311665, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.793010950088501, + "num_tokens": 262239499.0, + "step": 6874 + }, + { + "epoch": 0.874570665309757, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7996225357055664, + "num_tokens": 262281324.0, + "step": 6875 + }, + { + "epoch": 0.8746978755883476, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7838029861450195, + "num_tokens": 262318772.0, + "step": 6876 + }, + { + "epoch": 0.8748250858669381, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8084593415260315, + "num_tokens": 262353264.0, + "step": 6877 + }, + { + "epoch": 0.8749522961455286, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7934530973434448, + "num_tokens": 262393360.0, + "step": 6878 + }, + { + "epoch": 0.875079506424119, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8083051443099976, + "num_tokens": 262432010.0, + "step": 6879 + }, + { + "epoch": 0.8752067167027096, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7962835431098938, + "num_tokens": 262468828.0, + "step": 6880 + }, + { + "epoch": 0.8753339269813001, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8078855276107788, + "num_tokens": 262515141.0, + "step": 6881 + }, + { + "epoch": 0.8754611372598906, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7943313717842102, + "num_tokens": 262553137.0, + "step": 6882 + }, + { + "epoch": 0.8755883475384811, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8217083215713501, + "num_tokens": 262593847.0, + "step": 6883 + }, + { + "epoch": 0.8757155578170717, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7875467538833618, + "num_tokens": 262635048.0, + "step": 6884 + }, + { + "epoch": 0.8758427680956621, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7908569574356079, + "num_tokens": 262671200.0, + "step": 6885 + }, + { + "epoch": 0.8759699783742526, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8028267025947571, + "num_tokens": 262708304.0, + "step": 6886 + }, + { + "epoch": 0.8760971886528431, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.799999475479126, + "num_tokens": 262750680.0, + "step": 6887 + }, + { + "epoch": 0.8762243989314337, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7954954504966736, + "num_tokens": 262784845.0, + "step": 6888 + }, + { + "epoch": 0.8763516092100242, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7991200685501099, + "num_tokens": 262822646.0, + "step": 6889 + }, + { + "epoch": 0.8764788194886147, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7984955906867981, + "num_tokens": 262863663.0, + "step": 6890 + }, + { + "epoch": 0.8766060297672051, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7819445729255676, + "num_tokens": 262903726.0, + "step": 6891 + }, + { + "epoch": 0.8767332400457957, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7809898853302002, + "num_tokens": 262942748.0, + "step": 6892 + }, + { + "epoch": 0.8768604503243862, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7728542685508728, + "num_tokens": 262977639.0, + "step": 6893 + }, + { + "epoch": 0.8769876606029767, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8013847470283508, + "num_tokens": 263015970.0, + "step": 6894 + }, + { + "epoch": 0.8771148708815673, + "grad_norm": 82.5, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8106187582015991, + "num_tokens": 263058618.0, + "step": 6895 + }, + { + "epoch": 0.8772420811601578, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7809343934059143, + "num_tokens": 263098628.0, + "step": 6896 + }, + { + "epoch": 0.8773692914387482, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8031398057937622, + "num_tokens": 263136793.0, + "step": 6897 + }, + { + "epoch": 0.8774965017173387, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7973217964172363, + "num_tokens": 263170996.0, + "step": 6898 + }, + { + "epoch": 0.8776237119959293, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7866067290306091, + "num_tokens": 263199961.0, + "step": 6899 + }, + { + "epoch": 0.8777509222745198, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.799976110458374, + "num_tokens": 263241541.0, + "step": 6900 + }, + { + "epoch": 0.8778781325531103, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8103371262550354, + "num_tokens": 263278778.0, + "step": 6901 + }, + { + "epoch": 0.8780053428317008, + "grad_norm": 115.0, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7985073328018188, + "num_tokens": 263310155.0, + "step": 6902 + }, + { + "epoch": 0.8781325531102913, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7983739376068115, + "num_tokens": 263355793.0, + "step": 6903 + }, + { + "epoch": 0.8782597633888818, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8064833283424377, + "num_tokens": 263394446.0, + "step": 6904 + }, + { + "epoch": 0.8783869736674723, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.8007181882858276, + "num_tokens": 263438479.0, + "step": 6905 + }, + { + "epoch": 0.8785141839460628, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7844947576522827, + "num_tokens": 263478514.0, + "step": 6906 + }, + { + "epoch": 0.8786413942246534, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7793689966201782, + "num_tokens": 263516734.0, + "step": 6907 + }, + { + "epoch": 0.8787686045032439, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7918922901153564, + "num_tokens": 263560105.0, + "step": 6908 + }, + { + "epoch": 0.8788958147818343, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7949222326278687, + "num_tokens": 263597033.0, + "step": 6909 + }, + { + "epoch": 0.8790230250604248, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8049571514129639, + "num_tokens": 263631106.0, + "step": 6910 + }, + { + "epoch": 0.8791502353390154, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.788520097732544, + "num_tokens": 263668227.0, + "step": 6911 + }, + { + "epoch": 0.8792774456176059, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8020765781402588, + "num_tokens": 263704335.0, + "step": 6912 + }, + { + "epoch": 0.8794046558961964, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.802526593208313, + "num_tokens": 263742532.0, + "step": 6913 + }, + { + "epoch": 0.879531866174787, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8044159412384033, + "num_tokens": 263782321.0, + "step": 6914 + }, + { + "epoch": 0.8796590764533775, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8090341091156006, + "num_tokens": 263818037.0, + "step": 6915 + }, + { + "epoch": 0.8797862867319679, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7843875288963318, + "num_tokens": 263854901.0, + "step": 6916 + }, + { + "epoch": 0.8799134970105584, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8008056879043579, + "num_tokens": 263896241.0, + "step": 6917 + }, + { + "epoch": 0.880040707289149, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7986828088760376, + "num_tokens": 263934673.0, + "step": 6918 + }, + { + "epoch": 0.8801679175677395, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7918667197227478, + "num_tokens": 263973456.0, + "step": 6919 + }, + { + "epoch": 0.88029512784633, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7798366546630859, + "num_tokens": 264006050.0, + "step": 6920 + }, + { + "epoch": 0.8804223381249205, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8033171892166138, + "num_tokens": 264042736.0, + "step": 6921 + }, + { + "epoch": 0.880549548403511, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7933284044265747, + "num_tokens": 264084433.0, + "step": 6922 + }, + { + "epoch": 0.8806767586821015, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.798513650894165, + "num_tokens": 264118331.0, + "step": 6923 + }, + { + "epoch": 0.880803968960692, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8028794527053833, + "num_tokens": 264152015.0, + "step": 6924 + }, + { + "epoch": 0.8809311792392825, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8017062544822693, + "num_tokens": 264187085.0, + "step": 6925 + }, + { + "epoch": 0.8810583895178731, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7943171858787537, + "num_tokens": 264225558.0, + "step": 6926 + }, + { + "epoch": 0.8811855997964636, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.793616771697998, + "num_tokens": 264264220.0, + "step": 6927 + }, + { + "epoch": 0.881312810075054, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8025951385498047, + "num_tokens": 264302097.0, + "step": 6928 + }, + { + "epoch": 0.8814400203536445, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8000234365463257, + "num_tokens": 264341966.0, + "step": 6929 + }, + { + "epoch": 0.8815672306322351, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7731252312660217, + "num_tokens": 264384298.0, + "step": 6930 + }, + { + "epoch": 0.8816944409108256, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7977613210678101, + "num_tokens": 264422513.0, + "step": 6931 + }, + { + "epoch": 0.8818216511894161, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8130796551704407, + "num_tokens": 264461819.0, + "step": 6932 + }, + { + "epoch": 0.8819488614680067, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7864823341369629, + "num_tokens": 264502618.0, + "step": 6933 + }, + { + "epoch": 0.8820760717465971, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7905926704406738, + "num_tokens": 264534382.0, + "step": 6934 + }, + { + "epoch": 0.8822032820251876, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7910189628601074, + "num_tokens": 264575297.0, + "step": 6935 + }, + { + "epoch": 0.8823304923037781, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.800986647605896, + "num_tokens": 264611822.0, + "step": 6936 + }, + { + "epoch": 0.8824577025823687, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7956912517547607, + "num_tokens": 264657483.0, + "step": 6937 + }, + { + "epoch": 0.8825849128609592, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7956114411354065, + "num_tokens": 264690633.0, + "step": 6938 + }, + { + "epoch": 0.8827121231395497, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8198150992393494, + "num_tokens": 264733268.0, + "step": 6939 + }, + { + "epoch": 0.8828393334181401, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7956555485725403, + "num_tokens": 264774769.0, + "step": 6940 + }, + { + "epoch": 0.8829665436967307, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7877097725868225, + "num_tokens": 264806658.0, + "step": 6941 + }, + { + "epoch": 0.8830937539753212, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7852129936218262, + "num_tokens": 264841054.0, + "step": 6942 + }, + { + "epoch": 0.8832209642539117, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7945935726165771, + "num_tokens": 264880241.0, + "step": 6943 + }, + { + "epoch": 0.8833481745325023, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8048836588859558, + "num_tokens": 264922449.0, + "step": 6944 + }, + { + "epoch": 0.8834753848110928, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.8018983602523804, + "num_tokens": 264961605.0, + "step": 6945 + }, + { + "epoch": 0.8836025950896832, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8036104440689087, + "num_tokens": 264992924.0, + "step": 6946 + }, + { + "epoch": 0.8837298053682737, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7947923541069031, + "num_tokens": 265031849.0, + "step": 6947 + }, + { + "epoch": 0.8838570156468643, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8124392032623291, + "num_tokens": 265069714.0, + "step": 6948 + }, + { + "epoch": 0.8839842259254548, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7906619310379028, + "num_tokens": 265103325.0, + "step": 6949 + }, + { + "epoch": 0.8841114362040453, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7741929888725281, + "num_tokens": 265139133.0, + "step": 6950 + }, + { + "epoch": 0.8842386464826358, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7732224464416504, + "num_tokens": 265183536.0, + "step": 6951 + }, + { + "epoch": 0.8843658567612263, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7910394072532654, + "num_tokens": 265226288.0, + "step": 6952 + }, + { + "epoch": 0.8844930670398168, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7946271896362305, + "num_tokens": 265265846.0, + "step": 6953 + }, + { + "epoch": 0.8846202773184073, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8111701607704163, + "num_tokens": 265300612.0, + "step": 6954 + }, + { + "epoch": 0.8847474875969978, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8047624826431274, + "num_tokens": 265341440.0, + "step": 6955 + }, + { + "epoch": 0.8848746978755884, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7870082855224609, + "num_tokens": 265384854.0, + "step": 6956 + }, + { + "epoch": 0.8850019081541789, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7882565259933472, + "num_tokens": 265420611.0, + "step": 6957 + }, + { + "epoch": 0.8851291184327693, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7949230670928955, + "num_tokens": 265457720.0, + "step": 6958 + }, + { + "epoch": 0.8852563287113598, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8058313727378845, + "num_tokens": 265493434.0, + "step": 6959 + }, + { + "epoch": 0.8853835389899504, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.796406626701355, + "num_tokens": 265529015.0, + "step": 6960 + }, + { + "epoch": 0.8855107492685409, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7828394174575806, + "num_tokens": 265568698.0, + "step": 6961 + }, + { + "epoch": 0.8856379595471314, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.797214150428772, + "num_tokens": 265605631.0, + "step": 6962 + }, + { + "epoch": 0.885765169825722, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.8002192974090576, + "num_tokens": 265647071.0, + "step": 6963 + }, + { + "epoch": 0.8858923801043125, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7994464635848999, + "num_tokens": 265688680.0, + "step": 6964 + }, + { + "epoch": 0.8860195903829029, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8019323945045471, + "num_tokens": 265727707.0, + "step": 6965 + }, + { + "epoch": 0.8861468006614934, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7947527170181274, + "num_tokens": 265765860.0, + "step": 6966 + }, + { + "epoch": 0.886274010940084, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8009295463562012, + "num_tokens": 265802988.0, + "step": 6967 + }, + { + "epoch": 0.8864012212186745, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7909620404243469, + "num_tokens": 265840614.0, + "step": 6968 + }, + { + "epoch": 0.886528431497265, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7698492407798767, + "num_tokens": 265883032.0, + "step": 6969 + }, + { + "epoch": 0.8866556417758555, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.811641275882721, + "num_tokens": 265918309.0, + "step": 6970 + }, + { + "epoch": 0.886782852054446, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7994882464408875, + "num_tokens": 265956406.0, + "step": 6971 + }, + { + "epoch": 0.8869100623330365, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8039960861206055, + "num_tokens": 265988413.0, + "step": 6972 + }, + { + "epoch": 0.887037272611627, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7932803630828857, + "num_tokens": 266022036.0, + "step": 6973 + }, + { + "epoch": 0.8871644828902175, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7898414134979248, + "num_tokens": 266053735.0, + "step": 6974 + }, + { + "epoch": 0.8872916931688081, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8122491240501404, + "num_tokens": 266094198.0, + "step": 6975 + }, + { + "epoch": 0.8874189034473986, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8063668012619019, + "num_tokens": 266132213.0, + "step": 6976 + }, + { + "epoch": 0.887546113725989, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7975406646728516, + "num_tokens": 266168001.0, + "step": 6977 + }, + { + "epoch": 0.8876733240045795, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7604314684867859, + "num_tokens": 266208771.0, + "step": 6978 + }, + { + "epoch": 0.8878005342831701, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7999675273895264, + "num_tokens": 266248090.0, + "step": 6979 + }, + { + "epoch": 0.8879277445617606, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.782730221748352, + "num_tokens": 266283748.0, + "step": 6980 + }, + { + "epoch": 0.8880549548403511, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8054665327072144, + "num_tokens": 266316613.0, + "step": 6981 + }, + { + "epoch": 0.8881821651189417, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8001387119293213, + "num_tokens": 266357515.0, + "step": 6982 + }, + { + "epoch": 0.8883093753975321, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8076211810112, + "num_tokens": 266393593.0, + "step": 6983 + }, + { + "epoch": 0.8884365856761226, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.8126776218414307, + "num_tokens": 266436104.0, + "step": 6984 + }, + { + "epoch": 0.8885637959547131, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8127459287643433, + "num_tokens": 266472809.0, + "step": 6985 + }, + { + "epoch": 0.8886910062333037, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8081142902374268, + "num_tokens": 266512726.0, + "step": 6986 + }, + { + "epoch": 0.8888182165118942, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7860959768295288, + "num_tokens": 266546098.0, + "step": 6987 + }, + { + "epoch": 0.8889454267904847, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7919428944587708, + "num_tokens": 266588478.0, + "step": 6988 + }, + { + "epoch": 0.8890726370690751, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7850952744483948, + "num_tokens": 266628975.0, + "step": 6989 + }, + { + "epoch": 0.8891998473476657, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8070254325866699, + "num_tokens": 266667533.0, + "step": 6990 + }, + { + "epoch": 0.8893270576262562, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7840476632118225, + "num_tokens": 266706990.0, + "step": 6991 + }, + { + "epoch": 0.8894542679048467, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7856732606887817, + "num_tokens": 266747314.0, + "step": 6992 + }, + { + "epoch": 0.8895814781834372, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.792415976524353, + "num_tokens": 266791031.0, + "step": 6993 + }, + { + "epoch": 0.8897086884620278, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7962510585784912, + "num_tokens": 266823236.0, + "step": 6994 + }, + { + "epoch": 0.8898358987406182, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7960432767868042, + "num_tokens": 266867993.0, + "step": 6995 + }, + { + "epoch": 0.8899631090192087, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7798361778259277, + "num_tokens": 266908573.0, + "step": 6996 + }, + { + "epoch": 0.8900903192977992, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7995195388793945, + "num_tokens": 266951167.0, + "step": 6997 + }, + { + "epoch": 0.8902175295763898, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7936300039291382, + "num_tokens": 266988976.0, + "step": 6998 + }, + { + "epoch": 0.8903447398549803, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7962396144866943, + "num_tokens": 267022450.0, + "step": 6999 + }, + { + "epoch": 0.8904719501335708, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7853744029998779, + "num_tokens": 267059933.0, + "step": 7000 + }, + { + "epoch": 0.8905991604121613, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7866250276565552, + "num_tokens": 267092518.0, + "step": 7001 + }, + { + "epoch": 0.8907263706907518, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7926127910614014, + "num_tokens": 267130008.0, + "step": 7002 + }, + { + "epoch": 0.8908535809693423, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8016664981842041, + "num_tokens": 267171039.0, + "step": 7003 + }, + { + "epoch": 0.8909807912479328, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7972323298454285, + "num_tokens": 267211197.0, + "step": 7004 + }, + { + "epoch": 0.8911080015265234, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8024505376815796, + "num_tokens": 267247471.0, + "step": 7005 + }, + { + "epoch": 0.8912352118051139, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7636436223983765, + "num_tokens": 267287276.0, + "step": 7006 + }, + { + "epoch": 0.8913624220837043, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7918685078620911, + "num_tokens": 267323827.0, + "step": 7007 + }, + { + "epoch": 0.8914896323622948, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8074522018432617, + "num_tokens": 267364053.0, + "step": 7008 + }, + { + "epoch": 0.8916168426408854, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7766542434692383, + "num_tokens": 267404393.0, + "step": 7009 + }, + { + "epoch": 0.8917440529194759, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7995330691337585, + "num_tokens": 267444589.0, + "step": 7010 + }, + { + "epoch": 0.8918712631980664, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7933642864227295, + "num_tokens": 267484460.0, + "step": 7011 + }, + { + "epoch": 0.891998473476657, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.792374849319458, + "num_tokens": 267517390.0, + "step": 7012 + }, + { + "epoch": 0.8921256837552475, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.792137861251831, + "num_tokens": 267555813.0, + "step": 7013 + }, + { + "epoch": 0.8922528940338379, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7960845828056335, + "num_tokens": 267595818.0, + "step": 7014 + }, + { + "epoch": 0.8923801043124284, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7701088190078735, + "num_tokens": 267633106.0, + "step": 7015 + }, + { + "epoch": 0.892507314591019, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8015276193618774, + "num_tokens": 267670166.0, + "step": 7016 + }, + { + "epoch": 0.8926345248696095, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.797264814376831, + "num_tokens": 267707284.0, + "step": 7017 + }, + { + "epoch": 0.8927617351482, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7933646440505981, + "num_tokens": 267742349.0, + "step": 7018 + }, + { + "epoch": 0.8928889454267905, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8021383881568909, + "num_tokens": 267776003.0, + "step": 7019 + }, + { + "epoch": 0.893016155705381, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.793542742729187, + "num_tokens": 267815887.0, + "step": 7020 + }, + { + "epoch": 0.8931433659839715, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8115497827529907, + "num_tokens": 267855874.0, + "step": 7021 + }, + { + "epoch": 0.893270576262562, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8100627660751343, + "num_tokens": 267897901.0, + "step": 7022 + }, + { + "epoch": 0.8933977865411525, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8063849210739136, + "num_tokens": 267938967.0, + "step": 7023 + }, + { + "epoch": 0.8935249968197431, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.789989709854126, + "num_tokens": 267975196.0, + "step": 7024 + }, + { + "epoch": 0.8936522070983336, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8065414428710938, + "num_tokens": 268014407.0, + "step": 7025 + }, + { + "epoch": 0.893779417376924, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.8094074130058289, + "num_tokens": 268043892.0, + "step": 7026 + }, + { + "epoch": 0.8939066276555145, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7939102053642273, + "num_tokens": 268079651.0, + "step": 7027 + }, + { + "epoch": 0.8940338379341051, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7938129901885986, + "num_tokens": 268114021.0, + "step": 7028 + }, + { + "epoch": 0.8941610482126956, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7938178181648254, + "num_tokens": 268155315.0, + "step": 7029 + }, + { + "epoch": 0.8942882584912861, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7951093316078186, + "num_tokens": 268199312.0, + "step": 7030 + }, + { + "epoch": 0.8944154687698767, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8017063140869141, + "num_tokens": 268236955.0, + "step": 7031 + }, + { + "epoch": 0.8945426790484671, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7949888706207275, + "num_tokens": 268272183.0, + "step": 7032 + }, + { + "epoch": 0.8946698893270576, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7876548767089844, + "num_tokens": 268305407.0, + "step": 7033 + }, + { + "epoch": 0.8947970996056481, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8053383827209473, + "num_tokens": 268340443.0, + "step": 7034 + }, + { + "epoch": 0.8949243098842387, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8181906938552856, + "num_tokens": 268375597.0, + "step": 7035 + }, + { + "epoch": 0.8950515201628292, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7934688925743103, + "num_tokens": 268412872.0, + "step": 7036 + }, + { + "epoch": 0.8951787304414197, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7956814765930176, + "num_tokens": 268452021.0, + "step": 7037 + }, + { + "epoch": 0.8953059407200101, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8062018156051636, + "num_tokens": 268486354.0, + "step": 7038 + }, + { + "epoch": 0.8954331509986007, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7927340269088745, + "num_tokens": 268519918.0, + "step": 7039 + }, + { + "epoch": 0.8955603612771912, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7906948328018188, + "num_tokens": 268554259.0, + "step": 7040 + }, + { + "epoch": 0.8956875715557817, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7929999828338623, + "num_tokens": 268595785.0, + "step": 7041 + }, + { + "epoch": 0.8958147818343722, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7875508069992065, + "num_tokens": 268640805.0, + "step": 7042 + }, + { + "epoch": 0.8959419921129628, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8052065968513489, + "num_tokens": 268682446.0, + "step": 7043 + }, + { + "epoch": 0.8960692023915532, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8029248714447021, + "num_tokens": 268720262.0, + "step": 7044 + }, + { + "epoch": 0.8961964126701437, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7851822376251221, + "num_tokens": 268760519.0, + "step": 7045 + }, + { + "epoch": 0.8963236229487342, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7857050895690918, + "num_tokens": 268796621.0, + "step": 7046 + }, + { + "epoch": 0.8964508332273248, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8054394721984863, + "num_tokens": 268830527.0, + "step": 7047 + }, + { + "epoch": 0.8965780435059153, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8049264550209045, + "num_tokens": 268862112.0, + "step": 7048 + }, + { + "epoch": 0.8967052537845058, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7985547780990601, + "num_tokens": 268901089.0, + "step": 7049 + }, + { + "epoch": 0.8968324640630962, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8044727444648743, + "num_tokens": 268942766.0, + "step": 7050 + }, + { + "epoch": 0.8969596743416868, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7968356609344482, + "num_tokens": 268978642.0, + "step": 7051 + }, + { + "epoch": 0.8970868846202773, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8059118390083313, + "num_tokens": 269016048.0, + "step": 7052 + }, + { + "epoch": 0.8972140948988678, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7975729703903198, + "num_tokens": 269054381.0, + "step": 7053 + }, + { + "epoch": 0.8973413051774584, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7968112230300903, + "num_tokens": 269090672.0, + "step": 7054 + }, + { + "epoch": 0.8974685154560489, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7961622476577759, + "num_tokens": 269130231.0, + "step": 7055 + }, + { + "epoch": 0.8975957257346393, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7971141338348389, + "num_tokens": 269166282.0, + "step": 7056 + }, + { + "epoch": 0.8977229360132298, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.801446795463562, + "num_tokens": 269205326.0, + "step": 7057 + }, + { + "epoch": 0.8978501462918204, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8005245923995972, + "num_tokens": 269243903.0, + "step": 7058 + }, + { + "epoch": 0.8979773565704109, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7948799133300781, + "num_tokens": 269288330.0, + "step": 7059 + }, + { + "epoch": 0.8981045668490014, + "grad_norm": 73.5, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7859196066856384, + "num_tokens": 269328731.0, + "step": 7060 + }, + { + "epoch": 0.898231777127592, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8007771968841553, + "num_tokens": 269369576.0, + "step": 7061 + }, + { + "epoch": 0.8983589874061825, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7955461740493774, + "num_tokens": 269407734.0, + "step": 7062 + }, + { + "epoch": 0.8984861976847729, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7751609683036804, + "num_tokens": 269444042.0, + "step": 7063 + }, + { + "epoch": 0.8986134079633634, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8252912759780884, + "num_tokens": 269480979.0, + "step": 7064 + }, + { + "epoch": 0.898740618241954, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7945870757102966, + "num_tokens": 269515594.0, + "step": 7065 + }, + { + "epoch": 0.8988678285205445, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7982161045074463, + "num_tokens": 269556974.0, + "step": 7066 + }, + { + "epoch": 0.898995038799135, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7876182198524475, + "num_tokens": 269591088.0, + "step": 7067 + }, + { + "epoch": 0.8991222490777255, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8081403970718384, + "num_tokens": 269627921.0, + "step": 7068 + }, + { + "epoch": 0.899249459356316, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7831500768661499, + "num_tokens": 269666586.0, + "step": 7069 + }, + { + "epoch": 0.8993766696349065, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7998327016830444, + "num_tokens": 269701205.0, + "step": 7070 + }, + { + "epoch": 0.899503879913497, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8116368055343628, + "num_tokens": 269739982.0, + "step": 7071 + }, + { + "epoch": 0.8996310901920875, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8010300397872925, + "num_tokens": 269773048.0, + "step": 7072 + }, + { + "epoch": 0.8997583004706781, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7922428846359253, + "num_tokens": 269811343.0, + "step": 7073 + }, + { + "epoch": 0.8998855107492686, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8039420247077942, + "num_tokens": 269851601.0, + "step": 7074 + }, + { + "epoch": 0.900012721027859, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8058213591575623, + "num_tokens": 269891396.0, + "step": 7075 + }, + { + "epoch": 0.9001399313064495, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7916089296340942, + "num_tokens": 269930292.0, + "step": 7076 + }, + { + "epoch": 0.9002671415850401, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.771271824836731, + "num_tokens": 269968464.0, + "step": 7077 + }, + { + "epoch": 0.9003943518636306, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.787725031375885, + "num_tokens": 270011569.0, + "step": 7078 + }, + { + "epoch": 0.9005215621422211, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7893364429473877, + "num_tokens": 270050608.0, + "step": 7079 + }, + { + "epoch": 0.9006487724208116, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7842007875442505, + "num_tokens": 270096100.0, + "step": 7080 + }, + { + "epoch": 0.9007759826994021, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8089536428451538, + "num_tokens": 270140915.0, + "step": 7081 + }, + { + "epoch": 0.9009031929779926, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7997937202453613, + "num_tokens": 270184233.0, + "step": 7082 + }, + { + "epoch": 0.9010304032565831, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7949376702308655, + "num_tokens": 270223889.0, + "step": 7083 + }, + { + "epoch": 0.9011576135351737, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8057013750076294, + "num_tokens": 270266357.0, + "step": 7084 + }, + { + "epoch": 0.9012848238137642, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8071214556694031, + "num_tokens": 270298252.0, + "step": 7085 + }, + { + "epoch": 0.9014120340923547, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.794912576675415, + "num_tokens": 270343187.0, + "step": 7086 + }, + { + "epoch": 0.9015392443709451, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7842223644256592, + "num_tokens": 270385094.0, + "step": 7087 + }, + { + "epoch": 0.9016664546495357, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7844940423965454, + "num_tokens": 270421772.0, + "step": 7088 + }, + { + "epoch": 0.9017936649281262, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7746934294700623, + "num_tokens": 270465033.0, + "step": 7089 + }, + { + "epoch": 0.9019208752067167, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8163638710975647, + "num_tokens": 270501370.0, + "step": 7090 + }, + { + "epoch": 0.9020480854853072, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8044523000717163, + "num_tokens": 270543571.0, + "step": 7091 + }, + { + "epoch": 0.9021752957638978, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8074352741241455, + "num_tokens": 270580635.0, + "step": 7092 + }, + { + "epoch": 0.9023025060424882, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8129535913467407, + "num_tokens": 270616534.0, + "step": 7093 + }, + { + "epoch": 0.9024297163210787, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7755381464958191, + "num_tokens": 270653998.0, + "step": 7094 + }, + { + "epoch": 0.9025569265996692, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7959798574447632, + "num_tokens": 270688023.0, + "step": 7095 + }, + { + "epoch": 0.9026841368782598, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7775281667709351, + "num_tokens": 270721298.0, + "step": 7096 + }, + { + "epoch": 0.9028113471568503, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.801438570022583, + "num_tokens": 270762182.0, + "step": 7097 + }, + { + "epoch": 0.9029385574354408, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8140497207641602, + "num_tokens": 270799577.0, + "step": 7098 + }, + { + "epoch": 0.9030657677140312, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7964726686477661, + "num_tokens": 270840218.0, + "step": 7099 + }, + { + "epoch": 0.9031929779926218, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8150467872619629, + "num_tokens": 270872982.0, + "step": 7100 + }, + { + "epoch": 0.9033201882712123, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8007117509841919, + "num_tokens": 270908381.0, + "step": 7101 + }, + { + "epoch": 0.9034473985498028, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8001275658607483, + "num_tokens": 270949217.0, + "step": 7102 + }, + { + "epoch": 0.9035746088283934, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.78055340051651, + "num_tokens": 270989667.0, + "step": 7103 + }, + { + "epoch": 0.9037018191069839, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7887192368507385, + "num_tokens": 271030395.0, + "step": 7104 + }, + { + "epoch": 0.9038290293855743, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7842534780502319, + "num_tokens": 271068239.0, + "step": 7105 + }, + { + "epoch": 0.9039562396641648, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7888561487197876, + "num_tokens": 271110294.0, + "step": 7106 + }, + { + "epoch": 0.9040834499427554, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.8196156620979309, + "num_tokens": 271143288.0, + "step": 7107 + }, + { + "epoch": 0.9042106602213459, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8049453496932983, + "num_tokens": 271181183.0, + "step": 7108 + }, + { + "epoch": 0.9043378704999364, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7953643798828125, + "num_tokens": 271216201.0, + "step": 7109 + }, + { + "epoch": 0.9044650807785269, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8049436807632446, + "num_tokens": 271250158.0, + "step": 7110 + }, + { + "epoch": 0.9045922910571175, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8117625713348389, + "num_tokens": 271289994.0, + "step": 7111 + }, + { + "epoch": 0.9047195013357079, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8025937080383301, + "num_tokens": 271325782.0, + "step": 7112 + }, + { + "epoch": 0.9048467116142984, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7795014381408691, + "num_tokens": 271361604.0, + "step": 7113 + }, + { + "epoch": 0.9049739218928889, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8071973323822021, + "num_tokens": 271398520.0, + "step": 7114 + }, + { + "epoch": 0.9051011321714795, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7935712337493896, + "num_tokens": 271445138.0, + "step": 7115 + }, + { + "epoch": 0.90522834245007, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7997308373451233, + "num_tokens": 271479147.0, + "step": 7116 + }, + { + "epoch": 0.9053555527286605, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8001052141189575, + "num_tokens": 271512477.0, + "step": 7117 + }, + { + "epoch": 0.905482763007251, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7829768061637878, + "num_tokens": 271552355.0, + "step": 7118 + }, + { + "epoch": 0.9056099732858415, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7961791157722473, + "num_tokens": 271595825.0, + "step": 7119 + }, + { + "epoch": 0.905737183564432, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7846441268920898, + "num_tokens": 271627384.0, + "step": 7120 + }, + { + "epoch": 0.9058643938430225, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8077109456062317, + "num_tokens": 271666711.0, + "step": 7121 + }, + { + "epoch": 0.9059916041216131, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8029476404190063, + "num_tokens": 271703145.0, + "step": 7122 + }, + { + "epoch": 0.9061188144002036, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7937744855880737, + "num_tokens": 271743131.0, + "step": 7123 + }, + { + "epoch": 0.906246024678794, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8014026880264282, + "num_tokens": 271778245.0, + "step": 7124 + }, + { + "epoch": 0.9063732349573845, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7874166369438171, + "num_tokens": 271817279.0, + "step": 7125 + }, + { + "epoch": 0.9065004452359751, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8055952191352844, + "num_tokens": 271857136.0, + "step": 7126 + }, + { + "epoch": 0.9066276555145656, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8042320013046265, + "num_tokens": 271891651.0, + "step": 7127 + }, + { + "epoch": 0.9067548657931561, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8021993637084961, + "num_tokens": 271925805.0, + "step": 7128 + }, + { + "epoch": 0.9068820760717466, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8081859946250916, + "num_tokens": 271964954.0, + "step": 7129 + }, + { + "epoch": 0.9070092863503371, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.802731990814209, + "num_tokens": 272000331.0, + "step": 7130 + }, + { + "epoch": 0.9071364966289276, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8116704225540161, + "num_tokens": 272037853.0, + "step": 7131 + }, + { + "epoch": 0.9072637069075181, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8096708059310913, + "num_tokens": 272077663.0, + "step": 7132 + }, + { + "epoch": 0.9073909171861086, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7901602387428284, + "num_tokens": 272116942.0, + "step": 7133 + }, + { + "epoch": 0.9075181274646992, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7904251217842102, + "num_tokens": 272151319.0, + "step": 7134 + }, + { + "epoch": 0.9076453377432897, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7846205234527588, + "num_tokens": 272190817.0, + "step": 7135 + }, + { + "epoch": 0.9077725480218801, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.790316641330719, + "num_tokens": 272230461.0, + "step": 7136 + }, + { + "epoch": 0.9078997583004706, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7911111116409302, + "num_tokens": 272268510.0, + "step": 7137 + }, + { + "epoch": 0.9080269685790612, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7981336116790771, + "num_tokens": 272312430.0, + "step": 7138 + }, + { + "epoch": 0.9081541788576517, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7905097603797913, + "num_tokens": 272348437.0, + "step": 7139 + }, + { + "epoch": 0.9082813891362422, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7847508192062378, + "num_tokens": 272378902.0, + "step": 7140 + }, + { + "epoch": 0.9084085994148328, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8093253970146179, + "num_tokens": 272421257.0, + "step": 7141 + }, + { + "epoch": 0.9085358096934232, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7942056655883789, + "num_tokens": 272459800.0, + "step": 7142 + }, + { + "epoch": 0.9086630199720137, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8178665041923523, + "num_tokens": 272501245.0, + "step": 7143 + }, + { + "epoch": 0.9087902302506042, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7970485687255859, + "num_tokens": 272535873.0, + "step": 7144 + }, + { + "epoch": 0.9089174405291948, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7984053492546082, + "num_tokens": 272575213.0, + "step": 7145 + }, + { + "epoch": 0.9090446508077853, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8049597144126892, + "num_tokens": 272611960.0, + "step": 7146 + }, + { + "epoch": 0.9091718610863758, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.795474648475647, + "num_tokens": 272650895.0, + "step": 7147 + }, + { + "epoch": 0.9092990713649662, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8006963729858398, + "num_tokens": 272690617.0, + "step": 7148 + }, + { + "epoch": 0.9094262816435568, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7937707901000977, + "num_tokens": 272732526.0, + "step": 7149 + }, + { + "epoch": 0.9095534919221473, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7937294840812683, + "num_tokens": 272769882.0, + "step": 7150 + }, + { + "epoch": 0.9096807022007378, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8006724119186401, + "num_tokens": 272809614.0, + "step": 7151 + }, + { + "epoch": 0.9098079124793284, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7928082942962646, + "num_tokens": 272844819.0, + "step": 7152 + }, + { + "epoch": 0.9099351227579189, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8034852743148804, + "num_tokens": 272882798.0, + "step": 7153 + }, + { + "epoch": 0.9100623330365093, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7933468222618103, + "num_tokens": 272917472.0, + "step": 7154 + }, + { + "epoch": 0.9101895433150998, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7873746752738953, + "num_tokens": 272959947.0, + "step": 7155 + }, + { + "epoch": 0.9103167535936904, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7767448425292969, + "num_tokens": 272998194.0, + "step": 7156 + }, + { + "epoch": 0.9104439638722809, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8169189691543579, + "num_tokens": 273034688.0, + "step": 7157 + }, + { + "epoch": 0.9105711741508714, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8111438751220703, + "num_tokens": 273071575.0, + "step": 7158 + }, + { + "epoch": 0.9106983844294619, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7745996117591858, + "num_tokens": 273106106.0, + "step": 7159 + }, + { + "epoch": 0.9108255947080524, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7903162240982056, + "num_tokens": 273142959.0, + "step": 7160 + }, + { + "epoch": 0.9109528049866429, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7814507484436035, + "num_tokens": 273182781.0, + "step": 7161 + }, + { + "epoch": 0.9110800152652334, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8088828325271606, + "num_tokens": 273225738.0, + "step": 7162 + }, + { + "epoch": 0.9112072255438239, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8149511814117432, + "num_tokens": 273269166.0, + "step": 7163 + }, + { + "epoch": 0.9113344358224145, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7911137938499451, + "num_tokens": 273309093.0, + "step": 7164 + }, + { + "epoch": 0.911461646101005, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7978077530860901, + "num_tokens": 273346857.0, + "step": 7165 + }, + { + "epoch": 0.9115888563795955, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7972962856292725, + "num_tokens": 273385720.0, + "step": 7166 + }, + { + "epoch": 0.9117160666581859, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7989661693572998, + "num_tokens": 273431079.0, + "step": 7167 + }, + { + "epoch": 0.9118432769367765, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.8016846776008606, + "num_tokens": 273477316.0, + "step": 7168 + }, + { + "epoch": 0.911970487215367, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.80662602186203, + "num_tokens": 273512831.0, + "step": 7169 + }, + { + "epoch": 0.9120976974939575, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7947005033493042, + "num_tokens": 273551927.0, + "step": 7170 + }, + { + "epoch": 0.912224907772548, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7970433235168457, + "num_tokens": 273583053.0, + "step": 7171 + }, + { + "epoch": 0.9123521180511386, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7836816906929016, + "num_tokens": 273617953.0, + "step": 7172 + }, + { + "epoch": 0.912479328329729, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7802459001541138, + "num_tokens": 273649959.0, + "step": 7173 + }, + { + "epoch": 0.9126065386083195, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.783694326877594, + "num_tokens": 273682666.0, + "step": 7174 + }, + { + "epoch": 0.9127337488869101, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8186657428741455, + "num_tokens": 273718641.0, + "step": 7175 + }, + { + "epoch": 0.9128609591655006, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8118250370025635, + "num_tokens": 273760502.0, + "step": 7176 + }, + { + "epoch": 0.9129881694440911, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7906707525253296, + "num_tokens": 273795068.0, + "step": 7177 + }, + { + "epoch": 0.9131153797226816, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.775007426738739, + "num_tokens": 273825099.0, + "step": 7178 + }, + { + "epoch": 0.9132425900012721, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8087863922119141, + "num_tokens": 273862566.0, + "step": 7179 + }, + { + "epoch": 0.9133698002798626, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7957242727279663, + "num_tokens": 273896107.0, + "step": 7180 + }, + { + "epoch": 0.9134970105584531, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8090564012527466, + "num_tokens": 273934066.0, + "step": 7181 + }, + { + "epoch": 0.9136242208370436, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7944984436035156, + "num_tokens": 273971843.0, + "step": 7182 + }, + { + "epoch": 0.9137514311156342, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.784900426864624, + "num_tokens": 274001646.0, + "step": 7183 + }, + { + "epoch": 0.9138786413942247, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.783196747303009, + "num_tokens": 274042656.0, + "step": 7184 + }, + { + "epoch": 0.9140058516728151, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7849445343017578, + "num_tokens": 274084609.0, + "step": 7185 + }, + { + "epoch": 0.9141330619514056, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7999985218048096, + "num_tokens": 274127503.0, + "step": 7186 + }, + { + "epoch": 0.9142602722299962, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8013902902603149, + "num_tokens": 274164126.0, + "step": 7187 + }, + { + "epoch": 0.9143874825085867, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7985005378723145, + "num_tokens": 274198104.0, + "step": 7188 + }, + { + "epoch": 0.9145146927871772, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7873190641403198, + "num_tokens": 274233487.0, + "step": 7189 + }, + { + "epoch": 0.9146419030657678, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7941220998764038, + "num_tokens": 274267317.0, + "step": 7190 + }, + { + "epoch": 0.9147691133443582, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7873315811157227, + "num_tokens": 274304169.0, + "step": 7191 + }, + { + "epoch": 0.9148963236229487, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.8004985451698303, + "num_tokens": 274343265.0, + "step": 7192 + }, + { + "epoch": 0.9150235339015392, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7777324318885803, + "num_tokens": 274380756.0, + "step": 7193 + }, + { + "epoch": 0.9151507441801298, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8106355667114258, + "num_tokens": 274422035.0, + "step": 7194 + }, + { + "epoch": 0.9152779544587203, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7862845659255981, + "num_tokens": 274460875.0, + "step": 7195 + }, + { + "epoch": 0.9154051647373108, + "grad_norm": 76.5, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7952498197555542, + "num_tokens": 274495807.0, + "step": 7196 + }, + { + "epoch": 0.9155323750159012, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8055530786514282, + "num_tokens": 274533098.0, + "step": 7197 + }, + { + "epoch": 0.9156595852944918, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7910060286521912, + "num_tokens": 274565282.0, + "step": 7198 + }, + { + "epoch": 0.9157867955730823, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8017406463623047, + "num_tokens": 274605401.0, + "step": 7199 + }, + { + "epoch": 0.9159140058516728, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7920044660568237, + "num_tokens": 274645018.0, + "step": 7200 + }, + { + "epoch": 0.9160412161302633, + "grad_norm": 90.5, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8140808343887329, + "num_tokens": 274681541.0, + "step": 7201 + }, + { + "epoch": 0.9161684264088539, + "grad_norm": 82.0, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7991687059402466, + "num_tokens": 274716883.0, + "step": 7202 + }, + { + "epoch": 0.9162956366874443, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7894514799118042, + "num_tokens": 274753038.0, + "step": 7203 + }, + { + "epoch": 0.9164228469660348, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8012703061103821, + "num_tokens": 274790760.0, + "step": 7204 + }, + { + "epoch": 0.9165500572446253, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7915736436843872, + "num_tokens": 274830872.0, + "step": 7205 + }, + { + "epoch": 0.9166772675232159, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7841362953186035, + "num_tokens": 274869832.0, + "step": 7206 + }, + { + "epoch": 0.9168044778018064, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7861223816871643, + "num_tokens": 274903853.0, + "step": 7207 + }, + { + "epoch": 0.9169316880803969, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7941731214523315, + "num_tokens": 274942911.0, + "step": 7208 + }, + { + "epoch": 0.9170588983589874, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.8020914793014526, + "num_tokens": 274986427.0, + "step": 7209 + }, + { + "epoch": 0.9171861086375779, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7975997924804688, + "num_tokens": 275021557.0, + "step": 7210 + }, + { + "epoch": 0.9173133189161684, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7836931943893433, + "num_tokens": 275061308.0, + "step": 7211 + }, + { + "epoch": 0.9174405291947589, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7911421656608582, + "num_tokens": 275101497.0, + "step": 7212 + }, + { + "epoch": 0.9175677394733495, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7814836502075195, + "num_tokens": 275138317.0, + "step": 7213 + }, + { + "epoch": 0.91769494975194, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7972639799118042, + "num_tokens": 275180119.0, + "step": 7214 + }, + { + "epoch": 0.9178221600305305, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7954614162445068, + "num_tokens": 275212083.0, + "step": 7215 + }, + { + "epoch": 0.9179493703091209, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7862516641616821, + "num_tokens": 275247872.0, + "step": 7216 + }, + { + "epoch": 0.9180765805877115, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8085120916366577, + "num_tokens": 275287022.0, + "step": 7217 + }, + { + "epoch": 0.918203790866302, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7959515452384949, + "num_tokens": 275322259.0, + "step": 7218 + }, + { + "epoch": 0.9183310011448925, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.809823751449585, + "num_tokens": 275357410.0, + "step": 7219 + }, + { + "epoch": 0.918458211423483, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8031173944473267, + "num_tokens": 275388720.0, + "step": 7220 + }, + { + "epoch": 0.9185854217020736, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8023263216018677, + "num_tokens": 275427699.0, + "step": 7221 + }, + { + "epoch": 0.918712631980664, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8031069040298462, + "num_tokens": 275472220.0, + "step": 7222 + }, + { + "epoch": 0.9188398422592545, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7854381203651428, + "num_tokens": 275506693.0, + "step": 7223 + }, + { + "epoch": 0.918967052537845, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.800174355506897, + "num_tokens": 275547295.0, + "step": 7224 + }, + { + "epoch": 0.9190942628164356, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7966921329498291, + "num_tokens": 275591552.0, + "step": 7225 + }, + { + "epoch": 0.9192214730950261, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7708589434623718, + "num_tokens": 275632197.0, + "step": 7226 + }, + { + "epoch": 0.9193486833736166, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7893351316452026, + "num_tokens": 275670680.0, + "step": 7227 + }, + { + "epoch": 0.919475893652207, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7950044870376587, + "num_tokens": 275709315.0, + "step": 7228 + }, + { + "epoch": 0.9196031039307976, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7969337701797485, + "num_tokens": 275740405.0, + "step": 7229 + }, + { + "epoch": 0.9197303142093881, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.8166688680648804, + "num_tokens": 275780915.0, + "step": 7230 + }, + { + "epoch": 0.9198575244879786, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7911679744720459, + "num_tokens": 275817832.0, + "step": 7231 + }, + { + "epoch": 0.9199847347665692, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8010533452033997, + "num_tokens": 275857621.0, + "step": 7232 + }, + { + "epoch": 0.9201119450451597, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7944698333740234, + "num_tokens": 275898917.0, + "step": 7233 + }, + { + "epoch": 0.9202391553237501, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7953651547431946, + "num_tokens": 275930151.0, + "step": 7234 + }, + { + "epoch": 0.9203663656023406, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8029356002807617, + "num_tokens": 275966901.0, + "step": 7235 + }, + { + "epoch": 0.9204935758809312, + "grad_norm": 85.0, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8159401416778564, + "num_tokens": 276002845.0, + "step": 7236 + }, + { + "epoch": 0.9206207861595217, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.788703978061676, + "num_tokens": 276038476.0, + "step": 7237 + }, + { + "epoch": 0.9207479964381122, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7940902709960938, + "num_tokens": 276077867.0, + "step": 7238 + }, + { + "epoch": 0.9208752067167028, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7895666360855103, + "num_tokens": 276123289.0, + "step": 7239 + }, + { + "epoch": 0.9210024169952932, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7794378995895386, + "num_tokens": 276157357.0, + "step": 7240 + }, + { + "epoch": 0.9211296272738837, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8210294246673584, + "num_tokens": 276195548.0, + "step": 7241 + }, + { + "epoch": 0.9212568375524742, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7962404489517212, + "num_tokens": 276233533.0, + "step": 7242 + }, + { + "epoch": 0.9213840478310648, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8002007603645325, + "num_tokens": 276274310.0, + "step": 7243 + }, + { + "epoch": 0.9215112581096553, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7959777116775513, + "num_tokens": 276310272.0, + "step": 7244 + }, + { + "epoch": 0.9216384683882458, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7898315191268921, + "num_tokens": 276350334.0, + "step": 7245 + }, + { + "epoch": 0.9217656786668362, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7941678166389465, + "num_tokens": 276393913.0, + "step": 7246 + }, + { + "epoch": 0.9218928889454268, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.790791392326355, + "num_tokens": 276430368.0, + "step": 7247 + }, + { + "epoch": 0.9220200992240173, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7913621664047241, + "num_tokens": 276470930.0, + "step": 7248 + }, + { + "epoch": 0.9221473095026078, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8045378923416138, + "num_tokens": 276509359.0, + "step": 7249 + }, + { + "epoch": 0.9222745197811983, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7889131903648376, + "num_tokens": 276546980.0, + "step": 7250 + }, + { + "epoch": 0.9224017300597889, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7830852270126343, + "num_tokens": 276586212.0, + "step": 7251 + }, + { + "epoch": 0.9225289403383793, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.80495285987854, + "num_tokens": 276623946.0, + "step": 7252 + }, + { + "epoch": 0.9226561506169698, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7936310768127441, + "num_tokens": 276667567.0, + "step": 7253 + }, + { + "epoch": 0.9227833608955603, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7913508415222168, + "num_tokens": 276708479.0, + "step": 7254 + }, + { + "epoch": 0.9229105711741509, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7942839860916138, + "num_tokens": 276744627.0, + "step": 7255 + }, + { + "epoch": 0.9230377814527414, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7997643947601318, + "num_tokens": 276782830.0, + "step": 7256 + }, + { + "epoch": 0.9231649917313319, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7974170446395874, + "num_tokens": 276825745.0, + "step": 7257 + }, + { + "epoch": 0.9232922020099223, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8030624389648438, + "num_tokens": 276866743.0, + "step": 7258 + }, + { + "epoch": 0.9234194122885129, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7890855669975281, + "num_tokens": 276903884.0, + "step": 7259 + }, + { + "epoch": 0.9235466225671034, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7991182208061218, + "num_tokens": 276939142.0, + "step": 7260 + }, + { + "epoch": 0.9236738328456939, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.8009969592094421, + "num_tokens": 276979041.0, + "step": 7261 + }, + { + "epoch": 0.9238010431242845, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8095449209213257, + "num_tokens": 277015590.0, + "step": 7262 + }, + { + "epoch": 0.923928253402875, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7871493697166443, + "num_tokens": 277055745.0, + "step": 7263 + }, + { + "epoch": 0.9240554636814655, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7925726175308228, + "num_tokens": 277098856.0, + "step": 7264 + }, + { + "epoch": 0.9241826739600559, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8077827095985413, + "num_tokens": 277139165.0, + "step": 7265 + }, + { + "epoch": 0.9243098842386465, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7971219420433044, + "num_tokens": 277171879.0, + "step": 7266 + }, + { + "epoch": 0.924437094517237, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.795423686504364, + "num_tokens": 277211719.0, + "step": 7267 + }, + { + "epoch": 0.9245643047958275, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7894642353057861, + "num_tokens": 277252675.0, + "step": 7268 + }, + { + "epoch": 0.924691515074418, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7793965935707092, + "num_tokens": 277287529.0, + "step": 7269 + }, + { + "epoch": 0.9248187253530086, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7941361665725708, + "num_tokens": 277328392.0, + "step": 7270 + }, + { + "epoch": 0.924945935631599, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8120821118354797, + "num_tokens": 277368412.0, + "step": 7271 + }, + { + "epoch": 0.9250731459101895, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8012129068374634, + "num_tokens": 277404373.0, + "step": 7272 + }, + { + "epoch": 0.92520035618878, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7953723669052124, + "num_tokens": 277444988.0, + "step": 7273 + }, + { + "epoch": 0.9253275664673706, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8067219257354736, + "num_tokens": 277487501.0, + "step": 7274 + }, + { + "epoch": 0.9254547767459611, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7907766103744507, + "num_tokens": 277529056.0, + "step": 7275 + }, + { + "epoch": 0.9255819870245516, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8055393695831299, + "num_tokens": 277563022.0, + "step": 7276 + }, + { + "epoch": 0.925709197303142, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.808222770690918, + "num_tokens": 277603879.0, + "step": 7277 + }, + { + "epoch": 0.9258364075817326, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7923395037651062, + "num_tokens": 277643711.0, + "step": 7278 + }, + { + "epoch": 0.9259636178603231, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7957378029823303, + "num_tokens": 277680667.0, + "step": 7279 + }, + { + "epoch": 0.9260908281389136, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7959369421005249, + "num_tokens": 277715548.0, + "step": 7280 + }, + { + "epoch": 0.9262180384175042, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7949917316436768, + "num_tokens": 277745287.0, + "step": 7281 + }, + { + "epoch": 0.9263452486960947, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7933753728866577, + "num_tokens": 277781484.0, + "step": 7282 + }, + { + "epoch": 0.9264724589746851, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8007130026817322, + "num_tokens": 277818429.0, + "step": 7283 + }, + { + "epoch": 0.9265996692532756, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.802531361579895, + "num_tokens": 277853077.0, + "step": 7284 + }, + { + "epoch": 0.9267268795318662, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8169729709625244, + "num_tokens": 277893997.0, + "step": 7285 + }, + { + "epoch": 0.9268540898104567, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7866714000701904, + "num_tokens": 277934680.0, + "step": 7286 + }, + { + "epoch": 0.9269813000890472, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7910480499267578, + "num_tokens": 277980370.0, + "step": 7287 + }, + { + "epoch": 0.9271085103676378, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8109543323516846, + "num_tokens": 278017257.0, + "step": 7288 + }, + { + "epoch": 0.9272357206462282, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7965771555900574, + "num_tokens": 278054835.0, + "step": 7289 + }, + { + "epoch": 0.9273629309248187, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7957292795181274, + "num_tokens": 278089510.0, + "step": 7290 + }, + { + "epoch": 0.9274901412034092, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7955540418624878, + "num_tokens": 278126179.0, + "step": 7291 + }, + { + "epoch": 0.9276173514819998, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.770426869392395, + "num_tokens": 278170837.0, + "step": 7292 + }, + { + "epoch": 0.9277445617605903, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7967936396598816, + "num_tokens": 278207410.0, + "step": 7293 + }, + { + "epoch": 0.9278717720391808, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8201436400413513, + "num_tokens": 278240495.0, + "step": 7294 + }, + { + "epoch": 0.9279989823177712, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7885192632675171, + "num_tokens": 278278449.0, + "step": 7295 + }, + { + "epoch": 0.9281261925963618, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.807585597038269, + "num_tokens": 278321434.0, + "step": 7296 + }, + { + "epoch": 0.9282534028749523, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7916783094406128, + "num_tokens": 278363730.0, + "step": 7297 + }, + { + "epoch": 0.9283806131535428, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7906561493873596, + "num_tokens": 278395493.0, + "step": 7298 + }, + { + "epoch": 0.9285078234321333, + "grad_norm": 82.0, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.8002028465270996, + "num_tokens": 278433024.0, + "step": 7299 + }, + { + "epoch": 0.9286350337107239, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7982051968574524, + "num_tokens": 278471311.0, + "step": 7300 + }, + { + "epoch": 0.9287622439893143, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8182963132858276, + "num_tokens": 278509151.0, + "step": 7301 + }, + { + "epoch": 0.9288894542679048, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8045421838760376, + "num_tokens": 278548743.0, + "step": 7302 + }, + { + "epoch": 0.9290166645464953, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8087252378463745, + "num_tokens": 278590722.0, + "step": 7303 + }, + { + "epoch": 0.9291438748250859, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7829847931861877, + "num_tokens": 278628150.0, + "step": 7304 + }, + { + "epoch": 0.9292710851036764, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7824146151542664, + "num_tokens": 278672543.0, + "step": 7305 + }, + { + "epoch": 0.9293982953822669, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8028020858764648, + "num_tokens": 278707234.0, + "step": 7306 + }, + { + "epoch": 0.9295255056608573, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7960905432701111, + "num_tokens": 278734678.0, + "step": 7307 + }, + { + "epoch": 0.9296527159394479, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8071870803833008, + "num_tokens": 278778284.0, + "step": 7308 + }, + { + "epoch": 0.9297799262180384, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7944331169128418, + "num_tokens": 278815363.0, + "step": 7309 + }, + { + "epoch": 0.9299071364966289, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.787452220916748, + "num_tokens": 278860777.0, + "step": 7310 + }, + { + "epoch": 0.9300343467752195, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.811944842338562, + "num_tokens": 278890552.0, + "step": 7311 + }, + { + "epoch": 0.93016155705381, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7867369651794434, + "num_tokens": 278924741.0, + "step": 7312 + }, + { + "epoch": 0.9302887673324005, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8083407878875732, + "num_tokens": 278965185.0, + "step": 7313 + }, + { + "epoch": 0.9304159776109909, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7867216467857361, + "num_tokens": 278998961.0, + "step": 7314 + }, + { + "epoch": 0.9305431878895815, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7857078313827515, + "num_tokens": 279032689.0, + "step": 7315 + }, + { + "epoch": 0.930670398168172, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7882497310638428, + "num_tokens": 279080305.0, + "step": 7316 + }, + { + "epoch": 0.9307976084467625, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7980937957763672, + "num_tokens": 279120581.0, + "step": 7317 + }, + { + "epoch": 0.930924818725353, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8120694160461426, + "num_tokens": 279165050.0, + "step": 7318 + }, + { + "epoch": 0.9310520290039436, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8025805950164795, + "num_tokens": 279209393.0, + "step": 7319 + }, + { + "epoch": 0.931179239282534, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7807326316833496, + "num_tokens": 279250569.0, + "step": 7320 + }, + { + "epoch": 0.9313064495611245, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8032426238059998, + "num_tokens": 279291154.0, + "step": 7321 + }, + { + "epoch": 0.931433659839715, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7943992018699646, + "num_tokens": 279326967.0, + "step": 7322 + }, + { + "epoch": 0.9315608701183056, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8079308271408081, + "num_tokens": 279367129.0, + "step": 7323 + }, + { + "epoch": 0.9316880803968961, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8008203506469727, + "num_tokens": 279403651.0, + "step": 7324 + }, + { + "epoch": 0.9318152906754866, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7896817922592163, + "num_tokens": 279439166.0, + "step": 7325 + }, + { + "epoch": 0.931942500954077, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.789310872554779, + "num_tokens": 279473489.0, + "step": 7326 + }, + { + "epoch": 0.9320697112326676, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8024626970291138, + "num_tokens": 279511709.0, + "step": 7327 + }, + { + "epoch": 0.9321969215112581, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7901484966278076, + "num_tokens": 279551638.0, + "step": 7328 + }, + { + "epoch": 0.9323241317898486, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.807655930519104, + "num_tokens": 279587304.0, + "step": 7329 + }, + { + "epoch": 0.9324513420684392, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7858943939208984, + "num_tokens": 279628138.0, + "step": 7330 + }, + { + "epoch": 0.9325785523470297, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.799237847328186, + "num_tokens": 279663064.0, + "step": 7331 + }, + { + "epoch": 0.9327057626256201, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7928178310394287, + "num_tokens": 279704937.0, + "step": 7332 + }, + { + "epoch": 0.9328329729042106, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7959853410720825, + "num_tokens": 279738215.0, + "step": 7333 + }, + { + "epoch": 0.9329601831828012, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7712880373001099, + "num_tokens": 279771996.0, + "step": 7334 + }, + { + "epoch": 0.9330873934613917, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8155093193054199, + "num_tokens": 279810037.0, + "step": 7335 + }, + { + "epoch": 0.9332146037399822, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7902541756629944, + "num_tokens": 279847394.0, + "step": 7336 + }, + { + "epoch": 0.9333418140185727, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7793480157852173, + "num_tokens": 279879057.0, + "step": 7337 + }, + { + "epoch": 0.9334690242971632, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7847532033920288, + "num_tokens": 279911775.0, + "step": 7338 + }, + { + "epoch": 0.9335962345757537, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8007916212081909, + "num_tokens": 279947661.0, + "step": 7339 + }, + { + "epoch": 0.9337234448543442, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7962678670883179, + "num_tokens": 279977507.0, + "step": 7340 + }, + { + "epoch": 0.9338506551329347, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.795608401298523, + "num_tokens": 280016749.0, + "step": 7341 + }, + { + "epoch": 0.9339778654115253, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7982890009880066, + "num_tokens": 280056092.0, + "step": 7342 + }, + { + "epoch": 0.9341050756901158, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8091241121292114, + "num_tokens": 280089803.0, + "step": 7343 + }, + { + "epoch": 0.9342322859687062, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7798400521278381, + "num_tokens": 280126190.0, + "step": 7344 + }, + { + "epoch": 0.9343594962472968, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8080986738204956, + "num_tokens": 280162743.0, + "step": 7345 + }, + { + "epoch": 0.9344867065258873, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8021426200866699, + "num_tokens": 280203676.0, + "step": 7346 + }, + { + "epoch": 0.9346139168044778, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8151578903198242, + "num_tokens": 280241765.0, + "step": 7347 + }, + { + "epoch": 0.9347411270830683, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7767001390457153, + "num_tokens": 280280103.0, + "step": 7348 + }, + { + "epoch": 0.9348683373616589, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7784867882728577, + "num_tokens": 280314674.0, + "step": 7349 + }, + { + "epoch": 0.9349955476402493, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8019422888755798, + "num_tokens": 280353890.0, + "step": 7350 + }, + { + "epoch": 0.9351227579188398, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7928904891014099, + "num_tokens": 280396403.0, + "step": 7351 + }, + { + "epoch": 0.9352499681974303, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.799920916557312, + "num_tokens": 280436154.0, + "step": 7352 + }, + { + "epoch": 0.9353771784760209, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8049733638763428, + "num_tokens": 280476480.0, + "step": 7353 + }, + { + "epoch": 0.9355043887546114, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7985728979110718, + "num_tokens": 280505976.0, + "step": 7354 + }, + { + "epoch": 0.9356315990332019, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.797941267490387, + "num_tokens": 280548365.0, + "step": 7355 + }, + { + "epoch": 0.9357588093117923, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.819176435470581, + "num_tokens": 280588074.0, + "step": 7356 + }, + { + "epoch": 0.9358860195903829, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8105950951576233, + "num_tokens": 280626915.0, + "step": 7357 + }, + { + "epoch": 0.9360132298689734, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7901625037193298, + "num_tokens": 280660088.0, + "step": 7358 + }, + { + "epoch": 0.9361404401475639, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7984955906867981, + "num_tokens": 280701326.0, + "step": 7359 + }, + { + "epoch": 0.9362676504261545, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8074411153793335, + "num_tokens": 280740256.0, + "step": 7360 + }, + { + "epoch": 0.936394860704745, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8092410564422607, + "num_tokens": 280773515.0, + "step": 7361 + }, + { + "epoch": 0.9365220709833355, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7734425067901611, + "num_tokens": 280817207.0, + "step": 7362 + }, + { + "epoch": 0.9366492812619259, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7866638898849487, + "num_tokens": 280860253.0, + "step": 7363 + }, + { + "epoch": 0.9367764915405165, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7885957956314087, + "num_tokens": 280892634.0, + "step": 7364 + }, + { + "epoch": 0.936903701819107, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8084971904754639, + "num_tokens": 280930912.0, + "step": 7365 + }, + { + "epoch": 0.9370309120976975, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8033994436264038, + "num_tokens": 280971588.0, + "step": 7366 + }, + { + "epoch": 0.937158122376288, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.805389404296875, + "num_tokens": 281009898.0, + "step": 7367 + }, + { + "epoch": 0.9372853326548786, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7880998849868774, + "num_tokens": 281046726.0, + "step": 7368 + }, + { + "epoch": 0.937412542933469, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8094644546508789, + "num_tokens": 281085163.0, + "step": 7369 + }, + { + "epoch": 0.9375397532120595, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8056570291519165, + "num_tokens": 281128660.0, + "step": 7370 + }, + { + "epoch": 0.93766696349065, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7707852125167847, + "num_tokens": 281163780.0, + "step": 7371 + }, + { + "epoch": 0.9377941737692406, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8049606084823608, + "num_tokens": 281204247.0, + "step": 7372 + }, + { + "epoch": 0.9379213840478311, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.8181349039077759, + "num_tokens": 281239518.0, + "step": 7373 + }, + { + "epoch": 0.9380485943264216, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8068867325782776, + "num_tokens": 281275916.0, + "step": 7374 + }, + { + "epoch": 0.938175804605012, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8064553737640381, + "num_tokens": 281307615.0, + "step": 7375 + }, + { + "epoch": 0.9383030148836026, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8050680756568909, + "num_tokens": 281342677.0, + "step": 7376 + }, + { + "epoch": 0.9384302251621931, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7917255163192749, + "num_tokens": 281379545.0, + "step": 7377 + }, + { + "epoch": 0.9385574354407836, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.776774525642395, + "num_tokens": 281424169.0, + "step": 7378 + }, + { + "epoch": 0.9386846457193742, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.8220773935317993, + "num_tokens": 281461306.0, + "step": 7379 + }, + { + "epoch": 0.9388118559979647, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8076735734939575, + "num_tokens": 281498753.0, + "step": 7380 + }, + { + "epoch": 0.9389390662765551, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8095863461494446, + "num_tokens": 281537163.0, + "step": 7381 + }, + { + "epoch": 0.9390662765551456, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7931010127067566, + "num_tokens": 281577926.0, + "step": 7382 + }, + { + "epoch": 0.9391934868337362, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8188099265098572, + "num_tokens": 281621283.0, + "step": 7383 + }, + { + "epoch": 0.9393206971123267, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7984958291053772, + "num_tokens": 281658336.0, + "step": 7384 + }, + { + "epoch": 0.9394479073909172, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8033734560012817, + "num_tokens": 281695137.0, + "step": 7385 + }, + { + "epoch": 0.9395751176695077, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.8184410929679871, + "num_tokens": 281733570.0, + "step": 7386 + }, + { + "epoch": 0.9397023279480982, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8026891946792603, + "num_tokens": 281771163.0, + "step": 7387 + }, + { + "epoch": 0.9398295382266887, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8041785955429077, + "num_tokens": 281813715.0, + "step": 7388 + }, + { + "epoch": 0.9399567485052792, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.797986626625061, + "num_tokens": 281856921.0, + "step": 7389 + }, + { + "epoch": 0.9400839587838697, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.787125825881958, + "num_tokens": 281893273.0, + "step": 7390 + }, + { + "epoch": 0.9402111690624603, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8048423528671265, + "num_tokens": 281927452.0, + "step": 7391 + }, + { + "epoch": 0.9403383793410508, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.5783, + "mean_token_accuracy": 0.8222160339355469, + "num_tokens": 281965359.0, + "step": 7392 + }, + { + "epoch": 0.9404655896196412, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7947843670845032, + "num_tokens": 281999209.0, + "step": 7393 + }, + { + "epoch": 0.9405927998982317, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7873773574829102, + "num_tokens": 282042129.0, + "step": 7394 + }, + { + "epoch": 0.9407200101768223, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7928155660629272, + "num_tokens": 282087941.0, + "step": 7395 + }, + { + "epoch": 0.9408472204554128, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8023412823677063, + "num_tokens": 282123130.0, + "step": 7396 + }, + { + "epoch": 0.9409744307340033, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7938012480735779, + "num_tokens": 282161142.0, + "step": 7397 + }, + { + "epoch": 0.9411016410125939, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.5839, + "mean_token_accuracy": 0.822456955909729, + "num_tokens": 282196363.0, + "step": 7398 + }, + { + "epoch": 0.9412288512911843, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.775111198425293, + "num_tokens": 282233202.0, + "step": 7399 + }, + { + "epoch": 0.9413560615697748, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7957776784896851, + "num_tokens": 282270976.0, + "step": 7400 + }, + { + "epoch": 0.9414832718483653, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8117759823799133, + "num_tokens": 282312223.0, + "step": 7401 + }, + { + "epoch": 0.9416104821269559, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.783758819103241, + "num_tokens": 282350418.0, + "step": 7402 + }, + { + "epoch": 0.9417376924055464, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7954798936843872, + "num_tokens": 282380751.0, + "step": 7403 + }, + { + "epoch": 0.9418649026841369, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8120864629745483, + "num_tokens": 282420248.0, + "step": 7404 + }, + { + "epoch": 0.9419921129627273, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8035413026809692, + "num_tokens": 282457951.0, + "step": 7405 + }, + { + "epoch": 0.9421193232413179, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.80339115858078, + "num_tokens": 282507379.0, + "step": 7406 + }, + { + "epoch": 0.9422465335199084, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.786821186542511, + "num_tokens": 282546905.0, + "step": 7407 + }, + { + "epoch": 0.9423737437984989, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7706880569458008, + "num_tokens": 282591368.0, + "step": 7408 + }, + { + "epoch": 0.9425009540770894, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7887680530548096, + "num_tokens": 282617331.0, + "step": 7409 + }, + { + "epoch": 0.94262816435568, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8132017850875854, + "num_tokens": 282652346.0, + "step": 7410 + }, + { + "epoch": 0.9427553746342705, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8031542301177979, + "num_tokens": 282691564.0, + "step": 7411 + }, + { + "epoch": 0.9428825849128609, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8049980401992798, + "num_tokens": 282728186.0, + "step": 7412 + }, + { + "epoch": 0.9430097951914514, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7900719046592712, + "num_tokens": 282770491.0, + "step": 7413 + }, + { + "epoch": 0.943137005470042, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7918623089790344, + "num_tokens": 282809824.0, + "step": 7414 + }, + { + "epoch": 0.9432642157486325, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8045835494995117, + "num_tokens": 282846735.0, + "step": 7415 + }, + { + "epoch": 0.943391426027223, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8230946660041809, + "num_tokens": 282882242.0, + "step": 7416 + }, + { + "epoch": 0.9435186363058136, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7874420881271362, + "num_tokens": 282918774.0, + "step": 7417 + }, + { + "epoch": 0.943645846584404, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7925902009010315, + "num_tokens": 282957766.0, + "step": 7418 + }, + { + "epoch": 0.9437730568629945, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7899794578552246, + "num_tokens": 282995980.0, + "step": 7419 + }, + { + "epoch": 0.943900267141585, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.800169825553894, + "num_tokens": 283033148.0, + "step": 7420 + }, + { + "epoch": 0.9440274774201756, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8019403219223022, + "num_tokens": 283069781.0, + "step": 7421 + }, + { + "epoch": 0.9441546876987661, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.790727436542511, + "num_tokens": 283107270.0, + "step": 7422 + }, + { + "epoch": 0.9442818979773566, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8112446069717407, + "num_tokens": 283143571.0, + "step": 7423 + }, + { + "epoch": 0.944409108255947, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7999203205108643, + "num_tokens": 283179995.0, + "step": 7424 + }, + { + "epoch": 0.9445363185345376, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7978829145431519, + "num_tokens": 283218370.0, + "step": 7425 + }, + { + "epoch": 0.9446635288131281, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7859197854995728, + "num_tokens": 283247879.0, + "step": 7426 + }, + { + "epoch": 0.9447907390917186, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8137454390525818, + "num_tokens": 283285027.0, + "step": 7427 + }, + { + "epoch": 0.9449179493703092, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8034849762916565, + "num_tokens": 283324934.0, + "step": 7428 + }, + { + "epoch": 0.9450451596488997, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8065898418426514, + "num_tokens": 283357540.0, + "step": 7429 + }, + { + "epoch": 0.9451723699274901, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7962313890457153, + "num_tokens": 283393470.0, + "step": 7430 + }, + { + "epoch": 0.9452995802060806, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8045543432235718, + "num_tokens": 283429916.0, + "step": 7431 + }, + { + "epoch": 0.9454267904846712, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7825624346733093, + "num_tokens": 283467997.0, + "step": 7432 + }, + { + "epoch": 0.9455540007632617, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7588435411453247, + "num_tokens": 283504741.0, + "step": 7433 + }, + { + "epoch": 0.9456812110418522, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8031606674194336, + "num_tokens": 283534535.0, + "step": 7434 + }, + { + "epoch": 0.9458084213204427, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8119744062423706, + "num_tokens": 283571600.0, + "step": 7435 + }, + { + "epoch": 0.9459356315990332, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7836509943008423, + "num_tokens": 283605662.0, + "step": 7436 + }, + { + "epoch": 0.9460628418776237, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7983510494232178, + "num_tokens": 283642868.0, + "step": 7437 + }, + { + "epoch": 0.9461900521562142, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7841572761535645, + "num_tokens": 283675604.0, + "step": 7438 + }, + { + "epoch": 0.9463172624348047, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8206084966659546, + "num_tokens": 283709459.0, + "step": 7439 + }, + { + "epoch": 0.9464444727133953, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.8005845546722412, + "num_tokens": 283748020.0, + "step": 7440 + }, + { + "epoch": 0.9465716829919858, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7923586368560791, + "num_tokens": 283778514.0, + "step": 7441 + }, + { + "epoch": 0.9466988932705762, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7993250489234924, + "num_tokens": 283819764.0, + "step": 7442 + }, + { + "epoch": 0.9468261035491667, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7848336100578308, + "num_tokens": 283853897.0, + "step": 7443 + }, + { + "epoch": 0.9469533138277573, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8139051198959351, + "num_tokens": 283891826.0, + "step": 7444 + }, + { + "epoch": 0.9470805241063478, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7973871827125549, + "num_tokens": 283931690.0, + "step": 7445 + }, + { + "epoch": 0.9472077343849383, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7983594536781311, + "num_tokens": 283971107.0, + "step": 7446 + }, + { + "epoch": 0.9473349446635289, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8039224743843079, + "num_tokens": 284005286.0, + "step": 7447 + }, + { + "epoch": 0.9474621549421193, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.8164125680923462, + "num_tokens": 284044089.0, + "step": 7448 + }, + { + "epoch": 0.9475893652207098, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7891018390655518, + "num_tokens": 284081172.0, + "step": 7449 + }, + { + "epoch": 0.9477165754993003, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.804935872554779, + "num_tokens": 284113362.0, + "step": 7450 + }, + { + "epoch": 0.9478437857778909, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8109045028686523, + "num_tokens": 284155272.0, + "step": 7451 + }, + { + "epoch": 0.9479709960564814, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8024306297302246, + "num_tokens": 284188218.0, + "step": 7452 + }, + { + "epoch": 0.9480982063350719, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.8056305050849915, + "num_tokens": 284228838.0, + "step": 7453 + }, + { + "epoch": 0.9482254166136623, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7944427728652954, + "num_tokens": 284265570.0, + "step": 7454 + }, + { + "epoch": 0.9483526268922529, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7932546734809875, + "num_tokens": 284307062.0, + "step": 7455 + }, + { + "epoch": 0.9484798371708434, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7934476137161255, + "num_tokens": 284340488.0, + "step": 7456 + }, + { + "epoch": 0.9486070474494339, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.8013010025024414, + "num_tokens": 284382399.0, + "step": 7457 + }, + { + "epoch": 0.9487342577280244, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.781004786491394, + "num_tokens": 284416391.0, + "step": 7458 + }, + { + "epoch": 0.948861468006615, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8063013553619385, + "num_tokens": 284457049.0, + "step": 7459 + }, + { + "epoch": 0.9489886782852054, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7973723411560059, + "num_tokens": 284494426.0, + "step": 7460 + }, + { + "epoch": 0.9491158885637959, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.8196715116500854, + "num_tokens": 284533349.0, + "step": 7461 + }, + { + "epoch": 0.9492430988423864, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8044753074645996, + "num_tokens": 284567777.0, + "step": 7462 + }, + { + "epoch": 0.949370309120977, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7953556776046753, + "num_tokens": 284609120.0, + "step": 7463 + }, + { + "epoch": 0.9494975193995675, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.794434666633606, + "num_tokens": 284649221.0, + "step": 7464 + }, + { + "epoch": 0.949624729678158, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8012210726737976, + "num_tokens": 284682595.0, + "step": 7465 + }, + { + "epoch": 0.9497519399567486, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.793084979057312, + "num_tokens": 284722903.0, + "step": 7466 + }, + { + "epoch": 0.949879150235339, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8027453422546387, + "num_tokens": 284759166.0, + "step": 7467 + }, + { + "epoch": 0.9500063605139295, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7977362871170044, + "num_tokens": 284803192.0, + "step": 7468 + }, + { + "epoch": 0.95013357079252, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8043766021728516, + "num_tokens": 284837088.0, + "step": 7469 + }, + { + "epoch": 0.9502607810711106, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7784621715545654, + "num_tokens": 284879641.0, + "step": 7470 + }, + { + "epoch": 0.9503879913497011, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7897065877914429, + "num_tokens": 284917519.0, + "step": 7471 + }, + { + "epoch": 0.9505152016282916, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.786286473274231, + "num_tokens": 284950422.0, + "step": 7472 + }, + { + "epoch": 0.950642411906882, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8053693771362305, + "num_tokens": 284986392.0, + "step": 7473 + }, + { + "epoch": 0.9507696221854726, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.802558958530426, + "num_tokens": 285027684.0, + "step": 7474 + }, + { + "epoch": 0.9508968324640631, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7897956371307373, + "num_tokens": 285059551.0, + "step": 7475 + }, + { + "epoch": 0.9510240427426536, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.804236650466919, + "num_tokens": 285100197.0, + "step": 7476 + }, + { + "epoch": 0.9511512530212441, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7951000332832336, + "num_tokens": 285130115.0, + "step": 7477 + }, + { + "epoch": 0.9512784632998347, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8016100525856018, + "num_tokens": 285169069.0, + "step": 7478 + }, + { + "epoch": 0.9514056735784251, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7755177617073059, + "num_tokens": 285209971.0, + "step": 7479 + }, + { + "epoch": 0.9515328838570156, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7936300039291382, + "num_tokens": 285250024.0, + "step": 7480 + }, + { + "epoch": 0.9516600941356061, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7896674871444702, + "num_tokens": 285284900.0, + "step": 7481 + }, + { + "epoch": 0.9517873044141967, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8068462610244751, + "num_tokens": 285325657.0, + "step": 7482 + }, + { + "epoch": 0.9519145146927872, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7954054474830627, + "num_tokens": 285356362.0, + "step": 7483 + }, + { + "epoch": 0.9520417249713777, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7964170575141907, + "num_tokens": 285394321.0, + "step": 7484 + }, + { + "epoch": 0.9521689352499682, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8172425031661987, + "num_tokens": 285435260.0, + "step": 7485 + }, + { + "epoch": 0.9522961455285587, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8082733154296875, + "num_tokens": 285479449.0, + "step": 7486 + }, + { + "epoch": 0.9524233558071492, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7793468236923218, + "num_tokens": 285520492.0, + "step": 7487 + }, + { + "epoch": 0.9525505660857397, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7986471652984619, + "num_tokens": 285558410.0, + "step": 7488 + }, + { + "epoch": 0.9526777763643303, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.777810275554657, + "num_tokens": 285598871.0, + "step": 7489 + }, + { + "epoch": 0.9528049866429208, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7929726839065552, + "num_tokens": 285641374.0, + "step": 7490 + }, + { + "epoch": 0.9529321969215112, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8120687007904053, + "num_tokens": 285675416.0, + "step": 7491 + }, + { + "epoch": 0.9530594072001017, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8011302947998047, + "num_tokens": 285707491.0, + "step": 7492 + }, + { + "epoch": 0.9531866174786923, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.787641167640686, + "num_tokens": 285747571.0, + "step": 7493 + }, + { + "epoch": 0.9533138277572828, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5828, + "mean_token_accuracy": 0.823516845703125, + "num_tokens": 285786362.0, + "step": 7494 + }, + { + "epoch": 0.9534410380358733, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8110843300819397, + "num_tokens": 285829401.0, + "step": 7495 + }, + { + "epoch": 0.9535682483144639, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7917495965957642, + "num_tokens": 285867853.0, + "step": 7496 + }, + { + "epoch": 0.9536954585930543, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8115284442901611, + "num_tokens": 285908497.0, + "step": 7497 + }, + { + "epoch": 0.9538226688716448, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7850052118301392, + "num_tokens": 285940528.0, + "step": 7498 + }, + { + "epoch": 0.9539498791502353, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.8000105023384094, + "num_tokens": 285980206.0, + "step": 7499 + }, + { + "epoch": 0.9540770894288259, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7908939123153687, + "num_tokens": 286019409.0, + "step": 7500 + }, + { + "epoch": 0.9542042997074164, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8065890073776245, + "num_tokens": 286050955.0, + "step": 7501 + }, + { + "epoch": 0.9543315099860069, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8021931648254395, + "num_tokens": 286087567.0, + "step": 7502 + }, + { + "epoch": 0.9544587202645973, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7765524387359619, + "num_tokens": 286126893.0, + "step": 7503 + }, + { + "epoch": 0.9545859305431879, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.790064811706543, + "num_tokens": 286165818.0, + "step": 7504 + }, + { + "epoch": 0.9547131408217784, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.798600435256958, + "num_tokens": 286208026.0, + "step": 7505 + }, + { + "epoch": 0.9548403511003689, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8001648187637329, + "num_tokens": 286245468.0, + "step": 7506 + }, + { + "epoch": 0.9549675613789594, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8009251356124878, + "num_tokens": 286283955.0, + "step": 7507 + }, + { + "epoch": 0.95509477165755, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8094139099121094, + "num_tokens": 286326560.0, + "step": 7508 + }, + { + "epoch": 0.9552219819361404, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8001145124435425, + "num_tokens": 286365791.0, + "step": 7509 + }, + { + "epoch": 0.9553491922147309, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8100936412811279, + "num_tokens": 286404132.0, + "step": 7510 + }, + { + "epoch": 0.9554764024933214, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7871084213256836, + "num_tokens": 286444828.0, + "step": 7511 + }, + { + "epoch": 0.955603612771912, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8015725612640381, + "num_tokens": 286481725.0, + "step": 7512 + }, + { + "epoch": 0.9557308230505025, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.798850953578949, + "num_tokens": 286520354.0, + "step": 7513 + }, + { + "epoch": 0.955858033329093, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7971863746643066, + "num_tokens": 286564314.0, + "step": 7514 + }, + { + "epoch": 0.9559852436076836, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.795772910118103, + "num_tokens": 286606955.0, + "step": 7515 + }, + { + "epoch": 0.956112453886274, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7977332472801208, + "num_tokens": 286640091.0, + "step": 7516 + }, + { + "epoch": 0.9562396641648645, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7981058359146118, + "num_tokens": 286679090.0, + "step": 7517 + }, + { + "epoch": 0.956366874443455, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8019986152648926, + "num_tokens": 286720260.0, + "step": 7518 + }, + { + "epoch": 0.9564940847220456, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8078482151031494, + "num_tokens": 286763992.0, + "step": 7519 + }, + { + "epoch": 0.9566212950006361, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7818703651428223, + "num_tokens": 286807068.0, + "step": 7520 + }, + { + "epoch": 0.9567485052792266, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8068986535072327, + "num_tokens": 286853020.0, + "step": 7521 + }, + { + "epoch": 0.956875715557817, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7873925566673279, + "num_tokens": 286896098.0, + "step": 7522 + }, + { + "epoch": 0.9570029258364076, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7855676412582397, + "num_tokens": 286938873.0, + "step": 7523 + }, + { + "epoch": 0.9571301361149981, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.799607515335083, + "num_tokens": 286972941.0, + "step": 7524 + }, + { + "epoch": 0.9572573463935886, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7824230194091797, + "num_tokens": 287010818.0, + "step": 7525 + }, + { + "epoch": 0.9573845566721791, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8011130094528198, + "num_tokens": 287053841.0, + "step": 7526 + }, + { + "epoch": 0.9575117669507697, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8026672005653381, + "num_tokens": 287090900.0, + "step": 7527 + }, + { + "epoch": 0.9576389772293601, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7957676649093628, + "num_tokens": 287131203.0, + "step": 7528 + }, + { + "epoch": 0.9577661875079506, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8079037070274353, + "num_tokens": 287165949.0, + "step": 7529 + }, + { + "epoch": 0.9578933977865411, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7981061935424805, + "num_tokens": 287203376.0, + "step": 7530 + }, + { + "epoch": 0.9580206080651317, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7903137803077698, + "num_tokens": 287240391.0, + "step": 7531 + }, + { + "epoch": 0.9581478183437222, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.795971155166626, + "num_tokens": 287278485.0, + "step": 7532 + }, + { + "epoch": 0.9582750286223127, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7945079803466797, + "num_tokens": 287313626.0, + "step": 7533 + }, + { + "epoch": 0.9584022389009031, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8137945532798767, + "num_tokens": 287344532.0, + "step": 7534 + }, + { + "epoch": 0.9585294491794937, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7802349925041199, + "num_tokens": 287383069.0, + "step": 7535 + }, + { + "epoch": 0.9586566594580842, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.8009132146835327, + "num_tokens": 287418554.0, + "step": 7536 + }, + { + "epoch": 0.9587838697366747, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7989406585693359, + "num_tokens": 287452449.0, + "step": 7537 + }, + { + "epoch": 0.9589110800152653, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7859939336776733, + "num_tokens": 287492247.0, + "step": 7538 + }, + { + "epoch": 0.9590382902938558, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7990911602973938, + "num_tokens": 287528651.0, + "step": 7539 + }, + { + "epoch": 0.9591655005724462, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7914324998855591, + "num_tokens": 287565837.0, + "step": 7540 + }, + { + "epoch": 0.9592927108510367, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7975567579269409, + "num_tokens": 287599469.0, + "step": 7541 + }, + { + "epoch": 0.9594199211296273, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7963228225708008, + "num_tokens": 287637104.0, + "step": 7542 + }, + { + "epoch": 0.9595471314082178, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8041256666183472, + "num_tokens": 287679134.0, + "step": 7543 + }, + { + "epoch": 0.9596743416868083, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8014050126075745, + "num_tokens": 287720340.0, + "step": 7544 + }, + { + "epoch": 0.9598015519653988, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.807917594909668, + "num_tokens": 287760931.0, + "step": 7545 + }, + { + "epoch": 0.9599287622439893, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7848331928253174, + "num_tokens": 287801723.0, + "step": 7546 + }, + { + "epoch": 0.9600559725225798, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7906589508056641, + "num_tokens": 287837010.0, + "step": 7547 + }, + { + "epoch": 0.9601831828011703, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7987433671951294, + "num_tokens": 287874381.0, + "step": 7548 + }, + { + "epoch": 0.9603103930797608, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7916392683982849, + "num_tokens": 287916673.0, + "step": 7549 + }, + { + "epoch": 0.9604376033583514, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7940798401832581, + "num_tokens": 287956764.0, + "step": 7550 + }, + { + "epoch": 0.9605648136369419, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8073020577430725, + "num_tokens": 287996373.0, + "step": 7551 + }, + { + "epoch": 0.9606920239155323, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7902816534042358, + "num_tokens": 288033671.0, + "step": 7552 + }, + { + "epoch": 0.9608192341941229, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.811848521232605, + "num_tokens": 288068254.0, + "step": 7553 + }, + { + "epoch": 0.9609464444727134, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.783135175704956, + "num_tokens": 288102833.0, + "step": 7554 + }, + { + "epoch": 0.9610736547513039, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8121975064277649, + "num_tokens": 288135927.0, + "step": 7555 + }, + { + "epoch": 0.9612008650298944, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8182582855224609, + "num_tokens": 288173842.0, + "step": 7556 + }, + { + "epoch": 0.961328075308485, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8051684498786926, + "num_tokens": 288215007.0, + "step": 7557 + }, + { + "epoch": 0.9614552855870754, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7845423221588135, + "num_tokens": 288251598.0, + "step": 7558 + }, + { + "epoch": 0.9615824958656659, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8051048517227173, + "num_tokens": 288292811.0, + "step": 7559 + }, + { + "epoch": 0.9617097061442564, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8077870607376099, + "num_tokens": 288337790.0, + "step": 7560 + }, + { + "epoch": 0.961836916422847, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.798573911190033, + "num_tokens": 288373347.0, + "step": 7561 + }, + { + "epoch": 0.9619641267014375, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.8193197250366211, + "num_tokens": 288405650.0, + "step": 7562 + }, + { + "epoch": 0.962091336980028, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8046625852584839, + "num_tokens": 288443849.0, + "step": 7563 + }, + { + "epoch": 0.9622185472586186, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7795843482017517, + "num_tokens": 288487557.0, + "step": 7564 + }, + { + "epoch": 0.962345757537209, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7906407117843628, + "num_tokens": 288529932.0, + "step": 7565 + }, + { + "epoch": 0.9624729678157995, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7931575775146484, + "num_tokens": 288568276.0, + "step": 7566 + }, + { + "epoch": 0.96260017809439, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7992357015609741, + "num_tokens": 288602387.0, + "step": 7567 + }, + { + "epoch": 0.9627273883729806, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8084553480148315, + "num_tokens": 288638157.0, + "step": 7568 + }, + { + "epoch": 0.9628545986515711, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7875601053237915, + "num_tokens": 288675054.0, + "step": 7569 + }, + { + "epoch": 0.9629818089301616, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8069497346878052, + "num_tokens": 288714222.0, + "step": 7570 + }, + { + "epoch": 0.963109019208752, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7867803573608398, + "num_tokens": 288752772.0, + "step": 7571 + }, + { + "epoch": 0.9632362294873426, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7795770764350891, + "num_tokens": 288792067.0, + "step": 7572 + }, + { + "epoch": 0.9633634397659331, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8056182861328125, + "num_tokens": 288826172.0, + "step": 7573 + }, + { + "epoch": 0.9634906500445236, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8088919520378113, + "num_tokens": 288863774.0, + "step": 7574 + }, + { + "epoch": 0.9636178603231141, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7932813167572021, + "num_tokens": 288901892.0, + "step": 7575 + }, + { + "epoch": 0.9637450706017047, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7895314693450928, + "num_tokens": 288941470.0, + "step": 7576 + }, + { + "epoch": 0.9638722808802951, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7833541631698608, + "num_tokens": 288988616.0, + "step": 7577 + }, + { + "epoch": 0.9639994911588856, + "grad_norm": 9.8125, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.8165005445480347, + "num_tokens": 289024589.0, + "step": 7578 + }, + { + "epoch": 0.9641267014374761, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.8033652305603027, + "num_tokens": 289064334.0, + "step": 7579 + }, + { + "epoch": 0.9642539117160667, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7811117172241211, + "num_tokens": 289099385.0, + "step": 7580 + }, + { + "epoch": 0.9643811219946572, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8020921945571899, + "num_tokens": 289139143.0, + "step": 7581 + }, + { + "epoch": 0.9645083322732477, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7969603538513184, + "num_tokens": 289179303.0, + "step": 7582 + }, + { + "epoch": 0.9646355425518381, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7835787534713745, + "num_tokens": 289214218.0, + "step": 7583 + }, + { + "epoch": 0.9647627528304287, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7878156900405884, + "num_tokens": 289240991.0, + "step": 7584 + }, + { + "epoch": 0.9648899631090192, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7918049097061157, + "num_tokens": 289273374.0, + "step": 7585 + }, + { + "epoch": 0.9650171733876097, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7858718633651733, + "num_tokens": 289314904.0, + "step": 7586 + }, + { + "epoch": 0.9651443836662003, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7921266555786133, + "num_tokens": 289358110.0, + "step": 7587 + }, + { + "epoch": 0.9652715939447908, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7931202054023743, + "num_tokens": 289391581.0, + "step": 7588 + }, + { + "epoch": 0.9653988042233812, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.769933819770813, + "num_tokens": 289432280.0, + "step": 7589 + }, + { + "epoch": 0.9655260145019717, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7829782366752625, + "num_tokens": 289473222.0, + "step": 7590 + }, + { + "epoch": 0.9656532247805623, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.789925217628479, + "num_tokens": 289512898.0, + "step": 7591 + }, + { + "epoch": 0.9657804350591528, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8035991787910461, + "num_tokens": 289554394.0, + "step": 7592 + }, + { + "epoch": 0.9659076453377433, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7950540781021118, + "num_tokens": 289594871.0, + "step": 7593 + }, + { + "epoch": 0.9660348556163338, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7850992679595947, + "num_tokens": 289629165.0, + "step": 7594 + }, + { + "epoch": 0.9661620658949243, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7961246967315674, + "num_tokens": 289670131.0, + "step": 7595 + }, + { + "epoch": 0.9662892761735148, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8001495003700256, + "num_tokens": 289705238.0, + "step": 7596 + }, + { + "epoch": 0.9664164864521053, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8019093871116638, + "num_tokens": 289740146.0, + "step": 7597 + }, + { + "epoch": 0.9665436967306958, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8045825958251953, + "num_tokens": 289781145.0, + "step": 7598 + }, + { + "epoch": 0.9666709070092864, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7765198945999146, + "num_tokens": 289816347.0, + "step": 7599 + }, + { + "epoch": 0.9667981172878769, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7808244228363037, + "num_tokens": 289852974.0, + "step": 7600 + }, + { + "epoch": 0.9669253275664673, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.78656005859375, + "num_tokens": 289894553.0, + "step": 7601 + }, + { + "epoch": 0.9670525378450578, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7765030860900879, + "num_tokens": 289930641.0, + "step": 7602 + }, + { + "epoch": 0.9671797481236484, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.795052170753479, + "num_tokens": 289972690.0, + "step": 7603 + }, + { + "epoch": 0.9673069584022389, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7893012762069702, + "num_tokens": 290006518.0, + "step": 7604 + }, + { + "epoch": 0.9674341686808294, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7818282842636108, + "num_tokens": 290045206.0, + "step": 7605 + }, + { + "epoch": 0.96756137895942, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8031859993934631, + "num_tokens": 290086976.0, + "step": 7606 + }, + { + "epoch": 0.9676885892380104, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8131245374679565, + "num_tokens": 290126894.0, + "step": 7607 + }, + { + "epoch": 0.9678157995166009, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8147022724151611, + "num_tokens": 290161684.0, + "step": 7608 + }, + { + "epoch": 0.9679430097951914, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7967296242713928, + "num_tokens": 290203407.0, + "step": 7609 + }, + { + "epoch": 0.968070220073782, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8050462007522583, + "num_tokens": 290238645.0, + "step": 7610 + }, + { + "epoch": 0.9681974303523725, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7991195321083069, + "num_tokens": 290280279.0, + "step": 7611 + }, + { + "epoch": 0.968324640630963, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7974011301994324, + "num_tokens": 290317739.0, + "step": 7612 + }, + { + "epoch": 0.9684518509095535, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7962572574615479, + "num_tokens": 290360384.0, + "step": 7613 + }, + { + "epoch": 0.968579061188144, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.8003378510475159, + "num_tokens": 290398706.0, + "step": 7614 + }, + { + "epoch": 0.9687062714667345, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8178043365478516, + "num_tokens": 290439524.0, + "step": 7615 + }, + { + "epoch": 0.968833481745325, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7880237698554993, + "num_tokens": 290472125.0, + "step": 7616 + }, + { + "epoch": 0.9689606920239155, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7860933542251587, + "num_tokens": 290514348.0, + "step": 7617 + }, + { + "epoch": 0.9690879023025061, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7914046049118042, + "num_tokens": 290551699.0, + "step": 7618 + }, + { + "epoch": 0.9692151125810966, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7988072037696838, + "num_tokens": 290588644.0, + "step": 7619 + }, + { + "epoch": 0.969342322859687, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7994124889373779, + "num_tokens": 290629546.0, + "step": 7620 + }, + { + "epoch": 0.9694695331382776, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.800693154335022, + "num_tokens": 290665527.0, + "step": 7621 + }, + { + "epoch": 0.9695967434168681, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7911844253540039, + "num_tokens": 290706667.0, + "step": 7622 + }, + { + "epoch": 0.9697239536954586, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7920801639556885, + "num_tokens": 290745698.0, + "step": 7623 + }, + { + "epoch": 0.9698511639740491, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.80092853307724, + "num_tokens": 290777983.0, + "step": 7624 + }, + { + "epoch": 0.9699783742526397, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7923489809036255, + "num_tokens": 290814184.0, + "step": 7625 + }, + { + "epoch": 0.9701055845312301, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7893004417419434, + "num_tokens": 290854054.0, + "step": 7626 + }, + { + "epoch": 0.9702327948098206, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7929763793945312, + "num_tokens": 290891325.0, + "step": 7627 + }, + { + "epoch": 0.9703600050884111, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7924865484237671, + "num_tokens": 290930101.0, + "step": 7628 + }, + { + "epoch": 0.9704872153670017, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7916769981384277, + "num_tokens": 290964691.0, + "step": 7629 + }, + { + "epoch": 0.9706144256455922, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.8000084161758423, + "num_tokens": 291008140.0, + "step": 7630 + }, + { + "epoch": 0.9707416359241827, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7897883653640747, + "num_tokens": 291041484.0, + "step": 7631 + }, + { + "epoch": 0.9708688462027731, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.783350944519043, + "num_tokens": 291082337.0, + "step": 7632 + }, + { + "epoch": 0.9709960564813637, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7932071685791016, + "num_tokens": 291115921.0, + "step": 7633 + }, + { + "epoch": 0.9711232667599542, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8053966760635376, + "num_tokens": 291154367.0, + "step": 7634 + }, + { + "epoch": 0.9712504770385447, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.797243595123291, + "num_tokens": 291198919.0, + "step": 7635 + }, + { + "epoch": 0.9713776873171353, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.78801429271698, + "num_tokens": 291237062.0, + "step": 7636 + }, + { + "epoch": 0.9715048975957258, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.812403678894043, + "num_tokens": 291275676.0, + "step": 7637 + }, + { + "epoch": 0.9716321078743162, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7889895439147949, + "num_tokens": 291316785.0, + "step": 7638 + }, + { + "epoch": 0.9717593181529067, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7951090931892395, + "num_tokens": 291360448.0, + "step": 7639 + }, + { + "epoch": 0.9718865284314973, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7955396175384521, + "num_tokens": 291401650.0, + "step": 7640 + }, + { + "epoch": 0.9720137387100878, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7975679039955139, + "num_tokens": 291439736.0, + "step": 7641 + }, + { + "epoch": 0.9721409489886783, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8028650283813477, + "num_tokens": 291479077.0, + "step": 7642 + }, + { + "epoch": 0.9722681592672688, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8046389222145081, + "num_tokens": 291512470.0, + "step": 7643 + }, + { + "epoch": 0.9723953695458593, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8128628730773926, + "num_tokens": 291552797.0, + "step": 7644 + }, + { + "epoch": 0.9725225798244498, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7992675304412842, + "num_tokens": 291589885.0, + "step": 7645 + }, + { + "epoch": 0.9726497901030403, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7990545034408569, + "num_tokens": 291633005.0, + "step": 7646 + }, + { + "epoch": 0.9727770003816308, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7924249172210693, + "num_tokens": 291666083.0, + "step": 7647 + }, + { + "epoch": 0.9729042106602214, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7985050678253174, + "num_tokens": 291706421.0, + "step": 7648 + }, + { + "epoch": 0.9730314209388119, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8015010356903076, + "num_tokens": 291744641.0, + "step": 7649 + }, + { + "epoch": 0.9731586312174023, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.579, + "mean_token_accuracy": 0.8232786655426025, + "num_tokens": 291780586.0, + "step": 7650 + }, + { + "epoch": 0.9732858414959928, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7940110564231873, + "num_tokens": 291820482.0, + "step": 7651 + }, + { + "epoch": 0.9734130517745834, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8091298341751099, + "num_tokens": 291858568.0, + "step": 7652 + }, + { + "epoch": 0.9735402620531739, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7882204651832581, + "num_tokens": 291889383.0, + "step": 7653 + }, + { + "epoch": 0.9736674723317644, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8111023306846619, + "num_tokens": 291927541.0, + "step": 7654 + }, + { + "epoch": 0.973794682610355, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.784964919090271, + "num_tokens": 291969074.0, + "step": 7655 + }, + { + "epoch": 0.9739218928889454, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7955480217933655, + "num_tokens": 292008311.0, + "step": 7656 + }, + { + "epoch": 0.9740491031675359, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7909698486328125, + "num_tokens": 292051810.0, + "step": 7657 + }, + { + "epoch": 0.9741763134461264, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8060831427574158, + "num_tokens": 292090411.0, + "step": 7658 + }, + { + "epoch": 0.974303523724717, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7867794036865234, + "num_tokens": 292124411.0, + "step": 7659 + }, + { + "epoch": 0.9744307340033075, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7976959943771362, + "num_tokens": 292166428.0, + "step": 7660 + }, + { + "epoch": 0.974557944281898, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8000197410583496, + "num_tokens": 292202687.0, + "step": 7661 + }, + { + "epoch": 0.9746851545604885, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8065792322158813, + "num_tokens": 292246364.0, + "step": 7662 + }, + { + "epoch": 0.974812364839079, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7945328950881958, + "num_tokens": 292285789.0, + "step": 7663 + }, + { + "epoch": 0.9749395751176695, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7882826328277588, + "num_tokens": 292320755.0, + "step": 7664 + }, + { + "epoch": 0.97506678539626, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8109136819839478, + "num_tokens": 292362818.0, + "step": 7665 + }, + { + "epoch": 0.9751939956748505, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7849811911582947, + "num_tokens": 292404016.0, + "step": 7666 + }, + { + "epoch": 0.9753212059534411, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7826195359230042, + "num_tokens": 292442117.0, + "step": 7667 + }, + { + "epoch": 0.9754484162320316, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7997288703918457, + "num_tokens": 292472523.0, + "step": 7668 + }, + { + "epoch": 0.975575626510622, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.8005244135856628, + "num_tokens": 292508571.0, + "step": 7669 + }, + { + "epoch": 0.9757028367892125, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7960528135299683, + "num_tokens": 292545358.0, + "step": 7670 + }, + { + "epoch": 0.9758300470678031, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7804010510444641, + "num_tokens": 292583962.0, + "step": 7671 + }, + { + "epoch": 0.9759572573463936, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.796087384223938, + "num_tokens": 292616186.0, + "step": 7672 + }, + { + "epoch": 0.9760844676249841, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7981452345848083, + "num_tokens": 292657378.0, + "step": 7673 + }, + { + "epoch": 0.9762116779035747, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8093254566192627, + "num_tokens": 292692041.0, + "step": 7674 + }, + { + "epoch": 0.9763388881821651, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.797937273979187, + "num_tokens": 292725062.0, + "step": 7675 + }, + { + "epoch": 0.9764660984607556, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8030574321746826, + "num_tokens": 292766522.0, + "step": 7676 + }, + { + "epoch": 0.9765933087393461, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7981929779052734, + "num_tokens": 292811371.0, + "step": 7677 + }, + { + "epoch": 0.9767205190179367, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8026863932609558, + "num_tokens": 292850237.0, + "step": 7678 + }, + { + "epoch": 0.9768477292965272, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7979101538658142, + "num_tokens": 292887291.0, + "step": 7679 + }, + { + "epoch": 0.9769749395751177, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8149052262306213, + "num_tokens": 292928396.0, + "step": 7680 + }, + { + "epoch": 0.9771021498537081, + "grad_norm": 10.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7987880706787109, + "num_tokens": 292962748.0, + "step": 7681 + }, + { + "epoch": 0.9772293601322987, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.809227466583252, + "num_tokens": 293002330.0, + "step": 7682 + }, + { + "epoch": 0.9773565704108892, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7963303923606873, + "num_tokens": 293038288.0, + "step": 7683 + }, + { + "epoch": 0.9774837806894797, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7999231815338135, + "num_tokens": 293078121.0, + "step": 7684 + }, + { + "epoch": 0.9776109909680702, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7946152687072754, + "num_tokens": 293116846.0, + "step": 7685 + }, + { + "epoch": 0.9777382012466608, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7941263914108276, + "num_tokens": 293162237.0, + "step": 7686 + }, + { + "epoch": 0.9778654115252512, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7922064065933228, + "num_tokens": 293196631.0, + "step": 7687 + }, + { + "epoch": 0.9779926218038417, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8118823766708374, + "num_tokens": 293238883.0, + "step": 7688 + }, + { + "epoch": 0.9781198320824323, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7859419584274292, + "num_tokens": 293270802.0, + "step": 7689 + }, + { + "epoch": 0.9782470423610228, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7869518995285034, + "num_tokens": 293314910.0, + "step": 7690 + }, + { + "epoch": 0.9783742526396133, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.814348578453064, + "num_tokens": 293353083.0, + "step": 7691 + }, + { + "epoch": 0.9785014629182038, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8037976026535034, + "num_tokens": 293389473.0, + "step": 7692 + }, + { + "epoch": 0.9786286731967943, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.8153020143508911, + "num_tokens": 293426893.0, + "step": 7693 + }, + { + "epoch": 0.9787558834753848, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8043785095214844, + "num_tokens": 293468116.0, + "step": 7694 + }, + { + "epoch": 0.9788830937539753, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7994802594184875, + "num_tokens": 293503050.0, + "step": 7695 + }, + { + "epoch": 0.9790103040325658, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7887275218963623, + "num_tokens": 293541673.0, + "step": 7696 + }, + { + "epoch": 0.9791375143111564, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7920432686805725, + "num_tokens": 293578472.0, + "step": 7697 + }, + { + "epoch": 0.9792647245897469, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7940159440040588, + "num_tokens": 293611204.0, + "step": 7698 + }, + { + "epoch": 0.9793919348683373, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7876691818237305, + "num_tokens": 293651352.0, + "step": 7699 + }, + { + "epoch": 0.9795191451469278, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8086024522781372, + "num_tokens": 293687396.0, + "step": 7700 + }, + { + "epoch": 0.9796463554255184, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7899044752120972, + "num_tokens": 293725083.0, + "step": 7701 + }, + { + "epoch": 0.9797735657041089, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.799854040145874, + "num_tokens": 293766136.0, + "step": 7702 + }, + { + "epoch": 0.9799007759826994, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7755488753318787, + "num_tokens": 293805810.0, + "step": 7703 + }, + { + "epoch": 0.98002798626129, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7862995862960815, + "num_tokens": 293839041.0, + "step": 7704 + }, + { + "epoch": 0.9801551965398804, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8126769661903381, + "num_tokens": 293876541.0, + "step": 7705 + }, + { + "epoch": 0.9802824068184709, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8155872821807861, + "num_tokens": 293918141.0, + "step": 7706 + }, + { + "epoch": 0.9804096170970614, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7845948338508606, + "num_tokens": 293958078.0, + "step": 7707 + }, + { + "epoch": 0.980536827375652, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8022335767745972, + "num_tokens": 293990347.0, + "step": 7708 + }, + { + "epoch": 0.9806640376542425, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8083836436271667, + "num_tokens": 294032136.0, + "step": 7709 + }, + { + "epoch": 0.980791247932833, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7836698889732361, + "num_tokens": 294065136.0, + "step": 7710 + }, + { + "epoch": 0.9809184582114235, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8138880729675293, + "num_tokens": 294104539.0, + "step": 7711 + }, + { + "epoch": 0.981045668490014, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8009701371192932, + "num_tokens": 294148962.0, + "step": 7712 + }, + { + "epoch": 0.9811728787686045, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8037033677101135, + "num_tokens": 294181866.0, + "step": 7713 + }, + { + "epoch": 0.981300089047195, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8100249767303467, + "num_tokens": 294223558.0, + "step": 7714 + }, + { + "epoch": 0.9814272993257855, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7979012131690979, + "num_tokens": 294260487.0, + "step": 7715 + }, + { + "epoch": 0.9815545096043761, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7946022152900696, + "num_tokens": 294302642.0, + "step": 7716 + }, + { + "epoch": 0.9816817198829666, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7886056900024414, + "num_tokens": 294338376.0, + "step": 7717 + }, + { + "epoch": 0.981808930161557, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7927411794662476, + "num_tokens": 294372089.0, + "step": 7718 + }, + { + "epoch": 0.9819361404401475, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8019720911979675, + "num_tokens": 294403801.0, + "step": 7719 + }, + { + "epoch": 0.9820633507187381, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7987173199653625, + "num_tokens": 294440301.0, + "step": 7720 + }, + { + "epoch": 0.9821905609973286, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7999889254570007, + "num_tokens": 294472967.0, + "step": 7721 + }, + { + "epoch": 0.9823177712759191, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7845160961151123, + "num_tokens": 294518248.0, + "step": 7722 + }, + { + "epoch": 0.9824449815545097, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7916526794433594, + "num_tokens": 294553899.0, + "step": 7723 + }, + { + "epoch": 0.9825721918331001, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7730320692062378, + "num_tokens": 294591382.0, + "step": 7724 + }, + { + "epoch": 0.9826994021116906, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7963869571685791, + "num_tokens": 294629606.0, + "step": 7725 + }, + { + "epoch": 0.9828266123902811, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8042479753494263, + "num_tokens": 294669417.0, + "step": 7726 + }, + { + "epoch": 0.9829538226688717, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8092438578605652, + "num_tokens": 294709876.0, + "step": 7727 + }, + { + "epoch": 0.9830810329474622, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7890704274177551, + "num_tokens": 294750565.0, + "step": 7728 + }, + { + "epoch": 0.9832082432260527, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7986544370651245, + "num_tokens": 294790586.0, + "step": 7729 + }, + { + "epoch": 0.9833354535046431, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7954046726226807, + "num_tokens": 294830176.0, + "step": 7730 + }, + { + "epoch": 0.9834626637832337, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7821947336196899, + "num_tokens": 294874199.0, + "step": 7731 + }, + { + "epoch": 0.9835898740618242, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8098005056381226, + "num_tokens": 294914219.0, + "step": 7732 + }, + { + "epoch": 0.9837170843404147, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.772577702999115, + "num_tokens": 294950726.0, + "step": 7733 + }, + { + "epoch": 0.9838442946190052, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7999553084373474, + "num_tokens": 294993815.0, + "step": 7734 + }, + { + "epoch": 0.9839715048975958, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8050873279571533, + "num_tokens": 295029275.0, + "step": 7735 + }, + { + "epoch": 0.9840987151761862, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7991881370544434, + "num_tokens": 295066166.0, + "step": 7736 + }, + { + "epoch": 0.9842259254547767, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7700741291046143, + "num_tokens": 295100436.0, + "step": 7737 + }, + { + "epoch": 0.9843531357333672, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7980397939682007, + "num_tokens": 295141756.0, + "step": 7738 + }, + { + "epoch": 0.9844803460119578, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8115619421005249, + "num_tokens": 295178894.0, + "step": 7739 + }, + { + "epoch": 0.9846075562905483, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7875680923461914, + "num_tokens": 295219133.0, + "step": 7740 + }, + { + "epoch": 0.9847347665691388, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7998042702674866, + "num_tokens": 295260322.0, + "step": 7741 + }, + { + "epoch": 0.9848619768477292, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7915778160095215, + "num_tokens": 295299675.0, + "step": 7742 + }, + { + "epoch": 0.9849891871263198, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8020104169845581, + "num_tokens": 295336281.0, + "step": 7743 + }, + { + "epoch": 0.9851163974049103, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8035920858383179, + "num_tokens": 295376932.0, + "step": 7744 + }, + { + "epoch": 0.9852436076835008, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7982771396636963, + "num_tokens": 295421904.0, + "step": 7745 + }, + { + "epoch": 0.9853708179620914, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7942463755607605, + "num_tokens": 295456688.0, + "step": 7746 + }, + { + "epoch": 0.9854980282406819, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8205596804618835, + "num_tokens": 295500022.0, + "step": 7747 + }, + { + "epoch": 0.9856252385192723, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.8185641765594482, + "num_tokens": 295538106.0, + "step": 7748 + }, + { + "epoch": 0.9857524487978628, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7949694395065308, + "num_tokens": 295572465.0, + "step": 7749 + }, + { + "epoch": 0.9858796590764534, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.803749680519104, + "num_tokens": 295614577.0, + "step": 7750 + }, + { + "epoch": 0.9860068693550439, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.803966760635376, + "num_tokens": 295654780.0, + "step": 7751 + }, + { + "epoch": 0.9861340796336344, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8052396178245544, + "num_tokens": 295695880.0, + "step": 7752 + }, + { + "epoch": 0.986261289912225, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7970490455627441, + "num_tokens": 295738176.0, + "step": 7753 + }, + { + "epoch": 0.9863885001908154, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7860543131828308, + "num_tokens": 295777562.0, + "step": 7754 + }, + { + "epoch": 0.9865157104694059, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8134676218032837, + "num_tokens": 295815013.0, + "step": 7755 + }, + { + "epoch": 0.9866429207479964, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7841817736625671, + "num_tokens": 295857751.0, + "step": 7756 + }, + { + "epoch": 0.986770131026587, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.805540919303894, + "num_tokens": 295896033.0, + "step": 7757 + }, + { + "epoch": 0.9868973413051775, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8043978214263916, + "num_tokens": 295936838.0, + "step": 7758 + }, + { + "epoch": 0.987024551583768, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.802544355392456, + "num_tokens": 295972307.0, + "step": 7759 + }, + { + "epoch": 0.9871517618623584, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7959684729576111, + "num_tokens": 296009966.0, + "step": 7760 + }, + { + "epoch": 0.987278972140949, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7955335378646851, + "num_tokens": 296047874.0, + "step": 7761 + }, + { + "epoch": 0.9874061824195395, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8031598329544067, + "num_tokens": 296090591.0, + "step": 7762 + }, + { + "epoch": 0.98753339269813, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7954328656196594, + "num_tokens": 296133024.0, + "step": 7763 + }, + { + "epoch": 0.9876606029767205, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7885956764221191, + "num_tokens": 296177347.0, + "step": 7764 + }, + { + "epoch": 0.9877878132553111, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.787956714630127, + "num_tokens": 296215532.0, + "step": 7765 + }, + { + "epoch": 0.9879150235339016, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7876092195510864, + "num_tokens": 296249494.0, + "step": 7766 + }, + { + "epoch": 0.988042233812492, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7913492918014526, + "num_tokens": 296289370.0, + "step": 7767 + }, + { + "epoch": 0.9881694440910825, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8016834259033203, + "num_tokens": 296329376.0, + "step": 7768 + }, + { + "epoch": 0.9882966543696731, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.8049595952033997, + "num_tokens": 296368875.0, + "step": 7769 + }, + { + "epoch": 0.9884238646482636, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7955154776573181, + "num_tokens": 296404766.0, + "step": 7770 + }, + { + "epoch": 0.9885510749268541, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7957340478897095, + "num_tokens": 296441705.0, + "step": 7771 + }, + { + "epoch": 0.9886782852054447, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8063730001449585, + "num_tokens": 296481075.0, + "step": 7772 + }, + { + "epoch": 0.9888054954840351, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7921626567840576, + "num_tokens": 296511404.0, + "step": 7773 + }, + { + "epoch": 0.9889327057626256, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7858436107635498, + "num_tokens": 296546080.0, + "step": 7774 + }, + { + "epoch": 0.9890599160412161, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8032942414283752, + "num_tokens": 296586119.0, + "step": 7775 + }, + { + "epoch": 0.9891871263198067, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7967102527618408, + "num_tokens": 296629602.0, + "step": 7776 + }, + { + "epoch": 0.9893143365983972, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7887866497039795, + "num_tokens": 296666709.0, + "step": 7777 + }, + { + "epoch": 0.9894415468769877, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7973393201828003, + "num_tokens": 296702611.0, + "step": 7778 + }, + { + "epoch": 0.9895687571555781, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8033818602561951, + "num_tokens": 296737649.0, + "step": 7779 + }, + { + "epoch": 0.9896959674341687, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8014213442802429, + "num_tokens": 296776915.0, + "step": 7780 + }, + { + "epoch": 0.9898231777127592, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7990556955337524, + "num_tokens": 296811491.0, + "step": 7781 + }, + { + "epoch": 0.9899503879913497, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8028044700622559, + "num_tokens": 296847602.0, + "step": 7782 + }, + { + "epoch": 0.9900775982699402, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.784607470035553, + "num_tokens": 296884737.0, + "step": 7783 + }, + { + "epoch": 0.9902048085485308, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7899479269981384, + "num_tokens": 296927780.0, + "step": 7784 + }, + { + "epoch": 0.9903320188271212, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.8103103041648865, + "num_tokens": 296964200.0, + "step": 7785 + }, + { + "epoch": 0.9904592291057117, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7900936603546143, + "num_tokens": 297009927.0, + "step": 7786 + }, + { + "epoch": 0.9905864393843022, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7814518809318542, + "num_tokens": 297050788.0, + "step": 7787 + }, + { + "epoch": 0.9907136496628928, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8056937456130981, + "num_tokens": 297092398.0, + "step": 7788 + }, + { + "epoch": 0.9908408599414833, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.794338583946228, + "num_tokens": 297130728.0, + "step": 7789 + }, + { + "epoch": 0.9909680702200738, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.8246309757232666, + "num_tokens": 297169124.0, + "step": 7790 + }, + { + "epoch": 0.9910952804986642, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.8050402402877808, + "num_tokens": 297204914.0, + "step": 7791 + }, + { + "epoch": 0.9912224907772548, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7965616583824158, + "num_tokens": 297240551.0, + "step": 7792 + }, + { + "epoch": 0.9913497010558453, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7964987754821777, + "num_tokens": 297287160.0, + "step": 7793 + }, + { + "epoch": 0.9914769113344358, + "grad_norm": 80.0, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7961109280586243, + "num_tokens": 297323394.0, + "step": 7794 + }, + { + "epoch": 0.9916041216130264, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7728821635246277, + "num_tokens": 297357808.0, + "step": 7795 + }, + { + "epoch": 0.9917313318916169, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7900667190551758, + "num_tokens": 297397300.0, + "step": 7796 + }, + { + "epoch": 0.9918585421702073, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8135347962379456, + "num_tokens": 297434699.0, + "step": 7797 + }, + { + "epoch": 0.9919857524487978, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7989906072616577, + "num_tokens": 297473939.0, + "step": 7798 + }, + { + "epoch": 0.9921129627273884, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7877707481384277, + "num_tokens": 297511154.0, + "step": 7799 + }, + { + "epoch": 0.9922401730059789, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7864314317703247, + "num_tokens": 297556092.0, + "step": 7800 + }, + { + "epoch": 0.9923673832845694, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8036636114120483, + "num_tokens": 297598160.0, + "step": 7801 + }, + { + "epoch": 0.9924945935631599, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8161261677742004, + "num_tokens": 297638706.0, + "step": 7802 + }, + { + "epoch": 0.9926218038417504, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8064408302307129, + "num_tokens": 297675955.0, + "step": 7803 + }, + { + "epoch": 0.9927490141203409, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.797327995300293, + "num_tokens": 297715362.0, + "step": 7804 + }, + { + "epoch": 0.9928762243989314, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7833147048950195, + "num_tokens": 297750931.0, + "step": 7805 + }, + { + "epoch": 0.993003434677522, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7893902063369751, + "num_tokens": 297786107.0, + "step": 7806 + }, + { + "epoch": 0.9931306449561125, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.796283483505249, + "num_tokens": 297824782.0, + "step": 7807 + }, + { + "epoch": 0.993257855234703, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.799543559551239, + "num_tokens": 297861748.0, + "step": 7808 + }, + { + "epoch": 0.9933850655132934, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7926033139228821, + "num_tokens": 297902722.0, + "step": 7809 + }, + { + "epoch": 0.993512275791884, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.8015617728233337, + "num_tokens": 297948709.0, + "step": 7810 + }, + { + "epoch": 0.9936394860704745, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7970782518386841, + "num_tokens": 297986446.0, + "step": 7811 + }, + { + "epoch": 0.993766696349065, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7941977977752686, + "num_tokens": 298024004.0, + "step": 7812 + }, + { + "epoch": 0.9938939066276555, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7936797738075256, + "num_tokens": 298065518.0, + "step": 7813 + }, + { + "epoch": 0.9940211169062461, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7934005260467529, + "num_tokens": 298108138.0, + "step": 7814 + }, + { + "epoch": 0.9941483271848366, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7933087348937988, + "num_tokens": 298149348.0, + "step": 7815 + }, + { + "epoch": 0.994275537463427, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7944971323013306, + "num_tokens": 298186347.0, + "step": 7816 + }, + { + "epoch": 0.9944027477420175, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7723497748374939, + "num_tokens": 298223322.0, + "step": 7817 + }, + { + "epoch": 0.9945299580206081, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7926735877990723, + "num_tokens": 298258794.0, + "step": 7818 + }, + { + "epoch": 0.9946571682991986, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7897316217422485, + "num_tokens": 298297510.0, + "step": 7819 + }, + { + "epoch": 0.9947843785777891, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8007129430770874, + "num_tokens": 298333784.0, + "step": 7820 + }, + { + "epoch": 0.9949115888563796, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7992600202560425, + "num_tokens": 298371241.0, + "step": 7821 + }, + { + "epoch": 0.9950387991349701, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7823777198791504, + "num_tokens": 298407902.0, + "step": 7822 + }, + { + "epoch": 0.9951660094135606, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7653974294662476, + "num_tokens": 298452339.0, + "step": 7823 + }, + { + "epoch": 0.9952932196921511, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7894929647445679, + "num_tokens": 298495854.0, + "step": 7824 + }, + { + "epoch": 0.9954204299707416, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7878481149673462, + "num_tokens": 298529642.0, + "step": 7825 + }, + { + "epoch": 0.9955476402493322, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8025137186050415, + "num_tokens": 298574396.0, + "step": 7826 + }, + { + "epoch": 0.9956748505279227, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7837218046188354, + "num_tokens": 298612457.0, + "step": 7827 + }, + { + "epoch": 0.9958020608065131, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7921159267425537, + "num_tokens": 298645061.0, + "step": 7828 + }, + { + "epoch": 0.9959292710851037, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.794987678527832, + "num_tokens": 298688014.0, + "step": 7829 + }, + { + "epoch": 0.9960564813636942, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8034387826919556, + "num_tokens": 298727815.0, + "step": 7830 + }, + { + "epoch": 0.9961836916422847, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.5965, + "mean_token_accuracy": 0.8168290853500366, + "num_tokens": 298765663.0, + "step": 7831 + }, + { + "epoch": 0.9963109019208752, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8005284667015076, + "num_tokens": 298799360.0, + "step": 7832 + }, + { + "epoch": 0.9964381121994658, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7998969554901123, + "num_tokens": 298835669.0, + "step": 7833 + }, + { + "epoch": 0.9965653224780562, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7956916093826294, + "num_tokens": 298866871.0, + "step": 7834 + }, + { + "epoch": 0.9966925327566467, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7879362106323242, + "num_tokens": 298904136.0, + "step": 7835 + }, + { + "epoch": 0.9968197430352372, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8031399846076965, + "num_tokens": 298937069.0, + "step": 7836 + }, + { + "epoch": 0.9969469533138278, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7896628379821777, + "num_tokens": 298976778.0, + "step": 7837 + }, + { + "epoch": 0.9970741635924183, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8082471489906311, + "num_tokens": 299010808.0, + "step": 7838 + }, + { + "epoch": 0.9972013738710088, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7822290658950806, + "num_tokens": 299047611.0, + "step": 7839 + }, + { + "epoch": 0.9973285841495992, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8076063990592957, + "num_tokens": 299088668.0, + "step": 7840 + }, + { + "epoch": 0.9974557944281898, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7744592428207397, + "num_tokens": 299126572.0, + "step": 7841 + }, + { + "epoch": 0.9975830047067803, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8058176040649414, + "num_tokens": 299164806.0, + "step": 7842 + }, + { + "epoch": 0.9977102149853708, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7962541580200195, + "num_tokens": 299207039.0, + "step": 7843 + }, + { + "epoch": 0.9978374252639614, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.808853268623352, + "num_tokens": 299243242.0, + "step": 7844 + }, + { + "epoch": 0.9979646355425519, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8000602126121521, + "num_tokens": 299282326.0, + "step": 7845 + }, + { + "epoch": 0.9980918458211423, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8269901275634766, + "num_tokens": 299316645.0, + "step": 7846 + }, + { + "epoch": 0.9982190560997328, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8001249432563782, + "num_tokens": 299349932.0, + "step": 7847 + }, + { + "epoch": 0.9983462663783234, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7735249400138855, + "num_tokens": 299392452.0, + "step": 7848 + }, + { + "epoch": 0.9984734766569139, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8025023937225342, + "num_tokens": 299431479.0, + "step": 7849 + }, + { + "epoch": 0.9986006869355044, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7954838275909424, + "num_tokens": 299475585.0, + "step": 7850 + }, + { + "epoch": 0.9987278972140949, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8134250640869141, + "num_tokens": 299514026.0, + "step": 7851 + }, + { + "epoch": 0.9988551074926854, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7889505624771118, + "num_tokens": 299550858.0, + "step": 7852 + }, + { + "epoch": 0.9989823177712759, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.788891613483429, + "num_tokens": 299586831.0, + "step": 7853 + }, + { + "epoch": 0.9991095280498664, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8107816576957703, + "num_tokens": 299623116.0, + "step": 7854 + }, + { + "epoch": 0.9992367383284569, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7987019419670105, + "num_tokens": 299662165.0, + "step": 7855 + }, + { + "epoch": 0.9993639486070475, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7911050319671631, + "num_tokens": 299702581.0, + "step": 7856 + }, + { + "epoch": 0.999491158885638, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7838500738143921, + "num_tokens": 299742559.0, + "step": 7857 + }, + { + "epoch": 0.9996183691642284, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8169207572937012, + "num_tokens": 299778637.0, + "step": 7858 + }, + { + "epoch": 0.9997455794428189, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8041622042655945, + "num_tokens": 299812808.0, + "step": 7859 + }, + { + "epoch": 0.9998727897214095, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7917956113815308, + "num_tokens": 299848987.0, + "step": 7860 + }, + { + "epoch": 1.0, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.773179829120636, + "num_tokens": 299886286.0, + "step": 7861 + }, + { + "epoch": 1.0001272102785905, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7844632267951965, + "num_tokens": 299925456.0, + "step": 7862 + }, + { + "epoch": 1.000254420557181, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7973433136940002, + "num_tokens": 299965936.0, + "step": 7863 + }, + { + "epoch": 1.0003816308357716, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8055946826934814, + "num_tokens": 300003181.0, + "step": 7864 + }, + { + "epoch": 1.0005088411143621, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7905470728874207, + "num_tokens": 300040502.0, + "step": 7865 + }, + { + "epoch": 1.0006360513929526, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8149909973144531, + "num_tokens": 300078795.0, + "step": 7866 + }, + { + "epoch": 1.0007632616715432, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8053759336471558, + "num_tokens": 300115188.0, + "step": 7867 + }, + { + "epoch": 1.0008904719501335, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7988924980163574, + "num_tokens": 300158583.0, + "step": 7868 + }, + { + "epoch": 1.001017682228724, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8021107912063599, + "num_tokens": 300194415.0, + "step": 7869 + }, + { + "epoch": 1.0011448925073145, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7958828210830688, + "num_tokens": 300235066.0, + "step": 7870 + }, + { + "epoch": 1.001272102785905, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8001808524131775, + "num_tokens": 300272167.0, + "step": 7871 + }, + { + "epoch": 1.0013993130644956, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7855126857757568, + "num_tokens": 300306032.0, + "step": 7872 + }, + { + "epoch": 1.0015265233430861, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.788435697555542, + "num_tokens": 300338433.0, + "step": 7873 + }, + { + "epoch": 1.0016537336216766, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7846777439117432, + "num_tokens": 300376527.0, + "step": 7874 + }, + { + "epoch": 1.0017809439002672, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8123669624328613, + "num_tokens": 300417309.0, + "step": 7875 + }, + { + "epoch": 1.0019081541788577, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7938199043273926, + "num_tokens": 300459438.0, + "step": 7876 + }, + { + "epoch": 1.0020353644574482, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7689061164855957, + "num_tokens": 300501443.0, + "step": 7877 + }, + { + "epoch": 1.0021625747360388, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.798804521560669, + "num_tokens": 300531918.0, + "step": 7878 + }, + { + "epoch": 1.0022897850146293, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7850304841995239, + "num_tokens": 300575084.0, + "step": 7879 + }, + { + "epoch": 1.0024169952932196, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.795356273651123, + "num_tokens": 300608737.0, + "step": 7880 + }, + { + "epoch": 1.0025442055718101, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7781227827072144, + "num_tokens": 300647855.0, + "step": 7881 + }, + { + "epoch": 1.0026714158504006, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7970656156539917, + "num_tokens": 300689763.0, + "step": 7882 + }, + { + "epoch": 1.0027986261289912, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.810691237449646, + "num_tokens": 300725838.0, + "step": 7883 + }, + { + "epoch": 1.0029258364075817, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8044328689575195, + "num_tokens": 300760423.0, + "step": 7884 + }, + { + "epoch": 1.0030530466861722, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7996306419372559, + "num_tokens": 300796236.0, + "step": 7885 + }, + { + "epoch": 1.0031802569647628, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7996090054512024, + "num_tokens": 300835213.0, + "step": 7886 + }, + { + "epoch": 1.0033074672433533, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8083335757255554, + "num_tokens": 300873363.0, + "step": 7887 + }, + { + "epoch": 1.0034346775219438, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7870444059371948, + "num_tokens": 300910233.0, + "step": 7888 + }, + { + "epoch": 1.0035618878005343, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8074190616607666, + "num_tokens": 300946986.0, + "step": 7889 + }, + { + "epoch": 1.0036890980791249, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8007314801216125, + "num_tokens": 300985750.0, + "step": 7890 + }, + { + "epoch": 1.0038163083577154, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8018268346786499, + "num_tokens": 301032820.0, + "step": 7891 + }, + { + "epoch": 1.0039435186363057, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7846700549125671, + "num_tokens": 301070825.0, + "step": 7892 + }, + { + "epoch": 1.0040707289148962, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7913112640380859, + "num_tokens": 301110764.0, + "step": 7893 + }, + { + "epoch": 1.0041979391934868, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8044417500495911, + "num_tokens": 301152421.0, + "step": 7894 + }, + { + "epoch": 1.0043251494720773, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7956174612045288, + "num_tokens": 301192780.0, + "step": 7895 + }, + { + "epoch": 1.0044523597506678, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7958394289016724, + "num_tokens": 301228147.0, + "step": 7896 + }, + { + "epoch": 1.0045795700292584, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8070066571235657, + "num_tokens": 301259568.0, + "step": 7897 + }, + { + "epoch": 1.0047067803078489, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7964490652084351, + "num_tokens": 301296483.0, + "step": 7898 + }, + { + "epoch": 1.0048339905864394, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8012561202049255, + "num_tokens": 301337437.0, + "step": 7899 + }, + { + "epoch": 1.00496120086503, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.797603964805603, + "num_tokens": 301368777.0, + "step": 7900 + }, + { + "epoch": 1.0050884111436205, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7969446778297424, + "num_tokens": 301404885.0, + "step": 7901 + }, + { + "epoch": 1.005215621422211, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7902688980102539, + "num_tokens": 301445308.0, + "step": 7902 + }, + { + "epoch": 1.0053428317008015, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7935922145843506, + "num_tokens": 301484845.0, + "step": 7903 + }, + { + "epoch": 1.0054700419793918, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7927854061126709, + "num_tokens": 301524498.0, + "step": 7904 + }, + { + "epoch": 1.0055972522579824, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8033366799354553, + "num_tokens": 301565522.0, + "step": 7905 + }, + { + "epoch": 1.0057244625365729, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7964158058166504, + "num_tokens": 301603724.0, + "step": 7906 + }, + { + "epoch": 1.0058516728151634, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7926455736160278, + "num_tokens": 301636650.0, + "step": 7907 + }, + { + "epoch": 1.005978883093754, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.786710262298584, + "num_tokens": 301675328.0, + "step": 7908 + }, + { + "epoch": 1.0061060933723445, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7871792316436768, + "num_tokens": 301710058.0, + "step": 7909 + }, + { + "epoch": 1.006233303650935, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7865927219390869, + "num_tokens": 301753567.0, + "step": 7910 + }, + { + "epoch": 1.0063605139295255, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8061152100563049, + "num_tokens": 301792895.0, + "step": 7911 + }, + { + "epoch": 1.006487724208116, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8138595819473267, + "num_tokens": 301833001.0, + "step": 7912 + }, + { + "epoch": 1.0066149344867066, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7994980812072754, + "num_tokens": 301871623.0, + "step": 7913 + }, + { + "epoch": 1.006742144765297, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8044224977493286, + "num_tokens": 301903750.0, + "step": 7914 + }, + { + "epoch": 1.0068693550438876, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8052706718444824, + "num_tokens": 301943079.0, + "step": 7915 + }, + { + "epoch": 1.0069965653224782, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7982286810874939, + "num_tokens": 301980031.0, + "step": 7916 + }, + { + "epoch": 1.0071237756010685, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7965666055679321, + "num_tokens": 302020454.0, + "step": 7917 + }, + { + "epoch": 1.007250985879659, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7937688827514648, + "num_tokens": 302064391.0, + "step": 7918 + }, + { + "epoch": 1.0073781961582495, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7983629107475281, + "num_tokens": 302101598.0, + "step": 7919 + }, + { + "epoch": 1.00750540643684, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.802291989326477, + "num_tokens": 302138803.0, + "step": 7920 + }, + { + "epoch": 1.0076326167154306, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7963170409202576, + "num_tokens": 302173682.0, + "step": 7921 + }, + { + "epoch": 1.0077598269940211, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7879703044891357, + "num_tokens": 302207390.0, + "step": 7922 + }, + { + "epoch": 1.0078870372726116, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7982969880104065, + "num_tokens": 302244440.0, + "step": 7923 + }, + { + "epoch": 1.0080142475512022, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7898998856544495, + "num_tokens": 302281058.0, + "step": 7924 + }, + { + "epoch": 1.0081414578297927, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8205952644348145, + "num_tokens": 302319586.0, + "step": 7925 + }, + { + "epoch": 1.0082686681083832, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7941714525222778, + "num_tokens": 302361356.0, + "step": 7926 + }, + { + "epoch": 1.0083958783869738, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8045647144317627, + "num_tokens": 302399974.0, + "step": 7927 + }, + { + "epoch": 1.0085230886655643, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8037627935409546, + "num_tokens": 302435709.0, + "step": 7928 + }, + { + "epoch": 1.0086502989441546, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7884844541549683, + "num_tokens": 302474151.0, + "step": 7929 + }, + { + "epoch": 1.0087775092227451, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7965323328971863, + "num_tokens": 302509877.0, + "step": 7930 + }, + { + "epoch": 1.0089047195013356, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8137076497077942, + "num_tokens": 302547689.0, + "step": 7931 + }, + { + "epoch": 1.0090319297799262, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.796779453754425, + "num_tokens": 302590728.0, + "step": 7932 + }, + { + "epoch": 1.0091591400585167, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7762202024459839, + "num_tokens": 302625709.0, + "step": 7933 + }, + { + "epoch": 1.0092863503371072, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8015682101249695, + "num_tokens": 302662086.0, + "step": 7934 + }, + { + "epoch": 1.0094135606156978, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8169736862182617, + "num_tokens": 302703680.0, + "step": 7935 + }, + { + "epoch": 1.0095407708942883, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7798154354095459, + "num_tokens": 302740426.0, + "step": 7936 + }, + { + "epoch": 1.0096679811728788, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8109492063522339, + "num_tokens": 302780778.0, + "step": 7937 + }, + { + "epoch": 1.0097951914514693, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7997005581855774, + "num_tokens": 302816012.0, + "step": 7938 + }, + { + "epoch": 1.0099224017300599, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7706418633460999, + "num_tokens": 302858325.0, + "step": 7939 + }, + { + "epoch": 1.0100496120086504, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7972111105918884, + "num_tokens": 302898210.0, + "step": 7940 + }, + { + "epoch": 1.0101768222872407, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7959368824958801, + "num_tokens": 302936072.0, + "step": 7941 + }, + { + "epoch": 1.0103040325658312, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7911663055419922, + "num_tokens": 302978941.0, + "step": 7942 + }, + { + "epoch": 1.0104312428444218, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7886675596237183, + "num_tokens": 303016664.0, + "step": 7943 + }, + { + "epoch": 1.0105584531230123, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.798651397228241, + "num_tokens": 303063313.0, + "step": 7944 + }, + { + "epoch": 1.0106856634016028, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7863737344741821, + "num_tokens": 303103786.0, + "step": 7945 + }, + { + "epoch": 1.0108128736801933, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7989204525947571, + "num_tokens": 303136928.0, + "step": 7946 + }, + { + "epoch": 1.0109400839587839, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.8013609051704407, + "num_tokens": 303174854.0, + "step": 7947 + }, + { + "epoch": 1.0110672942373744, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7956098318099976, + "num_tokens": 303214234.0, + "step": 7948 + }, + { + "epoch": 1.011194504515965, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8150820732116699, + "num_tokens": 303251830.0, + "step": 7949 + }, + { + "epoch": 1.0113217147945555, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7885780930519104, + "num_tokens": 303291564.0, + "step": 7950 + }, + { + "epoch": 1.011448925073146, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7917565703392029, + "num_tokens": 303328642.0, + "step": 7951 + }, + { + "epoch": 1.0115761353517365, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7992638945579529, + "num_tokens": 303373693.0, + "step": 7952 + }, + { + "epoch": 1.0117033456303268, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8010011911392212, + "num_tokens": 303407124.0, + "step": 7953 + }, + { + "epoch": 1.0118305559089174, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8089008927345276, + "num_tokens": 303441586.0, + "step": 7954 + }, + { + "epoch": 1.0119577661875079, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7926614880561829, + "num_tokens": 303473593.0, + "step": 7955 + }, + { + "epoch": 1.0120849764660984, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7931182384490967, + "num_tokens": 303510239.0, + "step": 7956 + }, + { + "epoch": 1.012212186744689, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7960672378540039, + "num_tokens": 303544709.0, + "step": 7957 + }, + { + "epoch": 1.0123393970232795, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8076112270355225, + "num_tokens": 303581990.0, + "step": 7958 + }, + { + "epoch": 1.01246660730187, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7999619245529175, + "num_tokens": 303625167.0, + "step": 7959 + }, + { + "epoch": 1.0125938175804605, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7988353967666626, + "num_tokens": 303664751.0, + "step": 7960 + }, + { + "epoch": 1.012721027859051, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.79515540599823, + "num_tokens": 303702190.0, + "step": 7961 + }, + { + "epoch": 1.0128482381376416, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8147733211517334, + "num_tokens": 303742872.0, + "step": 7962 + }, + { + "epoch": 1.012975448416232, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7917570471763611, + "num_tokens": 303781265.0, + "step": 7963 + }, + { + "epoch": 1.0131026586948226, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7940800189971924, + "num_tokens": 303820686.0, + "step": 7964 + }, + { + "epoch": 1.0132298689734132, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8007779717445374, + "num_tokens": 303863428.0, + "step": 7965 + }, + { + "epoch": 1.0133570792520035, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7942001223564148, + "num_tokens": 303903847.0, + "step": 7966 + }, + { + "epoch": 1.013484289530594, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7925134301185608, + "num_tokens": 303943083.0, + "step": 7967 + }, + { + "epoch": 1.0136114998091845, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8068211078643799, + "num_tokens": 303978773.0, + "step": 7968 + }, + { + "epoch": 1.013738710087775, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8061059713363647, + "num_tokens": 304017058.0, + "step": 7969 + }, + { + "epoch": 1.0138659203663656, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7872270345687866, + "num_tokens": 304054104.0, + "step": 7970 + }, + { + "epoch": 1.013993130644956, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8073438405990601, + "num_tokens": 304091100.0, + "step": 7971 + }, + { + "epoch": 1.0141203409235466, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7942070960998535, + "num_tokens": 304127747.0, + "step": 7972 + }, + { + "epoch": 1.0142475512021372, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7971705198287964, + "num_tokens": 304171089.0, + "step": 7973 + }, + { + "epoch": 1.0143747614807277, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7898539304733276, + "num_tokens": 304210207.0, + "step": 7974 + }, + { + "epoch": 1.0145019717593182, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8101444244384766, + "num_tokens": 304249423.0, + "step": 7975 + }, + { + "epoch": 1.0146291820379088, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7971838712692261, + "num_tokens": 304286872.0, + "step": 7976 + }, + { + "epoch": 1.0147563923164993, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7971527576446533, + "num_tokens": 304323033.0, + "step": 7977 + }, + { + "epoch": 1.0148836025950896, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.8001941442489624, + "num_tokens": 304360096.0, + "step": 7978 + }, + { + "epoch": 1.0150108128736801, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8234801292419434, + "num_tokens": 304396364.0, + "step": 7979 + }, + { + "epoch": 1.0151380231522706, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8006138801574707, + "num_tokens": 304431583.0, + "step": 7980 + }, + { + "epoch": 1.0152652334308612, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7940109968185425, + "num_tokens": 304470531.0, + "step": 7981 + }, + { + "epoch": 1.0153924437094517, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7992570996284485, + "num_tokens": 304505187.0, + "step": 7982 + }, + { + "epoch": 1.0155196539880422, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7748438119888306, + "num_tokens": 304542685.0, + "step": 7983 + }, + { + "epoch": 1.0156468642666328, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8012292385101318, + "num_tokens": 304579204.0, + "step": 7984 + }, + { + "epoch": 1.0157740745452233, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8050842881202698, + "num_tokens": 304615224.0, + "step": 7985 + }, + { + "epoch": 1.0159012848238138, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7938959002494812, + "num_tokens": 304653513.0, + "step": 7986 + }, + { + "epoch": 1.0160284951024043, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8049300909042358, + "num_tokens": 304686744.0, + "step": 7987 + }, + { + "epoch": 1.0161557053809949, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7971115112304688, + "num_tokens": 304727262.0, + "step": 7988 + }, + { + "epoch": 1.0162829156595854, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7971106767654419, + "num_tokens": 304763418.0, + "step": 7989 + }, + { + "epoch": 1.0164101259381757, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.801916241645813, + "num_tokens": 304800763.0, + "step": 7990 + }, + { + "epoch": 1.0165373362167662, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8142774105072021, + "num_tokens": 304833136.0, + "step": 7991 + }, + { + "epoch": 1.0166645464953568, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8059583902359009, + "num_tokens": 304873208.0, + "step": 7992 + }, + { + "epoch": 1.0167917567739473, + "grad_norm": 62.5, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7968500852584839, + "num_tokens": 304914237.0, + "step": 7993 + }, + { + "epoch": 1.0169189670525378, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8183735013008118, + "num_tokens": 304955453.0, + "step": 7994 + }, + { + "epoch": 1.0170461773311283, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7995257377624512, + "num_tokens": 304990533.0, + "step": 7995 + }, + { + "epoch": 1.0171733876097189, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7987176179885864, + "num_tokens": 305024849.0, + "step": 7996 + }, + { + "epoch": 1.0173005978883094, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7919578552246094, + "num_tokens": 305061615.0, + "step": 7997 + }, + { + "epoch": 1.0174278081669, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7828432321548462, + "num_tokens": 305102251.0, + "step": 7998 + }, + { + "epoch": 1.0175550184454905, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8068482279777527, + "num_tokens": 305135405.0, + "step": 7999 + }, + { + "epoch": 1.017682228724081, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7919609546661377, + "num_tokens": 305169005.0, + "step": 8000 + }, + { + "epoch": 1.0178094390026715, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7953531742095947, + "num_tokens": 305209050.0, + "step": 8001 + }, + { + "epoch": 1.0179366492812618, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7953183650970459, + "num_tokens": 305238250.0, + "step": 8002 + }, + { + "epoch": 1.0180638595598523, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8046094179153442, + "num_tokens": 305283544.0, + "step": 8003 + }, + { + "epoch": 1.0181910698384429, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7732148766517639, + "num_tokens": 305320548.0, + "step": 8004 + }, + { + "epoch": 1.0183182801170334, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8048480153083801, + "num_tokens": 305359560.0, + "step": 8005 + }, + { + "epoch": 1.018445490395624, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7976490259170532, + "num_tokens": 305396680.0, + "step": 8006 + }, + { + "epoch": 1.0185727006742145, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8070476055145264, + "num_tokens": 305432856.0, + "step": 8007 + }, + { + "epoch": 1.018699910952805, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7991894483566284, + "num_tokens": 305469973.0, + "step": 8008 + }, + { + "epoch": 1.0188271212313955, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8014264106750488, + "num_tokens": 305505576.0, + "step": 8009 + }, + { + "epoch": 1.018954331509986, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7697589993476868, + "num_tokens": 305542037.0, + "step": 8010 + }, + { + "epoch": 1.0190815417885766, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8104095458984375, + "num_tokens": 305587784.0, + "step": 8011 + }, + { + "epoch": 1.019208752067167, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7942547798156738, + "num_tokens": 305621888.0, + "step": 8012 + }, + { + "epoch": 1.0193359623457576, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8202136754989624, + "num_tokens": 305663844.0, + "step": 8013 + }, + { + "epoch": 1.0194631726243482, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8108163475990295, + "num_tokens": 305695913.0, + "step": 8014 + }, + { + "epoch": 1.0195903829029385, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.794542670249939, + "num_tokens": 305733632.0, + "step": 8015 + }, + { + "epoch": 1.019717593181529, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7862933278083801, + "num_tokens": 305766917.0, + "step": 8016 + }, + { + "epoch": 1.0198448034601195, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7953881621360779, + "num_tokens": 305805768.0, + "step": 8017 + }, + { + "epoch": 1.01997201373871, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8035328388214111, + "num_tokens": 305843412.0, + "step": 8018 + }, + { + "epoch": 1.0200992240173006, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8089690208435059, + "num_tokens": 305880523.0, + "step": 8019 + }, + { + "epoch": 1.020226434295891, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7987262010574341, + "num_tokens": 305915379.0, + "step": 8020 + }, + { + "epoch": 1.0203536445744816, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8165225982666016, + "num_tokens": 305952471.0, + "step": 8021 + }, + { + "epoch": 1.0204808548530722, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.8000725507736206, + "num_tokens": 305994023.0, + "step": 8022 + }, + { + "epoch": 1.0206080651316627, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7993156313896179, + "num_tokens": 306026563.0, + "step": 8023 + }, + { + "epoch": 1.0207352754102532, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7918693423271179, + "num_tokens": 306067913.0, + "step": 8024 + }, + { + "epoch": 1.0208624856888437, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7977892756462097, + "num_tokens": 306108692.0, + "step": 8025 + }, + { + "epoch": 1.0209896959674343, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7966113090515137, + "num_tokens": 306148082.0, + "step": 8026 + }, + { + "epoch": 1.0211169062460246, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8032146692276001, + "num_tokens": 306182800.0, + "step": 8027 + }, + { + "epoch": 1.021244116524615, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7944539785385132, + "num_tokens": 306226741.0, + "step": 8028 + }, + { + "epoch": 1.0213713268032056, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.794708251953125, + "num_tokens": 306265993.0, + "step": 8029 + }, + { + "epoch": 1.0214985370817962, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8054507970809937, + "num_tokens": 306306511.0, + "step": 8030 + }, + { + "epoch": 1.0216257473603867, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8083273768424988, + "num_tokens": 306337687.0, + "step": 8031 + }, + { + "epoch": 1.0217529576389772, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7581444978713989, + "num_tokens": 306379646.0, + "step": 8032 + }, + { + "epoch": 1.0218801679175677, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7946630716323853, + "num_tokens": 306415389.0, + "step": 8033 + }, + { + "epoch": 1.0220073781961583, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7907739877700806, + "num_tokens": 306452035.0, + "step": 8034 + }, + { + "epoch": 1.0221345884747488, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7967932820320129, + "num_tokens": 306496277.0, + "step": 8035 + }, + { + "epoch": 1.0222617987533393, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7996603846549988, + "num_tokens": 306528212.0, + "step": 8036 + }, + { + "epoch": 1.0223890090319299, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8011837005615234, + "num_tokens": 306562047.0, + "step": 8037 + }, + { + "epoch": 1.0225162193105204, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7873351573944092, + "num_tokens": 306601298.0, + "step": 8038 + }, + { + "epoch": 1.0226434295891107, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8024997711181641, + "num_tokens": 306640961.0, + "step": 8039 + }, + { + "epoch": 1.0227706398677012, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8192963600158691, + "num_tokens": 306679507.0, + "step": 8040 + }, + { + "epoch": 1.0228978501462918, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7887086272239685, + "num_tokens": 306717337.0, + "step": 8041 + }, + { + "epoch": 1.0230250604248823, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7714729309082031, + "num_tokens": 306754152.0, + "step": 8042 + }, + { + "epoch": 1.0231522707034728, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7900334596633911, + "num_tokens": 306794887.0, + "step": 8043 + }, + { + "epoch": 1.0232794809820633, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8053452372550964, + "num_tokens": 306828374.0, + "step": 8044 + }, + { + "epoch": 1.0234066912606539, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8012441992759705, + "num_tokens": 306865285.0, + "step": 8045 + }, + { + "epoch": 1.0235339015392444, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7993360161781311, + "num_tokens": 306897195.0, + "step": 8046 + }, + { + "epoch": 1.023661111817835, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7958859205245972, + "num_tokens": 306935352.0, + "step": 8047 + }, + { + "epoch": 1.0237883220964255, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8044066429138184, + "num_tokens": 306964621.0, + "step": 8048 + }, + { + "epoch": 1.023915532375016, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7958074808120728, + "num_tokens": 307003451.0, + "step": 8049 + }, + { + "epoch": 1.0240427426536065, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8007841110229492, + "num_tokens": 307044695.0, + "step": 8050 + }, + { + "epoch": 1.0241699529321968, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8044410943984985, + "num_tokens": 307080603.0, + "step": 8051 + }, + { + "epoch": 1.0242971632107873, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.79019695520401, + "num_tokens": 307122300.0, + "step": 8052 + }, + { + "epoch": 1.0244243734893779, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8067692518234253, + "num_tokens": 307158685.0, + "step": 8053 + }, + { + "epoch": 1.0245515837679684, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7976942658424377, + "num_tokens": 307199104.0, + "step": 8054 + }, + { + "epoch": 1.024678794046559, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7956643104553223, + "num_tokens": 307235840.0, + "step": 8055 + }, + { + "epoch": 1.0248060043251495, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7858095169067383, + "num_tokens": 307273089.0, + "step": 8056 + }, + { + "epoch": 1.02493321460374, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.8026034832000732, + "num_tokens": 307318825.0, + "step": 8057 + }, + { + "epoch": 1.0250604248823305, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8023704290390015, + "num_tokens": 307357006.0, + "step": 8058 + }, + { + "epoch": 1.025187635160921, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7949454188346863, + "num_tokens": 307390274.0, + "step": 8059 + }, + { + "epoch": 1.0253148454395116, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8080147504806519, + "num_tokens": 307427157.0, + "step": 8060 + }, + { + "epoch": 1.025442055718102, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7893478870391846, + "num_tokens": 307466555.0, + "step": 8061 + }, + { + "epoch": 1.0255692659966926, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7853045463562012, + "num_tokens": 307503375.0, + "step": 8062 + }, + { + "epoch": 1.0256964762752832, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8086693286895752, + "num_tokens": 307543170.0, + "step": 8063 + }, + { + "epoch": 1.0258236865538735, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7956341505050659, + "num_tokens": 307582890.0, + "step": 8064 + }, + { + "epoch": 1.025950896832464, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7900155782699585, + "num_tokens": 307617643.0, + "step": 8065 + }, + { + "epoch": 1.0260781071110545, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7931311726570129, + "num_tokens": 307652258.0, + "step": 8066 + }, + { + "epoch": 1.026205317389645, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7942572832107544, + "num_tokens": 307693245.0, + "step": 8067 + }, + { + "epoch": 1.0263325276682356, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8035496473312378, + "num_tokens": 307728751.0, + "step": 8068 + }, + { + "epoch": 1.026459737946826, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7883731722831726, + "num_tokens": 307767869.0, + "step": 8069 + }, + { + "epoch": 1.0265869482254166, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8090924024581909, + "num_tokens": 307809438.0, + "step": 8070 + }, + { + "epoch": 1.0267141585040072, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7769501209259033, + "num_tokens": 307848752.0, + "step": 8071 + }, + { + "epoch": 1.0268413687825977, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.789192795753479, + "num_tokens": 307886080.0, + "step": 8072 + }, + { + "epoch": 1.0269685790611882, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8121200799942017, + "num_tokens": 307922854.0, + "step": 8073 + }, + { + "epoch": 1.0270957893397787, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7929487824440002, + "num_tokens": 307960032.0, + "step": 8074 + }, + { + "epoch": 1.0272229996183693, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8035601377487183, + "num_tokens": 307996074.0, + "step": 8075 + }, + { + "epoch": 1.0273502098969596, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8107883930206299, + "num_tokens": 308030833.0, + "step": 8076 + }, + { + "epoch": 1.02747742017555, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8101940155029297, + "num_tokens": 308079620.0, + "step": 8077 + }, + { + "epoch": 1.0276046304541406, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7867692708969116, + "num_tokens": 308118555.0, + "step": 8078 + }, + { + "epoch": 1.0277318407327312, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7793007493019104, + "num_tokens": 308154410.0, + "step": 8079 + }, + { + "epoch": 1.0278590510113217, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7989224791526794, + "num_tokens": 308191990.0, + "step": 8080 + }, + { + "epoch": 1.0279862612899122, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.770889163017273, + "num_tokens": 308231047.0, + "step": 8081 + }, + { + "epoch": 1.0281134715685027, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8169443607330322, + "num_tokens": 308269487.0, + "step": 8082 + }, + { + "epoch": 1.0282406818470933, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7953572273254395, + "num_tokens": 308304974.0, + "step": 8083 + }, + { + "epoch": 1.0283678921256838, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8208891153335571, + "num_tokens": 308342072.0, + "step": 8084 + }, + { + "epoch": 1.0284951024042743, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7922477722167969, + "num_tokens": 308374293.0, + "step": 8085 + }, + { + "epoch": 1.0286223126828649, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8015691637992859, + "num_tokens": 308415872.0, + "step": 8086 + }, + { + "epoch": 1.0287495229614554, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7922623157501221, + "num_tokens": 308451198.0, + "step": 8087 + }, + { + "epoch": 1.0288767332400457, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8090118169784546, + "num_tokens": 308490554.0, + "step": 8088 + }, + { + "epoch": 1.0290039435186362, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7933627963066101, + "num_tokens": 308530221.0, + "step": 8089 + }, + { + "epoch": 1.0291311537972267, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7730951905250549, + "num_tokens": 308563634.0, + "step": 8090 + }, + { + "epoch": 1.0292583640758173, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7984285354614258, + "num_tokens": 308605033.0, + "step": 8091 + }, + { + "epoch": 1.0293855743544078, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7969664931297302, + "num_tokens": 308642603.0, + "step": 8092 + }, + { + "epoch": 1.0295127846329983, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7975788116455078, + "num_tokens": 308685189.0, + "step": 8093 + }, + { + "epoch": 1.0296399949115889, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7982683181762695, + "num_tokens": 308726052.0, + "step": 8094 + }, + { + "epoch": 1.0297672051901794, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7843766212463379, + "num_tokens": 308764828.0, + "step": 8095 + }, + { + "epoch": 1.02989441546877, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8063279390335083, + "num_tokens": 308799712.0, + "step": 8096 + }, + { + "epoch": 1.0300216257473604, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8000681400299072, + "num_tokens": 308839879.0, + "step": 8097 + }, + { + "epoch": 1.030148836025951, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7920395731925964, + "num_tokens": 308880362.0, + "step": 8098 + }, + { + "epoch": 1.0302760463045415, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8126598000526428, + "num_tokens": 308921395.0, + "step": 8099 + }, + { + "epoch": 1.0304032565831318, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8061205148696899, + "num_tokens": 308959976.0, + "step": 8100 + }, + { + "epoch": 1.0305304668617223, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8156100511550903, + "num_tokens": 309000702.0, + "step": 8101 + }, + { + "epoch": 1.0306576771403129, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8182947039604187, + "num_tokens": 309036540.0, + "step": 8102 + }, + { + "epoch": 1.0307848874189034, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7958407402038574, + "num_tokens": 309072408.0, + "step": 8103 + }, + { + "epoch": 1.030912097697494, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7844647169113159, + "num_tokens": 309110366.0, + "step": 8104 + }, + { + "epoch": 1.0310393079760845, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8060712814331055, + "num_tokens": 309155194.0, + "step": 8105 + }, + { + "epoch": 1.031166518254675, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8043176531791687, + "num_tokens": 309193747.0, + "step": 8106 + }, + { + "epoch": 1.0312937285332655, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7841441631317139, + "num_tokens": 309234161.0, + "step": 8107 + }, + { + "epoch": 1.031420938811856, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7871906757354736, + "num_tokens": 309272574.0, + "step": 8108 + }, + { + "epoch": 1.0315481490904466, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8011125326156616, + "num_tokens": 309306567.0, + "step": 8109 + }, + { + "epoch": 1.031675359369037, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8018620610237122, + "num_tokens": 309347218.0, + "step": 8110 + }, + { + "epoch": 1.0318025696476276, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8116810917854309, + "num_tokens": 309385701.0, + "step": 8111 + }, + { + "epoch": 1.0319297799262181, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7997630834579468, + "num_tokens": 309424600.0, + "step": 8112 + }, + { + "epoch": 1.0320569902048085, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7743633985519409, + "num_tokens": 309459929.0, + "step": 8113 + }, + { + "epoch": 1.032184200483399, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7956590056419373, + "num_tokens": 309505085.0, + "step": 8114 + }, + { + "epoch": 1.0323114107619895, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8062284588813782, + "num_tokens": 309544746.0, + "step": 8115 + }, + { + "epoch": 1.03243862104058, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7778100967407227, + "num_tokens": 309584301.0, + "step": 8116 + }, + { + "epoch": 1.0325658313191706, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7996796369552612, + "num_tokens": 309622570.0, + "step": 8117 + }, + { + "epoch": 1.032693041597761, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.782977819442749, + "num_tokens": 309666417.0, + "step": 8118 + }, + { + "epoch": 1.0328202518763516, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7852757573127747, + "num_tokens": 309705706.0, + "step": 8119 + }, + { + "epoch": 1.0329474621549422, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8068848252296448, + "num_tokens": 309744288.0, + "step": 8120 + }, + { + "epoch": 1.0330746724335327, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.780346155166626, + "num_tokens": 309780406.0, + "step": 8121 + }, + { + "epoch": 1.0332018827121232, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.759093165397644, + "num_tokens": 309811135.0, + "step": 8122 + }, + { + "epoch": 1.0333290929907137, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8013782501220703, + "num_tokens": 309847037.0, + "step": 8123 + }, + { + "epoch": 1.0334563032693043, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.795871376991272, + "num_tokens": 309884274.0, + "step": 8124 + }, + { + "epoch": 1.0335835135478946, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8099554777145386, + "num_tokens": 309915572.0, + "step": 8125 + }, + { + "epoch": 1.033710723826485, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8029942512512207, + "num_tokens": 309954502.0, + "step": 8126 + }, + { + "epoch": 1.0338379341050756, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.5716, + "mean_token_accuracy": 0.8285452723503113, + "num_tokens": 309993015.0, + "step": 8127 + }, + { + "epoch": 1.0339651443836662, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7875979542732239, + "num_tokens": 310030840.0, + "step": 8128 + }, + { + "epoch": 1.0340923546622567, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7941116094589233, + "num_tokens": 310067586.0, + "step": 8129 + }, + { + "epoch": 1.0342195649408472, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8087750673294067, + "num_tokens": 310104496.0, + "step": 8130 + }, + { + "epoch": 1.0343467752194377, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7776522636413574, + "num_tokens": 310141162.0, + "step": 8131 + }, + { + "epoch": 1.0344739854980283, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.795397162437439, + "num_tokens": 310182965.0, + "step": 8132 + }, + { + "epoch": 1.0346011957766188, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7781162858009338, + "num_tokens": 310219572.0, + "step": 8133 + }, + { + "epoch": 1.0347284060552093, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7787523865699768, + "num_tokens": 310248711.0, + "step": 8134 + }, + { + "epoch": 1.0348556163337999, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7804140448570251, + "num_tokens": 310287915.0, + "step": 8135 + }, + { + "epoch": 1.0349828266123904, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.794688880443573, + "num_tokens": 310325619.0, + "step": 8136 + }, + { + "epoch": 1.0351100368909807, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8033795356750488, + "num_tokens": 310362456.0, + "step": 8137 + }, + { + "epoch": 1.0352372471695712, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7951163053512573, + "num_tokens": 310399584.0, + "step": 8138 + }, + { + "epoch": 1.0353644574481617, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7887219786643982, + "num_tokens": 310437435.0, + "step": 8139 + }, + { + "epoch": 1.0354916677267523, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8047217726707458, + "num_tokens": 310476320.0, + "step": 8140 + }, + { + "epoch": 1.0356188780053428, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8151588439941406, + "num_tokens": 310512956.0, + "step": 8141 + }, + { + "epoch": 1.0357460882839333, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7925189137458801, + "num_tokens": 310547605.0, + "step": 8142 + }, + { + "epoch": 1.0358732985625239, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7955188751220703, + "num_tokens": 310586991.0, + "step": 8143 + }, + { + "epoch": 1.0360005088411144, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7962346076965332, + "num_tokens": 310621271.0, + "step": 8144 + }, + { + "epoch": 1.036127719119705, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8015655279159546, + "num_tokens": 310662154.0, + "step": 8145 + }, + { + "epoch": 1.0362549293982954, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8101310729980469, + "num_tokens": 310701924.0, + "step": 8146 + }, + { + "epoch": 1.036382139676886, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8028262257575989, + "num_tokens": 310739688.0, + "step": 8147 + }, + { + "epoch": 1.0365093499554765, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7939866781234741, + "num_tokens": 310773317.0, + "step": 8148 + }, + { + "epoch": 1.0366365602340668, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7948902249336243, + "num_tokens": 310810854.0, + "step": 8149 + }, + { + "epoch": 1.0367637705126573, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7852528095245361, + "num_tokens": 310846080.0, + "step": 8150 + }, + { + "epoch": 1.0368909807912479, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7778304815292358, + "num_tokens": 310885032.0, + "step": 8151 + }, + { + "epoch": 1.0370181910698384, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8031697273254395, + "num_tokens": 310924518.0, + "step": 8152 + }, + { + "epoch": 1.037145401348429, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7962859272956848, + "num_tokens": 310961734.0, + "step": 8153 + }, + { + "epoch": 1.0372726116270194, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7980678081512451, + "num_tokens": 311004541.0, + "step": 8154 + }, + { + "epoch": 1.03739982190561, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8057985305786133, + "num_tokens": 311043947.0, + "step": 8155 + }, + { + "epoch": 1.0375270321842005, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7874107360839844, + "num_tokens": 311083154.0, + "step": 8156 + }, + { + "epoch": 1.037654242462791, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8108854293823242, + "num_tokens": 311121429.0, + "step": 8157 + }, + { + "epoch": 1.0377814527413816, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7843573093414307, + "num_tokens": 311166578.0, + "step": 8158 + }, + { + "epoch": 1.037908663019972, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8027728796005249, + "num_tokens": 311209645.0, + "step": 8159 + }, + { + "epoch": 1.0380358732985626, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8012787103652954, + "num_tokens": 311248428.0, + "step": 8160 + }, + { + "epoch": 1.0381630835771531, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7956644296646118, + "num_tokens": 311282040.0, + "step": 8161 + }, + { + "epoch": 1.0382902938557435, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8050538897514343, + "num_tokens": 311320233.0, + "step": 8162 + }, + { + "epoch": 1.038417504134334, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7794451713562012, + "num_tokens": 311357586.0, + "step": 8163 + }, + { + "epoch": 1.0385447144129245, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8002887964248657, + "num_tokens": 311395545.0, + "step": 8164 + }, + { + "epoch": 1.038671924691515, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7934616804122925, + "num_tokens": 311428785.0, + "step": 8165 + }, + { + "epoch": 1.0387991349701056, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7897312641143799, + "num_tokens": 311469546.0, + "step": 8166 + }, + { + "epoch": 1.038926345248696, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7880217432975769, + "num_tokens": 311504064.0, + "step": 8167 + }, + { + "epoch": 1.0390535555272866, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7759587168693542, + "num_tokens": 311536600.0, + "step": 8168 + }, + { + "epoch": 1.0391807658058771, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8150838017463684, + "num_tokens": 311573638.0, + "step": 8169 + }, + { + "epoch": 1.0393079760844677, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7879518866539001, + "num_tokens": 311611085.0, + "step": 8170 + }, + { + "epoch": 1.0394351863630582, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7928299903869629, + "num_tokens": 311655497.0, + "step": 8171 + }, + { + "epoch": 1.0395623966416487, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7768765091896057, + "num_tokens": 311688167.0, + "step": 8172 + }, + { + "epoch": 1.0396896069202393, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7997685670852661, + "num_tokens": 311724926.0, + "step": 8173 + }, + { + "epoch": 1.0398168171988296, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.775760293006897, + "num_tokens": 311760729.0, + "step": 8174 + }, + { + "epoch": 1.03994402747742, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8048502206802368, + "num_tokens": 311802138.0, + "step": 8175 + }, + { + "epoch": 1.0400712377560106, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.804234504699707, + "num_tokens": 311838510.0, + "step": 8176 + }, + { + "epoch": 1.0401984480346012, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7909126877784729, + "num_tokens": 311870875.0, + "step": 8177 + }, + { + "epoch": 1.0403256583131917, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8056126236915588, + "num_tokens": 311905854.0, + "step": 8178 + }, + { + "epoch": 1.0404528685917822, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7920374870300293, + "num_tokens": 311936381.0, + "step": 8179 + }, + { + "epoch": 1.0405800788703727, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7873674631118774, + "num_tokens": 311972529.0, + "step": 8180 + }, + { + "epoch": 1.0407072891489633, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8214475512504578, + "num_tokens": 312007883.0, + "step": 8181 + }, + { + "epoch": 1.0408344994275538, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7913921475410461, + "num_tokens": 312041000.0, + "step": 8182 + }, + { + "epoch": 1.0409617097061443, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.798336386680603, + "num_tokens": 312076011.0, + "step": 8183 + }, + { + "epoch": 1.0410889199847349, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7917312383651733, + "num_tokens": 312115614.0, + "step": 8184 + }, + { + "epoch": 1.0412161302633254, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7890890836715698, + "num_tokens": 312154111.0, + "step": 8185 + }, + { + "epoch": 1.0413433405419157, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7927275896072388, + "num_tokens": 312191571.0, + "step": 8186 + }, + { + "epoch": 1.0414705508205062, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7903026342391968, + "num_tokens": 312229205.0, + "step": 8187 + }, + { + "epoch": 1.0415977610990967, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8121183514595032, + "num_tokens": 312266604.0, + "step": 8188 + }, + { + "epoch": 1.0417249713776873, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7833054065704346, + "num_tokens": 312305222.0, + "step": 8189 + }, + { + "epoch": 1.0418521816562778, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7737331390380859, + "num_tokens": 312345823.0, + "step": 8190 + }, + { + "epoch": 1.0419793919348683, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7880159616470337, + "num_tokens": 312383569.0, + "step": 8191 + }, + { + "epoch": 1.0421066022134589, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8005223274230957, + "num_tokens": 312423068.0, + "step": 8192 + }, + { + "epoch": 1.0422338124920494, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7892531156539917, + "num_tokens": 312460569.0, + "step": 8193 + }, + { + "epoch": 1.04236102277064, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7769278287887573, + "num_tokens": 312501654.0, + "step": 8194 + }, + { + "epoch": 1.0424882330492304, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7901864051818848, + "num_tokens": 312543630.0, + "step": 8195 + }, + { + "epoch": 1.042615443327821, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7944296598434448, + "num_tokens": 312579428.0, + "step": 8196 + }, + { + "epoch": 1.0427426536064115, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8109517097473145, + "num_tokens": 312621388.0, + "step": 8197 + }, + { + "epoch": 1.0428698638850018, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7971101999282837, + "num_tokens": 312665303.0, + "step": 8198 + }, + { + "epoch": 1.0429970741635923, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7855843305587769, + "num_tokens": 312702354.0, + "step": 8199 + }, + { + "epoch": 1.0431242844421829, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8022857904434204, + "num_tokens": 312737223.0, + "step": 8200 + }, + { + "epoch": 1.0432514947207734, + "grad_norm": 63.75, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.798609733581543, + "num_tokens": 312773353.0, + "step": 8201 + }, + { + "epoch": 1.043378704999364, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7769634127616882, + "num_tokens": 312809230.0, + "step": 8202 + }, + { + "epoch": 1.0435059152779544, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8142412900924683, + "num_tokens": 312841278.0, + "step": 8203 + }, + { + "epoch": 1.043633125556545, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8027036190032959, + "num_tokens": 312888469.0, + "step": 8204 + }, + { + "epoch": 1.0437603358351355, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8018711805343628, + "num_tokens": 312925446.0, + "step": 8205 + }, + { + "epoch": 1.043887546113726, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7974050641059875, + "num_tokens": 312965425.0, + "step": 8206 + }, + { + "epoch": 1.0440147563923166, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7851970195770264, + "num_tokens": 313000637.0, + "step": 8207 + }, + { + "epoch": 1.044141966670907, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7952127456665039, + "num_tokens": 313035986.0, + "step": 8208 + }, + { + "epoch": 1.0442691769494976, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7856255769729614, + "num_tokens": 313077027.0, + "step": 8209 + }, + { + "epoch": 1.0443963872280881, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8114440441131592, + "num_tokens": 313121733.0, + "step": 8210 + }, + { + "epoch": 1.0445235975066784, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7924163341522217, + "num_tokens": 313165542.0, + "step": 8211 + }, + { + "epoch": 1.044650807785269, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7851026654243469, + "num_tokens": 313201082.0, + "step": 8212 + }, + { + "epoch": 1.0447780180638595, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8097750544548035, + "num_tokens": 313241765.0, + "step": 8213 + }, + { + "epoch": 1.04490522834245, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7933356165885925, + "num_tokens": 313282471.0, + "step": 8214 + }, + { + "epoch": 1.0450324386210406, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7986699342727661, + "num_tokens": 313331505.0, + "step": 8215 + }, + { + "epoch": 1.045159648899631, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8013884425163269, + "num_tokens": 313369823.0, + "step": 8216 + }, + { + "epoch": 1.0452868591782216, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7940796613693237, + "num_tokens": 313404255.0, + "step": 8217 + }, + { + "epoch": 1.0454140694568121, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8092018961906433, + "num_tokens": 313439368.0, + "step": 8218 + }, + { + "epoch": 1.0455412797354027, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7841194868087769, + "num_tokens": 313479475.0, + "step": 8219 + }, + { + "epoch": 1.0456684900139932, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7851422429084778, + "num_tokens": 313514604.0, + "step": 8220 + }, + { + "epoch": 1.0457957002925837, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.791735827922821, + "num_tokens": 313548907.0, + "step": 8221 + }, + { + "epoch": 1.0459229105711743, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7936453819274902, + "num_tokens": 313586615.0, + "step": 8222 + }, + { + "epoch": 1.0460501208497646, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7824631929397583, + "num_tokens": 313625202.0, + "step": 8223 + }, + { + "epoch": 1.046177331128355, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7820935845375061, + "num_tokens": 313663257.0, + "step": 8224 + }, + { + "epoch": 1.0463045414069456, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.8002610802650452, + "num_tokens": 313701430.0, + "step": 8225 + }, + { + "epoch": 1.0464317516855361, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.804101824760437, + "num_tokens": 313741286.0, + "step": 8226 + }, + { + "epoch": 1.0465589619641267, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7767703533172607, + "num_tokens": 313779115.0, + "step": 8227 + }, + { + "epoch": 1.0466861722427172, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8086828589439392, + "num_tokens": 313822331.0, + "step": 8228 + }, + { + "epoch": 1.0468133825213077, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7952251434326172, + "num_tokens": 313863545.0, + "step": 8229 + }, + { + "epoch": 1.0469405927998983, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8165308833122253, + "num_tokens": 313894956.0, + "step": 8230 + }, + { + "epoch": 1.0470678030784888, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8038244247436523, + "num_tokens": 313927920.0, + "step": 8231 + }, + { + "epoch": 1.0471950133570793, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8016728162765503, + "num_tokens": 313963230.0, + "step": 8232 + }, + { + "epoch": 1.0473222236356698, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7824915647506714, + "num_tokens": 313995951.0, + "step": 8233 + }, + { + "epoch": 1.0474494339142604, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.802413821220398, + "num_tokens": 314040928.0, + "step": 8234 + }, + { + "epoch": 1.0475766441928507, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7868846654891968, + "num_tokens": 314080329.0, + "step": 8235 + }, + { + "epoch": 1.0477038544714412, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7934681177139282, + "num_tokens": 314117107.0, + "step": 8236 + }, + { + "epoch": 1.0478310647500317, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7863190770149231, + "num_tokens": 314148077.0, + "step": 8237 + }, + { + "epoch": 1.0479582750286223, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7835459113121033, + "num_tokens": 314182064.0, + "step": 8238 + }, + { + "epoch": 1.0480854853072128, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7855870723724365, + "num_tokens": 314221667.0, + "step": 8239 + }, + { + "epoch": 1.0482126955858033, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7922007441520691, + "num_tokens": 314255225.0, + "step": 8240 + }, + { + "epoch": 1.0483399058643939, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7920013666152954, + "num_tokens": 314288011.0, + "step": 8241 + }, + { + "epoch": 1.0484671161429844, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7970092296600342, + "num_tokens": 314332461.0, + "step": 8242 + }, + { + "epoch": 1.048594326421575, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7903704643249512, + "num_tokens": 314367967.0, + "step": 8243 + }, + { + "epoch": 1.0487215367001654, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8023445010185242, + "num_tokens": 314404487.0, + "step": 8244 + }, + { + "epoch": 1.048848746978756, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.8020135164260864, + "num_tokens": 314444731.0, + "step": 8245 + }, + { + "epoch": 1.0489759572573465, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7937905192375183, + "num_tokens": 314479720.0, + "step": 8246 + }, + { + "epoch": 1.0491031675359368, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8153641223907471, + "num_tokens": 314515613.0, + "step": 8247 + }, + { + "epoch": 1.0492303778145273, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7966621518135071, + "num_tokens": 314556519.0, + "step": 8248 + }, + { + "epoch": 1.0493575880931179, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8102334141731262, + "num_tokens": 314590803.0, + "step": 8249 + }, + { + "epoch": 1.0494847983717084, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7961121201515198, + "num_tokens": 314631347.0, + "step": 8250 + }, + { + "epoch": 1.049612008650299, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7952152490615845, + "num_tokens": 314673739.0, + "step": 8251 + }, + { + "epoch": 1.0497392189288894, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.8013111352920532, + "num_tokens": 314718498.0, + "step": 8252 + }, + { + "epoch": 1.04986642920748, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8059753179550171, + "num_tokens": 314758755.0, + "step": 8253 + }, + { + "epoch": 1.0499936394860705, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.796412467956543, + "num_tokens": 314793644.0, + "step": 8254 + }, + { + "epoch": 1.050120849764661, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.792119026184082, + "num_tokens": 314828045.0, + "step": 8255 + }, + { + "epoch": 1.0502480600432516, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8033009767532349, + "num_tokens": 314866088.0, + "step": 8256 + }, + { + "epoch": 1.050375270321842, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7906601428985596, + "num_tokens": 314904757.0, + "step": 8257 + }, + { + "epoch": 1.0505024806004326, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7867757678031921, + "num_tokens": 314941410.0, + "step": 8258 + }, + { + "epoch": 1.0506296908790231, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8091480135917664, + "num_tokens": 314982060.0, + "step": 8259 + }, + { + "epoch": 1.0507569011576134, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7859302163124084, + "num_tokens": 315021812.0, + "step": 8260 + }, + { + "epoch": 1.050884111436204, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7959440350532532, + "num_tokens": 315059520.0, + "step": 8261 + }, + { + "epoch": 1.0510113217147945, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7857824563980103, + "num_tokens": 315098392.0, + "step": 8262 + }, + { + "epoch": 1.051138531993385, + "grad_norm": 147.0, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7871028780937195, + "num_tokens": 315137375.0, + "step": 8263 + }, + { + "epoch": 1.0512657422719756, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7993048429489136, + "num_tokens": 315171308.0, + "step": 8264 + }, + { + "epoch": 1.051392952550566, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7991308569908142, + "num_tokens": 315212176.0, + "step": 8265 + }, + { + "epoch": 1.0515201628291566, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7842882871627808, + "num_tokens": 315251321.0, + "step": 8266 + }, + { + "epoch": 1.0516473731077471, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7843560576438904, + "num_tokens": 315288780.0, + "step": 8267 + }, + { + "epoch": 1.0517745833863377, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.5816, + "mean_token_accuracy": 0.8200606107711792, + "num_tokens": 315325630.0, + "step": 8268 + }, + { + "epoch": 1.0519017936649282, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7897053360939026, + "num_tokens": 315362887.0, + "step": 8269 + }, + { + "epoch": 1.0520290039435187, + "grad_norm": 99.5, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8053262233734131, + "num_tokens": 315397552.0, + "step": 8270 + }, + { + "epoch": 1.0521562142221093, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.804726243019104, + "num_tokens": 315436785.0, + "step": 8271 + }, + { + "epoch": 1.0522834245006996, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8055858016014099, + "num_tokens": 315467849.0, + "step": 8272 + }, + { + "epoch": 1.05241063477929, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8032028675079346, + "num_tokens": 315501496.0, + "step": 8273 + }, + { + "epoch": 1.0525378450578806, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8058448433876038, + "num_tokens": 315540878.0, + "step": 8274 + }, + { + "epoch": 1.0526650553364711, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8113553524017334, + "num_tokens": 315578139.0, + "step": 8275 + }, + { + "epoch": 1.0527922656150617, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8002523183822632, + "num_tokens": 315615845.0, + "step": 8276 + }, + { + "epoch": 1.0529194758936522, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7750177383422852, + "num_tokens": 315645407.0, + "step": 8277 + }, + { + "epoch": 1.0530466861722427, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8053028583526611, + "num_tokens": 315687922.0, + "step": 8278 + }, + { + "epoch": 1.0531738964508333, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7993085384368896, + "num_tokens": 315723730.0, + "step": 8279 + }, + { + "epoch": 1.0533011067294238, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8127865791320801, + "num_tokens": 315765662.0, + "step": 8280 + }, + { + "epoch": 1.0534283170080143, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7691900730133057, + "num_tokens": 315808668.0, + "step": 8281 + }, + { + "epoch": 1.0535555272866048, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7916007041931152, + "num_tokens": 315844878.0, + "step": 8282 + }, + { + "epoch": 1.0536827375651954, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8089486360549927, + "num_tokens": 315879472.0, + "step": 8283 + }, + { + "epoch": 1.0538099478437857, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.790311336517334, + "num_tokens": 315915496.0, + "step": 8284 + }, + { + "epoch": 1.0539371581223762, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7945728898048401, + "num_tokens": 315954317.0, + "step": 8285 + }, + { + "epoch": 1.0540643684009667, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.789114236831665, + "num_tokens": 315995518.0, + "step": 8286 + }, + { + "epoch": 1.0541915786795573, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8111895322799683, + "num_tokens": 316040475.0, + "step": 8287 + }, + { + "epoch": 1.0543187889581478, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7984918355941772, + "num_tokens": 316083003.0, + "step": 8288 + }, + { + "epoch": 1.0544459992367383, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.80302894115448, + "num_tokens": 316124687.0, + "step": 8289 + }, + { + "epoch": 1.0545732095153288, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7813358902931213, + "num_tokens": 316170052.0, + "step": 8290 + }, + { + "epoch": 1.0547004197939194, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7827165722846985, + "num_tokens": 316211941.0, + "step": 8291 + }, + { + "epoch": 1.05482763007251, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.8009359836578369, + "num_tokens": 316252592.0, + "step": 8292 + }, + { + "epoch": 1.0549548403511004, + "grad_norm": 75.0, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7934613227844238, + "num_tokens": 316287250.0, + "step": 8293 + }, + { + "epoch": 1.055082050629691, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.8023028373718262, + "num_tokens": 316326308.0, + "step": 8294 + }, + { + "epoch": 1.0552092609082815, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7983179688453674, + "num_tokens": 316364519.0, + "step": 8295 + }, + { + "epoch": 1.0553364711868718, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8040339946746826, + "num_tokens": 316401165.0, + "step": 8296 + }, + { + "epoch": 1.0554636814654623, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7936736941337585, + "num_tokens": 316440245.0, + "step": 8297 + }, + { + "epoch": 1.0555908917440529, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8087769746780396, + "num_tokens": 316484969.0, + "step": 8298 + }, + { + "epoch": 1.0557181020226434, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8137770295143127, + "num_tokens": 316520510.0, + "step": 8299 + }, + { + "epoch": 1.055845312301234, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7827863693237305, + "num_tokens": 316561946.0, + "step": 8300 + }, + { + "epoch": 1.0559725225798244, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.821224570274353, + "num_tokens": 316605489.0, + "step": 8301 + }, + { + "epoch": 1.056099732858415, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7842778563499451, + "num_tokens": 316648590.0, + "step": 8302 + }, + { + "epoch": 1.0562269431370055, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8134938478469849, + "num_tokens": 316688169.0, + "step": 8303 + }, + { + "epoch": 1.056354153415596, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.770570695400238, + "num_tokens": 316724031.0, + "step": 8304 + }, + { + "epoch": 1.0564813636941865, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7980266809463501, + "num_tokens": 316769762.0, + "step": 8305 + }, + { + "epoch": 1.056608573972777, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8066409826278687, + "num_tokens": 316814156.0, + "step": 8306 + }, + { + "epoch": 1.0567357842513676, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8120415806770325, + "num_tokens": 316852784.0, + "step": 8307 + }, + { + "epoch": 1.0568629945299581, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.789897084236145, + "num_tokens": 316891992.0, + "step": 8308 + }, + { + "epoch": 1.0569902048085484, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8136735558509827, + "num_tokens": 316932246.0, + "step": 8309 + }, + { + "epoch": 1.057117415087139, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7879658937454224, + "num_tokens": 316972017.0, + "step": 8310 + }, + { + "epoch": 1.0572446253657295, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7889392375946045, + "num_tokens": 317008631.0, + "step": 8311 + }, + { + "epoch": 1.05737183564432, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8036374449729919, + "num_tokens": 317048616.0, + "step": 8312 + }, + { + "epoch": 1.0574990459229106, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7864910364151001, + "num_tokens": 317088675.0, + "step": 8313 + }, + { + "epoch": 1.057626256201501, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8193190097808838, + "num_tokens": 317120351.0, + "step": 8314 + }, + { + "epoch": 1.0577534664800916, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7931253910064697, + "num_tokens": 317159027.0, + "step": 8315 + }, + { + "epoch": 1.0578806767586821, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7935883402824402, + "num_tokens": 317188892.0, + "step": 8316 + }, + { + "epoch": 1.0580078870372727, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8068318367004395, + "num_tokens": 317229026.0, + "step": 8317 + }, + { + "epoch": 1.0581350973158632, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7978299856185913, + "num_tokens": 317272946.0, + "step": 8318 + }, + { + "epoch": 1.0582623075944537, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7975950241088867, + "num_tokens": 317314625.0, + "step": 8319 + }, + { + "epoch": 1.058389517873044, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8072916269302368, + "num_tokens": 317351450.0, + "step": 8320 + }, + { + "epoch": 1.0585167281516346, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7802005410194397, + "num_tokens": 317391519.0, + "step": 8321 + }, + { + "epoch": 1.058643938430225, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7895179986953735, + "num_tokens": 317430672.0, + "step": 8322 + }, + { + "epoch": 1.0587711487088156, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8030811548233032, + "num_tokens": 317470203.0, + "step": 8323 + }, + { + "epoch": 1.0588983589874061, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8012597560882568, + "num_tokens": 317509594.0, + "step": 8324 + }, + { + "epoch": 1.0590255692659967, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7921285629272461, + "num_tokens": 317549735.0, + "step": 8325 + }, + { + "epoch": 1.0591527795445872, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7939465045928955, + "num_tokens": 317585064.0, + "step": 8326 + }, + { + "epoch": 1.0592799898231777, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8141177892684937, + "num_tokens": 317622289.0, + "step": 8327 + }, + { + "epoch": 1.0594072001017683, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7847201824188232, + "num_tokens": 317662872.0, + "step": 8328 + }, + { + "epoch": 1.0595344103803588, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.794013500213623, + "num_tokens": 317701418.0, + "step": 8329 + }, + { + "epoch": 1.0596616206589493, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8113270998001099, + "num_tokens": 317741416.0, + "step": 8330 + }, + { + "epoch": 1.0597888309375398, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8013758063316345, + "num_tokens": 317778623.0, + "step": 8331 + }, + { + "epoch": 1.0599160412161304, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7964290380477905, + "num_tokens": 317816577.0, + "step": 8332 + }, + { + "epoch": 1.0600432514947207, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7951955795288086, + "num_tokens": 317851532.0, + "step": 8333 + }, + { + "epoch": 1.0601704617733112, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7935070991516113, + "num_tokens": 317886789.0, + "step": 8334 + }, + { + "epoch": 1.0602976720519017, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8045830130577087, + "num_tokens": 317929511.0, + "step": 8335 + }, + { + "epoch": 1.0604248823304923, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7905963659286499, + "num_tokens": 317963767.0, + "step": 8336 + }, + { + "epoch": 1.0605520926090828, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8131232261657715, + "num_tokens": 318000458.0, + "step": 8337 + }, + { + "epoch": 1.0606793028876733, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8130329251289368, + "num_tokens": 318040394.0, + "step": 8338 + }, + { + "epoch": 1.0608065131662638, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7960900664329529, + "num_tokens": 318080859.0, + "step": 8339 + }, + { + "epoch": 1.0609337234448544, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8199528455734253, + "num_tokens": 318115378.0, + "step": 8340 + }, + { + "epoch": 1.061060933723445, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7863321304321289, + "num_tokens": 318150351.0, + "step": 8341 + }, + { + "epoch": 1.0611881440020354, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7999072074890137, + "num_tokens": 318192427.0, + "step": 8342 + }, + { + "epoch": 1.061315354280626, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7894337177276611, + "num_tokens": 318232150.0, + "step": 8343 + }, + { + "epoch": 1.0614425645592165, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7934905886650085, + "num_tokens": 318267742.0, + "step": 8344 + }, + { + "epoch": 1.0615697748378068, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8037342429161072, + "num_tokens": 318307974.0, + "step": 8345 + }, + { + "epoch": 1.0616969851163973, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.775062084197998, + "num_tokens": 318349880.0, + "step": 8346 + }, + { + "epoch": 1.0618241953949878, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7945972084999084, + "num_tokens": 318383869.0, + "step": 8347 + }, + { + "epoch": 1.0619514056735784, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8099429607391357, + "num_tokens": 318419659.0, + "step": 8348 + }, + { + "epoch": 1.062078615952169, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8070341348648071, + "num_tokens": 318460795.0, + "step": 8349 + }, + { + "epoch": 1.0622058262307594, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7921842336654663, + "num_tokens": 318499451.0, + "step": 8350 + }, + { + "epoch": 1.06233303650935, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8037924766540527, + "num_tokens": 318533413.0, + "step": 8351 + }, + { + "epoch": 1.0624602467879405, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7881124019622803, + "num_tokens": 318572533.0, + "step": 8352 + }, + { + "epoch": 1.062587457066531, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7977290153503418, + "num_tokens": 318609442.0, + "step": 8353 + }, + { + "epoch": 1.0627146673451215, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8081550002098083, + "num_tokens": 318643475.0, + "step": 8354 + }, + { + "epoch": 1.062841877623712, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8020492792129517, + "num_tokens": 318684223.0, + "step": 8355 + }, + { + "epoch": 1.0629690879023026, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7951595783233643, + "num_tokens": 318717148.0, + "step": 8356 + }, + { + "epoch": 1.0630962981808931, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8123841285705566, + "num_tokens": 318758929.0, + "step": 8357 + }, + { + "epoch": 1.0632235084594834, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7983700037002563, + "num_tokens": 318796438.0, + "step": 8358 + }, + { + "epoch": 1.063350718738074, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7891217470169067, + "num_tokens": 318835732.0, + "step": 8359 + }, + { + "epoch": 1.0634779290166645, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7942426800727844, + "num_tokens": 318872151.0, + "step": 8360 + }, + { + "epoch": 1.063605139295255, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7867692708969116, + "num_tokens": 318911956.0, + "step": 8361 + }, + { + "epoch": 1.0637323495738455, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7992371320724487, + "num_tokens": 318945593.0, + "step": 8362 + }, + { + "epoch": 1.063859559852436, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8036658763885498, + "num_tokens": 318976670.0, + "step": 8363 + }, + { + "epoch": 1.0639867701310266, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.8017910718917847, + "num_tokens": 319007798.0, + "step": 8364 + }, + { + "epoch": 1.0641139804096171, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7836170196533203, + "num_tokens": 319044808.0, + "step": 8365 + }, + { + "epoch": 1.0642411906882077, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8186606168746948, + "num_tokens": 319083387.0, + "step": 8366 + }, + { + "epoch": 1.0643684009667982, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8158905506134033, + "num_tokens": 319121897.0, + "step": 8367 + }, + { + "epoch": 1.0644956112453887, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7823259830474854, + "num_tokens": 319157190.0, + "step": 8368 + }, + { + "epoch": 1.064622821523979, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.804750919342041, + "num_tokens": 319188353.0, + "step": 8369 + }, + { + "epoch": 1.0647500318025696, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7963651418685913, + "num_tokens": 319225536.0, + "step": 8370 + }, + { + "epoch": 1.06487724208116, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8056027889251709, + "num_tokens": 319265114.0, + "step": 8371 + }, + { + "epoch": 1.0650044523597506, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.812774658203125, + "num_tokens": 319309280.0, + "step": 8372 + }, + { + "epoch": 1.0651316626383411, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8092005848884583, + "num_tokens": 319346114.0, + "step": 8373 + }, + { + "epoch": 1.0652588729169317, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8112829923629761, + "num_tokens": 319389001.0, + "step": 8374 + }, + { + "epoch": 1.0653860831955222, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8092458844184875, + "num_tokens": 319431464.0, + "step": 8375 + }, + { + "epoch": 1.0655132934741127, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.773230254650116, + "num_tokens": 319475638.0, + "step": 8376 + }, + { + "epoch": 1.0656405037527032, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8084856271743774, + "num_tokens": 319515983.0, + "step": 8377 + }, + { + "epoch": 1.0657677140312938, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7936159372329712, + "num_tokens": 319552152.0, + "step": 8378 + }, + { + "epoch": 1.0658949243098843, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7924891114234924, + "num_tokens": 319592963.0, + "step": 8379 + }, + { + "epoch": 1.0660221345884748, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7832019329071045, + "num_tokens": 319635136.0, + "step": 8380 + }, + { + "epoch": 1.0661493448670654, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8042739629745483, + "num_tokens": 319670118.0, + "step": 8381 + }, + { + "epoch": 1.0662765551456557, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8000880479812622, + "num_tokens": 319711057.0, + "step": 8382 + }, + { + "epoch": 1.0664037654242462, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7951329946517944, + "num_tokens": 319746451.0, + "step": 8383 + }, + { + "epoch": 1.0665309757028367, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7996790409088135, + "num_tokens": 319792149.0, + "step": 8384 + }, + { + "epoch": 1.0666581859814273, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7931360006332397, + "num_tokens": 319826508.0, + "step": 8385 + }, + { + "epoch": 1.0667853962600178, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8033902645111084, + "num_tokens": 319867000.0, + "step": 8386 + }, + { + "epoch": 1.0669126065386083, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7774166464805603, + "num_tokens": 319904476.0, + "step": 8387 + }, + { + "epoch": 1.0670398168171988, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7697427272796631, + "num_tokens": 319943279.0, + "step": 8388 + }, + { + "epoch": 1.0671670270957894, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7953950762748718, + "num_tokens": 319986610.0, + "step": 8389 + }, + { + "epoch": 1.06729423737438, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7910590767860413, + "num_tokens": 320022702.0, + "step": 8390 + }, + { + "epoch": 1.0674214476529704, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7990535497665405, + "num_tokens": 320067918.0, + "step": 8391 + }, + { + "epoch": 1.067548657931561, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7935935258865356, + "num_tokens": 320110455.0, + "step": 8392 + }, + { + "epoch": 1.0676758682101515, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8137471675872803, + "num_tokens": 320147675.0, + "step": 8393 + }, + { + "epoch": 1.0678030784887418, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7808644771575928, + "num_tokens": 320183882.0, + "step": 8394 + }, + { + "epoch": 1.0679302887673323, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7847256660461426, + "num_tokens": 320222215.0, + "step": 8395 + }, + { + "epoch": 1.0680574990459228, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7928762435913086, + "num_tokens": 320262418.0, + "step": 8396 + }, + { + "epoch": 1.0681847093245134, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.793643593788147, + "num_tokens": 320297304.0, + "step": 8397 + }, + { + "epoch": 1.068311919603104, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7969858646392822, + "num_tokens": 320330919.0, + "step": 8398 + }, + { + "epoch": 1.0684391298816944, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7957376837730408, + "num_tokens": 320367602.0, + "step": 8399 + }, + { + "epoch": 1.068566340160285, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8000218272209167, + "num_tokens": 320403875.0, + "step": 8400 + }, + { + "epoch": 1.0686935504388755, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7970243692398071, + "num_tokens": 320439149.0, + "step": 8401 + }, + { + "epoch": 1.068820760717466, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8104367256164551, + "num_tokens": 320476786.0, + "step": 8402 + }, + { + "epoch": 1.0689479709960565, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7784674763679504, + "num_tokens": 320516945.0, + "step": 8403 + }, + { + "epoch": 1.069075181274647, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.812416136264801, + "num_tokens": 320552801.0, + "step": 8404 + }, + { + "epoch": 1.0692023915532376, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.8005545735359192, + "num_tokens": 320593605.0, + "step": 8405 + }, + { + "epoch": 1.0693296018318281, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8072386980056763, + "num_tokens": 320624476.0, + "step": 8406 + }, + { + "epoch": 1.0694568121104184, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8078768253326416, + "num_tokens": 320667000.0, + "step": 8407 + }, + { + "epoch": 1.069584022389009, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7990123629570007, + "num_tokens": 320699286.0, + "step": 8408 + }, + { + "epoch": 1.0697112326675995, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8108776807785034, + "num_tokens": 320732520.0, + "step": 8409 + }, + { + "epoch": 1.06983844294619, + "grad_norm": 85.5, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7893165349960327, + "num_tokens": 320773425.0, + "step": 8410 + }, + { + "epoch": 1.0699656532247805, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7727195620536804, + "num_tokens": 320810859.0, + "step": 8411 + }, + { + "epoch": 1.070092863503371, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8026517033576965, + "num_tokens": 320850266.0, + "step": 8412 + }, + { + "epoch": 1.0702200737819616, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.799713134765625, + "num_tokens": 320892812.0, + "step": 8413 + }, + { + "epoch": 1.0703472840605521, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8042905330657959, + "num_tokens": 320929563.0, + "step": 8414 + }, + { + "epoch": 1.0704744943391427, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8063513040542603, + "num_tokens": 320964010.0, + "step": 8415 + }, + { + "epoch": 1.0706017046177332, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7916461229324341, + "num_tokens": 321006405.0, + "step": 8416 + }, + { + "epoch": 1.0707289148963237, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7903618812561035, + "num_tokens": 321050146.0, + "step": 8417 + }, + { + "epoch": 1.070856125174914, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7973508834838867, + "num_tokens": 321086857.0, + "step": 8418 + }, + { + "epoch": 1.0709833354535045, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8040969371795654, + "num_tokens": 321127310.0, + "step": 8419 + }, + { + "epoch": 1.071110545732095, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7911874055862427, + "num_tokens": 321165881.0, + "step": 8420 + }, + { + "epoch": 1.0712377560106856, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7882713675498962, + "num_tokens": 321204494.0, + "step": 8421 + }, + { + "epoch": 1.0713649662892761, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7836670279502869, + "num_tokens": 321243639.0, + "step": 8422 + }, + { + "epoch": 1.0714921765678667, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7903017997741699, + "num_tokens": 321281399.0, + "step": 8423 + }, + { + "epoch": 1.0716193868464572, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7813398838043213, + "num_tokens": 321316539.0, + "step": 8424 + }, + { + "epoch": 1.0717465971250477, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.791920006275177, + "num_tokens": 321348618.0, + "step": 8425 + }, + { + "epoch": 1.0718738074036382, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8007610440254211, + "num_tokens": 321381674.0, + "step": 8426 + }, + { + "epoch": 1.0720010176822288, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.8115840554237366, + "num_tokens": 321420342.0, + "step": 8427 + }, + { + "epoch": 1.0721282279608193, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7876449823379517, + "num_tokens": 321450770.0, + "step": 8428 + }, + { + "epoch": 1.0722554382394098, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8040717244148254, + "num_tokens": 321491450.0, + "step": 8429 + }, + { + "epoch": 1.0723826485180004, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7907751798629761, + "num_tokens": 321531774.0, + "step": 8430 + }, + { + "epoch": 1.0725098587965907, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7972297668457031, + "num_tokens": 321567692.0, + "step": 8431 + }, + { + "epoch": 1.0726370690751812, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.787440299987793, + "num_tokens": 321601168.0, + "step": 8432 + }, + { + "epoch": 1.0727642793537717, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7742346525192261, + "num_tokens": 321636100.0, + "step": 8433 + }, + { + "epoch": 1.0728914896323622, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.799677312374115, + "num_tokens": 321674365.0, + "step": 8434 + }, + { + "epoch": 1.0730186999109528, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8065187931060791, + "num_tokens": 321712267.0, + "step": 8435 + }, + { + "epoch": 1.0731459101895433, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7911937236785889, + "num_tokens": 321749955.0, + "step": 8436 + }, + { + "epoch": 1.0732731204681338, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8078736066818237, + "num_tokens": 321788522.0, + "step": 8437 + }, + { + "epoch": 1.0734003307467244, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7940791249275208, + "num_tokens": 321826314.0, + "step": 8438 + }, + { + "epoch": 1.073527541025315, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7925878763198853, + "num_tokens": 321864851.0, + "step": 8439 + }, + { + "epoch": 1.0736547513039054, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8069831132888794, + "num_tokens": 321901001.0, + "step": 8440 + }, + { + "epoch": 1.073781961582496, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7965629696846008, + "num_tokens": 321941368.0, + "step": 8441 + }, + { + "epoch": 1.0739091718610865, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.801474392414093, + "num_tokens": 321978268.0, + "step": 8442 + }, + { + "epoch": 1.0740363821396768, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.806350827217102, + "num_tokens": 322016969.0, + "step": 8443 + }, + { + "epoch": 1.0741635924182673, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7909931540489197, + "num_tokens": 322059362.0, + "step": 8444 + }, + { + "epoch": 1.0742908026968578, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7821372747421265, + "num_tokens": 322097334.0, + "step": 8445 + }, + { + "epoch": 1.0744180129754484, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8113759160041809, + "num_tokens": 322121988.0, + "step": 8446 + }, + { + "epoch": 1.074545223254039, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7936339378356934, + "num_tokens": 322156602.0, + "step": 8447 + }, + { + "epoch": 1.0746724335326294, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7948671579360962, + "num_tokens": 322194475.0, + "step": 8448 + }, + { + "epoch": 1.07479964381122, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8051550388336182, + "num_tokens": 322232158.0, + "step": 8449 + }, + { + "epoch": 1.0749268540898105, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8061873912811279, + "num_tokens": 322272170.0, + "step": 8450 + }, + { + "epoch": 1.075054064368401, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8043547868728638, + "num_tokens": 322305496.0, + "step": 8451 + }, + { + "epoch": 1.0751812746469915, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8001832962036133, + "num_tokens": 322347552.0, + "step": 8452 + }, + { + "epoch": 1.075308484925582, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7986553311347961, + "num_tokens": 322386471.0, + "step": 8453 + }, + { + "epoch": 1.0754356952041726, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7614105343818665, + "num_tokens": 322431019.0, + "step": 8454 + }, + { + "epoch": 1.0755629054827631, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.818779468536377, + "num_tokens": 322471525.0, + "step": 8455 + }, + { + "epoch": 1.0756901157613534, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8047528862953186, + "num_tokens": 322516594.0, + "step": 8456 + }, + { + "epoch": 1.075817326039944, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8093578815460205, + "num_tokens": 322557837.0, + "step": 8457 + }, + { + "epoch": 1.0759445363185345, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7956790924072266, + "num_tokens": 322594165.0, + "step": 8458 + }, + { + "epoch": 1.076071746597125, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7808523178100586, + "num_tokens": 322637083.0, + "step": 8459 + }, + { + "epoch": 1.0761989568757155, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7953672409057617, + "num_tokens": 322673216.0, + "step": 8460 + }, + { + "epoch": 1.076326167154306, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7981782555580139, + "num_tokens": 322709150.0, + "step": 8461 + }, + { + "epoch": 1.0764533774328966, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8055216073989868, + "num_tokens": 322747893.0, + "step": 8462 + }, + { + "epoch": 1.0765805877114871, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7711951732635498, + "num_tokens": 322781667.0, + "step": 8463 + }, + { + "epoch": 1.0767077979900777, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7978779673576355, + "num_tokens": 322818008.0, + "step": 8464 + }, + { + "epoch": 1.0768350082686682, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7744698524475098, + "num_tokens": 322856528.0, + "step": 8465 + }, + { + "epoch": 1.0769622185472587, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7938215136528015, + "num_tokens": 322888798.0, + "step": 8466 + }, + { + "epoch": 1.077089428825849, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.793295681476593, + "num_tokens": 322923301.0, + "step": 8467 + }, + { + "epoch": 1.0772166391044395, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8136827945709229, + "num_tokens": 322964253.0, + "step": 8468 + }, + { + "epoch": 1.07734384938303, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.799862802028656, + "num_tokens": 323006926.0, + "step": 8469 + }, + { + "epoch": 1.0774710596616206, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8127740621566772, + "num_tokens": 323047541.0, + "step": 8470 + }, + { + "epoch": 1.0775982699402111, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.786423921585083, + "num_tokens": 323084272.0, + "step": 8471 + }, + { + "epoch": 1.0777254802188017, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8063541054725647, + "num_tokens": 323116698.0, + "step": 8472 + }, + { + "epoch": 1.0778526904973922, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8063461780548096, + "num_tokens": 323158836.0, + "step": 8473 + }, + { + "epoch": 1.0779799007759827, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7898694276809692, + "num_tokens": 323193777.0, + "step": 8474 + }, + { + "epoch": 1.0781071110545732, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.799953818321228, + "num_tokens": 323232546.0, + "step": 8475 + }, + { + "epoch": 1.0782343213331638, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.797683596611023, + "num_tokens": 323269403.0, + "step": 8476 + }, + { + "epoch": 1.0783615316117543, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8126652240753174, + "num_tokens": 323305982.0, + "step": 8477 + }, + { + "epoch": 1.0784887418903448, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7851870059967041, + "num_tokens": 323341032.0, + "step": 8478 + }, + { + "epoch": 1.0786159521689354, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7952718734741211, + "num_tokens": 323381647.0, + "step": 8479 + }, + { + "epoch": 1.0787431624475257, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7976513504981995, + "num_tokens": 323417185.0, + "step": 8480 + }, + { + "epoch": 1.0788703727261162, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7984087467193604, + "num_tokens": 323454624.0, + "step": 8481 + }, + { + "epoch": 1.0789975830047067, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.801005482673645, + "num_tokens": 323492862.0, + "step": 8482 + }, + { + "epoch": 1.0791247932832972, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7957625389099121, + "num_tokens": 323536012.0, + "step": 8483 + }, + { + "epoch": 1.0792520035618878, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8011330366134644, + "num_tokens": 323573495.0, + "step": 8484 + }, + { + "epoch": 1.0793792138404783, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8086649179458618, + "num_tokens": 323610900.0, + "step": 8485 + }, + { + "epoch": 1.0795064241190688, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7906107306480408, + "num_tokens": 323650983.0, + "step": 8486 + }, + { + "epoch": 1.0796336343976594, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7903409004211426, + "num_tokens": 323690155.0, + "step": 8487 + }, + { + "epoch": 1.0797608446762499, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.8185713291168213, + "num_tokens": 323732212.0, + "step": 8488 + }, + { + "epoch": 1.0798880549548404, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8101334571838379, + "num_tokens": 323773400.0, + "step": 8489 + }, + { + "epoch": 1.080015265233431, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.799824595451355, + "num_tokens": 323812753.0, + "step": 8490 + }, + { + "epoch": 1.0801424755120215, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8023499846458435, + "num_tokens": 323855835.0, + "step": 8491 + }, + { + "epoch": 1.0802696857906118, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8181211352348328, + "num_tokens": 323896065.0, + "step": 8492 + }, + { + "epoch": 1.0803968960692023, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8008830547332764, + "num_tokens": 323934351.0, + "step": 8493 + }, + { + "epoch": 1.0805241063477928, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7800487279891968, + "num_tokens": 323968878.0, + "step": 8494 + }, + { + "epoch": 1.0806513166263834, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7957995533943176, + "num_tokens": 324004838.0, + "step": 8495 + }, + { + "epoch": 1.080778526904974, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8102245330810547, + "num_tokens": 324040851.0, + "step": 8496 + }, + { + "epoch": 1.0809057371835644, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7863685488700867, + "num_tokens": 324084675.0, + "step": 8497 + }, + { + "epoch": 1.081032947462155, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8065091967582703, + "num_tokens": 324122136.0, + "step": 8498 + }, + { + "epoch": 1.0811601577407455, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7928210496902466, + "num_tokens": 324153142.0, + "step": 8499 + }, + { + "epoch": 1.081287368019336, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7773652076721191, + "num_tokens": 324187244.0, + "step": 8500 + }, + { + "epoch": 1.0814145782979265, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7992420196533203, + "num_tokens": 324225644.0, + "step": 8501 + }, + { + "epoch": 1.081541788576517, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8072373270988464, + "num_tokens": 324270394.0, + "step": 8502 + }, + { + "epoch": 1.0816689988551076, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8107857704162598, + "num_tokens": 324311244.0, + "step": 8503 + }, + { + "epoch": 1.0817962091336981, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7780095338821411, + "num_tokens": 324350821.0, + "step": 8504 + }, + { + "epoch": 1.0819234194122884, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.80070960521698, + "num_tokens": 324388255.0, + "step": 8505 + }, + { + "epoch": 1.082050629690879, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7752572298049927, + "num_tokens": 324430068.0, + "step": 8506 + }, + { + "epoch": 1.0821778399694695, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8045228719711304, + "num_tokens": 324465341.0, + "step": 8507 + }, + { + "epoch": 1.08230505024806, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8063955307006836, + "num_tokens": 324502199.0, + "step": 8508 + }, + { + "epoch": 1.0824322605266505, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.798220157623291, + "num_tokens": 324535898.0, + "step": 8509 + }, + { + "epoch": 1.082559470805241, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7892876863479614, + "num_tokens": 324574404.0, + "step": 8510 + }, + { + "epoch": 1.0826866810838316, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7936522960662842, + "num_tokens": 324611743.0, + "step": 8511 + }, + { + "epoch": 1.0828138913624221, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7908843159675598, + "num_tokens": 324653126.0, + "step": 8512 + }, + { + "epoch": 1.0829411016410126, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8053262829780579, + "num_tokens": 324692957.0, + "step": 8513 + }, + { + "epoch": 1.0830683119196032, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7968244552612305, + "num_tokens": 324737214.0, + "step": 8514 + }, + { + "epoch": 1.0831955221981937, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8077377676963806, + "num_tokens": 324773098.0, + "step": 8515 + }, + { + "epoch": 1.083322732476784, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7858986854553223, + "num_tokens": 324815846.0, + "step": 8516 + }, + { + "epoch": 1.0834499427553745, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7966305017471313, + "num_tokens": 324855912.0, + "step": 8517 + }, + { + "epoch": 1.083577153033965, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7895828485488892, + "num_tokens": 324893090.0, + "step": 8518 + }, + { + "epoch": 1.0837043633125556, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7980138659477234, + "num_tokens": 324929779.0, + "step": 8519 + }, + { + "epoch": 1.0838315735911461, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8010517954826355, + "num_tokens": 324961349.0, + "step": 8520 + }, + { + "epoch": 1.0839587838697367, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7975367903709412, + "num_tokens": 324998688.0, + "step": 8521 + }, + { + "epoch": 1.0840859941483272, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7896120548248291, + "num_tokens": 325040601.0, + "step": 8522 + }, + { + "epoch": 1.0842132044269177, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8081909418106079, + "num_tokens": 325081729.0, + "step": 8523 + }, + { + "epoch": 1.0843404147055082, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7838222980499268, + "num_tokens": 325118886.0, + "step": 8524 + }, + { + "epoch": 1.0844676249840988, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8195855617523193, + "num_tokens": 325156417.0, + "step": 8525 + }, + { + "epoch": 1.0845948352626893, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8125860095024109, + "num_tokens": 325200902.0, + "step": 8526 + }, + { + "epoch": 1.0847220455412798, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7937067747116089, + "num_tokens": 325233362.0, + "step": 8527 + }, + { + "epoch": 1.0848492558198704, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7965548038482666, + "num_tokens": 325274646.0, + "step": 8528 + }, + { + "epoch": 1.0849764660984607, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8117783069610596, + "num_tokens": 325304735.0, + "step": 8529 + }, + { + "epoch": 1.0851036763770512, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.8007234334945679, + "num_tokens": 325339833.0, + "step": 8530 + }, + { + "epoch": 1.0852308866556417, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7794530391693115, + "num_tokens": 325373838.0, + "step": 8531 + }, + { + "epoch": 1.0853580969342322, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.5846, + "mean_token_accuracy": 0.8212205767631531, + "num_tokens": 325412913.0, + "step": 8532 + }, + { + "epoch": 1.0854853072128228, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8020929098129272, + "num_tokens": 325451830.0, + "step": 8533 + }, + { + "epoch": 1.0856125174914133, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7801029682159424, + "num_tokens": 325487162.0, + "step": 8534 + }, + { + "epoch": 1.0857397277700038, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8102201223373413, + "num_tokens": 325525906.0, + "step": 8535 + }, + { + "epoch": 1.0858669380485944, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.808225691318512, + "num_tokens": 325559925.0, + "step": 8536 + }, + { + "epoch": 1.0859941483271849, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.799238383769989, + "num_tokens": 325599980.0, + "step": 8537 + }, + { + "epoch": 1.0861213586057754, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7795202732086182, + "num_tokens": 325631745.0, + "step": 8538 + }, + { + "epoch": 1.086248568884366, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7916573286056519, + "num_tokens": 325673583.0, + "step": 8539 + }, + { + "epoch": 1.0863757791629565, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7947301864624023, + "num_tokens": 325712995.0, + "step": 8540 + }, + { + "epoch": 1.0865029894415468, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8034945130348206, + "num_tokens": 325748259.0, + "step": 8541 + }, + { + "epoch": 1.0866301997201373, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8069971203804016, + "num_tokens": 325787676.0, + "step": 8542 + }, + { + "epoch": 1.0867574099987278, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7721251249313354, + "num_tokens": 325827819.0, + "step": 8543 + }, + { + "epoch": 1.0868846202773184, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8104664087295532, + "num_tokens": 325866328.0, + "step": 8544 + }, + { + "epoch": 1.0870118305559089, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7949211001396179, + "num_tokens": 325906752.0, + "step": 8545 + }, + { + "epoch": 1.0871390408344994, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7985026836395264, + "num_tokens": 325943765.0, + "step": 8546 + }, + { + "epoch": 1.08726625111309, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8011685013771057, + "num_tokens": 325981471.0, + "step": 8547 + }, + { + "epoch": 1.0873934613916805, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8036967515945435, + "num_tokens": 326017253.0, + "step": 8548 + }, + { + "epoch": 1.087520671670271, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7954494953155518, + "num_tokens": 326053484.0, + "step": 8549 + }, + { + "epoch": 1.0876478819488615, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.8008874654769897, + "num_tokens": 326084779.0, + "step": 8550 + }, + { + "epoch": 1.087775092227452, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7928727865219116, + "num_tokens": 326120795.0, + "step": 8551 + }, + { + "epoch": 1.0879023025060426, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7958173751831055, + "num_tokens": 326155991.0, + "step": 8552 + }, + { + "epoch": 1.0880295127846331, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7947053909301758, + "num_tokens": 326195738.0, + "step": 8553 + }, + { + "epoch": 1.0881567230632234, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7983155250549316, + "num_tokens": 326237773.0, + "step": 8554 + }, + { + "epoch": 1.088283933341814, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8123724460601807, + "num_tokens": 326278589.0, + "step": 8555 + }, + { + "epoch": 1.0884111436204045, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7925213575363159, + "num_tokens": 326316862.0, + "step": 8556 + }, + { + "epoch": 1.088538353898995, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7926376461982727, + "num_tokens": 326353892.0, + "step": 8557 + }, + { + "epoch": 1.0886655641775855, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.8028864860534668, + "num_tokens": 326389478.0, + "step": 8558 + }, + { + "epoch": 1.088792774456176, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7902514934539795, + "num_tokens": 326428072.0, + "step": 8559 + }, + { + "epoch": 1.0889199847347666, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7988705039024353, + "num_tokens": 326466374.0, + "step": 8560 + }, + { + "epoch": 1.0890471950133571, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7968721389770508, + "num_tokens": 326501081.0, + "step": 8561 + }, + { + "epoch": 1.0891744052919476, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8004958629608154, + "num_tokens": 326532539.0, + "step": 8562 + }, + { + "epoch": 1.0893016155705382, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8041598796844482, + "num_tokens": 326569481.0, + "step": 8563 + }, + { + "epoch": 1.0894288258491287, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7966264486312866, + "num_tokens": 326609746.0, + "step": 8564 + }, + { + "epoch": 1.089556036127719, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7900819778442383, + "num_tokens": 326652379.0, + "step": 8565 + }, + { + "epoch": 1.0896832464063095, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7958191633224487, + "num_tokens": 326687617.0, + "step": 8566 + }, + { + "epoch": 1.0898104566849, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7846106886863708, + "num_tokens": 326724155.0, + "step": 8567 + }, + { + "epoch": 1.0899376669634906, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7961649298667908, + "num_tokens": 326768234.0, + "step": 8568 + }, + { + "epoch": 1.0900648772420811, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8035159111022949, + "num_tokens": 326810423.0, + "step": 8569 + }, + { + "epoch": 1.0901920875206716, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8010632991790771, + "num_tokens": 326851834.0, + "step": 8570 + }, + { + "epoch": 1.0903192977992622, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7814141511917114, + "num_tokens": 326889334.0, + "step": 8571 + }, + { + "epoch": 1.0904465080778527, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7822984457015991, + "num_tokens": 326931884.0, + "step": 8572 + }, + { + "epoch": 1.0905737183564432, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8042962551116943, + "num_tokens": 326969043.0, + "step": 8573 + }, + { + "epoch": 1.0907009286350338, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7988736629486084, + "num_tokens": 326998369.0, + "step": 8574 + }, + { + "epoch": 1.0908281389136243, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8001473546028137, + "num_tokens": 327036909.0, + "step": 8575 + }, + { + "epoch": 1.0909553491922148, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7895092964172363, + "num_tokens": 327076053.0, + "step": 8576 + }, + { + "epoch": 1.0910825594708053, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8042817115783691, + "num_tokens": 327112285.0, + "step": 8577 + }, + { + "epoch": 1.0912097697493957, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8045544624328613, + "num_tokens": 327151591.0, + "step": 8578 + }, + { + "epoch": 1.0913369800279862, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7787569761276245, + "num_tokens": 327193600.0, + "step": 8579 + }, + { + "epoch": 1.0914641903065767, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.783929705619812, + "num_tokens": 327230249.0, + "step": 8580 + }, + { + "epoch": 1.0915914005851672, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8110326528549194, + "num_tokens": 327270043.0, + "step": 8581 + }, + { + "epoch": 1.0917186108637578, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7971769571304321, + "num_tokens": 327309396.0, + "step": 8582 + }, + { + "epoch": 1.0918458211423483, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8097074031829834, + "num_tokens": 327349497.0, + "step": 8583 + }, + { + "epoch": 1.0919730314209388, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7898002862930298, + "num_tokens": 327385648.0, + "step": 8584 + }, + { + "epoch": 1.0921002416995294, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8024045825004578, + "num_tokens": 327426379.0, + "step": 8585 + }, + { + "epoch": 1.0922274519781199, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8048752546310425, + "num_tokens": 327467555.0, + "step": 8586 + }, + { + "epoch": 1.0923546622567104, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.797681987285614, + "num_tokens": 327505303.0, + "step": 8587 + }, + { + "epoch": 1.092481872535301, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8115159273147583, + "num_tokens": 327547854.0, + "step": 8588 + }, + { + "epoch": 1.0926090828138915, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7948476076126099, + "num_tokens": 327584132.0, + "step": 8589 + }, + { + "epoch": 1.0927362930924818, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7889425754547119, + "num_tokens": 327629250.0, + "step": 8590 + }, + { + "epoch": 1.0928635033710723, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.789341926574707, + "num_tokens": 327666988.0, + "step": 8591 + }, + { + "epoch": 1.0929907136496628, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7993471622467041, + "num_tokens": 327705955.0, + "step": 8592 + }, + { + "epoch": 1.0931179239282534, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7924656271934509, + "num_tokens": 327746362.0, + "step": 8593 + }, + { + "epoch": 1.0932451342068439, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8012678623199463, + "num_tokens": 327789034.0, + "step": 8594 + }, + { + "epoch": 1.0933723444854344, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7895001173019409, + "num_tokens": 327824203.0, + "step": 8595 + }, + { + "epoch": 1.093499554764025, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7994170784950256, + "num_tokens": 327860340.0, + "step": 8596 + }, + { + "epoch": 1.0936267650426155, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7907912135124207, + "num_tokens": 327901709.0, + "step": 8597 + }, + { + "epoch": 1.093753975321206, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7993971109390259, + "num_tokens": 327940327.0, + "step": 8598 + }, + { + "epoch": 1.0938811855997965, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8073740005493164, + "num_tokens": 327977966.0, + "step": 8599 + }, + { + "epoch": 1.094008395878387, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7885056734085083, + "num_tokens": 328021143.0, + "step": 8600 + }, + { + "epoch": 1.0941356061569776, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8042101263999939, + "num_tokens": 328060792.0, + "step": 8601 + }, + { + "epoch": 1.094262816435568, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.796278715133667, + "num_tokens": 328099698.0, + "step": 8602 + }, + { + "epoch": 1.0943900267141584, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7952145338058472, + "num_tokens": 328141075.0, + "step": 8603 + }, + { + "epoch": 1.094517236992749, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7862163782119751, + "num_tokens": 328178945.0, + "step": 8604 + }, + { + "epoch": 1.0946444472713395, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7669131755828857, + "num_tokens": 328214423.0, + "step": 8605 + }, + { + "epoch": 1.09477165754993, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7980859279632568, + "num_tokens": 328248640.0, + "step": 8606 + }, + { + "epoch": 1.0948988678285205, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7723257541656494, + "num_tokens": 328283016.0, + "step": 8607 + }, + { + "epoch": 1.095026078107111, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8031118512153625, + "num_tokens": 328322377.0, + "step": 8608 + }, + { + "epoch": 1.0951532883857016, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7937780618667603, + "num_tokens": 328359409.0, + "step": 8609 + }, + { + "epoch": 1.0952804986642921, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8069753646850586, + "num_tokens": 328401257.0, + "step": 8610 + }, + { + "epoch": 1.0954077089428826, + "grad_norm": 86.5, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.799561619758606, + "num_tokens": 328435886.0, + "step": 8611 + }, + { + "epoch": 1.0955349192214732, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7846112847328186, + "num_tokens": 328472091.0, + "step": 8612 + }, + { + "epoch": 1.0956621295000637, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7890979051589966, + "num_tokens": 328509453.0, + "step": 8613 + }, + { + "epoch": 1.095789339778654, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7992078065872192, + "num_tokens": 328545590.0, + "step": 8614 + }, + { + "epoch": 1.0959165500572445, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7863316535949707, + "num_tokens": 328583093.0, + "step": 8615 + }, + { + "epoch": 1.096043760335835, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7848957777023315, + "num_tokens": 328625312.0, + "step": 8616 + }, + { + "epoch": 1.0961709706144256, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7893826961517334, + "num_tokens": 328655392.0, + "step": 8617 + }, + { + "epoch": 1.0962981808930161, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7803640365600586, + "num_tokens": 328693686.0, + "step": 8618 + }, + { + "epoch": 1.0964253911716066, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8061454892158508, + "num_tokens": 328728051.0, + "step": 8619 + }, + { + "epoch": 1.0965526014501972, + "grad_norm": 10.4375, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.786660373210907, + "num_tokens": 328768091.0, + "step": 8620 + }, + { + "epoch": 1.0966798117287877, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7871588468551636, + "num_tokens": 328803851.0, + "step": 8621 + }, + { + "epoch": 1.0968070220073782, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8064604997634888, + "num_tokens": 328837429.0, + "step": 8622 + }, + { + "epoch": 1.0969342322859688, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7933762669563293, + "num_tokens": 328874264.0, + "step": 8623 + }, + { + "epoch": 1.0970614425645593, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7932265996932983, + "num_tokens": 328911627.0, + "step": 8624 + }, + { + "epoch": 1.0971886528431498, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7845954298973083, + "num_tokens": 328944471.0, + "step": 8625 + }, + { + "epoch": 1.0973158631217403, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8068221807479858, + "num_tokens": 328983826.0, + "step": 8626 + }, + { + "epoch": 1.0974430734003306, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7909302711486816, + "num_tokens": 329019385.0, + "step": 8627 + }, + { + "epoch": 1.0975702836789212, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7860919237136841, + "num_tokens": 329065179.0, + "step": 8628 + }, + { + "epoch": 1.0976974939575117, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.807044267654419, + "num_tokens": 329101077.0, + "step": 8629 + }, + { + "epoch": 1.0978247042361022, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8017462491989136, + "num_tokens": 329137538.0, + "step": 8630 + }, + { + "epoch": 1.0979519145146928, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8113898038864136, + "num_tokens": 329171952.0, + "step": 8631 + }, + { + "epoch": 1.0980791247932833, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8075999617576599, + "num_tokens": 329214956.0, + "step": 8632 + }, + { + "epoch": 1.0982063350718738, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7853146195411682, + "num_tokens": 329254228.0, + "step": 8633 + }, + { + "epoch": 1.0983335453504643, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7947590351104736, + "num_tokens": 329288337.0, + "step": 8634 + }, + { + "epoch": 1.0984607556290549, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7894359827041626, + "num_tokens": 329329118.0, + "step": 8635 + }, + { + "epoch": 1.0985879659076454, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7780282497406006, + "num_tokens": 329366301.0, + "step": 8636 + }, + { + "epoch": 1.098715176186236, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7796212434768677, + "num_tokens": 329401612.0, + "step": 8637 + }, + { + "epoch": 1.0988423864648265, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.798560380935669, + "num_tokens": 329440403.0, + "step": 8638 + }, + { + "epoch": 1.0989695967434168, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.806057870388031, + "num_tokens": 329478699.0, + "step": 8639 + }, + { + "epoch": 1.0990968070220073, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7997462749481201, + "num_tokens": 329517860.0, + "step": 8640 + }, + { + "epoch": 1.0992240173005978, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8008526563644409, + "num_tokens": 329555385.0, + "step": 8641 + }, + { + "epoch": 1.0993512275791884, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.791038990020752, + "num_tokens": 329594775.0, + "step": 8642 + }, + { + "epoch": 1.0994784378577789, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8023434281349182, + "num_tokens": 329629872.0, + "step": 8643 + }, + { + "epoch": 1.0996056481363694, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7904272079467773, + "num_tokens": 329665263.0, + "step": 8644 + }, + { + "epoch": 1.09973285841496, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7846330404281616, + "num_tokens": 329705853.0, + "step": 8645 + }, + { + "epoch": 1.0998600686935505, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8067764043807983, + "num_tokens": 329741690.0, + "step": 8646 + }, + { + "epoch": 1.099987278972141, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.792823076248169, + "num_tokens": 329781548.0, + "step": 8647 + }, + { + "epoch": 1.1001144892507315, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8067647218704224, + "num_tokens": 329822304.0, + "step": 8648 + }, + { + "epoch": 1.100241699529322, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7978816628456116, + "num_tokens": 329856065.0, + "step": 8649 + }, + { + "epoch": 1.1003689098079126, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7797194123268127, + "num_tokens": 329895002.0, + "step": 8650 + }, + { + "epoch": 1.100496120086503, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8013867735862732, + "num_tokens": 329938810.0, + "step": 8651 + }, + { + "epoch": 1.1006233303650934, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8101193904876709, + "num_tokens": 329976513.0, + "step": 8652 + }, + { + "epoch": 1.100750540643684, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8041049242019653, + "num_tokens": 330014013.0, + "step": 8653 + }, + { + "epoch": 1.1008777509222745, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7954232692718506, + "num_tokens": 330052577.0, + "step": 8654 + }, + { + "epoch": 1.101004961200865, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7952224016189575, + "num_tokens": 330091385.0, + "step": 8655 + }, + { + "epoch": 1.1011321714794555, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8075483441352844, + "num_tokens": 330130285.0, + "step": 8656 + }, + { + "epoch": 1.101259381758046, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7816414833068848, + "num_tokens": 330166823.0, + "step": 8657 + }, + { + "epoch": 1.1013865920366366, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8022751808166504, + "num_tokens": 330200655.0, + "step": 8658 + }, + { + "epoch": 1.101513802315227, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.8029299974441528, + "num_tokens": 330243656.0, + "step": 8659 + }, + { + "epoch": 1.1016410125938176, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7880589962005615, + "num_tokens": 330287200.0, + "step": 8660 + }, + { + "epoch": 1.1017682228724082, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8040866851806641, + "num_tokens": 330324794.0, + "step": 8661 + }, + { + "epoch": 1.1018954331509987, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7986341118812561, + "num_tokens": 330365297.0, + "step": 8662 + }, + { + "epoch": 1.102022643429589, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8160272240638733, + "num_tokens": 330408046.0, + "step": 8663 + }, + { + "epoch": 1.1021498537081795, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7968676090240479, + "num_tokens": 330447815.0, + "step": 8664 + }, + { + "epoch": 1.10227706398677, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7837210893630981, + "num_tokens": 330487796.0, + "step": 8665 + }, + { + "epoch": 1.1024042742653606, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8103702068328857, + "num_tokens": 330531155.0, + "step": 8666 + }, + { + "epoch": 1.1025314845439511, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8064829111099243, + "num_tokens": 330568611.0, + "step": 8667 + }, + { + "epoch": 1.1026586948225416, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7939034700393677, + "num_tokens": 330608085.0, + "step": 8668 + }, + { + "epoch": 1.1027859051011322, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7782608270645142, + "num_tokens": 330644776.0, + "step": 8669 + }, + { + "epoch": 1.1029131153797227, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7891892194747925, + "num_tokens": 330686401.0, + "step": 8670 + }, + { + "epoch": 1.1030403256583132, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7982831001281738, + "num_tokens": 330723115.0, + "step": 8671 + }, + { + "epoch": 1.1031675359369038, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8149523138999939, + "num_tokens": 330761170.0, + "step": 8672 + }, + { + "epoch": 1.1032947462154943, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7894521355628967, + "num_tokens": 330790230.0, + "step": 8673 + }, + { + "epoch": 1.1034219564940848, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8071396350860596, + "num_tokens": 330826178.0, + "step": 8674 + }, + { + "epoch": 1.1035491667726753, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8007118701934814, + "num_tokens": 330866341.0, + "step": 8675 + }, + { + "epoch": 1.1036763770512656, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7822108268737793, + "num_tokens": 330903955.0, + "step": 8676 + }, + { + "epoch": 1.1038035873298562, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.797720193862915, + "num_tokens": 330943024.0, + "step": 8677 + }, + { + "epoch": 1.1039307976084467, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7905763983726501, + "num_tokens": 330981004.0, + "step": 8678 + }, + { + "epoch": 1.1040580078870372, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8028275370597839, + "num_tokens": 331021532.0, + "step": 8679 + }, + { + "epoch": 1.1041852181656278, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8001373410224915, + "num_tokens": 331058010.0, + "step": 8680 + }, + { + "epoch": 1.1043124284442183, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8044629096984863, + "num_tokens": 331090713.0, + "step": 8681 + }, + { + "epoch": 1.1044396387228088, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8039788603782654, + "num_tokens": 331127532.0, + "step": 8682 + }, + { + "epoch": 1.1045668490013993, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7735955715179443, + "num_tokens": 331172088.0, + "step": 8683 + }, + { + "epoch": 1.1046940592799899, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7925492525100708, + "num_tokens": 331213419.0, + "step": 8684 + }, + { + "epoch": 1.1048212695585804, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.796772301197052, + "num_tokens": 331250102.0, + "step": 8685 + }, + { + "epoch": 1.104948479837171, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7955130338668823, + "num_tokens": 331284597.0, + "step": 8686 + }, + { + "epoch": 1.1050756901157615, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7896758913993835, + "num_tokens": 331330688.0, + "step": 8687 + }, + { + "epoch": 1.1052029003943518, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.802223801612854, + "num_tokens": 331368777.0, + "step": 8688 + }, + { + "epoch": 1.1053301106729423, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8041863441467285, + "num_tokens": 331414369.0, + "step": 8689 + }, + { + "epoch": 1.1054573209515328, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8013688921928406, + "num_tokens": 331452788.0, + "step": 8690 + }, + { + "epoch": 1.1055845312301233, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7810775637626648, + "num_tokens": 331489575.0, + "step": 8691 + }, + { + "epoch": 1.1057117415087139, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8153115510940552, + "num_tokens": 331529129.0, + "step": 8692 + }, + { + "epoch": 1.1058389517873044, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7939519882202148, + "num_tokens": 331564989.0, + "step": 8693 + }, + { + "epoch": 1.105966162065895, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7995589971542358, + "num_tokens": 331602561.0, + "step": 8694 + }, + { + "epoch": 1.1060933723444855, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7983462810516357, + "num_tokens": 331646902.0, + "step": 8695 + }, + { + "epoch": 1.106220582623076, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8013135194778442, + "num_tokens": 331682347.0, + "step": 8696 + }, + { + "epoch": 1.1063477929016665, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7904424071311951, + "num_tokens": 331725813.0, + "step": 8697 + }, + { + "epoch": 1.106475003180257, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7683656811714172, + "num_tokens": 331767856.0, + "step": 8698 + }, + { + "epoch": 1.1066022134588476, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8096227049827576, + "num_tokens": 331804238.0, + "step": 8699 + }, + { + "epoch": 1.106729423737438, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7941857576370239, + "num_tokens": 331839963.0, + "step": 8700 + }, + { + "epoch": 1.1068566340160284, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8040137887001038, + "num_tokens": 331881563.0, + "step": 8701 + }, + { + "epoch": 1.106983844294619, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7902675271034241, + "num_tokens": 331925540.0, + "step": 8702 + }, + { + "epoch": 1.1071110545732095, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8097425699234009, + "num_tokens": 331963171.0, + "step": 8703 + }, + { + "epoch": 1.1072382648518, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8150730133056641, + "num_tokens": 331998310.0, + "step": 8704 + }, + { + "epoch": 1.1073654751303905, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8090367317199707, + "num_tokens": 332034181.0, + "step": 8705 + }, + { + "epoch": 1.107492685408981, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8101025819778442, + "num_tokens": 332075368.0, + "step": 8706 + }, + { + "epoch": 1.1076198956875716, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8081193566322327, + "num_tokens": 332113281.0, + "step": 8707 + }, + { + "epoch": 1.107747105966162, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7889297008514404, + "num_tokens": 332154206.0, + "step": 8708 + }, + { + "epoch": 1.1078743162447526, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7792763113975525, + "num_tokens": 332195409.0, + "step": 8709 + }, + { + "epoch": 1.1080015265233432, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.80979323387146, + "num_tokens": 332233836.0, + "step": 8710 + }, + { + "epoch": 1.1081287368019337, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.802666187286377, + "num_tokens": 332274167.0, + "step": 8711 + }, + { + "epoch": 1.108255947080524, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7924715876579285, + "num_tokens": 332315856.0, + "step": 8712 + }, + { + "epoch": 1.1083831573591145, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8007767200469971, + "num_tokens": 332366365.0, + "step": 8713 + }, + { + "epoch": 1.108510367637705, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7924752235412598, + "num_tokens": 332400411.0, + "step": 8714 + }, + { + "epoch": 1.1086375779162956, + "grad_norm": 81.0, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.800166130065918, + "num_tokens": 332441959.0, + "step": 8715 + }, + { + "epoch": 1.108764788194886, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7863692045211792, + "num_tokens": 332479545.0, + "step": 8716 + }, + { + "epoch": 1.1088919984734766, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8102739453315735, + "num_tokens": 332516240.0, + "step": 8717 + }, + { + "epoch": 1.1090192087520672, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7992172241210938, + "num_tokens": 332553488.0, + "step": 8718 + }, + { + "epoch": 1.1091464190306577, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7854039669036865, + "num_tokens": 332588646.0, + "step": 8719 + }, + { + "epoch": 1.1092736293092482, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7885265350341797, + "num_tokens": 332629103.0, + "step": 8720 + }, + { + "epoch": 1.1094008395878387, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8007158637046814, + "num_tokens": 332664820.0, + "step": 8721 + }, + { + "epoch": 1.1095280498664293, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7983628511428833, + "num_tokens": 332699772.0, + "step": 8722 + }, + { + "epoch": 1.1096552601450198, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8083986639976501, + "num_tokens": 332733478.0, + "step": 8723 + }, + { + "epoch": 1.1097824704236103, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7972457408905029, + "num_tokens": 332774733.0, + "step": 8724 + }, + { + "epoch": 1.1099096807022006, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7698796987533569, + "num_tokens": 332812970.0, + "step": 8725 + }, + { + "epoch": 1.1100368909807912, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7917200326919556, + "num_tokens": 332854974.0, + "step": 8726 + }, + { + "epoch": 1.1101641012593817, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7820984125137329, + "num_tokens": 332886462.0, + "step": 8727 + }, + { + "epoch": 1.1102913115379722, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.794975996017456, + "num_tokens": 332928614.0, + "step": 8728 + }, + { + "epoch": 1.1104185218165628, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.796773374080658, + "num_tokens": 332962447.0, + "step": 8729 + }, + { + "epoch": 1.1105457320951533, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7819371223449707, + "num_tokens": 332991395.0, + "step": 8730 + }, + { + "epoch": 1.1106729423737438, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.788792610168457, + "num_tokens": 333027852.0, + "step": 8731 + }, + { + "epoch": 1.1108001526523343, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7984241247177124, + "num_tokens": 333063337.0, + "step": 8732 + }, + { + "epoch": 1.1109273629309249, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.781175434589386, + "num_tokens": 333094632.0, + "step": 8733 + }, + { + "epoch": 1.1110545732095154, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.8115354776382446, + "num_tokens": 333134071.0, + "step": 8734 + }, + { + "epoch": 1.111181783488106, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8000971078872681, + "num_tokens": 333171288.0, + "step": 8735 + }, + { + "epoch": 1.1113089937666965, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8038914203643799, + "num_tokens": 333212111.0, + "step": 8736 + }, + { + "epoch": 1.1114362040452868, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.8003189563751221, + "num_tokens": 333247557.0, + "step": 8737 + }, + { + "epoch": 1.1115634143238773, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7889812588691711, + "num_tokens": 333278705.0, + "step": 8738 + }, + { + "epoch": 1.1116906246024678, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7997866868972778, + "num_tokens": 333316493.0, + "step": 8739 + }, + { + "epoch": 1.1118178348810583, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8039253950119019, + "num_tokens": 333360384.0, + "step": 8740 + }, + { + "epoch": 1.1119450451596489, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7931910753250122, + "num_tokens": 333395568.0, + "step": 8741 + }, + { + "epoch": 1.1120722554382394, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8019489049911499, + "num_tokens": 333429267.0, + "step": 8742 + }, + { + "epoch": 1.11219946571683, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8031580448150635, + "num_tokens": 333469908.0, + "step": 8743 + }, + { + "epoch": 1.1123266759954205, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.803592324256897, + "num_tokens": 333507865.0, + "step": 8744 + }, + { + "epoch": 1.112453886274011, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8092140555381775, + "num_tokens": 333544011.0, + "step": 8745 + }, + { + "epoch": 1.1125810965526015, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8033215403556824, + "num_tokens": 333581949.0, + "step": 8746 + }, + { + "epoch": 1.112708306831192, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8027287125587463, + "num_tokens": 333615994.0, + "step": 8747 + }, + { + "epoch": 1.1128355171097826, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8110265135765076, + "num_tokens": 333654454.0, + "step": 8748 + }, + { + "epoch": 1.112962727388373, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.789442777633667, + "num_tokens": 333700206.0, + "step": 8749 + }, + { + "epoch": 1.1130899376669634, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.800571084022522, + "num_tokens": 333737538.0, + "step": 8750 + }, + { + "epoch": 1.113217147945554, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7965694665908813, + "num_tokens": 333771237.0, + "step": 8751 + }, + { + "epoch": 1.1133443582241445, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7954869866371155, + "num_tokens": 333806522.0, + "step": 8752 + }, + { + "epoch": 1.113471568502735, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7977883219718933, + "num_tokens": 333846656.0, + "step": 8753 + }, + { + "epoch": 1.1135987787813255, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7883200645446777, + "num_tokens": 333886294.0, + "step": 8754 + }, + { + "epoch": 1.113725989059916, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.795149564743042, + "num_tokens": 333923868.0, + "step": 8755 + }, + { + "epoch": 1.1138531993385066, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.802309513092041, + "num_tokens": 333957187.0, + "step": 8756 + }, + { + "epoch": 1.113980409617097, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8155274987220764, + "num_tokens": 333999061.0, + "step": 8757 + }, + { + "epoch": 1.1141076198956876, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7938715219497681, + "num_tokens": 334043595.0, + "step": 8758 + }, + { + "epoch": 1.1142348301742782, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7972445487976074, + "num_tokens": 334078487.0, + "step": 8759 + }, + { + "epoch": 1.1143620404528687, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8073227405548096, + "num_tokens": 334119610.0, + "step": 8760 + }, + { + "epoch": 1.114489250731459, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.792000949382782, + "num_tokens": 334160149.0, + "step": 8761 + }, + { + "epoch": 1.1146164610100495, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7955377101898193, + "num_tokens": 334198667.0, + "step": 8762 + }, + { + "epoch": 1.11474367128864, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7967982888221741, + "num_tokens": 334233838.0, + "step": 8763 + }, + { + "epoch": 1.1148708815672306, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7885251045227051, + "num_tokens": 334275703.0, + "step": 8764 + }, + { + "epoch": 1.114998091845821, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8122697472572327, + "num_tokens": 334311119.0, + "step": 8765 + }, + { + "epoch": 1.1151253021244116, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7843212485313416, + "num_tokens": 334348298.0, + "step": 8766 + }, + { + "epoch": 1.1152525124030022, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7828280925750732, + "num_tokens": 334386986.0, + "step": 8767 + }, + { + "epoch": 1.1153797226815927, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8018814325332642, + "num_tokens": 334423263.0, + "step": 8768 + }, + { + "epoch": 1.1155069329601832, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8126051425933838, + "num_tokens": 334460363.0, + "step": 8769 + }, + { + "epoch": 1.1156341432387737, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8061354160308838, + "num_tokens": 334501883.0, + "step": 8770 + }, + { + "epoch": 1.1157613535173643, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7971097230911255, + "num_tokens": 334536242.0, + "step": 8771 + }, + { + "epoch": 1.1158885637959548, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8088558912277222, + "num_tokens": 334569685.0, + "step": 8772 + }, + { + "epoch": 1.1160157740745453, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7976018190383911, + "num_tokens": 334604946.0, + "step": 8773 + }, + { + "epoch": 1.1161429843531356, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8040616512298584, + "num_tokens": 334645495.0, + "step": 8774 + }, + { + "epoch": 1.1162701946317262, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7940791845321655, + "num_tokens": 334684239.0, + "step": 8775 + }, + { + "epoch": 1.1163974049103167, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8021457195281982, + "num_tokens": 334713799.0, + "step": 8776 + }, + { + "epoch": 1.1165246151889072, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8190785646438599, + "num_tokens": 334753374.0, + "step": 8777 + }, + { + "epoch": 1.1166518254674977, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7960810661315918, + "num_tokens": 334792867.0, + "step": 8778 + }, + { + "epoch": 1.1167790357460883, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7963703274726868, + "num_tokens": 334832664.0, + "step": 8779 + }, + { + "epoch": 1.1169062460246788, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7934104204177856, + "num_tokens": 334876763.0, + "step": 8780 + }, + { + "epoch": 1.1170334563032693, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7982683777809143, + "num_tokens": 334909431.0, + "step": 8781 + }, + { + "epoch": 1.1171606665818599, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7923756837844849, + "num_tokens": 334941939.0, + "step": 8782 + }, + { + "epoch": 1.1172878768604504, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8118321895599365, + "num_tokens": 334981338.0, + "step": 8783 + }, + { + "epoch": 1.117415087139041, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8022595047950745, + "num_tokens": 335020660.0, + "step": 8784 + }, + { + "epoch": 1.1175422974176314, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7988077402114868, + "num_tokens": 335057377.0, + "step": 8785 + }, + { + "epoch": 1.1176695076962218, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7849441170692444, + "num_tokens": 335092912.0, + "step": 8786 + }, + { + "epoch": 1.1177967179748123, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6097, + "mean_token_accuracy": 0.814647912979126, + "num_tokens": 335129157.0, + "step": 8787 + }, + { + "epoch": 1.1179239282534028, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7988874912261963, + "num_tokens": 335168065.0, + "step": 8788 + }, + { + "epoch": 1.1180511385319933, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7989139556884766, + "num_tokens": 335218876.0, + "step": 8789 + }, + { + "epoch": 1.1181783488105839, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.806011438369751, + "num_tokens": 335256300.0, + "step": 8790 + }, + { + "epoch": 1.1183055590891744, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8084741830825806, + "num_tokens": 335291035.0, + "step": 8791 + }, + { + "epoch": 1.118432769367765, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.812664806842804, + "num_tokens": 335330092.0, + "step": 8792 + }, + { + "epoch": 1.1185599796463555, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.788499116897583, + "num_tokens": 335367146.0, + "step": 8793 + }, + { + "epoch": 1.118687189924946, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7967588901519775, + "num_tokens": 335405069.0, + "step": 8794 + }, + { + "epoch": 1.1188144002035365, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7883093357086182, + "num_tokens": 335443779.0, + "step": 8795 + }, + { + "epoch": 1.118941610482127, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7902917861938477, + "num_tokens": 335485978.0, + "step": 8796 + }, + { + "epoch": 1.1190688207607176, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8022717833518982, + "num_tokens": 335530375.0, + "step": 8797 + }, + { + "epoch": 1.119196031039308, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8024083375930786, + "num_tokens": 335569109.0, + "step": 8798 + }, + { + "epoch": 1.1193232413178984, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7798742055892944, + "num_tokens": 335599283.0, + "step": 8799 + }, + { + "epoch": 1.119450451596489, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8028249740600586, + "num_tokens": 335634792.0, + "step": 8800 + }, + { + "epoch": 1.1195776618750795, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8129568696022034, + "num_tokens": 335672869.0, + "step": 8801 + }, + { + "epoch": 1.11970487215367, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8017931580543518, + "num_tokens": 335714996.0, + "step": 8802 + }, + { + "epoch": 1.1198320824322605, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7904728651046753, + "num_tokens": 335756884.0, + "step": 8803 + }, + { + "epoch": 1.119959292710851, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.809277355670929, + "num_tokens": 335793935.0, + "step": 8804 + }, + { + "epoch": 1.1200865029894416, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8098412156105042, + "num_tokens": 335834169.0, + "step": 8805 + }, + { + "epoch": 1.120213713268032, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8157666325569153, + "num_tokens": 335872302.0, + "step": 8806 + }, + { + "epoch": 1.1203409235466226, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8035671710968018, + "num_tokens": 335908074.0, + "step": 8807 + }, + { + "epoch": 1.1204681338252132, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8124240636825562, + "num_tokens": 335946268.0, + "step": 8808 + }, + { + "epoch": 1.1205953441038037, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7986233830451965, + "num_tokens": 335985364.0, + "step": 8809 + }, + { + "epoch": 1.120722554382394, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8016635179519653, + "num_tokens": 336025300.0, + "step": 8810 + }, + { + "epoch": 1.1208497646609845, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8010287284851074, + "num_tokens": 336062105.0, + "step": 8811 + }, + { + "epoch": 1.120976974939575, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7875475287437439, + "num_tokens": 336102944.0, + "step": 8812 + }, + { + "epoch": 1.1211041852181656, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.5622, + "mean_token_accuracy": 0.8320585489273071, + "num_tokens": 336142964.0, + "step": 8813 + }, + { + "epoch": 1.121231395496756, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8027869462966919, + "num_tokens": 336186962.0, + "step": 8814 + }, + { + "epoch": 1.1213586057753466, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8064171075820923, + "num_tokens": 336220775.0, + "step": 8815 + }, + { + "epoch": 1.1214858160539372, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7969471216201782, + "num_tokens": 336263067.0, + "step": 8816 + }, + { + "epoch": 1.1216130263325277, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.806614875793457, + "num_tokens": 336308463.0, + "step": 8817 + }, + { + "epoch": 1.1217402366111182, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8148917555809021, + "num_tokens": 336346965.0, + "step": 8818 + }, + { + "epoch": 1.1218674468897087, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7982984781265259, + "num_tokens": 336382114.0, + "step": 8819 + }, + { + "epoch": 1.1219946571682993, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.799547553062439, + "num_tokens": 336414020.0, + "step": 8820 + }, + { + "epoch": 1.1221218674468898, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7977474927902222, + "num_tokens": 336452522.0, + "step": 8821 + }, + { + "epoch": 1.1222490777254803, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8078596591949463, + "num_tokens": 336484593.0, + "step": 8822 + }, + { + "epoch": 1.1223762880040706, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7784602642059326, + "num_tokens": 336518964.0, + "step": 8823 + }, + { + "epoch": 1.1225034982826612, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7841179966926575, + "num_tokens": 336556320.0, + "step": 8824 + }, + { + "epoch": 1.1226307085612517, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7856757640838623, + "num_tokens": 336595473.0, + "step": 8825 + }, + { + "epoch": 1.1227579188398422, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7969926595687866, + "num_tokens": 336631793.0, + "step": 8826 + }, + { + "epoch": 1.1228851291184327, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7800339460372925, + "num_tokens": 336671230.0, + "step": 8827 + }, + { + "epoch": 1.1230123393970233, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8030699491500854, + "num_tokens": 336712015.0, + "step": 8828 + }, + { + "epoch": 1.1231395496756138, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7743692994117737, + "num_tokens": 336741206.0, + "step": 8829 + }, + { + "epoch": 1.1232667599542043, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7934877872467041, + "num_tokens": 336778838.0, + "step": 8830 + }, + { + "epoch": 1.1233939702327949, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.80547034740448, + "num_tokens": 336815471.0, + "step": 8831 + }, + { + "epoch": 1.1235211805113854, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8004608154296875, + "num_tokens": 336850870.0, + "step": 8832 + }, + { + "epoch": 1.123648390789976, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8052436113357544, + "num_tokens": 336891779.0, + "step": 8833 + }, + { + "epoch": 1.1237756010685664, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8059133291244507, + "num_tokens": 336931171.0, + "step": 8834 + }, + { + "epoch": 1.1239028113471567, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8098188042640686, + "num_tokens": 336963080.0, + "step": 8835 + }, + { + "epoch": 1.1240300216257473, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7841326594352722, + "num_tokens": 337000817.0, + "step": 8836 + }, + { + "epoch": 1.1241572319043378, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7919885516166687, + "num_tokens": 337037249.0, + "step": 8837 + }, + { + "epoch": 1.1242844421829283, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.808053195476532, + "num_tokens": 337077752.0, + "step": 8838 + }, + { + "epoch": 1.1244116524615189, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7955793142318726, + "num_tokens": 337119649.0, + "step": 8839 + }, + { + "epoch": 1.1245388627401094, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7785686254501343, + "num_tokens": 337154376.0, + "step": 8840 + }, + { + "epoch": 1.1246660730187, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8103305101394653, + "num_tokens": 337193377.0, + "step": 8841 + }, + { + "epoch": 1.1247932832972904, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7853317856788635, + "num_tokens": 337233176.0, + "step": 8842 + }, + { + "epoch": 1.124920493575881, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7790104150772095, + "num_tokens": 337262995.0, + "step": 8843 + }, + { + "epoch": 1.1250477038544715, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7953442931175232, + "num_tokens": 337302466.0, + "step": 8844 + }, + { + "epoch": 1.125174914133062, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8098304867744446, + "num_tokens": 337341197.0, + "step": 8845 + }, + { + "epoch": 1.1253021244116526, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7997719049453735, + "num_tokens": 337375602.0, + "step": 8846 + }, + { + "epoch": 1.125429334690243, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7882506847381592, + "num_tokens": 337406685.0, + "step": 8847 + }, + { + "epoch": 1.1255565449688334, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7937172055244446, + "num_tokens": 337447364.0, + "step": 8848 + }, + { + "epoch": 1.125683755247424, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7964425086975098, + "num_tokens": 337487840.0, + "step": 8849 + }, + { + "epoch": 1.1258109655260145, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8029720783233643, + "num_tokens": 337525117.0, + "step": 8850 + }, + { + "epoch": 1.125938175804605, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7907816171646118, + "num_tokens": 337567024.0, + "step": 8851 + }, + { + "epoch": 1.1260653860831955, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7959536910057068, + "num_tokens": 337600117.0, + "step": 8852 + }, + { + "epoch": 1.126192596361786, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7836315631866455, + "num_tokens": 337637188.0, + "step": 8853 + }, + { + "epoch": 1.1263198066403766, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7919380068778992, + "num_tokens": 337676492.0, + "step": 8854 + }, + { + "epoch": 1.126447016918967, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8098292350769043, + "num_tokens": 337713468.0, + "step": 8855 + }, + { + "epoch": 1.1265742271975576, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7877181768417358, + "num_tokens": 337750944.0, + "step": 8856 + }, + { + "epoch": 1.1267014374761481, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8139978051185608, + "num_tokens": 337793802.0, + "step": 8857 + }, + { + "epoch": 1.1268286477547387, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8130010366439819, + "num_tokens": 337828550.0, + "step": 8858 + }, + { + "epoch": 1.126955858033329, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8060880303382874, + "num_tokens": 337862168.0, + "step": 8859 + }, + { + "epoch": 1.1270830683119195, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.798699140548706, + "num_tokens": 337898837.0, + "step": 8860 + }, + { + "epoch": 1.12721027859051, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7868164777755737, + "num_tokens": 337939608.0, + "step": 8861 + }, + { + "epoch": 1.1273374888691006, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8053488731384277, + "num_tokens": 337983412.0, + "step": 8862 + }, + { + "epoch": 1.127464699147691, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8081865906715393, + "num_tokens": 338022171.0, + "step": 8863 + }, + { + "epoch": 1.1275919094262816, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7955093383789062, + "num_tokens": 338058838.0, + "step": 8864 + }, + { + "epoch": 1.1277191197048722, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7920321226119995, + "num_tokens": 338095080.0, + "step": 8865 + }, + { + "epoch": 1.1278463299834627, + "grad_norm": 61.5, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7803217172622681, + "num_tokens": 338134206.0, + "step": 8866 + }, + { + "epoch": 1.1279735402620532, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8079332709312439, + "num_tokens": 338173794.0, + "step": 8867 + }, + { + "epoch": 1.1281007505406437, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7919527292251587, + "num_tokens": 338206591.0, + "step": 8868 + }, + { + "epoch": 1.1282279608192343, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8110718727111816, + "num_tokens": 338238728.0, + "step": 8869 + }, + { + "epoch": 1.1283551710978248, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8118222951889038, + "num_tokens": 338282372.0, + "step": 8870 + }, + { + "epoch": 1.1284823813764153, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7942936420440674, + "num_tokens": 338319175.0, + "step": 8871 + }, + { + "epoch": 1.1286095916550056, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7976949214935303, + "num_tokens": 338357967.0, + "step": 8872 + }, + { + "epoch": 1.1287368019335962, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8085358142852783, + "num_tokens": 338391431.0, + "step": 8873 + }, + { + "epoch": 1.1288640122121867, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8089679479598999, + "num_tokens": 338426298.0, + "step": 8874 + }, + { + "epoch": 1.1289912224907772, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7803958654403687, + "num_tokens": 338468147.0, + "step": 8875 + }, + { + "epoch": 1.1291184327693677, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7832050323486328, + "num_tokens": 338517095.0, + "step": 8876 + }, + { + "epoch": 1.1292456430479583, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7986763715744019, + "num_tokens": 338549925.0, + "step": 8877 + }, + { + "epoch": 1.1293728533265488, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8049875497817993, + "num_tokens": 338583966.0, + "step": 8878 + }, + { + "epoch": 1.1295000636051393, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8005259037017822, + "num_tokens": 338622715.0, + "step": 8879 + }, + { + "epoch": 1.1296272738837299, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7822808027267456, + "num_tokens": 338659920.0, + "step": 8880 + }, + { + "epoch": 1.1297544841623204, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7960268259048462, + "num_tokens": 338698183.0, + "step": 8881 + }, + { + "epoch": 1.129881694440911, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7840880751609802, + "num_tokens": 338735853.0, + "step": 8882 + }, + { + "epoch": 1.1300089047195012, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7866395711898804, + "num_tokens": 338771714.0, + "step": 8883 + }, + { + "epoch": 1.1301361149980917, + "grad_norm": 10.375, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8210099935531616, + "num_tokens": 338813823.0, + "step": 8884 + }, + { + "epoch": 1.1302633252766823, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8023631572723389, + "num_tokens": 338858137.0, + "step": 8885 + }, + { + "epoch": 1.1303905355552728, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7879182696342468, + "num_tokens": 338893653.0, + "step": 8886 + }, + { + "epoch": 1.1305177458338633, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.779194712638855, + "num_tokens": 338931230.0, + "step": 8887 + }, + { + "epoch": 1.1306449561124539, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7941781282424927, + "num_tokens": 338966921.0, + "step": 8888 + }, + { + "epoch": 1.1307721663910444, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.773736834526062, + "num_tokens": 339000957.0, + "step": 8889 + }, + { + "epoch": 1.130899376669635, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7939926981925964, + "num_tokens": 339035876.0, + "step": 8890 + }, + { + "epoch": 1.1310265869482254, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.800379753112793, + "num_tokens": 339073700.0, + "step": 8891 + }, + { + "epoch": 1.131153797226816, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7998441457748413, + "num_tokens": 339108222.0, + "step": 8892 + }, + { + "epoch": 1.1312810075054065, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.812048077583313, + "num_tokens": 339145749.0, + "step": 8893 + }, + { + "epoch": 1.131408217783997, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7839343547821045, + "num_tokens": 339183852.0, + "step": 8894 + }, + { + "epoch": 1.1315354280625876, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.8136802911758423, + "num_tokens": 339218766.0, + "step": 8895 + }, + { + "epoch": 1.131662638341178, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.797957181930542, + "num_tokens": 339259932.0, + "step": 8896 + }, + { + "epoch": 1.1317898486197684, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.803577184677124, + "num_tokens": 339302201.0, + "step": 8897 + }, + { + "epoch": 1.131917058898359, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7689133882522583, + "num_tokens": 339342091.0, + "step": 8898 + }, + { + "epoch": 1.1320442691769494, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7947725057601929, + "num_tokens": 339379951.0, + "step": 8899 + }, + { + "epoch": 1.13217147945554, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7899537682533264, + "num_tokens": 339422243.0, + "step": 8900 + }, + { + "epoch": 1.1322986897341305, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8109533190727234, + "num_tokens": 339464969.0, + "step": 8901 + }, + { + "epoch": 1.132425900012721, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8017067909240723, + "num_tokens": 339502382.0, + "step": 8902 + }, + { + "epoch": 1.1325531102913116, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.783645510673523, + "num_tokens": 339542875.0, + "step": 8903 + }, + { + "epoch": 1.132680320569902, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7819314002990723, + "num_tokens": 339580776.0, + "step": 8904 + }, + { + "epoch": 1.1328075308484926, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.8191821575164795, + "num_tokens": 339620195.0, + "step": 8905 + }, + { + "epoch": 1.1329347411270831, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8096817135810852, + "num_tokens": 339659234.0, + "step": 8906 + }, + { + "epoch": 1.1330619514056737, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7923182845115662, + "num_tokens": 339703259.0, + "step": 8907 + }, + { + "epoch": 1.133189161684264, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8057883977890015, + "num_tokens": 339745901.0, + "step": 8908 + }, + { + "epoch": 1.1333163719628545, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7905133962631226, + "num_tokens": 339790417.0, + "step": 8909 + }, + { + "epoch": 1.133443582241445, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7956007719039917, + "num_tokens": 339827224.0, + "step": 8910 + }, + { + "epoch": 1.1335707925200356, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7875246405601501, + "num_tokens": 339859932.0, + "step": 8911 + }, + { + "epoch": 1.133698002798626, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7928909063339233, + "num_tokens": 339902416.0, + "step": 8912 + }, + { + "epoch": 1.1338252130772166, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8062537312507629, + "num_tokens": 339940495.0, + "step": 8913 + }, + { + "epoch": 1.1339524233558071, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7807462215423584, + "num_tokens": 339983010.0, + "step": 8914 + }, + { + "epoch": 1.1340796336343977, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7970981001853943, + "num_tokens": 340020683.0, + "step": 8915 + }, + { + "epoch": 1.1342068439129882, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7900049090385437, + "num_tokens": 340063179.0, + "step": 8916 + }, + { + "epoch": 1.1343340541915787, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8024390339851379, + "num_tokens": 340100166.0, + "step": 8917 + }, + { + "epoch": 1.1344612644701693, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7988314628601074, + "num_tokens": 340144325.0, + "step": 8918 + }, + { + "epoch": 1.1345884747487598, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7973560094833374, + "num_tokens": 340180563.0, + "step": 8919 + }, + { + "epoch": 1.1347156850273503, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7861513495445251, + "num_tokens": 340224621.0, + "step": 8920 + }, + { + "epoch": 1.1348428953059406, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8081876635551453, + "num_tokens": 340264810.0, + "step": 8921 + }, + { + "epoch": 1.1349701055845312, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8109999895095825, + "num_tokens": 340301158.0, + "step": 8922 + }, + { + "epoch": 1.1350973158631217, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.8017563223838806, + "num_tokens": 340339806.0, + "step": 8923 + }, + { + "epoch": 1.1352245261417122, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8043357133865356, + "num_tokens": 340386412.0, + "step": 8924 + }, + { + "epoch": 1.1353517364203027, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8086906671524048, + "num_tokens": 340431391.0, + "step": 8925 + }, + { + "epoch": 1.1354789466988933, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8052657246589661, + "num_tokens": 340465904.0, + "step": 8926 + }, + { + "epoch": 1.1356061569774838, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8041678667068481, + "num_tokens": 340504290.0, + "step": 8927 + }, + { + "epoch": 1.1357333672560743, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7927284240722656, + "num_tokens": 340539100.0, + "step": 8928 + }, + { + "epoch": 1.1358605775346649, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8103101253509521, + "num_tokens": 340569005.0, + "step": 8929 + }, + { + "epoch": 1.1359877878132554, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7954921722412109, + "num_tokens": 340613476.0, + "step": 8930 + }, + { + "epoch": 1.136114998091846, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7822096347808838, + "num_tokens": 340647982.0, + "step": 8931 + }, + { + "epoch": 1.1362422083704362, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.773168683052063, + "num_tokens": 340683270.0, + "step": 8932 + }, + { + "epoch": 1.1363694186490267, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7996646761894226, + "num_tokens": 340719955.0, + "step": 8933 + }, + { + "epoch": 1.1364966289276173, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7942017912864685, + "num_tokens": 340753934.0, + "step": 8934 + }, + { + "epoch": 1.1366238392062078, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8058366179466248, + "num_tokens": 340787545.0, + "step": 8935 + }, + { + "epoch": 1.1367510494847983, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8123852014541626, + "num_tokens": 340826362.0, + "step": 8936 + }, + { + "epoch": 1.1368782597633889, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7975879311561584, + "num_tokens": 340857871.0, + "step": 8937 + }, + { + "epoch": 1.1370054700419794, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8046681880950928, + "num_tokens": 340897736.0, + "step": 8938 + }, + { + "epoch": 1.13713268032057, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7927041053771973, + "num_tokens": 340935774.0, + "step": 8939 + }, + { + "epoch": 1.1372598905991604, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7938369512557983, + "num_tokens": 340971497.0, + "step": 8940 + }, + { + "epoch": 1.137387100877751, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7991917133331299, + "num_tokens": 341016209.0, + "step": 8941 + }, + { + "epoch": 1.1375143111563415, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7880870699882507, + "num_tokens": 341052137.0, + "step": 8942 + }, + { + "epoch": 1.137641521434932, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8002351522445679, + "num_tokens": 341083417.0, + "step": 8943 + }, + { + "epoch": 1.1377687317135226, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8049777746200562, + "num_tokens": 341127798.0, + "step": 8944 + }, + { + "epoch": 1.137895941992113, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7814881801605225, + "num_tokens": 341173485.0, + "step": 8945 + }, + { + "epoch": 1.1380231522707034, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7862024307250977, + "num_tokens": 341212834.0, + "step": 8946 + }, + { + "epoch": 1.138150362549294, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.787607729434967, + "num_tokens": 341246497.0, + "step": 8947 + }, + { + "epoch": 1.1382775728278844, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7836350202560425, + "num_tokens": 341291498.0, + "step": 8948 + }, + { + "epoch": 1.138404783106475, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7878649234771729, + "num_tokens": 341328104.0, + "step": 8949 + }, + { + "epoch": 1.1385319933850655, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8082818984985352, + "num_tokens": 341359042.0, + "step": 8950 + }, + { + "epoch": 1.138659203663656, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7954302430152893, + "num_tokens": 341403529.0, + "step": 8951 + }, + { + "epoch": 1.1387864139422466, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8048677444458008, + "num_tokens": 341442509.0, + "step": 8952 + }, + { + "epoch": 1.138913624220837, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.796758234500885, + "num_tokens": 341483015.0, + "step": 8953 + }, + { + "epoch": 1.1390408344994276, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.797324538230896, + "num_tokens": 341519676.0, + "step": 8954 + }, + { + "epoch": 1.1391680447780181, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8019505739212036, + "num_tokens": 341563483.0, + "step": 8955 + }, + { + "epoch": 1.1392952550566087, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7845680117607117, + "num_tokens": 341598387.0, + "step": 8956 + }, + { + "epoch": 1.139422465335199, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8052645325660706, + "num_tokens": 341640253.0, + "step": 8957 + }, + { + "epoch": 1.1395496756137895, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.5815, + "mean_token_accuracy": 0.8263057470321655, + "num_tokens": 341675788.0, + "step": 8958 + }, + { + "epoch": 1.13967688589238, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7985868453979492, + "num_tokens": 341716917.0, + "step": 8959 + }, + { + "epoch": 1.1398040961709706, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.799108624458313, + "num_tokens": 341758936.0, + "step": 8960 + }, + { + "epoch": 1.139931306449561, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7699056267738342, + "num_tokens": 341797612.0, + "step": 8961 + }, + { + "epoch": 1.1400585167281516, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7951821088790894, + "num_tokens": 341836061.0, + "step": 8962 + }, + { + "epoch": 1.1401857270067421, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7923577427864075, + "num_tokens": 341875465.0, + "step": 8963 + }, + { + "epoch": 1.1403129372853327, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.796032190322876, + "num_tokens": 341919795.0, + "step": 8964 + }, + { + "epoch": 1.1404401475639232, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7919778823852539, + "num_tokens": 341953291.0, + "step": 8965 + }, + { + "epoch": 1.1405673578425137, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8147746920585632, + "num_tokens": 341988364.0, + "step": 8966 + }, + { + "epoch": 1.1406945681211043, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.8033074140548706, + "num_tokens": 342028478.0, + "step": 8967 + }, + { + "epoch": 1.1408217783996948, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8039041757583618, + "num_tokens": 342067348.0, + "step": 8968 + }, + { + "epoch": 1.1409489886782853, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7909865975379944, + "num_tokens": 342105845.0, + "step": 8969 + }, + { + "epoch": 1.1410761989568756, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7965173125267029, + "num_tokens": 342142072.0, + "step": 8970 + }, + { + "epoch": 1.1412034092354661, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8004870414733887, + "num_tokens": 342176592.0, + "step": 8971 + }, + { + "epoch": 1.1413306195140567, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8041427135467529, + "num_tokens": 342211896.0, + "step": 8972 + }, + { + "epoch": 1.1414578297926472, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8032923340797424, + "num_tokens": 342256078.0, + "step": 8973 + }, + { + "epoch": 1.1415850400712377, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8034244179725647, + "num_tokens": 342296766.0, + "step": 8974 + }, + { + "epoch": 1.1417122503498283, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.79132080078125, + "num_tokens": 342333689.0, + "step": 8975 + }, + { + "epoch": 1.1418394606284188, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8149135708808899, + "num_tokens": 342369605.0, + "step": 8976 + }, + { + "epoch": 1.1419666709070093, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8079830408096313, + "num_tokens": 342410806.0, + "step": 8977 + }, + { + "epoch": 1.1420938811855998, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.5839, + "mean_token_accuracy": 0.8240787386894226, + "num_tokens": 342446957.0, + "step": 8978 + }, + { + "epoch": 1.1422210914641904, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7841486930847168, + "num_tokens": 342480717.0, + "step": 8979 + }, + { + "epoch": 1.142348301742781, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8058716058731079, + "num_tokens": 342520431.0, + "step": 8980 + }, + { + "epoch": 1.1424755120213712, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7950776815414429, + "num_tokens": 342561735.0, + "step": 8981 + }, + { + "epoch": 1.1426027222999617, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.813038170337677, + "num_tokens": 342593899.0, + "step": 8982 + }, + { + "epoch": 1.1427299325785523, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7989641427993774, + "num_tokens": 342628044.0, + "step": 8983 + }, + { + "epoch": 1.1428571428571428, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7914265394210815, + "num_tokens": 342665018.0, + "step": 8984 + }, + { + "epoch": 1.1429843531357333, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7857692241668701, + "num_tokens": 342704832.0, + "step": 8985 + }, + { + "epoch": 1.1431115634143239, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8047285079956055, + "num_tokens": 342741598.0, + "step": 8986 + }, + { + "epoch": 1.1432387736929144, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7943075299263, + "num_tokens": 342777302.0, + "step": 8987 + }, + { + "epoch": 1.143365983971505, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7740879058837891, + "num_tokens": 342809578.0, + "step": 8988 + }, + { + "epoch": 1.1434931942500954, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8003664612770081, + "num_tokens": 342845809.0, + "step": 8989 + }, + { + "epoch": 1.143620404528686, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.785488486289978, + "num_tokens": 342886118.0, + "step": 8990 + }, + { + "epoch": 1.1437476148072765, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7887097001075745, + "num_tokens": 342922517.0, + "step": 8991 + }, + { + "epoch": 1.143874825085867, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7877241373062134, + "num_tokens": 342959741.0, + "step": 8992 + }, + { + "epoch": 1.1440020353644575, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8140762448310852, + "num_tokens": 342991418.0, + "step": 8993 + }, + { + "epoch": 1.144129245643048, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8145982623100281, + "num_tokens": 343024263.0, + "step": 8994 + }, + { + "epoch": 1.1442564559216384, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.8009657859802246, + "num_tokens": 343066388.0, + "step": 8995 + }, + { + "epoch": 1.144383666200229, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8000443577766418, + "num_tokens": 343107631.0, + "step": 8996 + }, + { + "epoch": 1.1445108764788194, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8067549467086792, + "num_tokens": 343143645.0, + "step": 8997 + }, + { + "epoch": 1.14463808675741, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7891382575035095, + "num_tokens": 343182076.0, + "step": 8998 + }, + { + "epoch": 1.1447652970360005, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.817229688167572, + "num_tokens": 343222808.0, + "step": 8999 + }, + { + "epoch": 1.144892507314591, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7951806783676147, + "num_tokens": 343268229.0, + "step": 9000 + }, + { + "epoch": 1.1450197175931816, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7901252508163452, + "num_tokens": 343304459.0, + "step": 9001 + }, + { + "epoch": 1.145146927871772, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7550997138023376, + "num_tokens": 343343211.0, + "step": 9002 + }, + { + "epoch": 1.1452741381503626, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7941503524780273, + "num_tokens": 343388498.0, + "step": 9003 + }, + { + "epoch": 1.1454013484289531, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.799514889717102, + "num_tokens": 343430101.0, + "step": 9004 + }, + { + "epoch": 1.1455285587075437, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7869799733161926, + "num_tokens": 343466454.0, + "step": 9005 + }, + { + "epoch": 1.145655768986134, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.812414288520813, + "num_tokens": 343506960.0, + "step": 9006 + }, + { + "epoch": 1.1457829792647245, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.8037382960319519, + "num_tokens": 343541133.0, + "step": 9007 + }, + { + "epoch": 1.145910189543315, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8044503927230835, + "num_tokens": 343580266.0, + "step": 9008 + }, + { + "epoch": 1.1460373998219056, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.790723443031311, + "num_tokens": 343611691.0, + "step": 9009 + }, + { + "epoch": 1.146164610100496, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8121060132980347, + "num_tokens": 343649355.0, + "step": 9010 + }, + { + "epoch": 1.1462918203790866, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8118556141853333, + "num_tokens": 343681606.0, + "step": 9011 + }, + { + "epoch": 1.1464190306576771, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8102123737335205, + "num_tokens": 343721592.0, + "step": 9012 + }, + { + "epoch": 1.1465462409362677, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.8004704117774963, + "num_tokens": 343764084.0, + "step": 9013 + }, + { + "epoch": 1.1466734512148582, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.803099513053894, + "num_tokens": 343801213.0, + "step": 9014 + }, + { + "epoch": 1.1468006614934487, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7995309829711914, + "num_tokens": 343839761.0, + "step": 9015 + }, + { + "epoch": 1.1469278717720393, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7997559309005737, + "num_tokens": 343877792.0, + "step": 9016 + }, + { + "epoch": 1.1470550820506298, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8033369779586792, + "num_tokens": 343915107.0, + "step": 9017 + }, + { + "epoch": 1.1471822923292203, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7977910041809082, + "num_tokens": 343949783.0, + "step": 9018 + }, + { + "epoch": 1.1473095026078106, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7755695581436157, + "num_tokens": 343991058.0, + "step": 9019 + }, + { + "epoch": 1.1474367128864011, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7988476753234863, + "num_tokens": 344030424.0, + "step": 9020 + }, + { + "epoch": 1.1475639231649917, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7966139316558838, + "num_tokens": 344071385.0, + "step": 9021 + }, + { + "epoch": 1.1476911334435822, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8099335432052612, + "num_tokens": 344109725.0, + "step": 9022 + }, + { + "epoch": 1.1478183437221727, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8020585179328918, + "num_tokens": 344143947.0, + "step": 9023 + }, + { + "epoch": 1.1479455540007633, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7832748889923096, + "num_tokens": 344179969.0, + "step": 9024 + }, + { + "epoch": 1.1480727642793538, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8072113990783691, + "num_tokens": 344212279.0, + "step": 9025 + }, + { + "epoch": 1.1481999745579443, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8039747476577759, + "num_tokens": 344248402.0, + "step": 9026 + }, + { + "epoch": 1.1483271848365348, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7926962971687317, + "num_tokens": 344286557.0, + "step": 9027 + }, + { + "epoch": 1.1484543951151254, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8010186553001404, + "num_tokens": 344331014.0, + "step": 9028 + }, + { + "epoch": 1.148581605393716, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7943096160888672, + "num_tokens": 344373239.0, + "step": 9029 + }, + { + "epoch": 1.1487088156723062, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.8100067377090454, + "num_tokens": 344409871.0, + "step": 9030 + }, + { + "epoch": 1.1488360259508967, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7931421399116516, + "num_tokens": 344452108.0, + "step": 9031 + }, + { + "epoch": 1.1489632362294873, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7972812652587891, + "num_tokens": 344489798.0, + "step": 9032 + }, + { + "epoch": 1.1490904465080778, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8059443235397339, + "num_tokens": 344528194.0, + "step": 9033 + }, + { + "epoch": 1.1492176567866683, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7775739431381226, + "num_tokens": 344559699.0, + "step": 9034 + }, + { + "epoch": 1.1493448670652588, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7912437319755554, + "num_tokens": 344598234.0, + "step": 9035 + }, + { + "epoch": 1.1494720773438494, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7903093099594116, + "num_tokens": 344634780.0, + "step": 9036 + }, + { + "epoch": 1.14959928762244, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8038953542709351, + "num_tokens": 344677167.0, + "step": 9037 + }, + { + "epoch": 1.1497264979010304, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7845648527145386, + "num_tokens": 344713824.0, + "step": 9038 + }, + { + "epoch": 1.149853708179621, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7818703055381775, + "num_tokens": 344755814.0, + "step": 9039 + }, + { + "epoch": 1.1499809184582115, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8163264989852905, + "num_tokens": 344795022.0, + "step": 9040 + }, + { + "epoch": 1.150108128736802, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.808682918548584, + "num_tokens": 344833906.0, + "step": 9041 + }, + { + "epoch": 1.1502353390153925, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8026790618896484, + "num_tokens": 344875975.0, + "step": 9042 + }, + { + "epoch": 1.150362549293983, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8021799921989441, + "num_tokens": 344915433.0, + "step": 9043 + }, + { + "epoch": 1.1504897595725734, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8028947114944458, + "num_tokens": 344949038.0, + "step": 9044 + }, + { + "epoch": 1.150616969851164, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7961224913597107, + "num_tokens": 344988260.0, + "step": 9045 + }, + { + "epoch": 1.1507441801297544, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8180291652679443, + "num_tokens": 345024355.0, + "step": 9046 + }, + { + "epoch": 1.150871390408345, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8077030777931213, + "num_tokens": 345063041.0, + "step": 9047 + }, + { + "epoch": 1.1509986006869355, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.8198409676551819, + "num_tokens": 345097813.0, + "step": 9048 + }, + { + "epoch": 1.151125810965526, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7991710901260376, + "num_tokens": 345134534.0, + "step": 9049 + }, + { + "epoch": 1.1512530212441165, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8074468970298767, + "num_tokens": 345171018.0, + "step": 9050 + }, + { + "epoch": 1.151380231522707, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8091698884963989, + "num_tokens": 345208006.0, + "step": 9051 + }, + { + "epoch": 1.1515074418012976, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8159542083740234, + "num_tokens": 345246616.0, + "step": 9052 + }, + { + "epoch": 1.1516346520798881, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8168583512306213, + "num_tokens": 345288095.0, + "step": 9053 + }, + { + "epoch": 1.1517618623584787, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7888598442077637, + "num_tokens": 345333291.0, + "step": 9054 + }, + { + "epoch": 1.151889072637069, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7996920943260193, + "num_tokens": 345375093.0, + "step": 9055 + }, + { + "epoch": 1.1520162829156595, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7972384691238403, + "num_tokens": 345415547.0, + "step": 9056 + }, + { + "epoch": 1.15214349319425, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8052436709403992, + "num_tokens": 345457340.0, + "step": 9057 + }, + { + "epoch": 1.1522707034728406, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7959995269775391, + "num_tokens": 345494711.0, + "step": 9058 + }, + { + "epoch": 1.152397913751431, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7877996563911438, + "num_tokens": 345534249.0, + "step": 9059 + }, + { + "epoch": 1.1525251240300216, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.799679160118103, + "num_tokens": 345570832.0, + "step": 9060 + }, + { + "epoch": 1.1526523343086121, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7950977683067322, + "num_tokens": 345609007.0, + "step": 9061 + }, + { + "epoch": 1.1527795445872027, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8128647804260254, + "num_tokens": 345647979.0, + "step": 9062 + }, + { + "epoch": 1.1529067548657932, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7900421619415283, + "num_tokens": 345681821.0, + "step": 9063 + }, + { + "epoch": 1.1530339651443837, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7893004417419434, + "num_tokens": 345724274.0, + "step": 9064 + }, + { + "epoch": 1.1531611754229742, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.5927, + "mean_token_accuracy": 0.8202905654907227, + "num_tokens": 345763558.0, + "step": 9065 + }, + { + "epoch": 1.1532883857015648, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7917778491973877, + "num_tokens": 345798486.0, + "step": 9066 + }, + { + "epoch": 1.1534155959801553, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7932507991790771, + "num_tokens": 345839379.0, + "step": 9067 + }, + { + "epoch": 1.1535428062587456, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7935061454772949, + "num_tokens": 345876675.0, + "step": 9068 + }, + { + "epoch": 1.1536700165373361, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6008, + "mean_token_accuracy": 0.8179640769958496, + "num_tokens": 345913482.0, + "step": 9069 + }, + { + "epoch": 1.1537972268159267, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7919149398803711, + "num_tokens": 345947671.0, + "step": 9070 + }, + { + "epoch": 1.1539244370945172, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8016788363456726, + "num_tokens": 345981520.0, + "step": 9071 + }, + { + "epoch": 1.1540516473731077, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.777331531047821, + "num_tokens": 346020050.0, + "step": 9072 + }, + { + "epoch": 1.1541788576516983, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7789955139160156, + "num_tokens": 346050000.0, + "step": 9073 + }, + { + "epoch": 1.1543060679302888, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7955111265182495, + "num_tokens": 346085038.0, + "step": 9074 + }, + { + "epoch": 1.1544332782088793, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8101121187210083, + "num_tokens": 346125871.0, + "step": 9075 + }, + { + "epoch": 1.1545604884874698, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8045326471328735, + "num_tokens": 346160460.0, + "step": 9076 + }, + { + "epoch": 1.1546876987660604, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7898910045623779, + "num_tokens": 346201621.0, + "step": 9077 + }, + { + "epoch": 1.154814909044651, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8026407957077026, + "num_tokens": 346242680.0, + "step": 9078 + }, + { + "epoch": 1.1549421193232412, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7909960150718689, + "num_tokens": 346275774.0, + "step": 9079 + }, + { + "epoch": 1.1550693296018317, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8015140295028687, + "num_tokens": 346317648.0, + "step": 9080 + }, + { + "epoch": 1.1551965398804223, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7988995909690857, + "num_tokens": 346357736.0, + "step": 9081 + }, + { + "epoch": 1.1553237501590128, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7958300113677979, + "num_tokens": 346397961.0, + "step": 9082 + }, + { + "epoch": 1.1554509604376033, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.810662031173706, + "num_tokens": 346438268.0, + "step": 9083 + }, + { + "epoch": 1.1555781707161938, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7842148542404175, + "num_tokens": 346479024.0, + "step": 9084 + }, + { + "epoch": 1.1557053809947844, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8061142563819885, + "num_tokens": 346523526.0, + "step": 9085 + }, + { + "epoch": 1.155832591273375, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.808086097240448, + "num_tokens": 346561985.0, + "step": 9086 + }, + { + "epoch": 1.1559598015519654, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7875845432281494, + "num_tokens": 346605600.0, + "step": 9087 + }, + { + "epoch": 1.156087011830556, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8145260214805603, + "num_tokens": 346640798.0, + "step": 9088 + }, + { + "epoch": 1.1562142221091465, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7923339605331421, + "num_tokens": 346678483.0, + "step": 9089 + }, + { + "epoch": 1.156341432387737, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8083943724632263, + "num_tokens": 346714604.0, + "step": 9090 + }, + { + "epoch": 1.1564686426663275, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.8101183176040649, + "num_tokens": 346747203.0, + "step": 9091 + }, + { + "epoch": 1.156595852944918, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.801500678062439, + "num_tokens": 346790650.0, + "step": 9092 + }, + { + "epoch": 1.1567230632235084, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8175468444824219, + "num_tokens": 346822872.0, + "step": 9093 + }, + { + "epoch": 1.156850273502099, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8105465173721313, + "num_tokens": 346857999.0, + "step": 9094 + }, + { + "epoch": 1.1569774837806894, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.806645929813385, + "num_tokens": 346891804.0, + "step": 9095 + }, + { + "epoch": 1.15710469405928, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7957663536071777, + "num_tokens": 346931554.0, + "step": 9096 + }, + { + "epoch": 1.1572319043378705, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7878000736236572, + "num_tokens": 346965214.0, + "step": 9097 + }, + { + "epoch": 1.157359114616461, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8107210993766785, + "num_tokens": 347001219.0, + "step": 9098 + }, + { + "epoch": 1.1574863248950515, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7778167724609375, + "num_tokens": 347043212.0, + "step": 9099 + }, + { + "epoch": 1.157613535173642, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8143858313560486, + "num_tokens": 347082249.0, + "step": 9100 + }, + { + "epoch": 1.1577407454522326, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.797217071056366, + "num_tokens": 347119286.0, + "step": 9101 + }, + { + "epoch": 1.1578679557308231, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8125565052032471, + "num_tokens": 347155549.0, + "step": 9102 + }, + { + "epoch": 1.1579951660094137, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8004310131072998, + "num_tokens": 347196846.0, + "step": 9103 + }, + { + "epoch": 1.158122376288004, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.81357342004776, + "num_tokens": 347234028.0, + "step": 9104 + }, + { + "epoch": 1.1582495865665945, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8127874135971069, + "num_tokens": 347276931.0, + "step": 9105 + }, + { + "epoch": 1.158376796845185, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7971364855766296, + "num_tokens": 347317139.0, + "step": 9106 + }, + { + "epoch": 1.1585040071237755, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8020836114883423, + "num_tokens": 347355718.0, + "step": 9107 + }, + { + "epoch": 1.158631217402366, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7885217666625977, + "num_tokens": 347393403.0, + "step": 9108 + }, + { + "epoch": 1.1587584276809566, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8049652576446533, + "num_tokens": 347433736.0, + "step": 9109 + }, + { + "epoch": 1.1588856379595471, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7985272407531738, + "num_tokens": 347474574.0, + "step": 9110 + }, + { + "epoch": 1.1590128482381377, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8049404621124268, + "num_tokens": 347518785.0, + "step": 9111 + }, + { + "epoch": 1.1591400585167282, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.784369945526123, + "num_tokens": 347557518.0, + "step": 9112 + }, + { + "epoch": 1.1592672687953187, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7727036476135254, + "num_tokens": 347600094.0, + "step": 9113 + }, + { + "epoch": 1.1593944790739092, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7987852096557617, + "num_tokens": 347633709.0, + "step": 9114 + }, + { + "epoch": 1.1595216893524998, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.805137574672699, + "num_tokens": 347670027.0, + "step": 9115 + }, + { + "epoch": 1.1596488996310903, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8044748306274414, + "num_tokens": 347707488.0, + "step": 9116 + }, + { + "epoch": 1.1597761099096806, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7979689836502075, + "num_tokens": 347746030.0, + "step": 9117 + }, + { + "epoch": 1.1599033201882711, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7900781035423279, + "num_tokens": 347783409.0, + "step": 9118 + }, + { + "epoch": 1.1600305304668617, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8109988570213318, + "num_tokens": 347821335.0, + "step": 9119 + }, + { + "epoch": 1.1601577407454522, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8099771738052368, + "num_tokens": 347857590.0, + "step": 9120 + }, + { + "epoch": 1.1602849510240427, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.811751663684845, + "num_tokens": 347897000.0, + "step": 9121 + }, + { + "epoch": 1.1604121613026332, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7958022356033325, + "num_tokens": 347928947.0, + "step": 9122 + }, + { + "epoch": 1.1605393715812238, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7617999315261841, + "num_tokens": 347959294.0, + "step": 9123 + }, + { + "epoch": 1.1606665818598143, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.803653359413147, + "num_tokens": 347993103.0, + "step": 9124 + }, + { + "epoch": 1.1607937921384048, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8128339052200317, + "num_tokens": 348028412.0, + "step": 9125 + }, + { + "epoch": 1.1609210024169954, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7868909239768982, + "num_tokens": 348071429.0, + "step": 9126 + }, + { + "epoch": 1.161048212695586, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8106534481048584, + "num_tokens": 348113222.0, + "step": 9127 + }, + { + "epoch": 1.1611754229741762, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.800697922706604, + "num_tokens": 348150188.0, + "step": 9128 + }, + { + "epoch": 1.1613026332527667, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7986558079719543, + "num_tokens": 348188547.0, + "step": 9129 + }, + { + "epoch": 1.1614298435313573, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7985963821411133, + "num_tokens": 348228624.0, + "step": 9130 + }, + { + "epoch": 1.1615570538099478, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7972471714019775, + "num_tokens": 348264932.0, + "step": 9131 + }, + { + "epoch": 1.1616842640885383, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8002699613571167, + "num_tokens": 348307680.0, + "step": 9132 + }, + { + "epoch": 1.1618114743671288, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8146659731864929, + "num_tokens": 348346449.0, + "step": 9133 + }, + { + "epoch": 1.1619386846457194, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.810992956161499, + "num_tokens": 348386393.0, + "step": 9134 + }, + { + "epoch": 1.16206589492431, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7950843572616577, + "num_tokens": 348429458.0, + "step": 9135 + }, + { + "epoch": 1.1621931052029004, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7854577898979187, + "num_tokens": 348471365.0, + "step": 9136 + }, + { + "epoch": 1.162320315481491, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7857729196548462, + "num_tokens": 348513425.0, + "step": 9137 + }, + { + "epoch": 1.1624475257600815, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7983705997467041, + "num_tokens": 348551181.0, + "step": 9138 + }, + { + "epoch": 1.162574736038672, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.5801, + "mean_token_accuracy": 0.8212428689002991, + "num_tokens": 348583611.0, + "step": 9139 + }, + { + "epoch": 1.1627019463172625, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7978561520576477, + "num_tokens": 348624281.0, + "step": 9140 + }, + { + "epoch": 1.162829156595853, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8000379204750061, + "num_tokens": 348661416.0, + "step": 9141 + }, + { + "epoch": 1.1629563668744434, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7807036638259888, + "num_tokens": 348703907.0, + "step": 9142 + }, + { + "epoch": 1.163083577153034, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8117523789405823, + "num_tokens": 348735617.0, + "step": 9143 + }, + { + "epoch": 1.1632107874316244, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.8007786870002747, + "num_tokens": 348773552.0, + "step": 9144 + }, + { + "epoch": 1.163337997710215, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.804390549659729, + "num_tokens": 348811274.0, + "step": 9145 + }, + { + "epoch": 1.1634652079888055, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7893300652503967, + "num_tokens": 348847773.0, + "step": 9146 + }, + { + "epoch": 1.163592418267396, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7928840517997742, + "num_tokens": 348886356.0, + "step": 9147 + }, + { + "epoch": 1.1637196285459865, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8105548024177551, + "num_tokens": 348922892.0, + "step": 9148 + }, + { + "epoch": 1.163846838824577, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8137531280517578, + "num_tokens": 348954924.0, + "step": 9149 + }, + { + "epoch": 1.1639740491031676, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.8012608885765076, + "num_tokens": 348997335.0, + "step": 9150 + }, + { + "epoch": 1.1641012593817581, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7849863767623901, + "num_tokens": 349038221.0, + "step": 9151 + }, + { + "epoch": 1.1642284696603487, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7933574914932251, + "num_tokens": 349076559.0, + "step": 9152 + }, + { + "epoch": 1.164355679938939, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7808691263198853, + "num_tokens": 349112264.0, + "step": 9153 + }, + { + "epoch": 1.1644828902175295, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8084385991096497, + "num_tokens": 349146571.0, + "step": 9154 + }, + { + "epoch": 1.16461010049612, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8046635389328003, + "num_tokens": 349183674.0, + "step": 9155 + }, + { + "epoch": 1.1647373107747105, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8081289529800415, + "num_tokens": 349227042.0, + "step": 9156 + }, + { + "epoch": 1.164864521053301, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7984325289726257, + "num_tokens": 349264072.0, + "step": 9157 + }, + { + "epoch": 1.1649917313318916, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7971392869949341, + "num_tokens": 349304427.0, + "step": 9158 + }, + { + "epoch": 1.1651189416104821, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.802420437335968, + "num_tokens": 349346708.0, + "step": 9159 + }, + { + "epoch": 1.1652461518890727, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7968525886535645, + "num_tokens": 349387716.0, + "step": 9160 + }, + { + "epoch": 1.1653733621676632, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.796563982963562, + "num_tokens": 349425757.0, + "step": 9161 + }, + { + "epoch": 1.1655005724462537, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8117519617080688, + "num_tokens": 349472097.0, + "step": 9162 + }, + { + "epoch": 1.1656277827248442, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7995045185089111, + "num_tokens": 349513809.0, + "step": 9163 + }, + { + "epoch": 1.1657549930034348, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8089115619659424, + "num_tokens": 349544906.0, + "step": 9164 + }, + { + "epoch": 1.1658822032820253, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.791050136089325, + "num_tokens": 349582529.0, + "step": 9165 + }, + { + "epoch": 1.1660094135606156, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7976325750350952, + "num_tokens": 349624871.0, + "step": 9166 + }, + { + "epoch": 1.1661366238392061, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7553524971008301, + "num_tokens": 349660783.0, + "step": 9167 + }, + { + "epoch": 1.1662638341177967, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8149917125701904, + "num_tokens": 349698430.0, + "step": 9168 + }, + { + "epoch": 1.1663910443963872, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7910149097442627, + "num_tokens": 349734981.0, + "step": 9169 + }, + { + "epoch": 1.1665182546749777, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7930140495300293, + "num_tokens": 349774267.0, + "step": 9170 + }, + { + "epoch": 1.1666454649535682, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7993505001068115, + "num_tokens": 349811400.0, + "step": 9171 + }, + { + "epoch": 1.1667726752321588, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7926959991455078, + "num_tokens": 349850031.0, + "step": 9172 + }, + { + "epoch": 1.1668998855107493, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8066260814666748, + "num_tokens": 349889967.0, + "step": 9173 + }, + { + "epoch": 1.1670270957893398, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7952975034713745, + "num_tokens": 349930276.0, + "step": 9174 + }, + { + "epoch": 1.1671543060679304, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7914048433303833, + "num_tokens": 349969889.0, + "step": 9175 + }, + { + "epoch": 1.1672815163465209, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.8026682734489441, + "num_tokens": 350005729.0, + "step": 9176 + }, + { + "epoch": 1.1674087266251112, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8028665781021118, + "num_tokens": 350043130.0, + "step": 9177 + }, + { + "epoch": 1.1675359369037017, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8192058801651001, + "num_tokens": 350081233.0, + "step": 9178 + }, + { + "epoch": 1.1676631471822922, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7865524888038635, + "num_tokens": 350119403.0, + "step": 9179 + }, + { + "epoch": 1.1677903574608828, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7855852842330933, + "num_tokens": 350159099.0, + "step": 9180 + }, + { + "epoch": 1.1679175677394733, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7983577847480774, + "num_tokens": 350195642.0, + "step": 9181 + }, + { + "epoch": 1.1680447780180638, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8078780174255371, + "num_tokens": 350236721.0, + "step": 9182 + }, + { + "epoch": 1.1681719882966544, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7860589623451233, + "num_tokens": 350272691.0, + "step": 9183 + }, + { + "epoch": 1.168299198575245, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7972170114517212, + "num_tokens": 350318299.0, + "step": 9184 + }, + { + "epoch": 1.1684264088538354, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.8015615344047546, + "num_tokens": 350360973.0, + "step": 9185 + }, + { + "epoch": 1.168553619132426, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8107867240905762, + "num_tokens": 350400516.0, + "step": 9186 + }, + { + "epoch": 1.1686808294110165, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7937997579574585, + "num_tokens": 350438147.0, + "step": 9187 + }, + { + "epoch": 1.168808039689607, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8029836416244507, + "num_tokens": 350481604.0, + "step": 9188 + }, + { + "epoch": 1.1689352499681975, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.8228452801704407, + "num_tokens": 350518225.0, + "step": 9189 + }, + { + "epoch": 1.169062460246788, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7993898987770081, + "num_tokens": 350553752.0, + "step": 9190 + }, + { + "epoch": 1.1691896705253784, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8066667318344116, + "num_tokens": 350593676.0, + "step": 9191 + }, + { + "epoch": 1.169316880803969, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7957375049591064, + "num_tokens": 350633618.0, + "step": 9192 + }, + { + "epoch": 1.1694440910825594, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8011112213134766, + "num_tokens": 350678221.0, + "step": 9193 + }, + { + "epoch": 1.16957130136115, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8089195489883423, + "num_tokens": 350718258.0, + "step": 9194 + }, + { + "epoch": 1.1696985116397405, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8146647214889526, + "num_tokens": 350757370.0, + "step": 9195 + }, + { + "epoch": 1.169825721918331, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8049173951148987, + "num_tokens": 350801394.0, + "step": 9196 + }, + { + "epoch": 1.1699529321969215, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7886347770690918, + "num_tokens": 350840635.0, + "step": 9197 + }, + { + "epoch": 1.170080142475512, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8138406276702881, + "num_tokens": 350879616.0, + "step": 9198 + }, + { + "epoch": 1.1702073527541026, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8036333918571472, + "num_tokens": 350921710.0, + "step": 9199 + }, + { + "epoch": 1.1703345630326931, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.784476101398468, + "num_tokens": 350964218.0, + "step": 9200 + }, + { + "epoch": 1.1704617733112836, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8042178750038147, + "num_tokens": 351004511.0, + "step": 9201 + }, + { + "epoch": 1.170588983589874, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7969970703125, + "num_tokens": 351044583.0, + "step": 9202 + }, + { + "epoch": 1.1707161938684645, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7829931378364563, + "num_tokens": 351083613.0, + "step": 9203 + }, + { + "epoch": 1.170843404147055, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7938974499702454, + "num_tokens": 351117412.0, + "step": 9204 + }, + { + "epoch": 1.1709706144256455, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7860996723175049, + "num_tokens": 351154779.0, + "step": 9205 + }, + { + "epoch": 1.171097824704236, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8113390207290649, + "num_tokens": 351191412.0, + "step": 9206 + }, + { + "epoch": 1.1712250349828266, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8029928207397461, + "num_tokens": 351235465.0, + "step": 9207 + }, + { + "epoch": 1.1713522452614171, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8050477504730225, + "num_tokens": 351279326.0, + "step": 9208 + }, + { + "epoch": 1.1714794555400077, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8155341148376465, + "num_tokens": 351316678.0, + "step": 9209 + }, + { + "epoch": 1.1716066658185982, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7977503538131714, + "num_tokens": 351352641.0, + "step": 9210 + }, + { + "epoch": 1.1717338760971887, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7866091728210449, + "num_tokens": 351385845.0, + "step": 9211 + }, + { + "epoch": 1.1718610863757792, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8034530878067017, + "num_tokens": 351425220.0, + "step": 9212 + }, + { + "epoch": 1.1719882966543698, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7983959913253784, + "num_tokens": 351467515.0, + "step": 9213 + }, + { + "epoch": 1.1721155069329603, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7950489521026611, + "num_tokens": 351507571.0, + "step": 9214 + }, + { + "epoch": 1.1722427172115506, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7968980073928833, + "num_tokens": 351546150.0, + "step": 9215 + }, + { + "epoch": 1.1723699274901411, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.791426420211792, + "num_tokens": 351588574.0, + "step": 9216 + }, + { + "epoch": 1.1724971377687317, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8009873032569885, + "num_tokens": 351626914.0, + "step": 9217 + }, + { + "epoch": 1.1726243480473222, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7856302857398987, + "num_tokens": 351663574.0, + "step": 9218 + }, + { + "epoch": 1.1727515583259127, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7791209816932678, + "num_tokens": 351703275.0, + "step": 9219 + }, + { + "epoch": 1.1728787686045032, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7909227609634399, + "num_tokens": 351745713.0, + "step": 9220 + }, + { + "epoch": 1.1730059788830938, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7862110137939453, + "num_tokens": 351780789.0, + "step": 9221 + }, + { + "epoch": 1.1731331891616843, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8115568161010742, + "num_tokens": 351820352.0, + "step": 9222 + }, + { + "epoch": 1.1732603994402748, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.8113340139389038, + "num_tokens": 351860011.0, + "step": 9223 + }, + { + "epoch": 1.1733876097188654, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7943851351737976, + "num_tokens": 351898742.0, + "step": 9224 + }, + { + "epoch": 1.1735148199974559, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.785260796546936, + "num_tokens": 351940201.0, + "step": 9225 + }, + { + "epoch": 1.1736420302760462, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.794015645980835, + "num_tokens": 351971041.0, + "step": 9226 + }, + { + "epoch": 1.1737692405546367, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7819767594337463, + "num_tokens": 352013002.0, + "step": 9227 + }, + { + "epoch": 1.1738964508332272, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7963563203811646, + "num_tokens": 352051435.0, + "step": 9228 + }, + { + "epoch": 1.1740236611118178, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8035566806793213, + "num_tokens": 352093295.0, + "step": 9229 + }, + { + "epoch": 1.1741508713904083, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7962567806243896, + "num_tokens": 352136401.0, + "step": 9230 + }, + { + "epoch": 1.1742780816689988, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8152527809143066, + "num_tokens": 352178696.0, + "step": 9231 + }, + { + "epoch": 1.1744052919475894, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7755197286605835, + "num_tokens": 352214759.0, + "step": 9232 + }, + { + "epoch": 1.1745325022261799, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8161112070083618, + "num_tokens": 352250520.0, + "step": 9233 + }, + { + "epoch": 1.1746597125047704, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7953886985778809, + "num_tokens": 352288773.0, + "step": 9234 + }, + { + "epoch": 1.174786922783361, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8061628341674805, + "num_tokens": 352326548.0, + "step": 9235 + }, + { + "epoch": 1.1749141330619515, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8148191571235657, + "num_tokens": 352366537.0, + "step": 9236 + }, + { + "epoch": 1.175041343340542, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7785928249359131, + "num_tokens": 352402390.0, + "step": 9237 + }, + { + "epoch": 1.1751685536191325, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7939159870147705, + "num_tokens": 352442357.0, + "step": 9238 + }, + { + "epoch": 1.175295763897723, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7918536067008972, + "num_tokens": 352476884.0, + "step": 9239 + }, + { + "epoch": 1.1754229741763134, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8000288009643555, + "num_tokens": 352513448.0, + "step": 9240 + }, + { + "epoch": 1.175550184454904, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.8205416202545166, + "num_tokens": 352548420.0, + "step": 9241 + }, + { + "epoch": 1.1756773947334944, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.8011852502822876, + "num_tokens": 352592157.0, + "step": 9242 + }, + { + "epoch": 1.175804605012085, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7868305444717407, + "num_tokens": 352626427.0, + "step": 9243 + }, + { + "epoch": 1.1759318152906755, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8131042122840881, + "num_tokens": 352667075.0, + "step": 9244 + }, + { + "epoch": 1.176059025569266, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7913715839385986, + "num_tokens": 352707114.0, + "step": 9245 + }, + { + "epoch": 1.1761862358478565, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8046473860740662, + "num_tokens": 352738064.0, + "step": 9246 + }, + { + "epoch": 1.176313446126447, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7799546718597412, + "num_tokens": 352779964.0, + "step": 9247 + }, + { + "epoch": 1.1764406564050376, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.808342456817627, + "num_tokens": 352819574.0, + "step": 9248 + }, + { + "epoch": 1.1765678666836281, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8090376853942871, + "num_tokens": 352859151.0, + "step": 9249 + }, + { + "epoch": 1.1766950769622184, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7794142365455627, + "num_tokens": 352901119.0, + "step": 9250 + }, + { + "epoch": 1.176822287240809, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7991755604743958, + "num_tokens": 352945878.0, + "step": 9251 + }, + { + "epoch": 1.1769494975193995, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.789442241191864, + "num_tokens": 352979819.0, + "step": 9252 + }, + { + "epoch": 1.17707670779799, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7877559661865234, + "num_tokens": 353023552.0, + "step": 9253 + }, + { + "epoch": 1.1772039180765805, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7967643141746521, + "num_tokens": 353058178.0, + "step": 9254 + }, + { + "epoch": 1.177331128355171, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8059298992156982, + "num_tokens": 353091428.0, + "step": 9255 + }, + { + "epoch": 1.1774583386337616, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.806649923324585, + "num_tokens": 353131943.0, + "step": 9256 + }, + { + "epoch": 1.1775855489123521, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7868841886520386, + "num_tokens": 353171213.0, + "step": 9257 + }, + { + "epoch": 1.1777127591909426, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8073102235794067, + "num_tokens": 353205057.0, + "step": 9258 + }, + { + "epoch": 1.1778399694695332, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7832706570625305, + "num_tokens": 353251164.0, + "step": 9259 + }, + { + "epoch": 1.1779671797481237, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8028926253318787, + "num_tokens": 353288202.0, + "step": 9260 + }, + { + "epoch": 1.1780943900267142, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7983264327049255, + "num_tokens": 353327396.0, + "step": 9261 + }, + { + "epoch": 1.1782216003053048, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7865676879882812, + "num_tokens": 353366512.0, + "step": 9262 + }, + { + "epoch": 1.1783488105838953, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7893682718276978, + "num_tokens": 353408546.0, + "step": 9263 + }, + { + "epoch": 1.1784760208624856, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8109331130981445, + "num_tokens": 353444950.0, + "step": 9264 + }, + { + "epoch": 1.1786032311410761, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8102273941040039, + "num_tokens": 353483306.0, + "step": 9265 + }, + { + "epoch": 1.1787304414196667, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7968397736549377, + "num_tokens": 353523038.0, + "step": 9266 + }, + { + "epoch": 1.1788576516982572, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8101299405097961, + "num_tokens": 353564266.0, + "step": 9267 + }, + { + "epoch": 1.1789848619768477, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8054006695747375, + "num_tokens": 353603558.0, + "step": 9268 + }, + { + "epoch": 1.1791120722554382, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8115748167037964, + "num_tokens": 353648097.0, + "step": 9269 + }, + { + "epoch": 1.1792392825340288, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7838760614395142, + "num_tokens": 353688970.0, + "step": 9270 + }, + { + "epoch": 1.1793664928126193, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7865222096443176, + "num_tokens": 353732505.0, + "step": 9271 + }, + { + "epoch": 1.1794937030912098, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8068827390670776, + "num_tokens": 353767460.0, + "step": 9272 + }, + { + "epoch": 1.1796209133698004, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8040889501571655, + "num_tokens": 353801951.0, + "step": 9273 + }, + { + "epoch": 1.1797481236483909, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8086414337158203, + "num_tokens": 353836815.0, + "step": 9274 + }, + { + "epoch": 1.1798753339269812, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7934516072273254, + "num_tokens": 353876296.0, + "step": 9275 + }, + { + "epoch": 1.1800025442055717, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.782863199710846, + "num_tokens": 353911792.0, + "step": 9276 + }, + { + "epoch": 1.1801297544841622, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7827945947647095, + "num_tokens": 353949532.0, + "step": 9277 + }, + { + "epoch": 1.1802569647627528, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.787203311920166, + "num_tokens": 353988622.0, + "step": 9278 + }, + { + "epoch": 1.1803841750413433, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8030083775520325, + "num_tokens": 354028136.0, + "step": 9279 + }, + { + "epoch": 1.1805113853199338, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8020866513252258, + "num_tokens": 354065029.0, + "step": 9280 + }, + { + "epoch": 1.1806385955985244, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.809378445148468, + "num_tokens": 354099279.0, + "step": 9281 + }, + { + "epoch": 1.1807658058771149, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7929574847221375, + "num_tokens": 354138870.0, + "step": 9282 + }, + { + "epoch": 1.1808930161557054, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7878431081771851, + "num_tokens": 354176498.0, + "step": 9283 + }, + { + "epoch": 1.181020226434296, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7799098491668701, + "num_tokens": 354210063.0, + "step": 9284 + }, + { + "epoch": 1.1811474367128865, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7960435152053833, + "num_tokens": 354248095.0, + "step": 9285 + }, + { + "epoch": 1.181274646991477, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7937468886375427, + "num_tokens": 354286897.0, + "step": 9286 + }, + { + "epoch": 1.1814018572700675, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8101720809936523, + "num_tokens": 354330685.0, + "step": 9287 + }, + { + "epoch": 1.181529067548658, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7908556461334229, + "num_tokens": 354371120.0, + "step": 9288 + }, + { + "epoch": 1.1816562778272484, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7963805198669434, + "num_tokens": 354410825.0, + "step": 9289 + }, + { + "epoch": 1.1817834881058389, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7968816757202148, + "num_tokens": 354453623.0, + "step": 9290 + }, + { + "epoch": 1.1819106983844294, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7966275811195374, + "num_tokens": 354491854.0, + "step": 9291 + }, + { + "epoch": 1.18203790866302, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7819312810897827, + "num_tokens": 354530788.0, + "step": 9292 + }, + { + "epoch": 1.1821651189416105, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8198316097259521, + "num_tokens": 354568576.0, + "step": 9293 + }, + { + "epoch": 1.182292329220201, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.8025053143501282, + "num_tokens": 354611307.0, + "step": 9294 + }, + { + "epoch": 1.1824195394987915, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8072489500045776, + "num_tokens": 354648810.0, + "step": 9295 + }, + { + "epoch": 1.182546749777382, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8023440837860107, + "num_tokens": 354688411.0, + "step": 9296 + }, + { + "epoch": 1.1826739600559726, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7949893474578857, + "num_tokens": 354734700.0, + "step": 9297 + }, + { + "epoch": 1.1828011703345631, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7910451292991638, + "num_tokens": 354765517.0, + "step": 9298 + }, + { + "epoch": 1.1829283806131534, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8057252764701843, + "num_tokens": 354808563.0, + "step": 9299 + }, + { + "epoch": 1.183055590891744, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7772518396377563, + "num_tokens": 354846618.0, + "step": 9300 + }, + { + "epoch": 1.1831828011703345, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.8003461360931396, + "num_tokens": 354881293.0, + "step": 9301 + }, + { + "epoch": 1.183310011448925, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7966079711914062, + "num_tokens": 354925239.0, + "step": 9302 + }, + { + "epoch": 1.1834372217275155, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7814154624938965, + "num_tokens": 354967836.0, + "step": 9303 + }, + { + "epoch": 1.183564432006106, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7777693271636963, + "num_tokens": 355003386.0, + "step": 9304 + }, + { + "epoch": 1.1836916422846966, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7766321897506714, + "num_tokens": 355043391.0, + "step": 9305 + }, + { + "epoch": 1.1838188525632871, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7991524338722229, + "num_tokens": 355082820.0, + "step": 9306 + }, + { + "epoch": 1.1839460628418776, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.821988046169281, + "num_tokens": 355117531.0, + "step": 9307 + }, + { + "epoch": 1.1840732731204682, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7864991426467896, + "num_tokens": 355159485.0, + "step": 9308 + }, + { + "epoch": 1.1842004833990587, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7966618537902832, + "num_tokens": 355206051.0, + "step": 9309 + }, + { + "epoch": 1.1843276936776492, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7911484241485596, + "num_tokens": 355242933.0, + "step": 9310 + }, + { + "epoch": 1.1844549039562398, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8140263557434082, + "num_tokens": 355286311.0, + "step": 9311 + }, + { + "epoch": 1.1845821142348303, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7989141941070557, + "num_tokens": 355327844.0, + "step": 9312 + }, + { + "epoch": 1.1847093245134206, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8059221506118774, + "num_tokens": 355368977.0, + "step": 9313 + }, + { + "epoch": 1.1848365347920111, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7919450998306274, + "num_tokens": 355411479.0, + "step": 9314 + }, + { + "epoch": 1.1849637450706016, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8199849724769592, + "num_tokens": 355447370.0, + "step": 9315 + }, + { + "epoch": 1.1850909553491922, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7975716590881348, + "num_tokens": 355490335.0, + "step": 9316 + }, + { + "epoch": 1.1852181656277827, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.802791178226471, + "num_tokens": 355527843.0, + "step": 9317 + }, + { + "epoch": 1.1853453759063732, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7949327230453491, + "num_tokens": 355566094.0, + "step": 9318 + }, + { + "epoch": 1.1854725861849638, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8110294342041016, + "num_tokens": 355609505.0, + "step": 9319 + }, + { + "epoch": 1.1855997964635543, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7750868797302246, + "num_tokens": 355650930.0, + "step": 9320 + }, + { + "epoch": 1.1857270067421448, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8059184551239014, + "num_tokens": 355684075.0, + "step": 9321 + }, + { + "epoch": 1.1858542170207353, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8064005970954895, + "num_tokens": 355723567.0, + "step": 9322 + }, + { + "epoch": 1.1859814272993259, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7849494218826294, + "num_tokens": 355764915.0, + "step": 9323 + }, + { + "epoch": 1.1861086375779162, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8027170300483704, + "num_tokens": 355804950.0, + "step": 9324 + }, + { + "epoch": 1.1862358478565067, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.805212140083313, + "num_tokens": 355844155.0, + "step": 9325 + }, + { + "epoch": 1.1863630581350972, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8179680705070496, + "num_tokens": 355883971.0, + "step": 9326 + }, + { + "epoch": 1.1864902684136878, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7915387749671936, + "num_tokens": 355922271.0, + "step": 9327 + }, + { + "epoch": 1.1866174786922783, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7935143709182739, + "num_tokens": 355961087.0, + "step": 9328 + }, + { + "epoch": 1.1867446889708688, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8051387667655945, + "num_tokens": 355999553.0, + "step": 9329 + }, + { + "epoch": 1.1868718992494594, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8036649227142334, + "num_tokens": 356038503.0, + "step": 9330 + }, + { + "epoch": 1.1869991095280499, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8104262351989746, + "num_tokens": 356078538.0, + "step": 9331 + }, + { + "epoch": 1.1871263198066404, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8094205856323242, + "num_tokens": 356115031.0, + "step": 9332 + }, + { + "epoch": 1.187253530085231, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8129335045814514, + "num_tokens": 356154751.0, + "step": 9333 + }, + { + "epoch": 1.1873807403638215, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7790015935897827, + "num_tokens": 356192105.0, + "step": 9334 + }, + { + "epoch": 1.187507950642412, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7842683792114258, + "num_tokens": 356224964.0, + "step": 9335 + }, + { + "epoch": 1.1876351609210025, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8033499121665955, + "num_tokens": 356261565.0, + "step": 9336 + }, + { + "epoch": 1.187762371199593, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8054778575897217, + "num_tokens": 356301886.0, + "step": 9337 + }, + { + "epoch": 1.1878895814781834, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8019907474517822, + "num_tokens": 356339481.0, + "step": 9338 + }, + { + "epoch": 1.1880167917567739, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.778616189956665, + "num_tokens": 356387038.0, + "step": 9339 + }, + { + "epoch": 1.1881440020353644, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7950830459594727, + "num_tokens": 356433256.0, + "step": 9340 + }, + { + "epoch": 1.188271212313955, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7845485806465149, + "num_tokens": 356471282.0, + "step": 9341 + }, + { + "epoch": 1.1883984225925455, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8042855858802795, + "num_tokens": 356508951.0, + "step": 9342 + }, + { + "epoch": 1.188525632871136, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7992485761642456, + "num_tokens": 356544487.0, + "step": 9343 + }, + { + "epoch": 1.1886528431497265, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8062257766723633, + "num_tokens": 356580873.0, + "step": 9344 + }, + { + "epoch": 1.188780053428317, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7904859781265259, + "num_tokens": 356613622.0, + "step": 9345 + }, + { + "epoch": 1.1889072637069076, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7762011289596558, + "num_tokens": 356660058.0, + "step": 9346 + }, + { + "epoch": 1.189034473985498, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8013488054275513, + "num_tokens": 356701614.0, + "step": 9347 + }, + { + "epoch": 1.1891616842640884, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.811428427696228, + "num_tokens": 356737105.0, + "step": 9348 + }, + { + "epoch": 1.189288894542679, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7916004657745361, + "num_tokens": 356779382.0, + "step": 9349 + }, + { + "epoch": 1.1894161048212695, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7888938784599304, + "num_tokens": 356813836.0, + "step": 9350 + }, + { + "epoch": 1.18954331509986, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.787908136844635, + "num_tokens": 356852591.0, + "step": 9351 + }, + { + "epoch": 1.1896705253784505, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7926434278488159, + "num_tokens": 356889940.0, + "step": 9352 + }, + { + "epoch": 1.189797735657041, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8009589910507202, + "num_tokens": 356927878.0, + "step": 9353 + }, + { + "epoch": 1.1899249459356316, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8117997646331787, + "num_tokens": 356963339.0, + "step": 9354 + }, + { + "epoch": 1.1900521562142221, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8065069913864136, + "num_tokens": 357002999.0, + "step": 9355 + }, + { + "epoch": 1.1901793664928126, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8021134734153748, + "num_tokens": 357044305.0, + "step": 9356 + }, + { + "epoch": 1.1903065767714032, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8006967306137085, + "num_tokens": 357082197.0, + "step": 9357 + }, + { + "epoch": 1.1904337870499937, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.777907133102417, + "num_tokens": 357119541.0, + "step": 9358 + }, + { + "epoch": 1.1905609973285842, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8178961277008057, + "num_tokens": 357159176.0, + "step": 9359 + }, + { + "epoch": 1.1906882076071748, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7982245683670044, + "num_tokens": 357199204.0, + "step": 9360 + }, + { + "epoch": 1.1908154178857653, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8122445940971375, + "num_tokens": 357236900.0, + "step": 9361 + }, + { + "epoch": 1.1909426281643556, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8060319423675537, + "num_tokens": 357271622.0, + "step": 9362 + }, + { + "epoch": 1.1910698384429461, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8129900693893433, + "num_tokens": 357311145.0, + "step": 9363 + }, + { + "epoch": 1.1911970487215366, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8164474964141846, + "num_tokens": 357344609.0, + "step": 9364 + }, + { + "epoch": 1.1913242590001272, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.788253128528595, + "num_tokens": 357378896.0, + "step": 9365 + }, + { + "epoch": 1.1914514692787177, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8092355728149414, + "num_tokens": 357424730.0, + "step": 9366 + }, + { + "epoch": 1.1915786795573082, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7744170427322388, + "num_tokens": 357461152.0, + "step": 9367 + }, + { + "epoch": 1.1917058898358988, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7959215641021729, + "num_tokens": 357502305.0, + "step": 9368 + }, + { + "epoch": 1.1918331001144893, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7927108407020569, + "num_tokens": 357539187.0, + "step": 9369 + }, + { + "epoch": 1.1919603103930798, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.782717227935791, + "num_tokens": 357569520.0, + "step": 9370 + }, + { + "epoch": 1.1920875206716703, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7990646958351135, + "num_tokens": 357605253.0, + "step": 9371 + }, + { + "epoch": 1.1922147309502609, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8098515272140503, + "num_tokens": 357639701.0, + "step": 9372 + }, + { + "epoch": 1.1923419412288512, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8064864277839661, + "num_tokens": 357675611.0, + "step": 9373 + }, + { + "epoch": 1.1924691515074417, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.808724045753479, + "num_tokens": 357708521.0, + "step": 9374 + }, + { + "epoch": 1.1925963617860322, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7891683578491211, + "num_tokens": 357749970.0, + "step": 9375 + }, + { + "epoch": 1.1927235720646228, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8115604519844055, + "num_tokens": 357783287.0, + "step": 9376 + }, + { + "epoch": 1.1928507823432133, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7967361211776733, + "num_tokens": 357818883.0, + "step": 9377 + }, + { + "epoch": 1.1929779926218038, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8112655282020569, + "num_tokens": 357862862.0, + "step": 9378 + }, + { + "epoch": 1.1931052029003943, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7887210249900818, + "num_tokens": 357901574.0, + "step": 9379 + }, + { + "epoch": 1.1932324131789849, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7958701252937317, + "num_tokens": 357941822.0, + "step": 9380 + }, + { + "epoch": 1.1933596234575754, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7909630537033081, + "num_tokens": 357982820.0, + "step": 9381 + }, + { + "epoch": 1.193486833736166, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.791316032409668, + "num_tokens": 358020491.0, + "step": 9382 + }, + { + "epoch": 1.1936140440147565, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8018788695335388, + "num_tokens": 358059775.0, + "step": 9383 + }, + { + "epoch": 1.193741254293347, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.809342622756958, + "num_tokens": 358098672.0, + "step": 9384 + }, + { + "epoch": 1.1938684645719375, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7892237305641174, + "num_tokens": 358140850.0, + "step": 9385 + }, + { + "epoch": 1.193995674850528, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8062703013420105, + "num_tokens": 358175503.0, + "step": 9386 + }, + { + "epoch": 1.1941228851291183, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8023536205291748, + "num_tokens": 358210615.0, + "step": 9387 + }, + { + "epoch": 1.1942500954077089, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.804790198802948, + "num_tokens": 358246605.0, + "step": 9388 + }, + { + "epoch": 1.1943773056862994, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7935174703598022, + "num_tokens": 358283136.0, + "step": 9389 + }, + { + "epoch": 1.19450451596489, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8142983913421631, + "num_tokens": 358325437.0, + "step": 9390 + }, + { + "epoch": 1.1946317262434805, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.79591965675354, + "num_tokens": 358365360.0, + "step": 9391 + }, + { + "epoch": 1.194758936522071, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8217340707778931, + "num_tokens": 358406502.0, + "step": 9392 + }, + { + "epoch": 1.1948861468006615, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7919948101043701, + "num_tokens": 358437220.0, + "step": 9393 + }, + { + "epoch": 1.195013357079252, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7884010076522827, + "num_tokens": 358475314.0, + "step": 9394 + }, + { + "epoch": 1.1951405673578426, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8148726224899292, + "num_tokens": 358520777.0, + "step": 9395 + }, + { + "epoch": 1.195267777636433, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8178832530975342, + "num_tokens": 358557973.0, + "step": 9396 + }, + { + "epoch": 1.1953949879150234, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8028301000595093, + "num_tokens": 358596795.0, + "step": 9397 + }, + { + "epoch": 1.195522198193614, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7983881235122681, + "num_tokens": 358632146.0, + "step": 9398 + }, + { + "epoch": 1.1956494084722045, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8095017075538635, + "num_tokens": 358672194.0, + "step": 9399 + }, + { + "epoch": 1.195776618750795, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8056488037109375, + "num_tokens": 358710192.0, + "step": 9400 + }, + { + "epoch": 1.1959038290293855, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7887840270996094, + "num_tokens": 358749284.0, + "step": 9401 + }, + { + "epoch": 1.196031039307976, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7910577058792114, + "num_tokens": 358782830.0, + "step": 9402 + }, + { + "epoch": 1.1961582495865666, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7978527545928955, + "num_tokens": 358820696.0, + "step": 9403 + }, + { + "epoch": 1.196285459865157, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.801666796207428, + "num_tokens": 358853616.0, + "step": 9404 + }, + { + "epoch": 1.1964126701437476, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8063883185386658, + "num_tokens": 358893420.0, + "step": 9405 + }, + { + "epoch": 1.1965398804223382, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.803464412689209, + "num_tokens": 358930704.0, + "step": 9406 + }, + { + "epoch": 1.1966670907009287, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8141927719116211, + "num_tokens": 358967730.0, + "step": 9407 + }, + { + "epoch": 1.1967943009795192, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8132111430168152, + "num_tokens": 359007241.0, + "step": 9408 + }, + { + "epoch": 1.1969215112581097, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7838312983512878, + "num_tokens": 359045573.0, + "step": 9409 + }, + { + "epoch": 1.1970487215367003, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8013558387756348, + "num_tokens": 359083420.0, + "step": 9410 + }, + { + "epoch": 1.1971759318152906, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8087981939315796, + "num_tokens": 359117555.0, + "step": 9411 + }, + { + "epoch": 1.1973031420938811, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7877883911132812, + "num_tokens": 359154597.0, + "step": 9412 + }, + { + "epoch": 1.1974303523724716, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7897096872329712, + "num_tokens": 359190074.0, + "step": 9413 + }, + { + "epoch": 1.1975575626510622, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.788119912147522, + "num_tokens": 359232711.0, + "step": 9414 + }, + { + "epoch": 1.1976847729296527, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7955648899078369, + "num_tokens": 359265582.0, + "step": 9415 + }, + { + "epoch": 1.1978119832082432, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8079172968864441, + "num_tokens": 359303177.0, + "step": 9416 + }, + { + "epoch": 1.1979391934868338, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.8108071088790894, + "num_tokens": 359338712.0, + "step": 9417 + }, + { + "epoch": 1.1980664037654243, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8089048862457275, + "num_tokens": 359372849.0, + "step": 9418 + }, + { + "epoch": 1.1981936140440148, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8010786771774292, + "num_tokens": 359407981.0, + "step": 9419 + }, + { + "epoch": 1.1983208243226053, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7804646492004395, + "num_tokens": 359450180.0, + "step": 9420 + }, + { + "epoch": 1.1984480346011959, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7984843254089355, + "num_tokens": 359487909.0, + "step": 9421 + }, + { + "epoch": 1.1985752448797862, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8063849806785583, + "num_tokens": 359522609.0, + "step": 9422 + }, + { + "epoch": 1.1987024551583767, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8005549907684326, + "num_tokens": 359560735.0, + "step": 9423 + }, + { + "epoch": 1.1988296654369672, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7916685342788696, + "num_tokens": 359602994.0, + "step": 9424 + }, + { + "epoch": 1.1989568757155578, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.800118625164032, + "num_tokens": 359638864.0, + "step": 9425 + }, + { + "epoch": 1.1990840859941483, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8052152991294861, + "num_tokens": 359674202.0, + "step": 9426 + }, + { + "epoch": 1.1992112962727388, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8039495944976807, + "num_tokens": 359710408.0, + "step": 9427 + }, + { + "epoch": 1.1993385065513293, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8075211048126221, + "num_tokens": 359749068.0, + "step": 9428 + }, + { + "epoch": 1.1994657168299199, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8030508756637573, + "num_tokens": 359788449.0, + "step": 9429 + }, + { + "epoch": 1.1995929271085104, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.794601559638977, + "num_tokens": 359824167.0, + "step": 9430 + }, + { + "epoch": 1.199720137387101, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7897449731826782, + "num_tokens": 359861476.0, + "step": 9431 + }, + { + "epoch": 1.1998473476656915, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7999249696731567, + "num_tokens": 359906463.0, + "step": 9432 + }, + { + "epoch": 1.199974557944282, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7989427447319031, + "num_tokens": 359940576.0, + "step": 9433 + }, + { + "epoch": 1.2001017682228725, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.811470627784729, + "num_tokens": 359973616.0, + "step": 9434 + }, + { + "epoch": 1.200228978501463, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8134534358978271, + "num_tokens": 360011059.0, + "step": 9435 + }, + { + "epoch": 1.2003561887800533, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8081792593002319, + "num_tokens": 360048307.0, + "step": 9436 + }, + { + "epoch": 1.2004833990586439, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8016370534896851, + "num_tokens": 360086946.0, + "step": 9437 + }, + { + "epoch": 1.2006106093372344, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8050739169120789, + "num_tokens": 360121908.0, + "step": 9438 + }, + { + "epoch": 1.200737819615825, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7934526801109314, + "num_tokens": 360164349.0, + "step": 9439 + }, + { + "epoch": 1.2008650298944155, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.81183922290802, + "num_tokens": 360197660.0, + "step": 9440 + }, + { + "epoch": 1.200992240173006, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7789003849029541, + "num_tokens": 360240297.0, + "step": 9441 + }, + { + "epoch": 1.2011194504515965, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7899640798568726, + "num_tokens": 360281566.0, + "step": 9442 + }, + { + "epoch": 1.201246660730187, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7985810041427612, + "num_tokens": 360313927.0, + "step": 9443 + }, + { + "epoch": 1.2013738710087776, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8012531399726868, + "num_tokens": 360349369.0, + "step": 9444 + }, + { + "epoch": 1.201501081287368, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7981760501861572, + "num_tokens": 360386996.0, + "step": 9445 + }, + { + "epoch": 1.2016282915659584, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8062682151794434, + "num_tokens": 360424883.0, + "step": 9446 + }, + { + "epoch": 1.201755501844549, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.801865816116333, + "num_tokens": 360457992.0, + "step": 9447 + }, + { + "epoch": 1.2018827121231395, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7806527614593506, + "num_tokens": 360497934.0, + "step": 9448 + }, + { + "epoch": 1.20200992240173, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7969115972518921, + "num_tokens": 360533928.0, + "step": 9449 + }, + { + "epoch": 1.2021371326803205, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8023401498794556, + "num_tokens": 360571076.0, + "step": 9450 + }, + { + "epoch": 1.202264342958911, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.795286238193512, + "num_tokens": 360611957.0, + "step": 9451 + }, + { + "epoch": 1.2023915532375016, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7959882020950317, + "num_tokens": 360647823.0, + "step": 9452 + }, + { + "epoch": 1.202518763516092, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7888542413711548, + "num_tokens": 360688341.0, + "step": 9453 + }, + { + "epoch": 1.2026459737946826, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7917003631591797, + "num_tokens": 360730417.0, + "step": 9454 + }, + { + "epoch": 1.2027731840732732, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7774958610534668, + "num_tokens": 360771931.0, + "step": 9455 + }, + { + "epoch": 1.2029003943518637, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7955754995346069, + "num_tokens": 360811460.0, + "step": 9456 + }, + { + "epoch": 1.2030276046304542, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8174580335617065, + "num_tokens": 360849476.0, + "step": 9457 + }, + { + "epoch": 1.2031548149090447, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7887783050537109, + "num_tokens": 360891178.0, + "step": 9458 + }, + { + "epoch": 1.2032820251876353, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.792170524597168, + "num_tokens": 360930503.0, + "step": 9459 + }, + { + "epoch": 1.2034092354662256, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7919380068778992, + "num_tokens": 360963730.0, + "step": 9460 + }, + { + "epoch": 1.203536445744816, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7882958054542542, + "num_tokens": 361003796.0, + "step": 9461 + }, + { + "epoch": 1.2036636560234066, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7982659339904785, + "num_tokens": 361044818.0, + "step": 9462 + }, + { + "epoch": 1.2037908663019972, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8098672032356262, + "num_tokens": 361080692.0, + "step": 9463 + }, + { + "epoch": 1.2039180765805877, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7903499603271484, + "num_tokens": 361116609.0, + "step": 9464 + }, + { + "epoch": 1.2040452868591782, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7850303053855896, + "num_tokens": 361154559.0, + "step": 9465 + }, + { + "epoch": 1.2041724971377687, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8065974712371826, + "num_tokens": 361191005.0, + "step": 9466 + }, + { + "epoch": 1.2042997074163593, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8017945289611816, + "num_tokens": 361230191.0, + "step": 9467 + }, + { + "epoch": 1.2044269176949498, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.794664740562439, + "num_tokens": 361269132.0, + "step": 9468 + }, + { + "epoch": 1.2045541279735403, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7900367975234985, + "num_tokens": 361306674.0, + "step": 9469 + }, + { + "epoch": 1.2046813382521309, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8028163313865662, + "num_tokens": 361348309.0, + "step": 9470 + }, + { + "epoch": 1.2048085485307212, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7917948961257935, + "num_tokens": 361386824.0, + "step": 9471 + }, + { + "epoch": 1.2049357588093117, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7824334502220154, + "num_tokens": 361426590.0, + "step": 9472 + }, + { + "epoch": 1.2050629690879022, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7839823365211487, + "num_tokens": 361463259.0, + "step": 9473 + }, + { + "epoch": 1.2051901793664928, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8093945980072021, + "num_tokens": 361498750.0, + "step": 9474 + }, + { + "epoch": 1.2053173896450833, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7871772646903992, + "num_tokens": 361537101.0, + "step": 9475 + }, + { + "epoch": 1.2054445999236738, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7974529266357422, + "num_tokens": 361572130.0, + "step": 9476 + }, + { + "epoch": 1.2055718102022643, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8140499591827393, + "num_tokens": 361608484.0, + "step": 9477 + }, + { + "epoch": 1.2056990204808549, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8194595575332642, + "num_tokens": 361650361.0, + "step": 9478 + }, + { + "epoch": 1.2058262307594454, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8075010776519775, + "num_tokens": 361680670.0, + "step": 9479 + }, + { + "epoch": 1.205953441038036, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8098488450050354, + "num_tokens": 361723504.0, + "step": 9480 + }, + { + "epoch": 1.2060806513166265, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8082641363143921, + "num_tokens": 361765850.0, + "step": 9481 + }, + { + "epoch": 1.206207861595217, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7896488308906555, + "num_tokens": 361804375.0, + "step": 9482 + }, + { + "epoch": 1.2063350718738075, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7893175482749939, + "num_tokens": 361834714.0, + "step": 9483 + }, + { + "epoch": 1.206462282152398, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.8025562763214111, + "num_tokens": 361877568.0, + "step": 9484 + }, + { + "epoch": 1.2065894924309883, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7953886985778809, + "num_tokens": 361914785.0, + "step": 9485 + }, + { + "epoch": 1.2067167027095789, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.81513512134552, + "num_tokens": 361956781.0, + "step": 9486 + }, + { + "epoch": 1.2068439129881694, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8168830871582031, + "num_tokens": 361997408.0, + "step": 9487 + }, + { + "epoch": 1.20697112326676, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7930608987808228, + "num_tokens": 362030226.0, + "step": 9488 + }, + { + "epoch": 1.2070983335453505, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8111525177955627, + "num_tokens": 362068370.0, + "step": 9489 + }, + { + "epoch": 1.207225543823941, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8119237422943115, + "num_tokens": 362108569.0, + "step": 9490 + }, + { + "epoch": 1.2073527541025315, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7902160286903381, + "num_tokens": 362140476.0, + "step": 9491 + }, + { + "epoch": 1.207479964381122, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8052881956100464, + "num_tokens": 362171132.0, + "step": 9492 + }, + { + "epoch": 1.2076071746597126, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.792785108089447, + "num_tokens": 362204791.0, + "step": 9493 + }, + { + "epoch": 1.207734384938303, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8031035661697388, + "num_tokens": 362242360.0, + "step": 9494 + }, + { + "epoch": 1.2078615952168934, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7970234155654907, + "num_tokens": 362276558.0, + "step": 9495 + }, + { + "epoch": 1.207988805495484, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7810949087142944, + "num_tokens": 362307055.0, + "step": 9496 + }, + { + "epoch": 1.2081160157740745, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8092222213745117, + "num_tokens": 362342433.0, + "step": 9497 + }, + { + "epoch": 1.208243226052665, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5791, + "mean_token_accuracy": 0.8232519626617432, + "num_tokens": 362377008.0, + "step": 9498 + }, + { + "epoch": 1.2083704363312555, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8131608963012695, + "num_tokens": 362412089.0, + "step": 9499 + }, + { + "epoch": 1.208497646609846, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7941687107086182, + "num_tokens": 362455817.0, + "step": 9500 + }, + { + "epoch": 1.2086248568884366, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.8050042390823364, + "num_tokens": 362498859.0, + "step": 9501 + }, + { + "epoch": 1.208752067167027, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8012011051177979, + "num_tokens": 362542346.0, + "step": 9502 + }, + { + "epoch": 1.2088792774456176, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8168890476226807, + "num_tokens": 362575360.0, + "step": 9503 + }, + { + "epoch": 1.2090064877242082, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8022217750549316, + "num_tokens": 362614737.0, + "step": 9504 + }, + { + "epoch": 1.2091336980027987, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7989053726196289, + "num_tokens": 362653756.0, + "step": 9505 + }, + { + "epoch": 1.2092609082813892, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8043488264083862, + "num_tokens": 362690745.0, + "step": 9506 + }, + { + "epoch": 1.2093881185599797, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8086694478988647, + "num_tokens": 362726328.0, + "step": 9507 + }, + { + "epoch": 1.2095153288385703, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8028151988983154, + "num_tokens": 362764285.0, + "step": 9508 + }, + { + "epoch": 1.2096425391171606, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7767655253410339, + "num_tokens": 362794697.0, + "step": 9509 + }, + { + "epoch": 1.209769749395751, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7994961738586426, + "num_tokens": 362831618.0, + "step": 9510 + }, + { + "epoch": 1.2098969596743416, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8096241354942322, + "num_tokens": 362866404.0, + "step": 9511 + }, + { + "epoch": 1.2100241699529322, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8157844543457031, + "num_tokens": 362902554.0, + "step": 9512 + }, + { + "epoch": 1.2101513802315227, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7948544025421143, + "num_tokens": 362944169.0, + "step": 9513 + }, + { + "epoch": 1.2102785905101132, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7764732837677002, + "num_tokens": 362983789.0, + "step": 9514 + }, + { + "epoch": 1.2104058007887037, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8108651638031006, + "num_tokens": 363021551.0, + "step": 9515 + }, + { + "epoch": 1.2105330110672943, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7843389511108398, + "num_tokens": 363064951.0, + "step": 9516 + }, + { + "epoch": 1.2106602213458848, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8006367087364197, + "num_tokens": 363110876.0, + "step": 9517 + }, + { + "epoch": 1.2107874316244753, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7985217571258545, + "num_tokens": 363155310.0, + "step": 9518 + }, + { + "epoch": 1.2109146419030659, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8142526745796204, + "num_tokens": 363193166.0, + "step": 9519 + }, + { + "epoch": 1.2110418521816562, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7757754325866699, + "num_tokens": 363225163.0, + "step": 9520 + }, + { + "epoch": 1.2111690624602467, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7974185943603516, + "num_tokens": 363264720.0, + "step": 9521 + }, + { + "epoch": 1.2112962727388372, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7867632508277893, + "num_tokens": 363305029.0, + "step": 9522 + }, + { + "epoch": 1.2114234830174277, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.809261679649353, + "num_tokens": 363344128.0, + "step": 9523 + }, + { + "epoch": 1.2115506932960183, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.8019341826438904, + "num_tokens": 363385903.0, + "step": 9524 + }, + { + "epoch": 1.2116779035746088, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8051787614822388, + "num_tokens": 363425216.0, + "step": 9525 + }, + { + "epoch": 1.2118051138531993, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8083134889602661, + "num_tokens": 363462691.0, + "step": 9526 + }, + { + "epoch": 1.2119323241317899, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7819135189056396, + "num_tokens": 363495849.0, + "step": 9527 + }, + { + "epoch": 1.2120595344103804, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7949155569076538, + "num_tokens": 363535268.0, + "step": 9528 + }, + { + "epoch": 1.212186744688971, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7788106203079224, + "num_tokens": 363573523.0, + "step": 9529 + }, + { + "epoch": 1.2123139549675614, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7992739081382751, + "num_tokens": 363607545.0, + "step": 9530 + }, + { + "epoch": 1.212441165246152, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7990126013755798, + "num_tokens": 363648456.0, + "step": 9531 + }, + { + "epoch": 1.2125683755247425, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8095007538795471, + "num_tokens": 363688726.0, + "step": 9532 + }, + { + "epoch": 1.212695585803333, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7925440073013306, + "num_tokens": 363728557.0, + "step": 9533 + }, + { + "epoch": 1.2128227960819233, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7977138757705688, + "num_tokens": 363764191.0, + "step": 9534 + }, + { + "epoch": 1.2129500063605139, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7827212810516357, + "num_tokens": 363798126.0, + "step": 9535 + }, + { + "epoch": 1.2130772166391044, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7835936546325684, + "num_tokens": 363836037.0, + "step": 9536 + }, + { + "epoch": 1.213204426917695, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.5697, + "mean_token_accuracy": 0.8301755785942078, + "num_tokens": 363872359.0, + "step": 9537 + }, + { + "epoch": 1.2133316371962855, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.796707272529602, + "num_tokens": 363914206.0, + "step": 9538 + }, + { + "epoch": 1.213458847474876, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8034416437149048, + "num_tokens": 363956147.0, + "step": 9539 + }, + { + "epoch": 1.2135860577534665, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8135765790939331, + "num_tokens": 363986328.0, + "step": 9540 + }, + { + "epoch": 1.213713268032057, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.798781156539917, + "num_tokens": 364024905.0, + "step": 9541 + }, + { + "epoch": 1.2138404783106476, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8035391569137573, + "num_tokens": 364068628.0, + "step": 9542 + }, + { + "epoch": 1.213967688589238, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7924999594688416, + "num_tokens": 364104261.0, + "step": 9543 + }, + { + "epoch": 1.2140948988678284, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7946053743362427, + "num_tokens": 364147524.0, + "step": 9544 + }, + { + "epoch": 1.214222109146419, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.801764726638794, + "num_tokens": 364187715.0, + "step": 9545 + }, + { + "epoch": 1.2143493194250095, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8080788254737854, + "num_tokens": 364226081.0, + "step": 9546 + }, + { + "epoch": 1.2144765297036, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7997754812240601, + "num_tokens": 364268390.0, + "step": 9547 + }, + { + "epoch": 1.2146037399821905, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8054289817810059, + "num_tokens": 364310176.0, + "step": 9548 + }, + { + "epoch": 1.214730950260781, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8089305758476257, + "num_tokens": 364345652.0, + "step": 9549 + }, + { + "epoch": 1.2148581605393716, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7979110479354858, + "num_tokens": 364383810.0, + "step": 9550 + }, + { + "epoch": 1.214985370817962, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8060653805732727, + "num_tokens": 364421169.0, + "step": 9551 + }, + { + "epoch": 1.2151125810965526, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7880762219429016, + "num_tokens": 364452552.0, + "step": 9552 + }, + { + "epoch": 1.2152397913751432, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7916061878204346, + "num_tokens": 364486779.0, + "step": 9553 + }, + { + "epoch": 1.2153670016537337, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8109080791473389, + "num_tokens": 364526237.0, + "step": 9554 + }, + { + "epoch": 1.2154942119323242, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8108311295509338, + "num_tokens": 364568446.0, + "step": 9555 + }, + { + "epoch": 1.2156214222109147, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8081712126731873, + "num_tokens": 364604448.0, + "step": 9556 + }, + { + "epoch": 1.2157486324895053, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7958053946495056, + "num_tokens": 364642707.0, + "step": 9557 + }, + { + "epoch": 1.2158758427680956, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7918803691864014, + "num_tokens": 364678428.0, + "step": 9558 + }, + { + "epoch": 1.216003053046686, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8189507722854614, + "num_tokens": 364718915.0, + "step": 9559 + }, + { + "epoch": 1.2161302633252766, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8193371295928955, + "num_tokens": 364756030.0, + "step": 9560 + }, + { + "epoch": 1.2162574736038672, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7991079092025757, + "num_tokens": 364796158.0, + "step": 9561 + }, + { + "epoch": 1.2163846838824577, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8085025548934937, + "num_tokens": 364832722.0, + "step": 9562 + }, + { + "epoch": 1.2165118941610482, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8032140135765076, + "num_tokens": 364874120.0, + "step": 9563 + }, + { + "epoch": 1.2166391044396387, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8019665479660034, + "num_tokens": 364913166.0, + "step": 9564 + }, + { + "epoch": 1.2167663147182293, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7767062187194824, + "num_tokens": 364949586.0, + "step": 9565 + }, + { + "epoch": 1.2168935249968198, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7854003310203552, + "num_tokens": 364989070.0, + "step": 9566 + }, + { + "epoch": 1.2170207352754103, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8013238906860352, + "num_tokens": 365027050.0, + "step": 9567 + }, + { + "epoch": 1.2171479455540009, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7975832223892212, + "num_tokens": 365060050.0, + "step": 9568 + }, + { + "epoch": 1.2172751558325912, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7711914777755737, + "num_tokens": 365098030.0, + "step": 9569 + }, + { + "epoch": 1.2174023661111817, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8073000311851501, + "num_tokens": 365141143.0, + "step": 9570 + }, + { + "epoch": 1.2175295763897722, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7746397852897644, + "num_tokens": 365174204.0, + "step": 9571 + }, + { + "epoch": 1.2176567866683627, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7973447442054749, + "num_tokens": 365207170.0, + "step": 9572 + }, + { + "epoch": 1.2177839969469533, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8067705631256104, + "num_tokens": 365248926.0, + "step": 9573 + }, + { + "epoch": 1.2179112072255438, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7979105114936829, + "num_tokens": 365283588.0, + "step": 9574 + }, + { + "epoch": 1.2180384175041343, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8002008199691772, + "num_tokens": 365324869.0, + "step": 9575 + }, + { + "epoch": 1.2181656277827249, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7958155870437622, + "num_tokens": 365366832.0, + "step": 9576 + }, + { + "epoch": 1.2182928380613154, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7892538905143738, + "num_tokens": 365401148.0, + "step": 9577 + }, + { + "epoch": 1.218420048339906, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8060179352760315, + "num_tokens": 365442303.0, + "step": 9578 + }, + { + "epoch": 1.2185472586184964, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8106734752655029, + "num_tokens": 365477877.0, + "step": 9579 + }, + { + "epoch": 1.218674468897087, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7833089828491211, + "num_tokens": 365520648.0, + "step": 9580 + }, + { + "epoch": 1.2188016791756775, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7764303684234619, + "num_tokens": 365555129.0, + "step": 9581 + }, + { + "epoch": 1.218928889454268, + "grad_norm": 9.1875, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8053978681564331, + "num_tokens": 365594368.0, + "step": 9582 + }, + { + "epoch": 1.2190560997328583, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7991745471954346, + "num_tokens": 365634763.0, + "step": 9583 + }, + { + "epoch": 1.2191833100114489, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8183594346046448, + "num_tokens": 365675951.0, + "step": 9584 + }, + { + "epoch": 1.2193105202900394, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8035389184951782, + "num_tokens": 365710348.0, + "step": 9585 + }, + { + "epoch": 1.21943773056863, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7989858388900757, + "num_tokens": 365745166.0, + "step": 9586 + }, + { + "epoch": 1.2195649408472204, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8091776967048645, + "num_tokens": 365780122.0, + "step": 9587 + }, + { + "epoch": 1.219692151125811, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7962453365325928, + "num_tokens": 365816326.0, + "step": 9588 + }, + { + "epoch": 1.2198193614044015, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7697998285293579, + "num_tokens": 365858938.0, + "step": 9589 + }, + { + "epoch": 1.219946571682992, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7898301482200623, + "num_tokens": 365893892.0, + "step": 9590 + }, + { + "epoch": 1.2200737819615826, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7884291410446167, + "num_tokens": 365938787.0, + "step": 9591 + }, + { + "epoch": 1.220200992240173, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8017237186431885, + "num_tokens": 365972664.0, + "step": 9592 + }, + { + "epoch": 1.2203282025187634, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8120793104171753, + "num_tokens": 366013367.0, + "step": 9593 + }, + { + "epoch": 1.220455412797354, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8081341981887817, + "num_tokens": 366051637.0, + "step": 9594 + }, + { + "epoch": 1.2205826230759445, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7933976054191589, + "num_tokens": 366092713.0, + "step": 9595 + }, + { + "epoch": 1.220709833354535, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7988862991333008, + "num_tokens": 366131594.0, + "step": 9596 + }, + { + "epoch": 1.2208370436331255, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7974516153335571, + "num_tokens": 366170871.0, + "step": 9597 + }, + { + "epoch": 1.220964253911716, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7870806455612183, + "num_tokens": 366212974.0, + "step": 9598 + }, + { + "epoch": 1.2210914641903066, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7862449884414673, + "num_tokens": 366255904.0, + "step": 9599 + }, + { + "epoch": 1.221218674468897, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8031688332557678, + "num_tokens": 366288887.0, + "step": 9600 + }, + { + "epoch": 1.2213458847474876, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.8148592710494995, + "num_tokens": 366328796.0, + "step": 9601 + }, + { + "epoch": 1.2214730950260781, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7975319623947144, + "num_tokens": 366373093.0, + "step": 9602 + }, + { + "epoch": 1.2216003053046687, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8150079250335693, + "num_tokens": 366414353.0, + "step": 9603 + }, + { + "epoch": 1.2217275155832592, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8140314817428589, + "num_tokens": 366450683.0, + "step": 9604 + }, + { + "epoch": 1.2218547258618497, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8004273176193237, + "num_tokens": 366487158.0, + "step": 9605 + }, + { + "epoch": 1.2219819361404403, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8023952841758728, + "num_tokens": 366529984.0, + "step": 9606 + }, + { + "epoch": 1.2221091464190306, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8047109842300415, + "num_tokens": 366564100.0, + "step": 9607 + }, + { + "epoch": 1.222236356697621, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8102800846099854, + "num_tokens": 366603752.0, + "step": 9608 + }, + { + "epoch": 1.2223635669762116, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7997338771820068, + "num_tokens": 366642170.0, + "step": 9609 + }, + { + "epoch": 1.2224907772548022, + "grad_norm": 60.5, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7825082540512085, + "num_tokens": 366684038.0, + "step": 9610 + }, + { + "epoch": 1.2226179875333927, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7931140661239624, + "num_tokens": 366717832.0, + "step": 9611 + }, + { + "epoch": 1.2227451978119832, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7934336066246033, + "num_tokens": 366757999.0, + "step": 9612 + }, + { + "epoch": 1.2228724080905737, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8128198981285095, + "num_tokens": 366798533.0, + "step": 9613 + }, + { + "epoch": 1.2229996183691643, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7991466522216797, + "num_tokens": 366843066.0, + "step": 9614 + }, + { + "epoch": 1.2231268286477548, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.820022702217102, + "num_tokens": 366883190.0, + "step": 9615 + }, + { + "epoch": 1.2232540389263453, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.804444432258606, + "num_tokens": 366920642.0, + "step": 9616 + }, + { + "epoch": 1.2233812492049359, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8039549589157104, + "num_tokens": 366954468.0, + "step": 9617 + }, + { + "epoch": 1.2235084594835262, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7842179536819458, + "num_tokens": 366991170.0, + "step": 9618 + }, + { + "epoch": 1.2236356697621167, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8067697882652283, + "num_tokens": 367031833.0, + "step": 9619 + }, + { + "epoch": 1.2237628800407072, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.794198751449585, + "num_tokens": 367074709.0, + "step": 9620 + }, + { + "epoch": 1.2238900903192977, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7947019934654236, + "num_tokens": 367106820.0, + "step": 9621 + }, + { + "epoch": 1.2240173005978883, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8137341737747192, + "num_tokens": 367142298.0, + "step": 9622 + }, + { + "epoch": 1.2241445108764788, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7860770225524902, + "num_tokens": 367179957.0, + "step": 9623 + }, + { + "epoch": 1.2242717211550693, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8042867183685303, + "num_tokens": 367226800.0, + "step": 9624 + }, + { + "epoch": 1.2243989314336599, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8024181127548218, + "num_tokens": 367257168.0, + "step": 9625 + }, + { + "epoch": 1.2245261417122504, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.799047589302063, + "num_tokens": 367295254.0, + "step": 9626 + }, + { + "epoch": 1.224653351990841, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7852655649185181, + "num_tokens": 367335814.0, + "step": 9627 + }, + { + "epoch": 1.2247805622694314, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7967488765716553, + "num_tokens": 367377592.0, + "step": 9628 + }, + { + "epoch": 1.224907772548022, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7853916883468628, + "num_tokens": 367408069.0, + "step": 9629 + }, + { + "epoch": 1.2250349828266125, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7978463172912598, + "num_tokens": 367444787.0, + "step": 9630 + }, + { + "epoch": 1.225162193105203, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7829872965812683, + "num_tokens": 367486848.0, + "step": 9631 + }, + { + "epoch": 1.2252894033837933, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8133233785629272, + "num_tokens": 367530456.0, + "step": 9632 + }, + { + "epoch": 1.2254166136623839, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8091574907302856, + "num_tokens": 367566513.0, + "step": 9633 + }, + { + "epoch": 1.2255438239409744, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7753121852874756, + "num_tokens": 367607757.0, + "step": 9634 + }, + { + "epoch": 1.225671034219565, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8013453483581543, + "num_tokens": 367645525.0, + "step": 9635 + }, + { + "epoch": 1.2257982444981554, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8060276508331299, + "num_tokens": 367680840.0, + "step": 9636 + }, + { + "epoch": 1.225925454776746, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8085857033729553, + "num_tokens": 367717444.0, + "step": 9637 + }, + { + "epoch": 1.2260526650553365, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7787553071975708, + "num_tokens": 367756548.0, + "step": 9638 + }, + { + "epoch": 1.226179875333927, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7898679375648499, + "num_tokens": 367798341.0, + "step": 9639 + }, + { + "epoch": 1.2263070856125176, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8091865181922913, + "num_tokens": 367835087.0, + "step": 9640 + }, + { + "epoch": 1.226434295891108, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.804661214351654, + "num_tokens": 367873816.0, + "step": 9641 + }, + { + "epoch": 1.2265615061696984, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7944332957267761, + "num_tokens": 367913621.0, + "step": 9642 + }, + { + "epoch": 1.226688716448289, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7895222902297974, + "num_tokens": 367946881.0, + "step": 9643 + }, + { + "epoch": 1.2268159267268794, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8040971159934998, + "num_tokens": 367980806.0, + "step": 9644 + }, + { + "epoch": 1.22694313700547, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8090510964393616, + "num_tokens": 368016041.0, + "step": 9645 + }, + { + "epoch": 1.2270703472840605, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.794445276260376, + "num_tokens": 368052331.0, + "step": 9646 + }, + { + "epoch": 1.227197557562651, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8112417459487915, + "num_tokens": 368091458.0, + "step": 9647 + }, + { + "epoch": 1.2273247678412416, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7851896286010742, + "num_tokens": 368132433.0, + "step": 9648 + }, + { + "epoch": 1.227451978119832, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7997108697891235, + "num_tokens": 368171030.0, + "step": 9649 + }, + { + "epoch": 1.2275791883984226, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7915858030319214, + "num_tokens": 368211555.0, + "step": 9650 + }, + { + "epoch": 1.2277063986770131, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.809200644493103, + "num_tokens": 368252076.0, + "step": 9651 + }, + { + "epoch": 1.2278336089556037, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7894086837768555, + "num_tokens": 368291265.0, + "step": 9652 + }, + { + "epoch": 1.2279608192341942, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7931777238845825, + "num_tokens": 368332178.0, + "step": 9653 + }, + { + "epoch": 1.2280880295127847, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8001551628112793, + "num_tokens": 368374103.0, + "step": 9654 + }, + { + "epoch": 1.2282152397913753, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8040642738342285, + "num_tokens": 368411928.0, + "step": 9655 + }, + { + "epoch": 1.2283424500699656, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7702694535255432, + "num_tokens": 368449200.0, + "step": 9656 + }, + { + "epoch": 1.228469660348556, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7816351652145386, + "num_tokens": 368489935.0, + "step": 9657 + }, + { + "epoch": 1.2285968706271466, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8071666955947876, + "num_tokens": 368522227.0, + "step": 9658 + }, + { + "epoch": 1.2287240809057371, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7961591482162476, + "num_tokens": 368555084.0, + "step": 9659 + }, + { + "epoch": 1.2288512911843277, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7918946743011475, + "num_tokens": 368591845.0, + "step": 9660 + }, + { + "epoch": 1.2289785014629182, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7881937623023987, + "num_tokens": 368637841.0, + "step": 9661 + }, + { + "epoch": 1.2291057117415087, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.8161256313323975, + "num_tokens": 368677470.0, + "step": 9662 + }, + { + "epoch": 1.2292329220200993, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8161886930465698, + "num_tokens": 368718412.0, + "step": 9663 + }, + { + "epoch": 1.2293601322986898, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7913305759429932, + "num_tokens": 368759267.0, + "step": 9664 + }, + { + "epoch": 1.2294873425772803, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7886520624160767, + "num_tokens": 368797586.0, + "step": 9665 + }, + { + "epoch": 1.2296145528558708, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7895263433456421, + "num_tokens": 368833801.0, + "step": 9666 + }, + { + "epoch": 1.2297417631344612, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7841849327087402, + "num_tokens": 368878513.0, + "step": 9667 + }, + { + "epoch": 1.2298689734130517, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7933014631271362, + "num_tokens": 368917815.0, + "step": 9668 + }, + { + "epoch": 1.2299961836916422, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7879598140716553, + "num_tokens": 368954971.0, + "step": 9669 + }, + { + "epoch": 1.2301233939702327, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8018604516983032, + "num_tokens": 368996212.0, + "step": 9670 + }, + { + "epoch": 1.2302506042488233, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.810194730758667, + "num_tokens": 369032186.0, + "step": 9671 + }, + { + "epoch": 1.2303778145274138, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8036563396453857, + "num_tokens": 369077565.0, + "step": 9672 + }, + { + "epoch": 1.2305050248060043, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8065629005432129, + "num_tokens": 369111935.0, + "step": 9673 + }, + { + "epoch": 1.2306322350845948, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7995456457138062, + "num_tokens": 369154374.0, + "step": 9674 + }, + { + "epoch": 1.2307594453631854, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7902332544326782, + "num_tokens": 369198557.0, + "step": 9675 + }, + { + "epoch": 1.230886655641776, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7840487360954285, + "num_tokens": 369237756.0, + "step": 9676 + }, + { + "epoch": 1.2310138659203664, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7871066927909851, + "num_tokens": 369270155.0, + "step": 9677 + }, + { + "epoch": 1.231141076198957, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7935462594032288, + "num_tokens": 369310870.0, + "step": 9678 + }, + { + "epoch": 1.2312682864775475, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7815160751342773, + "num_tokens": 369344679.0, + "step": 9679 + }, + { + "epoch": 1.231395496756138, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7988733053207397, + "num_tokens": 369380388.0, + "step": 9680 + }, + { + "epoch": 1.2315227070347283, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8115914463996887, + "num_tokens": 369422327.0, + "step": 9681 + }, + { + "epoch": 1.2316499173133189, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8041328191757202, + "num_tokens": 369457160.0, + "step": 9682 + }, + { + "epoch": 1.2317771275919094, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.8008707761764526, + "num_tokens": 369497883.0, + "step": 9683 + }, + { + "epoch": 1.2319043378705, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.5628, + "mean_token_accuracy": 0.8280711770057678, + "num_tokens": 369540014.0, + "step": 9684 + }, + { + "epoch": 1.2320315481490904, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7975739240646362, + "num_tokens": 369580424.0, + "step": 9685 + }, + { + "epoch": 1.232158758427681, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7951244115829468, + "num_tokens": 369623615.0, + "step": 9686 + }, + { + "epoch": 1.2322859687062715, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7823748588562012, + "num_tokens": 369658408.0, + "step": 9687 + }, + { + "epoch": 1.232413178984862, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7952723503112793, + "num_tokens": 369696311.0, + "step": 9688 + }, + { + "epoch": 1.2325403892634526, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7841655015945435, + "num_tokens": 369737473.0, + "step": 9689 + }, + { + "epoch": 1.232667599542043, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7921273708343506, + "num_tokens": 369770151.0, + "step": 9690 + }, + { + "epoch": 1.2327948098206334, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8110204935073853, + "num_tokens": 369813503.0, + "step": 9691 + }, + { + "epoch": 1.232922020099224, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8017845153808594, + "num_tokens": 369848768.0, + "step": 9692 + }, + { + "epoch": 1.2330492303778144, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.804315984249115, + "num_tokens": 369892510.0, + "step": 9693 + }, + { + "epoch": 1.233176440656405, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8058944940567017, + "num_tokens": 369929638.0, + "step": 9694 + }, + { + "epoch": 1.2333036509349955, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.8212814331054688, + "num_tokens": 369971201.0, + "step": 9695 + }, + { + "epoch": 1.233430861213586, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7888011932373047, + "num_tokens": 370011713.0, + "step": 9696 + }, + { + "epoch": 1.2335580714921766, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7678261995315552, + "num_tokens": 370042647.0, + "step": 9697 + }, + { + "epoch": 1.233685281770767, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7983826398849487, + "num_tokens": 370083976.0, + "step": 9698 + }, + { + "epoch": 1.2338124920493576, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7904052734375, + "num_tokens": 370126435.0, + "step": 9699 + }, + { + "epoch": 1.2339397023279481, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8054732084274292, + "num_tokens": 370165038.0, + "step": 9700 + }, + { + "epoch": 1.2340669126065387, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7865760922431946, + "num_tokens": 370203438.0, + "step": 9701 + }, + { + "epoch": 1.2341941228851292, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8115165829658508, + "num_tokens": 370235144.0, + "step": 9702 + }, + { + "epoch": 1.2343213331637197, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7976900339126587, + "num_tokens": 370278936.0, + "step": 9703 + }, + { + "epoch": 1.2344485434423103, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8002952933311462, + "num_tokens": 370312595.0, + "step": 9704 + }, + { + "epoch": 1.2345757537209006, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7931694984436035, + "num_tokens": 370351697.0, + "step": 9705 + }, + { + "epoch": 1.234702963999491, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8094510436058044, + "num_tokens": 370391264.0, + "step": 9706 + }, + { + "epoch": 1.2348301742780816, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8042829632759094, + "num_tokens": 370425248.0, + "step": 9707 + }, + { + "epoch": 1.2349573845566721, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7743470668792725, + "num_tokens": 370462670.0, + "step": 9708 + }, + { + "epoch": 1.2350845948352627, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7719894051551819, + "num_tokens": 370506126.0, + "step": 9709 + }, + { + "epoch": 1.2352118051138532, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8088055849075317, + "num_tokens": 370546895.0, + "step": 9710 + }, + { + "epoch": 1.2353390153924437, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8058512210845947, + "num_tokens": 370590645.0, + "step": 9711 + }, + { + "epoch": 1.2354662256710343, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8021913766860962, + "num_tokens": 370628614.0, + "step": 9712 + }, + { + "epoch": 1.2355934359496248, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8032418489456177, + "num_tokens": 370666082.0, + "step": 9713 + }, + { + "epoch": 1.2357206462282153, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7805780172348022, + "num_tokens": 370705780.0, + "step": 9714 + }, + { + "epoch": 1.2358478565068058, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8048740029335022, + "num_tokens": 370742550.0, + "step": 9715 + }, + { + "epoch": 1.2359750667853961, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8082621097564697, + "num_tokens": 370784406.0, + "step": 9716 + }, + { + "epoch": 1.2361022770639867, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7925115823745728, + "num_tokens": 370820911.0, + "step": 9717 + }, + { + "epoch": 1.2362294873425772, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.804266095161438, + "num_tokens": 370857427.0, + "step": 9718 + }, + { + "epoch": 1.2363566976211677, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7880110740661621, + "num_tokens": 370897989.0, + "step": 9719 + }, + { + "epoch": 1.2364839078997583, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7888979315757751, + "num_tokens": 370932672.0, + "step": 9720 + }, + { + "epoch": 1.2366111181783488, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7874656319618225, + "num_tokens": 370971027.0, + "step": 9721 + }, + { + "epoch": 1.2367383284569393, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8144564032554626, + "num_tokens": 371014742.0, + "step": 9722 + }, + { + "epoch": 1.2368655387355298, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7826749682426453, + "num_tokens": 371052065.0, + "step": 9723 + }, + { + "epoch": 1.2369927490141204, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7812316417694092, + "num_tokens": 371093897.0, + "step": 9724 + }, + { + "epoch": 1.237119959292711, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7983945608139038, + "num_tokens": 371132390.0, + "step": 9725 + }, + { + "epoch": 1.2372471695713014, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8088260889053345, + "num_tokens": 371170468.0, + "step": 9726 + }, + { + "epoch": 1.237374379849892, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7953410744667053, + "num_tokens": 371209323.0, + "step": 9727 + }, + { + "epoch": 1.2375015901284825, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.797276496887207, + "num_tokens": 371246526.0, + "step": 9728 + }, + { + "epoch": 1.237628800407073, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7994939684867859, + "num_tokens": 371282850.0, + "step": 9729 + }, + { + "epoch": 1.2377560106856633, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.806588888168335, + "num_tokens": 371312614.0, + "step": 9730 + }, + { + "epoch": 1.2378832209642538, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7992846369743347, + "num_tokens": 371350170.0, + "step": 9731 + }, + { + "epoch": 1.2380104312428444, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8065049648284912, + "num_tokens": 371389295.0, + "step": 9732 + }, + { + "epoch": 1.238137641521435, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.797766387462616, + "num_tokens": 371430579.0, + "step": 9733 + }, + { + "epoch": 1.2382648518000254, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7885698080062866, + "num_tokens": 371465087.0, + "step": 9734 + }, + { + "epoch": 1.238392062078616, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8050054907798767, + "num_tokens": 371503549.0, + "step": 9735 + }, + { + "epoch": 1.2385192723572065, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8195000886917114, + "num_tokens": 371542953.0, + "step": 9736 + }, + { + "epoch": 1.238646482635797, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8013341426849365, + "num_tokens": 371582541.0, + "step": 9737 + }, + { + "epoch": 1.2387736929143875, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8067938089370728, + "num_tokens": 371617221.0, + "step": 9738 + }, + { + "epoch": 1.238900903192978, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8063837885856628, + "num_tokens": 371660773.0, + "step": 9739 + }, + { + "epoch": 1.2390281134715684, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7971814274787903, + "num_tokens": 371694763.0, + "step": 9740 + }, + { + "epoch": 1.239155323750159, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8019452691078186, + "num_tokens": 371734793.0, + "step": 9741 + }, + { + "epoch": 1.2392825340287494, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7926886081695557, + "num_tokens": 371775691.0, + "step": 9742 + }, + { + "epoch": 1.23940974430734, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7947463989257812, + "num_tokens": 371817319.0, + "step": 9743 + }, + { + "epoch": 1.2395369545859305, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7978495359420776, + "num_tokens": 371847431.0, + "step": 9744 + }, + { + "epoch": 1.239664164864521, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8132829666137695, + "num_tokens": 371881788.0, + "step": 9745 + }, + { + "epoch": 1.2397913751431116, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7948867678642273, + "num_tokens": 371917388.0, + "step": 9746 + }, + { + "epoch": 1.239918585421702, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8049265146255493, + "num_tokens": 371957561.0, + "step": 9747 + }, + { + "epoch": 1.2400457957002926, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7859917879104614, + "num_tokens": 371987322.0, + "step": 9748 + }, + { + "epoch": 1.2401730059788831, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7992820143699646, + "num_tokens": 372027263.0, + "step": 9749 + }, + { + "epoch": 1.2403002162574737, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7977645397186279, + "num_tokens": 372061441.0, + "step": 9750 + }, + { + "epoch": 1.2404274265360642, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8053047060966492, + "num_tokens": 372099085.0, + "step": 9751 + }, + { + "epoch": 1.2405546368146547, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.791948139667511, + "num_tokens": 372134106.0, + "step": 9752 + }, + { + "epoch": 1.2406818470932452, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8036028146743774, + "num_tokens": 372164635.0, + "step": 9753 + }, + { + "epoch": 1.2408090573718356, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8075035214424133, + "num_tokens": 372204924.0, + "step": 9754 + }, + { + "epoch": 1.240936267650426, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8010760545730591, + "num_tokens": 372240939.0, + "step": 9755 + }, + { + "epoch": 1.2410634779290166, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7901686429977417, + "num_tokens": 372272355.0, + "step": 9756 + }, + { + "epoch": 1.2411906882076071, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7992903590202332, + "num_tokens": 372309566.0, + "step": 9757 + }, + { + "epoch": 1.2413178984861977, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7926881909370422, + "num_tokens": 372347768.0, + "step": 9758 + }, + { + "epoch": 1.2414451087647882, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8034096956253052, + "num_tokens": 372381020.0, + "step": 9759 + }, + { + "epoch": 1.2415723190433787, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.78618323802948, + "num_tokens": 372422434.0, + "step": 9760 + }, + { + "epoch": 1.2416995293219693, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7867776155471802, + "num_tokens": 372459342.0, + "step": 9761 + }, + { + "epoch": 1.2418267396005598, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8039072751998901, + "num_tokens": 372496637.0, + "step": 9762 + }, + { + "epoch": 1.2419539498791503, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8100401163101196, + "num_tokens": 372532068.0, + "step": 9763 + }, + { + "epoch": 1.2420811601577408, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.808883786201477, + "num_tokens": 372565460.0, + "step": 9764 + }, + { + "epoch": 1.2422083704363311, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8062046766281128, + "num_tokens": 372605998.0, + "step": 9765 + }, + { + "epoch": 1.2423355807149217, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7906432151794434, + "num_tokens": 372642038.0, + "step": 9766 + }, + { + "epoch": 1.2424627909935122, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7925034761428833, + "num_tokens": 372678192.0, + "step": 9767 + }, + { + "epoch": 1.2425900012721027, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7900675535202026, + "num_tokens": 372712867.0, + "step": 9768 + }, + { + "epoch": 1.2427172115506933, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7824915647506714, + "num_tokens": 372745790.0, + "step": 9769 + }, + { + "epoch": 1.2428444218292838, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8069103956222534, + "num_tokens": 372777634.0, + "step": 9770 + }, + { + "epoch": 1.2429716321078743, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7837411165237427, + "num_tokens": 372820527.0, + "step": 9771 + }, + { + "epoch": 1.2430988423864648, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8077148795127869, + "num_tokens": 372858479.0, + "step": 9772 + }, + { + "epoch": 1.2432260526650554, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8030308485031128, + "num_tokens": 372895219.0, + "step": 9773 + }, + { + "epoch": 1.243353262943646, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.8177074790000916, + "num_tokens": 372934491.0, + "step": 9774 + }, + { + "epoch": 1.2434804732222364, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7903684973716736, + "num_tokens": 372972892.0, + "step": 9775 + }, + { + "epoch": 1.243607683500827, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8031626343727112, + "num_tokens": 373011564.0, + "step": 9776 + }, + { + "epoch": 1.2437348937794175, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8001720905303955, + "num_tokens": 373049444.0, + "step": 9777 + }, + { + "epoch": 1.243862104058008, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.8009153604507446, + "num_tokens": 373085466.0, + "step": 9778 + }, + { + "epoch": 1.2439893143365983, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7787078022956848, + "num_tokens": 373124712.0, + "step": 9779 + }, + { + "epoch": 1.2441165246151888, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8015292882919312, + "num_tokens": 373160008.0, + "step": 9780 + }, + { + "epoch": 1.2442437348937794, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8000447154045105, + "num_tokens": 373192110.0, + "step": 9781 + }, + { + "epoch": 1.24437094517237, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8046475052833557, + "num_tokens": 373234432.0, + "step": 9782 + }, + { + "epoch": 1.2444981554509604, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7889735102653503, + "num_tokens": 373270725.0, + "step": 9783 + }, + { + "epoch": 1.244625365729551, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7737329602241516, + "num_tokens": 373307531.0, + "step": 9784 + }, + { + "epoch": 1.2447525760081415, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8089486360549927, + "num_tokens": 373351657.0, + "step": 9785 + }, + { + "epoch": 1.244879786286732, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.803115963935852, + "num_tokens": 373388811.0, + "step": 9786 + }, + { + "epoch": 1.2450069965653225, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.811774492263794, + "num_tokens": 373423378.0, + "step": 9787 + }, + { + "epoch": 1.245134206843913, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7992832660675049, + "num_tokens": 373460982.0, + "step": 9788 + }, + { + "epoch": 1.2452614171225034, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8100863695144653, + "num_tokens": 373501474.0, + "step": 9789 + }, + { + "epoch": 1.245388627401094, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7986149191856384, + "num_tokens": 373537830.0, + "step": 9790 + }, + { + "epoch": 1.2455158376796844, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7920933961868286, + "num_tokens": 373573975.0, + "step": 9791 + }, + { + "epoch": 1.245643047958275, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7776639461517334, + "num_tokens": 373612157.0, + "step": 9792 + }, + { + "epoch": 1.2457702582368655, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8152967095375061, + "num_tokens": 373645065.0, + "step": 9793 + }, + { + "epoch": 1.245897468515456, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7943319082260132, + "num_tokens": 373687658.0, + "step": 9794 + }, + { + "epoch": 1.2460246787940465, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7979884743690491, + "num_tokens": 373732262.0, + "step": 9795 + }, + { + "epoch": 1.246151889072637, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7836176156997681, + "num_tokens": 373772410.0, + "step": 9796 + }, + { + "epoch": 1.2462790993512276, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7956761121749878, + "num_tokens": 373808773.0, + "step": 9797 + }, + { + "epoch": 1.2464063096298181, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7975589632987976, + "num_tokens": 373848945.0, + "step": 9798 + }, + { + "epoch": 1.2465335199084087, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7925812005996704, + "num_tokens": 373892716.0, + "step": 9799 + }, + { + "epoch": 1.2466607301869992, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8176708221435547, + "num_tokens": 373929443.0, + "step": 9800 + }, + { + "epoch": 1.2467879404655897, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8010938763618469, + "num_tokens": 373968882.0, + "step": 9801 + }, + { + "epoch": 1.2469151507441802, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.8180606365203857, + "num_tokens": 374009265.0, + "step": 9802 + }, + { + "epoch": 1.2470423610227706, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7770792245864868, + "num_tokens": 374047152.0, + "step": 9803 + }, + { + "epoch": 1.247169571301361, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7941807508468628, + "num_tokens": 374083127.0, + "step": 9804 + }, + { + "epoch": 1.2472967815799516, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8097939491271973, + "num_tokens": 374127086.0, + "step": 9805 + }, + { + "epoch": 1.2474239918585421, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7933958172798157, + "num_tokens": 374163879.0, + "step": 9806 + }, + { + "epoch": 1.2475512021371327, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8087548613548279, + "num_tokens": 374206639.0, + "step": 9807 + }, + { + "epoch": 1.2476784124157232, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.8001916408538818, + "num_tokens": 374244106.0, + "step": 9808 + }, + { + "epoch": 1.2478056226943137, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7962289452552795, + "num_tokens": 374285776.0, + "step": 9809 + }, + { + "epoch": 1.2479328329729042, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8120807409286499, + "num_tokens": 374323036.0, + "step": 9810 + }, + { + "epoch": 1.2480600432514948, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8055059313774109, + "num_tokens": 374360204.0, + "step": 9811 + }, + { + "epoch": 1.2481872535300853, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8099960088729858, + "num_tokens": 374394961.0, + "step": 9812 + }, + { + "epoch": 1.2483144638086758, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8026741743087769, + "num_tokens": 374436142.0, + "step": 9813 + }, + { + "epoch": 1.2484416740872661, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7956628203392029, + "num_tokens": 374474263.0, + "step": 9814 + }, + { + "epoch": 1.2485688843658567, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7888441681861877, + "num_tokens": 374510250.0, + "step": 9815 + }, + { + "epoch": 1.2486960946444472, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8190751075744629, + "num_tokens": 374549579.0, + "step": 9816 + }, + { + "epoch": 1.2488233049230377, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7995480298995972, + "num_tokens": 374588631.0, + "step": 9817 + }, + { + "epoch": 1.2489505152016283, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7957699298858643, + "num_tokens": 374627092.0, + "step": 9818 + }, + { + "epoch": 1.2490777254802188, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7998186945915222, + "num_tokens": 374657797.0, + "step": 9819 + }, + { + "epoch": 1.2492049357588093, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.804124653339386, + "num_tokens": 374694011.0, + "step": 9820 + }, + { + "epoch": 1.2493321460373998, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8055359721183777, + "num_tokens": 374735127.0, + "step": 9821 + }, + { + "epoch": 1.2494593563159904, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7940887212753296, + "num_tokens": 374774170.0, + "step": 9822 + }, + { + "epoch": 1.249586566594581, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7946966290473938, + "num_tokens": 374809162.0, + "step": 9823 + }, + { + "epoch": 1.2497137768731714, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7920693755149841, + "num_tokens": 374847974.0, + "step": 9824 + }, + { + "epoch": 1.249840987151762, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.8191072940826416, + "num_tokens": 374894911.0, + "step": 9825 + }, + { + "epoch": 1.2499681974303525, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7917401790618896, + "num_tokens": 374933534.0, + "step": 9826 + }, + { + "epoch": 1.250095407708943, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7817814350128174, + "num_tokens": 374968183.0, + "step": 9827 + }, + { + "epoch": 1.2502226179875333, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7928544282913208, + "num_tokens": 375005973.0, + "step": 9828 + }, + { + "epoch": 1.2503498282661238, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.794750988483429, + "num_tokens": 375043446.0, + "step": 9829 + }, + { + "epoch": 1.2504770385447144, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8039751648902893, + "num_tokens": 375081518.0, + "step": 9830 + }, + { + "epoch": 1.250604248823305, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7872390747070312, + "num_tokens": 375117767.0, + "step": 9831 + }, + { + "epoch": 1.2507314591018954, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7720175385475159, + "num_tokens": 375152061.0, + "step": 9832 + }, + { + "epoch": 1.250858669380486, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7939097285270691, + "num_tokens": 375195677.0, + "step": 9833 + }, + { + "epoch": 1.2509858796590765, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7965476512908936, + "num_tokens": 375230319.0, + "step": 9834 + }, + { + "epoch": 1.251113089937667, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8002553582191467, + "num_tokens": 375268970.0, + "step": 9835 + }, + { + "epoch": 1.2512403002162575, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8085750341415405, + "num_tokens": 375303661.0, + "step": 9836 + }, + { + "epoch": 1.2513675104948478, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8061288595199585, + "num_tokens": 375343278.0, + "step": 9837 + }, + { + "epoch": 1.2514947207734384, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8035119771957397, + "num_tokens": 375378574.0, + "step": 9838 + }, + { + "epoch": 1.251621931052029, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7931143045425415, + "num_tokens": 375412781.0, + "step": 9839 + }, + { + "epoch": 1.2517491413306194, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8094775676727295, + "num_tokens": 375450431.0, + "step": 9840 + }, + { + "epoch": 1.25187635160921, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7916190028190613, + "num_tokens": 375490692.0, + "step": 9841 + }, + { + "epoch": 1.2520035618878005, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8044461011886597, + "num_tokens": 375535278.0, + "step": 9842 + }, + { + "epoch": 1.252130772166391, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7967571020126343, + "num_tokens": 375575859.0, + "step": 9843 + }, + { + "epoch": 1.2522579824449815, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7902098298072815, + "num_tokens": 375611084.0, + "step": 9844 + }, + { + "epoch": 1.252385192723572, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.807549238204956, + "num_tokens": 375644926.0, + "step": 9845 + }, + { + "epoch": 1.2525124030021626, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8082632422447205, + "num_tokens": 375686970.0, + "step": 9846 + }, + { + "epoch": 1.2526396132807531, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8114534616470337, + "num_tokens": 375727372.0, + "step": 9847 + }, + { + "epoch": 1.2527668235593437, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7876838445663452, + "num_tokens": 375763568.0, + "step": 9848 + }, + { + "epoch": 1.2528940338379342, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7822225093841553, + "num_tokens": 375797954.0, + "step": 9849 + }, + { + "epoch": 1.2530212441165247, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7927300930023193, + "num_tokens": 375836362.0, + "step": 9850 + }, + { + "epoch": 1.2531484543951152, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.796757161617279, + "num_tokens": 375879403.0, + "step": 9851 + }, + { + "epoch": 1.2532756646737058, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7968341112136841, + "num_tokens": 375916728.0, + "step": 9852 + }, + { + "epoch": 1.253402874952296, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7862880229949951, + "num_tokens": 375951743.0, + "step": 9853 + }, + { + "epoch": 1.2535300852308866, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7668444514274597, + "num_tokens": 375988872.0, + "step": 9854 + }, + { + "epoch": 1.2536572955094771, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7869542837142944, + "num_tokens": 376030388.0, + "step": 9855 + }, + { + "epoch": 1.2537845057880677, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.809520959854126, + "num_tokens": 376062975.0, + "step": 9856 + }, + { + "epoch": 1.2539117160666582, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7907239198684692, + "num_tokens": 376101873.0, + "step": 9857 + }, + { + "epoch": 1.2540389263452487, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7794278860092163, + "num_tokens": 376138774.0, + "step": 9858 + }, + { + "epoch": 1.2541661366238392, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7854235768318176, + "num_tokens": 376172357.0, + "step": 9859 + }, + { + "epoch": 1.2542933469024298, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.790310800075531, + "num_tokens": 376212977.0, + "step": 9860 + }, + { + "epoch": 1.2544205571810203, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7921174764633179, + "num_tokens": 376256989.0, + "step": 9861 + }, + { + "epoch": 1.2545477674596106, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8023814558982849, + "num_tokens": 376300130.0, + "step": 9862 + }, + { + "epoch": 1.2546749777382011, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7999340295791626, + "num_tokens": 376337475.0, + "step": 9863 + }, + { + "epoch": 1.2548021880167917, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7883070707321167, + "num_tokens": 376376485.0, + "step": 9864 + }, + { + "epoch": 1.2549293982953822, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8052809834480286, + "num_tokens": 376413393.0, + "step": 9865 + }, + { + "epoch": 1.2550566085739727, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7857672572135925, + "num_tokens": 376448192.0, + "step": 9866 + }, + { + "epoch": 1.2551838188525632, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8005996942520142, + "num_tokens": 376489382.0, + "step": 9867 + }, + { + "epoch": 1.2553110291311538, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.807262122631073, + "num_tokens": 376527221.0, + "step": 9868 + }, + { + "epoch": 1.2554382394097443, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7934638857841492, + "num_tokens": 376567045.0, + "step": 9869 + }, + { + "epoch": 1.2555654496883348, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8072944283485413, + "num_tokens": 376607039.0, + "step": 9870 + }, + { + "epoch": 1.2556926599669254, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7995802760124207, + "num_tokens": 376643220.0, + "step": 9871 + }, + { + "epoch": 1.255819870245516, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8159732818603516, + "num_tokens": 376681417.0, + "step": 9872 + }, + { + "epoch": 1.2559470805241064, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.8201254606246948, + "num_tokens": 376722836.0, + "step": 9873 + }, + { + "epoch": 1.256074290802697, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7882247567176819, + "num_tokens": 376763194.0, + "step": 9874 + }, + { + "epoch": 1.2562015010812875, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7971811294555664, + "num_tokens": 376803701.0, + "step": 9875 + }, + { + "epoch": 1.256328711359878, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8011126518249512, + "num_tokens": 376835296.0, + "step": 9876 + }, + { + "epoch": 1.2564559216384683, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.801535964012146, + "num_tokens": 376876317.0, + "step": 9877 + }, + { + "epoch": 1.2565831319170588, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8001584410667419, + "num_tokens": 376916403.0, + "step": 9878 + }, + { + "epoch": 1.2567103421956494, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8122431039810181, + "num_tokens": 376956079.0, + "step": 9879 + }, + { + "epoch": 1.25683755247424, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7893853187561035, + "num_tokens": 376991618.0, + "step": 9880 + }, + { + "epoch": 1.2569647627528304, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8097397089004517, + "num_tokens": 377033964.0, + "step": 9881 + }, + { + "epoch": 1.257091973031421, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8046433925628662, + "num_tokens": 377070091.0, + "step": 9882 + }, + { + "epoch": 1.2572191833100115, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8069275617599487, + "num_tokens": 377111185.0, + "step": 9883 + }, + { + "epoch": 1.257346393588602, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8024448156356812, + "num_tokens": 377147637.0, + "step": 9884 + }, + { + "epoch": 1.2574736038671925, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7974978685379028, + "num_tokens": 377181918.0, + "step": 9885 + }, + { + "epoch": 1.2576008141457828, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7908353805541992, + "num_tokens": 377222900.0, + "step": 9886 + }, + { + "epoch": 1.2577280244243734, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.8171603679656982, + "num_tokens": 377263107.0, + "step": 9887 + }, + { + "epoch": 1.257855234702964, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8173447847366333, + "num_tokens": 377301384.0, + "step": 9888 + }, + { + "epoch": 1.2579824449815544, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7896823883056641, + "num_tokens": 377339865.0, + "step": 9889 + }, + { + "epoch": 1.258109655260145, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8034932613372803, + "num_tokens": 377381582.0, + "step": 9890 + }, + { + "epoch": 1.2582368655387355, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7706847786903381, + "num_tokens": 377425548.0, + "step": 9891 + }, + { + "epoch": 1.258364075817326, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.794897735118866, + "num_tokens": 377462298.0, + "step": 9892 + }, + { + "epoch": 1.2584912860959165, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.789153516292572, + "num_tokens": 377503873.0, + "step": 9893 + }, + { + "epoch": 1.258618496374507, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.811302900314331, + "num_tokens": 377540497.0, + "step": 9894 + }, + { + "epoch": 1.2587457066530976, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7948703169822693, + "num_tokens": 377582242.0, + "step": 9895 + }, + { + "epoch": 1.2588729169316881, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8108617067337036, + "num_tokens": 377614972.0, + "step": 9896 + }, + { + "epoch": 1.2590001272102787, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7926018834114075, + "num_tokens": 377645392.0, + "step": 9897 + }, + { + "epoch": 1.2591273374888692, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8062905073165894, + "num_tokens": 377680317.0, + "step": 9898 + }, + { + "epoch": 1.2592545477674597, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8013262748718262, + "num_tokens": 377714879.0, + "step": 9899 + }, + { + "epoch": 1.2593817580460502, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7726433277130127, + "num_tokens": 377751404.0, + "step": 9900 + }, + { + "epoch": 1.2595089683246408, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8028813004493713, + "num_tokens": 377788354.0, + "step": 9901 + }, + { + "epoch": 1.259636178603231, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7873920202255249, + "num_tokens": 377827259.0, + "step": 9902 + }, + { + "epoch": 1.2597633888818216, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7948852181434631, + "num_tokens": 377872477.0, + "step": 9903 + }, + { + "epoch": 1.2598905991604121, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7907733917236328, + "num_tokens": 377911159.0, + "step": 9904 + }, + { + "epoch": 1.2600178094390027, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7994693517684937, + "num_tokens": 377951919.0, + "step": 9905 + }, + { + "epoch": 1.2601450197175932, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.779718816280365, + "num_tokens": 377997567.0, + "step": 9906 + }, + { + "epoch": 1.2602722299961837, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7971947193145752, + "num_tokens": 378033811.0, + "step": 9907 + }, + { + "epoch": 1.2603994402747742, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7908841967582703, + "num_tokens": 378070136.0, + "step": 9908 + }, + { + "epoch": 1.2605266505533648, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7993989586830139, + "num_tokens": 378106902.0, + "step": 9909 + }, + { + "epoch": 1.2606538608319553, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7966586947441101, + "num_tokens": 378144638.0, + "step": 9910 + }, + { + "epoch": 1.2607810711105456, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.8123770356178284, + "num_tokens": 378179059.0, + "step": 9911 + }, + { + "epoch": 1.2609082813891361, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7792127728462219, + "num_tokens": 378210346.0, + "step": 9912 + }, + { + "epoch": 1.2610354916677267, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8065059185028076, + "num_tokens": 378242687.0, + "step": 9913 + }, + { + "epoch": 1.2611627019463172, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.804875373840332, + "num_tokens": 378279421.0, + "step": 9914 + }, + { + "epoch": 1.2612899122249077, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7930678129196167, + "num_tokens": 378318133.0, + "step": 9915 + }, + { + "epoch": 1.2614171225034982, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7854510545730591, + "num_tokens": 378362987.0, + "step": 9916 + }, + { + "epoch": 1.2615443327820888, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8145959377288818, + "num_tokens": 378396197.0, + "step": 9917 + }, + { + "epoch": 1.2616715430606793, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8059873580932617, + "num_tokens": 378433691.0, + "step": 9918 + }, + { + "epoch": 1.2617987533392698, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8025848865509033, + "num_tokens": 378470277.0, + "step": 9919 + }, + { + "epoch": 1.2619259636178604, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.792577862739563, + "num_tokens": 378510500.0, + "step": 9920 + }, + { + "epoch": 1.2620531738964509, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8012973070144653, + "num_tokens": 378552893.0, + "step": 9921 + }, + { + "epoch": 1.2621803841750414, + "grad_norm": 8.9375, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8085030317306519, + "num_tokens": 378590917.0, + "step": 9922 + }, + { + "epoch": 1.262307594453632, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8082834482192993, + "num_tokens": 378637764.0, + "step": 9923 + }, + { + "epoch": 1.2624348047322225, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7866011261940002, + "num_tokens": 378675497.0, + "step": 9924 + }, + { + "epoch": 1.262562015010813, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7869994640350342, + "num_tokens": 378710470.0, + "step": 9925 + }, + { + "epoch": 1.2626892252894033, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7945601940155029, + "num_tokens": 378758116.0, + "step": 9926 + }, + { + "epoch": 1.2628164355679938, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7952331304550171, + "num_tokens": 378798166.0, + "step": 9927 + }, + { + "epoch": 1.2629436458465844, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8158735632896423, + "num_tokens": 378839899.0, + "step": 9928 + }, + { + "epoch": 1.263070856125175, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8014093637466431, + "num_tokens": 378886022.0, + "step": 9929 + }, + { + "epoch": 1.2631980664037654, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7840127944946289, + "num_tokens": 378925097.0, + "step": 9930 + }, + { + "epoch": 1.263325276682356, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8172347545623779, + "num_tokens": 378965242.0, + "step": 9931 + }, + { + "epoch": 1.2634524869609465, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7757185697555542, + "num_tokens": 379005932.0, + "step": 9932 + }, + { + "epoch": 1.263579697239537, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7961947321891785, + "num_tokens": 379045817.0, + "step": 9933 + }, + { + "epoch": 1.2637069075181275, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7966439723968506, + "num_tokens": 379086481.0, + "step": 9934 + }, + { + "epoch": 1.2638341177967178, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7836824655532837, + "num_tokens": 379129782.0, + "step": 9935 + }, + { + "epoch": 1.2639613280753084, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7973862290382385, + "num_tokens": 379171525.0, + "step": 9936 + }, + { + "epoch": 1.264088538353899, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7902107238769531, + "num_tokens": 379213887.0, + "step": 9937 + }, + { + "epoch": 1.2642157486324894, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8008706569671631, + "num_tokens": 379252634.0, + "step": 9938 + }, + { + "epoch": 1.26434295891108, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7930702567100525, + "num_tokens": 379287402.0, + "step": 9939 + }, + { + "epoch": 1.2644701691896705, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8101096153259277, + "num_tokens": 379325592.0, + "step": 9940 + }, + { + "epoch": 1.264597379468261, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7820968627929688, + "num_tokens": 379359010.0, + "step": 9941 + }, + { + "epoch": 1.2647245897468515, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.792496919631958, + "num_tokens": 379399783.0, + "step": 9942 + }, + { + "epoch": 1.264851800025442, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8030819892883301, + "num_tokens": 379442207.0, + "step": 9943 + }, + { + "epoch": 1.2649790103040326, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7977714538574219, + "num_tokens": 379479561.0, + "step": 9944 + }, + { + "epoch": 1.2651062205826231, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8042725324630737, + "num_tokens": 379516413.0, + "step": 9945 + }, + { + "epoch": 1.2652334308612136, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7863390445709229, + "num_tokens": 379557993.0, + "step": 9946 + }, + { + "epoch": 1.2653606411398042, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7766067385673523, + "num_tokens": 379594028.0, + "step": 9947 + }, + { + "epoch": 1.2654878514183947, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7895699739456177, + "num_tokens": 379631235.0, + "step": 9948 + }, + { + "epoch": 1.2656150616969852, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7868943214416504, + "num_tokens": 379668714.0, + "step": 9949 + }, + { + "epoch": 1.2657422719755758, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.8089861273765564, + "num_tokens": 379710233.0, + "step": 9950 + }, + { + "epoch": 1.265869482254166, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7892049551010132, + "num_tokens": 379746208.0, + "step": 9951 + }, + { + "epoch": 1.2659966925327566, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8114269375801086, + "num_tokens": 379796420.0, + "step": 9952 + }, + { + "epoch": 1.2661239028113471, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7930406332015991, + "num_tokens": 379830027.0, + "step": 9953 + }, + { + "epoch": 1.2662511130899377, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7992267608642578, + "num_tokens": 379871751.0, + "step": 9954 + }, + { + "epoch": 1.2663783233685282, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7876813411712646, + "num_tokens": 379908255.0, + "step": 9955 + }, + { + "epoch": 1.2665055336471187, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7975184917449951, + "num_tokens": 379944502.0, + "step": 9956 + }, + { + "epoch": 1.2666327439257092, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7950143814086914, + "num_tokens": 379981929.0, + "step": 9957 + }, + { + "epoch": 1.2667599542042998, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7978146076202393, + "num_tokens": 380018876.0, + "step": 9958 + }, + { + "epoch": 1.2668871644828903, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7901857495307922, + "num_tokens": 380054153.0, + "step": 9959 + }, + { + "epoch": 1.2670143747614806, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7918240427970886, + "num_tokens": 380086550.0, + "step": 9960 + }, + { + "epoch": 1.2671415850400711, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7926210165023804, + "num_tokens": 380120308.0, + "step": 9961 + }, + { + "epoch": 1.2672687953186617, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7795805931091309, + "num_tokens": 380160204.0, + "step": 9962 + }, + { + "epoch": 1.2673960055972522, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7989670038223267, + "num_tokens": 380196687.0, + "step": 9963 + }, + { + "epoch": 1.2675232158758427, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8144754767417908, + "num_tokens": 380237451.0, + "step": 9964 + }, + { + "epoch": 1.2676504261544332, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7963625192642212, + "num_tokens": 380276631.0, + "step": 9965 + }, + { + "epoch": 1.2677776364330238, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8030251264572144, + "num_tokens": 380319432.0, + "step": 9966 + }, + { + "epoch": 1.2679048467116143, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8077307939529419, + "num_tokens": 380356666.0, + "step": 9967 + }, + { + "epoch": 1.2680320569902048, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.800855278968811, + "num_tokens": 380390347.0, + "step": 9968 + }, + { + "epoch": 1.2681592672687954, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7994486689567566, + "num_tokens": 380424083.0, + "step": 9969 + }, + { + "epoch": 1.2682864775473859, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7960606813430786, + "num_tokens": 380471179.0, + "step": 9970 + }, + { + "epoch": 1.2684136878259764, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7985020875930786, + "num_tokens": 380510291.0, + "step": 9971 + }, + { + "epoch": 1.268540898104567, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8067001104354858, + "num_tokens": 380556297.0, + "step": 9972 + }, + { + "epoch": 1.2686681083831575, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8057801127433777, + "num_tokens": 380592113.0, + "step": 9973 + }, + { + "epoch": 1.268795318661748, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8187806606292725, + "num_tokens": 380623942.0, + "step": 9974 + }, + { + "epoch": 1.2689225289403383, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7891198992729187, + "num_tokens": 380662300.0, + "step": 9975 + }, + { + "epoch": 1.2690497392189288, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7946710586547852, + "num_tokens": 380701452.0, + "step": 9976 + }, + { + "epoch": 1.2691769494975194, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8130511045455933, + "num_tokens": 380741248.0, + "step": 9977 + }, + { + "epoch": 1.2693041597761099, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7951750755310059, + "num_tokens": 380781455.0, + "step": 9978 + }, + { + "epoch": 1.2694313700547004, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7887364029884338, + "num_tokens": 380818139.0, + "step": 9979 + }, + { + "epoch": 1.269558580333291, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.803931474685669, + "num_tokens": 380860530.0, + "step": 9980 + }, + { + "epoch": 1.2696857906118815, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8135730028152466, + "num_tokens": 380900825.0, + "step": 9981 + }, + { + "epoch": 1.269813000890472, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7924550175666809, + "num_tokens": 380939270.0, + "step": 9982 + }, + { + "epoch": 1.2699402111690625, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7777851819992065, + "num_tokens": 380981708.0, + "step": 9983 + }, + { + "epoch": 1.2700674214476528, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7981546521186829, + "num_tokens": 381015186.0, + "step": 9984 + }, + { + "epoch": 1.2701946317262434, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8091024160385132, + "num_tokens": 381055503.0, + "step": 9985 + }, + { + "epoch": 1.270321842004834, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8155605792999268, + "num_tokens": 381091935.0, + "step": 9986 + }, + { + "epoch": 1.2704490522834244, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8037747144699097, + "num_tokens": 381127951.0, + "step": 9987 + }, + { + "epoch": 1.270576262562015, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7898263931274414, + "num_tokens": 381161918.0, + "step": 9988 + }, + { + "epoch": 1.2707034728406055, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7853788733482361, + "num_tokens": 381198747.0, + "step": 9989 + }, + { + "epoch": 1.270830683119196, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.800754725933075, + "num_tokens": 381242714.0, + "step": 9990 + }, + { + "epoch": 1.2709578933977865, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8037327527999878, + "num_tokens": 381274667.0, + "step": 9991 + }, + { + "epoch": 1.271085103676377, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8135804533958435, + "num_tokens": 381318110.0, + "step": 9992 + }, + { + "epoch": 1.2712123139549676, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.79575115442276, + "num_tokens": 381356694.0, + "step": 9993 + }, + { + "epoch": 1.2713395242335581, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7820844054222107, + "num_tokens": 381395218.0, + "step": 9994 + }, + { + "epoch": 1.2714667345121486, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8090964555740356, + "num_tokens": 381429195.0, + "step": 9995 + }, + { + "epoch": 1.2715939447907392, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7924507260322571, + "num_tokens": 381468276.0, + "step": 9996 + }, + { + "epoch": 1.2717211550693297, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8040629625320435, + "num_tokens": 381503400.0, + "step": 9997 + }, + { + "epoch": 1.2718483653479202, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8005448579788208, + "num_tokens": 381542126.0, + "step": 9998 + }, + { + "epoch": 1.2719755756265108, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8022209405899048, + "num_tokens": 381579169.0, + "step": 9999 + }, + { + "epoch": 1.272102785905101, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7840263843536377, + "num_tokens": 381615501.0, + "step": 10000 + }, + { + "epoch": 1.2722299961836916, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7915312051773071, + "num_tokens": 381649737.0, + "step": 10001 + }, + { + "epoch": 1.2723572064622821, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8004350066184998, + "num_tokens": 381693776.0, + "step": 10002 + }, + { + "epoch": 1.2724844167408726, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8085014820098877, + "num_tokens": 381733324.0, + "step": 10003 + }, + { + "epoch": 1.2726116270194632, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7911579608917236, + "num_tokens": 381765893.0, + "step": 10004 + }, + { + "epoch": 1.2727388372980537, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8008716106414795, + "num_tokens": 381802782.0, + "step": 10005 + }, + { + "epoch": 1.2728660475766442, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8075999617576599, + "num_tokens": 381843448.0, + "step": 10006 + }, + { + "epoch": 1.2729932578552348, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.785312294960022, + "num_tokens": 381878850.0, + "step": 10007 + }, + { + "epoch": 1.2731204681338253, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7768757343292236, + "num_tokens": 381915508.0, + "step": 10008 + }, + { + "epoch": 1.2732476784124156, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7903561592102051, + "num_tokens": 381952699.0, + "step": 10009 + }, + { + "epoch": 1.2733748886910061, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7910497784614563, + "num_tokens": 381989947.0, + "step": 10010 + }, + { + "epoch": 1.2735020989695967, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7979124784469604, + "num_tokens": 382027701.0, + "step": 10011 + }, + { + "epoch": 1.2736293092481872, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.8001147508621216, + "num_tokens": 382066415.0, + "step": 10012 + }, + { + "epoch": 1.2737565195267777, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7947565317153931, + "num_tokens": 382108402.0, + "step": 10013 + }, + { + "epoch": 1.2738837298053682, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8044360280036926, + "num_tokens": 382146865.0, + "step": 10014 + }, + { + "epoch": 1.2740109400839588, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7990487217903137, + "num_tokens": 382186746.0, + "step": 10015 + }, + { + "epoch": 1.2741381503625493, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8154312372207642, + "num_tokens": 382220693.0, + "step": 10016 + }, + { + "epoch": 1.2742653606411398, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7901020050048828, + "num_tokens": 382259549.0, + "step": 10017 + }, + { + "epoch": 1.2743925709197303, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.805182695388794, + "num_tokens": 382299164.0, + "step": 10018 + }, + { + "epoch": 1.2745197811983209, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8113188147544861, + "num_tokens": 382333596.0, + "step": 10019 + }, + { + "epoch": 1.2746469914769114, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.807388424873352, + "num_tokens": 382372321.0, + "step": 10020 + }, + { + "epoch": 1.274774201755502, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8020312786102295, + "num_tokens": 382411125.0, + "step": 10021 + }, + { + "epoch": 1.2749014120340925, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7895132303237915, + "num_tokens": 382450935.0, + "step": 10022 + }, + { + "epoch": 1.275028622312683, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7940368056297302, + "num_tokens": 382488232.0, + "step": 10023 + }, + { + "epoch": 1.2751558325912733, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7954905033111572, + "num_tokens": 382522280.0, + "step": 10024 + }, + { + "epoch": 1.2752830428698638, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8047066330909729, + "num_tokens": 382562788.0, + "step": 10025 + }, + { + "epoch": 1.2754102531484544, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7838230729103088, + "num_tokens": 382598534.0, + "step": 10026 + }, + { + "epoch": 1.2755374634270449, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7983396053314209, + "num_tokens": 382641803.0, + "step": 10027 + }, + { + "epoch": 1.2756646737056354, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8090983629226685, + "num_tokens": 382677871.0, + "step": 10028 + }, + { + "epoch": 1.275791883984226, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8067830801010132, + "num_tokens": 382714989.0, + "step": 10029 + }, + { + "epoch": 1.2759190942628165, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.8029947280883789, + "num_tokens": 382751919.0, + "step": 10030 + }, + { + "epoch": 1.276046304541407, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8131451606750488, + "num_tokens": 382793550.0, + "step": 10031 + }, + { + "epoch": 1.2761735148199975, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8120276927947998, + "num_tokens": 382833142.0, + "step": 10032 + }, + { + "epoch": 1.2763007250985878, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8063910007476807, + "num_tokens": 382869934.0, + "step": 10033 + }, + { + "epoch": 1.2764279353771784, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7852828502655029, + "num_tokens": 382909533.0, + "step": 10034 + }, + { + "epoch": 1.2765551456557689, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8032598495483398, + "num_tokens": 382947462.0, + "step": 10035 + }, + { + "epoch": 1.2766823559343594, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7994014024734497, + "num_tokens": 382989973.0, + "step": 10036 + }, + { + "epoch": 1.27680956621295, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7988885045051575, + "num_tokens": 383033052.0, + "step": 10037 + }, + { + "epoch": 1.2769367764915405, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7989580631256104, + "num_tokens": 383075981.0, + "step": 10038 + }, + { + "epoch": 1.277063986770131, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.8156408667564392, + "num_tokens": 383111984.0, + "step": 10039 + }, + { + "epoch": 1.2771911970487215, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8069241046905518, + "num_tokens": 383155323.0, + "step": 10040 + }, + { + "epoch": 1.277318407327312, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7987611293792725, + "num_tokens": 383190069.0, + "step": 10041 + }, + { + "epoch": 1.2774456176059026, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7891493439674377, + "num_tokens": 383224999.0, + "step": 10042 + }, + { + "epoch": 1.2775728278844931, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7909571528434753, + "num_tokens": 383267764.0, + "step": 10043 + }, + { + "epoch": 1.2777000381630836, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8080446124076843, + "num_tokens": 383306148.0, + "step": 10044 + }, + { + "epoch": 1.2778272484416742, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7999939322471619, + "num_tokens": 383343408.0, + "step": 10045 + }, + { + "epoch": 1.2779544587202647, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8207435607910156, + "num_tokens": 383383626.0, + "step": 10046 + }, + { + "epoch": 1.2780816689988552, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7865993976593018, + "num_tokens": 383423218.0, + "step": 10047 + }, + { + "epoch": 1.2782088792774458, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8055056929588318, + "num_tokens": 383461126.0, + "step": 10048 + }, + { + "epoch": 1.278336089556036, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.791727602481842, + "num_tokens": 383501808.0, + "step": 10049 + }, + { + "epoch": 1.2784632998346266, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.8225374221801758, + "num_tokens": 383542395.0, + "step": 10050 + }, + { + "epoch": 1.2785905101132171, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7925409078598022, + "num_tokens": 383579968.0, + "step": 10051 + }, + { + "epoch": 1.2787177203918076, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8020116686820984, + "num_tokens": 383617861.0, + "step": 10052 + }, + { + "epoch": 1.2788449306703982, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7903703451156616, + "num_tokens": 383652534.0, + "step": 10053 + }, + { + "epoch": 1.2789721409489887, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7994048595428467, + "num_tokens": 383693120.0, + "step": 10054 + }, + { + "epoch": 1.2790993512275792, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7921385765075684, + "num_tokens": 383734958.0, + "step": 10055 + }, + { + "epoch": 1.2792265615061698, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.789936900138855, + "num_tokens": 383768873.0, + "step": 10056 + }, + { + "epoch": 1.2793537717847603, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8022619485855103, + "num_tokens": 383804063.0, + "step": 10057 + }, + { + "epoch": 1.2794809820633506, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8113573789596558, + "num_tokens": 383843098.0, + "step": 10058 + }, + { + "epoch": 1.2796081923419411, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8082290291786194, + "num_tokens": 383879443.0, + "step": 10059 + }, + { + "epoch": 1.2797354026205316, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7941600680351257, + "num_tokens": 383918071.0, + "step": 10060 + }, + { + "epoch": 1.2798626128991222, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7880915999412537, + "num_tokens": 383953052.0, + "step": 10061 + }, + { + "epoch": 1.2799898231777127, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8033055663108826, + "num_tokens": 383992835.0, + "step": 10062 + }, + { + "epoch": 1.2801170334563032, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7834540605545044, + "num_tokens": 384031234.0, + "step": 10063 + }, + { + "epoch": 1.2802442437348938, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.784202516078949, + "num_tokens": 384063611.0, + "step": 10064 + }, + { + "epoch": 1.2803714540134843, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8123884201049805, + "num_tokens": 384106757.0, + "step": 10065 + }, + { + "epoch": 1.2804986642920748, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8012599945068359, + "num_tokens": 384143094.0, + "step": 10066 + }, + { + "epoch": 1.2806258745706653, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7973902225494385, + "num_tokens": 384181498.0, + "step": 10067 + }, + { + "epoch": 1.2807530848492559, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7958307266235352, + "num_tokens": 384209784.0, + "step": 10068 + }, + { + "epoch": 1.2808802951278464, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8009884357452393, + "num_tokens": 384248000.0, + "step": 10069 + }, + { + "epoch": 1.281007505406437, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7918310165405273, + "num_tokens": 384290897.0, + "step": 10070 + }, + { + "epoch": 1.2811347156850275, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7888269424438477, + "num_tokens": 384329464.0, + "step": 10071 + }, + { + "epoch": 1.281261925963618, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8062179684638977, + "num_tokens": 384368915.0, + "step": 10072 + }, + { + "epoch": 1.2813891362422083, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8083166480064392, + "num_tokens": 384407445.0, + "step": 10073 + }, + { + "epoch": 1.2815163465207988, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.792426586151123, + "num_tokens": 384444384.0, + "step": 10074 + }, + { + "epoch": 1.2816435567993893, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7957320809364319, + "num_tokens": 384479670.0, + "step": 10075 + }, + { + "epoch": 1.2817707670779799, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7909682393074036, + "num_tokens": 384522436.0, + "step": 10076 + }, + { + "epoch": 1.2818979773565704, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7872751951217651, + "num_tokens": 384559754.0, + "step": 10077 + }, + { + "epoch": 1.282025187635161, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7951453924179077, + "num_tokens": 384595153.0, + "step": 10078 + }, + { + "epoch": 1.2821523979137515, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7984715700149536, + "num_tokens": 384628345.0, + "step": 10079 + }, + { + "epoch": 1.282279608192342, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7891455888748169, + "num_tokens": 384666725.0, + "step": 10080 + }, + { + "epoch": 1.2824068184709325, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7891470193862915, + "num_tokens": 384706217.0, + "step": 10081 + }, + { + "epoch": 1.2825340287495228, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7941476106643677, + "num_tokens": 384741835.0, + "step": 10082 + }, + { + "epoch": 1.2826612390281134, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7982675433158875, + "num_tokens": 384777273.0, + "step": 10083 + }, + { + "epoch": 1.2827884493067039, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8110764026641846, + "num_tokens": 384814950.0, + "step": 10084 + }, + { + "epoch": 1.2829156595852944, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7837497591972351, + "num_tokens": 384848088.0, + "step": 10085 + }, + { + "epoch": 1.283042869863885, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.791265606880188, + "num_tokens": 384887361.0, + "step": 10086 + }, + { + "epoch": 1.2831700801424755, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8063338398933411, + "num_tokens": 384930459.0, + "step": 10087 + }, + { + "epoch": 1.283297290421066, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.797050952911377, + "num_tokens": 384969506.0, + "step": 10088 + }, + { + "epoch": 1.2834245006996565, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8035485744476318, + "num_tokens": 385004368.0, + "step": 10089 + }, + { + "epoch": 1.283551710978247, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8017730116844177, + "num_tokens": 385044553.0, + "step": 10090 + }, + { + "epoch": 1.2836789212568376, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8060213327407837, + "num_tokens": 385076873.0, + "step": 10091 + }, + { + "epoch": 1.283806131535428, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8007051944732666, + "num_tokens": 385113969.0, + "step": 10092 + }, + { + "epoch": 1.2839333418140186, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8034012317657471, + "num_tokens": 385149460.0, + "step": 10093 + }, + { + "epoch": 1.2840605520926092, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7817298173904419, + "num_tokens": 385187240.0, + "step": 10094 + }, + { + "epoch": 1.2841877623711997, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.791793942451477, + "num_tokens": 385225532.0, + "step": 10095 + }, + { + "epoch": 1.2843149726497902, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8025423288345337, + "num_tokens": 385267404.0, + "step": 10096 + }, + { + "epoch": 1.2844421829283807, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7893714308738708, + "num_tokens": 385314008.0, + "step": 10097 + }, + { + "epoch": 1.284569393206971, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7891613245010376, + "num_tokens": 385352204.0, + "step": 10098 + }, + { + "epoch": 1.2846966034855616, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7918189167976379, + "num_tokens": 385393307.0, + "step": 10099 + }, + { + "epoch": 1.2848238137641521, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7981526255607605, + "num_tokens": 385426124.0, + "step": 10100 + }, + { + "epoch": 1.2849510240427426, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.806243360042572, + "num_tokens": 385467471.0, + "step": 10101 + }, + { + "epoch": 1.2850782343213332, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7918310165405273, + "num_tokens": 385508780.0, + "step": 10102 + }, + { + "epoch": 1.2852054445999237, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8047806620597839, + "num_tokens": 385552646.0, + "step": 10103 + }, + { + "epoch": 1.2853326548785142, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7842282056808472, + "num_tokens": 385586676.0, + "step": 10104 + }, + { + "epoch": 1.2854598651571048, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7849935293197632, + "num_tokens": 385623201.0, + "step": 10105 + }, + { + "epoch": 1.2855870754356953, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8082488775253296, + "num_tokens": 385662716.0, + "step": 10106 + }, + { + "epoch": 1.2857142857142856, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7889896631240845, + "num_tokens": 385706986.0, + "step": 10107 + }, + { + "epoch": 1.2858414959928761, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7927898168563843, + "num_tokens": 385744240.0, + "step": 10108 + }, + { + "epoch": 1.2859687062714666, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7782629132270813, + "num_tokens": 385778583.0, + "step": 10109 + }, + { + "epoch": 1.2860959165500572, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7969631552696228, + "num_tokens": 385819910.0, + "step": 10110 + }, + { + "epoch": 1.2862231268286477, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7913135290145874, + "num_tokens": 385860068.0, + "step": 10111 + }, + { + "epoch": 1.2863503371072382, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7832478284835815, + "num_tokens": 385890460.0, + "step": 10112 + }, + { + "epoch": 1.2864775473858288, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7921116352081299, + "num_tokens": 385926174.0, + "step": 10113 + }, + { + "epoch": 1.2866047576644193, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8061782121658325, + "num_tokens": 385962928.0, + "step": 10114 + }, + { + "epoch": 1.2867319679430098, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8111903071403503, + "num_tokens": 386002509.0, + "step": 10115 + }, + { + "epoch": 1.2868591782216003, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8033249378204346, + "num_tokens": 386043304.0, + "step": 10116 + }, + { + "epoch": 1.2869863885001909, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8071740865707397, + "num_tokens": 386081607.0, + "step": 10117 + }, + { + "epoch": 1.2871135987787814, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8152170181274414, + "num_tokens": 386117762.0, + "step": 10118 + }, + { + "epoch": 1.287240809057372, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7888956069946289, + "num_tokens": 386155908.0, + "step": 10119 + }, + { + "epoch": 1.2873680193359625, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7969615459442139, + "num_tokens": 386199385.0, + "step": 10120 + }, + { + "epoch": 1.287495229614553, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7920299172401428, + "num_tokens": 386231892.0, + "step": 10121 + }, + { + "epoch": 1.2876224398931433, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7796573638916016, + "num_tokens": 386269657.0, + "step": 10122 + }, + { + "epoch": 1.2877496501717338, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8061742186546326, + "num_tokens": 386306936.0, + "step": 10123 + }, + { + "epoch": 1.2878768604503243, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.799460768699646, + "num_tokens": 386342260.0, + "step": 10124 + }, + { + "epoch": 1.2880040707289149, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7914514541625977, + "num_tokens": 386381784.0, + "step": 10125 + }, + { + "epoch": 1.2881312810075054, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7990010976791382, + "num_tokens": 386422350.0, + "step": 10126 + }, + { + "epoch": 1.288258491286096, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7945179343223572, + "num_tokens": 386467951.0, + "step": 10127 + }, + { + "epoch": 1.2883857015646865, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8090916872024536, + "num_tokens": 386511896.0, + "step": 10128 + }, + { + "epoch": 1.288512911843277, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7855443954467773, + "num_tokens": 386550704.0, + "step": 10129 + }, + { + "epoch": 1.2886401221218675, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7841446995735168, + "num_tokens": 386589295.0, + "step": 10130 + }, + { + "epoch": 1.2887673324004578, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7980958223342896, + "num_tokens": 386623462.0, + "step": 10131 + }, + { + "epoch": 1.2888945426790483, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7983893156051636, + "num_tokens": 386663036.0, + "step": 10132 + }, + { + "epoch": 1.2890217529576389, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7986822128295898, + "num_tokens": 386702530.0, + "step": 10133 + }, + { + "epoch": 1.2891489632362294, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8137174248695374, + "num_tokens": 386739758.0, + "step": 10134 + }, + { + "epoch": 1.28927617351482, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8075790405273438, + "num_tokens": 386778287.0, + "step": 10135 + }, + { + "epoch": 1.2894033837934105, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8047194480895996, + "num_tokens": 386812348.0, + "step": 10136 + }, + { + "epoch": 1.289530594072001, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8058093786239624, + "num_tokens": 386855540.0, + "step": 10137 + }, + { + "epoch": 1.2896578043505915, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7965395450592041, + "num_tokens": 386893775.0, + "step": 10138 + }, + { + "epoch": 1.289785014629182, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8012845516204834, + "num_tokens": 386929553.0, + "step": 10139 + }, + { + "epoch": 1.2899122249077726, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7905174493789673, + "num_tokens": 386968033.0, + "step": 10140 + }, + { + "epoch": 1.290039435186363, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7974329590797424, + "num_tokens": 387004967.0, + "step": 10141 + }, + { + "epoch": 1.2901666454649536, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7863414287567139, + "num_tokens": 387038086.0, + "step": 10142 + }, + { + "epoch": 1.2902938557435442, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7737718224525452, + "num_tokens": 387077112.0, + "step": 10143 + }, + { + "epoch": 1.2904210660221347, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7761679291725159, + "num_tokens": 387118800.0, + "step": 10144 + }, + { + "epoch": 1.2905482763007252, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.792695164680481, + "num_tokens": 387160321.0, + "step": 10145 + }, + { + "epoch": 1.2906754865793157, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7953454256057739, + "num_tokens": 387199387.0, + "step": 10146 + }, + { + "epoch": 1.290802696857906, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8070657849311829, + "num_tokens": 387238859.0, + "step": 10147 + }, + { + "epoch": 1.2909299071364966, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7998321056365967, + "num_tokens": 387272149.0, + "step": 10148 + }, + { + "epoch": 1.291057117415087, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7974604964256287, + "num_tokens": 387311617.0, + "step": 10149 + }, + { + "epoch": 1.2911843276936776, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7796695232391357, + "num_tokens": 387344281.0, + "step": 10150 + }, + { + "epoch": 1.2913115379722682, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7882275581359863, + "num_tokens": 387385959.0, + "step": 10151 + }, + { + "epoch": 1.2914387482508587, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7933580875396729, + "num_tokens": 387426197.0, + "step": 10152 + }, + { + "epoch": 1.2915659585294492, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8022044897079468, + "num_tokens": 387462250.0, + "step": 10153 + }, + { + "epoch": 1.2916931688080397, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.802366316318512, + "num_tokens": 387499903.0, + "step": 10154 + }, + { + "epoch": 1.2918203790866303, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8093703985214233, + "num_tokens": 387535361.0, + "step": 10155 + }, + { + "epoch": 1.2919475893652206, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7975998520851135, + "num_tokens": 387574300.0, + "step": 10156 + }, + { + "epoch": 1.2920747996438111, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7984541058540344, + "num_tokens": 387606860.0, + "step": 10157 + }, + { + "epoch": 1.2922020099224016, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7875890135765076, + "num_tokens": 387646519.0, + "step": 10158 + }, + { + "epoch": 1.2923292202009922, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8078505396842957, + "num_tokens": 387681100.0, + "step": 10159 + }, + { + "epoch": 1.2924564304795827, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8228684663772583, + "num_tokens": 387715962.0, + "step": 10160 + }, + { + "epoch": 1.2925836407581732, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.80024254322052, + "num_tokens": 387755856.0, + "step": 10161 + }, + { + "epoch": 1.2927108510367638, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.804210364818573, + "num_tokens": 387793256.0, + "step": 10162 + }, + { + "epoch": 1.2928380613153543, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7898606657981873, + "num_tokens": 387834696.0, + "step": 10163 + }, + { + "epoch": 1.2929652715939448, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7900352478027344, + "num_tokens": 387874973.0, + "step": 10164 + }, + { + "epoch": 1.2930924818725353, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7844749093055725, + "num_tokens": 387904887.0, + "step": 10165 + }, + { + "epoch": 1.2932196921511259, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8027138710021973, + "num_tokens": 387944676.0, + "step": 10166 + }, + { + "epoch": 1.2933469024297164, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.796576738357544, + "num_tokens": 387982234.0, + "step": 10167 + }, + { + "epoch": 1.293474112708307, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7991642355918884, + "num_tokens": 388018001.0, + "step": 10168 + }, + { + "epoch": 1.2936013229868975, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.791257917881012, + "num_tokens": 388057476.0, + "step": 10169 + }, + { + "epoch": 1.293728533265488, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8136608600616455, + "num_tokens": 388095351.0, + "step": 10170 + }, + { + "epoch": 1.2938557435440783, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7986596822738647, + "num_tokens": 388135366.0, + "step": 10171 + }, + { + "epoch": 1.2939829538226688, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7829256057739258, + "num_tokens": 388175003.0, + "step": 10172 + }, + { + "epoch": 1.2941101641012593, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7754203081130981, + "num_tokens": 388207083.0, + "step": 10173 + }, + { + "epoch": 1.2942373743798499, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8159531354904175, + "num_tokens": 388245038.0, + "step": 10174 + }, + { + "epoch": 1.2943645846584404, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7978127598762512, + "num_tokens": 388285035.0, + "step": 10175 + }, + { + "epoch": 1.294491794937031, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7992775440216064, + "num_tokens": 388323971.0, + "step": 10176 + }, + { + "epoch": 1.2946190052156215, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7916586399078369, + "num_tokens": 388362256.0, + "step": 10177 + }, + { + "epoch": 1.294746215494212, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8197290897369385, + "num_tokens": 388401120.0, + "step": 10178 + }, + { + "epoch": 1.2948734257728025, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8057559728622437, + "num_tokens": 388446833.0, + "step": 10179 + }, + { + "epoch": 1.2950006360513928, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.812691330909729, + "num_tokens": 388482266.0, + "step": 10180 + }, + { + "epoch": 1.2951278463299833, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7948609590530396, + "num_tokens": 388518024.0, + "step": 10181 + }, + { + "epoch": 1.2952550566085739, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8023624420166016, + "num_tokens": 388564246.0, + "step": 10182 + }, + { + "epoch": 1.2953822668871644, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8042296767234802, + "num_tokens": 388603556.0, + "step": 10183 + }, + { + "epoch": 1.295509477165755, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8117097616195679, + "num_tokens": 388643615.0, + "step": 10184 + }, + { + "epoch": 1.2956366874443455, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8037109375, + "num_tokens": 388682917.0, + "step": 10185 + }, + { + "epoch": 1.295763897722936, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8029502630233765, + "num_tokens": 388720269.0, + "step": 10186 + }, + { + "epoch": 1.2958911080015265, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7941790819168091, + "num_tokens": 388759245.0, + "step": 10187 + }, + { + "epoch": 1.296018318280117, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7995265126228333, + "num_tokens": 388795673.0, + "step": 10188 + }, + { + "epoch": 1.2961455285587076, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8064135909080505, + "num_tokens": 388832926.0, + "step": 10189 + }, + { + "epoch": 1.296272738837298, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8050172924995422, + "num_tokens": 388869767.0, + "step": 10190 + }, + { + "epoch": 1.2963999491158886, + "grad_norm": 10.125, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8030235767364502, + "num_tokens": 388913347.0, + "step": 10191 + }, + { + "epoch": 1.2965271593944792, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.808928370475769, + "num_tokens": 388948211.0, + "step": 10192 + }, + { + "epoch": 1.2966543696730697, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8089089393615723, + "num_tokens": 388989198.0, + "step": 10193 + }, + { + "epoch": 1.2967815799516602, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7908898591995239, + "num_tokens": 389029076.0, + "step": 10194 + }, + { + "epoch": 1.2969087902302507, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8020965456962585, + "num_tokens": 389066751.0, + "step": 10195 + }, + { + "epoch": 1.297036000508841, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7990436553955078, + "num_tokens": 389104207.0, + "step": 10196 + }, + { + "epoch": 1.2971632107874316, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8062360882759094, + "num_tokens": 389141941.0, + "step": 10197 + }, + { + "epoch": 1.297290421066022, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8092697262763977, + "num_tokens": 389185326.0, + "step": 10198 + }, + { + "epoch": 1.2974176313446126, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8068510293960571, + "num_tokens": 389221512.0, + "step": 10199 + }, + { + "epoch": 1.2975448416232032, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8084814548492432, + "num_tokens": 389259388.0, + "step": 10200 + }, + { + "epoch": 1.2976720519017937, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7910028100013733, + "num_tokens": 389298773.0, + "step": 10201 + }, + { + "epoch": 1.2977992621803842, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8005815744400024, + "num_tokens": 389339680.0, + "step": 10202 + }, + { + "epoch": 1.2979264724589747, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8104392290115356, + "num_tokens": 389380755.0, + "step": 10203 + }, + { + "epoch": 1.2980536827375653, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7962157726287842, + "num_tokens": 389412031.0, + "step": 10204 + }, + { + "epoch": 1.2981808930161556, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8129160404205322, + "num_tokens": 389448299.0, + "step": 10205 + }, + { + "epoch": 1.298308103294746, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7911206483840942, + "num_tokens": 389483962.0, + "step": 10206 + }, + { + "epoch": 1.2984353135733366, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7898776531219482, + "num_tokens": 389521989.0, + "step": 10207 + }, + { + "epoch": 1.2985625238519272, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7940410375595093, + "num_tokens": 389558228.0, + "step": 10208 + }, + { + "epoch": 1.2986897341305177, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7980226278305054, + "num_tokens": 389597164.0, + "step": 10209 + }, + { + "epoch": 1.2988169444091082, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7991600632667542, + "num_tokens": 389636435.0, + "step": 10210 + }, + { + "epoch": 1.2989441546876987, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.785271167755127, + "num_tokens": 389670498.0, + "step": 10211 + }, + { + "epoch": 1.2990713649662893, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8022754192352295, + "num_tokens": 389707433.0, + "step": 10212 + }, + { + "epoch": 1.2991985752448798, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8040245771408081, + "num_tokens": 389747300.0, + "step": 10213 + }, + { + "epoch": 1.2993257855234703, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7951147556304932, + "num_tokens": 389788381.0, + "step": 10214 + }, + { + "epoch": 1.2994529958020609, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8008956909179688, + "num_tokens": 389821792.0, + "step": 10215 + }, + { + "epoch": 1.2995802060806514, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8190778493881226, + "num_tokens": 389859370.0, + "step": 10216 + }, + { + "epoch": 1.299707416359242, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7871261239051819, + "num_tokens": 389897020.0, + "step": 10217 + }, + { + "epoch": 1.2998346266378324, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7723078727722168, + "num_tokens": 389927107.0, + "step": 10218 + }, + { + "epoch": 1.299961836916423, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7922910451889038, + "num_tokens": 389966323.0, + "step": 10219 + }, + { + "epoch": 1.3000890471950133, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7962023019790649, + "num_tokens": 390001416.0, + "step": 10220 + }, + { + "epoch": 1.3002162574736038, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8062402009963989, + "num_tokens": 390042326.0, + "step": 10221 + }, + { + "epoch": 1.3003434677521943, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7854714393615723, + "num_tokens": 390084443.0, + "step": 10222 + }, + { + "epoch": 1.3004706780307849, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.796425461769104, + "num_tokens": 390126309.0, + "step": 10223 + }, + { + "epoch": 1.3005978883093754, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8034974336624146, + "num_tokens": 390161463.0, + "step": 10224 + }, + { + "epoch": 1.300725098587966, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8137004375457764, + "num_tokens": 390203221.0, + "step": 10225 + }, + { + "epoch": 1.3008523088665565, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8149974346160889, + "num_tokens": 390243269.0, + "step": 10226 + }, + { + "epoch": 1.300979519145147, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.798701822757721, + "num_tokens": 390282488.0, + "step": 10227 + }, + { + "epoch": 1.3011067294237375, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7924925088882446, + "num_tokens": 390320268.0, + "step": 10228 + }, + { + "epoch": 1.3012339397023278, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.782991349697113, + "num_tokens": 390354733.0, + "step": 10229 + }, + { + "epoch": 1.3013611499809183, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7954564094543457, + "num_tokens": 390388161.0, + "step": 10230 + }, + { + "epoch": 1.3014883602595089, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8104919791221619, + "num_tokens": 390424609.0, + "step": 10231 + }, + { + "epoch": 1.3016155705380994, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8146554827690125, + "num_tokens": 390460809.0, + "step": 10232 + }, + { + "epoch": 1.30174278081669, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7918316721916199, + "num_tokens": 390500198.0, + "step": 10233 + }, + { + "epoch": 1.3018699910952805, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7928080558776855, + "num_tokens": 390545588.0, + "step": 10234 + }, + { + "epoch": 1.301997201373871, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7912455797195435, + "num_tokens": 390581681.0, + "step": 10235 + }, + { + "epoch": 1.3021244116524615, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8212801218032837, + "num_tokens": 390619122.0, + "step": 10236 + }, + { + "epoch": 1.302251621931052, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7885152101516724, + "num_tokens": 390667974.0, + "step": 10237 + }, + { + "epoch": 1.3023788322096426, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8040552139282227, + "num_tokens": 390705188.0, + "step": 10238 + }, + { + "epoch": 1.302506042488233, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8094221353530884, + "num_tokens": 390746892.0, + "step": 10239 + }, + { + "epoch": 1.3026332527668236, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.8138084411621094, + "num_tokens": 390787727.0, + "step": 10240 + }, + { + "epoch": 1.3027604630454142, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7927385568618774, + "num_tokens": 390823702.0, + "step": 10241 + }, + { + "epoch": 1.3028876733240047, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7925313115119934, + "num_tokens": 390860256.0, + "step": 10242 + }, + { + "epoch": 1.3030148836025952, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8176471590995789, + "num_tokens": 390901676.0, + "step": 10243 + }, + { + "epoch": 1.3031420938811857, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7741603255271912, + "num_tokens": 390936911.0, + "step": 10244 + }, + { + "epoch": 1.303269304159776, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7870819568634033, + "num_tokens": 390976185.0, + "step": 10245 + }, + { + "epoch": 1.3033965144383666, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8052329421043396, + "num_tokens": 391014661.0, + "step": 10246 + }, + { + "epoch": 1.303523724716957, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.791799783706665, + "num_tokens": 391056205.0, + "step": 10247 + }, + { + "epoch": 1.3036509349955476, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7975040078163147, + "num_tokens": 391092187.0, + "step": 10248 + }, + { + "epoch": 1.3037781452741382, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8135124444961548, + "num_tokens": 391130890.0, + "step": 10249 + }, + { + "epoch": 1.3039053555527287, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8091673851013184, + "num_tokens": 391172783.0, + "step": 10250 + }, + { + "epoch": 1.3040325658313192, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7803806066513062, + "num_tokens": 391215031.0, + "step": 10251 + }, + { + "epoch": 1.3041597761099097, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8036645650863647, + "num_tokens": 391251400.0, + "step": 10252 + }, + { + "epoch": 1.3042869863885003, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7862681150436401, + "num_tokens": 391285895.0, + "step": 10253 + }, + { + "epoch": 1.3044141966670906, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8006184101104736, + "num_tokens": 391328361.0, + "step": 10254 + }, + { + "epoch": 1.304541406945681, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7958948016166687, + "num_tokens": 391365019.0, + "step": 10255 + }, + { + "epoch": 1.3046686172242716, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7858452796936035, + "num_tokens": 391400407.0, + "step": 10256 + }, + { + "epoch": 1.3047958275028622, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7780974507331848, + "num_tokens": 391440660.0, + "step": 10257 + }, + { + "epoch": 1.3049230377814527, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8035553097724915, + "num_tokens": 391481096.0, + "step": 10258 + }, + { + "epoch": 1.3050502480600432, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7990261316299438, + "num_tokens": 391516356.0, + "step": 10259 + }, + { + "epoch": 1.3051774583386337, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.786777138710022, + "num_tokens": 391554691.0, + "step": 10260 + }, + { + "epoch": 1.3053046686172243, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7844682931900024, + "num_tokens": 391592473.0, + "step": 10261 + }, + { + "epoch": 1.3054318788958148, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8065622448921204, + "num_tokens": 391632091.0, + "step": 10262 + }, + { + "epoch": 1.3055590891744053, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7926519513130188, + "num_tokens": 391676121.0, + "step": 10263 + }, + { + "epoch": 1.3056862994529959, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7893978357315063, + "num_tokens": 391709664.0, + "step": 10264 + }, + { + "epoch": 1.3058135097315864, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7873988747596741, + "num_tokens": 391755828.0, + "step": 10265 + }, + { + "epoch": 1.305940720010177, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8223266005516052, + "num_tokens": 391800232.0, + "step": 10266 + }, + { + "epoch": 1.3060679302887674, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8026936650276184, + "num_tokens": 391837612.0, + "step": 10267 + }, + { + "epoch": 1.306195140567358, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.806279718875885, + "num_tokens": 391871193.0, + "step": 10268 + }, + { + "epoch": 1.3063223508459483, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7880076169967651, + "num_tokens": 391908203.0, + "step": 10269 + }, + { + "epoch": 1.3064495611245388, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7978816032409668, + "num_tokens": 391938744.0, + "step": 10270 + }, + { + "epoch": 1.3065767714031293, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7991031408309937, + "num_tokens": 391976020.0, + "step": 10271 + }, + { + "epoch": 1.3067039816817199, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7913869023323059, + "num_tokens": 392018006.0, + "step": 10272 + }, + { + "epoch": 1.3068311919603104, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7972613573074341, + "num_tokens": 392055358.0, + "step": 10273 + }, + { + "epoch": 1.306958402238901, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7998142242431641, + "num_tokens": 392097727.0, + "step": 10274 + }, + { + "epoch": 1.3070856125174914, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.805492103099823, + "num_tokens": 392144641.0, + "step": 10275 + }, + { + "epoch": 1.307212822796082, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8065124154090881, + "num_tokens": 392178810.0, + "step": 10276 + }, + { + "epoch": 1.3073400330746725, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7869453430175781, + "num_tokens": 392213571.0, + "step": 10277 + }, + { + "epoch": 1.3074672433532628, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7956860065460205, + "num_tokens": 392254628.0, + "step": 10278 + }, + { + "epoch": 1.3075944536318533, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7985950708389282, + "num_tokens": 392297243.0, + "step": 10279 + }, + { + "epoch": 1.3077216639104439, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7970473170280457, + "num_tokens": 392329116.0, + "step": 10280 + }, + { + "epoch": 1.3078488741890344, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7964380383491516, + "num_tokens": 392370542.0, + "step": 10281 + }, + { + "epoch": 1.307976084467625, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8020904660224915, + "num_tokens": 392411155.0, + "step": 10282 + }, + { + "epoch": 1.3081032947462155, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8019987344741821, + "num_tokens": 392450440.0, + "step": 10283 + }, + { + "epoch": 1.308230505024806, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8111340999603271, + "num_tokens": 392488545.0, + "step": 10284 + }, + { + "epoch": 1.3083577153033965, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7998762130737305, + "num_tokens": 392527138.0, + "step": 10285 + }, + { + "epoch": 1.308484925581987, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7959886789321899, + "num_tokens": 392568181.0, + "step": 10286 + }, + { + "epoch": 1.3086121358605776, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8153232336044312, + "num_tokens": 392610900.0, + "step": 10287 + }, + { + "epoch": 1.308739346139168, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7916533946990967, + "num_tokens": 392646624.0, + "step": 10288 + }, + { + "epoch": 1.3088665564177586, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7981367111206055, + "num_tokens": 392683103.0, + "step": 10289 + }, + { + "epoch": 1.3089937666963491, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7986038327217102, + "num_tokens": 392723169.0, + "step": 10290 + }, + { + "epoch": 1.3091209769749397, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8133496046066284, + "num_tokens": 392762323.0, + "step": 10291 + }, + { + "epoch": 1.3092481872535302, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.8182553648948669, + "num_tokens": 392797555.0, + "step": 10292 + }, + { + "epoch": 1.3093753975321207, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8122941851615906, + "num_tokens": 392840001.0, + "step": 10293 + }, + { + "epoch": 1.309502607810711, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.799321174621582, + "num_tokens": 392880871.0, + "step": 10294 + }, + { + "epoch": 1.3096298180893016, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8073873519897461, + "num_tokens": 392917554.0, + "step": 10295 + }, + { + "epoch": 1.309757028367892, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.585, + "mean_token_accuracy": 0.8208446502685547, + "num_tokens": 392958601.0, + "step": 10296 + }, + { + "epoch": 1.3098842386464826, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7993752956390381, + "num_tokens": 393000108.0, + "step": 10297 + }, + { + "epoch": 1.3100114489250732, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8136872053146362, + "num_tokens": 393032321.0, + "step": 10298 + }, + { + "epoch": 1.3101386592036637, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7939980626106262, + "num_tokens": 393068347.0, + "step": 10299 + }, + { + "epoch": 1.3102658694822542, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.812844455242157, + "num_tokens": 393102509.0, + "step": 10300 + }, + { + "epoch": 1.3103930797608447, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8115940690040588, + "num_tokens": 393140974.0, + "step": 10301 + }, + { + "epoch": 1.3105202900394353, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8131591081619263, + "num_tokens": 393177214.0, + "step": 10302 + }, + { + "epoch": 1.3106475003180256, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8004281520843506, + "num_tokens": 393215801.0, + "step": 10303 + }, + { + "epoch": 1.310774710596616, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8105270862579346, + "num_tokens": 393256444.0, + "step": 10304 + }, + { + "epoch": 1.3109019208752066, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7952525019645691, + "num_tokens": 393300238.0, + "step": 10305 + }, + { + "epoch": 1.3110291311537972, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7870817184448242, + "num_tokens": 393344087.0, + "step": 10306 + }, + { + "epoch": 1.3111563414323877, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8063457012176514, + "num_tokens": 393381678.0, + "step": 10307 + }, + { + "epoch": 1.3112835517109782, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7932436466217041, + "num_tokens": 393415740.0, + "step": 10308 + }, + { + "epoch": 1.3114107619895687, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7765165567398071, + "num_tokens": 393457692.0, + "step": 10309 + }, + { + "epoch": 1.3115379722681593, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.796859085559845, + "num_tokens": 393499003.0, + "step": 10310 + }, + { + "epoch": 1.3116651825467498, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7864854335784912, + "num_tokens": 393533842.0, + "step": 10311 + }, + { + "epoch": 1.3117923928253403, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8070824146270752, + "num_tokens": 393573906.0, + "step": 10312 + }, + { + "epoch": 1.3119196031039309, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8005072474479675, + "num_tokens": 393615298.0, + "step": 10313 + }, + { + "epoch": 1.3120468133825214, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7879939675331116, + "num_tokens": 393654690.0, + "step": 10314 + }, + { + "epoch": 1.312174023661112, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8012991547584534, + "num_tokens": 393696222.0, + "step": 10315 + }, + { + "epoch": 1.3123012339397024, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7981528639793396, + "num_tokens": 393736748.0, + "step": 10316 + }, + { + "epoch": 1.312428444218293, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7898390889167786, + "num_tokens": 393773077.0, + "step": 10317 + }, + { + "epoch": 1.3125556544968833, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8049018383026123, + "num_tokens": 393806466.0, + "step": 10318 + }, + { + "epoch": 1.3126828647754738, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7993096113204956, + "num_tokens": 393840922.0, + "step": 10319 + }, + { + "epoch": 1.3128100750540643, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8035622835159302, + "num_tokens": 393874544.0, + "step": 10320 + }, + { + "epoch": 1.3129372853326549, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7945302128791809, + "num_tokens": 393917773.0, + "step": 10321 + }, + { + "epoch": 1.3130644956112454, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.8011717796325684, + "num_tokens": 393951525.0, + "step": 10322 + }, + { + "epoch": 1.313191705889836, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7997832298278809, + "num_tokens": 393989796.0, + "step": 10323 + }, + { + "epoch": 1.3133189161684264, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7954003214836121, + "num_tokens": 394028746.0, + "step": 10324 + }, + { + "epoch": 1.313446126447017, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8120042085647583, + "num_tokens": 394065241.0, + "step": 10325 + }, + { + "epoch": 1.3135733367256075, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7945001125335693, + "num_tokens": 394110113.0, + "step": 10326 + }, + { + "epoch": 1.3137005470041978, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7822470664978027, + "num_tokens": 394146577.0, + "step": 10327 + }, + { + "epoch": 1.3138277572827883, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7998379468917847, + "num_tokens": 394185392.0, + "step": 10328 + }, + { + "epoch": 1.3139549675613789, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7933914065361023, + "num_tokens": 394226301.0, + "step": 10329 + }, + { + "epoch": 1.3140821778399694, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7926667928695679, + "num_tokens": 394262654.0, + "step": 10330 + }, + { + "epoch": 1.31420938811856, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8059777021408081, + "num_tokens": 394299185.0, + "step": 10331 + }, + { + "epoch": 1.3143365983971504, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.79815673828125, + "num_tokens": 394331364.0, + "step": 10332 + }, + { + "epoch": 1.314463808675741, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7847552299499512, + "num_tokens": 394375479.0, + "step": 10333 + }, + { + "epoch": 1.3145910189543315, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8041443824768066, + "num_tokens": 394414913.0, + "step": 10334 + }, + { + "epoch": 1.314718229232922, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7971827983856201, + "num_tokens": 394450889.0, + "step": 10335 + }, + { + "epoch": 1.3148454395115126, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7861158847808838, + "num_tokens": 394495280.0, + "step": 10336 + }, + { + "epoch": 1.314972649790103, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8111392259597778, + "num_tokens": 394536811.0, + "step": 10337 + }, + { + "epoch": 1.3150998600686936, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7926830053329468, + "num_tokens": 394573353.0, + "step": 10338 + }, + { + "epoch": 1.3152270703472841, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7860534191131592, + "num_tokens": 394603517.0, + "step": 10339 + }, + { + "epoch": 1.3153542806258747, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8088861107826233, + "num_tokens": 394650547.0, + "step": 10340 + }, + { + "epoch": 1.3154814909044652, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.804868757724762, + "num_tokens": 394684460.0, + "step": 10341 + }, + { + "epoch": 1.3156087011830557, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.5805, + "mean_token_accuracy": 0.8254446387290955, + "num_tokens": 394733520.0, + "step": 10342 + }, + { + "epoch": 1.315735911461646, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7986975312232971, + "num_tokens": 394775498.0, + "step": 10343 + }, + { + "epoch": 1.3158631217402366, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7831867337226868, + "num_tokens": 394815138.0, + "step": 10344 + }, + { + "epoch": 1.315990332018827, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7864596843719482, + "num_tokens": 394850447.0, + "step": 10345 + }, + { + "epoch": 1.3161175422974176, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7910252213478088, + "num_tokens": 394891800.0, + "step": 10346 + }, + { + "epoch": 1.3162447525760081, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.805173397064209, + "num_tokens": 394930087.0, + "step": 10347 + }, + { + "epoch": 1.3163719628545987, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8028533458709717, + "num_tokens": 394964497.0, + "step": 10348 + }, + { + "epoch": 1.3164991731331892, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7803565263748169, + "num_tokens": 395003127.0, + "step": 10349 + }, + { + "epoch": 1.3166263834117797, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7889590263366699, + "num_tokens": 395040570.0, + "step": 10350 + }, + { + "epoch": 1.3167535936903703, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7918589115142822, + "num_tokens": 395075205.0, + "step": 10351 + }, + { + "epoch": 1.3168808039689606, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8011685609817505, + "num_tokens": 395117492.0, + "step": 10352 + }, + { + "epoch": 1.317008014247551, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8000104427337646, + "num_tokens": 395162449.0, + "step": 10353 + }, + { + "epoch": 1.3171352245261416, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8010338544845581, + "num_tokens": 395197310.0, + "step": 10354 + }, + { + "epoch": 1.3172624348047322, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8102793097496033, + "num_tokens": 395232420.0, + "step": 10355 + }, + { + "epoch": 1.3173896450833227, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7724623680114746, + "num_tokens": 395269392.0, + "step": 10356 + }, + { + "epoch": 1.3175168553619132, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7842006683349609, + "num_tokens": 395305496.0, + "step": 10357 + }, + { + "epoch": 1.3176440656405037, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8109374046325684, + "num_tokens": 395347089.0, + "step": 10358 + }, + { + "epoch": 1.3177712759190943, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8095344305038452, + "num_tokens": 395382885.0, + "step": 10359 + }, + { + "epoch": 1.3178984861976848, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8103306889533997, + "num_tokens": 395416260.0, + "step": 10360 + }, + { + "epoch": 1.3180256964762753, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7853631973266602, + "num_tokens": 395454179.0, + "step": 10361 + }, + { + "epoch": 1.3181529067548658, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7888196110725403, + "num_tokens": 395489173.0, + "step": 10362 + }, + { + "epoch": 1.3182801170334564, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7977175712585449, + "num_tokens": 395532057.0, + "step": 10363 + }, + { + "epoch": 1.318407327312047, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8079615831375122, + "num_tokens": 395570321.0, + "step": 10364 + }, + { + "epoch": 1.3185345375906374, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8243249654769897, + "num_tokens": 395609041.0, + "step": 10365 + }, + { + "epoch": 1.318661747869228, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7733103632926941, + "num_tokens": 395649961.0, + "step": 10366 + }, + { + "epoch": 1.3187889581478183, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7999897003173828, + "num_tokens": 395691882.0, + "step": 10367 + }, + { + "epoch": 1.3189161684264088, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.785224974155426, + "num_tokens": 395731496.0, + "step": 10368 + }, + { + "epoch": 1.3190433787049993, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.807021975517273, + "num_tokens": 395772618.0, + "step": 10369 + }, + { + "epoch": 1.3191705889835899, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7795782089233398, + "num_tokens": 395811907.0, + "step": 10370 + }, + { + "epoch": 1.3192977992621804, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7961090803146362, + "num_tokens": 395851186.0, + "step": 10371 + }, + { + "epoch": 1.319425009540771, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.796822190284729, + "num_tokens": 395893868.0, + "step": 10372 + }, + { + "epoch": 1.3195522198193614, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8096363544464111, + "num_tokens": 395930519.0, + "step": 10373 + }, + { + "epoch": 1.319679430097952, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8136283755302429, + "num_tokens": 395965692.0, + "step": 10374 + }, + { + "epoch": 1.3198066403765425, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7926459312438965, + "num_tokens": 396002316.0, + "step": 10375 + }, + { + "epoch": 1.3199338506551328, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8043506145477295, + "num_tokens": 396043392.0, + "step": 10376 + }, + { + "epoch": 1.3200610609337233, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7972749471664429, + "num_tokens": 396078603.0, + "step": 10377 + }, + { + "epoch": 1.3201882712123139, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.791028618812561, + "num_tokens": 396121132.0, + "step": 10378 + }, + { + "epoch": 1.3203154814909044, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8143889904022217, + "num_tokens": 396155730.0, + "step": 10379 + }, + { + "epoch": 1.320442691769495, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7829858064651489, + "num_tokens": 396190976.0, + "step": 10380 + }, + { + "epoch": 1.3205699020480854, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7968794107437134, + "num_tokens": 396229571.0, + "step": 10381 + }, + { + "epoch": 1.320697112326676, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7973113656044006, + "num_tokens": 396265586.0, + "step": 10382 + }, + { + "epoch": 1.3208243226052665, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.802831768989563, + "num_tokens": 396303810.0, + "step": 10383 + }, + { + "epoch": 1.320951532883857, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8189059495925903, + "num_tokens": 396345671.0, + "step": 10384 + }, + { + "epoch": 1.3210787431624476, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8149484992027283, + "num_tokens": 396388062.0, + "step": 10385 + }, + { + "epoch": 1.321205953441038, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8169227838516235, + "num_tokens": 396425189.0, + "step": 10386 + }, + { + "epoch": 1.3213331637196286, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7870715856552124, + "num_tokens": 396461861.0, + "step": 10387 + }, + { + "epoch": 1.3214603739982191, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8168125152587891, + "num_tokens": 396506190.0, + "step": 10388 + }, + { + "epoch": 1.3215875842768097, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8006445169448853, + "num_tokens": 396538900.0, + "step": 10389 + }, + { + "epoch": 1.3217147945554002, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8047623634338379, + "num_tokens": 396577029.0, + "step": 10390 + }, + { + "epoch": 1.3218420048339907, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7848321795463562, + "num_tokens": 396616917.0, + "step": 10391 + }, + { + "epoch": 1.321969215112581, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8078674077987671, + "num_tokens": 396651726.0, + "step": 10392 + }, + { + "epoch": 1.3220964253911716, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7867788076400757, + "num_tokens": 396693796.0, + "step": 10393 + }, + { + "epoch": 1.322223635669762, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7871078252792358, + "num_tokens": 396731947.0, + "step": 10394 + }, + { + "epoch": 1.3223508459483526, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8025183081626892, + "num_tokens": 396767062.0, + "step": 10395 + }, + { + "epoch": 1.3224780562269431, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8035420775413513, + "num_tokens": 396803918.0, + "step": 10396 + }, + { + "epoch": 1.3226052665055337, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7754533290863037, + "num_tokens": 396850314.0, + "step": 10397 + }, + { + "epoch": 1.3227324767841242, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7855333089828491, + "num_tokens": 396887920.0, + "step": 10398 + }, + { + "epoch": 1.3228596870627147, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7930952906608582, + "num_tokens": 396923369.0, + "step": 10399 + }, + { + "epoch": 1.3229868973413053, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8000918626785278, + "num_tokens": 396962338.0, + "step": 10400 + }, + { + "epoch": 1.3231141076198956, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8108248114585876, + "num_tokens": 396998275.0, + "step": 10401 + }, + { + "epoch": 1.323241317898486, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7975694537162781, + "num_tokens": 397040034.0, + "step": 10402 + }, + { + "epoch": 1.3233685281770766, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.823590874671936, + "num_tokens": 397080549.0, + "step": 10403 + }, + { + "epoch": 1.3234957384556671, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8121005296707153, + "num_tokens": 397120803.0, + "step": 10404 + }, + { + "epoch": 1.3236229487342577, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8030828237533569, + "num_tokens": 397160463.0, + "step": 10405 + }, + { + "epoch": 1.3237501590128482, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7860533595085144, + "num_tokens": 397205425.0, + "step": 10406 + }, + { + "epoch": 1.3238773692914387, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.79558265209198, + "num_tokens": 397238907.0, + "step": 10407 + }, + { + "epoch": 1.3240045795700293, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8035756945610046, + "num_tokens": 397281639.0, + "step": 10408 + }, + { + "epoch": 1.3241317898486198, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7883279323577881, + "num_tokens": 397323350.0, + "step": 10409 + }, + { + "epoch": 1.3242590001272103, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7919843196868896, + "num_tokens": 397364438.0, + "step": 10410 + }, + { + "epoch": 1.3243862104058008, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7827924489974976, + "num_tokens": 397406881.0, + "step": 10411 + }, + { + "epoch": 1.3245134206843914, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.805042028427124, + "num_tokens": 397451190.0, + "step": 10412 + }, + { + "epoch": 1.324640630962982, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7999250292778015, + "num_tokens": 397491063.0, + "step": 10413 + }, + { + "epoch": 1.3247678412415724, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7949425578117371, + "num_tokens": 397529961.0, + "step": 10414 + }, + { + "epoch": 1.324895051520163, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8095558881759644, + "num_tokens": 397565616.0, + "step": 10415 + }, + { + "epoch": 1.3250222617987533, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.805010974407196, + "num_tokens": 397602223.0, + "step": 10416 + }, + { + "epoch": 1.3251494720773438, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.776615560054779, + "num_tokens": 397638910.0, + "step": 10417 + }, + { + "epoch": 1.3252766823559343, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7941334247589111, + "num_tokens": 397681200.0, + "step": 10418 + }, + { + "epoch": 1.3254038926345248, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.8004500865936279, + "num_tokens": 397724832.0, + "step": 10419 + }, + { + "epoch": 1.3255311029131154, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7908259034156799, + "num_tokens": 397765537.0, + "step": 10420 + }, + { + "epoch": 1.325658313191706, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7913620471954346, + "num_tokens": 397800945.0, + "step": 10421 + }, + { + "epoch": 1.3257855234702964, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8040701150894165, + "num_tokens": 397840786.0, + "step": 10422 + }, + { + "epoch": 1.325912733748887, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8024929761886597, + "num_tokens": 397875607.0, + "step": 10423 + }, + { + "epoch": 1.3260399440274775, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8126086592674255, + "num_tokens": 397909524.0, + "step": 10424 + }, + { + "epoch": 1.3261671543060678, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7766168713569641, + "num_tokens": 397947862.0, + "step": 10425 + }, + { + "epoch": 1.3262943645846583, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8108090162277222, + "num_tokens": 397980958.0, + "step": 10426 + }, + { + "epoch": 1.3264215748632489, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7921859622001648, + "num_tokens": 398021231.0, + "step": 10427 + }, + { + "epoch": 1.3265487851418394, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7970250844955444, + "num_tokens": 398059202.0, + "step": 10428 + }, + { + "epoch": 1.32667599542043, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.822442889213562, + "num_tokens": 398097121.0, + "step": 10429 + }, + { + "epoch": 1.3268032056990204, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7981566190719604, + "num_tokens": 398130413.0, + "step": 10430 + }, + { + "epoch": 1.326930415977611, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7986252307891846, + "num_tokens": 398172535.0, + "step": 10431 + }, + { + "epoch": 1.3270576262562015, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.804177463054657, + "num_tokens": 398206681.0, + "step": 10432 + }, + { + "epoch": 1.327184836534792, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8023151159286499, + "num_tokens": 398245206.0, + "step": 10433 + }, + { + "epoch": 1.3273120468133826, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7902353405952454, + "num_tokens": 398278902.0, + "step": 10434 + }, + { + "epoch": 1.327439257091973, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8012493252754211, + "num_tokens": 398315833.0, + "step": 10435 + }, + { + "epoch": 1.3275664673705636, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.784809410572052, + "num_tokens": 398356496.0, + "step": 10436 + }, + { + "epoch": 1.3276936776491541, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8065007925033569, + "num_tokens": 398393661.0, + "step": 10437 + }, + { + "epoch": 1.3278208879277447, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8159507513046265, + "num_tokens": 398430463.0, + "step": 10438 + }, + { + "epoch": 1.3279480982063352, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7961334586143494, + "num_tokens": 398470382.0, + "step": 10439 + }, + { + "epoch": 1.3280753084849257, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.776091992855072, + "num_tokens": 398504793.0, + "step": 10440 + }, + { + "epoch": 1.328202518763516, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7893388271331787, + "num_tokens": 398541038.0, + "step": 10441 + }, + { + "epoch": 1.3283297290421066, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7811977863311768, + "num_tokens": 398578047.0, + "step": 10442 + }, + { + "epoch": 1.328456939320697, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7892893552780151, + "num_tokens": 398618946.0, + "step": 10443 + }, + { + "epoch": 1.3285841495992876, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8023003339767456, + "num_tokens": 398656895.0, + "step": 10444 + }, + { + "epoch": 1.3287113598778781, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8005262613296509, + "num_tokens": 398696521.0, + "step": 10445 + }, + { + "epoch": 1.3288385701564687, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7977249026298523, + "num_tokens": 398735275.0, + "step": 10446 + }, + { + "epoch": 1.3289657804350592, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7794946432113647, + "num_tokens": 398774753.0, + "step": 10447 + }, + { + "epoch": 1.3290929907136497, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7914965152740479, + "num_tokens": 398810893.0, + "step": 10448 + }, + { + "epoch": 1.3292202009922403, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7901133298873901, + "num_tokens": 398849741.0, + "step": 10449 + }, + { + "epoch": 1.3293474112708306, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7894294261932373, + "num_tokens": 398881186.0, + "step": 10450 + }, + { + "epoch": 1.329474621549421, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.598, + "mean_token_accuracy": 0.8197717070579529, + "num_tokens": 398918677.0, + "step": 10451 + }, + { + "epoch": 1.3296018318280116, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.794460654258728, + "num_tokens": 398953435.0, + "step": 10452 + }, + { + "epoch": 1.3297290421066021, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8097562789916992, + "num_tokens": 398991799.0, + "step": 10453 + }, + { + "epoch": 1.3298562523851927, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8002519607543945, + "num_tokens": 399023211.0, + "step": 10454 + }, + { + "epoch": 1.3299834626637832, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7926782369613647, + "num_tokens": 399061431.0, + "step": 10455 + }, + { + "epoch": 1.3301106729423737, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7956923246383667, + "num_tokens": 399097384.0, + "step": 10456 + }, + { + "epoch": 1.3302378832209643, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8031110763549805, + "num_tokens": 399133935.0, + "step": 10457 + }, + { + "epoch": 1.3303650934995548, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8015761375427246, + "num_tokens": 399165709.0, + "step": 10458 + }, + { + "epoch": 1.3304923037781453, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7935377359390259, + "num_tokens": 399203548.0, + "step": 10459 + }, + { + "epoch": 1.3306195140567358, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7988623380661011, + "num_tokens": 399244341.0, + "step": 10460 + }, + { + "epoch": 1.3307467243353264, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.803537905216217, + "num_tokens": 399282284.0, + "step": 10461 + }, + { + "epoch": 1.330873934613917, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8138227462768555, + "num_tokens": 399320523.0, + "step": 10462 + }, + { + "epoch": 1.3310011448925074, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7987592816352844, + "num_tokens": 399357489.0, + "step": 10463 + }, + { + "epoch": 1.331128355171098, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7829365730285645, + "num_tokens": 399393561.0, + "step": 10464 + }, + { + "epoch": 1.3312555654496883, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7956595420837402, + "num_tokens": 399424514.0, + "step": 10465 + }, + { + "epoch": 1.3313827757282788, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8063240647315979, + "num_tokens": 399461372.0, + "step": 10466 + }, + { + "epoch": 1.3315099860068693, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.811586856842041, + "num_tokens": 399502503.0, + "step": 10467 + }, + { + "epoch": 1.3316371962854598, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7917659282684326, + "num_tokens": 399537836.0, + "step": 10468 + }, + { + "epoch": 1.3317644065640504, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7926942110061646, + "num_tokens": 399575189.0, + "step": 10469 + }, + { + "epoch": 1.331891616842641, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7831217050552368, + "num_tokens": 399614406.0, + "step": 10470 + }, + { + "epoch": 1.3320188271212314, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8037764430046082, + "num_tokens": 399652847.0, + "step": 10471 + }, + { + "epoch": 1.332146037399822, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7883849740028381, + "num_tokens": 399691628.0, + "step": 10472 + }, + { + "epoch": 1.3322732476784125, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8166587352752686, + "num_tokens": 399732875.0, + "step": 10473 + }, + { + "epoch": 1.3324004579570028, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8051003217697144, + "num_tokens": 399769537.0, + "step": 10474 + }, + { + "epoch": 1.3325276682355933, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7803962230682373, + "num_tokens": 399807757.0, + "step": 10475 + }, + { + "epoch": 1.3326548785141838, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8011053800582886, + "num_tokens": 399837351.0, + "step": 10476 + }, + { + "epoch": 1.3327820887927744, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8184080123901367, + "num_tokens": 399873678.0, + "step": 10477 + }, + { + "epoch": 1.332909299071365, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8020703792572021, + "num_tokens": 399916206.0, + "step": 10478 + }, + { + "epoch": 1.3330365093499554, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.812114953994751, + "num_tokens": 399954144.0, + "step": 10479 + }, + { + "epoch": 1.333163719628546, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8093730211257935, + "num_tokens": 399995865.0, + "step": 10480 + }, + { + "epoch": 1.3332909299071365, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7961276173591614, + "num_tokens": 400033234.0, + "step": 10481 + }, + { + "epoch": 1.333418140185727, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8063507080078125, + "num_tokens": 400066842.0, + "step": 10482 + }, + { + "epoch": 1.3335453504643175, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8075957298278809, + "num_tokens": 400099380.0, + "step": 10483 + }, + { + "epoch": 1.333672560742908, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8100555539131165, + "num_tokens": 400137419.0, + "step": 10484 + }, + { + "epoch": 1.3337997710214986, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.803024172782898, + "num_tokens": 400171086.0, + "step": 10485 + }, + { + "epoch": 1.3339269813000891, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.793980598449707, + "num_tokens": 400206763.0, + "step": 10486 + }, + { + "epoch": 1.3340541915786797, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7815276980400085, + "num_tokens": 400242708.0, + "step": 10487 + }, + { + "epoch": 1.3341814018572702, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7721102237701416, + "num_tokens": 400277417.0, + "step": 10488 + }, + { + "epoch": 1.3343086121358605, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7953240871429443, + "num_tokens": 400315811.0, + "step": 10489 + }, + { + "epoch": 1.334435822414451, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7833682298660278, + "num_tokens": 400356321.0, + "step": 10490 + }, + { + "epoch": 1.3345630326930416, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7966191172599792, + "num_tokens": 400392382.0, + "step": 10491 + }, + { + "epoch": 1.334690242971632, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8100016117095947, + "num_tokens": 400434947.0, + "step": 10492 + }, + { + "epoch": 1.3348174532502226, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.778715193271637, + "num_tokens": 400472742.0, + "step": 10493 + }, + { + "epoch": 1.3349446635288131, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8194552659988403, + "num_tokens": 400510029.0, + "step": 10494 + }, + { + "epoch": 1.3350718738074037, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7993303537368774, + "num_tokens": 400547383.0, + "step": 10495 + }, + { + "epoch": 1.3351990840859942, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8110001087188721, + "num_tokens": 400585242.0, + "step": 10496 + }, + { + "epoch": 1.3353262943645847, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8151757717132568, + "num_tokens": 400623117.0, + "step": 10497 + }, + { + "epoch": 1.3354535046431752, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7939474582672119, + "num_tokens": 400660170.0, + "step": 10498 + }, + { + "epoch": 1.3355807149217656, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8039017915725708, + "num_tokens": 400700467.0, + "step": 10499 + }, + { + "epoch": 1.335707925200356, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7905310392379761, + "num_tokens": 400740269.0, + "step": 10500 + }, + { + "epoch": 1.3358351354789466, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8137504458427429, + "num_tokens": 400776465.0, + "step": 10501 + }, + { + "epoch": 1.3359623457575371, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.795371413230896, + "num_tokens": 400811210.0, + "step": 10502 + }, + { + "epoch": 1.3360895560361277, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7927790880203247, + "num_tokens": 400848820.0, + "step": 10503 + }, + { + "epoch": 1.3362167663147182, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.801099419593811, + "num_tokens": 400888774.0, + "step": 10504 + }, + { + "epoch": 1.3363439765933087, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8054202198982239, + "num_tokens": 400935067.0, + "step": 10505 + }, + { + "epoch": 1.3364711868718993, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8126230239868164, + "num_tokens": 400974122.0, + "step": 10506 + }, + { + "epoch": 1.3365983971504898, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.80183345079422, + "num_tokens": 401008911.0, + "step": 10507 + }, + { + "epoch": 1.3367256074290803, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.8019958734512329, + "num_tokens": 401039453.0, + "step": 10508 + }, + { + "epoch": 1.3368528177076708, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8107342720031738, + "num_tokens": 401075854.0, + "step": 10509 + }, + { + "epoch": 1.3369800279862614, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8000432252883911, + "num_tokens": 401114922.0, + "step": 10510 + }, + { + "epoch": 1.337107238264852, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8054229021072388, + "num_tokens": 401155833.0, + "step": 10511 + }, + { + "epoch": 1.3372344485434424, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8076770305633545, + "num_tokens": 401196971.0, + "step": 10512 + }, + { + "epoch": 1.337361658822033, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7662872076034546, + "num_tokens": 401237312.0, + "step": 10513 + }, + { + "epoch": 1.3374888691006233, + "grad_norm": 75.5, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7913536429405212, + "num_tokens": 401274630.0, + "step": 10514 + }, + { + "epoch": 1.3376160793792138, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8046505451202393, + "num_tokens": 401311780.0, + "step": 10515 + }, + { + "epoch": 1.3377432896578043, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7989522218704224, + "num_tokens": 401352614.0, + "step": 10516 + }, + { + "epoch": 1.3378704999363948, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8067014217376709, + "num_tokens": 401394495.0, + "step": 10517 + }, + { + "epoch": 1.3379977102149854, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7850360870361328, + "num_tokens": 401427758.0, + "step": 10518 + }, + { + "epoch": 1.338124920493576, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7849230766296387, + "num_tokens": 401473868.0, + "step": 10519 + }, + { + "epoch": 1.3382521307721664, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7786083817481995, + "num_tokens": 401510459.0, + "step": 10520 + }, + { + "epoch": 1.338379341050757, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7704721689224243, + "num_tokens": 401545283.0, + "step": 10521 + }, + { + "epoch": 1.3385065513293475, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7973294258117676, + "num_tokens": 401577166.0, + "step": 10522 + }, + { + "epoch": 1.3386337616079378, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8096192479133606, + "num_tokens": 401619075.0, + "step": 10523 + }, + { + "epoch": 1.3387609718865283, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7952960729598999, + "num_tokens": 401661286.0, + "step": 10524 + }, + { + "epoch": 1.3388881821651188, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7853066921234131, + "num_tokens": 401699672.0, + "step": 10525 + }, + { + "epoch": 1.3390153924437094, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.784447431564331, + "num_tokens": 401732676.0, + "step": 10526 + }, + { + "epoch": 1.3391426027223, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7965382933616638, + "num_tokens": 401771593.0, + "step": 10527 + }, + { + "epoch": 1.3392698130008904, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7827849388122559, + "num_tokens": 401809605.0, + "step": 10528 + }, + { + "epoch": 1.339397023279481, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7913271188735962, + "num_tokens": 401845349.0, + "step": 10529 + }, + { + "epoch": 1.3395242335580715, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5767, + "mean_token_accuracy": 0.8223866820335388, + "num_tokens": 401883186.0, + "step": 10530 + }, + { + "epoch": 1.339651443836662, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7836883068084717, + "num_tokens": 401921149.0, + "step": 10531 + }, + { + "epoch": 1.3397786541152525, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7927888631820679, + "num_tokens": 401961813.0, + "step": 10532 + }, + { + "epoch": 1.339905864393843, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7804961204528809, + "num_tokens": 402001212.0, + "step": 10533 + }, + { + "epoch": 1.3400330746724336, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7890530228614807, + "num_tokens": 402044648.0, + "step": 10534 + }, + { + "epoch": 1.3401602849510241, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7983230948448181, + "num_tokens": 402087391.0, + "step": 10535 + }, + { + "epoch": 1.3402874952296147, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8164982795715332, + "num_tokens": 402119598.0, + "step": 10536 + }, + { + "epoch": 1.3404147055082052, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7683454751968384, + "num_tokens": 402153628.0, + "step": 10537 + }, + { + "epoch": 1.3405419157867955, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8153256773948669, + "num_tokens": 402190876.0, + "step": 10538 + }, + { + "epoch": 1.340669126065386, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8048515319824219, + "num_tokens": 402227168.0, + "step": 10539 + }, + { + "epoch": 1.3407963363439765, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8078857660293579, + "num_tokens": 402262731.0, + "step": 10540 + }, + { + "epoch": 1.340923546622567, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8009458780288696, + "num_tokens": 402299992.0, + "step": 10541 + }, + { + "epoch": 1.3410507569011576, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7944756746292114, + "num_tokens": 402339129.0, + "step": 10542 + }, + { + "epoch": 1.3411779671797481, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7810932397842407, + "num_tokens": 402375380.0, + "step": 10543 + }, + { + "epoch": 1.3413051774583387, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7906566858291626, + "num_tokens": 402412159.0, + "step": 10544 + }, + { + "epoch": 1.3414323877369292, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7936556935310364, + "num_tokens": 402446676.0, + "step": 10545 + }, + { + "epoch": 1.3415595980155197, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8058249354362488, + "num_tokens": 402490741.0, + "step": 10546 + }, + { + "epoch": 1.3416868082941102, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8086862564086914, + "num_tokens": 402527398.0, + "step": 10547 + }, + { + "epoch": 1.3418140185727006, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7967624664306641, + "num_tokens": 402565914.0, + "step": 10548 + }, + { + "epoch": 1.341941228851291, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8024867177009583, + "num_tokens": 402600671.0, + "step": 10549 + }, + { + "epoch": 1.3420684391298816, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8075882792472839, + "num_tokens": 402637381.0, + "step": 10550 + }, + { + "epoch": 1.3421956494084721, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8064334988594055, + "num_tokens": 402671190.0, + "step": 10551 + }, + { + "epoch": 1.3423228596870627, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8030576705932617, + "num_tokens": 402721518.0, + "step": 10552 + }, + { + "epoch": 1.3424500699656532, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7952725291252136, + "num_tokens": 402758949.0, + "step": 10553 + }, + { + "epoch": 1.3425772802442437, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7949571013450623, + "num_tokens": 402794342.0, + "step": 10554 + }, + { + "epoch": 1.3427044905228342, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8138852715492249, + "num_tokens": 402831031.0, + "step": 10555 + }, + { + "epoch": 1.3428317008014248, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8115813732147217, + "num_tokens": 402871486.0, + "step": 10556 + }, + { + "epoch": 1.3429589110800153, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8103933334350586, + "num_tokens": 402916024.0, + "step": 10557 + }, + { + "epoch": 1.3430861213586058, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7825745940208435, + "num_tokens": 402952116.0, + "step": 10558 + }, + { + "epoch": 1.3432133316371964, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.8005931377410889, + "num_tokens": 402998064.0, + "step": 10559 + }, + { + "epoch": 1.343340541915787, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8065321445465088, + "num_tokens": 403041763.0, + "step": 10560 + }, + { + "epoch": 1.3434677521943774, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8086900115013123, + "num_tokens": 403080956.0, + "step": 10561 + }, + { + "epoch": 1.343594962472968, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8107808232307434, + "num_tokens": 403117200.0, + "step": 10562 + }, + { + "epoch": 1.3437221727515583, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8074316382408142, + "num_tokens": 403154112.0, + "step": 10563 + }, + { + "epoch": 1.3438493830301488, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7941877245903015, + "num_tokens": 403188265.0, + "step": 10564 + }, + { + "epoch": 1.3439765933087393, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7759025692939758, + "num_tokens": 403231236.0, + "step": 10565 + }, + { + "epoch": 1.3441038035873298, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8020584583282471, + "num_tokens": 403268642.0, + "step": 10566 + }, + { + "epoch": 1.3442310138659204, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8105306625366211, + "num_tokens": 403307679.0, + "step": 10567 + }, + { + "epoch": 1.344358224144511, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8060204982757568, + "num_tokens": 403346395.0, + "step": 10568 + }, + { + "epoch": 1.3444854344231014, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8061892986297607, + "num_tokens": 403383001.0, + "step": 10569 + }, + { + "epoch": 1.344612644701692, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7906602621078491, + "num_tokens": 403426090.0, + "step": 10570 + }, + { + "epoch": 1.3447398549802825, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7986545562744141, + "num_tokens": 403470231.0, + "step": 10571 + }, + { + "epoch": 1.3448670652588728, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.808585524559021, + "num_tokens": 403508677.0, + "step": 10572 + }, + { + "epoch": 1.3449942755374633, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7777082920074463, + "num_tokens": 403545175.0, + "step": 10573 + }, + { + "epoch": 1.3451214858160538, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8159209489822388, + "num_tokens": 403583321.0, + "step": 10574 + }, + { + "epoch": 1.3452486960946444, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7902486324310303, + "num_tokens": 403618547.0, + "step": 10575 + }, + { + "epoch": 1.345375906373235, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8074500560760498, + "num_tokens": 403657552.0, + "step": 10576 + }, + { + "epoch": 1.3455031166518254, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8025271892547607, + "num_tokens": 403697021.0, + "step": 10577 + }, + { + "epoch": 1.345630326930416, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.789201557636261, + "num_tokens": 403736480.0, + "step": 10578 + }, + { + "epoch": 1.3457575372090065, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8096311092376709, + "num_tokens": 403772056.0, + "step": 10579 + }, + { + "epoch": 1.345884747487597, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7955425381660461, + "num_tokens": 403813646.0, + "step": 10580 + }, + { + "epoch": 1.3460119577661875, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8008366823196411, + "num_tokens": 403853904.0, + "step": 10581 + }, + { + "epoch": 1.346139168044778, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.784813642501831, + "num_tokens": 403898528.0, + "step": 10582 + }, + { + "epoch": 1.3462663783233686, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7818266153335571, + "num_tokens": 403929518.0, + "step": 10583 + }, + { + "epoch": 1.3463935886019591, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.805065929889679, + "num_tokens": 403970323.0, + "step": 10584 + }, + { + "epoch": 1.3465207988805497, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8079316020011902, + "num_tokens": 404012499.0, + "step": 10585 + }, + { + "epoch": 1.3466480091591402, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7950586080551147, + "num_tokens": 404055370.0, + "step": 10586 + }, + { + "epoch": 1.3467752194377305, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.800308346748352, + "num_tokens": 404092498.0, + "step": 10587 + }, + { + "epoch": 1.346902429716321, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8088560700416565, + "num_tokens": 404132466.0, + "step": 10588 + }, + { + "epoch": 1.3470296399949115, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.801554799079895, + "num_tokens": 404171913.0, + "step": 10589 + }, + { + "epoch": 1.347156850273502, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8170331716537476, + "num_tokens": 404205817.0, + "step": 10590 + }, + { + "epoch": 1.3472840605520926, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8051918745040894, + "num_tokens": 404243922.0, + "step": 10591 + }, + { + "epoch": 1.3474112708306831, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8047007918357849, + "num_tokens": 404283392.0, + "step": 10592 + }, + { + "epoch": 1.3475384811092737, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7756121158599854, + "num_tokens": 404323044.0, + "step": 10593 + }, + { + "epoch": 1.3476656913878642, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8009495139122009, + "num_tokens": 404359761.0, + "step": 10594 + }, + { + "epoch": 1.3477929016664547, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7835900783538818, + "num_tokens": 404395354.0, + "step": 10595 + }, + { + "epoch": 1.3479201119450452, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7737205028533936, + "num_tokens": 404432052.0, + "step": 10596 + }, + { + "epoch": 1.3480473222236355, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7959780693054199, + "num_tokens": 404472689.0, + "step": 10597 + }, + { + "epoch": 1.348174532502226, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7852649092674255, + "num_tokens": 404515916.0, + "step": 10598 + }, + { + "epoch": 1.3483017427808166, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7995171546936035, + "num_tokens": 404557399.0, + "step": 10599 + }, + { + "epoch": 1.3484289530594071, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.795599102973938, + "num_tokens": 404592573.0, + "step": 10600 + }, + { + "epoch": 1.3485561633379977, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8016043305397034, + "num_tokens": 404635266.0, + "step": 10601 + }, + { + "epoch": 1.3486833736165882, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8082126975059509, + "num_tokens": 404673004.0, + "step": 10602 + }, + { + "epoch": 1.3488105838951787, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7808387279510498, + "num_tokens": 404705153.0, + "step": 10603 + }, + { + "epoch": 1.3489377941737692, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.788487434387207, + "num_tokens": 404737068.0, + "step": 10604 + }, + { + "epoch": 1.3490650044523598, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7793310880661011, + "num_tokens": 404773680.0, + "step": 10605 + }, + { + "epoch": 1.3491922147309503, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8065418601036072, + "num_tokens": 404809937.0, + "step": 10606 + }, + { + "epoch": 1.3493194250095408, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7883248329162598, + "num_tokens": 404847146.0, + "step": 10607 + }, + { + "epoch": 1.3494466352881314, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7988943457603455, + "num_tokens": 404882952.0, + "step": 10608 + }, + { + "epoch": 1.3495738455667219, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8103578686714172, + "num_tokens": 404925347.0, + "step": 10609 + }, + { + "epoch": 1.3497010558453124, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.788212239742279, + "num_tokens": 404963411.0, + "step": 10610 + }, + { + "epoch": 1.349828266123903, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7963846921920776, + "num_tokens": 404997522.0, + "step": 10611 + }, + { + "epoch": 1.3499554764024932, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8004947900772095, + "num_tokens": 405034049.0, + "step": 10612 + }, + { + "epoch": 1.3500826866810838, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7944437265396118, + "num_tokens": 405074985.0, + "step": 10613 + }, + { + "epoch": 1.3502098969596743, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8193206787109375, + "num_tokens": 405116133.0, + "step": 10614 + }, + { + "epoch": 1.3503371072382648, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7990978360176086, + "num_tokens": 405155572.0, + "step": 10615 + }, + { + "epoch": 1.3504643175168554, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8059571981430054, + "num_tokens": 405194700.0, + "step": 10616 + }, + { + "epoch": 1.350591527795446, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7985201478004456, + "num_tokens": 405233582.0, + "step": 10617 + }, + { + "epoch": 1.3507187380740364, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8123791217803955, + "num_tokens": 405274987.0, + "step": 10618 + }, + { + "epoch": 1.350845948352627, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7922521829605103, + "num_tokens": 405308151.0, + "step": 10619 + }, + { + "epoch": 1.3509731586312175, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7938344478607178, + "num_tokens": 405343939.0, + "step": 10620 + }, + { + "epoch": 1.3511003689098078, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7974556088447571, + "num_tokens": 405384211.0, + "step": 10621 + }, + { + "epoch": 1.3512275791883983, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8042132258415222, + "num_tokens": 405419971.0, + "step": 10622 + }, + { + "epoch": 1.3513547894669888, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7851753234863281, + "num_tokens": 405451322.0, + "step": 10623 + }, + { + "epoch": 1.3514819997455794, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7914415597915649, + "num_tokens": 405496101.0, + "step": 10624 + }, + { + "epoch": 1.35160921002417, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7814401984214783, + "num_tokens": 405530034.0, + "step": 10625 + }, + { + "epoch": 1.3517364203027604, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8149881958961487, + "num_tokens": 405570893.0, + "step": 10626 + }, + { + "epoch": 1.351863630581351, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7855022549629211, + "num_tokens": 405610120.0, + "step": 10627 + }, + { + "epoch": 1.3519908408599415, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8002885580062866, + "num_tokens": 405651878.0, + "step": 10628 + }, + { + "epoch": 1.352118051138532, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8046975135803223, + "num_tokens": 405694578.0, + "step": 10629 + }, + { + "epoch": 1.3522452614171225, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8054561614990234, + "num_tokens": 405734128.0, + "step": 10630 + }, + { + "epoch": 1.352372471695713, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.8030795454978943, + "num_tokens": 405776588.0, + "step": 10631 + }, + { + "epoch": 1.3524996819743036, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7929608821868896, + "num_tokens": 405810867.0, + "step": 10632 + }, + { + "epoch": 1.3526268922528941, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8231980800628662, + "num_tokens": 405848884.0, + "step": 10633 + }, + { + "epoch": 1.3527541025314846, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6032, + "mean_token_accuracy": 0.818711519241333, + "num_tokens": 405888759.0, + "step": 10634 + }, + { + "epoch": 1.3528813128100752, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8109769225120544, + "num_tokens": 405925686.0, + "step": 10635 + }, + { + "epoch": 1.3530085230886655, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8112011551856995, + "num_tokens": 405960836.0, + "step": 10636 + }, + { + "epoch": 1.353135733367256, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8120258450508118, + "num_tokens": 406000314.0, + "step": 10637 + }, + { + "epoch": 1.3532629436458465, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8070703744888306, + "num_tokens": 406039468.0, + "step": 10638 + }, + { + "epoch": 1.353390153924437, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.5981, + "mean_token_accuracy": 0.8190819025039673, + "num_tokens": 406076828.0, + "step": 10639 + }, + { + "epoch": 1.3535173642030276, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7982909679412842, + "num_tokens": 406116504.0, + "step": 10640 + }, + { + "epoch": 1.3536445744816181, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7959193587303162, + "num_tokens": 406155184.0, + "step": 10641 + }, + { + "epoch": 1.3537717847602087, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7790953516960144, + "num_tokens": 406195411.0, + "step": 10642 + }, + { + "epoch": 1.3538989950387992, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7862551212310791, + "num_tokens": 406230159.0, + "step": 10643 + }, + { + "epoch": 1.3540262053173897, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7831829786300659, + "num_tokens": 406264541.0, + "step": 10644 + }, + { + "epoch": 1.3541534155959802, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8079184293746948, + "num_tokens": 406301869.0, + "step": 10645 + }, + { + "epoch": 1.3542806258745705, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8051602840423584, + "num_tokens": 406338819.0, + "step": 10646 + }, + { + "epoch": 1.354407836153161, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8021712303161621, + "num_tokens": 406375914.0, + "step": 10647 + }, + { + "epoch": 1.3545350464317516, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8083590865135193, + "num_tokens": 406418543.0, + "step": 10648 + }, + { + "epoch": 1.3546622567103421, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.796764612197876, + "num_tokens": 406455731.0, + "step": 10649 + }, + { + "epoch": 1.3547894669889327, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8002501726150513, + "num_tokens": 406496266.0, + "step": 10650 + }, + { + "epoch": 1.3549166772675232, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7951633334159851, + "num_tokens": 406541676.0, + "step": 10651 + }, + { + "epoch": 1.3550438875461137, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7906855344772339, + "num_tokens": 406578997.0, + "step": 10652 + }, + { + "epoch": 1.3551710978247042, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8049832582473755, + "num_tokens": 406619404.0, + "step": 10653 + }, + { + "epoch": 1.3552983081032948, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.8219537138938904, + "num_tokens": 406656092.0, + "step": 10654 + }, + { + "epoch": 1.3554255183818853, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7859481573104858, + "num_tokens": 406701600.0, + "step": 10655 + }, + { + "epoch": 1.3555527286604758, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7995102405548096, + "num_tokens": 406739649.0, + "step": 10656 + }, + { + "epoch": 1.3556799389390664, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8149707913398743, + "num_tokens": 406774962.0, + "step": 10657 + }, + { + "epoch": 1.3558071492176569, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7975521087646484, + "num_tokens": 406818047.0, + "step": 10658 + }, + { + "epoch": 1.3559343594962474, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7857459783554077, + "num_tokens": 406859464.0, + "step": 10659 + }, + { + "epoch": 1.356061569774838, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7986611127853394, + "num_tokens": 406896444.0, + "step": 10660 + }, + { + "epoch": 1.3561887800534282, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7961921691894531, + "num_tokens": 406934623.0, + "step": 10661 + }, + { + "epoch": 1.3563159903320188, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7952467203140259, + "num_tokens": 406974290.0, + "step": 10662 + }, + { + "epoch": 1.3564432006106093, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8208223581314087, + "num_tokens": 407011279.0, + "step": 10663 + }, + { + "epoch": 1.3565704108891998, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7911471128463745, + "num_tokens": 407054733.0, + "step": 10664 + }, + { + "epoch": 1.3566976211677904, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8020823001861572, + "num_tokens": 407098410.0, + "step": 10665 + }, + { + "epoch": 1.3568248314463809, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.792750895023346, + "num_tokens": 407134231.0, + "step": 10666 + }, + { + "epoch": 1.3569520417249714, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7805166244506836, + "num_tokens": 407167231.0, + "step": 10667 + }, + { + "epoch": 1.357079252003562, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7993800044059753, + "num_tokens": 407206604.0, + "step": 10668 + }, + { + "epoch": 1.3572064622821525, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8035461902618408, + "num_tokens": 407248247.0, + "step": 10669 + }, + { + "epoch": 1.3573336725607428, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8074244856834412, + "num_tokens": 407282870.0, + "step": 10670 + }, + { + "epoch": 1.3574608828393333, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8045032620429993, + "num_tokens": 407318704.0, + "step": 10671 + }, + { + "epoch": 1.3575880931179238, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8025687336921692, + "num_tokens": 407360003.0, + "step": 10672 + }, + { + "epoch": 1.3577153033965144, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7923876047134399, + "num_tokens": 407399473.0, + "step": 10673 + }, + { + "epoch": 1.357842513675105, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7642600536346436, + "num_tokens": 407440576.0, + "step": 10674 + }, + { + "epoch": 1.3579697239536954, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7922041416168213, + "num_tokens": 407481057.0, + "step": 10675 + }, + { + "epoch": 1.358096934232286, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8122414946556091, + "num_tokens": 407517104.0, + "step": 10676 + }, + { + "epoch": 1.3582241445108765, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8034389019012451, + "num_tokens": 407552604.0, + "step": 10677 + }, + { + "epoch": 1.358351354789467, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8141871690750122, + "num_tokens": 407588282.0, + "step": 10678 + }, + { + "epoch": 1.3584785650680575, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7969603538513184, + "num_tokens": 407619837.0, + "step": 10679 + }, + { + "epoch": 1.358605775346648, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.8007903099060059, + "num_tokens": 407656939.0, + "step": 10680 + }, + { + "epoch": 1.3587329856252386, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7909371852874756, + "num_tokens": 407702756.0, + "step": 10681 + }, + { + "epoch": 1.3588601959038291, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8066959381103516, + "num_tokens": 407736142.0, + "step": 10682 + }, + { + "epoch": 1.3589874061824196, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7965481281280518, + "num_tokens": 407773184.0, + "step": 10683 + }, + { + "epoch": 1.3591146164610102, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.774268627166748, + "num_tokens": 407806543.0, + "step": 10684 + }, + { + "epoch": 1.3592418267396005, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7961830496788025, + "num_tokens": 407841371.0, + "step": 10685 + }, + { + "epoch": 1.359369037018191, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8030200600624084, + "num_tokens": 407879862.0, + "step": 10686 + }, + { + "epoch": 1.3594962472967815, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7924696803092957, + "num_tokens": 407917947.0, + "step": 10687 + }, + { + "epoch": 1.359623457575372, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8019518852233887, + "num_tokens": 407953987.0, + "step": 10688 + }, + { + "epoch": 1.3597506678539626, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.783779501914978, + "num_tokens": 407993166.0, + "step": 10689 + }, + { + "epoch": 1.3598778781325531, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8125557899475098, + "num_tokens": 408037063.0, + "step": 10690 + }, + { + "epoch": 1.3600050884111436, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8042083382606506, + "num_tokens": 408078526.0, + "step": 10691 + }, + { + "epoch": 1.3601322986897342, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8126932978630066, + "num_tokens": 408116002.0, + "step": 10692 + }, + { + "epoch": 1.3602595089683247, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7954124212265015, + "num_tokens": 408155885.0, + "step": 10693 + }, + { + "epoch": 1.3603867192469152, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8013773560523987, + "num_tokens": 408190085.0, + "step": 10694 + }, + { + "epoch": 1.3605139295255055, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7997374534606934, + "num_tokens": 408227289.0, + "step": 10695 + }, + { + "epoch": 1.360641139804096, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8106318116188049, + "num_tokens": 408271853.0, + "step": 10696 + }, + { + "epoch": 1.3607683500826866, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8147591352462769, + "num_tokens": 408311133.0, + "step": 10697 + }, + { + "epoch": 1.3608955603612771, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.797220766544342, + "num_tokens": 408342775.0, + "step": 10698 + }, + { + "epoch": 1.3610227706398677, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8152902722358704, + "num_tokens": 408379063.0, + "step": 10699 + }, + { + "epoch": 1.3611499809184582, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7836491465568542, + "num_tokens": 408415856.0, + "step": 10700 + }, + { + "epoch": 1.3612771911970487, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.806770920753479, + "num_tokens": 408451522.0, + "step": 10701 + }, + { + "epoch": 1.3614044014756392, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8103614449501038, + "num_tokens": 408494605.0, + "step": 10702 + }, + { + "epoch": 1.3615316117542298, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.796205461025238, + "num_tokens": 408536373.0, + "step": 10703 + }, + { + "epoch": 1.3616588220328203, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.782375156879425, + "num_tokens": 408574724.0, + "step": 10704 + }, + { + "epoch": 1.3617860323114108, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7945334315299988, + "num_tokens": 408615057.0, + "step": 10705 + }, + { + "epoch": 1.3619132425900013, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7851021885871887, + "num_tokens": 408651324.0, + "step": 10706 + }, + { + "epoch": 1.3620404528685919, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.806698739528656, + "num_tokens": 408690533.0, + "step": 10707 + }, + { + "epoch": 1.3621676631471824, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7880970239639282, + "num_tokens": 408725970.0, + "step": 10708 + }, + { + "epoch": 1.362294873425773, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8083051443099976, + "num_tokens": 408766550.0, + "step": 10709 + }, + { + "epoch": 1.3624220837043632, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8056871891021729, + "num_tokens": 408804320.0, + "step": 10710 + }, + { + "epoch": 1.3625492939829538, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.80211341381073, + "num_tokens": 408840902.0, + "step": 10711 + }, + { + "epoch": 1.3626765042615443, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7998263835906982, + "num_tokens": 408879346.0, + "step": 10712 + }, + { + "epoch": 1.3628037145401348, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.8175287842750549, + "num_tokens": 408919336.0, + "step": 10713 + }, + { + "epoch": 1.3629309248187254, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.803805947303772, + "num_tokens": 408961547.0, + "step": 10714 + }, + { + "epoch": 1.3630581350973159, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.8197550177574158, + "num_tokens": 409003573.0, + "step": 10715 + }, + { + "epoch": 1.3631853453759064, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7948148846626282, + "num_tokens": 409045560.0, + "step": 10716 + }, + { + "epoch": 1.363312555654497, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8012990355491638, + "num_tokens": 409087245.0, + "step": 10717 + }, + { + "epoch": 1.3634397659330875, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8113468289375305, + "num_tokens": 409134177.0, + "step": 10718 + }, + { + "epoch": 1.3635669762116778, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7799282073974609, + "num_tokens": 409169495.0, + "step": 10719 + }, + { + "epoch": 1.3636941864902683, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7761465311050415, + "num_tokens": 409207781.0, + "step": 10720 + }, + { + "epoch": 1.3638213967688588, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7955526113510132, + "num_tokens": 409245411.0, + "step": 10721 + }, + { + "epoch": 1.3639486070474494, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.8098877668380737, + "num_tokens": 409281757.0, + "step": 10722 + }, + { + "epoch": 1.3640758173260399, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7854768633842468, + "num_tokens": 409321317.0, + "step": 10723 + }, + { + "epoch": 1.3642030276046304, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8113393187522888, + "num_tokens": 409358919.0, + "step": 10724 + }, + { + "epoch": 1.364330237883221, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.790393590927124, + "num_tokens": 409394355.0, + "step": 10725 + }, + { + "epoch": 1.3644574481618115, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8139747977256775, + "num_tokens": 409433781.0, + "step": 10726 + }, + { + "epoch": 1.364584658440402, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.790777862071991, + "num_tokens": 409471728.0, + "step": 10727 + }, + { + "epoch": 1.3647118687189925, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8139216899871826, + "num_tokens": 409502294.0, + "step": 10728 + }, + { + "epoch": 1.364839078997583, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.8007934093475342, + "num_tokens": 409545878.0, + "step": 10729 + }, + { + "epoch": 1.3649662892761736, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8015186786651611, + "num_tokens": 409584753.0, + "step": 10730 + }, + { + "epoch": 1.3650934995547641, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8017043471336365, + "num_tokens": 409613725.0, + "step": 10731 + }, + { + "epoch": 1.3652207098333546, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7940443754196167, + "num_tokens": 409652443.0, + "step": 10732 + }, + { + "epoch": 1.3653479201119452, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8064924478530884, + "num_tokens": 409690683.0, + "step": 10733 + }, + { + "epoch": 1.3654751303905355, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7976769208908081, + "num_tokens": 409727502.0, + "step": 10734 + }, + { + "epoch": 1.365602340669126, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7950162887573242, + "num_tokens": 409764592.0, + "step": 10735 + }, + { + "epoch": 1.3657295509477165, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7854996919631958, + "num_tokens": 409809645.0, + "step": 10736 + }, + { + "epoch": 1.365856761226307, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7992214560508728, + "num_tokens": 409848048.0, + "step": 10737 + }, + { + "epoch": 1.3659839715048976, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7883337140083313, + "num_tokens": 409884176.0, + "step": 10738 + }, + { + "epoch": 1.3661111817834881, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.8017944693565369, + "num_tokens": 409922630.0, + "step": 10739 + }, + { + "epoch": 1.3662383920620786, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8032000660896301, + "num_tokens": 409961876.0, + "step": 10740 + }, + { + "epoch": 1.3663656023406692, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8056666851043701, + "num_tokens": 409995432.0, + "step": 10741 + }, + { + "epoch": 1.3664928126192597, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7995097637176514, + "num_tokens": 410034876.0, + "step": 10742 + }, + { + "epoch": 1.3666200228978502, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8009523153305054, + "num_tokens": 410079190.0, + "step": 10743 + }, + { + "epoch": 1.3667472331764405, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7986202239990234, + "num_tokens": 410112475.0, + "step": 10744 + }, + { + "epoch": 1.366874443455031, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.820504903793335, + "num_tokens": 410154583.0, + "step": 10745 + }, + { + "epoch": 1.3670016537336216, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.8005908727645874, + "num_tokens": 410191631.0, + "step": 10746 + }, + { + "epoch": 1.3671288640122121, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7767761945724487, + "num_tokens": 410232435.0, + "step": 10747 + }, + { + "epoch": 1.3672560742908026, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8050733804702759, + "num_tokens": 410272940.0, + "step": 10748 + }, + { + "epoch": 1.3673832845693932, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.797034740447998, + "num_tokens": 410312807.0, + "step": 10749 + }, + { + "epoch": 1.3675104948479837, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8056049942970276, + "num_tokens": 410352433.0, + "step": 10750 + }, + { + "epoch": 1.3676377051265742, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8010512590408325, + "num_tokens": 410394843.0, + "step": 10751 + }, + { + "epoch": 1.3677649154051648, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8032799959182739, + "num_tokens": 410439353.0, + "step": 10752 + }, + { + "epoch": 1.3678921256837553, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7771976590156555, + "num_tokens": 410485789.0, + "step": 10753 + }, + { + "epoch": 1.3680193359623458, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7843447923660278, + "num_tokens": 410523385.0, + "step": 10754 + }, + { + "epoch": 1.3681465462409363, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.790608823299408, + "num_tokens": 410553340.0, + "step": 10755 + }, + { + "epoch": 1.3682737565195269, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8095979690551758, + "num_tokens": 410590041.0, + "step": 10756 + }, + { + "epoch": 1.3684009667981174, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7991336584091187, + "num_tokens": 410627703.0, + "step": 10757 + }, + { + "epoch": 1.368528177076708, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7851356863975525, + "num_tokens": 410662555.0, + "step": 10758 + }, + { + "epoch": 1.3686553873552982, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7847360372543335, + "num_tokens": 410706026.0, + "step": 10759 + }, + { + "epoch": 1.3687825976338888, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7944812774658203, + "num_tokens": 410743275.0, + "step": 10760 + }, + { + "epoch": 1.3689098079124793, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8128466606140137, + "num_tokens": 410781220.0, + "step": 10761 + }, + { + "epoch": 1.3690370181910698, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8069092631340027, + "num_tokens": 410820014.0, + "step": 10762 + }, + { + "epoch": 1.3691642284696603, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7946978807449341, + "num_tokens": 410858721.0, + "step": 10763 + }, + { + "epoch": 1.3692914387482509, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8060556054115295, + "num_tokens": 410898527.0, + "step": 10764 + }, + { + "epoch": 1.3694186490268414, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8022772669792175, + "num_tokens": 410938426.0, + "step": 10765 + }, + { + "epoch": 1.369545859305432, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8017700910568237, + "num_tokens": 410969940.0, + "step": 10766 + }, + { + "epoch": 1.3696730695840225, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7849045991897583, + "num_tokens": 411008859.0, + "step": 10767 + }, + { + "epoch": 1.3698002798626128, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8076093196868896, + "num_tokens": 411043516.0, + "step": 10768 + }, + { + "epoch": 1.3699274901412033, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7901737689971924, + "num_tokens": 411079092.0, + "step": 10769 + }, + { + "epoch": 1.3700547004197938, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8006993532180786, + "num_tokens": 411117360.0, + "step": 10770 + }, + { + "epoch": 1.3701819106983844, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7884030938148499, + "num_tokens": 411153615.0, + "step": 10771 + }, + { + "epoch": 1.3703091209769749, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7952982783317566, + "num_tokens": 411193571.0, + "step": 10772 + }, + { + "epoch": 1.3704363312555654, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.780066967010498, + "num_tokens": 411230427.0, + "step": 10773 + }, + { + "epoch": 1.370563541534156, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.799321174621582, + "num_tokens": 411271251.0, + "step": 10774 + }, + { + "epoch": 1.3706907518127465, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8016656041145325, + "num_tokens": 411304906.0, + "step": 10775 + }, + { + "epoch": 1.370817962091337, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.819266676902771, + "num_tokens": 411341417.0, + "step": 10776 + }, + { + "epoch": 1.3709451723699275, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8048772811889648, + "num_tokens": 411382118.0, + "step": 10777 + }, + { + "epoch": 1.371072382648518, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.788554847240448, + "num_tokens": 411426702.0, + "step": 10778 + }, + { + "epoch": 1.3711995929271086, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7974926829338074, + "num_tokens": 411461689.0, + "step": 10779 + }, + { + "epoch": 1.371326803205699, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8042237758636475, + "num_tokens": 411499613.0, + "step": 10780 + }, + { + "epoch": 1.3714540134842896, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7975990176200867, + "num_tokens": 411538422.0, + "step": 10781 + }, + { + "epoch": 1.3715812237628802, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.81294184923172, + "num_tokens": 411572401.0, + "step": 10782 + }, + { + "epoch": 1.3717084340414705, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8076695203781128, + "num_tokens": 411611268.0, + "step": 10783 + }, + { + "epoch": 1.371835644320061, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7967877388000488, + "num_tokens": 411648837.0, + "step": 10784 + }, + { + "epoch": 1.3719628545986515, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8041435480117798, + "num_tokens": 411690327.0, + "step": 10785 + }, + { + "epoch": 1.372090064877242, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7985314726829529, + "num_tokens": 411729665.0, + "step": 10786 + }, + { + "epoch": 1.3722172751558326, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8161818981170654, + "num_tokens": 411764716.0, + "step": 10787 + }, + { + "epoch": 1.3723444854344231, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8013452291488647, + "num_tokens": 411805560.0, + "step": 10788 + }, + { + "epoch": 1.3724716957130136, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8037304878234863, + "num_tokens": 411841151.0, + "step": 10789 + }, + { + "epoch": 1.3725989059916042, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8027824759483337, + "num_tokens": 411875088.0, + "step": 10790 + }, + { + "epoch": 1.3727261162701947, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7976568937301636, + "num_tokens": 411914554.0, + "step": 10791 + }, + { + "epoch": 1.3728533265487852, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8007299900054932, + "num_tokens": 411952237.0, + "step": 10792 + }, + { + "epoch": 1.3729805368273755, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7972752451896667, + "num_tokens": 411989676.0, + "step": 10793 + }, + { + "epoch": 1.373107747105966, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8112602233886719, + "num_tokens": 412025804.0, + "step": 10794 + }, + { + "epoch": 1.3732349573845566, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8136162757873535, + "num_tokens": 412060085.0, + "step": 10795 + }, + { + "epoch": 1.3733621676631471, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8048593997955322, + "num_tokens": 412100750.0, + "step": 10796 + }, + { + "epoch": 1.3734893779417376, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8081627488136292, + "num_tokens": 412145829.0, + "step": 10797 + }, + { + "epoch": 1.3736165882203282, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8089865446090698, + "num_tokens": 412187607.0, + "step": 10798 + }, + { + "epoch": 1.3737437984989187, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.786432147026062, + "num_tokens": 412227214.0, + "step": 10799 + }, + { + "epoch": 1.3738710087775092, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7969541549682617, + "num_tokens": 412261929.0, + "step": 10800 + }, + { + "epoch": 1.3739982190560998, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7960627675056458, + "num_tokens": 412299441.0, + "step": 10801 + }, + { + "epoch": 1.3741254293346903, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.8002912998199463, + "num_tokens": 412339248.0, + "step": 10802 + }, + { + "epoch": 1.3742526396132808, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8018792867660522, + "num_tokens": 412377224.0, + "step": 10803 + }, + { + "epoch": 1.3743798498918713, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7720038294792175, + "num_tokens": 412415331.0, + "step": 10804 + }, + { + "epoch": 1.3745070601704619, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7990903854370117, + "num_tokens": 412455604.0, + "step": 10805 + }, + { + "epoch": 1.3746342704490524, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8088781237602234, + "num_tokens": 412489295.0, + "step": 10806 + }, + { + "epoch": 1.374761480727643, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7948189377784729, + "num_tokens": 412525750.0, + "step": 10807 + }, + { + "epoch": 1.3748886910062332, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8095647096633911, + "num_tokens": 412567163.0, + "step": 10808 + }, + { + "epoch": 1.3750159012848238, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8032410144805908, + "num_tokens": 412598789.0, + "step": 10809 + }, + { + "epoch": 1.3751431115634143, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8068543672561646, + "num_tokens": 412632588.0, + "step": 10810 + }, + { + "epoch": 1.3752703218420048, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7980457544326782, + "num_tokens": 412676369.0, + "step": 10811 + }, + { + "epoch": 1.3753975321205953, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8135668635368347, + "num_tokens": 412712674.0, + "step": 10812 + }, + { + "epoch": 1.3755247423991859, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.8006402850151062, + "num_tokens": 412753531.0, + "step": 10813 + }, + { + "epoch": 1.3756519526777764, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8008767366409302, + "num_tokens": 412799450.0, + "step": 10814 + }, + { + "epoch": 1.375779162956367, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8195078372955322, + "num_tokens": 412836858.0, + "step": 10815 + }, + { + "epoch": 1.3759063732349575, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8040765523910522, + "num_tokens": 412871169.0, + "step": 10816 + }, + { + "epoch": 1.3760335835135478, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7989721298217773, + "num_tokens": 412909281.0, + "step": 10817 + }, + { + "epoch": 1.3761607937921383, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7999402284622192, + "num_tokens": 412950126.0, + "step": 10818 + }, + { + "epoch": 1.3762880040707288, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7904139161109924, + "num_tokens": 412988644.0, + "step": 10819 + }, + { + "epoch": 1.3764152143493193, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7818911075592041, + "num_tokens": 413028403.0, + "step": 10820 + }, + { + "epoch": 1.3765424246279099, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7903904914855957, + "num_tokens": 413067429.0, + "step": 10821 + }, + { + "epoch": 1.3766696349065004, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.793938398361206, + "num_tokens": 413101630.0, + "step": 10822 + }, + { + "epoch": 1.376796845185091, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7946821451187134, + "num_tokens": 413147757.0, + "step": 10823 + }, + { + "epoch": 1.3769240554636815, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8044431209564209, + "num_tokens": 413186803.0, + "step": 10824 + }, + { + "epoch": 1.377051265742272, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7977306842803955, + "num_tokens": 413227201.0, + "step": 10825 + }, + { + "epoch": 1.3771784760208625, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8001251220703125, + "num_tokens": 413270594.0, + "step": 10826 + }, + { + "epoch": 1.377305686299453, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7786877155303955, + "num_tokens": 413307345.0, + "step": 10827 + }, + { + "epoch": 1.3774328965780436, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8088141083717346, + "num_tokens": 413344310.0, + "step": 10828 + }, + { + "epoch": 1.377560106856634, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8177998065948486, + "num_tokens": 413382047.0, + "step": 10829 + }, + { + "epoch": 1.3776873171352246, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8011727333068848, + "num_tokens": 413424658.0, + "step": 10830 + }, + { + "epoch": 1.3778145274138152, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8091747760772705, + "num_tokens": 413458386.0, + "step": 10831 + }, + { + "epoch": 1.3779417376924055, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7851277589797974, + "num_tokens": 413492530.0, + "step": 10832 + }, + { + "epoch": 1.378068947970996, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8123550415039062, + "num_tokens": 413528894.0, + "step": 10833 + }, + { + "epoch": 1.3781961582495865, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7961613535881042, + "num_tokens": 413564765.0, + "step": 10834 + }, + { + "epoch": 1.378323368528177, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.770512580871582, + "num_tokens": 413598122.0, + "step": 10835 + }, + { + "epoch": 1.3784505788067676, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7769526243209839, + "num_tokens": 413634519.0, + "step": 10836 + }, + { + "epoch": 1.378577789085358, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7907841205596924, + "num_tokens": 413674021.0, + "step": 10837 + }, + { + "epoch": 1.3787049993639486, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8040857911109924, + "num_tokens": 413713804.0, + "step": 10838 + }, + { + "epoch": 1.3788322096425392, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8133951425552368, + "num_tokens": 413755665.0, + "step": 10839 + }, + { + "epoch": 1.3789594199211297, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8006172180175781, + "num_tokens": 413790872.0, + "step": 10840 + }, + { + "epoch": 1.3790866301997202, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.8228397965431213, + "num_tokens": 413830798.0, + "step": 10841 + }, + { + "epoch": 1.3792138404783105, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.801767885684967, + "num_tokens": 413865960.0, + "step": 10842 + }, + { + "epoch": 1.379341050756901, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8064531683921814, + "num_tokens": 413898222.0, + "step": 10843 + }, + { + "epoch": 1.3794682610354916, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7915695905685425, + "num_tokens": 413931839.0, + "step": 10844 + }, + { + "epoch": 1.3795954713140821, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.786827027797699, + "num_tokens": 413963815.0, + "step": 10845 + }, + { + "epoch": 1.3797226815926726, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8113909959793091, + "num_tokens": 414002476.0, + "step": 10846 + }, + { + "epoch": 1.3798498918712632, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7969304323196411, + "num_tokens": 414041121.0, + "step": 10847 + }, + { + "epoch": 1.3799771021498537, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.783747136592865, + "num_tokens": 414080447.0, + "step": 10848 + }, + { + "epoch": 1.3801043124284442, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7872581481933594, + "num_tokens": 414108925.0, + "step": 10849 + }, + { + "epoch": 1.3802315227070348, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8061714172363281, + "num_tokens": 414154296.0, + "step": 10850 + }, + { + "epoch": 1.3803587329856253, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7911379933357239, + "num_tokens": 414194612.0, + "step": 10851 + }, + { + "epoch": 1.3804859432642158, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.802699089050293, + "num_tokens": 414229939.0, + "step": 10852 + }, + { + "epoch": 1.3806131535428063, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.79510498046875, + "num_tokens": 414265193.0, + "step": 10853 + }, + { + "epoch": 1.3807403638213969, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7950539588928223, + "num_tokens": 414301471.0, + "step": 10854 + }, + { + "epoch": 1.3808675740999874, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7713813781738281, + "num_tokens": 414338755.0, + "step": 10855 + }, + { + "epoch": 1.380994784378578, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8068991899490356, + "num_tokens": 414379996.0, + "step": 10856 + }, + { + "epoch": 1.3811219946571682, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8022246956825256, + "num_tokens": 414421905.0, + "step": 10857 + }, + { + "epoch": 1.3812492049357588, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.805514395236969, + "num_tokens": 414465149.0, + "step": 10858 + }, + { + "epoch": 1.3813764152143493, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8091562986373901, + "num_tokens": 414504086.0, + "step": 10859 + }, + { + "epoch": 1.3815036254929398, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7820327281951904, + "num_tokens": 414545392.0, + "step": 10860 + }, + { + "epoch": 1.3816308357715303, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7624428868293762, + "num_tokens": 414586445.0, + "step": 10861 + }, + { + "epoch": 1.3817580460501209, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8046860098838806, + "num_tokens": 414618467.0, + "step": 10862 + }, + { + "epoch": 1.3818852563287114, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.8121922016143799, + "num_tokens": 414651253.0, + "step": 10863 + }, + { + "epoch": 1.382012466607302, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8020381331443787, + "num_tokens": 414692016.0, + "step": 10864 + }, + { + "epoch": 1.3821396768858925, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8052644729614258, + "num_tokens": 414731840.0, + "step": 10865 + }, + { + "epoch": 1.3822668871644828, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7972867488861084, + "num_tokens": 414767110.0, + "step": 10866 + }, + { + "epoch": 1.3823940974430733, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8008255958557129, + "num_tokens": 414799340.0, + "step": 10867 + }, + { + "epoch": 1.3825213077216638, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8054636716842651, + "num_tokens": 414840915.0, + "step": 10868 + }, + { + "epoch": 1.3826485180002543, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7898419499397278, + "num_tokens": 414877298.0, + "step": 10869 + }, + { + "epoch": 1.3827757282788449, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8042507171630859, + "num_tokens": 414916431.0, + "step": 10870 + }, + { + "epoch": 1.3829029385574354, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.800629734992981, + "num_tokens": 414958148.0, + "step": 10871 + }, + { + "epoch": 1.383030148836026, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.5786, + "mean_token_accuracy": 0.8236194849014282, + "num_tokens": 414997008.0, + "step": 10872 + }, + { + "epoch": 1.3831573591146165, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7877655625343323, + "num_tokens": 415042893.0, + "step": 10873 + }, + { + "epoch": 1.383284569393207, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7965196371078491, + "num_tokens": 415077476.0, + "step": 10874 + }, + { + "epoch": 1.3834117796717975, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7980502843856812, + "num_tokens": 415124329.0, + "step": 10875 + }, + { + "epoch": 1.383538989950388, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8163632154464722, + "num_tokens": 415164101.0, + "step": 10876 + }, + { + "epoch": 1.3836662002289786, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7997819185256958, + "num_tokens": 415203792.0, + "step": 10877 + }, + { + "epoch": 1.383793410507569, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7935992479324341, + "num_tokens": 415233754.0, + "step": 10878 + }, + { + "epoch": 1.3839206207861596, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8069404363632202, + "num_tokens": 415269567.0, + "step": 10879 + }, + { + "epoch": 1.3840478310647502, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8084771037101746, + "num_tokens": 415312660.0, + "step": 10880 + }, + { + "epoch": 1.3841750413433405, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8169984817504883, + "num_tokens": 415355377.0, + "step": 10881 + }, + { + "epoch": 1.384302251621931, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.8102500438690186, + "num_tokens": 415393109.0, + "step": 10882 + }, + { + "epoch": 1.3844294619005215, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8155391216278076, + "num_tokens": 415428650.0, + "step": 10883 + }, + { + "epoch": 1.384556672179112, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7981718182563782, + "num_tokens": 415467166.0, + "step": 10884 + }, + { + "epoch": 1.3846838824577026, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8164769411087036, + "num_tokens": 415505633.0, + "step": 10885 + }, + { + "epoch": 1.384811092736293, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7980324029922485, + "num_tokens": 415546254.0, + "step": 10886 + }, + { + "epoch": 1.3849383030148836, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7921284437179565, + "num_tokens": 415584999.0, + "step": 10887 + }, + { + "epoch": 1.3850655132934742, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7993164658546448, + "num_tokens": 415624608.0, + "step": 10888 + }, + { + "epoch": 1.3851927235720647, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7832367420196533, + "num_tokens": 415663405.0, + "step": 10889 + }, + { + "epoch": 1.385319933850655, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.8142581582069397, + "num_tokens": 415698934.0, + "step": 10890 + }, + { + "epoch": 1.3854471441292455, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7675638198852539, + "num_tokens": 415736973.0, + "step": 10891 + }, + { + "epoch": 1.385574354407836, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8044700622558594, + "num_tokens": 415778364.0, + "step": 10892 + }, + { + "epoch": 1.3857015646864266, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8137412667274475, + "num_tokens": 415817502.0, + "step": 10893 + }, + { + "epoch": 1.385828774965017, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7961322665214539, + "num_tokens": 415860364.0, + "step": 10894 + }, + { + "epoch": 1.3859559852436076, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8027865886688232, + "num_tokens": 415896816.0, + "step": 10895 + }, + { + "epoch": 1.3860831955221982, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7831494808197021, + "num_tokens": 415935173.0, + "step": 10896 + }, + { + "epoch": 1.3862104058007887, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8141880035400391, + "num_tokens": 415977043.0, + "step": 10897 + }, + { + "epoch": 1.3863376160793792, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8103252649307251, + "num_tokens": 416014376.0, + "step": 10898 + }, + { + "epoch": 1.3864648263579697, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7876784801483154, + "num_tokens": 416055648.0, + "step": 10899 + }, + { + "epoch": 1.3865920366365603, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7942112684249878, + "num_tokens": 416087062.0, + "step": 10900 + }, + { + "epoch": 1.3867192469151508, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8102378249168396, + "num_tokens": 416122559.0, + "step": 10901 + }, + { + "epoch": 1.3868464571937413, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7900545597076416, + "num_tokens": 416155228.0, + "step": 10902 + }, + { + "epoch": 1.3869736674723319, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7818859815597534, + "num_tokens": 416191190.0, + "step": 10903 + }, + { + "epoch": 1.3871008777509224, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7991244792938232, + "num_tokens": 416228009.0, + "step": 10904 + }, + { + "epoch": 1.387228088029513, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7975258231163025, + "num_tokens": 416263657.0, + "step": 10905 + }, + { + "epoch": 1.3873552983081032, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.790071964263916, + "num_tokens": 416305187.0, + "step": 10906 + }, + { + "epoch": 1.3874825085866938, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7987727522850037, + "num_tokens": 416342578.0, + "step": 10907 + }, + { + "epoch": 1.3876097188652843, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7946568131446838, + "num_tokens": 416378969.0, + "step": 10908 + }, + { + "epoch": 1.3877369291438748, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.800118088722229, + "num_tokens": 416412958.0, + "step": 10909 + }, + { + "epoch": 1.3878641394224653, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7746940851211548, + "num_tokens": 416445643.0, + "step": 10910 + }, + { + "epoch": 1.3879913497010559, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7863630652427673, + "num_tokens": 416484891.0, + "step": 10911 + }, + { + "epoch": 1.3881185599796464, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.786975622177124, + "num_tokens": 416520504.0, + "step": 10912 + }, + { + "epoch": 1.388245770258237, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.801971435546875, + "num_tokens": 416561877.0, + "step": 10913 + }, + { + "epoch": 1.3883729805368275, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.798910915851593, + "num_tokens": 416597676.0, + "step": 10914 + }, + { + "epoch": 1.3885001908154178, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8057416677474976, + "num_tokens": 416630557.0, + "step": 10915 + }, + { + "epoch": 1.3886274010940083, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8009203672409058, + "num_tokens": 416668604.0, + "step": 10916 + }, + { + "epoch": 1.3887546113725988, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7805149555206299, + "num_tokens": 416710246.0, + "step": 10917 + }, + { + "epoch": 1.3888818216511893, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7997831106185913, + "num_tokens": 416744962.0, + "step": 10918 + }, + { + "epoch": 1.3890090319297799, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8018105626106262, + "num_tokens": 416779483.0, + "step": 10919 + }, + { + "epoch": 1.3891362422083704, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8147045373916626, + "num_tokens": 416822135.0, + "step": 10920 + }, + { + "epoch": 1.389263452486961, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8024053573608398, + "num_tokens": 416860383.0, + "step": 10921 + }, + { + "epoch": 1.3893906627655515, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7921478152275085, + "num_tokens": 416896960.0, + "step": 10922 + }, + { + "epoch": 1.389517873044142, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8136675953865051, + "num_tokens": 416931981.0, + "step": 10923 + }, + { + "epoch": 1.3896450833227325, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8081945180892944, + "num_tokens": 416968179.0, + "step": 10924 + }, + { + "epoch": 1.389772293601323, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8051747679710388, + "num_tokens": 417008879.0, + "step": 10925 + }, + { + "epoch": 1.3898995038799136, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8127326965332031, + "num_tokens": 417042497.0, + "step": 10926 + }, + { + "epoch": 1.390026714158504, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7997235059738159, + "num_tokens": 417084396.0, + "step": 10927 + }, + { + "epoch": 1.3901539244370946, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8146711587905884, + "num_tokens": 417121014.0, + "step": 10928 + }, + { + "epoch": 1.3902811347156852, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.797282874584198, + "num_tokens": 417160169.0, + "step": 10929 + }, + { + "epoch": 1.3904083449942755, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7824915647506714, + "num_tokens": 417195024.0, + "step": 10930 + }, + { + "epoch": 1.390535555272866, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8098267316818237, + "num_tokens": 417231344.0, + "step": 10931 + }, + { + "epoch": 1.3906627655514565, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7956321835517883, + "num_tokens": 417270266.0, + "step": 10932 + }, + { + "epoch": 1.390789975830047, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7971525192260742, + "num_tokens": 417308292.0, + "step": 10933 + }, + { + "epoch": 1.3909171861086376, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7914217710494995, + "num_tokens": 417346358.0, + "step": 10934 + }, + { + "epoch": 1.391044396387228, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7968100309371948, + "num_tokens": 417381924.0, + "step": 10935 + }, + { + "epoch": 1.3911716066658186, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7728598117828369, + "num_tokens": 417417779.0, + "step": 10936 + }, + { + "epoch": 1.3912988169444092, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7895703315734863, + "num_tokens": 417459076.0, + "step": 10937 + }, + { + "epoch": 1.3914260272229997, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8009594082832336, + "num_tokens": 417499422.0, + "step": 10938 + }, + { + "epoch": 1.39155323750159, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8011852502822876, + "num_tokens": 417541776.0, + "step": 10939 + }, + { + "epoch": 1.3916804477801805, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7813702821731567, + "num_tokens": 417576469.0, + "step": 10940 + }, + { + "epoch": 1.391807658058771, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8004453182220459, + "num_tokens": 417611726.0, + "step": 10941 + }, + { + "epoch": 1.3919348683373616, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7903261184692383, + "num_tokens": 417650135.0, + "step": 10942 + }, + { + "epoch": 1.392062078615952, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7813873291015625, + "num_tokens": 417690981.0, + "step": 10943 + }, + { + "epoch": 1.3921892888945426, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7884924411773682, + "num_tokens": 417732152.0, + "step": 10944 + }, + { + "epoch": 1.3923164991731332, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7854830026626587, + "num_tokens": 417771814.0, + "step": 10945 + }, + { + "epoch": 1.3924437094517237, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7870442867279053, + "num_tokens": 417810663.0, + "step": 10946 + }, + { + "epoch": 1.3925709197303142, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8056812286376953, + "num_tokens": 417845664.0, + "step": 10947 + }, + { + "epoch": 1.3926981300089047, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7927228808403015, + "num_tokens": 417889617.0, + "step": 10948 + }, + { + "epoch": 1.3928253402874953, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7995304465293884, + "num_tokens": 417930820.0, + "step": 10949 + }, + { + "epoch": 1.3929525505660858, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7974939346313477, + "num_tokens": 417966799.0, + "step": 10950 + }, + { + "epoch": 1.3930797608446763, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7910687923431396, + "num_tokens": 418009328.0, + "step": 10951 + }, + { + "epoch": 1.3932069711232669, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8009271025657654, + "num_tokens": 418048548.0, + "step": 10952 + }, + { + "epoch": 1.3933341814018574, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7949787378311157, + "num_tokens": 418094947.0, + "step": 10953 + }, + { + "epoch": 1.393461391680448, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7914585471153259, + "num_tokens": 418137144.0, + "step": 10954 + }, + { + "epoch": 1.3935886019590382, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.819151759147644, + "num_tokens": 418171449.0, + "step": 10955 + }, + { + "epoch": 1.3937158122376287, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.804540753364563, + "num_tokens": 418209854.0, + "step": 10956 + }, + { + "epoch": 1.3938430225162193, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8107929229736328, + "num_tokens": 418245121.0, + "step": 10957 + }, + { + "epoch": 1.3939702327948098, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.805300772190094, + "num_tokens": 418284158.0, + "step": 10958 + }, + { + "epoch": 1.3940974430734003, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8099485635757446, + "num_tokens": 418324091.0, + "step": 10959 + }, + { + "epoch": 1.3942246533519909, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8028784394264221, + "num_tokens": 418364764.0, + "step": 10960 + }, + { + "epoch": 1.3943518636305814, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.800070583820343, + "num_tokens": 418403974.0, + "step": 10961 + }, + { + "epoch": 1.394479073909172, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7899011373519897, + "num_tokens": 418446341.0, + "step": 10962 + }, + { + "epoch": 1.3946062841877624, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8024967312812805, + "num_tokens": 418490989.0, + "step": 10963 + }, + { + "epoch": 1.3947334944663528, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7975568175315857, + "num_tokens": 418524633.0, + "step": 10964 + }, + { + "epoch": 1.3948607047449433, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7948185205459595, + "num_tokens": 418562576.0, + "step": 10965 + }, + { + "epoch": 1.3949879150235338, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7961333990097046, + "num_tokens": 418596537.0, + "step": 10966 + }, + { + "epoch": 1.3951151253021243, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7900866270065308, + "num_tokens": 418625800.0, + "step": 10967 + }, + { + "epoch": 1.3952423355807149, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7848840951919556, + "num_tokens": 418663818.0, + "step": 10968 + }, + { + "epoch": 1.3953695458593054, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8056396245956421, + "num_tokens": 418700607.0, + "step": 10969 + }, + { + "epoch": 1.395496756137896, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7893510460853577, + "num_tokens": 418739357.0, + "step": 10970 + }, + { + "epoch": 1.3956239664164865, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7933441400527954, + "num_tokens": 418769752.0, + "step": 10971 + }, + { + "epoch": 1.395751176695077, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7751442790031433, + "num_tokens": 418808196.0, + "step": 10972 + }, + { + "epoch": 1.3958783869736675, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8174554109573364, + "num_tokens": 418843335.0, + "step": 10973 + }, + { + "epoch": 1.396005597252258, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8033316135406494, + "num_tokens": 418876591.0, + "step": 10974 + }, + { + "epoch": 1.3961328075308486, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7823902368545532, + "num_tokens": 418910482.0, + "step": 10975 + }, + { + "epoch": 1.396260017809439, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.786064863204956, + "num_tokens": 418942248.0, + "step": 10976 + }, + { + "epoch": 1.3963872280880296, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.806079626083374, + "num_tokens": 418977770.0, + "step": 10977 + }, + { + "epoch": 1.3965144383666201, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7938632369041443, + "num_tokens": 419016464.0, + "step": 10978 + }, + { + "epoch": 1.3966416486452105, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7895796298980713, + "num_tokens": 419053254.0, + "step": 10979 + }, + { + "epoch": 1.396768858923801, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.786300539970398, + "num_tokens": 419088940.0, + "step": 10980 + }, + { + "epoch": 1.3968960692023915, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8002065420150757, + "num_tokens": 419127897.0, + "step": 10981 + }, + { + "epoch": 1.397023279480982, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8008778095245361, + "num_tokens": 419170921.0, + "step": 10982 + }, + { + "epoch": 1.3971504897595726, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8054658770561218, + "num_tokens": 419206032.0, + "step": 10983 + }, + { + "epoch": 1.397277700038163, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8154737949371338, + "num_tokens": 419244240.0, + "step": 10984 + }, + { + "epoch": 1.3974049103167536, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7992525100708008, + "num_tokens": 419274902.0, + "step": 10985 + }, + { + "epoch": 1.3975321205953442, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8138666152954102, + "num_tokens": 419312295.0, + "step": 10986 + }, + { + "epoch": 1.3976593308739347, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7788299918174744, + "num_tokens": 419350999.0, + "step": 10987 + }, + { + "epoch": 1.397786541152525, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7937748432159424, + "num_tokens": 419386739.0, + "step": 10988 + }, + { + "epoch": 1.3979137514311155, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8102121353149414, + "num_tokens": 419430353.0, + "step": 10989 + }, + { + "epoch": 1.398040961709706, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7900874614715576, + "num_tokens": 419462910.0, + "step": 10990 + }, + { + "epoch": 1.3981681719882966, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8083219528198242, + "num_tokens": 419498727.0, + "step": 10991 + }, + { + "epoch": 1.398295382266887, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7875818014144897, + "num_tokens": 419534496.0, + "step": 10992 + }, + { + "epoch": 1.3984225925454776, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7981677651405334, + "num_tokens": 419575760.0, + "step": 10993 + }, + { + "epoch": 1.3985498028240682, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.810238778591156, + "num_tokens": 419616822.0, + "step": 10994 + }, + { + "epoch": 1.3986770131026587, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8063614964485168, + "num_tokens": 419650425.0, + "step": 10995 + }, + { + "epoch": 1.3988042233812492, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.805094838142395, + "num_tokens": 419685501.0, + "step": 10996 + }, + { + "epoch": 1.3989314336598397, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8040249347686768, + "num_tokens": 419724471.0, + "step": 10997 + }, + { + "epoch": 1.3990586439384303, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8071959018707275, + "num_tokens": 419763683.0, + "step": 10998 + }, + { + "epoch": 1.3991858542170208, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8032181262969971, + "num_tokens": 419801238.0, + "step": 10999 + }, + { + "epoch": 1.3993130644956113, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7896393537521362, + "num_tokens": 419832787.0, + "step": 11000 + }, + { + "epoch": 1.3994402747742019, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7828873991966248, + "num_tokens": 419864294.0, + "step": 11001 + }, + { + "epoch": 1.3995674850527924, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8079408407211304, + "num_tokens": 419900493.0, + "step": 11002 + }, + { + "epoch": 1.399694695331383, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.8219573497772217, + "num_tokens": 419938917.0, + "step": 11003 + }, + { + "epoch": 1.3998219056099732, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.8126789927482605, + "num_tokens": 419978771.0, + "step": 11004 + }, + { + "epoch": 1.3999491158885637, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7964159250259399, + "num_tokens": 420011836.0, + "step": 11005 + }, + { + "epoch": 1.4000763261671543, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7765181064605713, + "num_tokens": 420051618.0, + "step": 11006 + }, + { + "epoch": 1.4002035364457448, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7857495546340942, + "num_tokens": 420089119.0, + "step": 11007 + }, + { + "epoch": 1.4003307467243353, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8064160346984863, + "num_tokens": 420123706.0, + "step": 11008 + }, + { + "epoch": 1.4004579570029259, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7960584163665771, + "num_tokens": 420156723.0, + "step": 11009 + }, + { + "epoch": 1.4005851672815164, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7981494665145874, + "num_tokens": 420198519.0, + "step": 11010 + }, + { + "epoch": 1.400712377560107, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6008, + "mean_token_accuracy": 0.8155965209007263, + "num_tokens": 420231722.0, + "step": 11011 + }, + { + "epoch": 1.4008395878386974, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8036460876464844, + "num_tokens": 420272059.0, + "step": 11012 + }, + { + "epoch": 1.4009667981172877, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8093553781509399, + "num_tokens": 420307695.0, + "step": 11013 + }, + { + "epoch": 1.4010940083958783, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.813689649105072, + "num_tokens": 420345989.0, + "step": 11014 + }, + { + "epoch": 1.4012212186744688, + "grad_norm": 9.4375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8070345520973206, + "num_tokens": 420387194.0, + "step": 11015 + }, + { + "epoch": 1.4013484289530593, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8149409294128418, + "num_tokens": 420427608.0, + "step": 11016 + }, + { + "epoch": 1.4014756392316499, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8049582242965698, + "num_tokens": 420461363.0, + "step": 11017 + }, + { + "epoch": 1.4016028495102404, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7827513217926025, + "num_tokens": 420503577.0, + "step": 11018 + }, + { + "epoch": 1.401730059788831, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8100665807723999, + "num_tokens": 420538852.0, + "step": 11019 + }, + { + "epoch": 1.4018572700674214, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7985532283782959, + "num_tokens": 420577495.0, + "step": 11020 + }, + { + "epoch": 1.401984480346012, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8040995597839355, + "num_tokens": 420616123.0, + "step": 11021 + }, + { + "epoch": 1.4021116906246025, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8116393089294434, + "num_tokens": 420653904.0, + "step": 11022 + }, + { + "epoch": 1.402238900903193, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8226068019866943, + "num_tokens": 420693539.0, + "step": 11023 + }, + { + "epoch": 1.4023661111817836, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8032946586608887, + "num_tokens": 420729046.0, + "step": 11024 + }, + { + "epoch": 1.402493321460374, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.8020451664924622, + "num_tokens": 420766143.0, + "step": 11025 + }, + { + "epoch": 1.4026205317389646, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7982844114303589, + "num_tokens": 420806367.0, + "step": 11026 + }, + { + "epoch": 1.4027477420175551, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8072942495346069, + "num_tokens": 420845081.0, + "step": 11027 + }, + { + "epoch": 1.4028749522961454, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8074818849563599, + "num_tokens": 420885833.0, + "step": 11028 + }, + { + "epoch": 1.403002162574736, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8012586832046509, + "num_tokens": 420916313.0, + "step": 11029 + }, + { + "epoch": 1.4031293728533265, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.786918044090271, + "num_tokens": 420957011.0, + "step": 11030 + }, + { + "epoch": 1.403256583131917, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8051052689552307, + "num_tokens": 420994815.0, + "step": 11031 + }, + { + "epoch": 1.4033837934105076, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.8144235610961914, + "num_tokens": 421033977.0, + "step": 11032 + }, + { + "epoch": 1.403511003689098, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8031256794929504, + "num_tokens": 421068871.0, + "step": 11033 + }, + { + "epoch": 1.4036382139676886, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8032763600349426, + "num_tokens": 421105680.0, + "step": 11034 + }, + { + "epoch": 1.4037654242462791, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8162596821784973, + "num_tokens": 421149519.0, + "step": 11035 + }, + { + "epoch": 1.4038926345248697, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.8065197467803955, + "num_tokens": 421190195.0, + "step": 11036 + }, + { + "epoch": 1.40401984480346, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.789484441280365, + "num_tokens": 421218637.0, + "step": 11037 + }, + { + "epoch": 1.4041470550820505, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8058545589447021, + "num_tokens": 421261300.0, + "step": 11038 + }, + { + "epoch": 1.404274265360641, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7941359281539917, + "num_tokens": 421297250.0, + "step": 11039 + }, + { + "epoch": 1.4044014756392316, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7896977066993713, + "num_tokens": 421334213.0, + "step": 11040 + }, + { + "epoch": 1.404528685917822, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8073719143867493, + "num_tokens": 421368341.0, + "step": 11041 + }, + { + "epoch": 1.4046558961964126, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7885236740112305, + "num_tokens": 421408278.0, + "step": 11042 + }, + { + "epoch": 1.4047831064750032, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.791236162185669, + "num_tokens": 421442559.0, + "step": 11043 + }, + { + "epoch": 1.4049103167535937, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7916747331619263, + "num_tokens": 421474014.0, + "step": 11044 + }, + { + "epoch": 1.4050375270321842, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.776745080947876, + "num_tokens": 421511193.0, + "step": 11045 + }, + { + "epoch": 1.4051647373107747, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8001923561096191, + "num_tokens": 421547895.0, + "step": 11046 + }, + { + "epoch": 1.4052919475893653, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7945246696472168, + "num_tokens": 421582883.0, + "step": 11047 + }, + { + "epoch": 1.4054191578679558, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7988667488098145, + "num_tokens": 421619838.0, + "step": 11048 + }, + { + "epoch": 1.4055463681465463, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8121247887611389, + "num_tokens": 421659451.0, + "step": 11049 + }, + { + "epoch": 1.4056735784251368, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.794620156288147, + "num_tokens": 421704826.0, + "step": 11050 + }, + { + "epoch": 1.4058007887037274, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8001841306686401, + "num_tokens": 421748953.0, + "step": 11051 + }, + { + "epoch": 1.405927998982318, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7824387550354004, + "num_tokens": 421786614.0, + "step": 11052 + }, + { + "epoch": 1.4060552092609082, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7929078936576843, + "num_tokens": 421817291.0, + "step": 11053 + }, + { + "epoch": 1.4061824195394987, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8011487722396851, + "num_tokens": 421858200.0, + "step": 11054 + }, + { + "epoch": 1.4063096298180893, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8013274073600769, + "num_tokens": 421903343.0, + "step": 11055 + }, + { + "epoch": 1.4064368400966798, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.8194409608840942, + "num_tokens": 421941426.0, + "step": 11056 + }, + { + "epoch": 1.4065640503752703, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7963598966598511, + "num_tokens": 421983992.0, + "step": 11057 + }, + { + "epoch": 1.4066912606538609, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8087993860244751, + "num_tokens": 422015066.0, + "step": 11058 + }, + { + "epoch": 1.4068184709324514, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8047971725463867, + "num_tokens": 422052392.0, + "step": 11059 + }, + { + "epoch": 1.406945681211042, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7998476028442383, + "num_tokens": 422091453.0, + "step": 11060 + }, + { + "epoch": 1.4070728914896324, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7819291353225708, + "num_tokens": 422131019.0, + "step": 11061 + }, + { + "epoch": 1.4072001017682227, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7932102680206299, + "num_tokens": 422172139.0, + "step": 11062 + }, + { + "epoch": 1.4073273120468133, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8058655261993408, + "num_tokens": 422209708.0, + "step": 11063 + }, + { + "epoch": 1.4074545223254038, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7986223101615906, + "num_tokens": 422245091.0, + "step": 11064 + }, + { + "epoch": 1.4075817326039943, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7946300506591797, + "num_tokens": 422280054.0, + "step": 11065 + }, + { + "epoch": 1.4077089428825849, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.804379940032959, + "num_tokens": 422310036.0, + "step": 11066 + }, + { + "epoch": 1.4078361531611754, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7857173681259155, + "num_tokens": 422347914.0, + "step": 11067 + }, + { + "epoch": 1.407963363439766, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7982552647590637, + "num_tokens": 422383207.0, + "step": 11068 + }, + { + "epoch": 1.4080905737183564, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8015725612640381, + "num_tokens": 422414675.0, + "step": 11069 + }, + { + "epoch": 1.408217783996947, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7883262634277344, + "num_tokens": 422449746.0, + "step": 11070 + }, + { + "epoch": 1.4083449942755375, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7906602025032043, + "num_tokens": 422487789.0, + "step": 11071 + }, + { + "epoch": 1.408472204554128, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7992759943008423, + "num_tokens": 422521750.0, + "step": 11072 + }, + { + "epoch": 1.4085994148327186, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8127173185348511, + "num_tokens": 422562923.0, + "step": 11073 + }, + { + "epoch": 1.408726625111309, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7916048765182495, + "num_tokens": 422606832.0, + "step": 11074 + }, + { + "epoch": 1.4088538353898996, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8016533255577087, + "num_tokens": 422644465.0, + "step": 11075 + }, + { + "epoch": 1.4089810456684901, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8062300682067871, + "num_tokens": 422681204.0, + "step": 11076 + }, + { + "epoch": 1.4091082559470804, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7842885255813599, + "num_tokens": 422716325.0, + "step": 11077 + }, + { + "epoch": 1.409235466225671, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7885730266571045, + "num_tokens": 422753644.0, + "step": 11078 + }, + { + "epoch": 1.4093626765042615, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7991807460784912, + "num_tokens": 422790893.0, + "step": 11079 + }, + { + "epoch": 1.409489886782852, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.791563093662262, + "num_tokens": 422829400.0, + "step": 11080 + }, + { + "epoch": 1.4096170970614426, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8102630376815796, + "num_tokens": 422867386.0, + "step": 11081 + }, + { + "epoch": 1.409744307340033, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8046159744262695, + "num_tokens": 422910736.0, + "step": 11082 + }, + { + "epoch": 1.4098715176186236, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.80219566822052, + "num_tokens": 422950302.0, + "step": 11083 + }, + { + "epoch": 1.4099987278972141, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7996922731399536, + "num_tokens": 422985892.0, + "step": 11084 + }, + { + "epoch": 1.4101259381758047, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7892976999282837, + "num_tokens": 423020075.0, + "step": 11085 + }, + { + "epoch": 1.410253148454395, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8005204796791077, + "num_tokens": 423058260.0, + "step": 11086 + }, + { + "epoch": 1.4103803587329855, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.5684, + "mean_token_accuracy": 0.8264284133911133, + "num_tokens": 423092158.0, + "step": 11087 + }, + { + "epoch": 1.410507569011576, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7949341535568237, + "num_tokens": 423123999.0, + "step": 11088 + }, + { + "epoch": 1.4106347792901666, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7983378171920776, + "num_tokens": 423167111.0, + "step": 11089 + }, + { + "epoch": 1.410761989568757, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7928674221038818, + "num_tokens": 423202060.0, + "step": 11090 + }, + { + "epoch": 1.4108891998473476, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7885318994522095, + "num_tokens": 423241200.0, + "step": 11091 + }, + { + "epoch": 1.4110164101259381, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8153986930847168, + "num_tokens": 423284713.0, + "step": 11092 + }, + { + "epoch": 1.4111436204045287, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.799323320388794, + "num_tokens": 423322524.0, + "step": 11093 + }, + { + "epoch": 1.4112708306831192, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7976245880126953, + "num_tokens": 423360551.0, + "step": 11094 + }, + { + "epoch": 1.4113980409617097, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8034391403198242, + "num_tokens": 423395227.0, + "step": 11095 + }, + { + "epoch": 1.4115252512403003, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7859854698181152, + "num_tokens": 423434757.0, + "step": 11096 + }, + { + "epoch": 1.4116524615188908, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.5931, + "mean_token_accuracy": 0.8194390535354614, + "num_tokens": 423464599.0, + "step": 11097 + }, + { + "epoch": 1.4117796717974813, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7943053245544434, + "num_tokens": 423504327.0, + "step": 11098 + }, + { + "epoch": 1.4119068820760718, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.781879723072052, + "num_tokens": 423543114.0, + "step": 11099 + }, + { + "epoch": 1.4120340923546624, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7888991236686707, + "num_tokens": 423581727.0, + "step": 11100 + }, + { + "epoch": 1.412161302633253, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8046729564666748, + "num_tokens": 423617073.0, + "step": 11101 + }, + { + "epoch": 1.4122885129118432, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8040114641189575, + "num_tokens": 423664922.0, + "step": 11102 + }, + { + "epoch": 1.4124157231904337, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7953615784645081, + "num_tokens": 423706990.0, + "step": 11103 + }, + { + "epoch": 1.4125429334690243, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7782038450241089, + "num_tokens": 423745945.0, + "step": 11104 + }, + { + "epoch": 1.4126701437476148, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8033155202865601, + "num_tokens": 423780824.0, + "step": 11105 + }, + { + "epoch": 1.4127973540262053, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8168419599533081, + "num_tokens": 423818640.0, + "step": 11106 + }, + { + "epoch": 1.4129245643047958, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8044916391372681, + "num_tokens": 423856657.0, + "step": 11107 + }, + { + "epoch": 1.4130517745833864, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8150169849395752, + "num_tokens": 423900288.0, + "step": 11108 + }, + { + "epoch": 1.413178984861977, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8034380674362183, + "num_tokens": 423935929.0, + "step": 11109 + }, + { + "epoch": 1.4133061951405674, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8014739751815796, + "num_tokens": 423970719.0, + "step": 11110 + }, + { + "epoch": 1.4134334054191577, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7998796701431274, + "num_tokens": 424012338.0, + "step": 11111 + }, + { + "epoch": 1.4135606156977483, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7770214676856995, + "num_tokens": 424050112.0, + "step": 11112 + }, + { + "epoch": 1.4136878259763388, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.8016172647476196, + "num_tokens": 424092688.0, + "step": 11113 + }, + { + "epoch": 1.4138150362549293, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7979316711425781, + "num_tokens": 424131355.0, + "step": 11114 + }, + { + "epoch": 1.4139422465335199, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7910904288291931, + "num_tokens": 424177768.0, + "step": 11115 + }, + { + "epoch": 1.4140694568121104, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8139516115188599, + "num_tokens": 424215834.0, + "step": 11116 + }, + { + "epoch": 1.414196667090701, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8050526976585388, + "num_tokens": 424260002.0, + "step": 11117 + }, + { + "epoch": 1.4143238773692914, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.796689510345459, + "num_tokens": 424299343.0, + "step": 11118 + }, + { + "epoch": 1.414451087647882, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7917873859405518, + "num_tokens": 424340312.0, + "step": 11119 + }, + { + "epoch": 1.4145782979264725, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8131599426269531, + "num_tokens": 424376658.0, + "step": 11120 + }, + { + "epoch": 1.414705508205063, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.80704665184021, + "num_tokens": 424415499.0, + "step": 11121 + }, + { + "epoch": 1.4148327184836536, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7981499433517456, + "num_tokens": 424453401.0, + "step": 11122 + }, + { + "epoch": 1.414959928762244, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8047330379486084, + "num_tokens": 424490654.0, + "step": 11123 + }, + { + "epoch": 1.4150871390408346, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8115038871765137, + "num_tokens": 424534469.0, + "step": 11124 + }, + { + "epoch": 1.4152143493194251, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7863354682922363, + "num_tokens": 424570187.0, + "step": 11125 + }, + { + "epoch": 1.4153415595980154, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7744510173797607, + "num_tokens": 424607898.0, + "step": 11126 + }, + { + "epoch": 1.415468769876606, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8058308362960815, + "num_tokens": 424652156.0, + "step": 11127 + }, + { + "epoch": 1.4155959801551965, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7881591320037842, + "num_tokens": 424688735.0, + "step": 11128 + }, + { + "epoch": 1.415723190433787, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7978364825248718, + "num_tokens": 424732859.0, + "step": 11129 + }, + { + "epoch": 1.4158504007123776, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7930917739868164, + "num_tokens": 424769381.0, + "step": 11130 + }, + { + "epoch": 1.415977610990968, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7987160682678223, + "num_tokens": 424802362.0, + "step": 11131 + }, + { + "epoch": 1.4161048212695586, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.798019528388977, + "num_tokens": 424837897.0, + "step": 11132 + }, + { + "epoch": 1.4162320315481491, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8095676898956299, + "num_tokens": 424878154.0, + "step": 11133 + }, + { + "epoch": 1.4163592418267397, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7980425357818604, + "num_tokens": 424915063.0, + "step": 11134 + }, + { + "epoch": 1.41648645210533, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7994217276573181, + "num_tokens": 424955159.0, + "step": 11135 + }, + { + "epoch": 1.4166136623839205, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7942627668380737, + "num_tokens": 424992573.0, + "step": 11136 + }, + { + "epoch": 1.416740872662511, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8032013773918152, + "num_tokens": 425030452.0, + "step": 11137 + }, + { + "epoch": 1.4168680829411016, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8167059421539307, + "num_tokens": 425067118.0, + "step": 11138 + }, + { + "epoch": 1.416995293219692, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7920571565628052, + "num_tokens": 425101684.0, + "step": 11139 + }, + { + "epoch": 1.4171225034982826, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8106886744499207, + "num_tokens": 425143127.0, + "step": 11140 + }, + { + "epoch": 1.4172497137768731, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7930578589439392, + "num_tokens": 425173912.0, + "step": 11141 + }, + { + "epoch": 1.4173769240554637, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.801880955696106, + "num_tokens": 425210146.0, + "step": 11142 + }, + { + "epoch": 1.4175041343340542, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.816930890083313, + "num_tokens": 425246082.0, + "step": 11143 + }, + { + "epoch": 1.4176313446126447, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.800653338432312, + "num_tokens": 425287470.0, + "step": 11144 + }, + { + "epoch": 1.4177585548912353, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7963513731956482, + "num_tokens": 425326553.0, + "step": 11145 + }, + { + "epoch": 1.4178857651698258, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7945325374603271, + "num_tokens": 425363105.0, + "step": 11146 + }, + { + "epoch": 1.4180129754484163, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.797376811504364, + "num_tokens": 425402513.0, + "step": 11147 + }, + { + "epoch": 1.4181401857270068, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7998015880584717, + "num_tokens": 425440132.0, + "step": 11148 + }, + { + "epoch": 1.4182673960055974, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.798839807510376, + "num_tokens": 425481738.0, + "step": 11149 + }, + { + "epoch": 1.418394606284188, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8026136159896851, + "num_tokens": 425516916.0, + "step": 11150 + }, + { + "epoch": 1.4185218165627782, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.803127110004425, + "num_tokens": 425555951.0, + "step": 11151 + }, + { + "epoch": 1.4186490268413687, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.800855278968811, + "num_tokens": 425595302.0, + "step": 11152 + }, + { + "epoch": 1.4187762371199593, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8018274903297424, + "num_tokens": 425623281.0, + "step": 11153 + }, + { + "epoch": 1.4189034473985498, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8120144009590149, + "num_tokens": 425659124.0, + "step": 11154 + }, + { + "epoch": 1.4190306576771403, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7890186309814453, + "num_tokens": 425695809.0, + "step": 11155 + }, + { + "epoch": 1.4191578679557308, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7894316911697388, + "num_tokens": 425731890.0, + "step": 11156 + }, + { + "epoch": 1.4192850782343214, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8098248839378357, + "num_tokens": 425770899.0, + "step": 11157 + }, + { + "epoch": 1.419412288512912, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.8143662810325623, + "num_tokens": 425807812.0, + "step": 11158 + }, + { + "epoch": 1.4195394987915024, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7902839183807373, + "num_tokens": 425848997.0, + "step": 11159 + }, + { + "epoch": 1.4196667090700927, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8105157613754272, + "num_tokens": 425887816.0, + "step": 11160 + }, + { + "epoch": 1.4197939193486833, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7939125299453735, + "num_tokens": 425925488.0, + "step": 11161 + }, + { + "epoch": 1.4199211296272738, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8032273054122925, + "num_tokens": 425965616.0, + "step": 11162 + }, + { + "epoch": 1.4200483399058643, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7985928654670715, + "num_tokens": 426005308.0, + "step": 11163 + }, + { + "epoch": 1.4201755501844548, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7929956912994385, + "num_tokens": 426042339.0, + "step": 11164 + }, + { + "epoch": 1.4203027604630454, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8062078356742859, + "num_tokens": 426081925.0, + "step": 11165 + }, + { + "epoch": 1.420429970741636, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7954379320144653, + "num_tokens": 426120918.0, + "step": 11166 + }, + { + "epoch": 1.4205571810202264, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7874128818511963, + "num_tokens": 426158711.0, + "step": 11167 + }, + { + "epoch": 1.420684391298817, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7949373722076416, + "num_tokens": 426197034.0, + "step": 11168 + }, + { + "epoch": 1.4208116015774075, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7933001518249512, + "num_tokens": 426239390.0, + "step": 11169 + }, + { + "epoch": 1.420938811855998, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7904164791107178, + "num_tokens": 426285488.0, + "step": 11170 + }, + { + "epoch": 1.4210660221345885, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7844799160957336, + "num_tokens": 426323955.0, + "step": 11171 + }, + { + "epoch": 1.421193232413179, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.5929, + "mean_token_accuracy": 0.8234502673149109, + "num_tokens": 426359460.0, + "step": 11172 + }, + { + "epoch": 1.4213204426917696, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7913164496421814, + "num_tokens": 426397104.0, + "step": 11173 + }, + { + "epoch": 1.4214476529703601, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8118414878845215, + "num_tokens": 426441349.0, + "step": 11174 + }, + { + "epoch": 1.4215748632489504, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8045440912246704, + "num_tokens": 426478410.0, + "step": 11175 + }, + { + "epoch": 1.421702073527541, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7964395880699158, + "num_tokens": 426519874.0, + "step": 11176 + }, + { + "epoch": 1.4218292838061315, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7916278839111328, + "num_tokens": 426559970.0, + "step": 11177 + }, + { + "epoch": 1.421956494084722, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7935464382171631, + "num_tokens": 426593897.0, + "step": 11178 + }, + { + "epoch": 1.4220837043633126, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8017492294311523, + "num_tokens": 426633712.0, + "step": 11179 + }, + { + "epoch": 1.422210914641903, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8141415119171143, + "num_tokens": 426669085.0, + "step": 11180 + }, + { + "epoch": 1.4223381249204936, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.820284366607666, + "num_tokens": 426709260.0, + "step": 11181 + }, + { + "epoch": 1.4224653351990841, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7906501293182373, + "num_tokens": 426747747.0, + "step": 11182 + }, + { + "epoch": 1.4225925454776747, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7870575189590454, + "num_tokens": 426788185.0, + "step": 11183 + }, + { + "epoch": 1.422719755756265, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8008455038070679, + "num_tokens": 426828663.0, + "step": 11184 + }, + { + "epoch": 1.4228469660348555, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7886245250701904, + "num_tokens": 426869205.0, + "step": 11185 + }, + { + "epoch": 1.422974176313446, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8124374747276306, + "num_tokens": 426900561.0, + "step": 11186 + }, + { + "epoch": 1.4231013865920366, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.822691798210144, + "num_tokens": 426940618.0, + "step": 11187 + }, + { + "epoch": 1.423228596870627, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7913678884506226, + "num_tokens": 426975509.0, + "step": 11188 + }, + { + "epoch": 1.4233558071492176, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8098663687705994, + "num_tokens": 427011647.0, + "step": 11189 + }, + { + "epoch": 1.4234830174278081, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.793104350566864, + "num_tokens": 427046631.0, + "step": 11190 + }, + { + "epoch": 1.4236102277063987, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8010766506195068, + "num_tokens": 427081364.0, + "step": 11191 + }, + { + "epoch": 1.4237374379849892, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8052731156349182, + "num_tokens": 427118667.0, + "step": 11192 + }, + { + "epoch": 1.4238646482635797, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8016782999038696, + "num_tokens": 427155017.0, + "step": 11193 + }, + { + "epoch": 1.4239918585421703, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.792843222618103, + "num_tokens": 427191741.0, + "step": 11194 + }, + { + "epoch": 1.4241190688207608, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8039605617523193, + "num_tokens": 427226210.0, + "step": 11195 + }, + { + "epoch": 1.4242462790993513, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7976863980293274, + "num_tokens": 427261699.0, + "step": 11196 + }, + { + "epoch": 1.4243734893779418, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7922093272209167, + "num_tokens": 427297158.0, + "step": 11197 + }, + { + "epoch": 1.4245006996565324, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7810647487640381, + "num_tokens": 427338234.0, + "step": 11198 + }, + { + "epoch": 1.424627909935123, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7941526174545288, + "num_tokens": 427376727.0, + "step": 11199 + }, + { + "epoch": 1.4247551202137132, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8181899785995483, + "num_tokens": 427417095.0, + "step": 11200 + }, + { + "epoch": 1.4248823304923037, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7848433256149292, + "num_tokens": 427453243.0, + "step": 11201 + }, + { + "epoch": 1.4250095407708943, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8040740489959717, + "num_tokens": 427490961.0, + "step": 11202 + }, + { + "epoch": 1.4251367510494848, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8091059327125549, + "num_tokens": 427521487.0, + "step": 11203 + }, + { + "epoch": 1.4252639613280753, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7909700870513916, + "num_tokens": 427556607.0, + "step": 11204 + }, + { + "epoch": 1.4253911716066658, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8014524579048157, + "num_tokens": 427598940.0, + "step": 11205 + }, + { + "epoch": 1.4255183818852564, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7860262393951416, + "num_tokens": 427641337.0, + "step": 11206 + }, + { + "epoch": 1.425645592163847, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8179875612258911, + "num_tokens": 427680235.0, + "step": 11207 + }, + { + "epoch": 1.4257728024424374, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7984717488288879, + "num_tokens": 427719918.0, + "step": 11208 + }, + { + "epoch": 1.4259000127210277, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.802453875541687, + "num_tokens": 427758042.0, + "step": 11209 + }, + { + "epoch": 1.4260272229996183, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8045845031738281, + "num_tokens": 427794591.0, + "step": 11210 + }, + { + "epoch": 1.4261544332782088, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8046597838401794, + "num_tokens": 427828190.0, + "step": 11211 + }, + { + "epoch": 1.4262816435567993, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7924090623855591, + "num_tokens": 427870363.0, + "step": 11212 + }, + { + "epoch": 1.4264088538353898, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8176842927932739, + "num_tokens": 427910037.0, + "step": 11213 + }, + { + "epoch": 1.4265360641139804, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8042236566543579, + "num_tokens": 427945782.0, + "step": 11214 + }, + { + "epoch": 1.426663274392571, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8081265091896057, + "num_tokens": 427984635.0, + "step": 11215 + }, + { + "epoch": 1.4267904846711614, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8168481588363647, + "num_tokens": 428024289.0, + "step": 11216 + }, + { + "epoch": 1.426917694949752, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7964068055152893, + "num_tokens": 428067119.0, + "step": 11217 + }, + { + "epoch": 1.4270449052283425, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7958990335464478, + "num_tokens": 428108150.0, + "step": 11218 + }, + { + "epoch": 1.427172115506933, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8133598566055298, + "num_tokens": 428150839.0, + "step": 11219 + }, + { + "epoch": 1.4272993257855235, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7914072275161743, + "num_tokens": 428194325.0, + "step": 11220 + }, + { + "epoch": 1.427426536064114, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7969366908073425, + "num_tokens": 428230879.0, + "step": 11221 + }, + { + "epoch": 1.4275537463427046, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8095009922981262, + "num_tokens": 428265795.0, + "step": 11222 + }, + { + "epoch": 1.4276809566212951, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8051800727844238, + "num_tokens": 428300413.0, + "step": 11223 + }, + { + "epoch": 1.4278081668998854, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.8209056854248047, + "num_tokens": 428339293.0, + "step": 11224 + }, + { + "epoch": 1.427935377178476, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8091498613357544, + "num_tokens": 428374158.0, + "step": 11225 + }, + { + "epoch": 1.4280625874570665, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.801967442035675, + "num_tokens": 428411715.0, + "step": 11226 + }, + { + "epoch": 1.428189797735657, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7858906984329224, + "num_tokens": 428450000.0, + "step": 11227 + }, + { + "epoch": 1.4283170080142475, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8049805760383606, + "num_tokens": 428484546.0, + "step": 11228 + }, + { + "epoch": 1.428444218292838, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7989651560783386, + "num_tokens": 428527519.0, + "step": 11229 + }, + { + "epoch": 1.4285714285714286, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7983900308609009, + "num_tokens": 428568744.0, + "step": 11230 + }, + { + "epoch": 1.4286986388500191, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8016107678413391, + "num_tokens": 428613050.0, + "step": 11231 + }, + { + "epoch": 1.4288258491286097, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7992014288902283, + "num_tokens": 428647334.0, + "step": 11232 + }, + { + "epoch": 1.4289530594072, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.803169846534729, + "num_tokens": 428681015.0, + "step": 11233 + }, + { + "epoch": 1.4290802696857905, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7959386706352234, + "num_tokens": 428718903.0, + "step": 11234 + }, + { + "epoch": 1.429207479964381, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8141621947288513, + "num_tokens": 428756038.0, + "step": 11235 + }, + { + "epoch": 1.4293346902429716, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8177357912063599, + "num_tokens": 428793908.0, + "step": 11236 + }, + { + "epoch": 1.429461900521562, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7930682897567749, + "num_tokens": 428830924.0, + "step": 11237 + }, + { + "epoch": 1.4295891108001526, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7968630790710449, + "num_tokens": 428869083.0, + "step": 11238 + }, + { + "epoch": 1.4297163210787431, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7693898677825928, + "num_tokens": 428903970.0, + "step": 11239 + }, + { + "epoch": 1.4298435313573337, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.7996854782104492, + "num_tokens": 428940710.0, + "step": 11240 + }, + { + "epoch": 1.4299707416359242, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8051514625549316, + "num_tokens": 428981160.0, + "step": 11241 + }, + { + "epoch": 1.4300979519145147, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8049713373184204, + "num_tokens": 429022464.0, + "step": 11242 + }, + { + "epoch": 1.4302251621931052, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8041833639144897, + "num_tokens": 429060964.0, + "step": 11243 + }, + { + "epoch": 1.4303523724716958, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.803483247756958, + "num_tokens": 429101389.0, + "step": 11244 + }, + { + "epoch": 1.4304795827502863, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7967584729194641, + "num_tokens": 429135547.0, + "step": 11245 + }, + { + "epoch": 1.4306067930288768, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8062616586685181, + "num_tokens": 429177504.0, + "step": 11246 + }, + { + "epoch": 1.4307340033074674, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8015962839126587, + "num_tokens": 429217923.0, + "step": 11247 + }, + { + "epoch": 1.430861213586058, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7956501245498657, + "num_tokens": 429251514.0, + "step": 11248 + }, + { + "epoch": 1.4309884238646482, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.782259464263916, + "num_tokens": 429291747.0, + "step": 11249 + }, + { + "epoch": 1.4311156341432387, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.798820972442627, + "num_tokens": 429333480.0, + "step": 11250 + }, + { + "epoch": 1.4312428444218293, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8157604932785034, + "num_tokens": 429366482.0, + "step": 11251 + }, + { + "epoch": 1.4313700547004198, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7932518124580383, + "num_tokens": 429400704.0, + "step": 11252 + }, + { + "epoch": 1.4314972649790103, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7863229513168335, + "num_tokens": 429437463.0, + "step": 11253 + }, + { + "epoch": 1.4316244752576008, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7869335412979126, + "num_tokens": 429474496.0, + "step": 11254 + }, + { + "epoch": 1.4317516855361914, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7864207029342651, + "num_tokens": 429514641.0, + "step": 11255 + }, + { + "epoch": 1.431878895814782, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8008894920349121, + "num_tokens": 429554054.0, + "step": 11256 + }, + { + "epoch": 1.4320061060933724, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7927930355072021, + "num_tokens": 429594004.0, + "step": 11257 + }, + { + "epoch": 1.4321333163719627, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.81053626537323, + "num_tokens": 429633019.0, + "step": 11258 + }, + { + "epoch": 1.4322605266505533, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7967009544372559, + "num_tokens": 429666869.0, + "step": 11259 + }, + { + "epoch": 1.4323877369291438, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8130906224250793, + "num_tokens": 429700049.0, + "step": 11260 + }, + { + "epoch": 1.4325149472077343, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7932577133178711, + "num_tokens": 429737517.0, + "step": 11261 + }, + { + "epoch": 1.4326421574863248, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8166736364364624, + "num_tokens": 429778697.0, + "step": 11262 + }, + { + "epoch": 1.4327693677649154, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8021726608276367, + "num_tokens": 429811883.0, + "step": 11263 + }, + { + "epoch": 1.432896578043506, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8060214519500732, + "num_tokens": 429851748.0, + "step": 11264 + }, + { + "epoch": 1.4330237883220964, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7900742888450623, + "num_tokens": 429891033.0, + "step": 11265 + }, + { + "epoch": 1.433150998600687, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8036550879478455, + "num_tokens": 429935980.0, + "step": 11266 + }, + { + "epoch": 1.4332782088792775, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7896325588226318, + "num_tokens": 429970874.0, + "step": 11267 + }, + { + "epoch": 1.433405419157868, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7870721817016602, + "num_tokens": 430017909.0, + "step": 11268 + }, + { + "epoch": 1.4335326294364585, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7739178538322449, + "num_tokens": 430050641.0, + "step": 11269 + }, + { + "epoch": 1.433659839715049, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7997976541519165, + "num_tokens": 430082985.0, + "step": 11270 + }, + { + "epoch": 1.4337870499936396, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8012422323226929, + "num_tokens": 430122340.0, + "step": 11271 + }, + { + "epoch": 1.4339142602722301, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7814205884933472, + "num_tokens": 430162140.0, + "step": 11272 + }, + { + "epoch": 1.4340414705508204, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8159155249595642, + "num_tokens": 430201672.0, + "step": 11273 + }, + { + "epoch": 1.434168680829411, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7910157442092896, + "num_tokens": 430239760.0, + "step": 11274 + }, + { + "epoch": 1.4342958911080015, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7816664576530457, + "num_tokens": 430280527.0, + "step": 11275 + }, + { + "epoch": 1.434423101386592, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.796504020690918, + "num_tokens": 430312788.0, + "step": 11276 + }, + { + "epoch": 1.4345503116651825, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8060293197631836, + "num_tokens": 430348044.0, + "step": 11277 + }, + { + "epoch": 1.434677521943773, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8185300827026367, + "num_tokens": 430385676.0, + "step": 11278 + }, + { + "epoch": 1.4348047322223636, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7924661636352539, + "num_tokens": 430416976.0, + "step": 11279 + }, + { + "epoch": 1.4349319425009541, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8126760125160217, + "num_tokens": 430460002.0, + "step": 11280 + }, + { + "epoch": 1.4350591527795447, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7772403359413147, + "num_tokens": 430503172.0, + "step": 11281 + }, + { + "epoch": 1.435186363058135, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7862387895584106, + "num_tokens": 430541221.0, + "step": 11282 + }, + { + "epoch": 1.4353135733367255, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7948583364486694, + "num_tokens": 430579527.0, + "step": 11283 + }, + { + "epoch": 1.435440783615316, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7986890077590942, + "num_tokens": 430617263.0, + "step": 11284 + }, + { + "epoch": 1.4355679938939065, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.811699628829956, + "num_tokens": 430649961.0, + "step": 11285 + }, + { + "epoch": 1.435695204172497, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7930759191513062, + "num_tokens": 430682599.0, + "step": 11286 + }, + { + "epoch": 1.4358224144510876, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7694166302680969, + "num_tokens": 430721014.0, + "step": 11287 + }, + { + "epoch": 1.4359496247296781, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7995754480361938, + "num_tokens": 430757014.0, + "step": 11288 + }, + { + "epoch": 1.4360768350082687, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.796224057674408, + "num_tokens": 430796169.0, + "step": 11289 + }, + { + "epoch": 1.4362040452868592, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7983784675598145, + "num_tokens": 430832726.0, + "step": 11290 + }, + { + "epoch": 1.4363312555654497, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7913714647293091, + "num_tokens": 430866343.0, + "step": 11291 + }, + { + "epoch": 1.4364584658440402, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8052960634231567, + "num_tokens": 430901622.0, + "step": 11292 + }, + { + "epoch": 1.4365856761226308, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8139243721961975, + "num_tokens": 430937499.0, + "step": 11293 + }, + { + "epoch": 1.4367128864012213, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7887071371078491, + "num_tokens": 430974048.0, + "step": 11294 + }, + { + "epoch": 1.4368400966798118, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.815139651298523, + "num_tokens": 431010927.0, + "step": 11295 + }, + { + "epoch": 1.4369673069584024, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7841206789016724, + "num_tokens": 431051731.0, + "step": 11296 + }, + { + "epoch": 1.4370945172369929, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8186457753181458, + "num_tokens": 431087876.0, + "step": 11297 + }, + { + "epoch": 1.4372217275155832, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7887881994247437, + "num_tokens": 431126217.0, + "step": 11298 + }, + { + "epoch": 1.4373489377941737, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7971670031547546, + "num_tokens": 431168049.0, + "step": 11299 + }, + { + "epoch": 1.4374761480727642, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7953311204910278, + "num_tokens": 431211143.0, + "step": 11300 + }, + { + "epoch": 1.4376033583513548, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.809904158115387, + "num_tokens": 431255747.0, + "step": 11301 + }, + { + "epoch": 1.4377305686299453, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8043828010559082, + "num_tokens": 431296618.0, + "step": 11302 + }, + { + "epoch": 1.4378577789085358, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7971819043159485, + "num_tokens": 431336493.0, + "step": 11303 + }, + { + "epoch": 1.4379849891871264, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7929548025131226, + "num_tokens": 431376200.0, + "step": 11304 + }, + { + "epoch": 1.438112199465717, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7815050482749939, + "num_tokens": 431414015.0, + "step": 11305 + }, + { + "epoch": 1.4382394097443074, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.792312741279602, + "num_tokens": 431453498.0, + "step": 11306 + }, + { + "epoch": 1.4383666200228977, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8074172735214233, + "num_tokens": 431491126.0, + "step": 11307 + }, + { + "epoch": 1.4384938303014883, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.5736, + "mean_token_accuracy": 0.8265913128852844, + "num_tokens": 431528549.0, + "step": 11308 + }, + { + "epoch": 1.4386210405800788, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.814010500907898, + "num_tokens": 431569435.0, + "step": 11309 + }, + { + "epoch": 1.4387482508586693, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8051449060440063, + "num_tokens": 431606125.0, + "step": 11310 + }, + { + "epoch": 1.4388754611372598, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.792730450630188, + "num_tokens": 431643517.0, + "step": 11311 + }, + { + "epoch": 1.4390026714158504, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8089828491210938, + "num_tokens": 431682058.0, + "step": 11312 + }, + { + "epoch": 1.439129881694441, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8132360577583313, + "num_tokens": 431721241.0, + "step": 11313 + }, + { + "epoch": 1.4392570919730314, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8153600692749023, + "num_tokens": 431757809.0, + "step": 11314 + }, + { + "epoch": 1.439384302251622, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7895455360412598, + "num_tokens": 431795931.0, + "step": 11315 + }, + { + "epoch": 1.4395115125302125, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8100522756576538, + "num_tokens": 431830517.0, + "step": 11316 + }, + { + "epoch": 1.439638722808803, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8010649085044861, + "num_tokens": 431866047.0, + "step": 11317 + }, + { + "epoch": 1.4397659330873935, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7869752049446106, + "num_tokens": 431901602.0, + "step": 11318 + }, + { + "epoch": 1.439893143365984, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7952278256416321, + "num_tokens": 431934989.0, + "step": 11319 + }, + { + "epoch": 1.4400203536445746, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8024499416351318, + "num_tokens": 431968664.0, + "step": 11320 + }, + { + "epoch": 1.4401475639231651, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8127896785736084, + "num_tokens": 432008025.0, + "step": 11321 + }, + { + "epoch": 1.4402747742017554, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.800083339214325, + "num_tokens": 432043526.0, + "step": 11322 + }, + { + "epoch": 1.440401984480346, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8026379346847534, + "num_tokens": 432085175.0, + "step": 11323 + }, + { + "epoch": 1.4405291947589365, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.791077733039856, + "num_tokens": 432122379.0, + "step": 11324 + }, + { + "epoch": 1.440656405037527, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8023274540901184, + "num_tokens": 432157690.0, + "step": 11325 + }, + { + "epoch": 1.4407836153161175, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8076715469360352, + "num_tokens": 432188885.0, + "step": 11326 + }, + { + "epoch": 1.440910825594708, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8024735450744629, + "num_tokens": 432228804.0, + "step": 11327 + }, + { + "epoch": 1.4410380358732986, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7817173600196838, + "num_tokens": 432264150.0, + "step": 11328 + }, + { + "epoch": 1.4411652461518891, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7995786666870117, + "num_tokens": 432306759.0, + "step": 11329 + }, + { + "epoch": 1.4412924564304797, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8105307817459106, + "num_tokens": 432346928.0, + "step": 11330 + }, + { + "epoch": 1.44141966670907, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.809903621673584, + "num_tokens": 432386630.0, + "step": 11331 + }, + { + "epoch": 1.4415468769876605, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7897592186927795, + "num_tokens": 432427331.0, + "step": 11332 + }, + { + "epoch": 1.441674087266251, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7928162217140198, + "num_tokens": 432462901.0, + "step": 11333 + }, + { + "epoch": 1.4418012975448415, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7942070960998535, + "num_tokens": 432494232.0, + "step": 11334 + }, + { + "epoch": 1.441928507823432, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8143938183784485, + "num_tokens": 432534276.0, + "step": 11335 + }, + { + "epoch": 1.4420557181020226, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7944445610046387, + "num_tokens": 432577337.0, + "step": 11336 + }, + { + "epoch": 1.4421829283806131, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7932112216949463, + "num_tokens": 432614290.0, + "step": 11337 + }, + { + "epoch": 1.4423101386592037, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7972538471221924, + "num_tokens": 432648558.0, + "step": 11338 + }, + { + "epoch": 1.4424373489377942, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7703671455383301, + "num_tokens": 432684243.0, + "step": 11339 + }, + { + "epoch": 1.4425645592163847, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7986963987350464, + "num_tokens": 432719355.0, + "step": 11340 + }, + { + "epoch": 1.4426917694949752, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.8016350269317627, + "num_tokens": 432758675.0, + "step": 11341 + }, + { + "epoch": 1.4428189797735658, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7993135452270508, + "num_tokens": 432801677.0, + "step": 11342 + }, + { + "epoch": 1.4429461900521563, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7979596257209778, + "num_tokens": 432836299.0, + "step": 11343 + }, + { + "epoch": 1.4430734003307468, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7965505123138428, + "num_tokens": 432868838.0, + "step": 11344 + }, + { + "epoch": 1.4432006106093374, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8032629489898682, + "num_tokens": 432904895.0, + "step": 11345 + }, + { + "epoch": 1.4433278208879279, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7992146611213684, + "num_tokens": 432944785.0, + "step": 11346 + }, + { + "epoch": 1.4434550311665182, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7852931022644043, + "num_tokens": 432975645.0, + "step": 11347 + }, + { + "epoch": 1.4435822414451087, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7950747013092041, + "num_tokens": 433016887.0, + "step": 11348 + }, + { + "epoch": 1.4437094517236992, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.799647331237793, + "num_tokens": 433053666.0, + "step": 11349 + }, + { + "epoch": 1.4438366620022898, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.801118016242981, + "num_tokens": 433101131.0, + "step": 11350 + }, + { + "epoch": 1.4439638722808803, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8043560981750488, + "num_tokens": 433132535.0, + "step": 11351 + }, + { + "epoch": 1.4440910825594708, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7794850468635559, + "num_tokens": 433168229.0, + "step": 11352 + }, + { + "epoch": 1.4442182928380614, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8007369637489319, + "num_tokens": 433206094.0, + "step": 11353 + }, + { + "epoch": 1.4443455031166519, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7827588319778442, + "num_tokens": 433241038.0, + "step": 11354 + }, + { + "epoch": 1.4444727133952424, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7886669635772705, + "num_tokens": 433274674.0, + "step": 11355 + }, + { + "epoch": 1.4445999236738327, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8045939803123474, + "num_tokens": 433314511.0, + "step": 11356 + }, + { + "epoch": 1.4447271339524232, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8029722571372986, + "num_tokens": 433349338.0, + "step": 11357 + }, + { + "epoch": 1.4448543442310138, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8063899278640747, + "num_tokens": 433382609.0, + "step": 11358 + }, + { + "epoch": 1.4449815545096043, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8093628883361816, + "num_tokens": 433411553.0, + "step": 11359 + }, + { + "epoch": 1.4451087647881948, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7938904762268066, + "num_tokens": 433450571.0, + "step": 11360 + }, + { + "epoch": 1.4452359750667854, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.792708158493042, + "num_tokens": 433486994.0, + "step": 11361 + }, + { + "epoch": 1.445363185345376, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.812889039516449, + "num_tokens": 433521407.0, + "step": 11362 + }, + { + "epoch": 1.4454903956239664, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8149735927581787, + "num_tokens": 433561338.0, + "step": 11363 + }, + { + "epoch": 1.445617605902557, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8107757568359375, + "num_tokens": 433598892.0, + "step": 11364 + }, + { + "epoch": 1.4457448161811475, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7896554470062256, + "num_tokens": 433641423.0, + "step": 11365 + }, + { + "epoch": 1.445872026459738, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8007320165634155, + "num_tokens": 433677385.0, + "step": 11366 + }, + { + "epoch": 1.4459992367383285, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8032429218292236, + "num_tokens": 433716901.0, + "step": 11367 + }, + { + "epoch": 1.446126447016919, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7932690382003784, + "num_tokens": 433755099.0, + "step": 11368 + }, + { + "epoch": 1.4462536572955096, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7997093200683594, + "num_tokens": 433794012.0, + "step": 11369 + }, + { + "epoch": 1.4463808675741001, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8020935654640198, + "num_tokens": 433831965.0, + "step": 11370 + }, + { + "epoch": 1.4465080778526904, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.8016355633735657, + "num_tokens": 433870819.0, + "step": 11371 + }, + { + "epoch": 1.446635288131281, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.792646050453186, + "num_tokens": 433909197.0, + "step": 11372 + }, + { + "epoch": 1.4467624984098715, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8045858144760132, + "num_tokens": 433946026.0, + "step": 11373 + }, + { + "epoch": 1.446889708688462, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8233495950698853, + "num_tokens": 433986239.0, + "step": 11374 + }, + { + "epoch": 1.4470169189670525, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8037842512130737, + "num_tokens": 434022557.0, + "step": 11375 + }, + { + "epoch": 1.447144129245643, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8019196391105652, + "num_tokens": 434059258.0, + "step": 11376 + }, + { + "epoch": 1.4472713395242336, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7759308218955994, + "num_tokens": 434094872.0, + "step": 11377 + }, + { + "epoch": 1.4473985498028241, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7729848623275757, + "num_tokens": 434134842.0, + "step": 11378 + }, + { + "epoch": 1.4475257600814146, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8098490238189697, + "num_tokens": 434173152.0, + "step": 11379 + }, + { + "epoch": 1.447652970360005, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8055434226989746, + "num_tokens": 434208318.0, + "step": 11380 + }, + { + "epoch": 1.4477801806385955, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7958920001983643, + "num_tokens": 434243693.0, + "step": 11381 + }, + { + "epoch": 1.447907390917186, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7905054092407227, + "num_tokens": 434283172.0, + "step": 11382 + }, + { + "epoch": 1.4480346011957765, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7920278906822205, + "num_tokens": 434318789.0, + "step": 11383 + }, + { + "epoch": 1.448161811474367, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7942631840705872, + "num_tokens": 434358943.0, + "step": 11384 + }, + { + "epoch": 1.4482890217529576, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7978514432907104, + "num_tokens": 434398264.0, + "step": 11385 + }, + { + "epoch": 1.4484162320315481, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8061126470565796, + "num_tokens": 434433602.0, + "step": 11386 + }, + { + "epoch": 1.4485434423101387, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7997984290122986, + "num_tokens": 434480001.0, + "step": 11387 + }, + { + "epoch": 1.4486706525887292, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8175453543663025, + "num_tokens": 434519285.0, + "step": 11388 + }, + { + "epoch": 1.4487978628673197, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8003277778625488, + "num_tokens": 434556118.0, + "step": 11389 + }, + { + "epoch": 1.4489250731459102, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7806988954544067, + "num_tokens": 434595350.0, + "step": 11390 + }, + { + "epoch": 1.4490522834245008, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7990660071372986, + "num_tokens": 434632988.0, + "step": 11391 + }, + { + "epoch": 1.4491794937030913, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8053605556488037, + "num_tokens": 434673040.0, + "step": 11392 + }, + { + "epoch": 1.4493067039816818, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8173519372940063, + "num_tokens": 434712201.0, + "step": 11393 + }, + { + "epoch": 1.4494339142602723, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.817436158657074, + "num_tokens": 434756676.0, + "step": 11394 + }, + { + "epoch": 1.4495611245388629, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8021848201751709, + "num_tokens": 434797890.0, + "step": 11395 + }, + { + "epoch": 1.4496883348174532, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8143081665039062, + "num_tokens": 434836933.0, + "step": 11396 + }, + { + "epoch": 1.4498155450960437, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8058800101280212, + "num_tokens": 434868098.0, + "step": 11397 + }, + { + "epoch": 1.4499427553746342, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7951224446296692, + "num_tokens": 434907239.0, + "step": 11398 + }, + { + "epoch": 1.4500699656532248, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.8054453134536743, + "num_tokens": 434945672.0, + "step": 11399 + }, + { + "epoch": 1.4501971759318153, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7888662219047546, + "num_tokens": 434981693.0, + "step": 11400 + }, + { + "epoch": 1.4503243862104058, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8186678290367126, + "num_tokens": 435012452.0, + "step": 11401 + }, + { + "epoch": 1.4504515964889964, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7906770706176758, + "num_tokens": 435052711.0, + "step": 11402 + }, + { + "epoch": 1.4505788067675869, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7981500625610352, + "num_tokens": 435092766.0, + "step": 11403 + }, + { + "epoch": 1.4507060170461774, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7826426029205322, + "num_tokens": 435132093.0, + "step": 11404 + }, + { + "epoch": 1.4508332273247677, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7945255637168884, + "num_tokens": 435169071.0, + "step": 11405 + }, + { + "epoch": 1.4509604376033582, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8079808950424194, + "num_tokens": 435202665.0, + "step": 11406 + }, + { + "epoch": 1.4510876478819488, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.812159538269043, + "num_tokens": 435234719.0, + "step": 11407 + }, + { + "epoch": 1.4512148581605393, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8151828050613403, + "num_tokens": 435267090.0, + "step": 11408 + }, + { + "epoch": 1.4513420684391298, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8050402402877808, + "num_tokens": 435307243.0, + "step": 11409 + }, + { + "epoch": 1.4514692787177204, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.810492753982544, + "num_tokens": 435347507.0, + "step": 11410 + }, + { + "epoch": 1.4515964889963109, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7917523384094238, + "num_tokens": 435383499.0, + "step": 11411 + }, + { + "epoch": 1.4517236992749014, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7997334003448486, + "num_tokens": 435423500.0, + "step": 11412 + }, + { + "epoch": 1.451850909553492, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7997207045555115, + "num_tokens": 435457075.0, + "step": 11413 + }, + { + "epoch": 1.4519781198320825, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8007889986038208, + "num_tokens": 435498933.0, + "step": 11414 + }, + { + "epoch": 1.452105330110673, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7893972396850586, + "num_tokens": 435535949.0, + "step": 11415 + }, + { + "epoch": 1.4522325403892635, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7819275259971619, + "num_tokens": 435565252.0, + "step": 11416 + }, + { + "epoch": 1.452359750667854, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7813977599143982, + "num_tokens": 435604663.0, + "step": 11417 + }, + { + "epoch": 1.4524869609464446, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7892130017280579, + "num_tokens": 435640883.0, + "step": 11418 + }, + { + "epoch": 1.4526141712250351, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8023480176925659, + "num_tokens": 435682787.0, + "step": 11419 + }, + { + "epoch": 1.4527413815036254, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7951670289039612, + "num_tokens": 435719982.0, + "step": 11420 + }, + { + "epoch": 1.452868591782216, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7831442356109619, + "num_tokens": 435755153.0, + "step": 11421 + }, + { + "epoch": 1.4529958020608065, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8103103637695312, + "num_tokens": 435797513.0, + "step": 11422 + }, + { + "epoch": 1.453123012339397, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7967755794525146, + "num_tokens": 435836835.0, + "step": 11423 + }, + { + "epoch": 1.4532502226179875, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7946192026138306, + "num_tokens": 435875124.0, + "step": 11424 + }, + { + "epoch": 1.453377432896578, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8117334842681885, + "num_tokens": 435914993.0, + "step": 11425 + }, + { + "epoch": 1.4535046431751686, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8080819249153137, + "num_tokens": 435950916.0, + "step": 11426 + }, + { + "epoch": 1.4536318534537591, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7933452725410461, + "num_tokens": 435991986.0, + "step": 11427 + }, + { + "epoch": 1.4537590637323496, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.802932620048523, + "num_tokens": 436024603.0, + "step": 11428 + }, + { + "epoch": 1.45388627401094, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.797478973865509, + "num_tokens": 436065540.0, + "step": 11429 + }, + { + "epoch": 1.4540134842895305, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7867533564567566, + "num_tokens": 436104394.0, + "step": 11430 + }, + { + "epoch": 1.454140694568121, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7824534177780151, + "num_tokens": 436139977.0, + "step": 11431 + }, + { + "epoch": 1.4542679048467115, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.8171151280403137, + "num_tokens": 436178047.0, + "step": 11432 + }, + { + "epoch": 1.454395115125302, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8028380870819092, + "num_tokens": 436218331.0, + "step": 11433 + }, + { + "epoch": 1.4545223254038926, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7993730306625366, + "num_tokens": 436255663.0, + "step": 11434 + }, + { + "epoch": 1.4546495356824831, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8024574518203735, + "num_tokens": 436301328.0, + "step": 11435 + }, + { + "epoch": 1.4547767459610736, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8064994812011719, + "num_tokens": 436336487.0, + "step": 11436 + }, + { + "epoch": 1.4549039562396642, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8018635511398315, + "num_tokens": 436377629.0, + "step": 11437 + }, + { + "epoch": 1.4550311665182547, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8177420496940613, + "num_tokens": 436421590.0, + "step": 11438 + }, + { + "epoch": 1.4551583767968452, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7871675491333008, + "num_tokens": 436461172.0, + "step": 11439 + }, + { + "epoch": 1.4552855870754358, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8113007545471191, + "num_tokens": 436497536.0, + "step": 11440 + }, + { + "epoch": 1.4554127973540263, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8018006682395935, + "num_tokens": 436543906.0, + "step": 11441 + }, + { + "epoch": 1.4555400076326168, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8015692830085754, + "num_tokens": 436579462.0, + "step": 11442 + }, + { + "epoch": 1.4556672179112073, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7938531041145325, + "num_tokens": 436618840.0, + "step": 11443 + }, + { + "epoch": 1.4557944281897979, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7864913940429688, + "num_tokens": 436654176.0, + "step": 11444 + }, + { + "epoch": 1.4559216384683882, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.8150234222412109, + "num_tokens": 436685507.0, + "step": 11445 + }, + { + "epoch": 1.4560488487469787, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8087478876113892, + "num_tokens": 436729881.0, + "step": 11446 + }, + { + "epoch": 1.4561760590255692, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.8013290166854858, + "num_tokens": 436772355.0, + "step": 11447 + }, + { + "epoch": 1.4563032693041598, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7911519408226013, + "num_tokens": 436808701.0, + "step": 11448 + }, + { + "epoch": 1.4564304795827503, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8068503737449646, + "num_tokens": 436845441.0, + "step": 11449 + }, + { + "epoch": 1.4565576898613408, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8068440556526184, + "num_tokens": 436885316.0, + "step": 11450 + }, + { + "epoch": 1.4566849001399313, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7897685170173645, + "num_tokens": 436918315.0, + "step": 11451 + }, + { + "epoch": 1.4568121104185219, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7898736000061035, + "num_tokens": 436956039.0, + "step": 11452 + }, + { + "epoch": 1.4569393206971124, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7825651168823242, + "num_tokens": 436991274.0, + "step": 11453 + }, + { + "epoch": 1.4570665309757027, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.813561737537384, + "num_tokens": 437029963.0, + "step": 11454 + }, + { + "epoch": 1.4571937412542932, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7941149473190308, + "num_tokens": 437067030.0, + "step": 11455 + }, + { + "epoch": 1.4573209515328838, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8032869100570679, + "num_tokens": 437103363.0, + "step": 11456 + }, + { + "epoch": 1.4574481618114743, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8023786544799805, + "num_tokens": 437131574.0, + "step": 11457 + }, + { + "epoch": 1.4575753720900648, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7969828844070435, + "num_tokens": 437169537.0, + "step": 11458 + }, + { + "epoch": 1.4577025823686554, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.802993655204773, + "num_tokens": 437208025.0, + "step": 11459 + }, + { + "epoch": 1.4578297926472459, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.8124727606773376, + "num_tokens": 437246642.0, + "step": 11460 + }, + { + "epoch": 1.4579570029258364, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7919856905937195, + "num_tokens": 437285274.0, + "step": 11461 + }, + { + "epoch": 1.458084213204427, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8147187232971191, + "num_tokens": 437325865.0, + "step": 11462 + }, + { + "epoch": 1.4582114234830175, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8101704120635986, + "num_tokens": 437364093.0, + "step": 11463 + }, + { + "epoch": 1.458338633761608, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7879637479782104, + "num_tokens": 437402413.0, + "step": 11464 + }, + { + "epoch": 1.4584658440401985, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7837402820587158, + "num_tokens": 437442583.0, + "step": 11465 + }, + { + "epoch": 1.458593054318789, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7964991331100464, + "num_tokens": 437485682.0, + "step": 11466 + }, + { + "epoch": 1.4587202645973796, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7876536846160889, + "num_tokens": 437524859.0, + "step": 11467 + }, + { + "epoch": 1.45884747487597, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7896907329559326, + "num_tokens": 437563590.0, + "step": 11468 + }, + { + "epoch": 1.4589746851545604, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8204122185707092, + "num_tokens": 437601429.0, + "step": 11469 + }, + { + "epoch": 1.459101895433151, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7852215766906738, + "num_tokens": 437639144.0, + "step": 11470 + }, + { + "epoch": 1.4592291057117415, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8057632446289062, + "num_tokens": 437680846.0, + "step": 11471 + }, + { + "epoch": 1.459356315990332, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8012433052062988, + "num_tokens": 437722156.0, + "step": 11472 + }, + { + "epoch": 1.4594835262689225, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7811496257781982, + "num_tokens": 437766905.0, + "step": 11473 + }, + { + "epoch": 1.459610736547513, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7987057566642761, + "num_tokens": 437801425.0, + "step": 11474 + }, + { + "epoch": 1.4597379468261036, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7899467945098877, + "num_tokens": 437838126.0, + "step": 11475 + }, + { + "epoch": 1.4598651571046941, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7984971404075623, + "num_tokens": 437876597.0, + "step": 11476 + }, + { + "epoch": 1.4599923673832846, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8146626353263855, + "num_tokens": 437911852.0, + "step": 11477 + }, + { + "epoch": 1.460119577661875, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7996937036514282, + "num_tokens": 437953418.0, + "step": 11478 + }, + { + "epoch": 1.4602467879404655, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8058825731277466, + "num_tokens": 437989561.0, + "step": 11479 + }, + { + "epoch": 1.460373998219056, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8059951066970825, + "num_tokens": 438033464.0, + "step": 11480 + }, + { + "epoch": 1.4605012084976465, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.80599045753479, + "num_tokens": 438068676.0, + "step": 11481 + }, + { + "epoch": 1.460628418776237, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7814530730247498, + "num_tokens": 438107867.0, + "step": 11482 + }, + { + "epoch": 1.4607556290548276, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.8159310221672058, + "num_tokens": 438144256.0, + "step": 11483 + }, + { + "epoch": 1.4608828393334181, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8054596781730652, + "num_tokens": 438183458.0, + "step": 11484 + }, + { + "epoch": 1.4610100496120086, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7846250534057617, + "num_tokens": 438216025.0, + "step": 11485 + }, + { + "epoch": 1.4611372598905992, + "grad_norm": 10.125, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8168646097183228, + "num_tokens": 438258185.0, + "step": 11486 + }, + { + "epoch": 1.4612644701691897, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7898820638656616, + "num_tokens": 438298189.0, + "step": 11487 + }, + { + "epoch": 1.4613916804477802, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7896435260772705, + "num_tokens": 438336593.0, + "step": 11488 + }, + { + "epoch": 1.4615188907263708, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8048040866851807, + "num_tokens": 438374110.0, + "step": 11489 + }, + { + "epoch": 1.4616461010049613, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7820016741752625, + "num_tokens": 438413062.0, + "step": 11490 + }, + { + "epoch": 1.4617733112835518, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8018076419830322, + "num_tokens": 438452055.0, + "step": 11491 + }, + { + "epoch": 1.4619005215621423, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8052169680595398, + "num_tokens": 438494224.0, + "step": 11492 + }, + { + "epoch": 1.4620277318407329, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7954800724983215, + "num_tokens": 438529790.0, + "step": 11493 + }, + { + "epoch": 1.4621549421193232, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8051867485046387, + "num_tokens": 438571371.0, + "step": 11494 + }, + { + "epoch": 1.4622821523979137, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7862523198127747, + "num_tokens": 438612383.0, + "step": 11495 + }, + { + "epoch": 1.4624093626765042, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7902156114578247, + "num_tokens": 438653443.0, + "step": 11496 + }, + { + "epoch": 1.4625365729550948, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7832013368606567, + "num_tokens": 438689565.0, + "step": 11497 + }, + { + "epoch": 1.4626637832336853, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.8009445071220398, + "num_tokens": 438727753.0, + "step": 11498 + }, + { + "epoch": 1.4627909935122758, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8007072806358337, + "num_tokens": 438764667.0, + "step": 11499 + }, + { + "epoch": 1.4629182037908663, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8041032552719116, + "num_tokens": 438801739.0, + "step": 11500 + }, + { + "epoch": 1.4630454140694569, + "grad_norm": 60.0, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.797539472579956, + "num_tokens": 438840028.0, + "step": 11501 + }, + { + "epoch": 1.4631726243480474, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7988982200622559, + "num_tokens": 438880632.0, + "step": 11502 + }, + { + "epoch": 1.4632998346266377, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7943847179412842, + "num_tokens": 438923040.0, + "step": 11503 + }, + { + "epoch": 1.4634270449052282, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7992880344390869, + "num_tokens": 438958597.0, + "step": 11504 + }, + { + "epoch": 1.4635542551838188, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7871305346488953, + "num_tokens": 438997909.0, + "step": 11505 + }, + { + "epoch": 1.4636814654624093, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8081597089767456, + "num_tokens": 439030084.0, + "step": 11506 + }, + { + "epoch": 1.4638086757409998, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.8014270067214966, + "num_tokens": 439067770.0, + "step": 11507 + }, + { + "epoch": 1.4639358860195903, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8024166226387024, + "num_tokens": 439102476.0, + "step": 11508 + }, + { + "epoch": 1.4640630962981809, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8108757734298706, + "num_tokens": 439138660.0, + "step": 11509 + }, + { + "epoch": 1.4641903065767714, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8062881231307983, + "num_tokens": 439181690.0, + "step": 11510 + }, + { + "epoch": 1.464317516855362, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8009557127952576, + "num_tokens": 439218825.0, + "step": 11511 + }, + { + "epoch": 1.4644447271339525, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8054232001304626, + "num_tokens": 439253505.0, + "step": 11512 + }, + { + "epoch": 1.464571937412543, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7930627465248108, + "num_tokens": 439292764.0, + "step": 11513 + }, + { + "epoch": 1.4646991476911335, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8087112903594971, + "num_tokens": 439335269.0, + "step": 11514 + }, + { + "epoch": 1.464826357969724, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8061849474906921, + "num_tokens": 439371977.0, + "step": 11515 + }, + { + "epoch": 1.4649535682483146, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7993120551109314, + "num_tokens": 439412710.0, + "step": 11516 + }, + { + "epoch": 1.465080778526905, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.8001517057418823, + "num_tokens": 439449637.0, + "step": 11517 + }, + { + "epoch": 1.4652079888054954, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7973904609680176, + "num_tokens": 439495143.0, + "step": 11518 + }, + { + "epoch": 1.465335199084086, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7864432334899902, + "num_tokens": 439532721.0, + "step": 11519 + }, + { + "epoch": 1.4654624093626765, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8093712329864502, + "num_tokens": 439567938.0, + "step": 11520 + }, + { + "epoch": 1.465589619641267, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8016703128814697, + "num_tokens": 439595720.0, + "step": 11521 + }, + { + "epoch": 1.4657168299198575, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.8005363941192627, + "num_tokens": 439634091.0, + "step": 11522 + }, + { + "epoch": 1.465844040198448, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7868604063987732, + "num_tokens": 439671978.0, + "step": 11523 + }, + { + "epoch": 1.4659712504770386, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7969250679016113, + "num_tokens": 439707748.0, + "step": 11524 + }, + { + "epoch": 1.466098460755629, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8053208589553833, + "num_tokens": 439744060.0, + "step": 11525 + }, + { + "epoch": 1.4662256710342196, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7921251058578491, + "num_tokens": 439789277.0, + "step": 11526 + }, + { + "epoch": 1.46635288131281, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.820136308670044, + "num_tokens": 439827524.0, + "step": 11527 + }, + { + "epoch": 1.4664800915914005, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.794658899307251, + "num_tokens": 439865181.0, + "step": 11528 + }, + { + "epoch": 1.466607301869991, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7888634204864502, + "num_tokens": 439900071.0, + "step": 11529 + }, + { + "epoch": 1.4667345121485815, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8022764325141907, + "num_tokens": 439940839.0, + "step": 11530 + }, + { + "epoch": 1.466861722427172, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7987314462661743, + "num_tokens": 439978908.0, + "step": 11531 + }, + { + "epoch": 1.4669889327057626, + "grad_norm": 78.5, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7978392243385315, + "num_tokens": 440013448.0, + "step": 11532 + }, + { + "epoch": 1.467116142984353, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8021830320358276, + "num_tokens": 440048377.0, + "step": 11533 + }, + { + "epoch": 1.4672433532629436, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7973459959030151, + "num_tokens": 440089806.0, + "step": 11534 + }, + { + "epoch": 1.4673705635415342, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8014743328094482, + "num_tokens": 440124209.0, + "step": 11535 + }, + { + "epoch": 1.4674977738201247, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7957990169525146, + "num_tokens": 440158349.0, + "step": 11536 + }, + { + "epoch": 1.4676249840987152, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7980808019638062, + "num_tokens": 440195220.0, + "step": 11537 + }, + { + "epoch": 1.4677521943773058, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8013214468955994, + "num_tokens": 440235158.0, + "step": 11538 + }, + { + "epoch": 1.4678794046558963, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.796019434928894, + "num_tokens": 440271879.0, + "step": 11539 + }, + { + "epoch": 1.4680066149344868, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7953996062278748, + "num_tokens": 440306454.0, + "step": 11540 + }, + { + "epoch": 1.4681338252130773, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7919538021087646, + "num_tokens": 440341185.0, + "step": 11541 + }, + { + "epoch": 1.4682610354916679, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7965063452720642, + "num_tokens": 440379168.0, + "step": 11542 + }, + { + "epoch": 1.4683882457702582, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8009518384933472, + "num_tokens": 440417945.0, + "step": 11543 + }, + { + "epoch": 1.4685154560488487, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7781539559364319, + "num_tokens": 440457864.0, + "step": 11544 + }, + { + "epoch": 1.4686426663274392, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7968071699142456, + "num_tokens": 440491709.0, + "step": 11545 + }, + { + "epoch": 1.4687698766060298, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8028500080108643, + "num_tokens": 440533570.0, + "step": 11546 + }, + { + "epoch": 1.4688970868846203, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8053492307662964, + "num_tokens": 440568360.0, + "step": 11547 + }, + { + "epoch": 1.4690242971632108, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8103739023208618, + "num_tokens": 440606848.0, + "step": 11548 + }, + { + "epoch": 1.4691515074418013, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8124922513961792, + "num_tokens": 440638254.0, + "step": 11549 + }, + { + "epoch": 1.4692787177203919, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8089982867240906, + "num_tokens": 440674122.0, + "step": 11550 + }, + { + "epoch": 1.4694059279989824, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8156357407569885, + "num_tokens": 440718362.0, + "step": 11551 + }, + { + "epoch": 1.4695331382775727, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8093318343162537, + "num_tokens": 440755971.0, + "step": 11552 + }, + { + "epoch": 1.4696603485561632, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7978368401527405, + "num_tokens": 440794328.0, + "step": 11553 + }, + { + "epoch": 1.4697875588347538, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7911713719367981, + "num_tokens": 440833383.0, + "step": 11554 + }, + { + "epoch": 1.4699147691133443, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.808218240737915, + "num_tokens": 440875502.0, + "step": 11555 + }, + { + "epoch": 1.4700419793919348, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8068459630012512, + "num_tokens": 440915259.0, + "step": 11556 + }, + { + "epoch": 1.4701691896705253, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7964796423912048, + "num_tokens": 440951660.0, + "step": 11557 + }, + { + "epoch": 1.4702963999491159, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8016462922096252, + "num_tokens": 440992533.0, + "step": 11558 + }, + { + "epoch": 1.4704236102277064, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7861403226852417, + "num_tokens": 441027360.0, + "step": 11559 + }, + { + "epoch": 1.470550820506297, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8097163438796997, + "num_tokens": 441071746.0, + "step": 11560 + }, + { + "epoch": 1.4706780307848875, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8007076978683472, + "num_tokens": 441105378.0, + "step": 11561 + }, + { + "epoch": 1.470805241063478, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8065685033798218, + "num_tokens": 441150000.0, + "step": 11562 + }, + { + "epoch": 1.4709324513420685, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7965922355651855, + "num_tokens": 441188426.0, + "step": 11563 + }, + { + "epoch": 1.471059661620659, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8082051277160645, + "num_tokens": 441226028.0, + "step": 11564 + }, + { + "epoch": 1.4711868718992496, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8004403710365295, + "num_tokens": 441265788.0, + "step": 11565 + }, + { + "epoch": 1.47131408217784, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8044748902320862, + "num_tokens": 441297280.0, + "step": 11566 + }, + { + "epoch": 1.4714412924564304, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7996724247932434, + "num_tokens": 441331229.0, + "step": 11567 + }, + { + "epoch": 1.471568502735021, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8040416836738586, + "num_tokens": 441371064.0, + "step": 11568 + }, + { + "epoch": 1.4716957130136115, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7803084850311279, + "num_tokens": 441405706.0, + "step": 11569 + }, + { + "epoch": 1.471822923292202, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7904489040374756, + "num_tokens": 441445966.0, + "step": 11570 + }, + { + "epoch": 1.4719501335707925, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7864615321159363, + "num_tokens": 441478916.0, + "step": 11571 + }, + { + "epoch": 1.472077343849383, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7951112985610962, + "num_tokens": 441519245.0, + "step": 11572 + }, + { + "epoch": 1.4722045541279736, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7901821732521057, + "num_tokens": 441557485.0, + "step": 11573 + }, + { + "epoch": 1.472331764406564, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.811771035194397, + "num_tokens": 441596042.0, + "step": 11574 + }, + { + "epoch": 1.4724589746851546, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8097661733627319, + "num_tokens": 441634391.0, + "step": 11575 + }, + { + "epoch": 1.472586184963745, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7860230207443237, + "num_tokens": 441672201.0, + "step": 11576 + }, + { + "epoch": 1.4727133952423355, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7964646220207214, + "num_tokens": 441713350.0, + "step": 11577 + }, + { + "epoch": 1.472840605520926, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7991117238998413, + "num_tokens": 441754677.0, + "step": 11578 + }, + { + "epoch": 1.4729678157995165, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7900549173355103, + "num_tokens": 441797037.0, + "step": 11579 + }, + { + "epoch": 1.473095026078107, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7938075065612793, + "num_tokens": 441831661.0, + "step": 11580 + }, + { + "epoch": 1.4732222363566976, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8065553903579712, + "num_tokens": 441876933.0, + "step": 11581 + }, + { + "epoch": 1.473349446635288, + "grad_norm": 10.375, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8198413252830505, + "num_tokens": 441918015.0, + "step": 11582 + }, + { + "epoch": 1.4734766569138786, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7930304408073425, + "num_tokens": 441956573.0, + "step": 11583 + }, + { + "epoch": 1.4736038671924692, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8189191818237305, + "num_tokens": 442002064.0, + "step": 11584 + }, + { + "epoch": 1.4737310774710597, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8047714233398438, + "num_tokens": 442040122.0, + "step": 11585 + }, + { + "epoch": 1.4738582877496502, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.802929699420929, + "num_tokens": 442079698.0, + "step": 11586 + }, + { + "epoch": 1.4739854980282407, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.8190138339996338, + "num_tokens": 442118252.0, + "step": 11587 + }, + { + "epoch": 1.4741127083068313, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8059049844741821, + "num_tokens": 442154675.0, + "step": 11588 + }, + { + "epoch": 1.4742399185854218, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.806612491607666, + "num_tokens": 442189449.0, + "step": 11589 + }, + { + "epoch": 1.4743671288640123, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8061724901199341, + "num_tokens": 442222844.0, + "step": 11590 + }, + { + "epoch": 1.4744943391426026, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8092741966247559, + "num_tokens": 442258599.0, + "step": 11591 + }, + { + "epoch": 1.4746215494211932, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8052608966827393, + "num_tokens": 442294775.0, + "step": 11592 + }, + { + "epoch": 1.4747487596997837, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8092668652534485, + "num_tokens": 442330313.0, + "step": 11593 + }, + { + "epoch": 1.4748759699783742, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7963629961013794, + "num_tokens": 442364988.0, + "step": 11594 + }, + { + "epoch": 1.4750031802569648, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.8008632063865662, + "num_tokens": 442406927.0, + "step": 11595 + }, + { + "epoch": 1.4751303905355553, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8043269515037537, + "num_tokens": 442442092.0, + "step": 11596 + }, + { + "epoch": 1.4752576008141458, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8113623857498169, + "num_tokens": 442483013.0, + "step": 11597 + }, + { + "epoch": 1.4753848110927363, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8049778342247009, + "num_tokens": 442525052.0, + "step": 11598 + }, + { + "epoch": 1.4755120213713269, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7879520654678345, + "num_tokens": 442562543.0, + "step": 11599 + }, + { + "epoch": 1.4756392316499174, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8049326539039612, + "num_tokens": 442604976.0, + "step": 11600 + }, + { + "epoch": 1.4757664419285077, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8141254782676697, + "num_tokens": 442642271.0, + "step": 11601 + }, + { + "epoch": 1.4758936522070982, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7979515790939331, + "num_tokens": 442680508.0, + "step": 11602 + }, + { + "epoch": 1.4760208624856888, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7866218090057373, + "num_tokens": 442718391.0, + "step": 11603 + }, + { + "epoch": 1.4761480727642793, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8108684420585632, + "num_tokens": 442757688.0, + "step": 11604 + }, + { + "epoch": 1.4762752830428698, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8186138868331909, + "num_tokens": 442796704.0, + "step": 11605 + }, + { + "epoch": 1.4764024933214603, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8077206611633301, + "num_tokens": 442838791.0, + "step": 11606 + }, + { + "epoch": 1.4765297036000509, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7992421388626099, + "num_tokens": 442882368.0, + "step": 11607 + }, + { + "epoch": 1.4766569138786414, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8037301301956177, + "num_tokens": 442923768.0, + "step": 11608 + }, + { + "epoch": 1.476784124157232, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8177199363708496, + "num_tokens": 442955656.0, + "step": 11609 + }, + { + "epoch": 1.4769113344358225, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.790585458278656, + "num_tokens": 442997323.0, + "step": 11610 + }, + { + "epoch": 1.477038544714413, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7819563150405884, + "num_tokens": 443033454.0, + "step": 11611 + }, + { + "epoch": 1.4771657549930035, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7976477146148682, + "num_tokens": 443077902.0, + "step": 11612 + }, + { + "epoch": 1.477292965271594, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7946079969406128, + "num_tokens": 443116515.0, + "step": 11613 + }, + { + "epoch": 1.4774201755501846, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.5606, + "mean_token_accuracy": 0.8279644846916199, + "num_tokens": 443154124.0, + "step": 11614 + }, + { + "epoch": 1.477547385828775, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8114254474639893, + "num_tokens": 443195852.0, + "step": 11615 + }, + { + "epoch": 1.4776745961073654, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7894306778907776, + "num_tokens": 443229850.0, + "step": 11616 + }, + { + "epoch": 1.477801806385956, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7957763671875, + "num_tokens": 443274481.0, + "step": 11617 + }, + { + "epoch": 1.4779290166645465, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8101723790168762, + "num_tokens": 443311737.0, + "step": 11618 + }, + { + "epoch": 1.478056226943137, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7941128015518188, + "num_tokens": 443345310.0, + "step": 11619 + }, + { + "epoch": 1.4781834372217275, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8022660613059998, + "num_tokens": 443392869.0, + "step": 11620 + }, + { + "epoch": 1.478310647500318, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7999716997146606, + "num_tokens": 443432842.0, + "step": 11621 + }, + { + "epoch": 1.4784378577789086, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8063844442367554, + "num_tokens": 443465954.0, + "step": 11622 + }, + { + "epoch": 1.478565068057499, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7912116050720215, + "num_tokens": 443508506.0, + "step": 11623 + }, + { + "epoch": 1.4786922783360896, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7963218092918396, + "num_tokens": 443543575.0, + "step": 11624 + }, + { + "epoch": 1.47881948861468, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8015838861465454, + "num_tokens": 443578504.0, + "step": 11625 + }, + { + "epoch": 1.4789466988932705, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8153366446495056, + "num_tokens": 443614647.0, + "step": 11626 + }, + { + "epoch": 1.479073909171861, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7920385003089905, + "num_tokens": 443649244.0, + "step": 11627 + }, + { + "epoch": 1.4792011194504515, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7963153719902039, + "num_tokens": 443688374.0, + "step": 11628 + }, + { + "epoch": 1.479328329729042, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7687796354293823, + "num_tokens": 443730929.0, + "step": 11629 + }, + { + "epoch": 1.4794555400076326, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7888214588165283, + "num_tokens": 443763067.0, + "step": 11630 + }, + { + "epoch": 1.479582750286223, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7892731428146362, + "num_tokens": 443801273.0, + "step": 11631 + }, + { + "epoch": 1.4797099605648136, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7892823219299316, + "num_tokens": 443840394.0, + "step": 11632 + }, + { + "epoch": 1.4798371708434042, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7893390655517578, + "num_tokens": 443877486.0, + "step": 11633 + }, + { + "epoch": 1.4799643811219947, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8015873432159424, + "num_tokens": 443913763.0, + "step": 11634 + }, + { + "epoch": 1.4800915914005852, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8106621503829956, + "num_tokens": 443946840.0, + "step": 11635 + }, + { + "epoch": 1.4802188016791757, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7862803936004639, + "num_tokens": 443992439.0, + "step": 11636 + }, + { + "epoch": 1.4803460119577663, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8102383613586426, + "num_tokens": 444029348.0, + "step": 11637 + }, + { + "epoch": 1.4804732222363568, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8092149496078491, + "num_tokens": 444068386.0, + "step": 11638 + }, + { + "epoch": 1.4806004325149473, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.794625997543335, + "num_tokens": 444110307.0, + "step": 11639 + }, + { + "epoch": 1.4807276427935376, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7988477945327759, + "num_tokens": 444153919.0, + "step": 11640 + }, + { + "epoch": 1.4808548530721282, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.797313928604126, + "num_tokens": 444190575.0, + "step": 11641 + }, + { + "epoch": 1.4809820633507187, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7899975180625916, + "num_tokens": 444226108.0, + "step": 11642 + }, + { + "epoch": 1.4811092736293092, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8012427687644958, + "num_tokens": 444268162.0, + "step": 11643 + }, + { + "epoch": 1.4812364839078997, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7866030931472778, + "num_tokens": 444309273.0, + "step": 11644 + }, + { + "epoch": 1.4813636941864903, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7987282276153564, + "num_tokens": 444338697.0, + "step": 11645 + }, + { + "epoch": 1.4814909044650808, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.80849289894104, + "num_tokens": 444377954.0, + "step": 11646 + }, + { + "epoch": 1.4816181147436713, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8005658388137817, + "num_tokens": 444416996.0, + "step": 11647 + }, + { + "epoch": 1.4817453250222619, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8070703744888306, + "num_tokens": 444455550.0, + "step": 11648 + }, + { + "epoch": 1.4818725353008524, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.799150824546814, + "num_tokens": 444493979.0, + "step": 11649 + }, + { + "epoch": 1.4819997455794427, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7980989813804626, + "num_tokens": 444540622.0, + "step": 11650 + }, + { + "epoch": 1.4821269558580332, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7961546778678894, + "num_tokens": 444577772.0, + "step": 11651 + }, + { + "epoch": 1.4822541661366238, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.78725665807724, + "num_tokens": 444615523.0, + "step": 11652 + }, + { + "epoch": 1.4823813764152143, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7981617450714111, + "num_tokens": 444659785.0, + "step": 11653 + }, + { + "epoch": 1.4825085866938048, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8000310063362122, + "num_tokens": 444700806.0, + "step": 11654 + }, + { + "epoch": 1.4826357969723953, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.80190509557724, + "num_tokens": 444734320.0, + "step": 11655 + }, + { + "epoch": 1.4827630072509859, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7930538058280945, + "num_tokens": 444770215.0, + "step": 11656 + }, + { + "epoch": 1.4828902175295764, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8011488914489746, + "num_tokens": 444805432.0, + "step": 11657 + }, + { + "epoch": 1.483017427808167, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7941907644271851, + "num_tokens": 444832398.0, + "step": 11658 + }, + { + "epoch": 1.4831446380867574, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8085720539093018, + "num_tokens": 444876128.0, + "step": 11659 + }, + { + "epoch": 1.483271848365348, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7912551164627075, + "num_tokens": 444915188.0, + "step": 11660 + }, + { + "epoch": 1.4833990586439385, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8087579011917114, + "num_tokens": 444953528.0, + "step": 11661 + }, + { + "epoch": 1.483526268922529, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7954239845275879, + "num_tokens": 444996469.0, + "step": 11662 + }, + { + "epoch": 1.4836534792011196, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.8158913850784302, + "num_tokens": 445034298.0, + "step": 11663 + }, + { + "epoch": 1.48378068947971, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8032411336898804, + "num_tokens": 445079010.0, + "step": 11664 + }, + { + "epoch": 1.4839078997583004, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8126672506332397, + "num_tokens": 445116009.0, + "step": 11665 + }, + { + "epoch": 1.484035110036891, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.8140408992767334, + "num_tokens": 445156313.0, + "step": 11666 + }, + { + "epoch": 1.4841623203154815, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7973901033401489, + "num_tokens": 445187424.0, + "step": 11667 + }, + { + "epoch": 1.484289530594072, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8028348684310913, + "num_tokens": 445233448.0, + "step": 11668 + }, + { + "epoch": 1.4844167408726625, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8057265281677246, + "num_tokens": 445273663.0, + "step": 11669 + }, + { + "epoch": 1.484543951151253, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8052761554718018, + "num_tokens": 445313790.0, + "step": 11670 + }, + { + "epoch": 1.4846711614298436, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8107717037200928, + "num_tokens": 445351818.0, + "step": 11671 + }, + { + "epoch": 1.484798371708434, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7762447595596313, + "num_tokens": 445391763.0, + "step": 11672 + }, + { + "epoch": 1.4849255819870246, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7961086630821228, + "num_tokens": 445426723.0, + "step": 11673 + }, + { + "epoch": 1.485052792265615, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8032283186912537, + "num_tokens": 445465635.0, + "step": 11674 + }, + { + "epoch": 1.4851800025442055, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8049461841583252, + "num_tokens": 445502819.0, + "step": 11675 + }, + { + "epoch": 1.485307212822796, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8154022693634033, + "num_tokens": 445541861.0, + "step": 11676 + }, + { + "epoch": 1.4854344231013865, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8169418573379517, + "num_tokens": 445578058.0, + "step": 11677 + }, + { + "epoch": 1.485561633379977, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7864254117012024, + "num_tokens": 445612896.0, + "step": 11678 + }, + { + "epoch": 1.4856888436585676, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7956862449645996, + "num_tokens": 445647849.0, + "step": 11679 + }, + { + "epoch": 1.485816053937158, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7860605716705322, + "num_tokens": 445689199.0, + "step": 11680 + }, + { + "epoch": 1.4859432642157486, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7961224317550659, + "num_tokens": 445727869.0, + "step": 11681 + }, + { + "epoch": 1.4860704744943392, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8065299987792969, + "num_tokens": 445770105.0, + "step": 11682 + }, + { + "epoch": 1.4861976847729297, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7796872854232788, + "num_tokens": 445812950.0, + "step": 11683 + }, + { + "epoch": 1.4863248950515202, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.798803985118866, + "num_tokens": 445850811.0, + "step": 11684 + }, + { + "epoch": 1.4864521053301107, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8055353164672852, + "num_tokens": 445891967.0, + "step": 11685 + }, + { + "epoch": 1.4865793156087013, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.8020195364952087, + "num_tokens": 445927936.0, + "step": 11686 + }, + { + "epoch": 1.4867065258872918, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8095098733901978, + "num_tokens": 445962481.0, + "step": 11687 + }, + { + "epoch": 1.4868337361658823, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.814640998840332, + "num_tokens": 446001445.0, + "step": 11688 + }, + { + "epoch": 1.4869609464444726, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.819206178188324, + "num_tokens": 446037904.0, + "step": 11689 + }, + { + "epoch": 1.4870881567230632, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8091555237770081, + "num_tokens": 446077583.0, + "step": 11690 + }, + { + "epoch": 1.4872153670016537, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7990424633026123, + "num_tokens": 446110123.0, + "step": 11691 + }, + { + "epoch": 1.4873425772802442, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7981469631195068, + "num_tokens": 446149832.0, + "step": 11692 + }, + { + "epoch": 1.4874697875588347, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8037459850311279, + "num_tokens": 446184211.0, + "step": 11693 + }, + { + "epoch": 1.4875969978374253, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7994081377983093, + "num_tokens": 446229620.0, + "step": 11694 + }, + { + "epoch": 1.4877242081160158, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.8027539253234863, + "num_tokens": 446267134.0, + "step": 11695 + }, + { + "epoch": 1.4878514183946063, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7856582999229431, + "num_tokens": 446303001.0, + "step": 11696 + }, + { + "epoch": 1.4879786286731969, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7954079508781433, + "num_tokens": 446334286.0, + "step": 11697 + }, + { + "epoch": 1.4881058389517874, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8211688995361328, + "num_tokens": 446369547.0, + "step": 11698 + }, + { + "epoch": 1.4882330492303777, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7827287912368774, + "num_tokens": 446408179.0, + "step": 11699 + }, + { + "epoch": 1.4883602595089682, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8030317425727844, + "num_tokens": 446448533.0, + "step": 11700 + }, + { + "epoch": 1.4884874697875587, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7962738275527954, + "num_tokens": 446489445.0, + "step": 11701 + }, + { + "epoch": 1.4886146800661493, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8191837072372437, + "num_tokens": 446526530.0, + "step": 11702 + }, + { + "epoch": 1.4887418903447398, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7985661029815674, + "num_tokens": 446558315.0, + "step": 11703 + }, + { + "epoch": 1.4888691006233303, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.804924726486206, + "num_tokens": 446598548.0, + "step": 11704 + }, + { + "epoch": 1.4889963109019209, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7871941328048706, + "num_tokens": 446632111.0, + "step": 11705 + }, + { + "epoch": 1.4891235211805114, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8045949935913086, + "num_tokens": 446669296.0, + "step": 11706 + }, + { + "epoch": 1.489250731459102, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8007792234420776, + "num_tokens": 446707891.0, + "step": 11707 + }, + { + "epoch": 1.4893779417376924, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.785614013671875, + "num_tokens": 446749455.0, + "step": 11708 + }, + { + "epoch": 1.489505152016283, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8154877424240112, + "num_tokens": 446787001.0, + "step": 11709 + }, + { + "epoch": 1.4896323622948735, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8005087375640869, + "num_tokens": 446827262.0, + "step": 11710 + }, + { + "epoch": 1.489759572573464, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8192291259765625, + "num_tokens": 446860514.0, + "step": 11711 + }, + { + "epoch": 1.4898867828520546, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7900601625442505, + "num_tokens": 446902605.0, + "step": 11712 + }, + { + "epoch": 1.490013993130645, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8069803714752197, + "num_tokens": 446938638.0, + "step": 11713 + }, + { + "epoch": 1.4901412034092354, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8081660270690918, + "num_tokens": 446979286.0, + "step": 11714 + }, + { + "epoch": 1.490268413687826, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.78349369764328, + "num_tokens": 447017405.0, + "step": 11715 + }, + { + "epoch": 1.4903956239664164, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8018561005592346, + "num_tokens": 447056733.0, + "step": 11716 + }, + { + "epoch": 1.490522834245007, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7908289432525635, + "num_tokens": 447094493.0, + "step": 11717 + }, + { + "epoch": 1.4906500445235975, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8210667967796326, + "num_tokens": 447134039.0, + "step": 11718 + }, + { + "epoch": 1.490777254802188, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8040364980697632, + "num_tokens": 447168320.0, + "step": 11719 + }, + { + "epoch": 1.4909044650807786, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7769914865493774, + "num_tokens": 447205640.0, + "step": 11720 + }, + { + "epoch": 1.491031675359369, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.8162388801574707, + "num_tokens": 447239831.0, + "step": 11721 + }, + { + "epoch": 1.4911588856379596, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7871681451797485, + "num_tokens": 447281100.0, + "step": 11722 + }, + { + "epoch": 1.49128609591655, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.781402587890625, + "num_tokens": 447316137.0, + "step": 11723 + }, + { + "epoch": 1.4914133061951405, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7935433387756348, + "num_tokens": 447357922.0, + "step": 11724 + }, + { + "epoch": 1.491540516473731, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7941954135894775, + "num_tokens": 447394905.0, + "step": 11725 + }, + { + "epoch": 1.4916677267523215, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8165843486785889, + "num_tokens": 447435379.0, + "step": 11726 + }, + { + "epoch": 1.491794937030912, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7886269688606262, + "num_tokens": 447473058.0, + "step": 11727 + }, + { + "epoch": 1.4919221473095026, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8086796402931213, + "num_tokens": 447510966.0, + "step": 11728 + }, + { + "epoch": 1.492049357588093, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8055214285850525, + "num_tokens": 447554747.0, + "step": 11729 + }, + { + "epoch": 1.4921765678666836, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8027092814445496, + "num_tokens": 447594426.0, + "step": 11730 + }, + { + "epoch": 1.4923037781452742, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7916945219039917, + "num_tokens": 447634786.0, + "step": 11731 + }, + { + "epoch": 1.4924309884238647, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7948610782623291, + "num_tokens": 447668682.0, + "step": 11732 + }, + { + "epoch": 1.4925581987024552, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7888052463531494, + "num_tokens": 447705566.0, + "step": 11733 + }, + { + "epoch": 1.4926854089810457, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7975097298622131, + "num_tokens": 447746385.0, + "step": 11734 + }, + { + "epoch": 1.4928126192596363, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8055926561355591, + "num_tokens": 447790483.0, + "step": 11735 + }, + { + "epoch": 1.4929398295382268, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7983955144882202, + "num_tokens": 447826872.0, + "step": 11736 + }, + { + "epoch": 1.4930670398168173, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8162299394607544, + "num_tokens": 447864071.0, + "step": 11737 + }, + { + "epoch": 1.4931942500954076, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7832797169685364, + "num_tokens": 447903320.0, + "step": 11738 + }, + { + "epoch": 1.4933214603739982, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.811089277267456, + "num_tokens": 447940253.0, + "step": 11739 + }, + { + "epoch": 1.4934486706525887, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8066907525062561, + "num_tokens": 447978277.0, + "step": 11740 + }, + { + "epoch": 1.4935758809311792, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7929143905639648, + "num_tokens": 448018798.0, + "step": 11741 + }, + { + "epoch": 1.4937030912097697, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7877307534217834, + "num_tokens": 448057469.0, + "step": 11742 + }, + { + "epoch": 1.4938303014883603, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7788081169128418, + "num_tokens": 448089032.0, + "step": 11743 + }, + { + "epoch": 1.4939575117669508, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7963482141494751, + "num_tokens": 448122438.0, + "step": 11744 + }, + { + "epoch": 1.4940847220455413, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8033034801483154, + "num_tokens": 448166844.0, + "step": 11745 + }, + { + "epoch": 1.4942119323241319, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7890189290046692, + "num_tokens": 448197766.0, + "step": 11746 + }, + { + "epoch": 1.4943391426027224, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7866039276123047, + "num_tokens": 448240850.0, + "step": 11747 + }, + { + "epoch": 1.4944663528813127, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8078337907791138, + "num_tokens": 448274187.0, + "step": 11748 + }, + { + "epoch": 1.4945935631599032, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.804244875907898, + "num_tokens": 448311069.0, + "step": 11749 + }, + { + "epoch": 1.4947207734384937, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7990694046020508, + "num_tokens": 448351536.0, + "step": 11750 + }, + { + "epoch": 1.4948479837170843, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7973700761795044, + "num_tokens": 448385108.0, + "step": 11751 + }, + { + "epoch": 1.4949751939956748, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8120907545089722, + "num_tokens": 448423272.0, + "step": 11752 + }, + { + "epoch": 1.4951024042742653, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8090766072273254, + "num_tokens": 448462904.0, + "step": 11753 + }, + { + "epoch": 1.4952296145528559, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7947953939437866, + "num_tokens": 448503257.0, + "step": 11754 + }, + { + "epoch": 1.4953568248314464, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8026531934738159, + "num_tokens": 448535299.0, + "step": 11755 + }, + { + "epoch": 1.495484035110037, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8036965131759644, + "num_tokens": 448568878.0, + "step": 11756 + }, + { + "epoch": 1.4956112453886274, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.8229413628578186, + "num_tokens": 448604959.0, + "step": 11757 + }, + { + "epoch": 1.495738455667218, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8170862793922424, + "num_tokens": 448638487.0, + "step": 11758 + }, + { + "epoch": 1.4958656659458085, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8039434552192688, + "num_tokens": 448677212.0, + "step": 11759 + }, + { + "epoch": 1.495992876224399, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7989428043365479, + "num_tokens": 448719654.0, + "step": 11760 + }, + { + "epoch": 1.4961200865029896, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8043079972267151, + "num_tokens": 448760783.0, + "step": 11761 + }, + { + "epoch": 1.49624729678158, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7915013432502747, + "num_tokens": 448802843.0, + "step": 11762 + }, + { + "epoch": 1.4963745070601704, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7933090925216675, + "num_tokens": 448838430.0, + "step": 11763 + }, + { + "epoch": 1.496501717338761, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7909940481185913, + "num_tokens": 448880450.0, + "step": 11764 + }, + { + "epoch": 1.4966289276173514, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8001171350479126, + "num_tokens": 448919352.0, + "step": 11765 + }, + { + "epoch": 1.496756137895942, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.775617241859436, + "num_tokens": 448965402.0, + "step": 11766 + }, + { + "epoch": 1.4968833481745325, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8116191625595093, + "num_tokens": 448996983.0, + "step": 11767 + }, + { + "epoch": 1.497010558453123, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.788886308670044, + "num_tokens": 449036555.0, + "step": 11768 + }, + { + "epoch": 1.4971377687317136, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8149417638778687, + "num_tokens": 449070545.0, + "step": 11769 + }, + { + "epoch": 1.497264979010304, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7813495397567749, + "num_tokens": 449108859.0, + "step": 11770 + }, + { + "epoch": 1.4973921892888946, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7971090078353882, + "num_tokens": 449145333.0, + "step": 11771 + }, + { + "epoch": 1.497519399567485, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7911937236785889, + "num_tokens": 449181292.0, + "step": 11772 + }, + { + "epoch": 1.4976466098460754, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8023970127105713, + "num_tokens": 449213940.0, + "step": 11773 + }, + { + "epoch": 1.497773820124666, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8171307444572449, + "num_tokens": 449252487.0, + "step": 11774 + }, + { + "epoch": 1.4979010304032565, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7884235382080078, + "num_tokens": 449287268.0, + "step": 11775 + }, + { + "epoch": 1.498028240681847, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7996556758880615, + "num_tokens": 449326975.0, + "step": 11776 + }, + { + "epoch": 1.4981554509604376, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8007046580314636, + "num_tokens": 449366061.0, + "step": 11777 + }, + { + "epoch": 1.498282661239028, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8083245754241943, + "num_tokens": 449401763.0, + "step": 11778 + }, + { + "epoch": 1.4984098715176186, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8225035667419434, + "num_tokens": 449436605.0, + "step": 11779 + }, + { + "epoch": 1.4985370817962091, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.795928418636322, + "num_tokens": 449474523.0, + "step": 11780 + }, + { + "epoch": 1.4986642920747997, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.812410295009613, + "num_tokens": 449513301.0, + "step": 11781 + }, + { + "epoch": 1.4987915023533902, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7935341596603394, + "num_tokens": 449547826.0, + "step": 11782 + }, + { + "epoch": 1.4989187126319807, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7989339232444763, + "num_tokens": 449584818.0, + "step": 11783 + }, + { + "epoch": 1.4990459229105713, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7795305252075195, + "num_tokens": 449625368.0, + "step": 11784 + }, + { + "epoch": 1.4991731331891618, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7810099124908447, + "num_tokens": 449666320.0, + "step": 11785 + }, + { + "epoch": 1.4993003434677523, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7875728011131287, + "num_tokens": 449705993.0, + "step": 11786 + }, + { + "epoch": 1.4994275537463426, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7928074598312378, + "num_tokens": 449747088.0, + "step": 11787 + }, + { + "epoch": 1.4995547640249332, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8036512136459351, + "num_tokens": 449789840.0, + "step": 11788 + }, + { + "epoch": 1.4996819743035237, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7990756630897522, + "num_tokens": 449830599.0, + "step": 11789 + }, + { + "epoch": 1.4998091845821142, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8019061088562012, + "num_tokens": 449864476.0, + "step": 11790 + }, + { + "epoch": 1.4999363948607047, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7832282781600952, + "num_tokens": 449903791.0, + "step": 11791 + }, + { + "epoch": 1.5000636051392953, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8015429377555847, + "num_tokens": 449945478.0, + "step": 11792 + }, + { + "epoch": 1.5001908154178858, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7982526421546936, + "num_tokens": 449986175.0, + "step": 11793 + }, + { + "epoch": 1.5003180256964763, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8006123304367065, + "num_tokens": 450025537.0, + "step": 11794 + }, + { + "epoch": 1.5004452359750666, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7879300117492676, + "num_tokens": 450068269.0, + "step": 11795 + }, + { + "epoch": 1.5005724462536572, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8195168972015381, + "num_tokens": 450106234.0, + "step": 11796 + }, + { + "epoch": 1.5006996565322477, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8119324445724487, + "num_tokens": 450141916.0, + "step": 11797 + }, + { + "epoch": 1.5008268668108382, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7820810675621033, + "num_tokens": 450175427.0, + "step": 11798 + }, + { + "epoch": 1.5009540770894287, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.8000558614730835, + "num_tokens": 450211104.0, + "step": 11799 + }, + { + "epoch": 1.5010812873680193, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7890504002571106, + "num_tokens": 450244844.0, + "step": 11800 + }, + { + "epoch": 1.5012084976466098, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7988260984420776, + "num_tokens": 450286094.0, + "step": 11801 + }, + { + "epoch": 1.5013357079252003, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7938320636749268, + "num_tokens": 450324938.0, + "step": 11802 + }, + { + "epoch": 1.5014629182037909, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.8008166551589966, + "num_tokens": 450368572.0, + "step": 11803 + }, + { + "epoch": 1.5015901284823814, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7738800048828125, + "num_tokens": 450410926.0, + "step": 11804 + }, + { + "epoch": 1.501717338760972, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8037732839584351, + "num_tokens": 450450206.0, + "step": 11805 + }, + { + "epoch": 1.5018445490395624, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7813437581062317, + "num_tokens": 450488998.0, + "step": 11806 + }, + { + "epoch": 1.501971759318153, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8128257989883423, + "num_tokens": 450530476.0, + "step": 11807 + }, + { + "epoch": 1.5020989695967435, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8049799799919128, + "num_tokens": 450560655.0, + "step": 11808 + }, + { + "epoch": 1.502226179875334, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8018088340759277, + "num_tokens": 450601816.0, + "step": 11809 + }, + { + "epoch": 1.5023533901539246, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7991073727607727, + "num_tokens": 450634182.0, + "step": 11810 + }, + { + "epoch": 1.502480600432515, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7930954098701477, + "num_tokens": 450672705.0, + "step": 11811 + }, + { + "epoch": 1.5026078107111056, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7863938808441162, + "num_tokens": 450705636.0, + "step": 11812 + }, + { + "epoch": 1.502735020989696, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.786886990070343, + "num_tokens": 450736165.0, + "step": 11813 + }, + { + "epoch": 1.5028622312682864, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7983824014663696, + "num_tokens": 450776939.0, + "step": 11814 + }, + { + "epoch": 1.502989441546877, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8004242777824402, + "num_tokens": 450811994.0, + "step": 11815 + }, + { + "epoch": 1.5031166518254675, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8016153573989868, + "num_tokens": 450853688.0, + "step": 11816 + }, + { + "epoch": 1.503243862104058, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8014513850212097, + "num_tokens": 450892100.0, + "step": 11817 + }, + { + "epoch": 1.5033710723826486, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8194695115089417, + "num_tokens": 450936383.0, + "step": 11818 + }, + { + "epoch": 1.503498282661239, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7975452542304993, + "num_tokens": 450965525.0, + "step": 11819 + }, + { + "epoch": 1.5036254929398294, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7884097099304199, + "num_tokens": 451005874.0, + "step": 11820 + }, + { + "epoch": 1.50375270321842, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7788076400756836, + "num_tokens": 451040668.0, + "step": 11821 + }, + { + "epoch": 1.5038799134970104, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7971034049987793, + "num_tokens": 451082747.0, + "step": 11822 + }, + { + "epoch": 1.504007123775601, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8100878596305847, + "num_tokens": 451120344.0, + "step": 11823 + }, + { + "epoch": 1.5041343340541915, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.787299394607544, + "num_tokens": 451156564.0, + "step": 11824 + }, + { + "epoch": 1.504261544332782, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7982930541038513, + "num_tokens": 451197252.0, + "step": 11825 + }, + { + "epoch": 1.5043887546113726, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7906697988510132, + "num_tokens": 451236150.0, + "step": 11826 + }, + { + "epoch": 1.504515964889963, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7923415899276733, + "num_tokens": 451275751.0, + "step": 11827 + }, + { + "epoch": 1.5046431751685536, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8222982883453369, + "num_tokens": 451316778.0, + "step": 11828 + }, + { + "epoch": 1.5047703854471441, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8034964799880981, + "num_tokens": 451359921.0, + "step": 11829 + }, + { + "epoch": 1.5048975957257347, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8056039214134216, + "num_tokens": 451395616.0, + "step": 11830 + }, + { + "epoch": 1.5050248060043252, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8023860454559326, + "num_tokens": 451427766.0, + "step": 11831 + }, + { + "epoch": 1.5051520162829157, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8073762655258179, + "num_tokens": 451463073.0, + "step": 11832 + }, + { + "epoch": 1.5052792265615063, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8209153413772583, + "num_tokens": 451503039.0, + "step": 11833 + }, + { + "epoch": 1.5054064368400968, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7811833620071411, + "num_tokens": 451541599.0, + "step": 11834 + }, + { + "epoch": 1.5055336471186873, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7890971899032593, + "num_tokens": 451579278.0, + "step": 11835 + }, + { + "epoch": 1.5056608573972778, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8022712469100952, + "num_tokens": 451618591.0, + "step": 11836 + }, + { + "epoch": 1.5057880676758684, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7958921194076538, + "num_tokens": 451654453.0, + "step": 11837 + }, + { + "epoch": 1.5059152779544587, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7914641499519348, + "num_tokens": 451688302.0, + "step": 11838 + }, + { + "epoch": 1.5060424882330492, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8099792003631592, + "num_tokens": 451727301.0, + "step": 11839 + }, + { + "epoch": 1.5061696985116397, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7856010794639587, + "num_tokens": 451770471.0, + "step": 11840 + }, + { + "epoch": 1.5062969087902303, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7898175716400146, + "num_tokens": 451809413.0, + "step": 11841 + }, + { + "epoch": 1.5064241190688208, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7959133386611938, + "num_tokens": 451850177.0, + "step": 11842 + }, + { + "epoch": 1.5065513293474113, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7872087359428406, + "num_tokens": 451887977.0, + "step": 11843 + }, + { + "epoch": 1.5066785396260016, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8118025064468384, + "num_tokens": 451927403.0, + "step": 11844 + }, + { + "epoch": 1.5068057499045922, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7984751462936401, + "num_tokens": 451967350.0, + "step": 11845 + }, + { + "epoch": 1.5069329601831827, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.797736406326294, + "num_tokens": 452008664.0, + "step": 11846 + }, + { + "epoch": 1.5070601704617732, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7977641820907593, + "num_tokens": 452047303.0, + "step": 11847 + }, + { + "epoch": 1.5071873807403637, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7984650135040283, + "num_tokens": 452087359.0, + "step": 11848 + }, + { + "epoch": 1.5073145910189543, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7805372476577759, + "num_tokens": 452129426.0, + "step": 11849 + }, + { + "epoch": 1.5074418012975448, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8046349287033081, + "num_tokens": 452169134.0, + "step": 11850 + }, + { + "epoch": 1.5075690115761353, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7839587330818176, + "num_tokens": 452209962.0, + "step": 11851 + }, + { + "epoch": 1.5076962218547258, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.80363929271698, + "num_tokens": 452245978.0, + "step": 11852 + }, + { + "epoch": 1.5078234321333164, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8031879663467407, + "num_tokens": 452281013.0, + "step": 11853 + }, + { + "epoch": 1.507950642411907, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8139309287071228, + "num_tokens": 452317174.0, + "step": 11854 + }, + { + "epoch": 1.5080778526904974, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8005245923995972, + "num_tokens": 452362974.0, + "step": 11855 + }, + { + "epoch": 1.508205062969088, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.813234269618988, + "num_tokens": 452400701.0, + "step": 11856 + }, + { + "epoch": 1.5083322732476785, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8083411455154419, + "num_tokens": 452438541.0, + "step": 11857 + }, + { + "epoch": 1.508459483526269, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7951173782348633, + "num_tokens": 452476918.0, + "step": 11858 + }, + { + "epoch": 1.5085866938048595, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8018109798431396, + "num_tokens": 452509231.0, + "step": 11859 + }, + { + "epoch": 1.50871390408345, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7955151200294495, + "num_tokens": 452547079.0, + "step": 11860 + }, + { + "epoch": 1.5088411143620406, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8042020201683044, + "num_tokens": 452580215.0, + "step": 11861 + }, + { + "epoch": 1.508968324640631, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.801668643951416, + "num_tokens": 452615778.0, + "step": 11862 + }, + { + "epoch": 1.5090955349192214, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.801837682723999, + "num_tokens": 452652184.0, + "step": 11863 + }, + { + "epoch": 1.509222745197812, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7867542505264282, + "num_tokens": 452691345.0, + "step": 11864 + }, + { + "epoch": 1.5093499554764025, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7992327213287354, + "num_tokens": 452729982.0, + "step": 11865 + }, + { + "epoch": 1.509477165754993, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7967758178710938, + "num_tokens": 452771272.0, + "step": 11866 + }, + { + "epoch": 1.5096043760335836, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8004996180534363, + "num_tokens": 452812231.0, + "step": 11867 + }, + { + "epoch": 1.509731586312174, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7940904498100281, + "num_tokens": 452858406.0, + "step": 11868 + }, + { + "epoch": 1.5098587965907644, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7897083759307861, + "num_tokens": 452898793.0, + "step": 11869 + }, + { + "epoch": 1.509986006869355, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8014740347862244, + "num_tokens": 452937929.0, + "step": 11870 + }, + { + "epoch": 1.5101132171479454, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8042048215866089, + "num_tokens": 452975436.0, + "step": 11871 + }, + { + "epoch": 1.510240427426536, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8032981157302856, + "num_tokens": 453016334.0, + "step": 11872 + }, + { + "epoch": 1.5103676377051265, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8092461824417114, + "num_tokens": 453053630.0, + "step": 11873 + }, + { + "epoch": 1.510494847983717, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7790547013282776, + "num_tokens": 453092341.0, + "step": 11874 + }, + { + "epoch": 1.5106220582623076, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.806495726108551, + "num_tokens": 453126859.0, + "step": 11875 + }, + { + "epoch": 1.510749268540898, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7952854633331299, + "num_tokens": 453167327.0, + "step": 11876 + }, + { + "epoch": 1.5108764788194886, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8082701563835144, + "num_tokens": 453205901.0, + "step": 11877 + }, + { + "epoch": 1.5110036890980791, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7948891520500183, + "num_tokens": 453241856.0, + "step": 11878 + }, + { + "epoch": 1.5111308993766697, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8082858324050903, + "num_tokens": 453282528.0, + "step": 11879 + }, + { + "epoch": 1.5112581096552602, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7860836982727051, + "num_tokens": 453318762.0, + "step": 11880 + }, + { + "epoch": 1.5113853199338507, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.812059223651886, + "num_tokens": 453352482.0, + "step": 11881 + }, + { + "epoch": 1.5115125302124413, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.8003922700881958, + "num_tokens": 453393692.0, + "step": 11882 + }, + { + "epoch": 1.5116397404910318, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8048765063285828, + "num_tokens": 453429217.0, + "step": 11883 + }, + { + "epoch": 1.5117669507696223, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7831147909164429, + "num_tokens": 453468190.0, + "step": 11884 + }, + { + "epoch": 1.5118941610482128, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7921301126480103, + "num_tokens": 453509893.0, + "step": 11885 + }, + { + "epoch": 1.5120213713268034, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8021563291549683, + "num_tokens": 453546972.0, + "step": 11886 + }, + { + "epoch": 1.5121485816053937, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8098998069763184, + "num_tokens": 453578480.0, + "step": 11887 + }, + { + "epoch": 1.5122757918839842, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8032865524291992, + "num_tokens": 453623248.0, + "step": 11888 + }, + { + "epoch": 1.5124030021625747, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.820435106754303, + "num_tokens": 453660088.0, + "step": 11889 + }, + { + "epoch": 1.5125302124411653, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8068462610244751, + "num_tokens": 453698931.0, + "step": 11890 + }, + { + "epoch": 1.5126574227197558, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8034465312957764, + "num_tokens": 453734220.0, + "step": 11891 + }, + { + "epoch": 1.5127846329983463, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8017798066139221, + "num_tokens": 453774056.0, + "step": 11892 + }, + { + "epoch": 1.5129118432769366, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7998274564743042, + "num_tokens": 453806616.0, + "step": 11893 + }, + { + "epoch": 1.5130390535555271, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8111521005630493, + "num_tokens": 453843838.0, + "step": 11894 + }, + { + "epoch": 1.5131662638341177, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8000144362449646, + "num_tokens": 453876888.0, + "step": 11895 + }, + { + "epoch": 1.5132934741127082, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8168734312057495, + "num_tokens": 453915334.0, + "step": 11896 + }, + { + "epoch": 1.5134206843912987, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8105717301368713, + "num_tokens": 453950590.0, + "step": 11897 + }, + { + "epoch": 1.5135478946698893, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8027183413505554, + "num_tokens": 453992590.0, + "step": 11898 + }, + { + "epoch": 1.5136751049484798, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8078404664993286, + "num_tokens": 454038619.0, + "step": 11899 + }, + { + "epoch": 1.5138023152270703, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7904585003852844, + "num_tokens": 454081564.0, + "step": 11900 + }, + { + "epoch": 1.5139295255056608, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7909160256385803, + "num_tokens": 454125581.0, + "step": 11901 + }, + { + "epoch": 1.5140567357842514, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8124749660491943, + "num_tokens": 454163961.0, + "step": 11902 + }, + { + "epoch": 1.514183946062842, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7938175797462463, + "num_tokens": 454200410.0, + "step": 11903 + }, + { + "epoch": 1.5143111563414324, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7861166000366211, + "num_tokens": 454237143.0, + "step": 11904 + }, + { + "epoch": 1.514438366620023, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8063552379608154, + "num_tokens": 454274022.0, + "step": 11905 + }, + { + "epoch": 1.5145655768986135, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7971998453140259, + "num_tokens": 454307955.0, + "step": 11906 + }, + { + "epoch": 1.514692787177204, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8080412149429321, + "num_tokens": 454345449.0, + "step": 11907 + }, + { + "epoch": 1.5148199974557945, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7828667163848877, + "num_tokens": 454383096.0, + "step": 11908 + }, + { + "epoch": 1.514947207734385, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7854031920433044, + "num_tokens": 454420142.0, + "step": 11909 + }, + { + "epoch": 1.5150744180129756, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.797717809677124, + "num_tokens": 454454991.0, + "step": 11910 + }, + { + "epoch": 1.515201628291566, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7982774972915649, + "num_tokens": 454490161.0, + "step": 11911 + }, + { + "epoch": 1.5153288385701564, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7901384234428406, + "num_tokens": 454534767.0, + "step": 11912 + }, + { + "epoch": 1.515456048848747, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8067173957824707, + "num_tokens": 454573246.0, + "step": 11913 + }, + { + "epoch": 1.5155832591273375, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8136701583862305, + "num_tokens": 454607626.0, + "step": 11914 + }, + { + "epoch": 1.515710469405928, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7896261215209961, + "num_tokens": 454639840.0, + "step": 11915 + }, + { + "epoch": 1.5158376796845185, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.795407235622406, + "num_tokens": 454681445.0, + "step": 11916 + }, + { + "epoch": 1.515964889963109, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7974700927734375, + "num_tokens": 454726191.0, + "step": 11917 + }, + { + "epoch": 1.5160921002416994, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8078311681747437, + "num_tokens": 454765112.0, + "step": 11918 + }, + { + "epoch": 1.51621931052029, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7935265898704529, + "num_tokens": 454805254.0, + "step": 11919 + }, + { + "epoch": 1.5163465207988804, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7914208173751831, + "num_tokens": 454841344.0, + "step": 11920 + }, + { + "epoch": 1.516473731077471, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8131948709487915, + "num_tokens": 454879986.0, + "step": 11921 + }, + { + "epoch": 1.5166009413560615, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7938810586929321, + "num_tokens": 454917731.0, + "step": 11922 + }, + { + "epoch": 1.516728151634652, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8046674132347107, + "num_tokens": 454952837.0, + "step": 11923 + }, + { + "epoch": 1.5168553619132426, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8000518083572388, + "num_tokens": 454992751.0, + "step": 11924 + }, + { + "epoch": 1.516982572191833, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7987900972366333, + "num_tokens": 455029829.0, + "step": 11925 + }, + { + "epoch": 1.5171097824704236, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7969539761543274, + "num_tokens": 455072167.0, + "step": 11926 + }, + { + "epoch": 1.5172369927490141, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8050904273986816, + "num_tokens": 455108433.0, + "step": 11927 + }, + { + "epoch": 1.5173642030276047, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8047716617584229, + "num_tokens": 455148943.0, + "step": 11928 + }, + { + "epoch": 1.5174914133061952, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7967879772186279, + "num_tokens": 455185099.0, + "step": 11929 + }, + { + "epoch": 1.5176186235847857, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.8204835653305054, + "num_tokens": 455225322.0, + "step": 11930 + }, + { + "epoch": 1.5177458338633762, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.797382116317749, + "num_tokens": 455262830.0, + "step": 11931 + }, + { + "epoch": 1.5178730441419668, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.809234619140625, + "num_tokens": 455300825.0, + "step": 11932 + }, + { + "epoch": 1.5180002544205573, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.795677661895752, + "num_tokens": 455336209.0, + "step": 11933 + }, + { + "epoch": 1.5181274646991478, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7946110963821411, + "num_tokens": 455374909.0, + "step": 11934 + }, + { + "epoch": 1.5182546749777384, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.813762903213501, + "num_tokens": 455417179.0, + "step": 11935 + }, + { + "epoch": 1.5183818852563287, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7844128608703613, + "num_tokens": 455457923.0, + "step": 11936 + }, + { + "epoch": 1.5185090955349192, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7924777269363403, + "num_tokens": 455492932.0, + "step": 11937 + }, + { + "epoch": 1.5186363058135097, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7918845415115356, + "num_tokens": 455540392.0, + "step": 11938 + }, + { + "epoch": 1.5187635160921003, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7898827195167542, + "num_tokens": 455574461.0, + "step": 11939 + }, + { + "epoch": 1.5188907263706908, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7944286465644836, + "num_tokens": 455608315.0, + "step": 11940 + }, + { + "epoch": 1.5190179366492813, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.8138073682785034, + "num_tokens": 455654927.0, + "step": 11941 + }, + { + "epoch": 1.5191451469278716, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7830352783203125, + "num_tokens": 455695745.0, + "step": 11942 + }, + { + "epoch": 1.5192723572064621, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.799942135810852, + "num_tokens": 455734652.0, + "step": 11943 + }, + { + "epoch": 1.5193995674850527, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7780337333679199, + "num_tokens": 455776063.0, + "step": 11944 + }, + { + "epoch": 1.5195267777636432, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7861344814300537, + "num_tokens": 455820595.0, + "step": 11945 + }, + { + "epoch": 1.5196539880422337, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8028395175933838, + "num_tokens": 455859553.0, + "step": 11946 + }, + { + "epoch": 1.5197811983208243, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7951406240463257, + "num_tokens": 455892676.0, + "step": 11947 + }, + { + "epoch": 1.5199084085994148, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7883139848709106, + "num_tokens": 455932106.0, + "step": 11948 + }, + { + "epoch": 1.5200356188780053, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8072746396064758, + "num_tokens": 455971555.0, + "step": 11949 + }, + { + "epoch": 1.5201628291565958, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.793758749961853, + "num_tokens": 456014177.0, + "step": 11950 + }, + { + "epoch": 1.5202900394351864, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8005056381225586, + "num_tokens": 456048229.0, + "step": 11951 + }, + { + "epoch": 1.520417249713777, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.5834, + "mean_token_accuracy": 0.8238251209259033, + "num_tokens": 456085815.0, + "step": 11952 + }, + { + "epoch": 1.5205444599923674, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8028416633605957, + "num_tokens": 456128971.0, + "step": 11953 + }, + { + "epoch": 1.520671670270958, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.777432918548584, + "num_tokens": 456159705.0, + "step": 11954 + }, + { + "epoch": 1.5207988805495485, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7885140180587769, + "num_tokens": 456198251.0, + "step": 11955 + }, + { + "epoch": 1.520926090828139, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7930777072906494, + "num_tokens": 456237612.0, + "step": 11956 + }, + { + "epoch": 1.5210533011067295, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7970232963562012, + "num_tokens": 456273665.0, + "step": 11957 + }, + { + "epoch": 1.52118051138532, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8003162741661072, + "num_tokens": 456305467.0, + "step": 11958 + }, + { + "epoch": 1.5213077216639106, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8029896020889282, + "num_tokens": 456342951.0, + "step": 11959 + }, + { + "epoch": 1.521434931942501, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.79485023021698, + "num_tokens": 456375846.0, + "step": 11960 + }, + { + "epoch": 1.5215621422210914, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8106177449226379, + "num_tokens": 456407993.0, + "step": 11961 + }, + { + "epoch": 1.521689352499682, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7765827178955078, + "num_tokens": 456444278.0, + "step": 11962 + }, + { + "epoch": 1.5218165627782725, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8118974566459656, + "num_tokens": 456483950.0, + "step": 11963 + }, + { + "epoch": 1.521943773056863, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7996646761894226, + "num_tokens": 456527438.0, + "step": 11964 + }, + { + "epoch": 1.5220709833354535, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8109054565429688, + "num_tokens": 456560572.0, + "step": 11965 + }, + { + "epoch": 1.5221981936140438, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7904934883117676, + "num_tokens": 456604781.0, + "step": 11966 + }, + { + "epoch": 1.5223254038926344, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8127432465553284, + "num_tokens": 456639415.0, + "step": 11967 + }, + { + "epoch": 1.522452614171225, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8104898929595947, + "num_tokens": 456671183.0, + "step": 11968 + }, + { + "epoch": 1.5225798244498154, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7842078804969788, + "num_tokens": 456712995.0, + "step": 11969 + }, + { + "epoch": 1.522707034728406, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7944071292877197, + "num_tokens": 456750587.0, + "step": 11970 + }, + { + "epoch": 1.5228342450069965, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8051526546478271, + "num_tokens": 456787503.0, + "step": 11971 + }, + { + "epoch": 1.522961455285587, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.776954174041748, + "num_tokens": 456831645.0, + "step": 11972 + }, + { + "epoch": 1.5230886655641775, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8024504780769348, + "num_tokens": 456869143.0, + "step": 11973 + }, + { + "epoch": 1.523215875842768, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7968814969062805, + "num_tokens": 456906465.0, + "step": 11974 + }, + { + "epoch": 1.5233430861213586, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7981048822402954, + "num_tokens": 456942048.0, + "step": 11975 + }, + { + "epoch": 1.5234702963999491, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.8139525055885315, + "num_tokens": 456975710.0, + "step": 11976 + }, + { + "epoch": 1.5235975066785397, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7899346351623535, + "num_tokens": 457022295.0, + "step": 11977 + }, + { + "epoch": 1.5237247169571302, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7931604385375977, + "num_tokens": 457060433.0, + "step": 11978 + }, + { + "epoch": 1.5238519272357207, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8105693459510803, + "num_tokens": 457101760.0, + "step": 11979 + }, + { + "epoch": 1.5239791375143112, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8175448179244995, + "num_tokens": 457142676.0, + "step": 11980 + }, + { + "epoch": 1.5241063477929018, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8050838708877563, + "num_tokens": 457179507.0, + "step": 11981 + }, + { + "epoch": 1.5242335580714923, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8038156628608704, + "num_tokens": 457221563.0, + "step": 11982 + }, + { + "epoch": 1.5243607683500828, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.799604058265686, + "num_tokens": 457258634.0, + "step": 11983 + }, + { + "epoch": 1.5244879786286734, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.813753068447113, + "num_tokens": 457293097.0, + "step": 11984 + }, + { + "epoch": 1.5246151889072637, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8151494264602661, + "num_tokens": 457333525.0, + "step": 11985 + }, + { + "epoch": 1.5247423991858542, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7883334755897522, + "num_tokens": 457379772.0, + "step": 11986 + }, + { + "epoch": 1.5248696094644447, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8016873598098755, + "num_tokens": 457415852.0, + "step": 11987 + }, + { + "epoch": 1.5249968197430352, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8026204705238342, + "num_tokens": 457449771.0, + "step": 11988 + }, + { + "epoch": 1.5251240300216258, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7946197986602783, + "num_tokens": 457489601.0, + "step": 11989 + }, + { + "epoch": 1.5252512403002163, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8148956298828125, + "num_tokens": 457530516.0, + "step": 11990 + }, + { + "epoch": 1.5253784505788066, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8030048608779907, + "num_tokens": 457568299.0, + "step": 11991 + }, + { + "epoch": 1.5255056608573971, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7963268160820007, + "num_tokens": 457600476.0, + "step": 11992 + }, + { + "epoch": 1.5256328711359877, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8168385028839111, + "num_tokens": 457632024.0, + "step": 11993 + }, + { + "epoch": 1.5257600814145782, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8107023239135742, + "num_tokens": 457670805.0, + "step": 11994 + }, + { + "epoch": 1.5258872916931687, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7986078262329102, + "num_tokens": 457702468.0, + "step": 11995 + }, + { + "epoch": 1.5260145019717593, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8113038539886475, + "num_tokens": 457740443.0, + "step": 11996 + }, + { + "epoch": 1.5261417122503498, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8031288385391235, + "num_tokens": 457775833.0, + "step": 11997 + }, + { + "epoch": 1.5262689225289403, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8061448931694031, + "num_tokens": 457815850.0, + "step": 11998 + }, + { + "epoch": 1.5263961328075308, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.8107330203056335, + "num_tokens": 457854946.0, + "step": 11999 + }, + { + "epoch": 1.5265233430861214, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8056795597076416, + "num_tokens": 457890775.0, + "step": 12000 + }, + { + "epoch": 1.526650553364712, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8042784929275513, + "num_tokens": 457928143.0, + "step": 12001 + }, + { + "epoch": 1.5267777636433024, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8105377554893494, + "num_tokens": 457965999.0, + "step": 12002 + }, + { + "epoch": 1.526904973921893, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8031890392303467, + "num_tokens": 458006638.0, + "step": 12003 + }, + { + "epoch": 1.5270321842004835, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.803680419921875, + "num_tokens": 458047522.0, + "step": 12004 + }, + { + "epoch": 1.527159394479074, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8016807436943054, + "num_tokens": 458091219.0, + "step": 12005 + }, + { + "epoch": 1.5272866047576645, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7917171716690063, + "num_tokens": 458129208.0, + "step": 12006 + }, + { + "epoch": 1.527413815036255, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7974094748497009, + "num_tokens": 458163632.0, + "step": 12007 + }, + { + "epoch": 1.5275410253148456, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7927920818328857, + "num_tokens": 458202377.0, + "step": 12008 + }, + { + "epoch": 1.527668235593436, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7977138757705688, + "num_tokens": 458239901.0, + "step": 12009 + }, + { + "epoch": 1.5277954458720264, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7956556081771851, + "num_tokens": 458280291.0, + "step": 12010 + }, + { + "epoch": 1.527922656150617, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8042683005332947, + "num_tokens": 458316931.0, + "step": 12011 + }, + { + "epoch": 1.5280498664292075, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7977632880210876, + "num_tokens": 458356477.0, + "step": 12012 + }, + { + "epoch": 1.528177076707798, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.785533607006073, + "num_tokens": 458401340.0, + "step": 12013 + }, + { + "epoch": 1.5283042869863885, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8036285638809204, + "num_tokens": 458436190.0, + "step": 12014 + }, + { + "epoch": 1.5284314972649788, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7994533181190491, + "num_tokens": 458465513.0, + "step": 12015 + }, + { + "epoch": 1.5285587075435694, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7998954653739929, + "num_tokens": 458505465.0, + "step": 12016 + }, + { + "epoch": 1.52868591782216, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8009226322174072, + "num_tokens": 458545087.0, + "step": 12017 + }, + { + "epoch": 1.5288131281007504, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8155645132064819, + "num_tokens": 458583274.0, + "step": 12018 + }, + { + "epoch": 1.528940338379341, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7987151145935059, + "num_tokens": 458626026.0, + "step": 12019 + }, + { + "epoch": 1.5290675486579315, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.8180570602416992, + "num_tokens": 458668774.0, + "step": 12020 + }, + { + "epoch": 1.529194758936522, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8056197166442871, + "num_tokens": 458702256.0, + "step": 12021 + }, + { + "epoch": 1.5293219692151125, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7711275815963745, + "num_tokens": 458741429.0, + "step": 12022 + }, + { + "epoch": 1.529449179493703, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8008975982666016, + "num_tokens": 458775608.0, + "step": 12023 + }, + { + "epoch": 1.5295763897722936, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8040419816970825, + "num_tokens": 458811451.0, + "step": 12024 + }, + { + "epoch": 1.5297036000508841, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8022047877311707, + "num_tokens": 458851761.0, + "step": 12025 + }, + { + "epoch": 1.5298308103294747, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7952764630317688, + "num_tokens": 458888948.0, + "step": 12026 + }, + { + "epoch": 1.5299580206080652, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7961864471435547, + "num_tokens": 458922466.0, + "step": 12027 + }, + { + "epoch": 1.5300852308866557, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7925347089767456, + "num_tokens": 458956947.0, + "step": 12028 + }, + { + "epoch": 1.5302124411652462, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8233655095100403, + "num_tokens": 458993634.0, + "step": 12029 + }, + { + "epoch": 1.5303396514438368, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8099890351295471, + "num_tokens": 459034314.0, + "step": 12030 + }, + { + "epoch": 1.5304668617224273, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8088458180427551, + "num_tokens": 459072117.0, + "step": 12031 + }, + { + "epoch": 1.5305940720010178, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.794293224811554, + "num_tokens": 459112382.0, + "step": 12032 + }, + { + "epoch": 1.5307212822796084, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8103642463684082, + "num_tokens": 459148973.0, + "step": 12033 + }, + { + "epoch": 1.5308484925581987, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7967013716697693, + "num_tokens": 459183242.0, + "step": 12034 + }, + { + "epoch": 1.5309757028367892, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.819909930229187, + "num_tokens": 459219606.0, + "step": 12035 + }, + { + "epoch": 1.5311029131153797, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8212196826934814, + "num_tokens": 459253763.0, + "step": 12036 + }, + { + "epoch": 1.5312301233939702, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8000907301902771, + "num_tokens": 459291187.0, + "step": 12037 + }, + { + "epoch": 1.5313573336725608, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7931366562843323, + "num_tokens": 459327423.0, + "step": 12038 + }, + { + "epoch": 1.5314845439511513, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7912693023681641, + "num_tokens": 459366207.0, + "step": 12039 + }, + { + "epoch": 1.5316117542297416, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7965646982192993, + "num_tokens": 459408338.0, + "step": 12040 + }, + { + "epoch": 1.5317389645083321, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.81147301197052, + "num_tokens": 459443420.0, + "step": 12041 + }, + { + "epoch": 1.5318661747869227, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7815343141555786, + "num_tokens": 459475835.0, + "step": 12042 + }, + { + "epoch": 1.5319933850655132, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8006701469421387, + "num_tokens": 459511993.0, + "step": 12043 + }, + { + "epoch": 1.5321205953441037, + "grad_norm": 11.125, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8100776672363281, + "num_tokens": 459546221.0, + "step": 12044 + }, + { + "epoch": 1.5322478056226942, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.808465301990509, + "num_tokens": 459587271.0, + "step": 12045 + }, + { + "epoch": 1.5323750159012848, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7890691161155701, + "num_tokens": 459623388.0, + "step": 12046 + }, + { + "epoch": 1.5325022261798753, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8040045499801636, + "num_tokens": 459666440.0, + "step": 12047 + }, + { + "epoch": 1.5326294364584658, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7888411283493042, + "num_tokens": 459708373.0, + "step": 12048 + }, + { + "epoch": 1.5327566467370564, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8143351674079895, + "num_tokens": 459746707.0, + "step": 12049 + }, + { + "epoch": 1.532883857015647, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8061201572418213, + "num_tokens": 459792493.0, + "step": 12050 + }, + { + "epoch": 1.5330110672942374, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.8136221766471863, + "num_tokens": 459828897.0, + "step": 12051 + }, + { + "epoch": 1.533138277572828, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.792284369468689, + "num_tokens": 459866274.0, + "step": 12052 + }, + { + "epoch": 1.5332654878514185, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8149418234825134, + "num_tokens": 459902442.0, + "step": 12053 + }, + { + "epoch": 1.533392698130009, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8070021271705627, + "num_tokens": 459943331.0, + "step": 12054 + }, + { + "epoch": 1.5335199084085995, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8094611763954163, + "num_tokens": 459975349.0, + "step": 12055 + }, + { + "epoch": 1.53364711868719, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5981, + "mean_token_accuracy": 0.8156983852386475, + "num_tokens": 460008295.0, + "step": 12056 + }, + { + "epoch": 1.5337743289657806, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8046674728393555, + "num_tokens": 460045892.0, + "step": 12057 + }, + { + "epoch": 1.533901539244371, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8017024397850037, + "num_tokens": 460084255.0, + "step": 12058 + }, + { + "epoch": 1.5340287495229614, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.788304328918457, + "num_tokens": 460123706.0, + "step": 12059 + }, + { + "epoch": 1.534155959801552, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7882788181304932, + "num_tokens": 460162351.0, + "step": 12060 + }, + { + "epoch": 1.5342831700801425, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7916358113288879, + "num_tokens": 460200135.0, + "step": 12061 + }, + { + "epoch": 1.534410380358733, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.793144941329956, + "num_tokens": 460243585.0, + "step": 12062 + }, + { + "epoch": 1.5345375906373235, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7936888337135315, + "num_tokens": 460283695.0, + "step": 12063 + }, + { + "epoch": 1.5346648009159138, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8071858286857605, + "num_tokens": 460317241.0, + "step": 12064 + }, + { + "epoch": 1.5347920111945044, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.8005053997039795, + "num_tokens": 460351900.0, + "step": 12065 + }, + { + "epoch": 1.534919221473095, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7996770143508911, + "num_tokens": 460391672.0, + "step": 12066 + }, + { + "epoch": 1.5350464317516854, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7934855818748474, + "num_tokens": 460434120.0, + "step": 12067 + }, + { + "epoch": 1.535173642030276, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8034549951553345, + "num_tokens": 460474214.0, + "step": 12068 + }, + { + "epoch": 1.5353008523088665, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8036850690841675, + "num_tokens": 460507549.0, + "step": 12069 + }, + { + "epoch": 1.535428062587457, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7928106784820557, + "num_tokens": 460546391.0, + "step": 12070 + }, + { + "epoch": 1.5355552728660475, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7947431206703186, + "num_tokens": 460585232.0, + "step": 12071 + }, + { + "epoch": 1.535682483144638, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8049396276473999, + "num_tokens": 460624398.0, + "step": 12072 + }, + { + "epoch": 1.5358096934232286, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7951480150222778, + "num_tokens": 460667193.0, + "step": 12073 + }, + { + "epoch": 1.5359369037018191, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.776503324508667, + "num_tokens": 460710614.0, + "step": 12074 + }, + { + "epoch": 1.5360641139804097, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7993199825286865, + "num_tokens": 460751079.0, + "step": 12075 + }, + { + "epoch": 1.5361913242590002, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8011611700057983, + "num_tokens": 460786768.0, + "step": 12076 + }, + { + "epoch": 1.5363185345375907, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7888548970222473, + "num_tokens": 460824757.0, + "step": 12077 + }, + { + "epoch": 1.5364457448161812, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7938018441200256, + "num_tokens": 460865024.0, + "step": 12078 + }, + { + "epoch": 1.5365729550947718, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7915665507316589, + "num_tokens": 460900895.0, + "step": 12079 + }, + { + "epoch": 1.5367001653733623, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8060621023178101, + "num_tokens": 460936876.0, + "step": 12080 + }, + { + "epoch": 1.5368273756519528, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7805898189544678, + "num_tokens": 460975297.0, + "step": 12081 + }, + { + "epoch": 1.5369545859305433, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.785402774810791, + "num_tokens": 461008603.0, + "step": 12082 + }, + { + "epoch": 1.5370817962091337, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7890172600746155, + "num_tokens": 461047770.0, + "step": 12083 + }, + { + "epoch": 1.5372090064877242, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7843531370162964, + "num_tokens": 461088554.0, + "step": 12084 + }, + { + "epoch": 1.5373362167663147, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7774724960327148, + "num_tokens": 461126070.0, + "step": 12085 + }, + { + "epoch": 1.5374634270449052, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7971450090408325, + "num_tokens": 461166394.0, + "step": 12086 + }, + { + "epoch": 1.5375906373234958, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7925139665603638, + "num_tokens": 461208263.0, + "step": 12087 + }, + { + "epoch": 1.5377178476020863, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.808961033821106, + "num_tokens": 461245054.0, + "step": 12088 + }, + { + "epoch": 1.5378450578806766, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.789057195186615, + "num_tokens": 461281584.0, + "step": 12089 + }, + { + "epoch": 1.5379722681592671, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8135829567909241, + "num_tokens": 461322247.0, + "step": 12090 + }, + { + "epoch": 1.5380994784378577, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7874140739440918, + "num_tokens": 461358371.0, + "step": 12091 + }, + { + "epoch": 1.5382266887164482, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8085323572158813, + "num_tokens": 461394666.0, + "step": 12092 + }, + { + "epoch": 1.5383538989950387, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8129132390022278, + "num_tokens": 461430875.0, + "step": 12093 + }, + { + "epoch": 1.5384811092736292, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7985683083534241, + "num_tokens": 461467412.0, + "step": 12094 + }, + { + "epoch": 1.5386083195522198, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7797852754592896, + "num_tokens": 461501970.0, + "step": 12095 + }, + { + "epoch": 1.5387355298308103, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7898871898651123, + "num_tokens": 461542735.0, + "step": 12096 + }, + { + "epoch": 1.5388627401094008, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8054141402244568, + "num_tokens": 461583883.0, + "step": 12097 + }, + { + "epoch": 1.5389899503879914, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7941915988922119, + "num_tokens": 461625124.0, + "step": 12098 + }, + { + "epoch": 1.5391171606665819, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7880697846412659, + "num_tokens": 461660273.0, + "step": 12099 + }, + { + "epoch": 1.5392443709451724, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7913544178009033, + "num_tokens": 461698890.0, + "step": 12100 + }, + { + "epoch": 1.539371581223763, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8028267025947571, + "num_tokens": 461736927.0, + "step": 12101 + }, + { + "epoch": 1.5394987915023535, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7967036366462708, + "num_tokens": 461777796.0, + "step": 12102 + }, + { + "epoch": 1.539626001780944, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8084360361099243, + "num_tokens": 461814656.0, + "step": 12103 + }, + { + "epoch": 1.5397532120595345, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7795060873031616, + "num_tokens": 461852186.0, + "step": 12104 + }, + { + "epoch": 1.539880422338125, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8150991201400757, + "num_tokens": 461893724.0, + "step": 12105 + }, + { + "epoch": 1.5400076326167156, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7900035381317139, + "num_tokens": 461935380.0, + "step": 12106 + }, + { + "epoch": 1.5401348428953059, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8032979965209961, + "num_tokens": 461971463.0, + "step": 12107 + }, + { + "epoch": 1.5402620531738964, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8090585470199585, + "num_tokens": 462007514.0, + "step": 12108 + }, + { + "epoch": 1.540389263452487, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.8207449913024902, + "num_tokens": 462046022.0, + "step": 12109 + }, + { + "epoch": 1.5405164737310775, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8070555925369263, + "num_tokens": 462089367.0, + "step": 12110 + }, + { + "epoch": 1.540643684009668, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8071619272232056, + "num_tokens": 462128558.0, + "step": 12111 + }, + { + "epoch": 1.5407708942882585, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8067978620529175, + "num_tokens": 462162639.0, + "step": 12112 + }, + { + "epoch": 1.5408981045668488, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8023000359535217, + "num_tokens": 462199303.0, + "step": 12113 + }, + { + "epoch": 1.5410253148454394, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7965009212493896, + "num_tokens": 462246095.0, + "step": 12114 + }, + { + "epoch": 1.54115252512403, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7946705222129822, + "num_tokens": 462286653.0, + "step": 12115 + }, + { + "epoch": 1.5412797354026204, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.797216534614563, + "num_tokens": 462328295.0, + "step": 12116 + }, + { + "epoch": 1.541406945681211, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8017964959144592, + "num_tokens": 462365175.0, + "step": 12117 + }, + { + "epoch": 1.5415341559598015, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7948405742645264, + "num_tokens": 462407044.0, + "step": 12118 + }, + { + "epoch": 1.541661366238392, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7889163494110107, + "num_tokens": 462443819.0, + "step": 12119 + }, + { + "epoch": 1.5417885765169825, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8046292066574097, + "num_tokens": 462487319.0, + "step": 12120 + }, + { + "epoch": 1.541915786795573, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8023434281349182, + "num_tokens": 462527026.0, + "step": 12121 + }, + { + "epoch": 1.5420429970741636, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7896975874900818, + "num_tokens": 462557552.0, + "step": 12122 + }, + { + "epoch": 1.5421702073527541, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8018704652786255, + "num_tokens": 462595243.0, + "step": 12123 + }, + { + "epoch": 1.5422974176313446, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7975364923477173, + "num_tokens": 462636077.0, + "step": 12124 + }, + { + "epoch": 1.5424246279099352, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.809456467628479, + "num_tokens": 462668407.0, + "step": 12125 + }, + { + "epoch": 1.5425518381885257, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8009135723114014, + "num_tokens": 462707678.0, + "step": 12126 + }, + { + "epoch": 1.5426790484671162, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7888941764831543, + "num_tokens": 462739722.0, + "step": 12127 + }, + { + "epoch": 1.5428062587457068, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.805368185043335, + "num_tokens": 462770369.0, + "step": 12128 + }, + { + "epoch": 1.5429334690242973, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8084517121315002, + "num_tokens": 462806126.0, + "step": 12129 + }, + { + "epoch": 1.5430606793028878, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.801891028881073, + "num_tokens": 462846585.0, + "step": 12130 + }, + { + "epoch": 1.5431878895814783, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.797364354133606, + "num_tokens": 462889892.0, + "step": 12131 + }, + { + "epoch": 1.5433150998600687, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7972999215126038, + "num_tokens": 462927612.0, + "step": 12132 + }, + { + "epoch": 1.5434423101386592, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8071539998054504, + "num_tokens": 462971661.0, + "step": 12133 + }, + { + "epoch": 1.5435695204172497, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8035121560096741, + "num_tokens": 463011260.0, + "step": 12134 + }, + { + "epoch": 1.5436967306958402, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8028662800788879, + "num_tokens": 463045395.0, + "step": 12135 + }, + { + "epoch": 1.5438239409744308, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7919641733169556, + "num_tokens": 463081015.0, + "step": 12136 + }, + { + "epoch": 1.5439511512530213, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7994876503944397, + "num_tokens": 463115658.0, + "step": 12137 + }, + { + "epoch": 1.5440783615316116, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.814041018486023, + "num_tokens": 463152084.0, + "step": 12138 + }, + { + "epoch": 1.5442055718102021, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.792073130607605, + "num_tokens": 463190137.0, + "step": 12139 + }, + { + "epoch": 1.5443327820887927, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7997696399688721, + "num_tokens": 463227626.0, + "step": 12140 + }, + { + "epoch": 1.5444599923673832, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8034095764160156, + "num_tokens": 463266451.0, + "step": 12141 + }, + { + "epoch": 1.5445872026459737, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7865468263626099, + "num_tokens": 463305327.0, + "step": 12142 + }, + { + "epoch": 1.5447144129245642, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7994523048400879, + "num_tokens": 463345890.0, + "step": 12143 + }, + { + "epoch": 1.5448416232031548, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8027249574661255, + "num_tokens": 463381583.0, + "step": 12144 + }, + { + "epoch": 1.5449688334817453, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8034322261810303, + "num_tokens": 463414442.0, + "step": 12145 + }, + { + "epoch": 1.5450960437603358, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.802903413772583, + "num_tokens": 463449517.0, + "step": 12146 + }, + { + "epoch": 1.5452232540389264, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8071796894073486, + "num_tokens": 463487952.0, + "step": 12147 + }, + { + "epoch": 1.5453504643175169, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7994173765182495, + "num_tokens": 463520724.0, + "step": 12148 + }, + { + "epoch": 1.5454776745961074, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7962152361869812, + "num_tokens": 463564691.0, + "step": 12149 + }, + { + "epoch": 1.545604884874698, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.79796302318573, + "num_tokens": 463599037.0, + "step": 12150 + }, + { + "epoch": 1.5457320951532885, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7783843874931335, + "num_tokens": 463630744.0, + "step": 12151 + }, + { + "epoch": 1.545859305431879, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7952224016189575, + "num_tokens": 463663551.0, + "step": 12152 + }, + { + "epoch": 1.5459865157104695, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8152807354927063, + "num_tokens": 463702446.0, + "step": 12153 + }, + { + "epoch": 1.54611372598906, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.791419506072998, + "num_tokens": 463739324.0, + "step": 12154 + }, + { + "epoch": 1.5462409362676506, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7855777740478516, + "num_tokens": 463780765.0, + "step": 12155 + }, + { + "epoch": 1.5463681465462409, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7775087356567383, + "num_tokens": 463817727.0, + "step": 12156 + }, + { + "epoch": 1.5464953568248314, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8196776509284973, + "num_tokens": 463865793.0, + "step": 12157 + }, + { + "epoch": 1.546622567103422, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7999033331871033, + "num_tokens": 463908299.0, + "step": 12158 + }, + { + "epoch": 1.5467497773820125, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7949596643447876, + "num_tokens": 463944513.0, + "step": 12159 + }, + { + "epoch": 1.546876987660603, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.793533205986023, + "num_tokens": 463982862.0, + "step": 12160 + }, + { + "epoch": 1.5470041979391935, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8127759695053101, + "num_tokens": 464023357.0, + "step": 12161 + }, + { + "epoch": 1.5471314082177838, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8014006614685059, + "num_tokens": 464071889.0, + "step": 12162 + }, + { + "epoch": 1.5472586184963744, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7879297733306885, + "num_tokens": 464112334.0, + "step": 12163 + }, + { + "epoch": 1.5473858287749649, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7986782193183899, + "num_tokens": 464156833.0, + "step": 12164 + }, + { + "epoch": 1.5475130390535554, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8015853762626648, + "num_tokens": 464192989.0, + "step": 12165 + }, + { + "epoch": 1.547640249332146, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8068872094154358, + "num_tokens": 464232855.0, + "step": 12166 + }, + { + "epoch": 1.5477674596107365, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7867233753204346, + "num_tokens": 464271860.0, + "step": 12167 + }, + { + "epoch": 1.547894669889327, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8091163635253906, + "num_tokens": 464312925.0, + "step": 12168 + }, + { + "epoch": 1.5480218801679175, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7901644706726074, + "num_tokens": 464349190.0, + "step": 12169 + }, + { + "epoch": 1.548149090446508, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7944130301475525, + "num_tokens": 464392031.0, + "step": 12170 + }, + { + "epoch": 1.5482763007250986, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7918872833251953, + "num_tokens": 464424884.0, + "step": 12171 + }, + { + "epoch": 1.5484035110036891, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.794664740562439, + "num_tokens": 464466047.0, + "step": 12172 + }, + { + "epoch": 1.5485307212822796, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7999805212020874, + "num_tokens": 464497293.0, + "step": 12173 + }, + { + "epoch": 1.5486579315608702, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.784390389919281, + "num_tokens": 464541873.0, + "step": 12174 + }, + { + "epoch": 1.5487851418394607, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.797645628452301, + "num_tokens": 464579712.0, + "step": 12175 + }, + { + "epoch": 1.5489123521180512, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7908895015716553, + "num_tokens": 464612277.0, + "step": 12176 + }, + { + "epoch": 1.5490395623966418, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7936174273490906, + "num_tokens": 464647062.0, + "step": 12177 + }, + { + "epoch": 1.5491667726752323, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8034650087356567, + "num_tokens": 464677998.0, + "step": 12178 + }, + { + "epoch": 1.5492939829538228, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7919270396232605, + "num_tokens": 464719205.0, + "step": 12179 + }, + { + "epoch": 1.5494211932324133, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8065835237503052, + "num_tokens": 464760758.0, + "step": 12180 + }, + { + "epoch": 1.5495484035110036, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8048327565193176, + "num_tokens": 464797422.0, + "step": 12181 + }, + { + "epoch": 1.5496756137895942, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7966247200965881, + "num_tokens": 464832889.0, + "step": 12182 + }, + { + "epoch": 1.5498028240681847, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7898026704788208, + "num_tokens": 464867743.0, + "step": 12183 + }, + { + "epoch": 1.5499300343467752, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7830468416213989, + "num_tokens": 464904666.0, + "step": 12184 + }, + { + "epoch": 1.5500572446253658, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8052501678466797, + "num_tokens": 464942886.0, + "step": 12185 + }, + { + "epoch": 1.5501844549039563, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7990235686302185, + "num_tokens": 464977463.0, + "step": 12186 + }, + { + "epoch": 1.5503116651825466, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8091025352478027, + "num_tokens": 465015732.0, + "step": 12187 + }, + { + "epoch": 1.5504388754611371, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8122562766075134, + "num_tokens": 465060789.0, + "step": 12188 + }, + { + "epoch": 1.5505660857397277, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7934998869895935, + "num_tokens": 465097907.0, + "step": 12189 + }, + { + "epoch": 1.5506932960183182, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8093949556350708, + "num_tokens": 465139401.0, + "step": 12190 + }, + { + "epoch": 1.5508205062969087, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7801904082298279, + "num_tokens": 465176531.0, + "step": 12191 + }, + { + "epoch": 1.5509477165754992, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8093712329864502, + "num_tokens": 465216074.0, + "step": 12192 + }, + { + "epoch": 1.5510749268540898, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.8176186084747314, + "num_tokens": 465257986.0, + "step": 12193 + }, + { + "epoch": 1.5512021371326803, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8010185360908508, + "num_tokens": 465296038.0, + "step": 12194 + }, + { + "epoch": 1.5513293474112708, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7697110176086426, + "num_tokens": 465334215.0, + "step": 12195 + }, + { + "epoch": 1.5514565576898613, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7978512048721313, + "num_tokens": 465373362.0, + "step": 12196 + }, + { + "epoch": 1.5515837679684519, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7961232662200928, + "num_tokens": 465410181.0, + "step": 12197 + }, + { + "epoch": 1.5517109782470424, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8062933683395386, + "num_tokens": 465443056.0, + "step": 12198 + }, + { + "epoch": 1.551838188525633, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.8009974956512451, + "num_tokens": 465484509.0, + "step": 12199 + }, + { + "epoch": 1.5519653988042235, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7989501953125, + "num_tokens": 465520017.0, + "step": 12200 + }, + { + "epoch": 1.552092609082814, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8119814991950989, + "num_tokens": 465557221.0, + "step": 12201 + }, + { + "epoch": 1.5522198193614045, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8084455728530884, + "num_tokens": 465595455.0, + "step": 12202 + }, + { + "epoch": 1.552347029639995, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8007023334503174, + "num_tokens": 465636436.0, + "step": 12203 + }, + { + "epoch": 1.5524742399185856, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7998601198196411, + "num_tokens": 465668956.0, + "step": 12204 + }, + { + "epoch": 1.5526014501971759, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7829225063323975, + "num_tokens": 465708812.0, + "step": 12205 + }, + { + "epoch": 1.5527286604757664, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7815839648246765, + "num_tokens": 465742610.0, + "step": 12206 + }, + { + "epoch": 1.552855870754357, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7933789491653442, + "num_tokens": 465776468.0, + "step": 12207 + }, + { + "epoch": 1.5529830810329475, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.8115882277488708, + "num_tokens": 465809300.0, + "step": 12208 + }, + { + "epoch": 1.553110291311538, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8038613200187683, + "num_tokens": 465849847.0, + "step": 12209 + }, + { + "epoch": 1.5532375015901285, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8091172575950623, + "num_tokens": 465884986.0, + "step": 12210 + }, + { + "epoch": 1.5533647118687188, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8089482188224792, + "num_tokens": 465919710.0, + "step": 12211 + }, + { + "epoch": 1.5534919221473094, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7940075397491455, + "num_tokens": 465956041.0, + "step": 12212 + }, + { + "epoch": 1.5536191324258999, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8142346739768982, + "num_tokens": 465994352.0, + "step": 12213 + }, + { + "epoch": 1.5537463427044904, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8012403845787048, + "num_tokens": 466032493.0, + "step": 12214 + }, + { + "epoch": 1.553873552983081, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8018123507499695, + "num_tokens": 466067048.0, + "step": 12215 + }, + { + "epoch": 1.5540007632616715, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7743431329727173, + "num_tokens": 466106895.0, + "step": 12216 + }, + { + "epoch": 1.554127973540262, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7840317487716675, + "num_tokens": 466146740.0, + "step": 12217 + }, + { + "epoch": 1.5542551838188525, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8052999973297119, + "num_tokens": 466187097.0, + "step": 12218 + }, + { + "epoch": 1.554382394097443, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7866665124893188, + "num_tokens": 466223969.0, + "step": 12219 + }, + { + "epoch": 1.5545096043760336, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8081652522087097, + "num_tokens": 466260127.0, + "step": 12220 + }, + { + "epoch": 1.554636814654624, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8084071278572083, + "num_tokens": 466297219.0, + "step": 12221 + }, + { + "epoch": 1.5547640249332146, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7752257585525513, + "num_tokens": 466329474.0, + "step": 12222 + }, + { + "epoch": 1.5548912352118052, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8002884984016418, + "num_tokens": 466372801.0, + "step": 12223 + }, + { + "epoch": 1.5550184454903957, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8103266358375549, + "num_tokens": 466412950.0, + "step": 12224 + }, + { + "epoch": 1.5551456557689862, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8043333292007446, + "num_tokens": 466448830.0, + "step": 12225 + }, + { + "epoch": 1.5552728660475768, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.786046028137207, + "num_tokens": 466489749.0, + "step": 12226 + }, + { + "epoch": 1.5554000763261673, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7812572717666626, + "num_tokens": 466531320.0, + "step": 12227 + }, + { + "epoch": 1.5555272866047578, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.792031466960907, + "num_tokens": 466570001.0, + "step": 12228 + }, + { + "epoch": 1.5556544968833483, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8039947152137756, + "num_tokens": 466607186.0, + "step": 12229 + }, + { + "epoch": 1.5557817071619386, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.788844883441925, + "num_tokens": 466649480.0, + "step": 12230 + }, + { + "epoch": 1.5559089174405292, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8112225532531738, + "num_tokens": 466688861.0, + "step": 12231 + }, + { + "epoch": 1.5560361277191197, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7780523300170898, + "num_tokens": 466723200.0, + "step": 12232 + }, + { + "epoch": 1.5561633379977102, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7790271043777466, + "num_tokens": 466759639.0, + "step": 12233 + }, + { + "epoch": 1.5562905482763008, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7901099324226379, + "num_tokens": 466795748.0, + "step": 12234 + }, + { + "epoch": 1.5564177585548913, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8175998330116272, + "num_tokens": 466832938.0, + "step": 12235 + }, + { + "epoch": 1.5565449688334816, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7902804613113403, + "num_tokens": 466868710.0, + "step": 12236 + }, + { + "epoch": 1.5566721791120721, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7952046394348145, + "num_tokens": 466903558.0, + "step": 12237 + }, + { + "epoch": 1.5567993893906626, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7925519943237305, + "num_tokens": 466946785.0, + "step": 12238 + }, + { + "epoch": 1.5569265996692532, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7934911251068115, + "num_tokens": 466988563.0, + "step": 12239 + }, + { + "epoch": 1.5570538099478437, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7908357381820679, + "num_tokens": 467025447.0, + "step": 12240 + }, + { + "epoch": 1.5571810202264342, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8068270683288574, + "num_tokens": 467060751.0, + "step": 12241 + }, + { + "epoch": 1.5573082305050248, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7950395345687866, + "num_tokens": 467101889.0, + "step": 12242 + }, + { + "epoch": 1.5574354407836153, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7977531552314758, + "num_tokens": 467136782.0, + "step": 12243 + }, + { + "epoch": 1.5575626510622058, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8063831329345703, + "num_tokens": 467174137.0, + "step": 12244 + }, + { + "epoch": 1.5576898613407963, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.791269838809967, + "num_tokens": 467213919.0, + "step": 12245 + }, + { + "epoch": 1.5578170716193869, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8028742671012878, + "num_tokens": 467254860.0, + "step": 12246 + }, + { + "epoch": 1.5579442818979774, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.820127010345459, + "num_tokens": 467296806.0, + "step": 12247 + }, + { + "epoch": 1.558071492176568, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8038449287414551, + "num_tokens": 467332882.0, + "step": 12248 + }, + { + "epoch": 1.5581987024551585, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8170409202575684, + "num_tokens": 467375088.0, + "step": 12249 + }, + { + "epoch": 1.558325912733749, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.795303463935852, + "num_tokens": 467413000.0, + "step": 12250 + }, + { + "epoch": 1.5584531230123395, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8158413171768188, + "num_tokens": 467450947.0, + "step": 12251 + }, + { + "epoch": 1.55858033329093, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7909727692604065, + "num_tokens": 467493770.0, + "step": 12252 + }, + { + "epoch": 1.5587075435695206, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7999169826507568, + "num_tokens": 467531882.0, + "step": 12253 + }, + { + "epoch": 1.5588347538481109, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5625, + "mean_token_accuracy": 0.823546290397644, + "num_tokens": 467566059.0, + "step": 12254 + }, + { + "epoch": 1.5589619641267014, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8169815540313721, + "num_tokens": 467603734.0, + "step": 12255 + }, + { + "epoch": 1.559089174405292, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8072059154510498, + "num_tokens": 467641224.0, + "step": 12256 + }, + { + "epoch": 1.5592163846838825, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7920446991920471, + "num_tokens": 467677576.0, + "step": 12257 + }, + { + "epoch": 1.559343594962473, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8092104196548462, + "num_tokens": 467717526.0, + "step": 12258 + }, + { + "epoch": 1.5594708052410635, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7924762964248657, + "num_tokens": 467757591.0, + "step": 12259 + }, + { + "epoch": 1.5595980155196538, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7821943759918213, + "num_tokens": 467792794.0, + "step": 12260 + }, + { + "epoch": 1.5597252257982444, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7933104038238525, + "num_tokens": 467829554.0, + "step": 12261 + }, + { + "epoch": 1.5598524360768349, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7840886116027832, + "num_tokens": 467871403.0, + "step": 12262 + }, + { + "epoch": 1.5599796463554254, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7744615077972412, + "num_tokens": 467906498.0, + "step": 12263 + }, + { + "epoch": 1.560106856634016, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7984057664871216, + "num_tokens": 467946426.0, + "step": 12264 + }, + { + "epoch": 1.5602340669126065, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8020864725112915, + "num_tokens": 467979930.0, + "step": 12265 + }, + { + "epoch": 1.560361277191197, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7914246916770935, + "num_tokens": 468016347.0, + "step": 12266 + }, + { + "epoch": 1.5604884874697875, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8069466948509216, + "num_tokens": 468051446.0, + "step": 12267 + }, + { + "epoch": 1.560615697748378, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8154425621032715, + "num_tokens": 468085713.0, + "step": 12268 + }, + { + "epoch": 1.5607429080269686, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8129544258117676, + "num_tokens": 468117863.0, + "step": 12269 + }, + { + "epoch": 1.560870118305559, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7990397214889526, + "num_tokens": 468153041.0, + "step": 12270 + }, + { + "epoch": 1.5609973285841496, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7922338247299194, + "num_tokens": 468193624.0, + "step": 12271 + }, + { + "epoch": 1.5611245388627402, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8127061128616333, + "num_tokens": 468227091.0, + "step": 12272 + }, + { + "epoch": 1.5612517491413307, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7912635207176208, + "num_tokens": 468265079.0, + "step": 12273 + }, + { + "epoch": 1.5613789594199212, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.805131733417511, + "num_tokens": 468307055.0, + "step": 12274 + }, + { + "epoch": 1.5615061696985117, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7936563491821289, + "num_tokens": 468344429.0, + "step": 12275 + }, + { + "epoch": 1.5616333799771023, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7950829863548279, + "num_tokens": 468379448.0, + "step": 12276 + }, + { + "epoch": 1.5617605902556928, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8012387752532959, + "num_tokens": 468423344.0, + "step": 12277 + }, + { + "epoch": 1.5618878005342833, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8059110641479492, + "num_tokens": 468461657.0, + "step": 12278 + }, + { + "epoch": 1.5620150108128736, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8102080225944519, + "num_tokens": 468500428.0, + "step": 12279 + }, + { + "epoch": 1.5621422210914642, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7863945960998535, + "num_tokens": 468536310.0, + "step": 12280 + }, + { + "epoch": 1.5622694313700547, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.800460934638977, + "num_tokens": 468577954.0, + "step": 12281 + }, + { + "epoch": 1.5623966416486452, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7877508401870728, + "num_tokens": 468612131.0, + "step": 12282 + }, + { + "epoch": 1.5625238519272358, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7946882247924805, + "num_tokens": 468654022.0, + "step": 12283 + }, + { + "epoch": 1.5626510622058263, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7898144125938416, + "num_tokens": 468693120.0, + "step": 12284 + }, + { + "epoch": 1.5627782724844166, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.8008648157119751, + "num_tokens": 468727695.0, + "step": 12285 + }, + { + "epoch": 1.5629054827630071, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7940440773963928, + "num_tokens": 468766705.0, + "step": 12286 + }, + { + "epoch": 1.5630326930415976, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7848279476165771, + "num_tokens": 468804555.0, + "step": 12287 + }, + { + "epoch": 1.5631599033201882, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7963378429412842, + "num_tokens": 468842495.0, + "step": 12288 + }, + { + "epoch": 1.5632871135987787, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7942631244659424, + "num_tokens": 468877574.0, + "step": 12289 + }, + { + "epoch": 1.5634143238773692, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8076248168945312, + "num_tokens": 468914573.0, + "step": 12290 + }, + { + "epoch": 1.5635415341559598, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8072429895401001, + "num_tokens": 468950943.0, + "step": 12291 + }, + { + "epoch": 1.5636687444345503, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8184919357299805, + "num_tokens": 468990007.0, + "step": 12292 + }, + { + "epoch": 1.5637959547131408, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.798008382320404, + "num_tokens": 469022461.0, + "step": 12293 + }, + { + "epoch": 1.5639231649917313, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8067864775657654, + "num_tokens": 469056396.0, + "step": 12294 + }, + { + "epoch": 1.5640503752703219, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.814521312713623, + "num_tokens": 469099464.0, + "step": 12295 + }, + { + "epoch": 1.5641775855489124, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8036215901374817, + "num_tokens": 469139830.0, + "step": 12296 + }, + { + "epoch": 1.564304795827503, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7989879846572876, + "num_tokens": 469177838.0, + "step": 12297 + }, + { + "epoch": 1.5644320061060935, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.5578, + "mean_token_accuracy": 0.8299446105957031, + "num_tokens": 469218727.0, + "step": 12298 + }, + { + "epoch": 1.564559216384684, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7810064554214478, + "num_tokens": 469261186.0, + "step": 12299 + }, + { + "epoch": 1.5646864266632745, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8101695775985718, + "num_tokens": 469301191.0, + "step": 12300 + }, + { + "epoch": 1.564813636941865, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8036264181137085, + "num_tokens": 469340132.0, + "step": 12301 + }, + { + "epoch": 1.5649408472204556, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8167062997817993, + "num_tokens": 469376211.0, + "step": 12302 + }, + { + "epoch": 1.5650680574990459, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7898736596107483, + "num_tokens": 469413961.0, + "step": 12303 + }, + { + "epoch": 1.5651952677776364, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7980437278747559, + "num_tokens": 469449507.0, + "step": 12304 + }, + { + "epoch": 1.565322478056227, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7832685708999634, + "num_tokens": 469486324.0, + "step": 12305 + }, + { + "epoch": 1.5654496883348175, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7955852746963501, + "num_tokens": 469523799.0, + "step": 12306 + }, + { + "epoch": 1.565576898613408, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7829371094703674, + "num_tokens": 469558661.0, + "step": 12307 + }, + { + "epoch": 1.5657041088919985, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7987474203109741, + "num_tokens": 469603087.0, + "step": 12308 + }, + { + "epoch": 1.5658313191705888, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7967998385429382, + "num_tokens": 469637344.0, + "step": 12309 + }, + { + "epoch": 1.5659585294491793, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8112273216247559, + "num_tokens": 469681127.0, + "step": 12310 + }, + { + "epoch": 1.5660857397277699, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7996730804443359, + "num_tokens": 469724238.0, + "step": 12311 + }, + { + "epoch": 1.5662129500063604, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7922044396400452, + "num_tokens": 469765967.0, + "step": 12312 + }, + { + "epoch": 1.566340160284951, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8026922941207886, + "num_tokens": 469801387.0, + "step": 12313 + }, + { + "epoch": 1.5664673705635415, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7946218252182007, + "num_tokens": 469835448.0, + "step": 12314 + }, + { + "epoch": 1.566594580842132, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7904987335205078, + "num_tokens": 469876647.0, + "step": 12315 + }, + { + "epoch": 1.5667217911207225, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8173561096191406, + "num_tokens": 469918158.0, + "step": 12316 + }, + { + "epoch": 1.566849001399313, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7809473276138306, + "num_tokens": 469953691.0, + "step": 12317 + }, + { + "epoch": 1.5669762116779036, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8072998523712158, + "num_tokens": 469996174.0, + "step": 12318 + }, + { + "epoch": 1.567103421956494, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7764357924461365, + "num_tokens": 470031371.0, + "step": 12319 + }, + { + "epoch": 1.5672306322350846, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7993218898773193, + "num_tokens": 470068783.0, + "step": 12320 + }, + { + "epoch": 1.5673578425136752, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7833622097969055, + "num_tokens": 470103490.0, + "step": 12321 + }, + { + "epoch": 1.5674850527922657, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7920947670936584, + "num_tokens": 470137213.0, + "step": 12322 + }, + { + "epoch": 1.5676122630708562, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7994499802589417, + "num_tokens": 470171816.0, + "step": 12323 + }, + { + "epoch": 1.5677394733494467, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.772435188293457, + "num_tokens": 470210517.0, + "step": 12324 + }, + { + "epoch": 1.5678666836280373, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7879847288131714, + "num_tokens": 470247085.0, + "step": 12325 + }, + { + "epoch": 1.5679938939066278, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8012319803237915, + "num_tokens": 470288483.0, + "step": 12326 + }, + { + "epoch": 1.5681211041852183, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8017340898513794, + "num_tokens": 470322173.0, + "step": 12327 + }, + { + "epoch": 1.5682483144638086, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.796722412109375, + "num_tokens": 470356617.0, + "step": 12328 + }, + { + "epoch": 1.5683755247423992, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8091701865196228, + "num_tokens": 470393095.0, + "step": 12329 + }, + { + "epoch": 1.5685027350209897, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8109791874885559, + "num_tokens": 470427785.0, + "step": 12330 + }, + { + "epoch": 1.5686299452995802, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7933093309402466, + "num_tokens": 470469971.0, + "step": 12331 + }, + { + "epoch": 1.5687571555781707, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7849057912826538, + "num_tokens": 470503935.0, + "step": 12332 + }, + { + "epoch": 1.5688843658567613, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8004986047744751, + "num_tokens": 470536014.0, + "step": 12333 + }, + { + "epoch": 1.5690115761353516, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8126914501190186, + "num_tokens": 470575632.0, + "step": 12334 + }, + { + "epoch": 1.569138786413942, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8078157901763916, + "num_tokens": 470611295.0, + "step": 12335 + }, + { + "epoch": 1.5692659966925326, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7953848838806152, + "num_tokens": 470646993.0, + "step": 12336 + }, + { + "epoch": 1.5693932069711232, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7988846302032471, + "num_tokens": 470680253.0, + "step": 12337 + }, + { + "epoch": 1.5695204172497137, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7973459959030151, + "num_tokens": 470716169.0, + "step": 12338 + }, + { + "epoch": 1.5696476275283042, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8138999938964844, + "num_tokens": 470754208.0, + "step": 12339 + }, + { + "epoch": 1.5697748378068948, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8053679466247559, + "num_tokens": 470801329.0, + "step": 12340 + }, + { + "epoch": 1.5699020480854853, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7939797639846802, + "num_tokens": 470838408.0, + "step": 12341 + }, + { + "epoch": 1.5700292583640758, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7960883975028992, + "num_tokens": 470874001.0, + "step": 12342 + }, + { + "epoch": 1.5701564686426663, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7949934005737305, + "num_tokens": 470916870.0, + "step": 12343 + }, + { + "epoch": 1.5702836789212569, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8018380403518677, + "num_tokens": 470953862.0, + "step": 12344 + }, + { + "epoch": 1.5704108891998474, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8128727078437805, + "num_tokens": 470991507.0, + "step": 12345 + }, + { + "epoch": 1.570538099478438, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8028497695922852, + "num_tokens": 471030109.0, + "step": 12346 + }, + { + "epoch": 1.5706653097570284, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8140649795532227, + "num_tokens": 471063829.0, + "step": 12347 + }, + { + "epoch": 1.570792520035619, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7909126877784729, + "num_tokens": 471107259.0, + "step": 12348 + }, + { + "epoch": 1.5709197303142095, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.8003352880477905, + "num_tokens": 471141516.0, + "step": 12349 + }, + { + "epoch": 1.5710469405928, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7980982065200806, + "num_tokens": 471178655.0, + "step": 12350 + }, + { + "epoch": 1.5711741508713906, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8065652847290039, + "num_tokens": 471209437.0, + "step": 12351 + }, + { + "epoch": 1.5713013611499809, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.8198309540748596, + "num_tokens": 471249559.0, + "step": 12352 + }, + { + "epoch": 1.5714285714285714, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.806972086429596, + "num_tokens": 471287179.0, + "step": 12353 + }, + { + "epoch": 1.571555781707162, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8024044036865234, + "num_tokens": 471328052.0, + "step": 12354 + }, + { + "epoch": 1.5716829919857525, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.795305073261261, + "num_tokens": 471362414.0, + "step": 12355 + }, + { + "epoch": 1.571810202264343, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7949157953262329, + "num_tokens": 471397910.0, + "step": 12356 + }, + { + "epoch": 1.5719374125429335, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8155485391616821, + "num_tokens": 471436951.0, + "step": 12357 + }, + { + "epoch": 1.5720646228215238, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7983852028846741, + "num_tokens": 471470139.0, + "step": 12358 + }, + { + "epoch": 1.5721918331001143, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8119292259216309, + "num_tokens": 471508591.0, + "step": 12359 + }, + { + "epoch": 1.5723190433787049, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8013105392456055, + "num_tokens": 471549165.0, + "step": 12360 + }, + { + "epoch": 1.5724462536572954, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7863413095474243, + "num_tokens": 471589021.0, + "step": 12361 + }, + { + "epoch": 1.572573463935886, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.790104866027832, + "num_tokens": 471624263.0, + "step": 12362 + }, + { + "epoch": 1.5727006742144765, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7976641654968262, + "num_tokens": 471663678.0, + "step": 12363 + }, + { + "epoch": 1.572827884493067, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8001454472541809, + "num_tokens": 471705887.0, + "step": 12364 + }, + { + "epoch": 1.5729550947716575, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8103941082954407, + "num_tokens": 471739796.0, + "step": 12365 + }, + { + "epoch": 1.573082305050248, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7944491505622864, + "num_tokens": 471774723.0, + "step": 12366 + }, + { + "epoch": 1.5732095153288386, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7934994697570801, + "num_tokens": 471814169.0, + "step": 12367 + }, + { + "epoch": 1.573336725607429, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7987980842590332, + "num_tokens": 471852367.0, + "step": 12368 + }, + { + "epoch": 1.5734639358860196, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8037352561950684, + "num_tokens": 471883438.0, + "step": 12369 + }, + { + "epoch": 1.5735911461646102, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7933605909347534, + "num_tokens": 471921051.0, + "step": 12370 + }, + { + "epoch": 1.5737183564432007, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8017357587814331, + "num_tokens": 471959685.0, + "step": 12371 + }, + { + "epoch": 1.5738455667217912, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.798888623714447, + "num_tokens": 471997345.0, + "step": 12372 + }, + { + "epoch": 1.5739727770003817, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8038216233253479, + "num_tokens": 472034455.0, + "step": 12373 + }, + { + "epoch": 1.5740999872789723, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8003041744232178, + "num_tokens": 472073630.0, + "step": 12374 + }, + { + "epoch": 1.5742271975575628, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7982505559921265, + "num_tokens": 472111556.0, + "step": 12375 + }, + { + "epoch": 1.5743544078361533, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8042474985122681, + "num_tokens": 472148276.0, + "step": 12376 + }, + { + "epoch": 1.5744816181147436, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.795748233795166, + "num_tokens": 472181665.0, + "step": 12377 + }, + { + "epoch": 1.5746088283933342, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7907165288925171, + "num_tokens": 472222759.0, + "step": 12378 + }, + { + "epoch": 1.5747360386719247, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7944090962409973, + "num_tokens": 472260867.0, + "step": 12379 + }, + { + "epoch": 1.5748632489505152, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7895205020904541, + "num_tokens": 472300068.0, + "step": 12380 + }, + { + "epoch": 1.5749904592291057, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8117856383323669, + "num_tokens": 472336459.0, + "step": 12381 + }, + { + "epoch": 1.5751176695076963, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8047177195549011, + "num_tokens": 472376696.0, + "step": 12382 + }, + { + "epoch": 1.5752448797862866, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8155190944671631, + "num_tokens": 472414491.0, + "step": 12383 + }, + { + "epoch": 1.575372090064877, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7997453808784485, + "num_tokens": 472452188.0, + "step": 12384 + }, + { + "epoch": 1.5754993003434676, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7994537949562073, + "num_tokens": 472489690.0, + "step": 12385 + }, + { + "epoch": 1.5756265106220582, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8028905987739563, + "num_tokens": 472529912.0, + "step": 12386 + }, + { + "epoch": 1.5757537209006487, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7966806888580322, + "num_tokens": 472562534.0, + "step": 12387 + }, + { + "epoch": 1.5758809311792392, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.798201322555542, + "num_tokens": 472598955.0, + "step": 12388 + }, + { + "epoch": 1.5760081414578297, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8017426133155823, + "num_tokens": 472638064.0, + "step": 12389 + }, + { + "epoch": 1.5761353517364203, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8038057684898376, + "num_tokens": 472676823.0, + "step": 12390 + }, + { + "epoch": 1.5762625620150108, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8019131422042847, + "num_tokens": 472718528.0, + "step": 12391 + }, + { + "epoch": 1.5763897722936013, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.8060322403907776, + "num_tokens": 472755634.0, + "step": 12392 + }, + { + "epoch": 1.5765169825721919, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8115314245223999, + "num_tokens": 472798060.0, + "step": 12393 + }, + { + "epoch": 1.5766441928507824, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8104028701782227, + "num_tokens": 472837628.0, + "step": 12394 + }, + { + "epoch": 1.576771403129373, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8073039054870605, + "num_tokens": 472877446.0, + "step": 12395 + }, + { + "epoch": 1.5768986134079634, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7946093082427979, + "num_tokens": 472918437.0, + "step": 12396 + }, + { + "epoch": 1.577025823686554, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8096240758895874, + "num_tokens": 472956242.0, + "step": 12397 + }, + { + "epoch": 1.5771530339651445, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8067131042480469, + "num_tokens": 472985620.0, + "step": 12398 + }, + { + "epoch": 1.577280244243735, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8013573884963989, + "num_tokens": 473019517.0, + "step": 12399 + }, + { + "epoch": 1.5774074545223256, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.5827, + "mean_token_accuracy": 0.8203414082527161, + "num_tokens": 473054624.0, + "step": 12400 + }, + { + "epoch": 1.5775346648009159, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7907983660697937, + "num_tokens": 473094140.0, + "step": 12401 + }, + { + "epoch": 1.5776618750795064, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.784784197807312, + "num_tokens": 473136886.0, + "step": 12402 + }, + { + "epoch": 1.577789085358097, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7839787006378174, + "num_tokens": 473173267.0, + "step": 12403 + }, + { + "epoch": 1.5779162956366874, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7753294706344604, + "num_tokens": 473216477.0, + "step": 12404 + }, + { + "epoch": 1.578043505915278, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7935469150543213, + "num_tokens": 473256986.0, + "step": 12405 + }, + { + "epoch": 1.5781707161938685, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8014289140701294, + "num_tokens": 473297967.0, + "step": 12406 + }, + { + "epoch": 1.5782979264724588, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7933217883110046, + "num_tokens": 473333580.0, + "step": 12407 + }, + { + "epoch": 1.5784251367510493, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8078721165657043, + "num_tokens": 473370553.0, + "step": 12408 + }, + { + "epoch": 1.5785523470296399, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7843862771987915, + "num_tokens": 473410705.0, + "step": 12409 + }, + { + "epoch": 1.5786795573082304, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8023121356964111, + "num_tokens": 473443315.0, + "step": 12410 + }, + { + "epoch": 1.578806767586821, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7960914373397827, + "num_tokens": 473488767.0, + "step": 12411 + }, + { + "epoch": 1.5789339778654115, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.813502311706543, + "num_tokens": 473525360.0, + "step": 12412 + }, + { + "epoch": 1.579061188144002, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7901701927185059, + "num_tokens": 473560625.0, + "step": 12413 + }, + { + "epoch": 1.5791883984225925, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8167940378189087, + "num_tokens": 473595484.0, + "step": 12414 + }, + { + "epoch": 1.579315608701183, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7970613241195679, + "num_tokens": 473627269.0, + "step": 12415 + }, + { + "epoch": 1.5794428189797736, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7854592204093933, + "num_tokens": 473669159.0, + "step": 12416 + }, + { + "epoch": 1.579570029258364, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8093118071556091, + "num_tokens": 473707739.0, + "step": 12417 + }, + { + "epoch": 1.5796972395369546, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8047793507575989, + "num_tokens": 473746908.0, + "step": 12418 + }, + { + "epoch": 1.5798244498155452, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8012644648551941, + "num_tokens": 473781480.0, + "step": 12419 + }, + { + "epoch": 1.5799516600941357, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8020327687263489, + "num_tokens": 473814676.0, + "step": 12420 + }, + { + "epoch": 1.5800788703727262, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7988815903663635, + "num_tokens": 473855487.0, + "step": 12421 + }, + { + "epoch": 1.5802060806513167, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8100122213363647, + "num_tokens": 473885648.0, + "step": 12422 + }, + { + "epoch": 1.5803332909299073, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7980791926383972, + "num_tokens": 473923862.0, + "step": 12423 + }, + { + "epoch": 1.5804605012084978, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8045953512191772, + "num_tokens": 473957725.0, + "step": 12424 + }, + { + "epoch": 1.5805877114870883, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7974318861961365, + "num_tokens": 473994939.0, + "step": 12425 + }, + { + "epoch": 1.5807149217656786, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7981345653533936, + "num_tokens": 474035996.0, + "step": 12426 + }, + { + "epoch": 1.5808421320442692, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8094011545181274, + "num_tokens": 474071693.0, + "step": 12427 + }, + { + "epoch": 1.5809693423228597, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8080399632453918, + "num_tokens": 474114415.0, + "step": 12428 + }, + { + "epoch": 1.5810965526014502, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7947331070899963, + "num_tokens": 474152060.0, + "step": 12429 + }, + { + "epoch": 1.5812237628800407, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.799630880355835, + "num_tokens": 474190605.0, + "step": 12430 + }, + { + "epoch": 1.5813509731586313, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7931513786315918, + "num_tokens": 474229447.0, + "step": 12431 + }, + { + "epoch": 1.5814781834372216, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8099128007888794, + "num_tokens": 474268088.0, + "step": 12432 + }, + { + "epoch": 1.581605393715812, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7905279397964478, + "num_tokens": 474304856.0, + "step": 12433 + }, + { + "epoch": 1.5817326039944026, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8097648620605469, + "num_tokens": 474349248.0, + "step": 12434 + }, + { + "epoch": 1.5818598142729932, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8162857294082642, + "num_tokens": 474384432.0, + "step": 12435 + }, + { + "epoch": 1.5819870245515837, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7975917458534241, + "num_tokens": 474415881.0, + "step": 12436 + }, + { + "epoch": 1.5821142348301742, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7997479438781738, + "num_tokens": 474452755.0, + "step": 12437 + }, + { + "epoch": 1.5822414451087647, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8011325597763062, + "num_tokens": 474493937.0, + "step": 12438 + }, + { + "epoch": 1.5823686553873553, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8098291158676147, + "num_tokens": 474534072.0, + "step": 12439 + }, + { + "epoch": 1.5824958656659458, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7841647863388062, + "num_tokens": 474569343.0, + "step": 12440 + }, + { + "epoch": 1.5826230759445363, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7825466394424438, + "num_tokens": 474606610.0, + "step": 12441 + }, + { + "epoch": 1.5827502862231269, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7898913621902466, + "num_tokens": 474645071.0, + "step": 12442 + }, + { + "epoch": 1.5828774965017174, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7884207367897034, + "num_tokens": 474683985.0, + "step": 12443 + }, + { + "epoch": 1.583004706780308, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8090425729751587, + "num_tokens": 474726102.0, + "step": 12444 + }, + { + "epoch": 1.5831319170588984, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8064743280410767, + "num_tokens": 474766430.0, + "step": 12445 + }, + { + "epoch": 1.583259127337489, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7946164011955261, + "num_tokens": 474800648.0, + "step": 12446 + }, + { + "epoch": 1.5833863376160795, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8067704439163208, + "num_tokens": 474845540.0, + "step": 12447 + }, + { + "epoch": 1.58351354789467, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7962833642959595, + "num_tokens": 474884296.0, + "step": 12448 + }, + { + "epoch": 1.5836407581732606, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8065564036369324, + "num_tokens": 474929344.0, + "step": 12449 + }, + { + "epoch": 1.5837679684518509, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8052726984024048, + "num_tokens": 474968035.0, + "step": 12450 + }, + { + "epoch": 1.5838951787304414, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7952090501785278, + "num_tokens": 475006189.0, + "step": 12451 + }, + { + "epoch": 1.584022389009032, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7931710481643677, + "num_tokens": 475039768.0, + "step": 12452 + }, + { + "epoch": 1.5841495992876224, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8073310256004333, + "num_tokens": 475079938.0, + "step": 12453 + }, + { + "epoch": 1.584276809566213, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.791854739189148, + "num_tokens": 475122405.0, + "step": 12454 + }, + { + "epoch": 1.5844040198448035, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7922372817993164, + "num_tokens": 475158237.0, + "step": 12455 + }, + { + "epoch": 1.5845312301233938, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8036195039749146, + "num_tokens": 475194966.0, + "step": 12456 + }, + { + "epoch": 1.5846584404019843, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7986983060836792, + "num_tokens": 475234483.0, + "step": 12457 + }, + { + "epoch": 1.5847856506805749, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7952957153320312, + "num_tokens": 475275172.0, + "step": 12458 + }, + { + "epoch": 1.5849128609591654, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8076668381690979, + "num_tokens": 475313550.0, + "step": 12459 + }, + { + "epoch": 1.585040071237756, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8093338012695312, + "num_tokens": 475357019.0, + "step": 12460 + }, + { + "epoch": 1.5851672815163464, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8029494285583496, + "num_tokens": 475394788.0, + "step": 12461 + }, + { + "epoch": 1.585294491794937, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7990797758102417, + "num_tokens": 475432354.0, + "step": 12462 + }, + { + "epoch": 1.5854217020735275, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8032383918762207, + "num_tokens": 475467284.0, + "step": 12463 + }, + { + "epoch": 1.585548912352118, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7902597784996033, + "num_tokens": 475505946.0, + "step": 12464 + }, + { + "epoch": 1.5856761226307086, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7898081541061401, + "num_tokens": 475542799.0, + "step": 12465 + }, + { + "epoch": 1.585803332909299, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7997642755508423, + "num_tokens": 475582064.0, + "step": 12466 + }, + { + "epoch": 1.5859305431878896, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7954140901565552, + "num_tokens": 475615521.0, + "step": 12467 + }, + { + "epoch": 1.5860577534664801, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8139253258705139, + "num_tokens": 475650500.0, + "step": 12468 + }, + { + "epoch": 1.5861849637450707, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8034957647323608, + "num_tokens": 475695824.0, + "step": 12469 + }, + { + "epoch": 1.5863121740236612, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.770524263381958, + "num_tokens": 475734015.0, + "step": 12470 + }, + { + "epoch": 1.5864393843022517, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7956252098083496, + "num_tokens": 475768961.0, + "step": 12471 + }, + { + "epoch": 1.5865665945808423, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.8152535557746887, + "num_tokens": 475806551.0, + "step": 12472 + }, + { + "epoch": 1.5866938048594328, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7974839806556702, + "num_tokens": 475844066.0, + "step": 12473 + }, + { + "epoch": 1.5868210151380233, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7986135482788086, + "num_tokens": 475874554.0, + "step": 12474 + }, + { + "epoch": 1.5869482254166136, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8103972673416138, + "num_tokens": 475915840.0, + "step": 12475 + }, + { + "epoch": 1.5870754356952042, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7976417541503906, + "num_tokens": 475951516.0, + "step": 12476 + }, + { + "epoch": 1.5872026459737947, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7992802858352661, + "num_tokens": 475984224.0, + "step": 12477 + }, + { + "epoch": 1.5873298562523852, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7766478657722473, + "num_tokens": 476015758.0, + "step": 12478 + }, + { + "epoch": 1.5874570665309757, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7941657304763794, + "num_tokens": 476051835.0, + "step": 12479 + }, + { + "epoch": 1.5875842768095663, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7978161573410034, + "num_tokens": 476090667.0, + "step": 12480 + }, + { + "epoch": 1.5877114870881566, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7993982434272766, + "num_tokens": 476130448.0, + "step": 12481 + }, + { + "epoch": 1.587838697366747, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7906422019004822, + "num_tokens": 476174495.0, + "step": 12482 + }, + { + "epoch": 1.5879659076453376, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8076242208480835, + "num_tokens": 476213412.0, + "step": 12483 + }, + { + "epoch": 1.5880931179239282, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8131375908851624, + "num_tokens": 476250441.0, + "step": 12484 + }, + { + "epoch": 1.5882203282025187, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8121486306190491, + "num_tokens": 476284466.0, + "step": 12485 + }, + { + "epoch": 1.5883475384811092, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8028352856636047, + "num_tokens": 476327760.0, + "step": 12486 + }, + { + "epoch": 1.5884747487596997, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7932064533233643, + "num_tokens": 476369259.0, + "step": 12487 + }, + { + "epoch": 1.5886019590382903, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8051097989082336, + "num_tokens": 476403915.0, + "step": 12488 + }, + { + "epoch": 1.5887291693168808, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7920541763305664, + "num_tokens": 476440837.0, + "step": 12489 + }, + { + "epoch": 1.5888563795954713, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7950183153152466, + "num_tokens": 476478478.0, + "step": 12490 + }, + { + "epoch": 1.5889835898740619, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8146345019340515, + "num_tokens": 476514675.0, + "step": 12491 + }, + { + "epoch": 1.5891108001526524, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7943428754806519, + "num_tokens": 476559015.0, + "step": 12492 + }, + { + "epoch": 1.589238010431243, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7859739065170288, + "num_tokens": 476601848.0, + "step": 12493 + }, + { + "epoch": 1.5893652207098334, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8044945597648621, + "num_tokens": 476636794.0, + "step": 12494 + }, + { + "epoch": 1.589492430988424, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7970873117446899, + "num_tokens": 476671593.0, + "step": 12495 + }, + { + "epoch": 1.5896196412670145, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8089745044708252, + "num_tokens": 476706776.0, + "step": 12496 + }, + { + "epoch": 1.589746851545605, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7990401387214661, + "num_tokens": 476746700.0, + "step": 12497 + }, + { + "epoch": 1.5898740618241956, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8080247640609741, + "num_tokens": 476785837.0, + "step": 12498 + }, + { + "epoch": 1.5900012721027859, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8037468194961548, + "num_tokens": 476818894.0, + "step": 12499 + }, + { + "epoch": 1.5901284823813764, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7761356830596924, + "num_tokens": 476861773.0, + "step": 12500 + }, + { + "epoch": 1.590255692659967, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8046069145202637, + "num_tokens": 476899522.0, + "step": 12501 + }, + { + "epoch": 1.5903829029385574, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7960474491119385, + "num_tokens": 476939923.0, + "step": 12502 + }, + { + "epoch": 1.590510113217148, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8031945824623108, + "num_tokens": 476979867.0, + "step": 12503 + }, + { + "epoch": 1.5906373234957385, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.807817816734314, + "num_tokens": 477016068.0, + "step": 12504 + }, + { + "epoch": 1.5907645337743288, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.798751950263977, + "num_tokens": 477054526.0, + "step": 12505 + }, + { + "epoch": 1.5908917440529193, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7954438328742981, + "num_tokens": 477098029.0, + "step": 12506 + }, + { + "epoch": 1.5910189543315099, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.808944821357727, + "num_tokens": 477137070.0, + "step": 12507 + }, + { + "epoch": 1.5911461646101004, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.801024854183197, + "num_tokens": 477175640.0, + "step": 12508 + }, + { + "epoch": 1.591273374888691, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7985090017318726, + "num_tokens": 477210935.0, + "step": 12509 + }, + { + "epoch": 1.5914005851672814, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8075923919677734, + "num_tokens": 477251567.0, + "step": 12510 + }, + { + "epoch": 1.591527795445872, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8028358221054077, + "num_tokens": 477287964.0, + "step": 12511 + }, + { + "epoch": 1.5916550057244625, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7721275091171265, + "num_tokens": 477328660.0, + "step": 12512 + }, + { + "epoch": 1.591782216003053, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8035798072814941, + "num_tokens": 477364111.0, + "step": 12513 + }, + { + "epoch": 1.5919094262816436, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8103835582733154, + "num_tokens": 477402691.0, + "step": 12514 + }, + { + "epoch": 1.592036636560234, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8116198778152466, + "num_tokens": 477437526.0, + "step": 12515 + }, + { + "epoch": 1.5921638468388246, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7981061935424805, + "num_tokens": 477479175.0, + "step": 12516 + }, + { + "epoch": 1.5922910571174151, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8004417419433594, + "num_tokens": 477516896.0, + "step": 12517 + }, + { + "epoch": 1.5924182673960057, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.790636420249939, + "num_tokens": 477555543.0, + "step": 12518 + }, + { + "epoch": 1.5925454776745962, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8114048838615417, + "num_tokens": 477592251.0, + "step": 12519 + }, + { + "epoch": 1.5926726879531867, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7860462665557861, + "num_tokens": 477630039.0, + "step": 12520 + }, + { + "epoch": 1.5927998982317773, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7938991785049438, + "num_tokens": 477661028.0, + "step": 12521 + }, + { + "epoch": 1.5929271085103678, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8161919116973877, + "num_tokens": 477699612.0, + "step": 12522 + }, + { + "epoch": 1.5930543187889583, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7927982807159424, + "num_tokens": 477737300.0, + "step": 12523 + }, + { + "epoch": 1.5931815290675486, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7922664880752563, + "num_tokens": 477776031.0, + "step": 12524 + }, + { + "epoch": 1.5933087393461391, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8023942112922668, + "num_tokens": 477813640.0, + "step": 12525 + }, + { + "epoch": 1.5934359496247297, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.797887921333313, + "num_tokens": 477848652.0, + "step": 12526 + }, + { + "epoch": 1.5935631599033202, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7976524829864502, + "num_tokens": 477883710.0, + "step": 12527 + }, + { + "epoch": 1.5936903701819107, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7892763614654541, + "num_tokens": 477920316.0, + "step": 12528 + }, + { + "epoch": 1.5938175804605013, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8081157207489014, + "num_tokens": 477955523.0, + "step": 12529 + }, + { + "epoch": 1.5939447907390916, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7928054332733154, + "num_tokens": 477995768.0, + "step": 12530 + }, + { + "epoch": 1.594072001017682, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.807060718536377, + "num_tokens": 478037718.0, + "step": 12531 + }, + { + "epoch": 1.5941992112962726, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7865507006645203, + "num_tokens": 478077388.0, + "step": 12532 + }, + { + "epoch": 1.5943264215748632, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.816016674041748, + "num_tokens": 478109494.0, + "step": 12533 + }, + { + "epoch": 1.5944536318534537, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7980059385299683, + "num_tokens": 478142420.0, + "step": 12534 + }, + { + "epoch": 1.5945808421320442, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8041156530380249, + "num_tokens": 478183848.0, + "step": 12535 + }, + { + "epoch": 1.5947080524106347, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8063228130340576, + "num_tokens": 478225242.0, + "step": 12536 + }, + { + "epoch": 1.5948352626892253, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8056787252426147, + "num_tokens": 478263167.0, + "step": 12537 + }, + { + "epoch": 1.5949624729678158, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.801106333732605, + "num_tokens": 478297752.0, + "step": 12538 + }, + { + "epoch": 1.5950896832464063, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7964198589324951, + "num_tokens": 478332955.0, + "step": 12539 + }, + { + "epoch": 1.5952168935249968, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7765516042709351, + "num_tokens": 478367298.0, + "step": 12540 + }, + { + "epoch": 1.5953441038035874, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8064252734184265, + "num_tokens": 478411183.0, + "step": 12541 + }, + { + "epoch": 1.595471314082178, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8015643954277039, + "num_tokens": 478451184.0, + "step": 12542 + }, + { + "epoch": 1.5955985243607684, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.795703649520874, + "num_tokens": 478490913.0, + "step": 12543 + }, + { + "epoch": 1.595725734639359, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7980142831802368, + "num_tokens": 478529769.0, + "step": 12544 + }, + { + "epoch": 1.5958529449179495, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.795654833316803, + "num_tokens": 478566665.0, + "step": 12545 + }, + { + "epoch": 1.59598015519654, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5619, + "mean_token_accuracy": 0.8289161920547485, + "num_tokens": 478600814.0, + "step": 12546 + }, + { + "epoch": 1.5961073654751305, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8057873249053955, + "num_tokens": 478641610.0, + "step": 12547 + }, + { + "epoch": 1.5962345757537209, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8027196526527405, + "num_tokens": 478678183.0, + "step": 12548 + }, + { + "epoch": 1.5963617860323114, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7997773885726929, + "num_tokens": 478710881.0, + "step": 12549 + }, + { + "epoch": 1.596488996310902, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8141976594924927, + "num_tokens": 478748550.0, + "step": 12550 + }, + { + "epoch": 1.5966162065894924, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7886747121810913, + "num_tokens": 478790417.0, + "step": 12551 + }, + { + "epoch": 1.596743416868083, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8069291710853577, + "num_tokens": 478828856.0, + "step": 12552 + }, + { + "epoch": 1.5968706271466735, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7997479438781738, + "num_tokens": 478870240.0, + "step": 12553 + }, + { + "epoch": 1.5969978374252638, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7920852899551392, + "num_tokens": 478911920.0, + "step": 12554 + }, + { + "epoch": 1.5971250477038543, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7920275926589966, + "num_tokens": 478949083.0, + "step": 12555 + }, + { + "epoch": 1.5972522579824449, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7951600551605225, + "num_tokens": 478984087.0, + "step": 12556 + }, + { + "epoch": 1.5973794682610354, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7723915576934814, + "num_tokens": 479019364.0, + "step": 12557 + }, + { + "epoch": 1.597506678539626, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8024353384971619, + "num_tokens": 479059002.0, + "step": 12558 + }, + { + "epoch": 1.5976338888182164, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7996142506599426, + "num_tokens": 479092676.0, + "step": 12559 + }, + { + "epoch": 1.597761099096807, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8014117479324341, + "num_tokens": 479125951.0, + "step": 12560 + }, + { + "epoch": 1.5978883093753975, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.8201912641525269, + "num_tokens": 479166230.0, + "step": 12561 + }, + { + "epoch": 1.598015519653988, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7936639785766602, + "num_tokens": 479205813.0, + "step": 12562 + }, + { + "epoch": 1.5981427299325786, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7961059212684631, + "num_tokens": 479244960.0, + "step": 12563 + }, + { + "epoch": 1.598269940211169, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8092015981674194, + "num_tokens": 479289617.0, + "step": 12564 + }, + { + "epoch": 1.5983971504897596, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7984541058540344, + "num_tokens": 479326377.0, + "step": 12565 + }, + { + "epoch": 1.5985243607683501, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8042342662811279, + "num_tokens": 479360563.0, + "step": 12566 + }, + { + "epoch": 1.5986515710469407, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.813825249671936, + "num_tokens": 479403042.0, + "step": 12567 + }, + { + "epoch": 1.5987787813255312, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8022907972335815, + "num_tokens": 479437099.0, + "step": 12568 + }, + { + "epoch": 1.5989059916041217, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7992710471153259, + "num_tokens": 479479326.0, + "step": 12569 + }, + { + "epoch": 1.5990332018827123, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7825682759284973, + "num_tokens": 479515320.0, + "step": 12570 + }, + { + "epoch": 1.5991604121613028, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7904387712478638, + "num_tokens": 479559960.0, + "step": 12571 + }, + { + "epoch": 1.5992876224398933, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8010132908821106, + "num_tokens": 479599115.0, + "step": 12572 + }, + { + "epoch": 1.5994148327184836, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8105347156524658, + "num_tokens": 479636116.0, + "step": 12573 + }, + { + "epoch": 1.5995420429970741, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7948659658432007, + "num_tokens": 479676880.0, + "step": 12574 + }, + { + "epoch": 1.5996692532756647, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8012628555297852, + "num_tokens": 479717424.0, + "step": 12575 + }, + { + "epoch": 1.5997964635542552, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8007287383079529, + "num_tokens": 479756240.0, + "step": 12576 + }, + { + "epoch": 1.5999236738328457, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7859890460968018, + "num_tokens": 479791863.0, + "step": 12577 + }, + { + "epoch": 1.6000508841114363, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8183229565620422, + "num_tokens": 479827252.0, + "step": 12578 + }, + { + "epoch": 1.6001780943900266, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.798156201839447, + "num_tokens": 479863100.0, + "step": 12579 + }, + { + "epoch": 1.600305304668617, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7890342473983765, + "num_tokens": 479899684.0, + "step": 12580 + }, + { + "epoch": 1.6004325149472076, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8040704727172852, + "num_tokens": 479937656.0, + "step": 12581 + }, + { + "epoch": 1.6005597252257981, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.791542649269104, + "num_tokens": 479973357.0, + "step": 12582 + }, + { + "epoch": 1.6006869355043887, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7997881770133972, + "num_tokens": 480017862.0, + "step": 12583 + }, + { + "epoch": 1.6008141457829792, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8084303140640259, + "num_tokens": 480057039.0, + "step": 12584 + }, + { + "epoch": 1.6009413560615697, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7980746030807495, + "num_tokens": 480093396.0, + "step": 12585 + }, + { + "epoch": 1.6010685663401603, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7904907464981079, + "num_tokens": 480125200.0, + "step": 12586 + }, + { + "epoch": 1.6011957766187508, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7938396334648132, + "num_tokens": 480158085.0, + "step": 12587 + }, + { + "epoch": 1.6013229868973413, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8104552626609802, + "num_tokens": 480192979.0, + "step": 12588 + }, + { + "epoch": 1.6014501971759318, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8066908121109009, + "num_tokens": 480227984.0, + "step": 12589 + }, + { + "epoch": 1.6015774074545224, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.8182117938995361, + "num_tokens": 480264280.0, + "step": 12590 + }, + { + "epoch": 1.601704617733113, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7791965007781982, + "num_tokens": 480300670.0, + "step": 12591 + }, + { + "epoch": 1.6018318280117034, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7978374361991882, + "num_tokens": 480334460.0, + "step": 12592 + }, + { + "epoch": 1.601959038290294, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8123224377632141, + "num_tokens": 480376066.0, + "step": 12593 + }, + { + "epoch": 1.6020862485688845, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8041831254959106, + "num_tokens": 480417575.0, + "step": 12594 + }, + { + "epoch": 1.602213458847475, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7878979444503784, + "num_tokens": 480454106.0, + "step": 12595 + }, + { + "epoch": 1.6023406691260655, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7852727770805359, + "num_tokens": 480493031.0, + "step": 12596 + }, + { + "epoch": 1.6024678794046558, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7969064712524414, + "num_tokens": 480528326.0, + "step": 12597 + }, + { + "epoch": 1.6025950896832464, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8114439249038696, + "num_tokens": 480565399.0, + "step": 12598 + }, + { + "epoch": 1.602722299961837, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.817584753036499, + "num_tokens": 480600028.0, + "step": 12599 + }, + { + "epoch": 1.6028495102404274, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8014931678771973, + "num_tokens": 480635759.0, + "step": 12600 + }, + { + "epoch": 1.602976720519018, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8016610741615295, + "num_tokens": 480669528.0, + "step": 12601 + }, + { + "epoch": 1.6031039307976085, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7966117858886719, + "num_tokens": 480709425.0, + "step": 12602 + }, + { + "epoch": 1.6032311410761988, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7791986465454102, + "num_tokens": 480745010.0, + "step": 12603 + }, + { + "epoch": 1.6033583513547893, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8050387501716614, + "num_tokens": 480780320.0, + "step": 12604 + }, + { + "epoch": 1.6034855616333799, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8101176023483276, + "num_tokens": 480819335.0, + "step": 12605 + }, + { + "epoch": 1.6036127719119704, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.815533459186554, + "num_tokens": 480852702.0, + "step": 12606 + }, + { + "epoch": 1.603739982190561, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7972933053970337, + "num_tokens": 480890258.0, + "step": 12607 + }, + { + "epoch": 1.6038671924691514, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.802289605140686, + "num_tokens": 480929555.0, + "step": 12608 + }, + { + "epoch": 1.603994402747742, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.5592, + "mean_token_accuracy": 0.8313581943511963, + "num_tokens": 480970295.0, + "step": 12609 + }, + { + "epoch": 1.6041216130263325, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8070492148399353, + "num_tokens": 481011213.0, + "step": 12610 + }, + { + "epoch": 1.604248823304923, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7952130436897278, + "num_tokens": 481045273.0, + "step": 12611 + }, + { + "epoch": 1.6043760335835135, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7793067097663879, + "num_tokens": 481084567.0, + "step": 12612 + }, + { + "epoch": 1.604503243862104, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8005233407020569, + "num_tokens": 481119455.0, + "step": 12613 + }, + { + "epoch": 1.6046304541406946, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7832826375961304, + "num_tokens": 481154718.0, + "step": 12614 + }, + { + "epoch": 1.6047576644192851, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7933146357536316, + "num_tokens": 481193716.0, + "step": 12615 + }, + { + "epoch": 1.6048848746978757, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7818045616149902, + "num_tokens": 481231378.0, + "step": 12616 + }, + { + "epoch": 1.6050120849764662, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7950975894927979, + "num_tokens": 481267907.0, + "step": 12617 + }, + { + "epoch": 1.6051392952550567, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8036412000656128, + "num_tokens": 481309899.0, + "step": 12618 + }, + { + "epoch": 1.6052665055336472, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.8166192770004272, + "num_tokens": 481348845.0, + "step": 12619 + }, + { + "epoch": 1.6053937158122378, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7884871959686279, + "num_tokens": 481384999.0, + "step": 12620 + }, + { + "epoch": 1.6055209260908283, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8003461360931396, + "num_tokens": 481417287.0, + "step": 12621 + }, + { + "epoch": 1.6056481363694186, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.80975341796875, + "num_tokens": 481455943.0, + "step": 12622 + }, + { + "epoch": 1.6057753466480091, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.8188960552215576, + "num_tokens": 481492017.0, + "step": 12623 + }, + { + "epoch": 1.6059025569265997, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7928280234336853, + "num_tokens": 481535115.0, + "step": 12624 + }, + { + "epoch": 1.6060297672051902, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7846050262451172, + "num_tokens": 481580048.0, + "step": 12625 + }, + { + "epoch": 1.6061569774837807, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8104540109634399, + "num_tokens": 481616748.0, + "step": 12626 + }, + { + "epoch": 1.6062841877623713, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7751879692077637, + "num_tokens": 481659123.0, + "step": 12627 + }, + { + "epoch": 1.6064113980409616, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7883005738258362, + "num_tokens": 481704791.0, + "step": 12628 + }, + { + "epoch": 1.606538608319552, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7855718731880188, + "num_tokens": 481742913.0, + "step": 12629 + }, + { + "epoch": 1.6066658185981426, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7892741560935974, + "num_tokens": 481783549.0, + "step": 12630 + }, + { + "epoch": 1.6067930288767331, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8041295409202576, + "num_tokens": 481818486.0, + "step": 12631 + }, + { + "epoch": 1.6069202391553237, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7985478043556213, + "num_tokens": 481849969.0, + "step": 12632 + }, + { + "epoch": 1.6070474494339142, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7829145789146423, + "num_tokens": 481886108.0, + "step": 12633 + }, + { + "epoch": 1.6071746597125047, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8089150190353394, + "num_tokens": 481919389.0, + "step": 12634 + }, + { + "epoch": 1.6073018699910953, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.8135594129562378, + "num_tokens": 481958531.0, + "step": 12635 + }, + { + "epoch": 1.6074290802696858, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7934529781341553, + "num_tokens": 481999864.0, + "step": 12636 + }, + { + "epoch": 1.6075562905482763, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.8230353593826294, + "num_tokens": 482033890.0, + "step": 12637 + }, + { + "epoch": 1.6076835008268668, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8028546571731567, + "num_tokens": 482071186.0, + "step": 12638 + }, + { + "epoch": 1.6078107111054574, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8017699718475342, + "num_tokens": 482107218.0, + "step": 12639 + }, + { + "epoch": 1.607937921384048, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7984451055526733, + "num_tokens": 482146101.0, + "step": 12640 + }, + { + "epoch": 1.6080651316626384, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8121094703674316, + "num_tokens": 482178531.0, + "step": 12641 + }, + { + "epoch": 1.608192341941229, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8101369142532349, + "num_tokens": 482217019.0, + "step": 12642 + }, + { + "epoch": 1.6083195522198195, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7993142604827881, + "num_tokens": 482250547.0, + "step": 12643 + }, + { + "epoch": 1.60844676249841, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7937955260276794, + "num_tokens": 482287009.0, + "step": 12644 + }, + { + "epoch": 1.6085739727770005, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7895054817199707, + "num_tokens": 482327373.0, + "step": 12645 + }, + { + "epoch": 1.6087011830555908, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7896472215652466, + "num_tokens": 482363409.0, + "step": 12646 + }, + { + "epoch": 1.6088283933341814, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7872950434684753, + "num_tokens": 482396499.0, + "step": 12647 + }, + { + "epoch": 1.608955603612772, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.803238034248352, + "num_tokens": 482435059.0, + "step": 12648 + }, + { + "epoch": 1.6090828138913624, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7895574569702148, + "num_tokens": 482472793.0, + "step": 12649 + }, + { + "epoch": 1.609210024169953, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7989547848701477, + "num_tokens": 482508857.0, + "step": 12650 + }, + { + "epoch": 1.6093372344485435, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7862197160720825, + "num_tokens": 482548526.0, + "step": 12651 + }, + { + "epoch": 1.6094644447271338, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7873460054397583, + "num_tokens": 482586461.0, + "step": 12652 + }, + { + "epoch": 1.6095916550057243, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7928924560546875, + "num_tokens": 482624642.0, + "step": 12653 + }, + { + "epoch": 1.6097188652843148, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7948845028877258, + "num_tokens": 482664096.0, + "step": 12654 + }, + { + "epoch": 1.6098460755629054, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8140516877174377, + "num_tokens": 482701230.0, + "step": 12655 + }, + { + "epoch": 1.609973285841496, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.801406741142273, + "num_tokens": 482741611.0, + "step": 12656 + }, + { + "epoch": 1.6101004961200864, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8135100603103638, + "num_tokens": 482779482.0, + "step": 12657 + }, + { + "epoch": 1.610227706398677, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8073983788490295, + "num_tokens": 482816279.0, + "step": 12658 + }, + { + "epoch": 1.6103549166772675, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7936530113220215, + "num_tokens": 482859941.0, + "step": 12659 + }, + { + "epoch": 1.610482126955858, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8086317777633667, + "num_tokens": 482894348.0, + "step": 12660 + }, + { + "epoch": 1.6106093372344485, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7947906255722046, + "num_tokens": 482930882.0, + "step": 12661 + }, + { + "epoch": 1.610736547513039, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7775640487670898, + "num_tokens": 482969515.0, + "step": 12662 + }, + { + "epoch": 1.6108637577916296, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8008809685707092, + "num_tokens": 483008074.0, + "step": 12663 + }, + { + "epoch": 1.6109909680702201, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7885643839836121, + "num_tokens": 483038920.0, + "step": 12664 + }, + { + "epoch": 1.6111181783488107, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.812152624130249, + "num_tokens": 483075559.0, + "step": 12665 + }, + { + "epoch": 1.6112453886274012, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.80457603931427, + "num_tokens": 483114378.0, + "step": 12666 + }, + { + "epoch": 1.6113725989059917, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8065433502197266, + "num_tokens": 483147911.0, + "step": 12667 + }, + { + "epoch": 1.6114998091845822, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8196671605110168, + "num_tokens": 483188275.0, + "step": 12668 + }, + { + "epoch": 1.6116270194631728, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.8169617652893066, + "num_tokens": 483230804.0, + "step": 12669 + }, + { + "epoch": 1.6117542297417633, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7997837662696838, + "num_tokens": 483262368.0, + "step": 12670 + }, + { + "epoch": 1.6118814400203536, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.788681149482727, + "num_tokens": 483300894.0, + "step": 12671 + }, + { + "epoch": 1.6120086502989441, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8040595650672913, + "num_tokens": 483337377.0, + "step": 12672 + }, + { + "epoch": 1.6121358605775347, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7996798753738403, + "num_tokens": 483376572.0, + "step": 12673 + }, + { + "epoch": 1.6122630708561252, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7912098169326782, + "num_tokens": 483415584.0, + "step": 12674 + }, + { + "epoch": 1.6123902811347157, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8007290363311768, + "num_tokens": 483459584.0, + "step": 12675 + }, + { + "epoch": 1.6125174914133062, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7926010489463806, + "num_tokens": 483504260.0, + "step": 12676 + }, + { + "epoch": 1.6126447016918966, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8008586168289185, + "num_tokens": 483547271.0, + "step": 12677 + }, + { + "epoch": 1.612771911970487, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8041291236877441, + "num_tokens": 483582255.0, + "step": 12678 + }, + { + "epoch": 1.6128991222490776, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7968778610229492, + "num_tokens": 483622290.0, + "step": 12679 + }, + { + "epoch": 1.6130263325276681, + "grad_norm": 10.625, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8141041994094849, + "num_tokens": 483657253.0, + "step": 12680 + }, + { + "epoch": 1.6131535428062587, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8058497309684753, + "num_tokens": 483693062.0, + "step": 12681 + }, + { + "epoch": 1.6132807530848492, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7944594621658325, + "num_tokens": 483735845.0, + "step": 12682 + }, + { + "epoch": 1.6134079633634397, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8036209344863892, + "num_tokens": 483783833.0, + "step": 12683 + }, + { + "epoch": 1.6135351736420303, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8025385141372681, + "num_tokens": 483820175.0, + "step": 12684 + }, + { + "epoch": 1.6136623839206208, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.814115583896637, + "num_tokens": 483863454.0, + "step": 12685 + }, + { + "epoch": 1.6137895941992113, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.5836, + "mean_token_accuracy": 0.8232473731040955, + "num_tokens": 483900657.0, + "step": 12686 + }, + { + "epoch": 1.6139168044778018, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7898552417755127, + "num_tokens": 483935912.0, + "step": 12687 + }, + { + "epoch": 1.6140440147563924, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.796821653842926, + "num_tokens": 483968594.0, + "step": 12688 + }, + { + "epoch": 1.614171225034983, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8101508617401123, + "num_tokens": 484010866.0, + "step": 12689 + }, + { + "epoch": 1.6142984353135734, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7983167171478271, + "num_tokens": 484051355.0, + "step": 12690 + }, + { + "epoch": 1.614425645592164, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.8093370795249939, + "num_tokens": 484087049.0, + "step": 12691 + }, + { + "epoch": 1.6145528558707545, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7899335622787476, + "num_tokens": 484124914.0, + "step": 12692 + }, + { + "epoch": 1.614680066149345, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.81204754114151, + "num_tokens": 484154524.0, + "step": 12693 + }, + { + "epoch": 1.6148072764279355, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7975270748138428, + "num_tokens": 484195091.0, + "step": 12694 + }, + { + "epoch": 1.6149344867065258, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8056590557098389, + "num_tokens": 484236787.0, + "step": 12695 + }, + { + "epoch": 1.6150616969851164, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7981615662574768, + "num_tokens": 484281748.0, + "step": 12696 + }, + { + "epoch": 1.615188907263707, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.807771623134613, + "num_tokens": 484321181.0, + "step": 12697 + }, + { + "epoch": 1.6153161175422974, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.8004874587059021, + "num_tokens": 484360476.0, + "step": 12698 + }, + { + "epoch": 1.615443327820888, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.802792489528656, + "num_tokens": 484395065.0, + "step": 12699 + }, + { + "epoch": 1.6155705380994785, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8084392547607422, + "num_tokens": 484439309.0, + "step": 12700 + }, + { + "epoch": 1.6156977483780688, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7995260953903198, + "num_tokens": 484473018.0, + "step": 12701 + }, + { + "epoch": 1.6158249586566593, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8088638186454773, + "num_tokens": 484518306.0, + "step": 12702 + }, + { + "epoch": 1.6159521689352498, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7859214544296265, + "num_tokens": 484556348.0, + "step": 12703 + }, + { + "epoch": 1.6160793792138404, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8108727931976318, + "num_tokens": 484593136.0, + "step": 12704 + }, + { + "epoch": 1.616206589492431, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.812152624130249, + "num_tokens": 484628025.0, + "step": 12705 + }, + { + "epoch": 1.6163337997710214, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8121006488800049, + "num_tokens": 484666703.0, + "step": 12706 + }, + { + "epoch": 1.616461010049612, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8105881810188293, + "num_tokens": 484703417.0, + "step": 12707 + }, + { + "epoch": 1.6165882203282025, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8140918016433716, + "num_tokens": 484741830.0, + "step": 12708 + }, + { + "epoch": 1.616715430606793, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8018536567687988, + "num_tokens": 484780140.0, + "step": 12709 + }, + { + "epoch": 1.6168426408853835, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7974008321762085, + "num_tokens": 484813856.0, + "step": 12710 + }, + { + "epoch": 1.616969851163974, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8129533529281616, + "num_tokens": 484851383.0, + "step": 12711 + }, + { + "epoch": 1.6170970614425646, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8065521717071533, + "num_tokens": 484884361.0, + "step": 12712 + }, + { + "epoch": 1.6172242717211551, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7789608240127563, + "num_tokens": 484922106.0, + "step": 12713 + }, + { + "epoch": 1.6173514819997457, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.797166109085083, + "num_tokens": 484963100.0, + "step": 12714 + }, + { + "epoch": 1.6174786922783362, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.813762366771698, + "num_tokens": 485006896.0, + "step": 12715 + }, + { + "epoch": 1.6176059025569267, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7994367480278015, + "num_tokens": 485045968.0, + "step": 12716 + }, + { + "epoch": 1.6177331128355172, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7986449003219604, + "num_tokens": 485084736.0, + "step": 12717 + }, + { + "epoch": 1.6178603231141078, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8048050999641418, + "num_tokens": 485123130.0, + "step": 12718 + }, + { + "epoch": 1.6179875333926983, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7960461378097534, + "num_tokens": 485158072.0, + "step": 12719 + }, + { + "epoch": 1.6181147436712886, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8160290718078613, + "num_tokens": 485196783.0, + "step": 12720 + }, + { + "epoch": 1.6182419539498791, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.80625319480896, + "num_tokens": 485231433.0, + "step": 12721 + }, + { + "epoch": 1.6183691642284697, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7929567098617554, + "num_tokens": 485269016.0, + "step": 12722 + }, + { + "epoch": 1.6184963745070602, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7909340858459473, + "num_tokens": 485308293.0, + "step": 12723 + }, + { + "epoch": 1.6186235847856507, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8068039417266846, + "num_tokens": 485340758.0, + "step": 12724 + }, + { + "epoch": 1.6187507950642412, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.815543532371521, + "num_tokens": 485384047.0, + "step": 12725 + }, + { + "epoch": 1.6188780053428315, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7989366054534912, + "num_tokens": 485416909.0, + "step": 12726 + }, + { + "epoch": 1.619005215621422, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8010509014129639, + "num_tokens": 485449532.0, + "step": 12727 + }, + { + "epoch": 1.6191324259000126, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8093330264091492, + "num_tokens": 485491171.0, + "step": 12728 + }, + { + "epoch": 1.6192596361786031, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8012160658836365, + "num_tokens": 485525511.0, + "step": 12729 + }, + { + "epoch": 1.6193868464571937, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7952470183372498, + "num_tokens": 485562517.0, + "step": 12730 + }, + { + "epoch": 1.6195140567357842, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7987453937530518, + "num_tokens": 485601058.0, + "step": 12731 + }, + { + "epoch": 1.6196412670143747, + "grad_norm": 10.25, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.819008469581604, + "num_tokens": 485634737.0, + "step": 12732 + }, + { + "epoch": 1.6197684772929652, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8047316074371338, + "num_tokens": 485671300.0, + "step": 12733 + }, + { + "epoch": 1.6198956875715558, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.809813380241394, + "num_tokens": 485708712.0, + "step": 12734 + }, + { + "epoch": 1.6200228978501463, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8116185665130615, + "num_tokens": 485746820.0, + "step": 12735 + }, + { + "epoch": 1.6201501081287368, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8026163578033447, + "num_tokens": 485785921.0, + "step": 12736 + }, + { + "epoch": 1.6202773184073274, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7945774793624878, + "num_tokens": 485825766.0, + "step": 12737 + }, + { + "epoch": 1.6204045286859179, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7962784767150879, + "num_tokens": 485862288.0, + "step": 12738 + }, + { + "epoch": 1.6205317389645084, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.8013111352920532, + "num_tokens": 485903582.0, + "step": 12739 + }, + { + "epoch": 1.620658949243099, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7985420227050781, + "num_tokens": 485950931.0, + "step": 12740 + }, + { + "epoch": 1.6207861595216895, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8038231134414673, + "num_tokens": 485984185.0, + "step": 12741 + }, + { + "epoch": 1.62091336980028, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.806441068649292, + "num_tokens": 486018796.0, + "step": 12742 + }, + { + "epoch": 1.6210405800788705, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7876468300819397, + "num_tokens": 486055328.0, + "step": 12743 + }, + { + "epoch": 1.6211677903574608, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7928544878959656, + "num_tokens": 486092340.0, + "step": 12744 + }, + { + "epoch": 1.6212950006360514, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7845314145088196, + "num_tokens": 486123195.0, + "step": 12745 + }, + { + "epoch": 1.621422210914642, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8185515403747559, + "num_tokens": 486167145.0, + "step": 12746 + }, + { + "epoch": 1.6215494211932324, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7939683794975281, + "num_tokens": 486207925.0, + "step": 12747 + }, + { + "epoch": 1.621676631471823, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8081806898117065, + "num_tokens": 486245446.0, + "step": 12748 + }, + { + "epoch": 1.6218038417504135, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.800182044506073, + "num_tokens": 486280461.0, + "step": 12749 + }, + { + "epoch": 1.6219310520290038, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7892799377441406, + "num_tokens": 486323926.0, + "step": 12750 + }, + { + "epoch": 1.6220582623075943, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7929592728614807, + "num_tokens": 486360016.0, + "step": 12751 + }, + { + "epoch": 1.6221854725861848, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8050570487976074, + "num_tokens": 486394994.0, + "step": 12752 + }, + { + "epoch": 1.6223126828647754, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8005768656730652, + "num_tokens": 486436909.0, + "step": 12753 + }, + { + "epoch": 1.622439893143366, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8084986209869385, + "num_tokens": 486473201.0, + "step": 12754 + }, + { + "epoch": 1.6225671034219564, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7793094515800476, + "num_tokens": 486510733.0, + "step": 12755 + }, + { + "epoch": 1.622694313700547, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.8218026161193848, + "num_tokens": 486546666.0, + "step": 12756 + }, + { + "epoch": 1.6228215239791375, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.782069206237793, + "num_tokens": 486585639.0, + "step": 12757 + }, + { + "epoch": 1.622948734257728, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8088350296020508, + "num_tokens": 486623028.0, + "step": 12758 + }, + { + "epoch": 1.6230759445363185, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7951415777206421, + "num_tokens": 486655906.0, + "step": 12759 + }, + { + "epoch": 1.623203154814909, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7947053909301758, + "num_tokens": 486689461.0, + "step": 12760 + }, + { + "epoch": 1.6233303650934996, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7871564030647278, + "num_tokens": 486728373.0, + "step": 12761 + }, + { + "epoch": 1.6234575753720901, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.815994381904602, + "num_tokens": 486769955.0, + "step": 12762 + }, + { + "epoch": 1.6235847856506807, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8074977397918701, + "num_tokens": 486806468.0, + "step": 12763 + }, + { + "epoch": 1.6237119959292712, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7824207544326782, + "num_tokens": 486846526.0, + "step": 12764 + }, + { + "epoch": 1.6238392062078617, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8215910196304321, + "num_tokens": 486886323.0, + "step": 12765 + }, + { + "epoch": 1.6239664164864522, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7754636406898499, + "num_tokens": 486918099.0, + "step": 12766 + }, + { + "epoch": 1.6240936267650428, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8159363865852356, + "num_tokens": 486953789.0, + "step": 12767 + }, + { + "epoch": 1.6242208370436333, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7975181937217712, + "num_tokens": 486996165.0, + "step": 12768 + }, + { + "epoch": 1.6243480473222236, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8110631704330444, + "num_tokens": 487028249.0, + "step": 12769 + }, + { + "epoch": 1.6244752576008141, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8043439388275146, + "num_tokens": 487070099.0, + "step": 12770 + }, + { + "epoch": 1.6246024678794047, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8039513826370239, + "num_tokens": 487107117.0, + "step": 12771 + }, + { + "epoch": 1.6247296781579952, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7872722148895264, + "num_tokens": 487146024.0, + "step": 12772 + }, + { + "epoch": 1.6248568884365857, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7932788133621216, + "num_tokens": 487189864.0, + "step": 12773 + }, + { + "epoch": 1.6249840987151762, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8045499324798584, + "num_tokens": 487230307.0, + "step": 12774 + }, + { + "epoch": 1.6251113089937665, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8022801876068115, + "num_tokens": 487261387.0, + "step": 12775 + }, + { + "epoch": 1.625238519272357, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7924126386642456, + "num_tokens": 487298611.0, + "step": 12776 + }, + { + "epoch": 1.6253657295509476, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.802229642868042, + "num_tokens": 487338172.0, + "step": 12777 + }, + { + "epoch": 1.6254929398295381, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.804570734500885, + "num_tokens": 487375126.0, + "step": 12778 + }, + { + "epoch": 1.6256201501081287, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.811870813369751, + "num_tokens": 487416539.0, + "step": 12779 + }, + { + "epoch": 1.6257473603867192, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7978181838989258, + "num_tokens": 487448429.0, + "step": 12780 + }, + { + "epoch": 1.6258745706653097, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.8031129240989685, + "num_tokens": 487486054.0, + "step": 12781 + }, + { + "epoch": 1.6260017809439002, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8118714094161987, + "num_tokens": 487523771.0, + "step": 12782 + }, + { + "epoch": 1.6261289912224908, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8049774169921875, + "num_tokens": 487561334.0, + "step": 12783 + }, + { + "epoch": 1.6262562015010813, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8053632378578186, + "num_tokens": 487606382.0, + "step": 12784 + }, + { + "epoch": 1.6263834117796718, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7974057197570801, + "num_tokens": 487647640.0, + "step": 12785 + }, + { + "epoch": 1.6265106220582624, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7959534525871277, + "num_tokens": 487682604.0, + "step": 12786 + }, + { + "epoch": 1.6266378323368529, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8023772239685059, + "num_tokens": 487718271.0, + "step": 12787 + }, + { + "epoch": 1.6267650426154434, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7870743870735168, + "num_tokens": 487752872.0, + "step": 12788 + }, + { + "epoch": 1.626892252894034, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7907198667526245, + "num_tokens": 487798309.0, + "step": 12789 + }, + { + "epoch": 1.6270194631726245, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7935959100723267, + "num_tokens": 487834052.0, + "step": 12790 + }, + { + "epoch": 1.627146673451215, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7888210415840149, + "num_tokens": 487868662.0, + "step": 12791 + }, + { + "epoch": 1.6272738837298055, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7968242764472961, + "num_tokens": 487907425.0, + "step": 12792 + }, + { + "epoch": 1.6274010940083958, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8115920424461365, + "num_tokens": 487942979.0, + "step": 12793 + }, + { + "epoch": 1.6275283042869864, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8036307692527771, + "num_tokens": 487982402.0, + "step": 12794 + }, + { + "epoch": 1.6276555145655769, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8101688623428345, + "num_tokens": 488020071.0, + "step": 12795 + }, + { + "epoch": 1.6277827248441674, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.786354660987854, + "num_tokens": 488057457.0, + "step": 12796 + }, + { + "epoch": 1.627909935122758, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7974328994750977, + "num_tokens": 488094002.0, + "step": 12797 + }, + { + "epoch": 1.6280371454013485, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7854203581809998, + "num_tokens": 488134995.0, + "step": 12798 + }, + { + "epoch": 1.6281643556799388, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8024762272834778, + "num_tokens": 488169942.0, + "step": 12799 + }, + { + "epoch": 1.6282915659585293, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.801735520362854, + "num_tokens": 488210852.0, + "step": 12800 + }, + { + "epoch": 1.6284187762371198, + "grad_norm": 11.125, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.811083197593689, + "num_tokens": 488253067.0, + "step": 12801 + }, + { + "epoch": 1.6285459865157104, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.810386061668396, + "num_tokens": 488293498.0, + "step": 12802 + }, + { + "epoch": 1.628673196794301, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8050643801689148, + "num_tokens": 488333931.0, + "step": 12803 + }, + { + "epoch": 1.6288004070728914, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8077145218849182, + "num_tokens": 488371724.0, + "step": 12804 + }, + { + "epoch": 1.628927617351482, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.798386812210083, + "num_tokens": 488415171.0, + "step": 12805 + }, + { + "epoch": 1.6290548276300725, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7982732057571411, + "num_tokens": 488451672.0, + "step": 12806 + }, + { + "epoch": 1.629182037908663, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.8194388747215271, + "num_tokens": 488492657.0, + "step": 12807 + }, + { + "epoch": 1.6293092481872535, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8078533411026001, + "num_tokens": 488530681.0, + "step": 12808 + }, + { + "epoch": 1.629436458465844, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8068581819534302, + "num_tokens": 488560542.0, + "step": 12809 + }, + { + "epoch": 1.6295636687444346, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8032511472702026, + "num_tokens": 488597491.0, + "step": 12810 + }, + { + "epoch": 1.6296908790230251, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8071513175964355, + "num_tokens": 488634556.0, + "step": 12811 + }, + { + "epoch": 1.6298180893016156, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7940274477005005, + "num_tokens": 488668174.0, + "step": 12812 + }, + { + "epoch": 1.6299452995802062, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8089063167572021, + "num_tokens": 488705994.0, + "step": 12813 + }, + { + "epoch": 1.6300725098587967, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7862640619277954, + "num_tokens": 488746908.0, + "step": 12814 + }, + { + "epoch": 1.6301997201373872, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8035615682601929, + "num_tokens": 488785607.0, + "step": 12815 + }, + { + "epoch": 1.6303269304159778, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.800588846206665, + "num_tokens": 488823831.0, + "step": 12816 + }, + { + "epoch": 1.630454140694568, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8091162443161011, + "num_tokens": 488859944.0, + "step": 12817 + }, + { + "epoch": 1.6305813509731586, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7846615314483643, + "num_tokens": 488903764.0, + "step": 12818 + }, + { + "epoch": 1.6307085612517491, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7986969351768494, + "num_tokens": 488946890.0, + "step": 12819 + }, + { + "epoch": 1.6308357715303397, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7930961847305298, + "num_tokens": 488981901.0, + "step": 12820 + }, + { + "epoch": 1.6309629818089302, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8050001263618469, + "num_tokens": 489024180.0, + "step": 12821 + }, + { + "epoch": 1.6310901920875207, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8070058822631836, + "num_tokens": 489070971.0, + "step": 12822 + }, + { + "epoch": 1.6312174023661112, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.8131063580513, + "num_tokens": 489111636.0, + "step": 12823 + }, + { + "epoch": 1.6313446126447015, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8011888265609741, + "num_tokens": 489148677.0, + "step": 12824 + }, + { + "epoch": 1.631471822923292, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7869254350662231, + "num_tokens": 489186222.0, + "step": 12825 + }, + { + "epoch": 1.6315990332018826, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7765302658081055, + "num_tokens": 489223173.0, + "step": 12826 + }, + { + "epoch": 1.6317262434804731, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7948493361473083, + "num_tokens": 489259236.0, + "step": 12827 + }, + { + "epoch": 1.6318534537590637, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7926573157310486, + "num_tokens": 489298449.0, + "step": 12828 + }, + { + "epoch": 1.6319806640376542, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7755654454231262, + "num_tokens": 489334800.0, + "step": 12829 + }, + { + "epoch": 1.6321078743162447, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8083665370941162, + "num_tokens": 489378510.0, + "step": 12830 + }, + { + "epoch": 1.6322350845948352, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7924926280975342, + "num_tokens": 489419267.0, + "step": 12831 + }, + { + "epoch": 1.6323622948734258, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8037100434303284, + "num_tokens": 489454772.0, + "step": 12832 + }, + { + "epoch": 1.6324895051520163, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7957870364189148, + "num_tokens": 489490146.0, + "step": 12833 + }, + { + "epoch": 1.6326167154306068, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7727091908454895, + "num_tokens": 489532754.0, + "step": 12834 + }, + { + "epoch": 1.6327439257091974, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8001320362091064, + "num_tokens": 489570378.0, + "step": 12835 + }, + { + "epoch": 1.6328711359877879, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7903246283531189, + "num_tokens": 489604562.0, + "step": 12836 + }, + { + "epoch": 1.6329983462663784, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8093959093093872, + "num_tokens": 489644060.0, + "step": 12837 + }, + { + "epoch": 1.633125556544969, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.804637610912323, + "num_tokens": 489682693.0, + "step": 12838 + }, + { + "epoch": 1.6332527668235595, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8069263696670532, + "num_tokens": 489720786.0, + "step": 12839 + }, + { + "epoch": 1.63337997710215, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8045153617858887, + "num_tokens": 489758773.0, + "step": 12840 + }, + { + "epoch": 1.6335071873807405, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7980067729949951, + "num_tokens": 489793436.0, + "step": 12841 + }, + { + "epoch": 1.6336343976593308, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.8199430704116821, + "num_tokens": 489828634.0, + "step": 12842 + }, + { + "epoch": 1.6337616079379214, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8058744072914124, + "num_tokens": 489866335.0, + "step": 12843 + }, + { + "epoch": 1.6338888182165119, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8123173713684082, + "num_tokens": 489903059.0, + "step": 12844 + }, + { + "epoch": 1.6340160284951024, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8066938519477844, + "num_tokens": 489940186.0, + "step": 12845 + }, + { + "epoch": 1.634143238773693, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.8174237012863159, + "num_tokens": 489984732.0, + "step": 12846 + }, + { + "epoch": 1.6342704490522835, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8020176887512207, + "num_tokens": 490016699.0, + "step": 12847 + }, + { + "epoch": 1.6343976593308738, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7901681065559387, + "num_tokens": 490051351.0, + "step": 12848 + }, + { + "epoch": 1.6345248696094643, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8077201843261719, + "num_tokens": 490089668.0, + "step": 12849 + }, + { + "epoch": 1.6346520798880548, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8059749603271484, + "num_tokens": 490129651.0, + "step": 12850 + }, + { + "epoch": 1.6347792901666454, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7942836284637451, + "num_tokens": 490168441.0, + "step": 12851 + }, + { + "epoch": 1.6349065004452359, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8120154142379761, + "num_tokens": 490204754.0, + "step": 12852 + }, + { + "epoch": 1.6350337107238264, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8193219900131226, + "num_tokens": 490244178.0, + "step": 12853 + }, + { + "epoch": 1.635160921002417, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.796575129032135, + "num_tokens": 490281161.0, + "step": 12854 + }, + { + "epoch": 1.6352881312810075, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7995401620864868, + "num_tokens": 490314892.0, + "step": 12855 + }, + { + "epoch": 1.635415341559598, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7858060598373413, + "num_tokens": 490352600.0, + "step": 12856 + }, + { + "epoch": 1.6355425518381885, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.8146364688873291, + "num_tokens": 490384487.0, + "step": 12857 + }, + { + "epoch": 1.635669762116779, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7823631763458252, + "num_tokens": 490422199.0, + "step": 12858 + }, + { + "epoch": 1.6357969723953696, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7943688631057739, + "num_tokens": 490458058.0, + "step": 12859 + }, + { + "epoch": 1.6359241826739601, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8130495548248291, + "num_tokens": 490498447.0, + "step": 12860 + }, + { + "epoch": 1.6360513929525506, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7892758846282959, + "num_tokens": 490538502.0, + "step": 12861 + }, + { + "epoch": 1.6361786032311412, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8048002123832703, + "num_tokens": 490575752.0, + "step": 12862 + }, + { + "epoch": 1.6363058135097317, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7973254919052124, + "num_tokens": 490613502.0, + "step": 12863 + }, + { + "epoch": 1.6364330237883222, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.795433521270752, + "num_tokens": 490653820.0, + "step": 12864 + }, + { + "epoch": 1.6365602340669128, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7785731554031372, + "num_tokens": 490693332.0, + "step": 12865 + }, + { + "epoch": 1.636687444345503, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8002331256866455, + "num_tokens": 490733135.0, + "step": 12866 + }, + { + "epoch": 1.6368146546240936, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8055768013000488, + "num_tokens": 490775505.0, + "step": 12867 + }, + { + "epoch": 1.6369418649026841, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8032634258270264, + "num_tokens": 490817783.0, + "step": 12868 + }, + { + "epoch": 1.6370690751812746, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7928223609924316, + "num_tokens": 490849464.0, + "step": 12869 + }, + { + "epoch": 1.6371962854598652, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7933365106582642, + "num_tokens": 490891925.0, + "step": 12870 + }, + { + "epoch": 1.6373234957384557, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7924981117248535, + "num_tokens": 490926221.0, + "step": 12871 + }, + { + "epoch": 1.6374507060170462, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.794887125492096, + "num_tokens": 490964141.0, + "step": 12872 + }, + { + "epoch": 1.6375779162956365, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7958328723907471, + "num_tokens": 490997917.0, + "step": 12873 + }, + { + "epoch": 1.637705126574227, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.8151829838752747, + "num_tokens": 491029244.0, + "step": 12874 + }, + { + "epoch": 1.6378323368528176, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7937477827072144, + "num_tokens": 491066591.0, + "step": 12875 + }, + { + "epoch": 1.6379595471314081, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8029694557189941, + "num_tokens": 491106546.0, + "step": 12876 + }, + { + "epoch": 1.6380867574099987, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7936551570892334, + "num_tokens": 491145512.0, + "step": 12877 + }, + { + "epoch": 1.6382139676885892, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7963211536407471, + "num_tokens": 491185624.0, + "step": 12878 + }, + { + "epoch": 1.6383411779671797, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8016155958175659, + "num_tokens": 491220982.0, + "step": 12879 + }, + { + "epoch": 1.6384683882457702, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8075347542762756, + "num_tokens": 491262501.0, + "step": 12880 + }, + { + "epoch": 1.6385955985243608, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8131476640701294, + "num_tokens": 491298078.0, + "step": 12881 + }, + { + "epoch": 1.6387228088029513, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8103829026222229, + "num_tokens": 491336203.0, + "step": 12882 + }, + { + "epoch": 1.6388500190815418, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7913392782211304, + "num_tokens": 491374723.0, + "step": 12883 + }, + { + "epoch": 1.6389772293601323, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.798555314540863, + "num_tokens": 491410546.0, + "step": 12884 + }, + { + "epoch": 1.6391044396387229, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7987396717071533, + "num_tokens": 491450642.0, + "step": 12885 + }, + { + "epoch": 1.6392316499173134, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7983108162879944, + "num_tokens": 491489893.0, + "step": 12886 + }, + { + "epoch": 1.639358860195904, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7882767915725708, + "num_tokens": 491535870.0, + "step": 12887 + }, + { + "epoch": 1.6394860704744945, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7995128631591797, + "num_tokens": 491575988.0, + "step": 12888 + }, + { + "epoch": 1.639613280753085, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8042020797729492, + "num_tokens": 491616822.0, + "step": 12889 + }, + { + "epoch": 1.6397404910316755, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.8237868547439575, + "num_tokens": 491654132.0, + "step": 12890 + }, + { + "epoch": 1.6398677013102658, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7887539863586426, + "num_tokens": 491697215.0, + "step": 12891 + }, + { + "epoch": 1.6399949115888564, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7759611010551453, + "num_tokens": 491731338.0, + "step": 12892 + }, + { + "epoch": 1.6401221218674469, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7948526740074158, + "num_tokens": 491770808.0, + "step": 12893 + }, + { + "epoch": 1.6402493321460374, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8080474138259888, + "num_tokens": 491804167.0, + "step": 12894 + }, + { + "epoch": 1.640376542424628, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7990829944610596, + "num_tokens": 491840094.0, + "step": 12895 + }, + { + "epoch": 1.6405037527032185, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.786535382270813, + "num_tokens": 491881643.0, + "step": 12896 + }, + { + "epoch": 1.6406309629818088, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7904373407363892, + "num_tokens": 491920951.0, + "step": 12897 + }, + { + "epoch": 1.6407581732603993, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7936210632324219, + "num_tokens": 491954599.0, + "step": 12898 + }, + { + "epoch": 1.6408853835389898, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8072022795677185, + "num_tokens": 491995375.0, + "step": 12899 + }, + { + "epoch": 1.6410125938175804, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.794702410697937, + "num_tokens": 492035297.0, + "step": 12900 + }, + { + "epoch": 1.6411398040961709, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7978376746177673, + "num_tokens": 492071003.0, + "step": 12901 + }, + { + "epoch": 1.6412670143747614, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7961528301239014, + "num_tokens": 492106046.0, + "step": 12902 + }, + { + "epoch": 1.641394224653352, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8131992220878601, + "num_tokens": 492147252.0, + "step": 12903 + }, + { + "epoch": 1.6415214349319425, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7921208143234253, + "num_tokens": 492188185.0, + "step": 12904 + }, + { + "epoch": 1.641648645210533, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8021214008331299, + "num_tokens": 492219630.0, + "step": 12905 + }, + { + "epoch": 1.6417758554891235, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.801232099533081, + "num_tokens": 492251955.0, + "step": 12906 + }, + { + "epoch": 1.641903065767714, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8099319934844971, + "num_tokens": 492293484.0, + "step": 12907 + }, + { + "epoch": 1.6420302760463046, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7900699377059937, + "num_tokens": 492331169.0, + "step": 12908 + }, + { + "epoch": 1.642157486324895, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8098951578140259, + "num_tokens": 492366071.0, + "step": 12909 + }, + { + "epoch": 1.6422846966034856, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8182128667831421, + "num_tokens": 492396820.0, + "step": 12910 + }, + { + "epoch": 1.6424119068820762, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8080253601074219, + "num_tokens": 492435282.0, + "step": 12911 + }, + { + "epoch": 1.6425391171606667, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8005373477935791, + "num_tokens": 492470195.0, + "step": 12912 + }, + { + "epoch": 1.6426663274392572, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7941297292709351, + "num_tokens": 492514265.0, + "step": 12913 + }, + { + "epoch": 1.6427935377178478, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8009867668151855, + "num_tokens": 492550348.0, + "step": 12914 + }, + { + "epoch": 1.642920747996438, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8095788955688477, + "num_tokens": 492590648.0, + "step": 12915 + }, + { + "epoch": 1.6430479582750286, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.807490348815918, + "num_tokens": 492629523.0, + "step": 12916 + }, + { + "epoch": 1.6431751685536191, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7959907054901123, + "num_tokens": 492668392.0, + "step": 12917 + }, + { + "epoch": 1.6433023788322096, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7796993255615234, + "num_tokens": 492696073.0, + "step": 12918 + }, + { + "epoch": 1.6434295891108002, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8019927740097046, + "num_tokens": 492728590.0, + "step": 12919 + }, + { + "epoch": 1.6435567993893907, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.794158935546875, + "num_tokens": 492767290.0, + "step": 12920 + }, + { + "epoch": 1.6436840096679812, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8141714334487915, + "num_tokens": 492801019.0, + "step": 12921 + }, + { + "epoch": 1.6438112199465715, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8037406802177429, + "num_tokens": 492841220.0, + "step": 12922 + }, + { + "epoch": 1.643938430225162, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7911801338195801, + "num_tokens": 492885076.0, + "step": 12923 + }, + { + "epoch": 1.6440656405037526, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7904906868934631, + "num_tokens": 492918554.0, + "step": 12924 + }, + { + "epoch": 1.6441928507823431, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8180110454559326, + "num_tokens": 492961890.0, + "step": 12925 + }, + { + "epoch": 1.6443200610609336, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8151600956916809, + "num_tokens": 492997768.0, + "step": 12926 + }, + { + "epoch": 1.6444472713395242, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.8288934230804443, + "num_tokens": 493037017.0, + "step": 12927 + }, + { + "epoch": 1.6445744816181147, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8133643865585327, + "num_tokens": 493074579.0, + "step": 12928 + }, + { + "epoch": 1.6447016918967052, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.803101122379303, + "num_tokens": 493111249.0, + "step": 12929 + }, + { + "epoch": 1.6448289021752958, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7832058072090149, + "num_tokens": 493150723.0, + "step": 12930 + }, + { + "epoch": 1.6449561124538863, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7933434844017029, + "num_tokens": 493189530.0, + "step": 12931 + }, + { + "epoch": 1.6450833227324768, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.807640016078949, + "num_tokens": 493231369.0, + "step": 12932 + }, + { + "epoch": 1.6452105330110673, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.793272078037262, + "num_tokens": 493278430.0, + "step": 12933 + }, + { + "epoch": 1.6453377432896579, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8025039434432983, + "num_tokens": 493312258.0, + "step": 12934 + }, + { + "epoch": 1.6454649535682484, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7948981523513794, + "num_tokens": 493347725.0, + "step": 12935 + }, + { + "epoch": 1.645592163846839, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8025979995727539, + "num_tokens": 493388119.0, + "step": 12936 + }, + { + "epoch": 1.6457193741254295, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.801379919052124, + "num_tokens": 493424607.0, + "step": 12937 + }, + { + "epoch": 1.64584658440402, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7757996916770935, + "num_tokens": 493459546.0, + "step": 12938 + }, + { + "epoch": 1.6459737946826105, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.803310751914978, + "num_tokens": 493494162.0, + "step": 12939 + }, + { + "epoch": 1.6461010049612008, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7997668981552124, + "num_tokens": 493537523.0, + "step": 12940 + }, + { + "epoch": 1.6462282152397913, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.8019322156906128, + "num_tokens": 493580065.0, + "step": 12941 + }, + { + "epoch": 1.6463554255183819, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8111850619316101, + "num_tokens": 493626717.0, + "step": 12942 + }, + { + "epoch": 1.6464826357969724, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8089548349380493, + "num_tokens": 493666266.0, + "step": 12943 + }, + { + "epoch": 1.646609846075563, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7837214469909668, + "num_tokens": 493703184.0, + "step": 12944 + }, + { + "epoch": 1.6467370563541535, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.5929, + "mean_token_accuracy": 0.8211358189582825, + "num_tokens": 493744061.0, + "step": 12945 + }, + { + "epoch": 1.6468642666327438, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8024617433547974, + "num_tokens": 493783585.0, + "step": 12946 + }, + { + "epoch": 1.6469914769113343, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8125004768371582, + "num_tokens": 493827781.0, + "step": 12947 + }, + { + "epoch": 1.6471186871899248, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8037866353988647, + "num_tokens": 493864418.0, + "step": 12948 + }, + { + "epoch": 1.6472458974685154, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.792251467704773, + "num_tokens": 493895226.0, + "step": 12949 + }, + { + "epoch": 1.6473731077471059, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.790328860282898, + "num_tokens": 493935210.0, + "step": 12950 + }, + { + "epoch": 1.6475003180256964, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7998292446136475, + "num_tokens": 493966368.0, + "step": 12951 + }, + { + "epoch": 1.647627528304287, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7910382747650146, + "num_tokens": 494004304.0, + "step": 12952 + }, + { + "epoch": 1.6477547385828775, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8131251335144043, + "num_tokens": 494044758.0, + "step": 12953 + }, + { + "epoch": 1.647881948861468, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7813252210617065, + "num_tokens": 494080624.0, + "step": 12954 + }, + { + "epoch": 1.6480091591400585, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8012940287590027, + "num_tokens": 494116215.0, + "step": 12955 + }, + { + "epoch": 1.648136369418649, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7960607409477234, + "num_tokens": 494150317.0, + "step": 12956 + }, + { + "epoch": 1.6482635796972396, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8140968680381775, + "num_tokens": 494187273.0, + "step": 12957 + }, + { + "epoch": 1.64839078997583, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7988232970237732, + "num_tokens": 494226921.0, + "step": 12958 + }, + { + "epoch": 1.6485180002544206, + "grad_norm": 67.0, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8031610250473022, + "num_tokens": 494263141.0, + "step": 12959 + }, + { + "epoch": 1.6486452105330112, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7966755628585815, + "num_tokens": 494307925.0, + "step": 12960 + }, + { + "epoch": 1.6487724208116017, + "grad_norm": 61.75, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7995069026947021, + "num_tokens": 494353428.0, + "step": 12961 + }, + { + "epoch": 1.6488996310901922, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7820526957511902, + "num_tokens": 494383559.0, + "step": 12962 + }, + { + "epoch": 1.6490268413687827, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7989825010299683, + "num_tokens": 494419564.0, + "step": 12963 + }, + { + "epoch": 1.649154051647373, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8089537620544434, + "num_tokens": 494466886.0, + "step": 12964 + }, + { + "epoch": 1.6492812619259636, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.783108115196228, + "num_tokens": 494506672.0, + "step": 12965 + }, + { + "epoch": 1.649408472204554, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7949479818344116, + "num_tokens": 494541881.0, + "step": 12966 + }, + { + "epoch": 1.6495356824831446, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8073608875274658, + "num_tokens": 494583546.0, + "step": 12967 + }, + { + "epoch": 1.6496628927617352, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7874562740325928, + "num_tokens": 494616625.0, + "step": 12968 + }, + { + "epoch": 1.6497901030403257, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.8186004757881165, + "num_tokens": 494654590.0, + "step": 12969 + }, + { + "epoch": 1.6499173133189162, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8163574934005737, + "num_tokens": 494692471.0, + "step": 12970 + }, + { + "epoch": 1.6500445235975065, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7907221913337708, + "num_tokens": 494730347.0, + "step": 12971 + }, + { + "epoch": 1.650171733876097, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7838883399963379, + "num_tokens": 494763753.0, + "step": 12972 + }, + { + "epoch": 1.6502989441546876, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7914623618125916, + "num_tokens": 494800416.0, + "step": 12973 + }, + { + "epoch": 1.6504261544332781, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7969282865524292, + "num_tokens": 494842733.0, + "step": 12974 + }, + { + "epoch": 1.6505533647118686, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7960587739944458, + "num_tokens": 494878661.0, + "step": 12975 + }, + { + "epoch": 1.6506805749904592, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7937971353530884, + "num_tokens": 494918142.0, + "step": 12976 + }, + { + "epoch": 1.6508077852690497, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7999660968780518, + "num_tokens": 494956947.0, + "step": 12977 + }, + { + "epoch": 1.6509349955476402, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7937487363815308, + "num_tokens": 494989850.0, + "step": 12978 + }, + { + "epoch": 1.6510622058262308, + "grad_norm": 97.5, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7892510890960693, + "num_tokens": 495026770.0, + "step": 12979 + }, + { + "epoch": 1.6511894161048213, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8086118698120117, + "num_tokens": 495065009.0, + "step": 12980 + }, + { + "epoch": 1.6513166263834118, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7930231690406799, + "num_tokens": 495096914.0, + "step": 12981 + }, + { + "epoch": 1.6514438366620023, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8138378262519836, + "num_tokens": 495138348.0, + "step": 12982 + }, + { + "epoch": 1.6515710469405929, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7916269302368164, + "num_tokens": 495177026.0, + "step": 12983 + }, + { + "epoch": 1.6516982572191834, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8070927858352661, + "num_tokens": 495213850.0, + "step": 12984 + }, + { + "epoch": 1.651825467497774, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8116734623908997, + "num_tokens": 495257213.0, + "step": 12985 + }, + { + "epoch": 1.6519526777763645, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7919214963912964, + "num_tokens": 495300018.0, + "step": 12986 + }, + { + "epoch": 1.652079888054955, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7990528345108032, + "num_tokens": 495336471.0, + "step": 12987 + }, + { + "epoch": 1.6522070983335455, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8039886951446533, + "num_tokens": 495374680.0, + "step": 12988 + }, + { + "epoch": 1.6523343086121358, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8027331829071045, + "num_tokens": 495414096.0, + "step": 12989 + }, + { + "epoch": 1.6524615188907263, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7854866981506348, + "num_tokens": 495452955.0, + "step": 12990 + }, + { + "epoch": 1.6525887291693169, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7964884042739868, + "num_tokens": 495490091.0, + "step": 12991 + }, + { + "epoch": 1.6527159394479074, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8104816675186157, + "num_tokens": 495524543.0, + "step": 12992 + }, + { + "epoch": 1.652843149726498, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7896900177001953, + "num_tokens": 495562068.0, + "step": 12993 + }, + { + "epoch": 1.6529703600050885, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8060653209686279, + "num_tokens": 495598355.0, + "step": 12994 + }, + { + "epoch": 1.6530975702836788, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7794054746627808, + "num_tokens": 495634735.0, + "step": 12995 + }, + { + "epoch": 1.6532247805622693, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7945747375488281, + "num_tokens": 495668804.0, + "step": 12996 + }, + { + "epoch": 1.6533519908408598, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.803969144821167, + "num_tokens": 495707129.0, + "step": 12997 + }, + { + "epoch": 1.6534792011194503, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7804125547409058, + "num_tokens": 495746501.0, + "step": 12998 + }, + { + "epoch": 1.6536064113980409, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6018, + "mean_token_accuracy": 0.8212299346923828, + "num_tokens": 495789723.0, + "step": 12999 + }, + { + "epoch": 1.6537336216766314, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7897478938102722, + "num_tokens": 495827073.0, + "step": 13000 + }, + { + "epoch": 1.653860831955222, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7861319780349731, + "num_tokens": 495862362.0, + "step": 13001 + }, + { + "epoch": 1.6539880422338125, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8070459365844727, + "num_tokens": 495905546.0, + "step": 13002 + }, + { + "epoch": 1.654115252512403, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7922484874725342, + "num_tokens": 495942291.0, + "step": 13003 + }, + { + "epoch": 1.6542424627909935, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7872244119644165, + "num_tokens": 495978631.0, + "step": 13004 + }, + { + "epoch": 1.654369673069584, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7934579253196716, + "num_tokens": 496016626.0, + "step": 13005 + }, + { + "epoch": 1.6544968833481746, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7835645079612732, + "num_tokens": 496051949.0, + "step": 13006 + }, + { + "epoch": 1.654624093626765, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8096058368682861, + "num_tokens": 496092693.0, + "step": 13007 + }, + { + "epoch": 1.6547513039053556, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7962072491645813, + "num_tokens": 496132853.0, + "step": 13008 + }, + { + "epoch": 1.6548785141839462, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7850399613380432, + "num_tokens": 496168303.0, + "step": 13009 + }, + { + "epoch": 1.6550057244625367, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7873484492301941, + "num_tokens": 496206132.0, + "step": 13010 + }, + { + "epoch": 1.6551329347411272, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7924124002456665, + "num_tokens": 496247542.0, + "step": 13011 + }, + { + "epoch": 1.6552601450197177, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.81438148021698, + "num_tokens": 496285698.0, + "step": 13012 + }, + { + "epoch": 1.655387355298308, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7988259792327881, + "num_tokens": 496322216.0, + "step": 13013 + }, + { + "epoch": 1.6555145655768986, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7944309711456299, + "num_tokens": 496357687.0, + "step": 13014 + }, + { + "epoch": 1.655641775855489, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8166307806968689, + "num_tokens": 496400875.0, + "step": 13015 + }, + { + "epoch": 1.6557689861340796, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7988947629928589, + "num_tokens": 496432076.0, + "step": 13016 + }, + { + "epoch": 1.6558961964126702, + "grad_norm": 59.25, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7980035543441772, + "num_tokens": 496466852.0, + "step": 13017 + }, + { + "epoch": 1.6560234066912607, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.8012320399284363, + "num_tokens": 496508890.0, + "step": 13018 + }, + { + "epoch": 1.6561506169698512, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.8148641586303711, + "num_tokens": 496537964.0, + "step": 13019 + }, + { + "epoch": 1.6562778272484415, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7829961776733398, + "num_tokens": 496571263.0, + "step": 13020 + }, + { + "epoch": 1.656405037527032, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8051134347915649, + "num_tokens": 496610058.0, + "step": 13021 + }, + { + "epoch": 1.6565322478056226, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7955564260482788, + "num_tokens": 496640542.0, + "step": 13022 + }, + { + "epoch": 1.656659458084213, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7917056083679199, + "num_tokens": 496678030.0, + "step": 13023 + }, + { + "epoch": 1.6567866683628036, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7966338396072388, + "num_tokens": 496714057.0, + "step": 13024 + }, + { + "epoch": 1.6569138786413942, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8159594535827637, + "num_tokens": 496744501.0, + "step": 13025 + }, + { + "epoch": 1.6570410889199847, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8095229864120483, + "num_tokens": 496780233.0, + "step": 13026 + }, + { + "epoch": 1.6571682991985752, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7951687574386597, + "num_tokens": 496809645.0, + "step": 13027 + }, + { + "epoch": 1.6572955094771658, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.8018529415130615, + "num_tokens": 496854871.0, + "step": 13028 + }, + { + "epoch": 1.6574227197557563, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.8172270059585571, + "num_tokens": 496891958.0, + "step": 13029 + }, + { + "epoch": 1.6575499300343468, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7871263027191162, + "num_tokens": 496928550.0, + "step": 13030 + }, + { + "epoch": 1.6576771403129373, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.801406979560852, + "num_tokens": 496972009.0, + "step": 13031 + }, + { + "epoch": 1.6578043505915279, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8007733821868896, + "num_tokens": 497013605.0, + "step": 13032 + }, + { + "epoch": 1.6579315608701184, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8034496903419495, + "num_tokens": 497048712.0, + "step": 13033 + }, + { + "epoch": 1.658058771148709, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8191503882408142, + "num_tokens": 497089161.0, + "step": 13034 + }, + { + "epoch": 1.6581859814272994, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7738485932350159, + "num_tokens": 497127253.0, + "step": 13035 + }, + { + "epoch": 1.65831319170589, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7955204844474792, + "num_tokens": 497167637.0, + "step": 13036 + }, + { + "epoch": 1.6584404019844805, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.800186276435852, + "num_tokens": 497205663.0, + "step": 13037 + }, + { + "epoch": 1.6585676122630708, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.8171640634536743, + "num_tokens": 497242146.0, + "step": 13038 + }, + { + "epoch": 1.6586948225416613, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8090094923973083, + "num_tokens": 497279339.0, + "step": 13039 + }, + { + "epoch": 1.6588220328202519, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7925026416778564, + "num_tokens": 497318530.0, + "step": 13040 + }, + { + "epoch": 1.6589492430988424, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.790469765663147, + "num_tokens": 497352943.0, + "step": 13041 + }, + { + "epoch": 1.659076453377433, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7911382913589478, + "num_tokens": 497387729.0, + "step": 13042 + }, + { + "epoch": 1.6592036636560235, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8013184070587158, + "num_tokens": 497427875.0, + "step": 13043 + }, + { + "epoch": 1.6593308739346138, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7934086918830872, + "num_tokens": 497469076.0, + "step": 13044 + }, + { + "epoch": 1.6594580842132043, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.8008689284324646, + "num_tokens": 497509556.0, + "step": 13045 + }, + { + "epoch": 1.6595852944917948, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7841238975524902, + "num_tokens": 497548151.0, + "step": 13046 + }, + { + "epoch": 1.6597125047703853, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.771561861038208, + "num_tokens": 497589088.0, + "step": 13047 + }, + { + "epoch": 1.6598397150489759, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8137370944023132, + "num_tokens": 497632095.0, + "step": 13048 + }, + { + "epoch": 1.6599669253275664, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8114433288574219, + "num_tokens": 497665148.0, + "step": 13049 + }, + { + "epoch": 1.660094135606157, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.794715166091919, + "num_tokens": 497700134.0, + "step": 13050 + }, + { + "epoch": 1.6602213458847475, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8068273067474365, + "num_tokens": 497743101.0, + "step": 13051 + }, + { + "epoch": 1.660348556163338, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.783128023147583, + "num_tokens": 497775112.0, + "step": 13052 + }, + { + "epoch": 1.6604757664419285, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8124898672103882, + "num_tokens": 497813320.0, + "step": 13053 + }, + { + "epoch": 1.660602976720519, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7979793548583984, + "num_tokens": 497852780.0, + "step": 13054 + }, + { + "epoch": 1.6607301869991096, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7971969842910767, + "num_tokens": 497886313.0, + "step": 13055 + }, + { + "epoch": 1.6608573972777, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.8003883957862854, + "num_tokens": 497924108.0, + "step": 13056 + }, + { + "epoch": 1.6609846075562906, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7888891696929932, + "num_tokens": 497964656.0, + "step": 13057 + }, + { + "epoch": 1.6611118178348812, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7837649583816528, + "num_tokens": 498005436.0, + "step": 13058 + }, + { + "epoch": 1.6612390281134717, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7914751768112183, + "num_tokens": 498042237.0, + "step": 13059 + }, + { + "epoch": 1.6613662383920622, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7970402240753174, + "num_tokens": 498083008.0, + "step": 13060 + }, + { + "epoch": 1.6614934486706527, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7700679302215576, + "num_tokens": 498120550.0, + "step": 13061 + }, + { + "epoch": 1.661620658949243, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7980122566223145, + "num_tokens": 498153033.0, + "step": 13062 + }, + { + "epoch": 1.6617478692278336, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.8024114370346069, + "num_tokens": 498190491.0, + "step": 13063 + }, + { + "epoch": 1.661875079506424, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7942603826522827, + "num_tokens": 498235889.0, + "step": 13064 + }, + { + "epoch": 1.6620022897850146, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.797850489616394, + "num_tokens": 498277191.0, + "step": 13065 + }, + { + "epoch": 1.6621295000636052, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7854518890380859, + "num_tokens": 498318443.0, + "step": 13066 + }, + { + "epoch": 1.6622567103421957, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7859420776367188, + "num_tokens": 498355955.0, + "step": 13067 + }, + { + "epoch": 1.662383920620786, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7959491014480591, + "num_tokens": 498388588.0, + "step": 13068 + }, + { + "epoch": 1.6625111308993765, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7976423501968384, + "num_tokens": 498430661.0, + "step": 13069 + }, + { + "epoch": 1.662638341177967, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7862842082977295, + "num_tokens": 498470428.0, + "step": 13070 + }, + { + "epoch": 1.6627655514565576, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7975882291793823, + "num_tokens": 498512031.0, + "step": 13071 + }, + { + "epoch": 1.662892761735148, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.8217180967330933, + "num_tokens": 498548344.0, + "step": 13072 + }, + { + "epoch": 1.6630199720137386, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8149263858795166, + "num_tokens": 498590115.0, + "step": 13073 + }, + { + "epoch": 1.6631471822923292, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7999908924102783, + "num_tokens": 498634034.0, + "step": 13074 + }, + { + "epoch": 1.6632743925709197, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7948108911514282, + "num_tokens": 498675086.0, + "step": 13075 + }, + { + "epoch": 1.6634016028495102, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.807098388671875, + "num_tokens": 498711374.0, + "step": 13076 + }, + { + "epoch": 1.6635288131281007, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8074386715888977, + "num_tokens": 498753050.0, + "step": 13077 + }, + { + "epoch": 1.6636560234066913, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8003258109092712, + "num_tokens": 498790406.0, + "step": 13078 + }, + { + "epoch": 1.6637832336852818, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8000187277793884, + "num_tokens": 498827712.0, + "step": 13079 + }, + { + "epoch": 1.6639104439638723, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8067023754119873, + "num_tokens": 498862283.0, + "step": 13080 + }, + { + "epoch": 1.6640376542424629, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7981299161911011, + "num_tokens": 498902686.0, + "step": 13081 + }, + { + "epoch": 1.6641648645210534, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7853959202766418, + "num_tokens": 498942785.0, + "step": 13082 + }, + { + "epoch": 1.664292074799644, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8068909645080566, + "num_tokens": 498982761.0, + "step": 13083 + }, + { + "epoch": 1.6644192850782344, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.783515214920044, + "num_tokens": 499025762.0, + "step": 13084 + }, + { + "epoch": 1.664546495356825, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7923834323883057, + "num_tokens": 499063394.0, + "step": 13085 + }, + { + "epoch": 1.6646737056354155, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8140771389007568, + "num_tokens": 499105824.0, + "step": 13086 + }, + { + "epoch": 1.6648009159140058, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8035705089569092, + "num_tokens": 499140935.0, + "step": 13087 + }, + { + "epoch": 1.6649281261925963, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7968564033508301, + "num_tokens": 499179492.0, + "step": 13088 + }, + { + "epoch": 1.6650553364711869, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8140251636505127, + "num_tokens": 499214716.0, + "step": 13089 + }, + { + "epoch": 1.6651825467497774, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8064224720001221, + "num_tokens": 499250337.0, + "step": 13090 + }, + { + "epoch": 1.665309757028368, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8057509660720825, + "num_tokens": 499283675.0, + "step": 13091 + }, + { + "epoch": 1.6654369673069584, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8092653751373291, + "num_tokens": 499319153.0, + "step": 13092 + }, + { + "epoch": 1.6655641775855488, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7916913628578186, + "num_tokens": 499356675.0, + "step": 13093 + }, + { + "epoch": 1.6656913878641393, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8144861459732056, + "num_tokens": 499392129.0, + "step": 13094 + }, + { + "epoch": 1.6658185981427298, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.808042585849762, + "num_tokens": 499424958.0, + "step": 13095 + }, + { + "epoch": 1.6659458084213203, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8075016140937805, + "num_tokens": 499467769.0, + "step": 13096 + }, + { + "epoch": 1.6660730186999109, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7893235683441162, + "num_tokens": 499513828.0, + "step": 13097 + }, + { + "epoch": 1.6662002289785014, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8050650358200073, + "num_tokens": 499548219.0, + "step": 13098 + }, + { + "epoch": 1.666327439257092, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7823474407196045, + "num_tokens": 499587158.0, + "step": 13099 + }, + { + "epoch": 1.6664546495356825, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8041568994522095, + "num_tokens": 499625235.0, + "step": 13100 + }, + { + "epoch": 1.666581859814273, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7769755125045776, + "num_tokens": 499662842.0, + "step": 13101 + }, + { + "epoch": 1.6667090700928635, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7752137780189514, + "num_tokens": 499703464.0, + "step": 13102 + }, + { + "epoch": 1.666836280371454, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8031166195869446, + "num_tokens": 499739160.0, + "step": 13103 + }, + { + "epoch": 1.6669634906500446, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.780685305595398, + "num_tokens": 499775024.0, + "step": 13104 + }, + { + "epoch": 1.667090700928635, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.8020722270011902, + "num_tokens": 499816479.0, + "step": 13105 + }, + { + "epoch": 1.6672179112072256, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8049662113189697, + "num_tokens": 499848790.0, + "step": 13106 + }, + { + "epoch": 1.6673451214858162, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8213387727737427, + "num_tokens": 499890955.0, + "step": 13107 + }, + { + "epoch": 1.6674723317644067, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7997289896011353, + "num_tokens": 499924813.0, + "step": 13108 + }, + { + "epoch": 1.6675995420429972, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7993939518928528, + "num_tokens": 499960032.0, + "step": 13109 + }, + { + "epoch": 1.6677267523215877, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8144936561584473, + "num_tokens": 499995868.0, + "step": 13110 + }, + { + "epoch": 1.667853962600178, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8029676079750061, + "num_tokens": 500031662.0, + "step": 13111 + }, + { + "epoch": 1.6679811728787686, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7946226000785828, + "num_tokens": 500070189.0, + "step": 13112 + }, + { + "epoch": 1.668108383157359, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7963790893554688, + "num_tokens": 500109999.0, + "step": 13113 + }, + { + "epoch": 1.6682355934359496, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7854104042053223, + "num_tokens": 500151473.0, + "step": 13114 + }, + { + "epoch": 1.6683628037145402, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7939039468765259, + "num_tokens": 500187436.0, + "step": 13115 + }, + { + "epoch": 1.6684900139931307, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7981431484222412, + "num_tokens": 500222433.0, + "step": 13116 + }, + { + "epoch": 1.668617224271721, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7958415746688843, + "num_tokens": 500257390.0, + "step": 13117 + }, + { + "epoch": 1.6687444345503115, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.801937460899353, + "num_tokens": 500301260.0, + "step": 13118 + }, + { + "epoch": 1.668871644828902, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8101720213890076, + "num_tokens": 500343453.0, + "step": 13119 + }, + { + "epoch": 1.6689988551074926, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.803816556930542, + "num_tokens": 500375133.0, + "step": 13120 + }, + { + "epoch": 1.669126065386083, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7885900735855103, + "num_tokens": 500408540.0, + "step": 13121 + }, + { + "epoch": 1.6692532756646736, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8125429153442383, + "num_tokens": 500446243.0, + "step": 13122 + }, + { + "epoch": 1.6693804859432642, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7690802812576294, + "num_tokens": 500484313.0, + "step": 13123 + }, + { + "epoch": 1.6695076962218547, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8111118078231812, + "num_tokens": 500525566.0, + "step": 13124 + }, + { + "epoch": 1.6696349065004452, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8104543685913086, + "num_tokens": 500566274.0, + "step": 13125 + }, + { + "epoch": 1.6697621167790357, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7913896441459656, + "num_tokens": 500602769.0, + "step": 13126 + }, + { + "epoch": 1.6698893270576263, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7835935354232788, + "num_tokens": 500642895.0, + "step": 13127 + }, + { + "epoch": 1.6700165373362168, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7978717088699341, + "num_tokens": 500681461.0, + "step": 13128 + }, + { + "epoch": 1.6701437476148073, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7868589162826538, + "num_tokens": 500723395.0, + "step": 13129 + }, + { + "epoch": 1.6702709578933979, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7990283370018005, + "num_tokens": 500760571.0, + "step": 13130 + }, + { + "epoch": 1.6703981681719884, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7854713201522827, + "num_tokens": 500797858.0, + "step": 13131 + }, + { + "epoch": 1.670525378450579, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8076118230819702, + "num_tokens": 500840199.0, + "step": 13132 + }, + { + "epoch": 1.6706525887291694, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7934812307357788, + "num_tokens": 500883290.0, + "step": 13133 + }, + { + "epoch": 1.67077979900776, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8035367131233215, + "num_tokens": 500923072.0, + "step": 13134 + }, + { + "epoch": 1.6709070092863505, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8026633262634277, + "num_tokens": 500958726.0, + "step": 13135 + }, + { + "epoch": 1.6710342195649408, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7945905327796936, + "num_tokens": 500996746.0, + "step": 13136 + }, + { + "epoch": 1.6711614298435313, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7896713018417358, + "num_tokens": 501034135.0, + "step": 13137 + }, + { + "epoch": 1.6712886401221219, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.8032232522964478, + "num_tokens": 501069810.0, + "step": 13138 + }, + { + "epoch": 1.6714158504007124, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7973827123641968, + "num_tokens": 501110597.0, + "step": 13139 + }, + { + "epoch": 1.671543060679303, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7852219939231873, + "num_tokens": 501145927.0, + "step": 13140 + }, + { + "epoch": 1.6716702709578934, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7794976234436035, + "num_tokens": 501188473.0, + "step": 13141 + }, + { + "epoch": 1.6717974812364838, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8136966228485107, + "num_tokens": 501226143.0, + "step": 13142 + }, + { + "epoch": 1.6719246915150743, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7926027774810791, + "num_tokens": 501266980.0, + "step": 13143 + }, + { + "epoch": 1.6720519017936648, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7978326082229614, + "num_tokens": 501305414.0, + "step": 13144 + }, + { + "epoch": 1.6721791120722553, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7931417226791382, + "num_tokens": 501343009.0, + "step": 13145 + }, + { + "epoch": 1.6723063223508459, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8029918670654297, + "num_tokens": 501383571.0, + "step": 13146 + }, + { + "epoch": 1.6724335326294364, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8077264428138733, + "num_tokens": 501425444.0, + "step": 13147 + }, + { + "epoch": 1.672560742908027, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8132323622703552, + "num_tokens": 501466536.0, + "step": 13148 + }, + { + "epoch": 1.6726879531866174, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8012720942497253, + "num_tokens": 501507346.0, + "step": 13149 + }, + { + "epoch": 1.672815163465208, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8066698908805847, + "num_tokens": 501545515.0, + "step": 13150 + }, + { + "epoch": 1.6729423737437985, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7949167490005493, + "num_tokens": 501588793.0, + "step": 13151 + }, + { + "epoch": 1.673069584022389, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7972908020019531, + "num_tokens": 501624384.0, + "step": 13152 + }, + { + "epoch": 1.6731967943009796, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.8000143766403198, + "num_tokens": 501663150.0, + "step": 13153 + }, + { + "epoch": 1.67332400457957, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7915184497833252, + "num_tokens": 501706177.0, + "step": 13154 + }, + { + "epoch": 1.6734512148581606, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8069483041763306, + "num_tokens": 501745267.0, + "step": 13155 + }, + { + "epoch": 1.6735784251367511, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7932143211364746, + "num_tokens": 501783591.0, + "step": 13156 + }, + { + "epoch": 1.6737056354153417, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7895464897155762, + "num_tokens": 501825190.0, + "step": 13157 + }, + { + "epoch": 1.6738328456939322, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7971311211585999, + "num_tokens": 501855813.0, + "step": 13158 + }, + { + "epoch": 1.6739600559725227, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7947613000869751, + "num_tokens": 501890106.0, + "step": 13159 + }, + { + "epoch": 1.674087266251113, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7870981097221375, + "num_tokens": 501935437.0, + "step": 13160 + }, + { + "epoch": 1.6742144765297036, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7918955087661743, + "num_tokens": 501980134.0, + "step": 13161 + }, + { + "epoch": 1.674341686808294, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7912675142288208, + "num_tokens": 502015977.0, + "step": 13162 + }, + { + "epoch": 1.6744688970868846, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8057124614715576, + "num_tokens": 502061717.0, + "step": 13163 + }, + { + "epoch": 1.6745961073654752, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7961170673370361, + "num_tokens": 502101594.0, + "step": 13164 + }, + { + "epoch": 1.6747233176440657, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7987640500068665, + "num_tokens": 502136521.0, + "step": 13165 + }, + { + "epoch": 1.674850527922656, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7794941663742065, + "num_tokens": 502174821.0, + "step": 13166 + }, + { + "epoch": 1.6749777382012465, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7912843227386475, + "num_tokens": 502210615.0, + "step": 13167 + }, + { + "epoch": 1.675104948479837, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8022405505180359, + "num_tokens": 502248220.0, + "step": 13168 + }, + { + "epoch": 1.6752321587584276, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7999625205993652, + "num_tokens": 502290686.0, + "step": 13169 + }, + { + "epoch": 1.675359369037018, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7880524396896362, + "num_tokens": 502331789.0, + "step": 13170 + }, + { + "epoch": 1.6754865793156086, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8039425611495972, + "num_tokens": 502371490.0, + "step": 13171 + }, + { + "epoch": 1.6756137895941992, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7974275946617126, + "num_tokens": 502415606.0, + "step": 13172 + }, + { + "epoch": 1.6757409998727897, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7996972799301147, + "num_tokens": 502456203.0, + "step": 13173 + }, + { + "epoch": 1.6758682101513802, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8144797086715698, + "num_tokens": 502491508.0, + "step": 13174 + }, + { + "epoch": 1.6759954204299707, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7848564386367798, + "num_tokens": 502533408.0, + "step": 13175 + }, + { + "epoch": 1.6761226307085613, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.788619875907898, + "num_tokens": 502569824.0, + "step": 13176 + }, + { + "epoch": 1.6762498409871518, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8095990419387817, + "num_tokens": 502604776.0, + "step": 13177 + }, + { + "epoch": 1.6763770512657423, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7804330587387085, + "num_tokens": 502645732.0, + "step": 13178 + }, + { + "epoch": 1.6765042615443329, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8013427257537842, + "num_tokens": 502686044.0, + "step": 13179 + }, + { + "epoch": 1.6766314718229234, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7877570986747742, + "num_tokens": 502728288.0, + "step": 13180 + }, + { + "epoch": 1.676758682101514, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7870272994041443, + "num_tokens": 502760772.0, + "step": 13181 + }, + { + "epoch": 1.6768858923801044, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8000551462173462, + "num_tokens": 502791585.0, + "step": 13182 + }, + { + "epoch": 1.677013102658695, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8148386478424072, + "num_tokens": 502826339.0, + "step": 13183 + }, + { + "epoch": 1.6771403129372855, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8143614530563354, + "num_tokens": 502868716.0, + "step": 13184 + }, + { + "epoch": 1.6772675232158758, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7955408692359924, + "num_tokens": 502907992.0, + "step": 13185 + }, + { + "epoch": 1.6773947334944663, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.791760265827179, + "num_tokens": 502947197.0, + "step": 13186 + }, + { + "epoch": 1.6775219437730569, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8069932460784912, + "num_tokens": 502985890.0, + "step": 13187 + }, + { + "epoch": 1.6776491540516474, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7933368682861328, + "num_tokens": 503033476.0, + "step": 13188 + }, + { + "epoch": 1.677776364330238, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8086044788360596, + "num_tokens": 503068002.0, + "step": 13189 + }, + { + "epoch": 1.6779035746088284, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7948157787322998, + "num_tokens": 503110059.0, + "step": 13190 + }, + { + "epoch": 1.6780307848874187, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7955377101898193, + "num_tokens": 503146617.0, + "step": 13191 + }, + { + "epoch": 1.6781579951660093, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7970174551010132, + "num_tokens": 503186436.0, + "step": 13192 + }, + { + "epoch": 1.6782852054445998, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8037519454956055, + "num_tokens": 503227983.0, + "step": 13193 + }, + { + "epoch": 1.6784124157231903, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7842071056365967, + "num_tokens": 503269301.0, + "step": 13194 + }, + { + "epoch": 1.6785396260017809, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7792817950248718, + "num_tokens": 503308919.0, + "step": 13195 + }, + { + "epoch": 1.6786668362803714, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8109090924263, + "num_tokens": 503345030.0, + "step": 13196 + }, + { + "epoch": 1.678794046558962, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7842170000076294, + "num_tokens": 503379142.0, + "step": 13197 + }, + { + "epoch": 1.6789212568375524, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7944306135177612, + "num_tokens": 503411810.0, + "step": 13198 + }, + { + "epoch": 1.679048467116143, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.796584963798523, + "num_tokens": 503450266.0, + "step": 13199 + }, + { + "epoch": 1.6791756773947335, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8066061735153198, + "num_tokens": 503491015.0, + "step": 13200 + }, + { + "epoch": 1.679302887673324, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8044637441635132, + "num_tokens": 503531964.0, + "step": 13201 + }, + { + "epoch": 1.6794300979519146, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7998288869857788, + "num_tokens": 503576069.0, + "step": 13202 + }, + { + "epoch": 1.679557308230505, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7974897027015686, + "num_tokens": 503607895.0, + "step": 13203 + }, + { + "epoch": 1.6796845185090956, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7819287180900574, + "num_tokens": 503636891.0, + "step": 13204 + }, + { + "epoch": 1.6798117287876861, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.8214256763458252, + "num_tokens": 503682037.0, + "step": 13205 + }, + { + "epoch": 1.6799389390662767, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.803747832775116, + "num_tokens": 503724008.0, + "step": 13206 + }, + { + "epoch": 1.6800661493448672, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8024773597717285, + "num_tokens": 503759008.0, + "step": 13207 + }, + { + "epoch": 1.6801933596234577, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.794292688369751, + "num_tokens": 503796599.0, + "step": 13208 + }, + { + "epoch": 1.680320569902048, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7993590235710144, + "num_tokens": 503835539.0, + "step": 13209 + }, + { + "epoch": 1.6804477801806386, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7954223155975342, + "num_tokens": 503875933.0, + "step": 13210 + }, + { + "epoch": 1.680574990459229, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7979952692985535, + "num_tokens": 503913740.0, + "step": 13211 + }, + { + "epoch": 1.6807022007378196, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8031762838363647, + "num_tokens": 503955105.0, + "step": 13212 + }, + { + "epoch": 1.6808294110164101, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7940163016319275, + "num_tokens": 503987641.0, + "step": 13213 + }, + { + "epoch": 1.6809566212950007, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7908347249031067, + "num_tokens": 504028699.0, + "step": 13214 + }, + { + "epoch": 1.681083831573591, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7912764549255371, + "num_tokens": 504065597.0, + "step": 13215 + }, + { + "epoch": 1.6812110418521815, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7852393388748169, + "num_tokens": 504105968.0, + "step": 13216 + }, + { + "epoch": 1.681338252130772, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8057752847671509, + "num_tokens": 504143743.0, + "step": 13217 + }, + { + "epoch": 1.6814654624093626, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7986481785774231, + "num_tokens": 504177023.0, + "step": 13218 + }, + { + "epoch": 1.681592672687953, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7942459583282471, + "num_tokens": 504210492.0, + "step": 13219 + }, + { + "epoch": 1.6817198829665436, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7995814681053162, + "num_tokens": 504248322.0, + "step": 13220 + }, + { + "epoch": 1.6818470932451342, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7844400405883789, + "num_tokens": 504285915.0, + "step": 13221 + }, + { + "epoch": 1.6819743035237247, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7949185371398926, + "num_tokens": 504325578.0, + "step": 13222 + }, + { + "epoch": 1.6821015138023152, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8165173530578613, + "num_tokens": 504362876.0, + "step": 13223 + }, + { + "epoch": 1.6822287240809057, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7960312962532043, + "num_tokens": 504402303.0, + "step": 13224 + }, + { + "epoch": 1.6823559343594963, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7976058721542358, + "num_tokens": 504440039.0, + "step": 13225 + }, + { + "epoch": 1.6824831446380868, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7905961275100708, + "num_tokens": 504485102.0, + "step": 13226 + }, + { + "epoch": 1.6826103549166773, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7983405590057373, + "num_tokens": 504522762.0, + "step": 13227 + }, + { + "epoch": 1.6827375651952678, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8034979104995728, + "num_tokens": 504562818.0, + "step": 13228 + }, + { + "epoch": 1.6828647754738584, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.784940242767334, + "num_tokens": 504602914.0, + "step": 13229 + }, + { + "epoch": 1.682991985752449, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8088465332984924, + "num_tokens": 504641376.0, + "step": 13230 + }, + { + "epoch": 1.6831191960310394, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8032113909721375, + "num_tokens": 504675031.0, + "step": 13231 + }, + { + "epoch": 1.68324640630963, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8106582164764404, + "num_tokens": 504709624.0, + "step": 13232 + }, + { + "epoch": 1.6833736165882205, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8047096729278564, + "num_tokens": 504749310.0, + "step": 13233 + }, + { + "epoch": 1.6835008268668108, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.8024571537971497, + "num_tokens": 504791700.0, + "step": 13234 + }, + { + "epoch": 1.6836280371454013, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7873164415359497, + "num_tokens": 504833246.0, + "step": 13235 + }, + { + "epoch": 1.6837552474239919, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8049834966659546, + "num_tokens": 504872901.0, + "step": 13236 + }, + { + "epoch": 1.6838824577025824, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7881575226783752, + "num_tokens": 504906138.0, + "step": 13237 + }, + { + "epoch": 1.684009667981173, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7751864790916443, + "num_tokens": 504947147.0, + "step": 13238 + }, + { + "epoch": 1.6841368782597634, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8020727634429932, + "num_tokens": 504985332.0, + "step": 13239 + }, + { + "epoch": 1.6842640885383537, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7916480302810669, + "num_tokens": 505027995.0, + "step": 13240 + }, + { + "epoch": 1.6843912988169443, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8011745810508728, + "num_tokens": 505065311.0, + "step": 13241 + }, + { + "epoch": 1.6845185090955348, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7898407578468323, + "num_tokens": 505105810.0, + "step": 13242 + }, + { + "epoch": 1.6846457193741253, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8098767399787903, + "num_tokens": 505141003.0, + "step": 13243 + }, + { + "epoch": 1.6847729296527159, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7904742956161499, + "num_tokens": 505188025.0, + "step": 13244 + }, + { + "epoch": 1.6849001399313064, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8108313679695129, + "num_tokens": 505226101.0, + "step": 13245 + }, + { + "epoch": 1.685027350209897, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7874513268470764, + "num_tokens": 505264880.0, + "step": 13246 + }, + { + "epoch": 1.6851545604884874, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7959017157554626, + "num_tokens": 505306208.0, + "step": 13247 + }, + { + "epoch": 1.685281770767078, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.592, + "mean_token_accuracy": 0.8254542350769043, + "num_tokens": 505346024.0, + "step": 13248 + }, + { + "epoch": 1.6854089810456685, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8238519430160522, + "num_tokens": 505383644.0, + "step": 13249 + }, + { + "epoch": 1.685536191324259, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8020367622375488, + "num_tokens": 505420365.0, + "step": 13250 + }, + { + "epoch": 1.6856634016028496, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.797784149646759, + "num_tokens": 505458211.0, + "step": 13251 + }, + { + "epoch": 1.68579061188144, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7939351797103882, + "num_tokens": 505501649.0, + "step": 13252 + }, + { + "epoch": 1.6859178221600306, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8028268218040466, + "num_tokens": 505538763.0, + "step": 13253 + }, + { + "epoch": 1.6860450324386211, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.8265857696533203, + "num_tokens": 505574562.0, + "step": 13254 + }, + { + "epoch": 1.6861722427172117, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.8155134916305542, + "num_tokens": 505609718.0, + "step": 13255 + }, + { + "epoch": 1.6862994529958022, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8135181665420532, + "num_tokens": 505644337.0, + "step": 13256 + }, + { + "epoch": 1.6864266632743927, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8059984445571899, + "num_tokens": 505681929.0, + "step": 13257 + }, + { + "epoch": 1.686553873552983, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7759413719177246, + "num_tokens": 505719373.0, + "step": 13258 + }, + { + "epoch": 1.6866810838315736, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7824841737747192, + "num_tokens": 505757767.0, + "step": 13259 + }, + { + "epoch": 1.686808294110164, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7662240862846375, + "num_tokens": 505794777.0, + "step": 13260 + }, + { + "epoch": 1.6869355043887546, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8100855350494385, + "num_tokens": 505832672.0, + "step": 13261 + }, + { + "epoch": 1.6870627146673451, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7850121259689331, + "num_tokens": 505873142.0, + "step": 13262 + }, + { + "epoch": 1.6871899249459357, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8008655309677124, + "num_tokens": 505911994.0, + "step": 13263 + }, + { + "epoch": 1.687317135224526, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8035430908203125, + "num_tokens": 505948874.0, + "step": 13264 + }, + { + "epoch": 1.6874443455031165, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.805253267288208, + "num_tokens": 505989679.0, + "step": 13265 + }, + { + "epoch": 1.687571555781707, + "grad_norm": 10.375, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8097482919692993, + "num_tokens": 506033209.0, + "step": 13266 + }, + { + "epoch": 1.6876987660602976, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8039782643318176, + "num_tokens": 506065379.0, + "step": 13267 + }, + { + "epoch": 1.687825976338888, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7894054651260376, + "num_tokens": 506099484.0, + "step": 13268 + }, + { + "epoch": 1.6879531866174786, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7852082848548889, + "num_tokens": 506136559.0, + "step": 13269 + }, + { + "epoch": 1.6880803968960691, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7960548996925354, + "num_tokens": 506172997.0, + "step": 13270 + }, + { + "epoch": 1.6882076071746597, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8151448965072632, + "num_tokens": 506209858.0, + "step": 13271 + }, + { + "epoch": 1.6883348174532502, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8124873638153076, + "num_tokens": 506242260.0, + "step": 13272 + }, + { + "epoch": 1.6884620277318407, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7858380675315857, + "num_tokens": 506276856.0, + "step": 13273 + }, + { + "epoch": 1.6885892380104313, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.8202890157699585, + "num_tokens": 506311469.0, + "step": 13274 + }, + { + "epoch": 1.6887164482890218, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8072695136070251, + "num_tokens": 506348240.0, + "step": 13275 + }, + { + "epoch": 1.6888436585676123, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8019356727600098, + "num_tokens": 506387203.0, + "step": 13276 + }, + { + "epoch": 1.6889708688462028, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7728105783462524, + "num_tokens": 506421869.0, + "step": 13277 + }, + { + "epoch": 1.6890980791247934, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7983267903327942, + "num_tokens": 506458873.0, + "step": 13278 + }, + { + "epoch": 1.689225289403384, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7868199944496155, + "num_tokens": 506495947.0, + "step": 13279 + }, + { + "epoch": 1.6893524996819744, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7879824638366699, + "num_tokens": 506533098.0, + "step": 13280 + }, + { + "epoch": 1.689479709960565, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8084419369697571, + "num_tokens": 506571346.0, + "step": 13281 + }, + { + "epoch": 1.6896069202391555, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8047415018081665, + "num_tokens": 506607912.0, + "step": 13282 + }, + { + "epoch": 1.6897341305177458, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7918134331703186, + "num_tokens": 506654258.0, + "step": 13283 + }, + { + "epoch": 1.6898613407963363, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.807933509349823, + "num_tokens": 506693436.0, + "step": 13284 + }, + { + "epoch": 1.6899885510749268, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8072993159294128, + "num_tokens": 506737383.0, + "step": 13285 + }, + { + "epoch": 1.6901157613535174, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8066204190254211, + "num_tokens": 506772282.0, + "step": 13286 + }, + { + "epoch": 1.690242971632108, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8082458972930908, + "num_tokens": 506807208.0, + "step": 13287 + }, + { + "epoch": 1.6903701819106984, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8080467581748962, + "num_tokens": 506847952.0, + "step": 13288 + }, + { + "epoch": 1.6904973921892887, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8023888468742371, + "num_tokens": 506888265.0, + "step": 13289 + }, + { + "epoch": 1.6906246024678793, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7987191677093506, + "num_tokens": 506926096.0, + "step": 13290 + }, + { + "epoch": 1.6907518127464698, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7890339493751526, + "num_tokens": 506962252.0, + "step": 13291 + }, + { + "epoch": 1.6908790230250603, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.802803635597229, + "num_tokens": 507004709.0, + "step": 13292 + }, + { + "epoch": 1.6910062333036509, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8155098557472229, + "num_tokens": 507041041.0, + "step": 13293 + }, + { + "epoch": 1.6911334435822414, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7935655117034912, + "num_tokens": 507080251.0, + "step": 13294 + }, + { + "epoch": 1.691260653860832, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.802280068397522, + "num_tokens": 507120813.0, + "step": 13295 + }, + { + "epoch": 1.6913878641394224, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7849611639976501, + "num_tokens": 507160284.0, + "step": 13296 + }, + { + "epoch": 1.691515074418013, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7919878959655762, + "num_tokens": 507205069.0, + "step": 13297 + }, + { + "epoch": 1.6916422846966035, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8108583688735962, + "num_tokens": 507244689.0, + "step": 13298 + }, + { + "epoch": 1.691769494975194, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7999738454818726, + "num_tokens": 507283917.0, + "step": 13299 + }, + { + "epoch": 1.6918967052537845, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8035682439804077, + "num_tokens": 507327937.0, + "step": 13300 + }, + { + "epoch": 1.692023915532375, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8152326941490173, + "num_tokens": 507369521.0, + "step": 13301 + }, + { + "epoch": 1.6921511258109656, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8132078647613525, + "num_tokens": 507408157.0, + "step": 13302 + }, + { + "epoch": 1.6922783360895561, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8066972494125366, + "num_tokens": 507448003.0, + "step": 13303 + }, + { + "epoch": 1.6924055463681467, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8086053133010864, + "num_tokens": 507489970.0, + "step": 13304 + }, + { + "epoch": 1.6925327566467372, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8028966188430786, + "num_tokens": 507524536.0, + "step": 13305 + }, + { + "epoch": 1.6926599669253277, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8270003795623779, + "num_tokens": 507561575.0, + "step": 13306 + }, + { + "epoch": 1.692787177203918, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7948838472366333, + "num_tokens": 507597212.0, + "step": 13307 + }, + { + "epoch": 1.6929143874825086, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8040357828140259, + "num_tokens": 507634251.0, + "step": 13308 + }, + { + "epoch": 1.693041597761099, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8034692406654358, + "num_tokens": 507671345.0, + "step": 13309 + }, + { + "epoch": 1.6931688080396896, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7865865230560303, + "num_tokens": 507709797.0, + "step": 13310 + }, + { + "epoch": 1.6932960183182801, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8037315607070923, + "num_tokens": 507747047.0, + "step": 13311 + }, + { + "epoch": 1.6934232285968707, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8001890182495117, + "num_tokens": 507782187.0, + "step": 13312 + }, + { + "epoch": 1.693550438875461, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.78510582447052, + "num_tokens": 507826714.0, + "step": 13313 + }, + { + "epoch": 1.6936776491540515, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7909889221191406, + "num_tokens": 507869423.0, + "step": 13314 + }, + { + "epoch": 1.693804859432642, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8069358468055725, + "num_tokens": 507904711.0, + "step": 13315 + }, + { + "epoch": 1.6939320697112326, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7965139150619507, + "num_tokens": 507947760.0, + "step": 13316 + }, + { + "epoch": 1.694059279989823, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7959282994270325, + "num_tokens": 507984249.0, + "step": 13317 + }, + { + "epoch": 1.6941864902684136, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7992165088653564, + "num_tokens": 508017206.0, + "step": 13318 + }, + { + "epoch": 1.6943137005470041, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7963832020759583, + "num_tokens": 508056357.0, + "step": 13319 + }, + { + "epoch": 1.6944409108255947, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7958114147186279, + "num_tokens": 508094870.0, + "step": 13320 + }, + { + "epoch": 1.6945681211041852, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7927001714706421, + "num_tokens": 508120323.0, + "step": 13321 + }, + { + "epoch": 1.6946953313827757, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7900048494338989, + "num_tokens": 508160018.0, + "step": 13322 + }, + { + "epoch": 1.6948225416613663, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.8257468342781067, + "num_tokens": 508200947.0, + "step": 13323 + }, + { + "epoch": 1.6949497519399568, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7888915538787842, + "num_tokens": 508246457.0, + "step": 13324 + }, + { + "epoch": 1.6950769622185473, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7902253866195679, + "num_tokens": 508281951.0, + "step": 13325 + }, + { + "epoch": 1.6952041724971378, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.787265419960022, + "num_tokens": 508322571.0, + "step": 13326 + }, + { + "epoch": 1.6953313827757284, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8089796304702759, + "num_tokens": 508357347.0, + "step": 13327 + }, + { + "epoch": 1.695458593054319, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8041343688964844, + "num_tokens": 508401813.0, + "step": 13328 + }, + { + "epoch": 1.6955858033329094, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7886313796043396, + "num_tokens": 508440457.0, + "step": 13329 + }, + { + "epoch": 1.6957130136115, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7855915427207947, + "num_tokens": 508472272.0, + "step": 13330 + }, + { + "epoch": 1.6958402238900905, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8134037852287292, + "num_tokens": 508509352.0, + "step": 13331 + }, + { + "epoch": 1.6959674341686808, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8078842759132385, + "num_tokens": 508552515.0, + "step": 13332 + }, + { + "epoch": 1.6960946444472713, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8131932020187378, + "num_tokens": 508591399.0, + "step": 13333 + }, + { + "epoch": 1.6962218547258618, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8065944910049438, + "num_tokens": 508630427.0, + "step": 13334 + }, + { + "epoch": 1.6963490650044524, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8098728656768799, + "num_tokens": 508667469.0, + "step": 13335 + }, + { + "epoch": 1.696476275283043, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8030115365982056, + "num_tokens": 508706308.0, + "step": 13336 + }, + { + "epoch": 1.6966034855616334, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.783641517162323, + "num_tokens": 508738968.0, + "step": 13337 + }, + { + "epoch": 1.6967306958402237, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8055651783943176, + "num_tokens": 508769879.0, + "step": 13338 + }, + { + "epoch": 1.6968579061188143, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7859185934066772, + "num_tokens": 508812277.0, + "step": 13339 + }, + { + "epoch": 1.6969851163974048, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7847568392753601, + "num_tokens": 508849293.0, + "step": 13340 + }, + { + "epoch": 1.6971123266759953, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8006291389465332, + "num_tokens": 508885798.0, + "step": 13341 + }, + { + "epoch": 1.6972395369545858, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7935183048248291, + "num_tokens": 508920491.0, + "step": 13342 + }, + { + "epoch": 1.6973667472331764, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7963435053825378, + "num_tokens": 508956161.0, + "step": 13343 + }, + { + "epoch": 1.697493957511767, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7977052927017212, + "num_tokens": 508995696.0, + "step": 13344 + }, + { + "epoch": 1.6976211677903574, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8071194887161255, + "num_tokens": 509029505.0, + "step": 13345 + }, + { + "epoch": 1.697748378068948, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8082245588302612, + "num_tokens": 509062815.0, + "step": 13346 + }, + { + "epoch": 1.6978755883475385, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8047871589660645, + "num_tokens": 509103376.0, + "step": 13347 + }, + { + "epoch": 1.698002798626129, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8113319873809814, + "num_tokens": 509142797.0, + "step": 13348 + }, + { + "epoch": 1.6981300089047195, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7902876138687134, + "num_tokens": 509182724.0, + "step": 13349 + }, + { + "epoch": 1.69825721918331, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8080335855484009, + "num_tokens": 509217856.0, + "step": 13350 + }, + { + "epoch": 1.6983844294619006, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7985386252403259, + "num_tokens": 509251581.0, + "step": 13351 + }, + { + "epoch": 1.6985116397404911, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.8161885738372803, + "num_tokens": 509286687.0, + "step": 13352 + }, + { + "epoch": 1.6986388500190817, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7924381494522095, + "num_tokens": 509323372.0, + "step": 13353 + }, + { + "epoch": 1.6987660602976722, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7951157093048096, + "num_tokens": 509362472.0, + "step": 13354 + }, + { + "epoch": 1.6988932705762627, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7982316017150879, + "num_tokens": 509397676.0, + "step": 13355 + }, + { + "epoch": 1.699020480854853, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8031870126724243, + "num_tokens": 509441100.0, + "step": 13356 + }, + { + "epoch": 1.6991476911334435, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7947683930397034, + "num_tokens": 509485430.0, + "step": 13357 + }, + { + "epoch": 1.699274901412034, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7688096761703491, + "num_tokens": 509520102.0, + "step": 13358 + }, + { + "epoch": 1.6994021116906246, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7977670431137085, + "num_tokens": 509556525.0, + "step": 13359 + }, + { + "epoch": 1.6995293219692151, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7929428219795227, + "num_tokens": 509596899.0, + "step": 13360 + }, + { + "epoch": 1.6996565322478057, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8044935464859009, + "num_tokens": 509632125.0, + "step": 13361 + }, + { + "epoch": 1.699783742526396, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7916132807731628, + "num_tokens": 509668131.0, + "step": 13362 + }, + { + "epoch": 1.6999109528049865, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8138477206230164, + "num_tokens": 509704078.0, + "step": 13363 + }, + { + "epoch": 1.700038163083577, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.8151911497116089, + "num_tokens": 509739464.0, + "step": 13364 + }, + { + "epoch": 1.7001653733621676, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8021906614303589, + "num_tokens": 509777970.0, + "step": 13365 + }, + { + "epoch": 1.700292583640758, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7895123958587646, + "num_tokens": 509817010.0, + "step": 13366 + }, + { + "epoch": 1.7004197939193486, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8012754917144775, + "num_tokens": 509854192.0, + "step": 13367 + }, + { + "epoch": 1.7005470041979391, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.811994194984436, + "num_tokens": 509888633.0, + "step": 13368 + }, + { + "epoch": 1.7006742144765297, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8048779368400574, + "num_tokens": 509928474.0, + "step": 13369 + }, + { + "epoch": 1.7008014247551202, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7904677391052246, + "num_tokens": 509968875.0, + "step": 13370 + }, + { + "epoch": 1.7009286350337107, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7998346090316772, + "num_tokens": 510006021.0, + "step": 13371 + }, + { + "epoch": 1.7010558453123013, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7999757528305054, + "num_tokens": 510042111.0, + "step": 13372 + }, + { + "epoch": 1.7011830555908918, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8032519817352295, + "num_tokens": 510074894.0, + "step": 13373 + }, + { + "epoch": 1.7013102658694823, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7879053354263306, + "num_tokens": 510111516.0, + "step": 13374 + }, + { + "epoch": 1.7014374761480728, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.804141104221344, + "num_tokens": 510151549.0, + "step": 13375 + }, + { + "epoch": 1.7015646864266634, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7967233061790466, + "num_tokens": 510189215.0, + "step": 13376 + }, + { + "epoch": 1.701691896705254, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8210318088531494, + "num_tokens": 510231189.0, + "step": 13377 + }, + { + "epoch": 1.7018191069838444, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8046901226043701, + "num_tokens": 510273956.0, + "step": 13378 + }, + { + "epoch": 1.701946317262435, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8186017870903015, + "num_tokens": 510315095.0, + "step": 13379 + }, + { + "epoch": 1.7020735275410255, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7729755640029907, + "num_tokens": 510350753.0, + "step": 13380 + }, + { + "epoch": 1.7022007378196158, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8049286007881165, + "num_tokens": 510389257.0, + "step": 13381 + }, + { + "epoch": 1.7023279480982063, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.800683319568634, + "num_tokens": 510428879.0, + "step": 13382 + }, + { + "epoch": 1.7024551583767968, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8022792339324951, + "num_tokens": 510468409.0, + "step": 13383 + }, + { + "epoch": 1.7025823686553874, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8092528581619263, + "num_tokens": 510503988.0, + "step": 13384 + }, + { + "epoch": 1.702709578933978, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.794927179813385, + "num_tokens": 510539009.0, + "step": 13385 + }, + { + "epoch": 1.7028367892125684, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8142610788345337, + "num_tokens": 510576177.0, + "step": 13386 + }, + { + "epoch": 1.7029639994911587, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8066169023513794, + "num_tokens": 510622194.0, + "step": 13387 + }, + { + "epoch": 1.7030912097697493, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.8219468593597412, + "num_tokens": 510663564.0, + "step": 13388 + }, + { + "epoch": 1.7032184200483398, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8085870742797852, + "num_tokens": 510702153.0, + "step": 13389 + }, + { + "epoch": 1.7033456303269303, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8015430569648743, + "num_tokens": 510742812.0, + "step": 13390 + }, + { + "epoch": 1.7034728406055208, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7894715070724487, + "num_tokens": 510783363.0, + "step": 13391 + }, + { + "epoch": 1.7036000508841114, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7861059904098511, + "num_tokens": 510818362.0, + "step": 13392 + }, + { + "epoch": 1.703727261162702, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8090344667434692, + "num_tokens": 510854732.0, + "step": 13393 + }, + { + "epoch": 1.7038544714412924, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7940694689750671, + "num_tokens": 510895107.0, + "step": 13394 + }, + { + "epoch": 1.703981681719883, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.797290027141571, + "num_tokens": 510930737.0, + "step": 13395 + }, + { + "epoch": 1.7041088919984735, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.812886655330658, + "num_tokens": 510964646.0, + "step": 13396 + }, + { + "epoch": 1.704236102277064, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8038012981414795, + "num_tokens": 511000583.0, + "step": 13397 + }, + { + "epoch": 1.7043633125556545, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8005143404006958, + "num_tokens": 511033729.0, + "step": 13398 + }, + { + "epoch": 1.704490522834245, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8016095161437988, + "num_tokens": 511074099.0, + "step": 13399 + }, + { + "epoch": 1.7046177331128356, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.789492130279541, + "num_tokens": 511112473.0, + "step": 13400 + }, + { + "epoch": 1.7047449433914261, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8007637858390808, + "num_tokens": 511149839.0, + "step": 13401 + }, + { + "epoch": 1.7048721536700167, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.790311336517334, + "num_tokens": 511183054.0, + "step": 13402 + }, + { + "epoch": 1.7049993639486072, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8118959665298462, + "num_tokens": 511218498.0, + "step": 13403 + }, + { + "epoch": 1.7051265742271977, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.8017799854278564, + "num_tokens": 511259776.0, + "step": 13404 + }, + { + "epoch": 1.705253784505788, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.809343695640564, + "num_tokens": 511297331.0, + "step": 13405 + }, + { + "epoch": 1.7053809947843785, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8046311736106873, + "num_tokens": 511331055.0, + "step": 13406 + }, + { + "epoch": 1.705508205062969, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8171014785766602, + "num_tokens": 511368651.0, + "step": 13407 + }, + { + "epoch": 1.7056354153415596, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8050534725189209, + "num_tokens": 511402163.0, + "step": 13408 + }, + { + "epoch": 1.7057626256201501, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7884277105331421, + "num_tokens": 511443769.0, + "step": 13409 + }, + { + "epoch": 1.7058898358987407, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7984063625335693, + "num_tokens": 511481038.0, + "step": 13410 + }, + { + "epoch": 1.706017046177331, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8043840527534485, + "num_tokens": 511522940.0, + "step": 13411 + }, + { + "epoch": 1.7061442564559215, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.799860954284668, + "num_tokens": 511562630.0, + "step": 13412 + }, + { + "epoch": 1.706271466734512, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7808986902236938, + "num_tokens": 511598986.0, + "step": 13413 + }, + { + "epoch": 1.7063986770131025, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7996249794960022, + "num_tokens": 511633582.0, + "step": 13414 + }, + { + "epoch": 1.706525887291693, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8026944994926453, + "num_tokens": 511673296.0, + "step": 13415 + }, + { + "epoch": 1.7066530975702836, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7982813119888306, + "num_tokens": 511707890.0, + "step": 13416 + }, + { + "epoch": 1.7067803078488741, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8090115785598755, + "num_tokens": 511741597.0, + "step": 13417 + }, + { + "epoch": 1.7069075181274647, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.809316873550415, + "num_tokens": 511776816.0, + "step": 13418 + }, + { + "epoch": 1.7070347284060552, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7953598499298096, + "num_tokens": 511814866.0, + "step": 13419 + }, + { + "epoch": 1.7071619386846457, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8026683926582336, + "num_tokens": 511852458.0, + "step": 13420 + }, + { + "epoch": 1.7072891489632362, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.8006542325019836, + "num_tokens": 511891201.0, + "step": 13421 + }, + { + "epoch": 1.7074163592418268, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.795256495475769, + "num_tokens": 511930862.0, + "step": 13422 + }, + { + "epoch": 1.7075435695204173, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.809038519859314, + "num_tokens": 511967695.0, + "step": 13423 + }, + { + "epoch": 1.7076707797990078, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8146481513977051, + "num_tokens": 512005976.0, + "step": 13424 + }, + { + "epoch": 1.7077979900775984, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8007961511611938, + "num_tokens": 512041575.0, + "step": 13425 + }, + { + "epoch": 1.7079252003561889, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7936475276947021, + "num_tokens": 512076171.0, + "step": 13426 + }, + { + "epoch": 1.7080524106347794, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7946861386299133, + "num_tokens": 512112082.0, + "step": 13427 + }, + { + "epoch": 1.70817962091337, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7871536016464233, + "num_tokens": 512149126.0, + "step": 13428 + }, + { + "epoch": 1.7083068311919605, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8051079511642456, + "num_tokens": 512182368.0, + "step": 13429 + }, + { + "epoch": 1.7084340414705508, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7820214629173279, + "num_tokens": 512218224.0, + "step": 13430 + }, + { + "epoch": 1.7085612517491413, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.800554633140564, + "num_tokens": 512255523.0, + "step": 13431 + }, + { + "epoch": 1.7086884620277318, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8078243732452393, + "num_tokens": 512290514.0, + "step": 13432 + }, + { + "epoch": 1.7088156723063224, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8135713338851929, + "num_tokens": 512328468.0, + "step": 13433 + }, + { + "epoch": 1.708942882584913, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7902966737747192, + "num_tokens": 512359786.0, + "step": 13434 + }, + { + "epoch": 1.7090700928635034, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7748305797576904, + "num_tokens": 512395679.0, + "step": 13435 + }, + { + "epoch": 1.7091973031420937, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.796674370765686, + "num_tokens": 512433491.0, + "step": 13436 + }, + { + "epoch": 1.7093245134206843, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7927579879760742, + "num_tokens": 512470215.0, + "step": 13437 + }, + { + "epoch": 1.7094517236992748, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7999237179756165, + "num_tokens": 512507055.0, + "step": 13438 + }, + { + "epoch": 1.7095789339778653, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.779050350189209, + "num_tokens": 512547877.0, + "step": 13439 + }, + { + "epoch": 1.7097061442564558, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7856363654136658, + "num_tokens": 512586558.0, + "step": 13440 + }, + { + "epoch": 1.7098333545350464, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7982323169708252, + "num_tokens": 512631181.0, + "step": 13441 + }, + { + "epoch": 1.709960564813637, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8105698227882385, + "num_tokens": 512667925.0, + "step": 13442 + }, + { + "epoch": 1.7100877750922274, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7801722288131714, + "num_tokens": 512703136.0, + "step": 13443 + }, + { + "epoch": 1.710214985370818, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8041520714759827, + "num_tokens": 512739001.0, + "step": 13444 + }, + { + "epoch": 1.7103421956494085, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8054810762405396, + "num_tokens": 512780517.0, + "step": 13445 + }, + { + "epoch": 1.710469405927999, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8090322017669678, + "num_tokens": 512815964.0, + "step": 13446 + }, + { + "epoch": 1.7105966162065895, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8033518195152283, + "num_tokens": 512855451.0, + "step": 13447 + }, + { + "epoch": 1.71072382648518, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8102025389671326, + "num_tokens": 512895809.0, + "step": 13448 + }, + { + "epoch": 1.7108510367637706, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8110864162445068, + "num_tokens": 512936424.0, + "step": 13449 + }, + { + "epoch": 1.7109782470423611, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8167821168899536, + "num_tokens": 512976486.0, + "step": 13450 + }, + { + "epoch": 1.7111054573209517, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7870342135429382, + "num_tokens": 513014131.0, + "step": 13451 + }, + { + "epoch": 1.7112326675995422, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7660240530967712, + "num_tokens": 513054706.0, + "step": 13452 + }, + { + "epoch": 1.7113598778781327, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8077441453933716, + "num_tokens": 513094089.0, + "step": 13453 + }, + { + "epoch": 1.711487088156723, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.8019984364509583, + "num_tokens": 513136306.0, + "step": 13454 + }, + { + "epoch": 1.7116142984353135, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7979762554168701, + "num_tokens": 513174376.0, + "step": 13455 + }, + { + "epoch": 1.711741508713904, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8054502606391907, + "num_tokens": 513211834.0, + "step": 13456 + }, + { + "epoch": 1.7118687189924946, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7875814437866211, + "num_tokens": 513241765.0, + "step": 13457 + }, + { + "epoch": 1.7119959292710851, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.78593510389328, + "num_tokens": 513283510.0, + "step": 13458 + }, + { + "epoch": 1.7121231395496757, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7962150573730469, + "num_tokens": 513320746.0, + "step": 13459 + }, + { + "epoch": 1.712250349828266, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7929612398147583, + "num_tokens": 513358036.0, + "step": 13460 + }, + { + "epoch": 1.7123775601068565, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8063104748725891, + "num_tokens": 513395156.0, + "step": 13461 + }, + { + "epoch": 1.712504770385447, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7935018539428711, + "num_tokens": 513438631.0, + "step": 13462 + }, + { + "epoch": 1.7126319806640375, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7887177467346191, + "num_tokens": 513471225.0, + "step": 13463 + }, + { + "epoch": 1.712759190942628, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7885907292366028, + "num_tokens": 513513824.0, + "step": 13464 + }, + { + "epoch": 1.7128864012212186, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7965876460075378, + "num_tokens": 513556263.0, + "step": 13465 + }, + { + "epoch": 1.7130136114998091, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8026962280273438, + "num_tokens": 513595093.0, + "step": 13466 + }, + { + "epoch": 1.7131408217783997, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7940646409988403, + "num_tokens": 513633761.0, + "step": 13467 + }, + { + "epoch": 1.7132680320569902, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7928608059883118, + "num_tokens": 513671404.0, + "step": 13468 + }, + { + "epoch": 1.7133952423355807, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7986197471618652, + "num_tokens": 513713752.0, + "step": 13469 + }, + { + "epoch": 1.7135224526141712, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7971135377883911, + "num_tokens": 513755261.0, + "step": 13470 + }, + { + "epoch": 1.7136496628927618, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8053498268127441, + "num_tokens": 513799042.0, + "step": 13471 + }, + { + "epoch": 1.7137768731713523, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7877082824707031, + "num_tokens": 513838148.0, + "step": 13472 + }, + { + "epoch": 1.7139040834499428, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8046293258666992, + "num_tokens": 513878960.0, + "step": 13473 + }, + { + "epoch": 1.7140312937285334, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7977725267410278, + "num_tokens": 513914083.0, + "step": 13474 + }, + { + "epoch": 1.7141585040071239, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7875330448150635, + "num_tokens": 513951591.0, + "step": 13475 + }, + { + "epoch": 1.7142857142857144, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.782641589641571, + "num_tokens": 513986127.0, + "step": 13476 + }, + { + "epoch": 1.714412924564305, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7950992584228516, + "num_tokens": 514028248.0, + "step": 13477 + }, + { + "epoch": 1.7145401348428955, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.797644853591919, + "num_tokens": 514062757.0, + "step": 13478 + }, + { + "epoch": 1.7146673451214858, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7891796827316284, + "num_tokens": 514102635.0, + "step": 13479 + }, + { + "epoch": 1.7147945554000763, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7947041988372803, + "num_tokens": 514140883.0, + "step": 13480 + }, + { + "epoch": 1.7149217656786668, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8111838698387146, + "num_tokens": 514176536.0, + "step": 13481 + }, + { + "epoch": 1.7150489759572574, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7915017604827881, + "num_tokens": 514209969.0, + "step": 13482 + }, + { + "epoch": 1.7151761862358479, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7985773682594299, + "num_tokens": 514249679.0, + "step": 13483 + }, + { + "epoch": 1.7153033965144384, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7961699962615967, + "num_tokens": 514292719.0, + "step": 13484 + }, + { + "epoch": 1.7154306067930287, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7921029329299927, + "num_tokens": 514323396.0, + "step": 13485 + }, + { + "epoch": 1.7155578170716193, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.807185709476471, + "num_tokens": 514360951.0, + "step": 13486 + }, + { + "epoch": 1.7156850273502098, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7847029566764832, + "num_tokens": 514403879.0, + "step": 13487 + }, + { + "epoch": 1.7158122376288003, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8079371452331543, + "num_tokens": 514446628.0, + "step": 13488 + }, + { + "epoch": 1.7159394479073908, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7998746633529663, + "num_tokens": 514486132.0, + "step": 13489 + }, + { + "epoch": 1.7160666581859814, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7878419160842896, + "num_tokens": 514526384.0, + "step": 13490 + }, + { + "epoch": 1.716193868464572, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7978017926216125, + "num_tokens": 514565828.0, + "step": 13491 + }, + { + "epoch": 1.7163210787431624, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7973001599311829, + "num_tokens": 514605260.0, + "step": 13492 + }, + { + "epoch": 1.716448289021753, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7993066906929016, + "num_tokens": 514643433.0, + "step": 13493 + }, + { + "epoch": 1.7165754993003435, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8024188280105591, + "num_tokens": 514683264.0, + "step": 13494 + }, + { + "epoch": 1.716702709578934, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.807782769203186, + "num_tokens": 514719775.0, + "step": 13495 + }, + { + "epoch": 1.7168299198575245, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7973683476448059, + "num_tokens": 514766287.0, + "step": 13496 + }, + { + "epoch": 1.716957130136115, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8044164180755615, + "num_tokens": 514806725.0, + "step": 13497 + }, + { + "epoch": 1.7170843404147056, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.804790735244751, + "num_tokens": 514844008.0, + "step": 13498 + }, + { + "epoch": 1.7172115506932961, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7885106801986694, + "num_tokens": 514880356.0, + "step": 13499 + }, + { + "epoch": 1.7173387609718866, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.811162531375885, + "num_tokens": 514915309.0, + "step": 13500 + }, + { + "epoch": 1.7174659712504772, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7982290983200073, + "num_tokens": 514956496.0, + "step": 13501 + }, + { + "epoch": 1.7175931815290677, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7995930910110474, + "num_tokens": 514992247.0, + "step": 13502 + }, + { + "epoch": 1.717720391807658, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.795924961566925, + "num_tokens": 515035443.0, + "step": 13503 + }, + { + "epoch": 1.7178476020862485, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8035129308700562, + "num_tokens": 515072279.0, + "step": 13504 + }, + { + "epoch": 1.717974812364839, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7989721298217773, + "num_tokens": 515112547.0, + "step": 13505 + }, + { + "epoch": 1.7181020226434296, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7838426232337952, + "num_tokens": 515146814.0, + "step": 13506 + }, + { + "epoch": 1.7182292329220201, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8036874532699585, + "num_tokens": 515184393.0, + "step": 13507 + }, + { + "epoch": 1.7183564432006107, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7921786308288574, + "num_tokens": 515225312.0, + "step": 13508 + }, + { + "epoch": 1.718483653479201, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7757067084312439, + "num_tokens": 515261330.0, + "step": 13509 + }, + { + "epoch": 1.7186108637577915, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8016382455825806, + "num_tokens": 515299325.0, + "step": 13510 + }, + { + "epoch": 1.718738074036382, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.801060676574707, + "num_tokens": 515335213.0, + "step": 13511 + }, + { + "epoch": 1.7188652843149725, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8078387975692749, + "num_tokens": 515365699.0, + "step": 13512 + }, + { + "epoch": 1.718992494593563, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7989200949668884, + "num_tokens": 515396025.0, + "step": 13513 + }, + { + "epoch": 1.7191197048721536, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7996283173561096, + "num_tokens": 515429724.0, + "step": 13514 + }, + { + "epoch": 1.7192469151507441, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7920688986778259, + "num_tokens": 515465894.0, + "step": 13515 + }, + { + "epoch": 1.7193741254293347, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.8177935481071472, + "num_tokens": 515498894.0, + "step": 13516 + }, + { + "epoch": 1.7195013357079252, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7983807325363159, + "num_tokens": 515538783.0, + "step": 13517 + }, + { + "epoch": 1.7196285459865157, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.798407793045044, + "num_tokens": 515573107.0, + "step": 13518 + }, + { + "epoch": 1.7197557562651062, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8102466464042664, + "num_tokens": 515614901.0, + "step": 13519 + }, + { + "epoch": 1.7198829665436968, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8020510673522949, + "num_tokens": 515654197.0, + "step": 13520 + }, + { + "epoch": 1.7200101768222873, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8202575445175171, + "num_tokens": 515697003.0, + "step": 13521 + }, + { + "epoch": 1.7201373871008778, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7945988178253174, + "num_tokens": 515737801.0, + "step": 13522 + }, + { + "epoch": 1.7202645973794684, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7890781760215759, + "num_tokens": 515774711.0, + "step": 13523 + }, + { + "epoch": 1.7203918076580589, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7858377695083618, + "num_tokens": 515811728.0, + "step": 13524 + }, + { + "epoch": 1.7205190179366494, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8022915124893188, + "num_tokens": 515844425.0, + "step": 13525 + }, + { + "epoch": 1.72064622821524, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7983769178390503, + "num_tokens": 515882439.0, + "step": 13526 + }, + { + "epoch": 1.7207734384938305, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7779749035835266, + "num_tokens": 515922168.0, + "step": 13527 + }, + { + "epoch": 1.7209006487724208, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8053937554359436, + "num_tokens": 515961840.0, + "step": 13528 + }, + { + "epoch": 1.7210278590510113, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7993906736373901, + "num_tokens": 515997586.0, + "step": 13529 + }, + { + "epoch": 1.7211550693296018, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.805993378162384, + "num_tokens": 516036660.0, + "step": 13530 + }, + { + "epoch": 1.7212822796081924, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7880109548568726, + "num_tokens": 516070846.0, + "step": 13531 + }, + { + "epoch": 1.7214094898867829, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8238383531570435, + "num_tokens": 516107027.0, + "step": 13532 + }, + { + "epoch": 1.7215367001653734, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7989606261253357, + "num_tokens": 516143971.0, + "step": 13533 + }, + { + "epoch": 1.7216639104439637, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8003067970275879, + "num_tokens": 516180617.0, + "step": 13534 + }, + { + "epoch": 1.7217911207225542, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8054044246673584, + "num_tokens": 516220371.0, + "step": 13535 + }, + { + "epoch": 1.7219183310011448, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8009281158447266, + "num_tokens": 516258642.0, + "step": 13536 + }, + { + "epoch": 1.7220455412797353, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7930933237075806, + "num_tokens": 516298971.0, + "step": 13537 + }, + { + "epoch": 1.7221727515583258, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8092895746231079, + "num_tokens": 516334457.0, + "step": 13538 + }, + { + "epoch": 1.7222999618369164, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8221949338912964, + "num_tokens": 516374585.0, + "step": 13539 + }, + { + "epoch": 1.7224271721155069, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7843197584152222, + "num_tokens": 516414180.0, + "step": 13540 + }, + { + "epoch": 1.7225543823940974, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8079918026924133, + "num_tokens": 516450189.0, + "step": 13541 + }, + { + "epoch": 1.722681592672688, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.801460862159729, + "num_tokens": 516487549.0, + "step": 13542 + }, + { + "epoch": 1.7228088029512785, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.787018895149231, + "num_tokens": 516526649.0, + "step": 13543 + }, + { + "epoch": 1.722936013229869, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8027776479721069, + "num_tokens": 516563829.0, + "step": 13544 + }, + { + "epoch": 1.7230632235084595, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7808071970939636, + "num_tokens": 516605179.0, + "step": 13545 + }, + { + "epoch": 1.72319043378705, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7977609634399414, + "num_tokens": 516641576.0, + "step": 13546 + }, + { + "epoch": 1.7233176440656406, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7944643497467041, + "num_tokens": 516678687.0, + "step": 13547 + }, + { + "epoch": 1.7234448543442311, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7918667793273926, + "num_tokens": 516711484.0, + "step": 13548 + }, + { + "epoch": 1.7235720646228216, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7948609590530396, + "num_tokens": 516750598.0, + "step": 13549 + }, + { + "epoch": 1.7236992749014122, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8077847361564636, + "num_tokens": 516793438.0, + "step": 13550 + }, + { + "epoch": 1.7238264851800027, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8131716847419739, + "num_tokens": 516828692.0, + "step": 13551 + }, + { + "epoch": 1.723953695458593, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8027083277702332, + "num_tokens": 516867592.0, + "step": 13552 + }, + { + "epoch": 1.7240809057371835, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7948268055915833, + "num_tokens": 516907668.0, + "step": 13553 + }, + { + "epoch": 1.724208116015774, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7968634366989136, + "num_tokens": 516944628.0, + "step": 13554 + }, + { + "epoch": 1.7243353262943646, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7944007515907288, + "num_tokens": 516989027.0, + "step": 13555 + }, + { + "epoch": 1.7244625365729551, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7898737788200378, + "num_tokens": 517024464.0, + "step": 13556 + }, + { + "epoch": 1.7245897468515456, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8048908710479736, + "num_tokens": 517067714.0, + "step": 13557 + }, + { + "epoch": 1.724716957130136, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8012422323226929, + "num_tokens": 517107564.0, + "step": 13558 + }, + { + "epoch": 1.7248441674087265, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8075016736984253, + "num_tokens": 517148240.0, + "step": 13559 + }, + { + "epoch": 1.724971377687317, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7954786419868469, + "num_tokens": 517187841.0, + "step": 13560 + }, + { + "epoch": 1.7250985879659075, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7830284833908081, + "num_tokens": 517225732.0, + "step": 13561 + }, + { + "epoch": 1.725225798244498, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7795321941375732, + "num_tokens": 517260423.0, + "step": 13562 + }, + { + "epoch": 1.7253530085230886, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7875900864601135, + "num_tokens": 517300862.0, + "step": 13563 + }, + { + "epoch": 1.7254802188016791, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8043825030326843, + "num_tokens": 517341010.0, + "step": 13564 + }, + { + "epoch": 1.7256074290802697, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7784024477005005, + "num_tokens": 517376853.0, + "step": 13565 + }, + { + "epoch": 1.7257346393588602, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7933589816093445, + "num_tokens": 517421854.0, + "step": 13566 + }, + { + "epoch": 1.7258618496374507, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7914643883705139, + "num_tokens": 517461711.0, + "step": 13567 + }, + { + "epoch": 1.7259890599160412, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.803846001625061, + "num_tokens": 517494755.0, + "step": 13568 + }, + { + "epoch": 1.7261162701946318, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.794681191444397, + "num_tokens": 517539061.0, + "step": 13569 + }, + { + "epoch": 1.7262434804732223, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7977570295333862, + "num_tokens": 517579768.0, + "step": 13570 + }, + { + "epoch": 1.7263706907518128, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.8090678453445435, + "num_tokens": 517612006.0, + "step": 13571 + }, + { + "epoch": 1.7264979010304033, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7962472438812256, + "num_tokens": 517647547.0, + "step": 13572 + }, + { + "epoch": 1.7266251113089939, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7890410423278809, + "num_tokens": 517689067.0, + "step": 13573 + }, + { + "epoch": 1.7267523215875844, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.803751528263092, + "num_tokens": 517729967.0, + "step": 13574 + }, + { + "epoch": 1.726879531866175, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7947347164154053, + "num_tokens": 517770427.0, + "step": 13575 + }, + { + "epoch": 1.7270067421447655, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7948509454727173, + "num_tokens": 517804202.0, + "step": 13576 + }, + { + "epoch": 1.7271339524233558, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7885135412216187, + "num_tokens": 517834825.0, + "step": 13577 + }, + { + "epoch": 1.7272611627019463, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7896180152893066, + "num_tokens": 517874600.0, + "step": 13578 + }, + { + "epoch": 1.7273883729805368, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7890167832374573, + "num_tokens": 517913488.0, + "step": 13579 + }, + { + "epoch": 1.7275155832591274, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8171104192733765, + "num_tokens": 517951094.0, + "step": 13580 + }, + { + "epoch": 1.7276427935377179, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7913767695426941, + "num_tokens": 517993811.0, + "step": 13581 + }, + { + "epoch": 1.7277700038163084, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7980681657791138, + "num_tokens": 518032860.0, + "step": 13582 + }, + { + "epoch": 1.7278972140948987, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7938207387924194, + "num_tokens": 518070256.0, + "step": 13583 + }, + { + "epoch": 1.7280244243734892, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8072978258132935, + "num_tokens": 518102828.0, + "step": 13584 + }, + { + "epoch": 1.7281516346520798, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8171491622924805, + "num_tokens": 518146859.0, + "step": 13585 + }, + { + "epoch": 1.7282788449306703, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8112274408340454, + "num_tokens": 518186173.0, + "step": 13586 + }, + { + "epoch": 1.7284060552092608, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7784889936447144, + "num_tokens": 518221903.0, + "step": 13587 + }, + { + "epoch": 1.7285332654878514, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.784612774848938, + "num_tokens": 518258048.0, + "step": 13588 + }, + { + "epoch": 1.7286604757664419, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7928450703620911, + "num_tokens": 518299127.0, + "step": 13589 + }, + { + "epoch": 1.7287876860450324, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8109789490699768, + "num_tokens": 518336520.0, + "step": 13590 + }, + { + "epoch": 1.728914896323623, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.790854811668396, + "num_tokens": 518368484.0, + "step": 13591 + }, + { + "epoch": 1.7290421066022135, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7983240485191345, + "num_tokens": 518408439.0, + "step": 13592 + }, + { + "epoch": 1.729169316880804, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7818030714988708, + "num_tokens": 518447414.0, + "step": 13593 + }, + { + "epoch": 1.7292965271593945, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7917855978012085, + "num_tokens": 518486455.0, + "step": 13594 + }, + { + "epoch": 1.729423737437985, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.818172812461853, + "num_tokens": 518527023.0, + "step": 13595 + }, + { + "epoch": 1.7295509477165756, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7917587757110596, + "num_tokens": 518563285.0, + "step": 13596 + }, + { + "epoch": 1.729678157995166, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8006744384765625, + "num_tokens": 518607610.0, + "step": 13597 + }, + { + "epoch": 1.7298053682737566, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8105959892272949, + "num_tokens": 518649434.0, + "step": 13598 + }, + { + "epoch": 1.7299325785523472, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8067223429679871, + "num_tokens": 518684508.0, + "step": 13599 + }, + { + "epoch": 1.7300597888309377, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8021019697189331, + "num_tokens": 518717883.0, + "step": 13600 + }, + { + "epoch": 1.730186999109528, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.802466869354248, + "num_tokens": 518756158.0, + "step": 13601 + }, + { + "epoch": 1.7303142093881185, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8004019260406494, + "num_tokens": 518793003.0, + "step": 13602 + }, + { + "epoch": 1.730441419666709, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7956539392471313, + "num_tokens": 518838485.0, + "step": 13603 + }, + { + "epoch": 1.7305686299452996, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7930620312690735, + "num_tokens": 518885936.0, + "step": 13604 + }, + { + "epoch": 1.7306958402238901, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7840057611465454, + "num_tokens": 518925684.0, + "step": 13605 + }, + { + "epoch": 1.7308230505024806, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7985102534294128, + "num_tokens": 518966438.0, + "step": 13606 + }, + { + "epoch": 1.730950260781071, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8015540838241577, + "num_tokens": 519013576.0, + "step": 13607 + }, + { + "epoch": 1.7310774710596615, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7915704250335693, + "num_tokens": 519045342.0, + "step": 13608 + }, + { + "epoch": 1.731204681338252, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7861655354499817, + "num_tokens": 519088566.0, + "step": 13609 + }, + { + "epoch": 1.7313318916168425, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8036394119262695, + "num_tokens": 519126516.0, + "step": 13610 + }, + { + "epoch": 1.731459101895433, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8105894923210144, + "num_tokens": 519165431.0, + "step": 13611 + }, + { + "epoch": 1.7315863121740236, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8168987035751343, + "num_tokens": 519200991.0, + "step": 13612 + }, + { + "epoch": 1.7317135224526141, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8115581274032593, + "num_tokens": 519245988.0, + "step": 13613 + }, + { + "epoch": 1.7318407327312046, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8043515682220459, + "num_tokens": 519282595.0, + "step": 13614 + }, + { + "epoch": 1.7319679430097952, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7933944463729858, + "num_tokens": 519319061.0, + "step": 13615 + }, + { + "epoch": 1.7320951532883857, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7950976490974426, + "num_tokens": 519358617.0, + "step": 13616 + }, + { + "epoch": 1.7322223635669762, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.793386697769165, + "num_tokens": 519392461.0, + "step": 13617 + }, + { + "epoch": 1.7323495738455668, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8071259260177612, + "num_tokens": 519432893.0, + "step": 13618 + }, + { + "epoch": 1.7324767841241573, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7977704405784607, + "num_tokens": 519473445.0, + "step": 13619 + }, + { + "epoch": 1.7326039944027478, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8005592823028564, + "num_tokens": 519518883.0, + "step": 13620 + }, + { + "epoch": 1.7327312046813383, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8000317215919495, + "num_tokens": 519550123.0, + "step": 13621 + }, + { + "epoch": 1.7328584149599289, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7863648533821106, + "num_tokens": 519592538.0, + "step": 13622 + }, + { + "epoch": 1.7329856252385194, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7968196868896484, + "num_tokens": 519633466.0, + "step": 13623 + }, + { + "epoch": 1.73311283551711, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7907471060752869, + "num_tokens": 519672569.0, + "step": 13624 + }, + { + "epoch": 1.7332400457957005, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8083165287971497, + "num_tokens": 519707383.0, + "step": 13625 + }, + { + "epoch": 1.7333672560742908, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7796981930732727, + "num_tokens": 519748187.0, + "step": 13626 + }, + { + "epoch": 1.7334944663528813, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7949334979057312, + "num_tokens": 519790215.0, + "step": 13627 + }, + { + "epoch": 1.7336216766314718, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7918076515197754, + "num_tokens": 519827278.0, + "step": 13628 + }, + { + "epoch": 1.7337488869100623, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8046249151229858, + "num_tokens": 519860216.0, + "step": 13629 + }, + { + "epoch": 1.7338760971886529, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.798234760761261, + "num_tokens": 519894442.0, + "step": 13630 + }, + { + "epoch": 1.7340033074672434, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7982673645019531, + "num_tokens": 519932524.0, + "step": 13631 + }, + { + "epoch": 1.7341305177458337, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7968178987503052, + "num_tokens": 519974366.0, + "step": 13632 + }, + { + "epoch": 1.7342577280244242, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8106536865234375, + "num_tokens": 520013983.0, + "step": 13633 + }, + { + "epoch": 1.7343849383030148, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8016507625579834, + "num_tokens": 520047671.0, + "step": 13634 + }, + { + "epoch": 1.7345121485816053, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7925065159797668, + "num_tokens": 520086574.0, + "step": 13635 + }, + { + "epoch": 1.7346393588601958, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7877278327941895, + "num_tokens": 520130613.0, + "step": 13636 + }, + { + "epoch": 1.7347665691387864, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8126338124275208, + "num_tokens": 520163763.0, + "step": 13637 + }, + { + "epoch": 1.7348937794173769, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.793867290019989, + "num_tokens": 520198663.0, + "step": 13638 + }, + { + "epoch": 1.7350209896959674, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.797012448310852, + "num_tokens": 520231737.0, + "step": 13639 + }, + { + "epoch": 1.735148199974558, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8052669763565063, + "num_tokens": 520275664.0, + "step": 13640 + }, + { + "epoch": 1.7352754102531485, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7872098684310913, + "num_tokens": 520316100.0, + "step": 13641 + }, + { + "epoch": 1.735402620531739, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7931442260742188, + "num_tokens": 520348167.0, + "step": 13642 + }, + { + "epoch": 1.7355298308103295, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7819272875785828, + "num_tokens": 520385234.0, + "step": 13643 + }, + { + "epoch": 1.73565704108892, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7873749136924744, + "num_tokens": 520423051.0, + "step": 13644 + }, + { + "epoch": 1.7357842513675106, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7886835336685181, + "num_tokens": 520467202.0, + "step": 13645 + }, + { + "epoch": 1.735911461646101, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8001722097396851, + "num_tokens": 520501434.0, + "step": 13646 + }, + { + "epoch": 1.7360386719246916, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7963012456893921, + "num_tokens": 520542757.0, + "step": 13647 + }, + { + "epoch": 1.7361658822032822, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7859615683555603, + "num_tokens": 520585216.0, + "step": 13648 + }, + { + "epoch": 1.7362930924818727, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7992985248565674, + "num_tokens": 520620066.0, + "step": 13649 + }, + { + "epoch": 1.736420302760463, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8033296465873718, + "num_tokens": 520660917.0, + "step": 13650 + }, + { + "epoch": 1.7365475130390535, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7898120880126953, + "num_tokens": 520704255.0, + "step": 13651 + }, + { + "epoch": 1.736674723317644, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7849410772323608, + "num_tokens": 520743036.0, + "step": 13652 + }, + { + "epoch": 1.7368019335962346, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.790949821472168, + "num_tokens": 520778325.0, + "step": 13653 + }, + { + "epoch": 1.736929143874825, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8168168067932129, + "num_tokens": 520815226.0, + "step": 13654 + }, + { + "epoch": 1.7370563541534156, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7768275737762451, + "num_tokens": 520856606.0, + "step": 13655 + }, + { + "epoch": 1.737183564432006, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7901606559753418, + "num_tokens": 520893493.0, + "step": 13656 + }, + { + "epoch": 1.7373107747105965, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8022278547286987, + "num_tokens": 520930391.0, + "step": 13657 + }, + { + "epoch": 1.737437984989187, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8026934862136841, + "num_tokens": 520970512.0, + "step": 13658 + }, + { + "epoch": 1.7375651952677775, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.784247636795044, + "num_tokens": 521011787.0, + "step": 13659 + }, + { + "epoch": 1.737692405546368, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8048567175865173, + "num_tokens": 521053494.0, + "step": 13660 + }, + { + "epoch": 1.7378196158249586, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7804552316665649, + "num_tokens": 521089291.0, + "step": 13661 + }, + { + "epoch": 1.7379468261035491, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8122521638870239, + "num_tokens": 521128910.0, + "step": 13662 + }, + { + "epoch": 1.7380740363821396, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7966477274894714, + "num_tokens": 521168175.0, + "step": 13663 + }, + { + "epoch": 1.7382012466607302, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7966933250427246, + "num_tokens": 521208020.0, + "step": 13664 + }, + { + "epoch": 1.7383284569393207, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8088210821151733, + "num_tokens": 521249828.0, + "step": 13665 + }, + { + "epoch": 1.7384556672179112, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8018191456794739, + "num_tokens": 521289940.0, + "step": 13666 + }, + { + "epoch": 1.7385828774965018, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8169278502464294, + "num_tokens": 521327426.0, + "step": 13667 + }, + { + "epoch": 1.7387100877750923, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.790314257144928, + "num_tokens": 521372112.0, + "step": 13668 + }, + { + "epoch": 1.7388372980536828, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7951548099517822, + "num_tokens": 521410842.0, + "step": 13669 + }, + { + "epoch": 1.7389645083322733, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8079620003700256, + "num_tokens": 521446854.0, + "step": 13670 + }, + { + "epoch": 1.7390917186108639, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7901763916015625, + "num_tokens": 521486284.0, + "step": 13671 + }, + { + "epoch": 1.7392189288894544, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7831434607505798, + "num_tokens": 521522716.0, + "step": 13672 + }, + { + "epoch": 1.739346139168045, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7875900268554688, + "num_tokens": 521551549.0, + "step": 13673 + }, + { + "epoch": 1.7394733494466355, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.803760290145874, + "num_tokens": 521589552.0, + "step": 13674 + }, + { + "epoch": 1.7396005597252258, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.8147284984588623, + "num_tokens": 521627906.0, + "step": 13675 + }, + { + "epoch": 1.7397277700038163, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7871875762939453, + "num_tokens": 521666615.0, + "step": 13676 + }, + { + "epoch": 1.7398549802824068, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.8007752895355225, + "num_tokens": 521707742.0, + "step": 13677 + }, + { + "epoch": 1.7399821905609973, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7911158204078674, + "num_tokens": 521746848.0, + "step": 13678 + }, + { + "epoch": 1.7401094008395879, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7944425344467163, + "num_tokens": 521782917.0, + "step": 13679 + }, + { + "epoch": 1.7402366111181784, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.795072078704834, + "num_tokens": 521825609.0, + "step": 13680 + }, + { + "epoch": 1.7403638213967687, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8131396174430847, + "num_tokens": 521862277.0, + "step": 13681 + }, + { + "epoch": 1.7404910316753592, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.817718505859375, + "num_tokens": 521894951.0, + "step": 13682 + }, + { + "epoch": 1.7406182419539498, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7974865436553955, + "num_tokens": 521932158.0, + "step": 13683 + }, + { + "epoch": 1.7407454522325403, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8059439659118652, + "num_tokens": 521971050.0, + "step": 13684 + }, + { + "epoch": 1.7408726625111308, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8129618167877197, + "num_tokens": 522007617.0, + "step": 13685 + }, + { + "epoch": 1.7409998727897213, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7915234565734863, + "num_tokens": 522053484.0, + "step": 13686 + }, + { + "epoch": 1.7411270830683119, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.790672779083252, + "num_tokens": 522088047.0, + "step": 13687 + }, + { + "epoch": 1.7412542933469024, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7989027500152588, + "num_tokens": 522129865.0, + "step": 13688 + }, + { + "epoch": 1.741381503625493, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7979143261909485, + "num_tokens": 522170154.0, + "step": 13689 + }, + { + "epoch": 1.7415087139040835, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8218488693237305, + "num_tokens": 522204586.0, + "step": 13690 + }, + { + "epoch": 1.741635924182674, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.803587794303894, + "num_tokens": 522241942.0, + "step": 13691 + }, + { + "epoch": 1.7417631344612645, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7958858013153076, + "num_tokens": 522275800.0, + "step": 13692 + }, + { + "epoch": 1.741890344739855, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7894362211227417, + "num_tokens": 522317024.0, + "step": 13693 + }, + { + "epoch": 1.7420175550184456, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8112512230873108, + "num_tokens": 522358510.0, + "step": 13694 + }, + { + "epoch": 1.742144765297036, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.790694534778595, + "num_tokens": 522395960.0, + "step": 13695 + }, + { + "epoch": 1.7422719755756266, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8074531555175781, + "num_tokens": 522428807.0, + "step": 13696 + }, + { + "epoch": 1.7423991858542172, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8018958568572998, + "num_tokens": 522464472.0, + "step": 13697 + }, + { + "epoch": 1.7425263961328077, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8170706033706665, + "num_tokens": 522509522.0, + "step": 13698 + }, + { + "epoch": 1.742653606411398, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7845619916915894, + "num_tokens": 522545581.0, + "step": 13699 + }, + { + "epoch": 1.7427808166899885, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7981555461883545, + "num_tokens": 522582072.0, + "step": 13700 + }, + { + "epoch": 1.742908026968579, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7997180819511414, + "num_tokens": 522618183.0, + "step": 13701 + }, + { + "epoch": 1.7430352372471696, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.784827709197998, + "num_tokens": 522660578.0, + "step": 13702 + }, + { + "epoch": 1.74316244752576, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.8133342266082764, + "num_tokens": 522701934.0, + "step": 13703 + }, + { + "epoch": 1.7432896578043506, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.820788562297821, + "num_tokens": 522737404.0, + "step": 13704 + }, + { + "epoch": 1.743416868082941, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7953589558601379, + "num_tokens": 522774188.0, + "step": 13705 + }, + { + "epoch": 1.7435440783615315, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8066672086715698, + "num_tokens": 522810467.0, + "step": 13706 + }, + { + "epoch": 1.743671288640122, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7833356857299805, + "num_tokens": 522845246.0, + "step": 13707 + }, + { + "epoch": 1.7437984989187125, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8073418140411377, + "num_tokens": 522877150.0, + "step": 13708 + }, + { + "epoch": 1.743925709197303, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8102703094482422, + "num_tokens": 522913161.0, + "step": 13709 + }, + { + "epoch": 1.7440529194758936, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7853917479515076, + "num_tokens": 522953464.0, + "step": 13710 + }, + { + "epoch": 1.744180129754484, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8012521862983704, + "num_tokens": 522996874.0, + "step": 13711 + }, + { + "epoch": 1.7443073400330746, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7967187166213989, + "num_tokens": 523031349.0, + "step": 13712 + }, + { + "epoch": 1.7444345503116652, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7887418270111084, + "num_tokens": 523068910.0, + "step": 13713 + }, + { + "epoch": 1.7445617605902557, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8135465383529663, + "num_tokens": 523106179.0, + "step": 13714 + }, + { + "epoch": 1.7446889708688462, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.80227130651474, + "num_tokens": 523140827.0, + "step": 13715 + }, + { + "epoch": 1.7448161811474368, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8134807348251343, + "num_tokens": 523182880.0, + "step": 13716 + }, + { + "epoch": 1.7449433914260273, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8045426607131958, + "num_tokens": 523213059.0, + "step": 13717 + }, + { + "epoch": 1.7450706017046178, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.8016552925109863, + "num_tokens": 523254398.0, + "step": 13718 + }, + { + "epoch": 1.7451978119832083, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8108529448509216, + "num_tokens": 523293325.0, + "step": 13719 + }, + { + "epoch": 1.7453250222617989, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7949099540710449, + "num_tokens": 523334402.0, + "step": 13720 + }, + { + "epoch": 1.7454522325403894, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8051033020019531, + "num_tokens": 523376238.0, + "step": 13721 + }, + { + "epoch": 1.74557944281898, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8213586807250977, + "num_tokens": 523417074.0, + "step": 13722 + }, + { + "epoch": 1.7457066530975704, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.809579074382782, + "num_tokens": 523457984.0, + "step": 13723 + }, + { + "epoch": 1.7458338633761608, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7960790395736694, + "num_tokens": 523490071.0, + "step": 13724 + }, + { + "epoch": 1.7459610736547513, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7895077466964722, + "num_tokens": 523530105.0, + "step": 13725 + }, + { + "epoch": 1.7460882839333418, + "grad_norm": 9.5625, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8173998594284058, + "num_tokens": 523576698.0, + "step": 13726 + }, + { + "epoch": 1.7462154942119323, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8037583827972412, + "num_tokens": 523614387.0, + "step": 13727 + }, + { + "epoch": 1.7463427044905229, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7915369868278503, + "num_tokens": 523651149.0, + "step": 13728 + }, + { + "epoch": 1.7464699147691134, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8137402534484863, + "num_tokens": 523688448.0, + "step": 13729 + }, + { + "epoch": 1.7465971250477037, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8040410280227661, + "num_tokens": 523725805.0, + "step": 13730 + }, + { + "epoch": 1.7467243353262942, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7984445095062256, + "num_tokens": 523763564.0, + "step": 13731 + }, + { + "epoch": 1.7468515456048848, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7976887226104736, + "num_tokens": 523804422.0, + "step": 13732 + }, + { + "epoch": 1.7469787558834753, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.809794545173645, + "num_tokens": 523845889.0, + "step": 13733 + }, + { + "epoch": 1.7471059661620658, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8127065896987915, + "num_tokens": 523883623.0, + "step": 13734 + }, + { + "epoch": 1.7472331764406563, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8083978891372681, + "num_tokens": 523917273.0, + "step": 13735 + }, + { + "epoch": 1.7473603867192469, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.79781174659729, + "num_tokens": 523956149.0, + "step": 13736 + }, + { + "epoch": 1.7474875969978374, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7973133325576782, + "num_tokens": 523997229.0, + "step": 13737 + }, + { + "epoch": 1.747614807276428, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8046542406082153, + "num_tokens": 524032921.0, + "step": 13738 + }, + { + "epoch": 1.7477420175550185, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.8007394671440125, + "num_tokens": 524065016.0, + "step": 13739 + }, + { + "epoch": 1.747869227833609, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.784986138343811, + "num_tokens": 524107001.0, + "step": 13740 + }, + { + "epoch": 1.7479964381121995, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7960549592971802, + "num_tokens": 524149283.0, + "step": 13741 + }, + { + "epoch": 1.74812364839079, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7956255078315735, + "num_tokens": 524187240.0, + "step": 13742 + }, + { + "epoch": 1.7482508586693806, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8064634799957275, + "num_tokens": 524216793.0, + "step": 13743 + }, + { + "epoch": 1.748378068947971, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.804365873336792, + "num_tokens": 524257629.0, + "step": 13744 + }, + { + "epoch": 1.7485052792265616, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8090173602104187, + "num_tokens": 524291090.0, + "step": 13745 + }, + { + "epoch": 1.7486324895051522, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8128218650817871, + "num_tokens": 524324882.0, + "step": 13746 + }, + { + "epoch": 1.7487596997837427, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7819594740867615, + "num_tokens": 524359422.0, + "step": 13747 + }, + { + "epoch": 1.748886910062333, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8070223927497864, + "num_tokens": 524398383.0, + "step": 13748 + }, + { + "epoch": 1.7490141203409235, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7968977689743042, + "num_tokens": 524437984.0, + "step": 13749 + }, + { + "epoch": 1.749141330619514, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8040433526039124, + "num_tokens": 524471324.0, + "step": 13750 + }, + { + "epoch": 1.7492685408981046, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7937625646591187, + "num_tokens": 524511471.0, + "step": 13751 + }, + { + "epoch": 1.749395751176695, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8009960055351257, + "num_tokens": 524541743.0, + "step": 13752 + }, + { + "epoch": 1.7495229614552856, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8014025688171387, + "num_tokens": 524581470.0, + "step": 13753 + }, + { + "epoch": 1.749650171733876, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8006798028945923, + "num_tokens": 524612722.0, + "step": 13754 + }, + { + "epoch": 1.7497773820124665, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.8105760812759399, + "num_tokens": 524647640.0, + "step": 13755 + }, + { + "epoch": 1.749904592291057, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7925266027450562, + "num_tokens": 524682366.0, + "step": 13756 + }, + { + "epoch": 1.7500318025696475, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.797997236251831, + "num_tokens": 524724242.0, + "step": 13757 + }, + { + "epoch": 1.750159012848238, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7798146605491638, + "num_tokens": 524757373.0, + "step": 13758 + }, + { + "epoch": 1.7502862231268286, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.788764238357544, + "num_tokens": 524797704.0, + "step": 13759 + }, + { + "epoch": 1.750413433405419, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8035087585449219, + "num_tokens": 524836710.0, + "step": 13760 + }, + { + "epoch": 1.7505406436840096, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7938781380653381, + "num_tokens": 524870224.0, + "step": 13761 + }, + { + "epoch": 1.7506678539626002, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8087409138679504, + "num_tokens": 524905345.0, + "step": 13762 + }, + { + "epoch": 1.7507950642411907, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7851456999778748, + "num_tokens": 524943893.0, + "step": 13763 + }, + { + "epoch": 1.7509222745197812, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7930593490600586, + "num_tokens": 524982924.0, + "step": 13764 + }, + { + "epoch": 1.7510494847983717, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7993550300598145, + "num_tokens": 525016749.0, + "step": 13765 + }, + { + "epoch": 1.7511766950769623, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8007606267929077, + "num_tokens": 525049178.0, + "step": 13766 + }, + { + "epoch": 1.7513039053555528, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7800390720367432, + "num_tokens": 525083456.0, + "step": 13767 + }, + { + "epoch": 1.7514311156341433, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8118483424186707, + "num_tokens": 525120159.0, + "step": 13768 + }, + { + "epoch": 1.7515583259127339, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8027092218399048, + "num_tokens": 525159174.0, + "step": 13769 + }, + { + "epoch": 1.7516855361913244, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7955981492996216, + "num_tokens": 525198373.0, + "step": 13770 + }, + { + "epoch": 1.751812746469915, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8017012476921082, + "num_tokens": 525240618.0, + "step": 13771 + }, + { + "epoch": 1.7519399567485054, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8164889812469482, + "num_tokens": 525274899.0, + "step": 13772 + }, + { + "epoch": 1.7520671670270958, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7836456298828125, + "num_tokens": 525310144.0, + "step": 13773 + }, + { + "epoch": 1.7521943773056863, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8096944093704224, + "num_tokens": 525348565.0, + "step": 13774 + }, + { + "epoch": 1.7523215875842768, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8064122200012207, + "num_tokens": 525383709.0, + "step": 13775 + }, + { + "epoch": 1.7524487978628673, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7887594103813171, + "num_tokens": 525423983.0, + "step": 13776 + }, + { + "epoch": 1.7525760081414579, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8050360083580017, + "num_tokens": 525457357.0, + "step": 13777 + }, + { + "epoch": 1.7527032184200484, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7975469827651978, + "num_tokens": 525497021.0, + "step": 13778 + }, + { + "epoch": 1.7528304286986387, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.815036416053772, + "num_tokens": 525533592.0, + "step": 13779 + }, + { + "epoch": 1.7529576389772292, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7946113348007202, + "num_tokens": 525572747.0, + "step": 13780 + }, + { + "epoch": 1.7530848492558198, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8052215576171875, + "num_tokens": 525613578.0, + "step": 13781 + }, + { + "epoch": 1.7532120595344103, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.80936598777771, + "num_tokens": 525655047.0, + "step": 13782 + }, + { + "epoch": 1.7533392698130008, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7940999865531921, + "num_tokens": 525689561.0, + "step": 13783 + }, + { + "epoch": 1.7534664800915913, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7962138056755066, + "num_tokens": 525726946.0, + "step": 13784 + }, + { + "epoch": 1.7535936903701819, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7895817160606384, + "num_tokens": 525762089.0, + "step": 13785 + }, + { + "epoch": 1.7537209006487724, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7994402647018433, + "num_tokens": 525796081.0, + "step": 13786 + }, + { + "epoch": 1.753848110927363, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7880010008811951, + "num_tokens": 525829945.0, + "step": 13787 + }, + { + "epoch": 1.7539753212059535, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8129472136497498, + "num_tokens": 525871289.0, + "step": 13788 + }, + { + "epoch": 1.754102531484544, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8062306046485901, + "num_tokens": 525908455.0, + "step": 13789 + }, + { + "epoch": 1.7542297417631345, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.5668, + "mean_token_accuracy": 0.8276541829109192, + "num_tokens": 525943734.0, + "step": 13790 + }, + { + "epoch": 1.754356952041725, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8269851207733154, + "num_tokens": 525988070.0, + "step": 13791 + }, + { + "epoch": 1.7544841623203156, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8190566897392273, + "num_tokens": 526027940.0, + "step": 13792 + }, + { + "epoch": 1.754611372598906, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8104026913642883, + "num_tokens": 526061357.0, + "step": 13793 + }, + { + "epoch": 1.7547385828774966, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8062887191772461, + "num_tokens": 526102157.0, + "step": 13794 + }, + { + "epoch": 1.7548657931560872, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7938051223754883, + "num_tokens": 526137241.0, + "step": 13795 + }, + { + "epoch": 1.7549930034346777, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7891460061073303, + "num_tokens": 526169079.0, + "step": 13796 + }, + { + "epoch": 1.755120213713268, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8041045665740967, + "num_tokens": 526212212.0, + "step": 13797 + }, + { + "epoch": 1.7552474239918585, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8099172115325928, + "num_tokens": 526250380.0, + "step": 13798 + }, + { + "epoch": 1.755374634270449, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.806829035282135, + "num_tokens": 526292525.0, + "step": 13799 + }, + { + "epoch": 1.7555018445490396, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8022153377532959, + "num_tokens": 526330427.0, + "step": 13800 + }, + { + "epoch": 1.75562905482763, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7976813316345215, + "num_tokens": 526362954.0, + "step": 13801 + }, + { + "epoch": 1.7557562651062206, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7873526811599731, + "num_tokens": 526394310.0, + "step": 13802 + }, + { + "epoch": 1.755883475384811, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7793734073638916, + "num_tokens": 526438507.0, + "step": 13803 + }, + { + "epoch": 1.7560106856634015, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.8024263381958008, + "num_tokens": 526475124.0, + "step": 13804 + }, + { + "epoch": 1.756137895941992, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8084374666213989, + "num_tokens": 526514936.0, + "step": 13805 + }, + { + "epoch": 1.7562651062205825, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7922993898391724, + "num_tokens": 526548649.0, + "step": 13806 + }, + { + "epoch": 1.756392316499173, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7838866710662842, + "num_tokens": 526589028.0, + "step": 13807 + }, + { + "epoch": 1.7565195267777636, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8011050224304199, + "num_tokens": 526626327.0, + "step": 13808 + }, + { + "epoch": 1.756646737056354, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7993593811988831, + "num_tokens": 526672264.0, + "step": 13809 + }, + { + "epoch": 1.7567739473349446, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7951838970184326, + "num_tokens": 526709979.0, + "step": 13810 + }, + { + "epoch": 1.7569011576135352, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8168730735778809, + "num_tokens": 526745002.0, + "step": 13811 + }, + { + "epoch": 1.7570283678921257, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8128126859664917, + "num_tokens": 526785684.0, + "step": 13812 + }, + { + "epoch": 1.7571555781707162, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8207734227180481, + "num_tokens": 526820246.0, + "step": 13813 + }, + { + "epoch": 1.7572827884493067, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7989161014556885, + "num_tokens": 526854815.0, + "step": 13814 + }, + { + "epoch": 1.7574099987278973, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8033113479614258, + "num_tokens": 526896336.0, + "step": 13815 + }, + { + "epoch": 1.7575372090064878, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8167489171028137, + "num_tokens": 526939586.0, + "step": 13816 + }, + { + "epoch": 1.7576644192850783, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8025541305541992, + "num_tokens": 526975027.0, + "step": 13817 + }, + { + "epoch": 1.7577916295636689, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8060563802719116, + "num_tokens": 527017820.0, + "step": 13818 + }, + { + "epoch": 1.7579188398422594, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8141483664512634, + "num_tokens": 527051647.0, + "step": 13819 + }, + { + "epoch": 1.75804605012085, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8047621250152588, + "num_tokens": 527089590.0, + "step": 13820 + }, + { + "epoch": 1.7581732603994404, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7968919277191162, + "num_tokens": 527126696.0, + "step": 13821 + }, + { + "epoch": 1.7583004706780307, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7905908823013306, + "num_tokens": 527164823.0, + "step": 13822 + }, + { + "epoch": 1.7584276809566213, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7812027335166931, + "num_tokens": 527208032.0, + "step": 13823 + }, + { + "epoch": 1.7585548912352118, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8023196458816528, + "num_tokens": 527242773.0, + "step": 13824 + }, + { + "epoch": 1.7586821015138023, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8057974576950073, + "num_tokens": 527277968.0, + "step": 13825 + }, + { + "epoch": 1.7588093117923929, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8117263317108154, + "num_tokens": 527311077.0, + "step": 13826 + }, + { + "epoch": 1.7589365220709834, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8002001047134399, + "num_tokens": 527340543.0, + "step": 13827 + }, + { + "epoch": 1.7590637323495737, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7998431921005249, + "num_tokens": 527379596.0, + "step": 13828 + }, + { + "epoch": 1.7591909426281642, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7994330525398254, + "num_tokens": 527417873.0, + "step": 13829 + }, + { + "epoch": 1.7593181529067548, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8043630123138428, + "num_tokens": 527458607.0, + "step": 13830 + }, + { + "epoch": 1.7594453631853453, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7844533324241638, + "num_tokens": 527493299.0, + "step": 13831 + }, + { + "epoch": 1.7595725734639358, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7919211387634277, + "num_tokens": 527536065.0, + "step": 13832 + }, + { + "epoch": 1.7596997837425263, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7862635254859924, + "num_tokens": 527576477.0, + "step": 13833 + }, + { + "epoch": 1.7598269940211169, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7977672219276428, + "num_tokens": 527611777.0, + "step": 13834 + }, + { + "epoch": 1.7599542042997074, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8179870247840881, + "num_tokens": 527650236.0, + "step": 13835 + }, + { + "epoch": 1.760081414578298, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8056897521018982, + "num_tokens": 527685486.0, + "step": 13836 + }, + { + "epoch": 1.7602086248568884, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7856740951538086, + "num_tokens": 527725386.0, + "step": 13837 + }, + { + "epoch": 1.760335835135479, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7917758226394653, + "num_tokens": 527756213.0, + "step": 13838 + }, + { + "epoch": 1.7604630454140695, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8017556667327881, + "num_tokens": 527793884.0, + "step": 13839 + }, + { + "epoch": 1.76059025569266, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7983154654502869, + "num_tokens": 527826428.0, + "step": 13840 + }, + { + "epoch": 1.7607174659712506, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7926476001739502, + "num_tokens": 527866691.0, + "step": 13841 + }, + { + "epoch": 1.760844676249841, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8162716031074524, + "num_tokens": 527909482.0, + "step": 13842 + }, + { + "epoch": 1.7609718865284316, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.790269136428833, + "num_tokens": 527947818.0, + "step": 13843 + }, + { + "epoch": 1.7610990968070221, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8040022253990173, + "num_tokens": 527986760.0, + "step": 13844 + }, + { + "epoch": 1.7612263070856127, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7970401048660278, + "num_tokens": 528024742.0, + "step": 13845 + }, + { + "epoch": 1.761353517364203, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7767325043678284, + "num_tokens": 528060781.0, + "step": 13846 + }, + { + "epoch": 1.7614807276427935, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8031973838806152, + "num_tokens": 528099051.0, + "step": 13847 + }, + { + "epoch": 1.761607937921384, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7956304550170898, + "num_tokens": 528138325.0, + "step": 13848 + }, + { + "epoch": 1.7617351481999746, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7896180152893066, + "num_tokens": 528177997.0, + "step": 13849 + }, + { + "epoch": 1.761862358478565, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7985531687736511, + "num_tokens": 528216360.0, + "step": 13850 + }, + { + "epoch": 1.7619895687571556, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8162896037101746, + "num_tokens": 528251100.0, + "step": 13851 + }, + { + "epoch": 1.762116779035746, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8033244013786316, + "num_tokens": 528286298.0, + "step": 13852 + }, + { + "epoch": 1.7622439893143365, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8171948194503784, + "num_tokens": 528322970.0, + "step": 13853 + }, + { + "epoch": 1.762371199592927, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8194773197174072, + "num_tokens": 528355191.0, + "step": 13854 + }, + { + "epoch": 1.7624984098715175, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8015035390853882, + "num_tokens": 528388975.0, + "step": 13855 + }, + { + "epoch": 1.762625620150108, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8072732090950012, + "num_tokens": 528432139.0, + "step": 13856 + }, + { + "epoch": 1.7627528304286986, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8113493919372559, + "num_tokens": 528468315.0, + "step": 13857 + }, + { + "epoch": 1.762880040707289, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7946161031723022, + "num_tokens": 528510368.0, + "step": 13858 + }, + { + "epoch": 1.7630072509858796, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8015475273132324, + "num_tokens": 528545637.0, + "step": 13859 + }, + { + "epoch": 1.7631344612644702, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8049965500831604, + "num_tokens": 528586267.0, + "step": 13860 + }, + { + "epoch": 1.7632616715430607, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8096644282341003, + "num_tokens": 528629640.0, + "step": 13861 + }, + { + "epoch": 1.7633888818216512, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.783521831035614, + "num_tokens": 528665373.0, + "step": 13862 + }, + { + "epoch": 1.7635160921002417, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.8197681903839111, + "num_tokens": 528703426.0, + "step": 13863 + }, + { + "epoch": 1.7636433023788323, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7845024466514587, + "num_tokens": 528744886.0, + "step": 13864 + }, + { + "epoch": 1.7637705126574228, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.8139228224754333, + "num_tokens": 528779773.0, + "step": 13865 + }, + { + "epoch": 1.7638977229360133, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8049737215042114, + "num_tokens": 528821878.0, + "step": 13866 + }, + { + "epoch": 1.7640249332146039, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8005056381225586, + "num_tokens": 528858056.0, + "step": 13867 + }, + { + "epoch": 1.7641521434931944, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8019118309020996, + "num_tokens": 528903116.0, + "step": 13868 + }, + { + "epoch": 1.764279353771785, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7951697111129761, + "num_tokens": 528941561.0, + "step": 13869 + }, + { + "epoch": 1.7644065640503754, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8201861381530762, + "num_tokens": 528976215.0, + "step": 13870 + }, + { + "epoch": 1.7645337743289657, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.813971221446991, + "num_tokens": 529019280.0, + "step": 13871 + }, + { + "epoch": 1.7646609846075563, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8035880327224731, + "num_tokens": 529060733.0, + "step": 13872 + }, + { + "epoch": 1.7647881948861468, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8000847697257996, + "num_tokens": 529094948.0, + "step": 13873 + }, + { + "epoch": 1.7649154051647373, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8150606155395508, + "num_tokens": 529131374.0, + "step": 13874 + }, + { + "epoch": 1.7650426154433279, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7959344983100891, + "num_tokens": 529169133.0, + "step": 13875 + }, + { + "epoch": 1.7651698257219184, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.793861985206604, + "num_tokens": 529206467.0, + "step": 13876 + }, + { + "epoch": 1.7652970360005087, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.793298602104187, + "num_tokens": 529244891.0, + "step": 13877 + }, + { + "epoch": 1.7654242462790992, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8036423921585083, + "num_tokens": 529277384.0, + "step": 13878 + }, + { + "epoch": 1.7655514565576897, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8076003193855286, + "num_tokens": 529322530.0, + "step": 13879 + }, + { + "epoch": 1.7656786668362803, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7967591285705566, + "num_tokens": 529356345.0, + "step": 13880 + }, + { + "epoch": 1.7658058771148708, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8056681156158447, + "num_tokens": 529391877.0, + "step": 13881 + }, + { + "epoch": 1.7659330873934613, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.804260790348053, + "num_tokens": 529425460.0, + "step": 13882 + }, + { + "epoch": 1.7660602976720519, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7884355783462524, + "num_tokens": 529459660.0, + "step": 13883 + }, + { + "epoch": 1.7661875079506424, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7944154739379883, + "num_tokens": 529501839.0, + "step": 13884 + }, + { + "epoch": 1.766314718229233, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8004419803619385, + "num_tokens": 529540828.0, + "step": 13885 + }, + { + "epoch": 1.7664419285078234, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7995990514755249, + "num_tokens": 529581578.0, + "step": 13886 + }, + { + "epoch": 1.766569138786414, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8063653707504272, + "num_tokens": 529624363.0, + "step": 13887 + }, + { + "epoch": 1.7666963490650045, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.810157299041748, + "num_tokens": 529665286.0, + "step": 13888 + }, + { + "epoch": 1.766823559343595, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.8000442385673523, + "num_tokens": 529704150.0, + "step": 13889 + }, + { + "epoch": 1.7669507696221856, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8048957586288452, + "num_tokens": 529742110.0, + "step": 13890 + }, + { + "epoch": 1.767077979900776, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8045150637626648, + "num_tokens": 529776158.0, + "step": 13891 + }, + { + "epoch": 1.7672051901793666, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.775533139705658, + "num_tokens": 529814002.0, + "step": 13892 + }, + { + "epoch": 1.7673324004579571, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8030054569244385, + "num_tokens": 529846696.0, + "step": 13893 + }, + { + "epoch": 1.7674596107365477, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7978417277336121, + "num_tokens": 529881557.0, + "step": 13894 + }, + { + "epoch": 1.767586821015138, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7884649038314819, + "num_tokens": 529924337.0, + "step": 13895 + }, + { + "epoch": 1.7677140312937285, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7993285655975342, + "num_tokens": 529967195.0, + "step": 13896 + }, + { + "epoch": 1.767841241572319, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7773557901382446, + "num_tokens": 530003454.0, + "step": 13897 + }, + { + "epoch": 1.7679684518509096, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7723400592803955, + "num_tokens": 530042577.0, + "step": 13898 + }, + { + "epoch": 1.7680956621295, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7868731021881104, + "num_tokens": 530084903.0, + "step": 13899 + }, + { + "epoch": 1.7682228724080906, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7944085597991943, + "num_tokens": 530117598.0, + "step": 13900 + }, + { + "epoch": 1.768350082686681, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8004506230354309, + "num_tokens": 530151891.0, + "step": 13901 + }, + { + "epoch": 1.7684772929652715, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7894811630249023, + "num_tokens": 530191263.0, + "step": 13902 + }, + { + "epoch": 1.768604503243862, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7951219081878662, + "num_tokens": 530231156.0, + "step": 13903 + }, + { + "epoch": 1.7687317135224525, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.8155390620231628, + "num_tokens": 530270089.0, + "step": 13904 + }, + { + "epoch": 1.768858923801043, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8049271702766418, + "num_tokens": 530309808.0, + "step": 13905 + }, + { + "epoch": 1.7689861340796336, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7889202833175659, + "num_tokens": 530357723.0, + "step": 13906 + }, + { + "epoch": 1.769113344358224, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7894264459609985, + "num_tokens": 530389586.0, + "step": 13907 + }, + { + "epoch": 1.7692405546368146, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7936155200004578, + "num_tokens": 530430356.0, + "step": 13908 + }, + { + "epoch": 1.7693677649154052, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8046975135803223, + "num_tokens": 530472859.0, + "step": 13909 + }, + { + "epoch": 1.7694949751939957, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7857332229614258, + "num_tokens": 530515422.0, + "step": 13910 + }, + { + "epoch": 1.7696221854725862, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8009030222892761, + "num_tokens": 530553842.0, + "step": 13911 + }, + { + "epoch": 1.7697493957511767, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7998563051223755, + "num_tokens": 530592593.0, + "step": 13912 + }, + { + "epoch": 1.7698766060297673, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7980880737304688, + "num_tokens": 530632309.0, + "step": 13913 + }, + { + "epoch": 1.7700038163083578, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7937614917755127, + "num_tokens": 530667415.0, + "step": 13914 + }, + { + "epoch": 1.7701310265869483, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.798697829246521, + "num_tokens": 530705736.0, + "step": 13915 + }, + { + "epoch": 1.7702582368655388, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8077905774116516, + "num_tokens": 530744931.0, + "step": 13916 + }, + { + "epoch": 1.7703854471441294, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8171247243881226, + "num_tokens": 530784700.0, + "step": 13917 + }, + { + "epoch": 1.77051265742272, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.82082200050354, + "num_tokens": 530821097.0, + "step": 13918 + }, + { + "epoch": 1.7706398677013102, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7970207333564758, + "num_tokens": 530860444.0, + "step": 13919 + }, + { + "epoch": 1.7707670779799007, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7981705069541931, + "num_tokens": 530907304.0, + "step": 13920 + }, + { + "epoch": 1.7708942882584913, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7863296270370483, + "num_tokens": 530945692.0, + "step": 13921 + }, + { + "epoch": 1.7710214985370818, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8142191171646118, + "num_tokens": 530986695.0, + "step": 13922 + }, + { + "epoch": 1.7711487088156723, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8094680309295654, + "num_tokens": 531022656.0, + "step": 13923 + }, + { + "epoch": 1.7712759190942629, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8051313161849976, + "num_tokens": 531056896.0, + "step": 13924 + }, + { + "epoch": 1.7714031293728534, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7756671905517578, + "num_tokens": 531097286.0, + "step": 13925 + }, + { + "epoch": 1.7715303396514437, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8095738291740417, + "num_tokens": 531136430.0, + "step": 13926 + }, + { + "epoch": 1.7716575499300342, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7907978296279907, + "num_tokens": 531176698.0, + "step": 13927 + }, + { + "epoch": 1.7717847602086247, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.794942319393158, + "num_tokens": 531214343.0, + "step": 13928 + }, + { + "epoch": 1.7719119704872153, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8102987408638, + "num_tokens": 531252325.0, + "step": 13929 + }, + { + "epoch": 1.7720391807658058, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7862993478775024, + "num_tokens": 531290378.0, + "step": 13930 + }, + { + "epoch": 1.7721663910443963, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7911210060119629, + "num_tokens": 531330406.0, + "step": 13931 + }, + { + "epoch": 1.7722936013229869, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7979939579963684, + "num_tokens": 531367425.0, + "step": 13932 + }, + { + "epoch": 1.7724208116015774, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.5946, + "mean_token_accuracy": 0.8228296041488647, + "num_tokens": 531406465.0, + "step": 13933 + }, + { + "epoch": 1.772548021880168, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7854524850845337, + "num_tokens": 531445356.0, + "step": 13934 + }, + { + "epoch": 1.7726752321587584, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7977328300476074, + "num_tokens": 531484737.0, + "step": 13935 + }, + { + "epoch": 1.772802442437349, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7960083484649658, + "num_tokens": 531523453.0, + "step": 13936 + }, + { + "epoch": 1.7729296527159395, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8103751540184021, + "num_tokens": 531559627.0, + "step": 13937 + }, + { + "epoch": 1.77305686299453, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8056075572967529, + "num_tokens": 531597677.0, + "step": 13938 + }, + { + "epoch": 1.7731840732731206, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8107693195343018, + "num_tokens": 531642753.0, + "step": 13939 + }, + { + "epoch": 1.773311283551711, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7937847375869751, + "num_tokens": 531688249.0, + "step": 13940 + }, + { + "epoch": 1.7734384938303016, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.816076397895813, + "num_tokens": 531727994.0, + "step": 13941 + }, + { + "epoch": 1.7735657041088921, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7995771765708923, + "num_tokens": 531764428.0, + "step": 13942 + }, + { + "epoch": 1.7736929143874827, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7977422475814819, + "num_tokens": 531806604.0, + "step": 13943 + }, + { + "epoch": 1.773820124666073, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.585, + "mean_token_accuracy": 0.8205647468566895, + "num_tokens": 531845504.0, + "step": 13944 + }, + { + "epoch": 1.7739473349446635, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8082088232040405, + "num_tokens": 531880962.0, + "step": 13945 + }, + { + "epoch": 1.774074545223254, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7964797019958496, + "num_tokens": 531918182.0, + "step": 13946 + }, + { + "epoch": 1.7742017555018446, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8038645386695862, + "num_tokens": 531958340.0, + "step": 13947 + }, + { + "epoch": 1.774328965780435, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.784353494644165, + "num_tokens": 531996951.0, + "step": 13948 + }, + { + "epoch": 1.7744561760590256, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.808007001876831, + "num_tokens": 532036963.0, + "step": 13949 + }, + { + "epoch": 1.774583386337616, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7943394184112549, + "num_tokens": 532070937.0, + "step": 13950 + }, + { + "epoch": 1.7747105966162064, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7857922911643982, + "num_tokens": 532107866.0, + "step": 13951 + }, + { + "epoch": 1.774837806894797, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7939885258674622, + "num_tokens": 532151485.0, + "step": 13952 + }, + { + "epoch": 1.7749650171733875, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8093658685684204, + "num_tokens": 532191209.0, + "step": 13953 + }, + { + "epoch": 1.775092227451978, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7899076342582703, + "num_tokens": 532232219.0, + "step": 13954 + }, + { + "epoch": 1.7752194377305686, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7883530259132385, + "num_tokens": 532265330.0, + "step": 13955 + }, + { + "epoch": 1.775346648009159, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7967191934585571, + "num_tokens": 532295964.0, + "step": 13956 + }, + { + "epoch": 1.7754738582877496, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8073900938034058, + "num_tokens": 532333804.0, + "step": 13957 + }, + { + "epoch": 1.7756010685663401, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8083910346031189, + "num_tokens": 532373088.0, + "step": 13958 + }, + { + "epoch": 1.7757282788449307, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7942924499511719, + "num_tokens": 532409473.0, + "step": 13959 + }, + { + "epoch": 1.7758554891235212, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8128865361213684, + "num_tokens": 532447883.0, + "step": 13960 + }, + { + "epoch": 1.7759826994021117, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.805340588092804, + "num_tokens": 532485783.0, + "step": 13961 + }, + { + "epoch": 1.7761099096807023, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7990366220474243, + "num_tokens": 532511082.0, + "step": 13962 + }, + { + "epoch": 1.7762371199592928, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6016, + "mean_token_accuracy": 0.8200185298919678, + "num_tokens": 532555516.0, + "step": 13963 + }, + { + "epoch": 1.7763643302378833, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7898315787315369, + "num_tokens": 532599200.0, + "step": 13964 + }, + { + "epoch": 1.7764915405164738, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8110170364379883, + "num_tokens": 532636335.0, + "step": 13965 + }, + { + "epoch": 1.7766187507950644, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.814498245716095, + "num_tokens": 532678337.0, + "step": 13966 + }, + { + "epoch": 1.776745961073655, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.799045979976654, + "num_tokens": 532715051.0, + "step": 13967 + }, + { + "epoch": 1.7768731713522452, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8103055357933044, + "num_tokens": 532750836.0, + "step": 13968 + }, + { + "epoch": 1.7770003816308357, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8068779110908508, + "num_tokens": 532790331.0, + "step": 13969 + }, + { + "epoch": 1.7771275919094263, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7991927862167358, + "num_tokens": 532830169.0, + "step": 13970 + }, + { + "epoch": 1.7772548021880168, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7943224906921387, + "num_tokens": 532865957.0, + "step": 13971 + }, + { + "epoch": 1.7773820124666073, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7943851947784424, + "num_tokens": 532906112.0, + "step": 13972 + }, + { + "epoch": 1.7775092227451978, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8072642087936401, + "num_tokens": 532939676.0, + "step": 13973 + }, + { + "epoch": 1.7776364330237884, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7829560041427612, + "num_tokens": 532977668.0, + "step": 13974 + }, + { + "epoch": 1.7777636433023787, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7976742386817932, + "num_tokens": 533013862.0, + "step": 13975 + }, + { + "epoch": 1.7778908535809692, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.785729169845581, + "num_tokens": 533051222.0, + "step": 13976 + }, + { + "epoch": 1.7780180638595597, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7993068695068359, + "num_tokens": 533090182.0, + "step": 13977 + }, + { + "epoch": 1.7781452741381503, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8167552351951599, + "num_tokens": 533129600.0, + "step": 13978 + }, + { + "epoch": 1.7782724844167408, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8044404983520508, + "num_tokens": 533172351.0, + "step": 13979 + }, + { + "epoch": 1.7783996946953313, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7980986833572388, + "num_tokens": 533211583.0, + "step": 13980 + }, + { + "epoch": 1.7785269049739219, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8066024780273438, + "num_tokens": 533249960.0, + "step": 13981 + }, + { + "epoch": 1.7786541152525124, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8089092969894409, + "num_tokens": 533287328.0, + "step": 13982 + }, + { + "epoch": 1.778781325531103, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8015496134757996, + "num_tokens": 533327772.0, + "step": 13983 + }, + { + "epoch": 1.7789085358096934, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7955935001373291, + "num_tokens": 533372004.0, + "step": 13984 + }, + { + "epoch": 1.779035746088284, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7968676090240479, + "num_tokens": 533409502.0, + "step": 13985 + }, + { + "epoch": 1.7791629563668745, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7866455316543579, + "num_tokens": 533451160.0, + "step": 13986 + }, + { + "epoch": 1.779290166645465, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8002414703369141, + "num_tokens": 533480413.0, + "step": 13987 + }, + { + "epoch": 1.7794173769240555, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7873137593269348, + "num_tokens": 533518826.0, + "step": 13988 + }, + { + "epoch": 1.779544587202646, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7941219806671143, + "num_tokens": 533563760.0, + "step": 13989 + }, + { + "epoch": 1.7796717974812366, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7919464111328125, + "num_tokens": 533601499.0, + "step": 13990 + }, + { + "epoch": 1.7797990077598271, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8040691018104553, + "num_tokens": 533640757.0, + "step": 13991 + }, + { + "epoch": 1.7799262180384177, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8078337907791138, + "num_tokens": 533678047.0, + "step": 13992 + }, + { + "epoch": 1.780053428317008, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8007861971855164, + "num_tokens": 533720903.0, + "step": 13993 + }, + { + "epoch": 1.7801806385955985, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7977811098098755, + "num_tokens": 533759720.0, + "step": 13994 + }, + { + "epoch": 1.780307848874189, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7899695634841919, + "num_tokens": 533796122.0, + "step": 13995 + }, + { + "epoch": 1.7804350591527796, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8091760873794556, + "num_tokens": 533836326.0, + "step": 13996 + }, + { + "epoch": 1.78056226943137, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8156391382217407, + "num_tokens": 533867874.0, + "step": 13997 + }, + { + "epoch": 1.7806894797099606, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8064597249031067, + "num_tokens": 533902877.0, + "step": 13998 + }, + { + "epoch": 1.780816689988551, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8027840852737427, + "num_tokens": 533937440.0, + "step": 13999 + }, + { + "epoch": 1.7809439002671414, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8110501766204834, + "num_tokens": 533972308.0, + "step": 14000 + }, + { + "epoch": 1.781071110545732, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7899128794670105, + "num_tokens": 534011913.0, + "step": 14001 + }, + { + "epoch": 1.7811983208243225, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8172865509986877, + "num_tokens": 534057235.0, + "step": 14002 + }, + { + "epoch": 1.781325531102913, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8030246496200562, + "num_tokens": 534092850.0, + "step": 14003 + }, + { + "epoch": 1.7814527413815036, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8088624477386475, + "num_tokens": 534135310.0, + "step": 14004 + }, + { + "epoch": 1.781579951660094, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7961262464523315, + "num_tokens": 534176008.0, + "step": 14005 + }, + { + "epoch": 1.7817071619386846, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7987156510353088, + "num_tokens": 534208217.0, + "step": 14006 + }, + { + "epoch": 1.7818343722172751, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8048298358917236, + "num_tokens": 534247564.0, + "step": 14007 + }, + { + "epoch": 1.7819615824958657, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7830622792243958, + "num_tokens": 534281144.0, + "step": 14008 + }, + { + "epoch": 1.7820887927744562, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7895646095275879, + "num_tokens": 534317893.0, + "step": 14009 + }, + { + "epoch": 1.7822160030530467, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7981400489807129, + "num_tokens": 534356116.0, + "step": 14010 + }, + { + "epoch": 1.7823432133316373, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.768054187297821, + "num_tokens": 534390499.0, + "step": 14011 + }, + { + "epoch": 1.7824704236102278, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.79059898853302, + "num_tokens": 534424242.0, + "step": 14012 + }, + { + "epoch": 1.7825976338888183, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7900757789611816, + "num_tokens": 534467465.0, + "step": 14013 + }, + { + "epoch": 1.7827248441674088, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7950242161750793, + "num_tokens": 534502959.0, + "step": 14014 + }, + { + "epoch": 1.7828520544459994, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8062047958374023, + "num_tokens": 534549888.0, + "step": 14015 + }, + { + "epoch": 1.78297926472459, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8053333759307861, + "num_tokens": 534591686.0, + "step": 14016 + }, + { + "epoch": 1.7831064750031802, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.817080557346344, + "num_tokens": 534628584.0, + "step": 14017 + }, + { + "epoch": 1.7832336852817707, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7847442030906677, + "num_tokens": 534665684.0, + "step": 14018 + }, + { + "epoch": 1.7833608955603613, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8017489314079285, + "num_tokens": 534701604.0, + "step": 14019 + }, + { + "epoch": 1.7834881058389518, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7814397811889648, + "num_tokens": 534741765.0, + "step": 14020 + }, + { + "epoch": 1.7836153161175423, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8077780604362488, + "num_tokens": 534780331.0, + "step": 14021 + }, + { + "epoch": 1.7837425263961328, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7884599566459656, + "num_tokens": 534816254.0, + "step": 14022 + }, + { + "epoch": 1.7838697366747234, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7941508293151855, + "num_tokens": 534851367.0, + "step": 14023 + }, + { + "epoch": 1.7839969469533137, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7976399660110474, + "num_tokens": 534889077.0, + "step": 14024 + }, + { + "epoch": 1.7841241572319042, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.790570080280304, + "num_tokens": 534924249.0, + "step": 14025 + }, + { + "epoch": 1.7842513675104947, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7858996987342834, + "num_tokens": 534966370.0, + "step": 14026 + }, + { + "epoch": 1.7843785777890853, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8044581413269043, + "num_tokens": 535005869.0, + "step": 14027 + }, + { + "epoch": 1.7845057880676758, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7946505546569824, + "num_tokens": 535042057.0, + "step": 14028 + }, + { + "epoch": 1.7846329983462663, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7903372645378113, + "num_tokens": 535074410.0, + "step": 14029 + }, + { + "epoch": 1.7847602086248568, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7950491905212402, + "num_tokens": 535120356.0, + "step": 14030 + }, + { + "epoch": 1.7848874189034474, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7851495742797852, + "num_tokens": 535156939.0, + "step": 14031 + }, + { + "epoch": 1.785014629182038, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7983589172363281, + "num_tokens": 535195553.0, + "step": 14032 + }, + { + "epoch": 1.7851418394606284, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8064363598823547, + "num_tokens": 535235692.0, + "step": 14033 + }, + { + "epoch": 1.785269049739219, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7949512004852295, + "num_tokens": 535269265.0, + "step": 14034 + }, + { + "epoch": 1.7853962600178095, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8096506595611572, + "num_tokens": 535303825.0, + "step": 14035 + }, + { + "epoch": 1.7855234702964, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7913355231285095, + "num_tokens": 535338802.0, + "step": 14036 + }, + { + "epoch": 1.7856506805749905, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8027043342590332, + "num_tokens": 535376592.0, + "step": 14037 + }, + { + "epoch": 1.785777890853581, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8009902238845825, + "num_tokens": 535416528.0, + "step": 14038 + }, + { + "epoch": 1.7859051011321716, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7935397624969482, + "num_tokens": 535450892.0, + "step": 14039 + }, + { + "epoch": 1.7860323114107621, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8010371923446655, + "num_tokens": 535488198.0, + "step": 14040 + }, + { + "epoch": 1.7861595216893527, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7960162162780762, + "num_tokens": 535526020.0, + "step": 14041 + }, + { + "epoch": 1.786286731967943, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8038920164108276, + "num_tokens": 535560400.0, + "step": 14042 + }, + { + "epoch": 1.7864139422465335, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7918067574501038, + "num_tokens": 535594699.0, + "step": 14043 + }, + { + "epoch": 1.786541152525124, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7998759150505066, + "num_tokens": 535627019.0, + "step": 14044 + }, + { + "epoch": 1.7866683628037145, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8100615739822388, + "num_tokens": 535667479.0, + "step": 14045 + }, + { + "epoch": 1.786795573082305, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7907497882843018, + "num_tokens": 535705211.0, + "step": 14046 + }, + { + "epoch": 1.7869227833608956, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7821036577224731, + "num_tokens": 535744907.0, + "step": 14047 + }, + { + "epoch": 1.787049993639486, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.794784426689148, + "num_tokens": 535782526.0, + "step": 14048 + }, + { + "epoch": 1.7871772039180764, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8165934085845947, + "num_tokens": 535824153.0, + "step": 14049 + }, + { + "epoch": 1.787304414196667, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8120328187942505, + "num_tokens": 535859019.0, + "step": 14050 + }, + { + "epoch": 1.7874316244752575, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8136613368988037, + "num_tokens": 535896866.0, + "step": 14051 + }, + { + "epoch": 1.787558834753848, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8034317493438721, + "num_tokens": 535937128.0, + "step": 14052 + }, + { + "epoch": 1.7876860450324386, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8176353573799133, + "num_tokens": 535972727.0, + "step": 14053 + }, + { + "epoch": 1.787813255311029, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8075177669525146, + "num_tokens": 536011491.0, + "step": 14054 + }, + { + "epoch": 1.7879404655896196, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7992050051689148, + "num_tokens": 536048074.0, + "step": 14055 + }, + { + "epoch": 1.7880676758682101, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8082325458526611, + "num_tokens": 536089554.0, + "step": 14056 + }, + { + "epoch": 1.7881948861468007, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8083245754241943, + "num_tokens": 536127672.0, + "step": 14057 + }, + { + "epoch": 1.7883220964253912, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.796566367149353, + "num_tokens": 536157517.0, + "step": 14058 + }, + { + "epoch": 1.7884493067039817, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8131934404373169, + "num_tokens": 536192855.0, + "step": 14059 + }, + { + "epoch": 1.7885765169825723, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.800505518913269, + "num_tokens": 536231727.0, + "step": 14060 + }, + { + "epoch": 1.7887037272611628, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8155332803726196, + "num_tokens": 536270738.0, + "step": 14061 + }, + { + "epoch": 1.7888309375397533, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.5974, + "mean_token_accuracy": 0.820380687713623, + "num_tokens": 536313006.0, + "step": 14062 + }, + { + "epoch": 1.7889581478183438, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.793505072593689, + "num_tokens": 536350697.0, + "step": 14063 + }, + { + "epoch": 1.7890853580969344, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8086802363395691, + "num_tokens": 536394252.0, + "step": 14064 + }, + { + "epoch": 1.789212568375525, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.792262613773346, + "num_tokens": 536426591.0, + "step": 14065 + }, + { + "epoch": 1.7893397786541152, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7893798351287842, + "num_tokens": 536468567.0, + "step": 14066 + }, + { + "epoch": 1.7894669889327057, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7897576093673706, + "num_tokens": 536506352.0, + "step": 14067 + }, + { + "epoch": 1.7895941992112963, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7932051420211792, + "num_tokens": 536538122.0, + "step": 14068 + }, + { + "epoch": 1.7897214094898868, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7903434634208679, + "num_tokens": 536574617.0, + "step": 14069 + }, + { + "epoch": 1.7898486197684773, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8114390969276428, + "num_tokens": 536611895.0, + "step": 14070 + }, + { + "epoch": 1.7899758300470678, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7980243563652039, + "num_tokens": 536651336.0, + "step": 14071 + }, + { + "epoch": 1.7901030403256584, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7980906963348389, + "num_tokens": 536692566.0, + "step": 14072 + }, + { + "epoch": 1.7902302506042487, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8079216480255127, + "num_tokens": 536730855.0, + "step": 14073 + }, + { + "epoch": 1.7903574608828392, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7958922386169434, + "num_tokens": 536766478.0, + "step": 14074 + }, + { + "epoch": 1.7904846711614297, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7894413471221924, + "num_tokens": 536799900.0, + "step": 14075 + }, + { + "epoch": 1.7906118814400203, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8090587854385376, + "num_tokens": 536843597.0, + "step": 14076 + }, + { + "epoch": 1.7907390917186108, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8034020662307739, + "num_tokens": 536879189.0, + "step": 14077 + }, + { + "epoch": 1.7908663019972013, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8120230436325073, + "num_tokens": 536918099.0, + "step": 14078 + }, + { + "epoch": 1.7909935122757918, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7922669649124146, + "num_tokens": 536954410.0, + "step": 14079 + }, + { + "epoch": 1.7911207225543824, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.8202654123306274, + "num_tokens": 536986688.0, + "step": 14080 + }, + { + "epoch": 1.791247932832973, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8008430004119873, + "num_tokens": 537024181.0, + "step": 14081 + }, + { + "epoch": 1.7913751431115634, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8080803155899048, + "num_tokens": 537064174.0, + "step": 14082 + }, + { + "epoch": 1.791502353390154, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7894130945205688, + "num_tokens": 537100279.0, + "step": 14083 + }, + { + "epoch": 1.7916295636687445, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7711621522903442, + "num_tokens": 537135339.0, + "step": 14084 + }, + { + "epoch": 1.791756773947335, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7935000658035278, + "num_tokens": 537169608.0, + "step": 14085 + }, + { + "epoch": 1.7918839842259255, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8034824728965759, + "num_tokens": 537205560.0, + "step": 14086 + }, + { + "epoch": 1.792011194504516, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7832703590393066, + "num_tokens": 537238697.0, + "step": 14087 + }, + { + "epoch": 1.7921384047831066, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7983986139297485, + "num_tokens": 537282365.0, + "step": 14088 + }, + { + "epoch": 1.7922656150616971, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8029625415802002, + "num_tokens": 537320805.0, + "step": 14089 + }, + { + "epoch": 1.7923928253402877, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.8011633157730103, + "num_tokens": 537354607.0, + "step": 14090 + }, + { + "epoch": 1.792520035618878, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8100488781929016, + "num_tokens": 537386672.0, + "step": 14091 + }, + { + "epoch": 1.7926472458974685, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8019922375679016, + "num_tokens": 537430882.0, + "step": 14092 + }, + { + "epoch": 1.792774456176059, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8059607148170471, + "num_tokens": 537465114.0, + "step": 14093 + }, + { + "epoch": 1.7929016664546495, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7923270463943481, + "num_tokens": 537502820.0, + "step": 14094 + }, + { + "epoch": 1.79302887673324, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8087460994720459, + "num_tokens": 537537422.0, + "step": 14095 + }, + { + "epoch": 1.7931560870118306, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8013925552368164, + "num_tokens": 537566828.0, + "step": 14096 + }, + { + "epoch": 1.793283297290421, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7989209890365601, + "num_tokens": 537606102.0, + "step": 14097 + }, + { + "epoch": 1.7934105075690114, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8078904151916504, + "num_tokens": 537638527.0, + "step": 14098 + }, + { + "epoch": 1.793537717847602, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8038316965103149, + "num_tokens": 537683375.0, + "step": 14099 + }, + { + "epoch": 1.7936649281261925, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.798590898513794, + "num_tokens": 537731354.0, + "step": 14100 + }, + { + "epoch": 1.793792138404783, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7942426204681396, + "num_tokens": 537770055.0, + "step": 14101 + }, + { + "epoch": 1.7939193486833735, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.782563328742981, + "num_tokens": 537807977.0, + "step": 14102 + }, + { + "epoch": 1.794046558961964, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7943329215049744, + "num_tokens": 537847809.0, + "step": 14103 + }, + { + "epoch": 1.7941737692405546, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7795487642288208, + "num_tokens": 537885506.0, + "step": 14104 + }, + { + "epoch": 1.7943009795191451, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7947080731391907, + "num_tokens": 537918936.0, + "step": 14105 + }, + { + "epoch": 1.7944281897977357, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8035051226615906, + "num_tokens": 537954205.0, + "step": 14106 + }, + { + "epoch": 1.7945554000763262, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8036882877349854, + "num_tokens": 537989190.0, + "step": 14107 + }, + { + "epoch": 1.7946826103549167, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7927308082580566, + "num_tokens": 538028852.0, + "step": 14108 + }, + { + "epoch": 1.7948098206335072, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8003467321395874, + "num_tokens": 538061245.0, + "step": 14109 + }, + { + "epoch": 1.7949370309120978, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8157117962837219, + "num_tokens": 538103238.0, + "step": 14110 + }, + { + "epoch": 1.7950642411906883, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7893420457839966, + "num_tokens": 538142692.0, + "step": 14111 + }, + { + "epoch": 1.7951914514692788, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.8000918626785278, + "num_tokens": 538183155.0, + "step": 14112 + }, + { + "epoch": 1.7953186617478694, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7989135980606079, + "num_tokens": 538222063.0, + "step": 14113 + }, + { + "epoch": 1.7954458720264599, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.801324725151062, + "num_tokens": 538259847.0, + "step": 14114 + }, + { + "epoch": 1.7955730823050502, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8056650161743164, + "num_tokens": 538298557.0, + "step": 14115 + }, + { + "epoch": 1.7957002925836407, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7995883226394653, + "num_tokens": 538337926.0, + "step": 14116 + }, + { + "epoch": 1.7958275028622313, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8109627962112427, + "num_tokens": 538382755.0, + "step": 14117 + }, + { + "epoch": 1.7959547131408218, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.774444043636322, + "num_tokens": 538417639.0, + "step": 14118 + }, + { + "epoch": 1.7960819234194123, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7956401109695435, + "num_tokens": 538458444.0, + "step": 14119 + }, + { + "epoch": 1.7962091336980028, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8043110370635986, + "num_tokens": 538496580.0, + "step": 14120 + }, + { + "epoch": 1.7963363439765934, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8111079931259155, + "num_tokens": 538538474.0, + "step": 14121 + }, + { + "epoch": 1.7964635542551837, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8006621599197388, + "num_tokens": 538576828.0, + "step": 14122 + }, + { + "epoch": 1.7965907645337742, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.8003689050674438, + "num_tokens": 538611630.0, + "step": 14123 + }, + { + "epoch": 1.7967179748123647, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7782771587371826, + "num_tokens": 538659537.0, + "step": 14124 + }, + { + "epoch": 1.7968451850909553, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8079733848571777, + "num_tokens": 538700216.0, + "step": 14125 + }, + { + "epoch": 1.7969723953695458, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.796913743019104, + "num_tokens": 538737719.0, + "step": 14126 + }, + { + "epoch": 1.7970996056481363, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8087268471717834, + "num_tokens": 538774723.0, + "step": 14127 + }, + { + "epoch": 1.7972268159267268, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8105825185775757, + "num_tokens": 538815288.0, + "step": 14128 + }, + { + "epoch": 1.7973540262053174, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8030821084976196, + "num_tokens": 538858239.0, + "step": 14129 + }, + { + "epoch": 1.797481236483908, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8023708462715149, + "num_tokens": 538896947.0, + "step": 14130 + }, + { + "epoch": 1.7976084467624984, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8050957918167114, + "num_tokens": 538936511.0, + "step": 14131 + }, + { + "epoch": 1.797735657041089, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7873022556304932, + "num_tokens": 538969198.0, + "step": 14132 + }, + { + "epoch": 1.7978628673196795, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8084718585014343, + "num_tokens": 539002258.0, + "step": 14133 + }, + { + "epoch": 1.79799007759827, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7832672595977783, + "num_tokens": 539039993.0, + "step": 14134 + }, + { + "epoch": 1.7981172878768605, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.798092246055603, + "num_tokens": 539080713.0, + "step": 14135 + }, + { + "epoch": 1.798244498155451, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8133304119110107, + "num_tokens": 539121115.0, + "step": 14136 + }, + { + "epoch": 1.7983717084340416, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8209792971611023, + "num_tokens": 539159889.0, + "step": 14137 + }, + { + "epoch": 1.7984989187126321, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.800047755241394, + "num_tokens": 539201508.0, + "step": 14138 + }, + { + "epoch": 1.7986261289912227, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7798298597335815, + "num_tokens": 539238640.0, + "step": 14139 + }, + { + "epoch": 1.798753339269813, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7987209558486938, + "num_tokens": 539274797.0, + "step": 14140 + }, + { + "epoch": 1.7988805495484035, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7869651317596436, + "num_tokens": 539308410.0, + "step": 14141 + }, + { + "epoch": 1.799007759826994, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7912839651107788, + "num_tokens": 539345578.0, + "step": 14142 + }, + { + "epoch": 1.7991349701055845, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7807083129882812, + "num_tokens": 539378946.0, + "step": 14143 + }, + { + "epoch": 1.799262180384175, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.8173425793647766, + "num_tokens": 539417663.0, + "step": 14144 + }, + { + "epoch": 1.7993893906627656, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.8150146007537842, + "num_tokens": 539459597.0, + "step": 14145 + }, + { + "epoch": 1.799516600941356, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.805294930934906, + "num_tokens": 539499579.0, + "step": 14146 + }, + { + "epoch": 1.7996438112199464, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.811083197593689, + "num_tokens": 539538505.0, + "step": 14147 + }, + { + "epoch": 1.799771021498537, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7947090864181519, + "num_tokens": 539574824.0, + "step": 14148 + }, + { + "epoch": 1.7998982317771275, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7887893915176392, + "num_tokens": 539614181.0, + "step": 14149 + }, + { + "epoch": 1.800025442055718, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7887269258499146, + "num_tokens": 539656713.0, + "step": 14150 + }, + { + "epoch": 1.8001526523343085, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7976852059364319, + "num_tokens": 539695436.0, + "step": 14151 + }, + { + "epoch": 1.800279862612899, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8069576621055603, + "num_tokens": 539736549.0, + "step": 14152 + }, + { + "epoch": 1.8004070728914896, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8025169372558594, + "num_tokens": 539776326.0, + "step": 14153 + }, + { + "epoch": 1.8005342831700801, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.812056839466095, + "num_tokens": 539819665.0, + "step": 14154 + }, + { + "epoch": 1.8006614934486707, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7987072467803955, + "num_tokens": 539858175.0, + "step": 14155 + }, + { + "epoch": 1.8007887037272612, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7976982593536377, + "num_tokens": 539895152.0, + "step": 14156 + }, + { + "epoch": 1.8009159140058517, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8071529865264893, + "num_tokens": 539941033.0, + "step": 14157 + }, + { + "epoch": 1.8010431242844422, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7979235649108887, + "num_tokens": 539977959.0, + "step": 14158 + }, + { + "epoch": 1.8011703345630328, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7969082593917847, + "num_tokens": 540014857.0, + "step": 14159 + }, + { + "epoch": 1.8012975448416233, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7994338274002075, + "num_tokens": 540055775.0, + "step": 14160 + }, + { + "epoch": 1.8014247551202138, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.811592161655426, + "num_tokens": 540090287.0, + "step": 14161 + }, + { + "epoch": 1.8015519653988044, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8106675148010254, + "num_tokens": 540129139.0, + "step": 14162 + }, + { + "epoch": 1.8016791756773949, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8031579256057739, + "num_tokens": 540165187.0, + "step": 14163 + }, + { + "epoch": 1.8018063859559852, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8091685771942139, + "num_tokens": 540201113.0, + "step": 14164 + }, + { + "epoch": 1.8019335962345757, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8013665676116943, + "num_tokens": 540237448.0, + "step": 14165 + }, + { + "epoch": 1.8020608065131662, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7993083596229553, + "num_tokens": 540279703.0, + "step": 14166 + }, + { + "epoch": 1.8021880167917568, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.776414155960083, + "num_tokens": 540313071.0, + "step": 14167 + }, + { + "epoch": 1.8023152270703473, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8086693286895752, + "num_tokens": 540353940.0, + "step": 14168 + }, + { + "epoch": 1.8024424373489378, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.792902946472168, + "num_tokens": 540391241.0, + "step": 14169 + }, + { + "epoch": 1.8025696476275284, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8093428015708923, + "num_tokens": 540420205.0, + "step": 14170 + }, + { + "epoch": 1.8026968579061187, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8002332448959351, + "num_tokens": 540452121.0, + "step": 14171 + }, + { + "epoch": 1.8028240681847092, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.804271399974823, + "num_tokens": 540487415.0, + "step": 14172 + }, + { + "epoch": 1.8029512784632997, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7866753339767456, + "num_tokens": 540522293.0, + "step": 14173 + }, + { + "epoch": 1.8030784887418903, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8083065152168274, + "num_tokens": 540560345.0, + "step": 14174 + }, + { + "epoch": 1.8032056990204808, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8121761679649353, + "num_tokens": 540598299.0, + "step": 14175 + }, + { + "epoch": 1.8033329092990713, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8110007047653198, + "num_tokens": 540634480.0, + "step": 14176 + }, + { + "epoch": 1.8034601195776618, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7907508611679077, + "num_tokens": 540679473.0, + "step": 14177 + }, + { + "epoch": 1.8035873298562524, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7935153245925903, + "num_tokens": 540722259.0, + "step": 14178 + }, + { + "epoch": 1.803714540134843, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8024856448173523, + "num_tokens": 540758058.0, + "step": 14179 + }, + { + "epoch": 1.8038417504134334, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7964414358139038, + "num_tokens": 540798535.0, + "step": 14180 + }, + { + "epoch": 1.803968960692024, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.798164427280426, + "num_tokens": 540839423.0, + "step": 14181 + }, + { + "epoch": 1.8040961709706145, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8088026642799377, + "num_tokens": 540887512.0, + "step": 14182 + }, + { + "epoch": 1.804223381249205, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8010836243629456, + "num_tokens": 540922292.0, + "step": 14183 + }, + { + "epoch": 1.8043505915277955, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8093957901000977, + "num_tokens": 540964105.0, + "step": 14184 + }, + { + "epoch": 1.804477801806386, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7980761528015137, + "num_tokens": 541001793.0, + "step": 14185 + }, + { + "epoch": 1.8046050120849766, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8054968118667603, + "num_tokens": 541041104.0, + "step": 14186 + }, + { + "epoch": 1.8047322223635671, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7991198301315308, + "num_tokens": 541079643.0, + "step": 14187 + }, + { + "epoch": 1.8048594326421576, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8118128180503845, + "num_tokens": 541118554.0, + "step": 14188 + }, + { + "epoch": 1.804986642920748, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7845104932785034, + "num_tokens": 541156893.0, + "step": 14189 + }, + { + "epoch": 1.8051138531993385, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8211719989776611, + "num_tokens": 541197283.0, + "step": 14190 + }, + { + "epoch": 1.805241063477929, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.8140952587127686, + "num_tokens": 541232343.0, + "step": 14191 + }, + { + "epoch": 1.8053682737565195, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7920455932617188, + "num_tokens": 541270274.0, + "step": 14192 + }, + { + "epoch": 1.80549548403511, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7906420826911926, + "num_tokens": 541306018.0, + "step": 14193 + }, + { + "epoch": 1.8056226943137006, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7981072068214417, + "num_tokens": 541344974.0, + "step": 14194 + }, + { + "epoch": 1.805749904592291, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8054254055023193, + "num_tokens": 541378590.0, + "step": 14195 + }, + { + "epoch": 1.8058771148708814, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8073490262031555, + "num_tokens": 541415945.0, + "step": 14196 + }, + { + "epoch": 1.806004325149472, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7992604970932007, + "num_tokens": 541459577.0, + "step": 14197 + }, + { + "epoch": 1.8061315354280625, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8009493350982666, + "num_tokens": 541501041.0, + "step": 14198 + }, + { + "epoch": 1.806258745706653, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7913974523544312, + "num_tokens": 541538587.0, + "step": 14199 + }, + { + "epoch": 1.8063859559852435, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7869330644607544, + "num_tokens": 541582225.0, + "step": 14200 + }, + { + "epoch": 1.806513166263834, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8107283115386963, + "num_tokens": 541616956.0, + "step": 14201 + }, + { + "epoch": 1.8066403765424246, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.786666214466095, + "num_tokens": 541650814.0, + "step": 14202 + }, + { + "epoch": 1.8067675868210151, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8125903606414795, + "num_tokens": 541692167.0, + "step": 14203 + }, + { + "epoch": 1.8068947970996057, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7897052764892578, + "num_tokens": 541732427.0, + "step": 14204 + }, + { + "epoch": 1.8070220073781962, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8038817048072815, + "num_tokens": 541773256.0, + "step": 14205 + }, + { + "epoch": 1.8071492176567867, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8072237968444824, + "num_tokens": 541815008.0, + "step": 14206 + }, + { + "epoch": 1.8072764279353772, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.804888129234314, + "num_tokens": 541853462.0, + "step": 14207 + }, + { + "epoch": 1.8074036382139678, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7962574362754822, + "num_tokens": 541890487.0, + "step": 14208 + }, + { + "epoch": 1.8075308484925583, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8037211298942566, + "num_tokens": 541929857.0, + "step": 14209 + }, + { + "epoch": 1.8076580587711488, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7959003448486328, + "num_tokens": 541969923.0, + "step": 14210 + }, + { + "epoch": 1.8077852690497394, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8121851682662964, + "num_tokens": 542008802.0, + "step": 14211 + }, + { + "epoch": 1.8079124793283299, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.819246768951416, + "num_tokens": 542049276.0, + "step": 14212 + }, + { + "epoch": 1.8080396896069202, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7941445708274841, + "num_tokens": 542086921.0, + "step": 14213 + }, + { + "epoch": 1.8081668998855107, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8016356229782104, + "num_tokens": 542124291.0, + "step": 14214 + }, + { + "epoch": 1.8082941101641012, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7883749008178711, + "num_tokens": 542167600.0, + "step": 14215 + }, + { + "epoch": 1.8084213204426918, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7898925542831421, + "num_tokens": 542201134.0, + "step": 14216 + }, + { + "epoch": 1.8085485307212823, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7852916121482849, + "num_tokens": 542242590.0, + "step": 14217 + }, + { + "epoch": 1.8086757409998728, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.801098108291626, + "num_tokens": 542281945.0, + "step": 14218 + }, + { + "epoch": 1.8088029512784631, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8073912858963013, + "num_tokens": 542319389.0, + "step": 14219 + }, + { + "epoch": 1.8089301615570537, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7865287661552429, + "num_tokens": 542356912.0, + "step": 14220 + }, + { + "epoch": 1.8090573718356442, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7950457334518433, + "num_tokens": 542394054.0, + "step": 14221 + }, + { + "epoch": 1.8091845821142347, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8049356937408447, + "num_tokens": 542431063.0, + "step": 14222 + }, + { + "epoch": 1.8093117923928252, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7907620072364807, + "num_tokens": 542471398.0, + "step": 14223 + }, + { + "epoch": 1.8094390026714158, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7864575982093811, + "num_tokens": 542516288.0, + "step": 14224 + }, + { + "epoch": 1.8095662129500063, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7974353432655334, + "num_tokens": 542552292.0, + "step": 14225 + }, + { + "epoch": 1.8096934232285968, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.800927460193634, + "num_tokens": 542592380.0, + "step": 14226 + }, + { + "epoch": 1.8098206335071874, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7905949354171753, + "num_tokens": 542632698.0, + "step": 14227 + }, + { + "epoch": 1.8099478437857779, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.786878764629364, + "num_tokens": 542668234.0, + "step": 14228 + }, + { + "epoch": 1.8100750540643684, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.804629921913147, + "num_tokens": 542706205.0, + "step": 14229 + }, + { + "epoch": 1.810202264342959, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7947423458099365, + "num_tokens": 542737068.0, + "step": 14230 + }, + { + "epoch": 1.8103294746215495, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8117513656616211, + "num_tokens": 542769776.0, + "step": 14231 + }, + { + "epoch": 1.81045668490014, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8113123178482056, + "num_tokens": 542802440.0, + "step": 14232 + }, + { + "epoch": 1.8105838951787305, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8148448467254639, + "num_tokens": 542841158.0, + "step": 14233 + }, + { + "epoch": 1.810711105457321, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7928498983383179, + "num_tokens": 542876637.0, + "step": 14234 + }, + { + "epoch": 1.8108383157359116, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8093748688697815, + "num_tokens": 542915087.0, + "step": 14235 + }, + { + "epoch": 1.8109655260145021, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.794448971748352, + "num_tokens": 542950635.0, + "step": 14236 + }, + { + "epoch": 1.8110927362930926, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7992351055145264, + "num_tokens": 542990078.0, + "step": 14237 + }, + { + "epoch": 1.811219946571683, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7749693393707275, + "num_tokens": 543030333.0, + "step": 14238 + }, + { + "epoch": 1.8113471568502735, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.793701171875, + "num_tokens": 543070537.0, + "step": 14239 + }, + { + "epoch": 1.811474367128864, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8004426956176758, + "num_tokens": 543109137.0, + "step": 14240 + }, + { + "epoch": 1.8116015774074545, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7925703525543213, + "num_tokens": 543146616.0, + "step": 14241 + }, + { + "epoch": 1.811728787686045, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8129728436470032, + "num_tokens": 543178433.0, + "step": 14242 + }, + { + "epoch": 1.8118559979646356, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.8158666491508484, + "num_tokens": 543215086.0, + "step": 14243 + }, + { + "epoch": 1.811983208243226, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7816038727760315, + "num_tokens": 543252111.0, + "step": 14244 + }, + { + "epoch": 1.8121104185218164, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8082392811775208, + "num_tokens": 543291938.0, + "step": 14245 + }, + { + "epoch": 1.812237628800407, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8072372674942017, + "num_tokens": 543335141.0, + "step": 14246 + }, + { + "epoch": 1.8123648390789975, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.800929844379425, + "num_tokens": 543371305.0, + "step": 14247 + }, + { + "epoch": 1.812492049357588, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8108828663825989, + "num_tokens": 543407431.0, + "step": 14248 + }, + { + "epoch": 1.8126192596361785, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7982897162437439, + "num_tokens": 543450287.0, + "step": 14249 + }, + { + "epoch": 1.812746469914769, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8153434991836548, + "num_tokens": 543486521.0, + "step": 14250 + }, + { + "epoch": 1.8128736801933596, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.8010059595108032, + "num_tokens": 543522157.0, + "step": 14251 + }, + { + "epoch": 1.8130008904719501, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.807197630405426, + "num_tokens": 543560503.0, + "step": 14252 + }, + { + "epoch": 1.8131281007505406, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7831474542617798, + "num_tokens": 543598943.0, + "step": 14253 + }, + { + "epoch": 1.8132553110291312, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8039818406105042, + "num_tokens": 543640466.0, + "step": 14254 + }, + { + "epoch": 1.8133825213077217, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7978004217147827, + "num_tokens": 543677581.0, + "step": 14255 + }, + { + "epoch": 1.8135097315863122, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7909978628158569, + "num_tokens": 543711220.0, + "step": 14256 + }, + { + "epoch": 1.8136369418649028, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7916882038116455, + "num_tokens": 543750109.0, + "step": 14257 + }, + { + "epoch": 1.8137641521434933, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8060014247894287, + "num_tokens": 543788040.0, + "step": 14258 + }, + { + "epoch": 1.8138913624220838, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7978460192680359, + "num_tokens": 543821037.0, + "step": 14259 + }, + { + "epoch": 1.8140185727006743, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8114242553710938, + "num_tokens": 543856331.0, + "step": 14260 + }, + { + "epoch": 1.8141457829792649, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8111444115638733, + "num_tokens": 543892898.0, + "step": 14261 + }, + { + "epoch": 1.8142729932578552, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7986743450164795, + "num_tokens": 543932858.0, + "step": 14262 + }, + { + "epoch": 1.8144002035364457, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8079568147659302, + "num_tokens": 543968145.0, + "step": 14263 + }, + { + "epoch": 1.8145274138150362, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7976964116096497, + "num_tokens": 544000495.0, + "step": 14264 + }, + { + "epoch": 1.8146546240936268, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7963083982467651, + "num_tokens": 544041345.0, + "step": 14265 + }, + { + "epoch": 1.8147818343722173, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.805504322052002, + "num_tokens": 544080084.0, + "step": 14266 + }, + { + "epoch": 1.8149090446508078, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7879823446273804, + "num_tokens": 544115828.0, + "step": 14267 + }, + { + "epoch": 1.8150362549293981, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7958613038063049, + "num_tokens": 544153213.0, + "step": 14268 + }, + { + "epoch": 1.8151634652079887, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8038330674171448, + "num_tokens": 544192383.0, + "step": 14269 + }, + { + "epoch": 1.8152906754865792, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8046431541442871, + "num_tokens": 544232034.0, + "step": 14270 + }, + { + "epoch": 1.8154178857651697, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.782629132270813, + "num_tokens": 544266242.0, + "step": 14271 + }, + { + "epoch": 1.8155450960437602, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7914170026779175, + "num_tokens": 544304995.0, + "step": 14272 + }, + { + "epoch": 1.8156723063223508, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.799445629119873, + "num_tokens": 544341950.0, + "step": 14273 + }, + { + "epoch": 1.8157995166009413, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7950402498245239, + "num_tokens": 544379535.0, + "step": 14274 + }, + { + "epoch": 1.8159267268795318, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.795624852180481, + "num_tokens": 544416939.0, + "step": 14275 + }, + { + "epoch": 1.8160539371581224, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8132504224777222, + "num_tokens": 544461476.0, + "step": 14276 + }, + { + "epoch": 1.8161811474367129, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.79864901304245, + "num_tokens": 544505051.0, + "step": 14277 + }, + { + "epoch": 1.8163083577153034, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8004491329193115, + "num_tokens": 544550408.0, + "step": 14278 + }, + { + "epoch": 1.816435567993894, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8035435676574707, + "num_tokens": 544586645.0, + "step": 14279 + }, + { + "epoch": 1.8165627782724845, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7949692010879517, + "num_tokens": 544628087.0, + "step": 14280 + }, + { + "epoch": 1.816689988551075, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8159958124160767, + "num_tokens": 544667961.0, + "step": 14281 + }, + { + "epoch": 1.8168171988296655, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7943019866943359, + "num_tokens": 544704919.0, + "step": 14282 + }, + { + "epoch": 1.816944409108256, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8057520389556885, + "num_tokens": 544741421.0, + "step": 14283 + }, + { + "epoch": 1.8170716193868466, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8002811074256897, + "num_tokens": 544776414.0, + "step": 14284 + }, + { + "epoch": 1.817198829665437, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7924355268478394, + "num_tokens": 544814964.0, + "step": 14285 + }, + { + "epoch": 1.8173260399440276, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7868305444717407, + "num_tokens": 544853381.0, + "step": 14286 + }, + { + "epoch": 1.817453250222618, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8045142889022827, + "num_tokens": 544892474.0, + "step": 14287 + }, + { + "epoch": 1.8175804605012085, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8083745241165161, + "num_tokens": 544927667.0, + "step": 14288 + }, + { + "epoch": 1.817707670779799, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8097267746925354, + "num_tokens": 544959780.0, + "step": 14289 + }, + { + "epoch": 1.8178348810583895, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7906419038772583, + "num_tokens": 545001832.0, + "step": 14290 + }, + { + "epoch": 1.81796209133698, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.805924654006958, + "num_tokens": 545038489.0, + "step": 14291 + }, + { + "epoch": 1.8180893016155706, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7805263996124268, + "num_tokens": 545069086.0, + "step": 14292 + }, + { + "epoch": 1.818216511894161, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8019198775291443, + "num_tokens": 545104564.0, + "step": 14293 + }, + { + "epoch": 1.8183437221727514, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.811943531036377, + "num_tokens": 545149036.0, + "step": 14294 + }, + { + "epoch": 1.818470932451342, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8091185092926025, + "num_tokens": 545183825.0, + "step": 14295 + }, + { + "epoch": 1.8185981427299325, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8039268851280212, + "num_tokens": 545221274.0, + "step": 14296 + }, + { + "epoch": 1.818725353008523, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.8095122575759888, + "num_tokens": 545259242.0, + "step": 14297 + }, + { + "epoch": 1.8188525632871135, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.782475471496582, + "num_tokens": 545289380.0, + "step": 14298 + }, + { + "epoch": 1.818979773565704, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8007405996322632, + "num_tokens": 545321935.0, + "step": 14299 + }, + { + "epoch": 1.8191069838442946, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8085230588912964, + "num_tokens": 545365714.0, + "step": 14300 + }, + { + "epoch": 1.8192341941228851, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7896298170089722, + "num_tokens": 545406838.0, + "step": 14301 + }, + { + "epoch": 1.8193614044014756, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7951149940490723, + "num_tokens": 545443265.0, + "step": 14302 + }, + { + "epoch": 1.8194886146800662, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8097215890884399, + "num_tokens": 545479903.0, + "step": 14303 + }, + { + "epoch": 1.8196158249586567, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8043062090873718, + "num_tokens": 545513767.0, + "step": 14304 + }, + { + "epoch": 1.8197430352372472, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7842466235160828, + "num_tokens": 545550541.0, + "step": 14305 + }, + { + "epoch": 1.8198702455158378, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8032100796699524, + "num_tokens": 545590739.0, + "step": 14306 + }, + { + "epoch": 1.8199974557944283, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7847797870635986, + "num_tokens": 545627127.0, + "step": 14307 + }, + { + "epoch": 1.8201246660730188, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.80636066198349, + "num_tokens": 545665473.0, + "step": 14308 + }, + { + "epoch": 1.8202518763516093, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8156657218933105, + "num_tokens": 545708018.0, + "step": 14309 + }, + { + "epoch": 1.8203790866301999, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.805042564868927, + "num_tokens": 545745635.0, + "step": 14310 + }, + { + "epoch": 1.8205062969087902, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.788202166557312, + "num_tokens": 545785260.0, + "step": 14311 + }, + { + "epoch": 1.8206335071873807, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7945599555969238, + "num_tokens": 545821696.0, + "step": 14312 + }, + { + "epoch": 1.8207607174659712, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8186227083206177, + "num_tokens": 545857971.0, + "step": 14313 + }, + { + "epoch": 1.8208879277445618, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7888427972793579, + "num_tokens": 545891000.0, + "step": 14314 + }, + { + "epoch": 1.8210151380231523, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7852685451507568, + "num_tokens": 545932825.0, + "step": 14315 + }, + { + "epoch": 1.8211423483017428, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7897273898124695, + "num_tokens": 545972849.0, + "step": 14316 + }, + { + "epoch": 1.8212695585803331, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8104506731033325, + "num_tokens": 546011492.0, + "step": 14317 + }, + { + "epoch": 1.8213967688589237, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.8008236885070801, + "num_tokens": 546048461.0, + "step": 14318 + }, + { + "epoch": 1.8215239791375142, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8109715580940247, + "num_tokens": 546087113.0, + "step": 14319 + }, + { + "epoch": 1.8216511894161047, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8045692443847656, + "num_tokens": 546124413.0, + "step": 14320 + }, + { + "epoch": 1.8217783996946952, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8152481913566589, + "num_tokens": 546165763.0, + "step": 14321 + }, + { + "epoch": 1.8219056099732858, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7974656820297241, + "num_tokens": 546203218.0, + "step": 14322 + }, + { + "epoch": 1.8220328202518763, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7937732934951782, + "num_tokens": 546247610.0, + "step": 14323 + }, + { + "epoch": 1.8221600305304668, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7955269813537598, + "num_tokens": 546288952.0, + "step": 14324 + }, + { + "epoch": 1.8222872408090574, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.8115893602371216, + "num_tokens": 546322449.0, + "step": 14325 + }, + { + "epoch": 1.8224144510876479, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8067411184310913, + "num_tokens": 546362692.0, + "step": 14326 + }, + { + "epoch": 1.8225416613662384, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7855557203292847, + "num_tokens": 546398856.0, + "step": 14327 + }, + { + "epoch": 1.822668871644829, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7938738465309143, + "num_tokens": 546439633.0, + "step": 14328 + }, + { + "epoch": 1.8227960819234195, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8119131326675415, + "num_tokens": 546477282.0, + "step": 14329 + }, + { + "epoch": 1.82292329220201, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8134115934371948, + "num_tokens": 546514584.0, + "step": 14330 + }, + { + "epoch": 1.8230505024806005, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8045384883880615, + "num_tokens": 546550761.0, + "step": 14331 + }, + { + "epoch": 1.823177712759191, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7861436605453491, + "num_tokens": 546594157.0, + "step": 14332 + }, + { + "epoch": 1.8233049230377816, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.8149700164794922, + "num_tokens": 546634671.0, + "step": 14333 + }, + { + "epoch": 1.823432133316372, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7986627817153931, + "num_tokens": 546672084.0, + "step": 14334 + }, + { + "epoch": 1.8235593435949626, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7872508764266968, + "num_tokens": 546701920.0, + "step": 14335 + }, + { + "epoch": 1.823686553873553, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8061418533325195, + "num_tokens": 546735776.0, + "step": 14336 + }, + { + "epoch": 1.8238137641521435, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8078099489212036, + "num_tokens": 546772991.0, + "step": 14337 + }, + { + "epoch": 1.823940974430734, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7915866374969482, + "num_tokens": 546805310.0, + "step": 14338 + }, + { + "epoch": 1.8240681847093245, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8025410175323486, + "num_tokens": 546838240.0, + "step": 14339 + }, + { + "epoch": 1.824195394987915, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7932823896408081, + "num_tokens": 546878484.0, + "step": 14340 + }, + { + "epoch": 1.8243226052665056, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8159334063529968, + "num_tokens": 546919747.0, + "step": 14341 + }, + { + "epoch": 1.8244498155450959, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.79859858751297, + "num_tokens": 546962614.0, + "step": 14342 + }, + { + "epoch": 1.8245770258236864, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7977752089500427, + "num_tokens": 547000059.0, + "step": 14343 + }, + { + "epoch": 1.824704236102277, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7987195253372192, + "num_tokens": 547037372.0, + "step": 14344 + }, + { + "epoch": 1.8248314463808675, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7835251092910767, + "num_tokens": 547071840.0, + "step": 14345 + }, + { + "epoch": 1.824958656659458, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7663432955741882, + "num_tokens": 547108916.0, + "step": 14346 + }, + { + "epoch": 1.8250858669380485, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8235402703285217, + "num_tokens": 547146991.0, + "step": 14347 + }, + { + "epoch": 1.825213077216639, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8175133466720581, + "num_tokens": 547184093.0, + "step": 14348 + }, + { + "epoch": 1.8253402874952296, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.805509090423584, + "num_tokens": 547217570.0, + "step": 14349 + }, + { + "epoch": 1.8254674977738201, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.798090934753418, + "num_tokens": 547259498.0, + "step": 14350 + }, + { + "epoch": 1.8255947080524106, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7966826558113098, + "num_tokens": 547298419.0, + "step": 14351 + }, + { + "epoch": 1.8257219183310012, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8120944499969482, + "num_tokens": 547342984.0, + "step": 14352 + }, + { + "epoch": 1.8258491286095917, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8031984567642212, + "num_tokens": 547383162.0, + "step": 14353 + }, + { + "epoch": 1.8259763388881822, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8238195776939392, + "num_tokens": 547416967.0, + "step": 14354 + }, + { + "epoch": 1.8261035491667728, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8023502230644226, + "num_tokens": 547454695.0, + "step": 14355 + }, + { + "epoch": 1.8262307594453633, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8092951774597168, + "num_tokens": 547494258.0, + "step": 14356 + }, + { + "epoch": 1.8263579697239538, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.793999433517456, + "num_tokens": 547531404.0, + "step": 14357 + }, + { + "epoch": 1.8264851800025443, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8033926486968994, + "num_tokens": 547566553.0, + "step": 14358 + }, + { + "epoch": 1.8266123902811349, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7950791120529175, + "num_tokens": 547605505.0, + "step": 14359 + }, + { + "epoch": 1.8267396005597252, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8012385368347168, + "num_tokens": 547643146.0, + "step": 14360 + }, + { + "epoch": 1.8268668108383157, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.80350261926651, + "num_tokens": 547683373.0, + "step": 14361 + }, + { + "epoch": 1.8269940211169062, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7912389039993286, + "num_tokens": 547716762.0, + "step": 14362 + }, + { + "epoch": 1.8271212313954968, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8020309805870056, + "num_tokens": 547756907.0, + "step": 14363 + }, + { + "epoch": 1.8272484416740873, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7973231077194214, + "num_tokens": 547793545.0, + "step": 14364 + }, + { + "epoch": 1.8273756519526778, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7896906137466431, + "num_tokens": 547836845.0, + "step": 14365 + }, + { + "epoch": 1.8275028622312681, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8126287460327148, + "num_tokens": 547875849.0, + "step": 14366 + }, + { + "epoch": 1.8276300725098586, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7947516441345215, + "num_tokens": 547918672.0, + "step": 14367 + }, + { + "epoch": 1.8277572827884492, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7913240194320679, + "num_tokens": 547958681.0, + "step": 14368 + }, + { + "epoch": 1.8278844930670397, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7898063063621521, + "num_tokens": 547996940.0, + "step": 14369 + }, + { + "epoch": 1.8280117033456302, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8139581680297852, + "num_tokens": 548030188.0, + "step": 14370 + }, + { + "epoch": 1.8281389136242208, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.821017861366272, + "num_tokens": 548066891.0, + "step": 14371 + }, + { + "epoch": 1.8282661239028113, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8081227540969849, + "num_tokens": 548104664.0, + "step": 14372 + }, + { + "epoch": 1.8283933341814018, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.801038384437561, + "num_tokens": 548143938.0, + "step": 14373 + }, + { + "epoch": 1.8285205444599923, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8070193529129028, + "num_tokens": 548183213.0, + "step": 14374 + }, + { + "epoch": 1.8286477547385829, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7825024127960205, + "num_tokens": 548221304.0, + "step": 14375 + }, + { + "epoch": 1.8287749650171734, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8013527393341064, + "num_tokens": 548261123.0, + "step": 14376 + }, + { + "epoch": 1.828902175295764, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7832600474357605, + "num_tokens": 548305040.0, + "step": 14377 + }, + { + "epoch": 1.8290293855743545, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8049678802490234, + "num_tokens": 548338538.0, + "step": 14378 + }, + { + "epoch": 1.829156595852945, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8067525625228882, + "num_tokens": 548378510.0, + "step": 14379 + }, + { + "epoch": 1.8292838061315355, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7907977104187012, + "num_tokens": 548415790.0, + "step": 14380 + }, + { + "epoch": 1.829411016410126, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7880309224128723, + "num_tokens": 548447294.0, + "step": 14381 + }, + { + "epoch": 1.8295382266887166, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8131903409957886, + "num_tokens": 548484617.0, + "step": 14382 + }, + { + "epoch": 1.829665436967307, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7984633445739746, + "num_tokens": 548527552.0, + "step": 14383 + }, + { + "epoch": 1.8297926472458976, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8202683925628662, + "num_tokens": 548561322.0, + "step": 14384 + }, + { + "epoch": 1.829919857524488, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8092175722122192, + "num_tokens": 548594810.0, + "step": 14385 + }, + { + "epoch": 1.8300470678030785, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7899764776229858, + "num_tokens": 548631094.0, + "step": 14386 + }, + { + "epoch": 1.830174278081669, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7981839776039124, + "num_tokens": 548669162.0, + "step": 14387 + }, + { + "epoch": 1.8303014883602595, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8068896532058716, + "num_tokens": 548704555.0, + "step": 14388 + }, + { + "epoch": 1.83042869863885, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8156110644340515, + "num_tokens": 548741852.0, + "step": 14389 + }, + { + "epoch": 1.8305559089174406, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7917494773864746, + "num_tokens": 548788455.0, + "step": 14390 + }, + { + "epoch": 1.8306831191960309, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8098161220550537, + "num_tokens": 548826347.0, + "step": 14391 + }, + { + "epoch": 1.8308103294746214, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.797340989112854, + "num_tokens": 548865428.0, + "step": 14392 + }, + { + "epoch": 1.830937539753212, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7821304798126221, + "num_tokens": 548901671.0, + "step": 14393 + }, + { + "epoch": 1.8310647500318025, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7982597351074219, + "num_tokens": 548939234.0, + "step": 14394 + }, + { + "epoch": 1.831191960310393, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8021388649940491, + "num_tokens": 548975334.0, + "step": 14395 + }, + { + "epoch": 1.8313191705889835, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7953723669052124, + "num_tokens": 549016372.0, + "step": 14396 + }, + { + "epoch": 1.831446380867574, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8082410097122192, + "num_tokens": 549058049.0, + "step": 14397 + }, + { + "epoch": 1.8315735911461646, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7867159247398376, + "num_tokens": 549097855.0, + "step": 14398 + }, + { + "epoch": 1.831700801424755, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8211749792098999, + "num_tokens": 549130906.0, + "step": 14399 + }, + { + "epoch": 1.8318280117033456, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7818525433540344, + "num_tokens": 549176574.0, + "step": 14400 + }, + { + "epoch": 1.8319552219819362, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7964081764221191, + "num_tokens": 549213945.0, + "step": 14401 + }, + { + "epoch": 1.8320824322605267, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7927565574645996, + "num_tokens": 549255843.0, + "step": 14402 + }, + { + "epoch": 1.8322096425391172, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.804793119430542, + "num_tokens": 549292573.0, + "step": 14403 + }, + { + "epoch": 1.8323368528177078, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8170804381370544, + "num_tokens": 549335291.0, + "step": 14404 + }, + { + "epoch": 1.8324640630962983, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7893770933151245, + "num_tokens": 549381337.0, + "step": 14405 + }, + { + "epoch": 1.8325912733748888, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8086795806884766, + "num_tokens": 549418180.0, + "step": 14406 + }, + { + "epoch": 1.8327184836534793, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8101295828819275, + "num_tokens": 549461623.0, + "step": 14407 + }, + { + "epoch": 1.8328456939320699, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7998408675193787, + "num_tokens": 549500791.0, + "step": 14408 + }, + { + "epoch": 1.8329729042106602, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7980374097824097, + "num_tokens": 549533343.0, + "step": 14409 + }, + { + "epoch": 1.8331001144892507, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7891036868095398, + "num_tokens": 549572707.0, + "step": 14410 + }, + { + "epoch": 1.8332273247678412, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8055422306060791, + "num_tokens": 549609772.0, + "step": 14411 + }, + { + "epoch": 1.8333545350464318, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.788424015045166, + "num_tokens": 549648626.0, + "step": 14412 + }, + { + "epoch": 1.8334817453250223, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7890172004699707, + "num_tokens": 549693211.0, + "step": 14413 + }, + { + "epoch": 1.8336089556036128, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7945241928100586, + "num_tokens": 549729463.0, + "step": 14414 + }, + { + "epoch": 1.8337361658822031, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8077123165130615, + "num_tokens": 549769139.0, + "step": 14415 + }, + { + "epoch": 1.8338633761607936, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8030781745910645, + "num_tokens": 549807067.0, + "step": 14416 + }, + { + "epoch": 1.8339905864393842, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7947417497634888, + "num_tokens": 549842689.0, + "step": 14417 + }, + { + "epoch": 1.8341177967179747, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8196694850921631, + "num_tokens": 549876108.0, + "step": 14418 + }, + { + "epoch": 1.8342450069965652, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.8186262249946594, + "num_tokens": 549909656.0, + "step": 14419 + }, + { + "epoch": 1.8343722172751558, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7883573174476624, + "num_tokens": 549949156.0, + "step": 14420 + }, + { + "epoch": 1.8344994275537463, + "grad_norm": 65.5, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8066677451133728, + "num_tokens": 549990256.0, + "step": 14421 + }, + { + "epoch": 1.8346266378323368, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8052356243133545, + "num_tokens": 550023557.0, + "step": 14422 + }, + { + "epoch": 1.8347538481109273, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8019710183143616, + "num_tokens": 550069402.0, + "step": 14423 + }, + { + "epoch": 1.8348810583895179, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8137507438659668, + "num_tokens": 550110860.0, + "step": 14424 + }, + { + "epoch": 1.8350082686681084, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7905032634735107, + "num_tokens": 550151736.0, + "step": 14425 + }, + { + "epoch": 1.835135478946699, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7966958284378052, + "num_tokens": 550187731.0, + "step": 14426 + }, + { + "epoch": 1.8352626892252895, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7989016175270081, + "num_tokens": 550220545.0, + "step": 14427 + }, + { + "epoch": 1.83538989950388, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8042840957641602, + "num_tokens": 550254755.0, + "step": 14428 + }, + { + "epoch": 1.8355171097824705, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8068901300430298, + "num_tokens": 550295191.0, + "step": 14429 + }, + { + "epoch": 1.835644320061061, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7907350063323975, + "num_tokens": 550332984.0, + "step": 14430 + }, + { + "epoch": 1.8357715303396516, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8098637461662292, + "num_tokens": 550377470.0, + "step": 14431 + }, + { + "epoch": 1.835898740618242, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8053096532821655, + "num_tokens": 550416123.0, + "step": 14432 + }, + { + "epoch": 1.8360259508968326, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8078700304031372, + "num_tokens": 550458062.0, + "step": 14433 + }, + { + "epoch": 1.836153161175423, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7905678749084473, + "num_tokens": 550496485.0, + "step": 14434 + }, + { + "epoch": 1.8362803714540135, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.8028743863105774, + "num_tokens": 550536500.0, + "step": 14435 + }, + { + "epoch": 1.836407581732604, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7957850098609924, + "num_tokens": 550575611.0, + "step": 14436 + }, + { + "epoch": 1.8365347920111945, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7907397747039795, + "num_tokens": 550616431.0, + "step": 14437 + }, + { + "epoch": 1.836662002289785, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.806005597114563, + "num_tokens": 550653626.0, + "step": 14438 + }, + { + "epoch": 1.8367892125683756, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7936574220657349, + "num_tokens": 550690899.0, + "step": 14439 + }, + { + "epoch": 1.8369164228469659, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8032674789428711, + "num_tokens": 550734408.0, + "step": 14440 + }, + { + "epoch": 1.8370436331255564, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7878026366233826, + "num_tokens": 550767888.0, + "step": 14441 + }, + { + "epoch": 1.837170843404147, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7992593050003052, + "num_tokens": 550807192.0, + "step": 14442 + }, + { + "epoch": 1.8372980536827375, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8112789988517761, + "num_tokens": 550848006.0, + "step": 14443 + }, + { + "epoch": 1.837425263961328, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7992261648178101, + "num_tokens": 550888350.0, + "step": 14444 + }, + { + "epoch": 1.8375524742399185, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8054317235946655, + "num_tokens": 550925389.0, + "step": 14445 + }, + { + "epoch": 1.837679684518509, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8066912889480591, + "num_tokens": 550962610.0, + "step": 14446 + }, + { + "epoch": 1.8378068947970996, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.8009618520736694, + "num_tokens": 551001018.0, + "step": 14447 + }, + { + "epoch": 1.83793410507569, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8020881414413452, + "num_tokens": 551038505.0, + "step": 14448 + }, + { + "epoch": 1.8380613153542806, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.8124549388885498, + "num_tokens": 551072235.0, + "step": 14449 + }, + { + "epoch": 1.8381885256328712, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.816392183303833, + "num_tokens": 551110129.0, + "step": 14450 + }, + { + "epoch": 1.8383157359114617, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8039307594299316, + "num_tokens": 551152380.0, + "step": 14451 + }, + { + "epoch": 1.8384429461900522, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8184627294540405, + "num_tokens": 551195942.0, + "step": 14452 + }, + { + "epoch": 1.8385701564686427, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7992838025093079, + "num_tokens": 551238502.0, + "step": 14453 + }, + { + "epoch": 1.8386973667472333, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6097, + "mean_token_accuracy": 0.8150379657745361, + "num_tokens": 551278091.0, + "step": 14454 + }, + { + "epoch": 1.8388245770258238, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7785364389419556, + "num_tokens": 551314940.0, + "step": 14455 + }, + { + "epoch": 1.8389517873044143, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.784909188747406, + "num_tokens": 551356427.0, + "step": 14456 + }, + { + "epoch": 1.8390789975830049, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8071883916854858, + "num_tokens": 551400436.0, + "step": 14457 + }, + { + "epoch": 1.8392062078615952, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7977598905563354, + "num_tokens": 551438335.0, + "step": 14458 + }, + { + "epoch": 1.8393334181401857, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8033762574195862, + "num_tokens": 551471214.0, + "step": 14459 + }, + { + "epoch": 1.8394606284187762, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7837461233139038, + "num_tokens": 551504071.0, + "step": 14460 + }, + { + "epoch": 1.8395878386973668, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7943683862686157, + "num_tokens": 551538456.0, + "step": 14461 + }, + { + "epoch": 1.8397150489759573, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8096393346786499, + "num_tokens": 551575334.0, + "step": 14462 + }, + { + "epoch": 1.8398422592545478, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8137143850326538, + "num_tokens": 551617127.0, + "step": 14463 + }, + { + "epoch": 1.8399694695331381, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8089367151260376, + "num_tokens": 551655548.0, + "step": 14464 + }, + { + "epoch": 1.8400966798117286, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7994783520698547, + "num_tokens": 551691807.0, + "step": 14465 + }, + { + "epoch": 1.8402238900903192, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8090649247169495, + "num_tokens": 551729805.0, + "step": 14466 + }, + { + "epoch": 1.8403511003689097, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7798629999160767, + "num_tokens": 551775220.0, + "step": 14467 + }, + { + "epoch": 1.8404783106475002, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7910008430480957, + "num_tokens": 551816242.0, + "step": 14468 + }, + { + "epoch": 1.8406055209260908, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7943713665008545, + "num_tokens": 551854929.0, + "step": 14469 + }, + { + "epoch": 1.8407327312046813, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8042256832122803, + "num_tokens": 551898877.0, + "step": 14470 + }, + { + "epoch": 1.8408599414832718, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8170716166496277, + "num_tokens": 551934252.0, + "step": 14471 + }, + { + "epoch": 1.8409871517618623, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8151426911354065, + "num_tokens": 551974779.0, + "step": 14472 + }, + { + "epoch": 1.8411143620404529, + "grad_norm": 184.0, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7837894558906555, + "num_tokens": 552011627.0, + "step": 14473 + }, + { + "epoch": 1.8412415723190434, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7733754515647888, + "num_tokens": 552045333.0, + "step": 14474 + }, + { + "epoch": 1.841368782597634, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7881156206130981, + "num_tokens": 552079974.0, + "step": 14475 + }, + { + "epoch": 1.8414959928762245, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8077566027641296, + "num_tokens": 552117507.0, + "step": 14476 + }, + { + "epoch": 1.841623203154815, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.801668643951416, + "num_tokens": 552156009.0, + "step": 14477 + }, + { + "epoch": 1.8417504134334055, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8075225353240967, + "num_tokens": 552189092.0, + "step": 14478 + }, + { + "epoch": 1.841877623711996, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.799250602722168, + "num_tokens": 552228305.0, + "step": 14479 + }, + { + "epoch": 1.8420048339905866, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8034611344337463, + "num_tokens": 552269402.0, + "step": 14480 + }, + { + "epoch": 1.842132044269177, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7994848489761353, + "num_tokens": 552304870.0, + "step": 14481 + }, + { + "epoch": 1.8422592545477676, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7899924516677856, + "num_tokens": 552336802.0, + "step": 14482 + }, + { + "epoch": 1.842386464826358, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.813514769077301, + "num_tokens": 552376124.0, + "step": 14483 + }, + { + "epoch": 1.8425136751049485, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7903794050216675, + "num_tokens": 552412285.0, + "step": 14484 + }, + { + "epoch": 1.842640885383539, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7953670024871826, + "num_tokens": 552447527.0, + "step": 14485 + }, + { + "epoch": 1.8427680956621295, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8001214265823364, + "num_tokens": 552486569.0, + "step": 14486 + }, + { + "epoch": 1.84289530594072, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7658246755599976, + "num_tokens": 552529158.0, + "step": 14487 + }, + { + "epoch": 1.8430225162193106, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.796189546585083, + "num_tokens": 552572183.0, + "step": 14488 + }, + { + "epoch": 1.8431497264979009, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8012365102767944, + "num_tokens": 552609785.0, + "step": 14489 + }, + { + "epoch": 1.8432769367764914, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.804526686668396, + "num_tokens": 552641243.0, + "step": 14490 + }, + { + "epoch": 1.843404147055082, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7951140403747559, + "num_tokens": 552677595.0, + "step": 14491 + }, + { + "epoch": 1.8435313573336725, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8012503981590271, + "num_tokens": 552715492.0, + "step": 14492 + }, + { + "epoch": 1.843658567612263, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.80336594581604, + "num_tokens": 552757410.0, + "step": 14493 + }, + { + "epoch": 1.8437857778908535, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8219723701477051, + "num_tokens": 552793794.0, + "step": 14494 + }, + { + "epoch": 1.843912988169444, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7857893109321594, + "num_tokens": 552836010.0, + "step": 14495 + }, + { + "epoch": 1.8440401984480346, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8016867637634277, + "num_tokens": 552874136.0, + "step": 14496 + }, + { + "epoch": 1.844167408726625, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7971608638763428, + "num_tokens": 552912314.0, + "step": 14497 + }, + { + "epoch": 1.8442946190052156, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7859789133071899, + "num_tokens": 552953998.0, + "step": 14498 + }, + { + "epoch": 1.8444218292838062, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.8170586824417114, + "num_tokens": 552987310.0, + "step": 14499 + }, + { + "epoch": 1.8445490395623967, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7945838570594788, + "num_tokens": 553019830.0, + "step": 14500 + }, + { + "epoch": 1.8446762498409872, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8016512989997864, + "num_tokens": 553066695.0, + "step": 14501 + }, + { + "epoch": 1.8448034601195777, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7897334098815918, + "num_tokens": 553098943.0, + "step": 14502 + }, + { + "epoch": 1.8449306703981683, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8042318820953369, + "num_tokens": 553136106.0, + "step": 14503 + }, + { + "epoch": 1.8450578806767588, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.807462751865387, + "num_tokens": 553170014.0, + "step": 14504 + }, + { + "epoch": 1.8451850909553493, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7940871715545654, + "num_tokens": 553213297.0, + "step": 14505 + }, + { + "epoch": 1.8453123012339399, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7993768453598022, + "num_tokens": 553248559.0, + "step": 14506 + }, + { + "epoch": 1.8454395115125302, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8056556582450867, + "num_tokens": 553289140.0, + "step": 14507 + }, + { + "epoch": 1.8455667217911207, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7951139211654663, + "num_tokens": 553330619.0, + "step": 14508 + }, + { + "epoch": 1.8456939320697112, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8144835233688354, + "num_tokens": 553371295.0, + "step": 14509 + }, + { + "epoch": 1.8458211423483017, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7991953492164612, + "num_tokens": 553410237.0, + "step": 14510 + }, + { + "epoch": 1.8459483526268923, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8027827739715576, + "num_tokens": 553446727.0, + "step": 14511 + }, + { + "epoch": 1.8460755629054828, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8050336837768555, + "num_tokens": 553488626.0, + "step": 14512 + }, + { + "epoch": 1.846202773184073, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7919038534164429, + "num_tokens": 553528294.0, + "step": 14513 + }, + { + "epoch": 1.8463299834626636, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.810237467288971, + "num_tokens": 553563738.0, + "step": 14514 + }, + { + "epoch": 1.8464571937412542, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7629395723342896, + "num_tokens": 553603351.0, + "step": 14515 + }, + { + "epoch": 1.8465844040198447, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7966669797897339, + "num_tokens": 553643655.0, + "step": 14516 + }, + { + "epoch": 1.8467116142984352, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.8145596385002136, + "num_tokens": 553677944.0, + "step": 14517 + }, + { + "epoch": 1.8468388245770258, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7894963026046753, + "num_tokens": 553712120.0, + "step": 14518 + }, + { + "epoch": 1.8469660348556163, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8125284910202026, + "num_tokens": 553751442.0, + "step": 14519 + }, + { + "epoch": 1.8470932451342068, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8059878349304199, + "num_tokens": 553782924.0, + "step": 14520 + }, + { + "epoch": 1.8472204554127973, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7889759540557861, + "num_tokens": 553818861.0, + "step": 14521 + }, + { + "epoch": 1.8473476656913879, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7984952926635742, + "num_tokens": 553855593.0, + "step": 14522 + }, + { + "epoch": 1.8474748759699784, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8019078969955444, + "num_tokens": 553896028.0, + "step": 14523 + }, + { + "epoch": 1.847602086248569, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8084211349487305, + "num_tokens": 553935212.0, + "step": 14524 + }, + { + "epoch": 1.8477292965271594, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7976188063621521, + "num_tokens": 553971231.0, + "step": 14525 + }, + { + "epoch": 1.84785650680575, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.8099074959754944, + "num_tokens": 554006782.0, + "step": 14526 + }, + { + "epoch": 1.8479837170843405, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7951943874359131, + "num_tokens": 554043427.0, + "step": 14527 + }, + { + "epoch": 1.848110927362931, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7893732786178589, + "num_tokens": 554084187.0, + "step": 14528 + }, + { + "epoch": 1.8482381376415216, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8024797439575195, + "num_tokens": 554122186.0, + "step": 14529 + }, + { + "epoch": 1.848365347920112, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.797978937625885, + "num_tokens": 554161388.0, + "step": 14530 + }, + { + "epoch": 1.8484925581987026, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7926121950149536, + "num_tokens": 554199837.0, + "step": 14531 + }, + { + "epoch": 1.848619768477293, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7886426448822021, + "num_tokens": 554239567.0, + "step": 14532 + }, + { + "epoch": 1.8487469787558835, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8046159744262695, + "num_tokens": 554277040.0, + "step": 14533 + }, + { + "epoch": 1.848874189034474, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.803761899471283, + "num_tokens": 554318994.0, + "step": 14534 + }, + { + "epoch": 1.8490013993130645, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7953084707260132, + "num_tokens": 554355334.0, + "step": 14535 + }, + { + "epoch": 1.849128609591655, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.790973424911499, + "num_tokens": 554396504.0, + "step": 14536 + }, + { + "epoch": 1.8492558198702456, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8068897128105164, + "num_tokens": 554429772.0, + "step": 14537 + }, + { + "epoch": 1.8493830301488359, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8038715124130249, + "num_tokens": 554470102.0, + "step": 14538 + }, + { + "epoch": 1.8495102404274264, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7898809313774109, + "num_tokens": 554511632.0, + "step": 14539 + }, + { + "epoch": 1.849637450706017, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.796314001083374, + "num_tokens": 554553028.0, + "step": 14540 + }, + { + "epoch": 1.8497646609846075, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7949352860450745, + "num_tokens": 554585109.0, + "step": 14541 + }, + { + "epoch": 1.849891871263198, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7987838983535767, + "num_tokens": 554621141.0, + "step": 14542 + }, + { + "epoch": 1.8500190815417885, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.8003209829330444, + "num_tokens": 554660838.0, + "step": 14543 + }, + { + "epoch": 1.850146291820379, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8003323078155518, + "num_tokens": 554696338.0, + "step": 14544 + }, + { + "epoch": 1.8502735020989696, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8120619058609009, + "num_tokens": 554729782.0, + "step": 14545 + }, + { + "epoch": 1.85040071237756, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8171742558479309, + "num_tokens": 554767566.0, + "step": 14546 + }, + { + "epoch": 1.8505279226561506, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8074625730514526, + "num_tokens": 554806063.0, + "step": 14547 + }, + { + "epoch": 1.8506551329347412, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7960413098335266, + "num_tokens": 554839418.0, + "step": 14548 + }, + { + "epoch": 1.8507823432133317, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7966665625572205, + "num_tokens": 554881888.0, + "step": 14549 + }, + { + "epoch": 1.8509095534919222, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.5817, + "mean_token_accuracy": 0.8237590789794922, + "num_tokens": 554921561.0, + "step": 14550 + }, + { + "epoch": 1.8510367637705127, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7760117053985596, + "num_tokens": 554954380.0, + "step": 14551 + }, + { + "epoch": 1.8511639740491033, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7993966341018677, + "num_tokens": 554994781.0, + "step": 14552 + }, + { + "epoch": 1.8512911843276938, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.803413987159729, + "num_tokens": 555024938.0, + "step": 14553 + }, + { + "epoch": 1.8514183946062843, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7869617342948914, + "num_tokens": 555059785.0, + "step": 14554 + }, + { + "epoch": 1.8515456048848749, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.77704918384552, + "num_tokens": 555097502.0, + "step": 14555 + }, + { + "epoch": 1.8516728151634652, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.80194091796875, + "num_tokens": 555139976.0, + "step": 14556 + }, + { + "epoch": 1.8518000254420557, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8093079924583435, + "num_tokens": 555181250.0, + "step": 14557 + }, + { + "epoch": 1.8519272357206462, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7799960374832153, + "num_tokens": 555214927.0, + "step": 14558 + }, + { + "epoch": 1.8520544459992367, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7936598062515259, + "num_tokens": 555255795.0, + "step": 14559 + }, + { + "epoch": 1.8521816562778273, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8024035692214966, + "num_tokens": 555294177.0, + "step": 14560 + }, + { + "epoch": 1.8523088665564178, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.814653754234314, + "num_tokens": 555331548.0, + "step": 14561 + }, + { + "epoch": 1.852436076835008, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8118428587913513, + "num_tokens": 555368426.0, + "step": 14562 + }, + { + "epoch": 1.8525632871135986, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8190144300460815, + "num_tokens": 555411594.0, + "step": 14563 + }, + { + "epoch": 1.8526904973921892, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7931993007659912, + "num_tokens": 555450412.0, + "step": 14564 + }, + { + "epoch": 1.8528177076707797, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.8163251876831055, + "num_tokens": 555489067.0, + "step": 14565 + }, + { + "epoch": 1.8529449179493702, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8069295883178711, + "num_tokens": 555524552.0, + "step": 14566 + }, + { + "epoch": 1.8530721282279607, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7735881805419922, + "num_tokens": 555563766.0, + "step": 14567 + }, + { + "epoch": 1.8531993385065513, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8110184669494629, + "num_tokens": 555601410.0, + "step": 14568 + }, + { + "epoch": 1.8533265487851418, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.804608166217804, + "num_tokens": 555641346.0, + "step": 14569 + }, + { + "epoch": 1.8534537590637323, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8066732287406921, + "num_tokens": 555676537.0, + "step": 14570 + }, + { + "epoch": 1.8535809693423229, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7888526916503906, + "num_tokens": 555721634.0, + "step": 14571 + }, + { + "epoch": 1.8537081796209134, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8027899861335754, + "num_tokens": 555765447.0, + "step": 14572 + }, + { + "epoch": 1.853835389899504, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7961449027061462, + "num_tokens": 555798598.0, + "step": 14573 + }, + { + "epoch": 1.8539626001780944, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8026188611984253, + "num_tokens": 555829465.0, + "step": 14574 + }, + { + "epoch": 1.854089810456685, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8055602312088013, + "num_tokens": 555866577.0, + "step": 14575 + }, + { + "epoch": 1.8542170207352755, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7954084873199463, + "num_tokens": 555900472.0, + "step": 14576 + }, + { + "epoch": 1.854344231013866, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7863463759422302, + "num_tokens": 555936356.0, + "step": 14577 + }, + { + "epoch": 1.8544714412924566, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7913150787353516, + "num_tokens": 555975338.0, + "step": 14578 + }, + { + "epoch": 1.854598651571047, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.793546736240387, + "num_tokens": 556019651.0, + "step": 14579 + }, + { + "epoch": 1.8547258618496376, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7941475510597229, + "num_tokens": 556054530.0, + "step": 14580 + }, + { + "epoch": 1.854853072128228, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7956100106239319, + "num_tokens": 556091985.0, + "step": 14581 + }, + { + "epoch": 1.8549802824068184, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8053398132324219, + "num_tokens": 556128777.0, + "step": 14582 + }, + { + "epoch": 1.855107492685409, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8192180395126343, + "num_tokens": 556164286.0, + "step": 14583 + }, + { + "epoch": 1.8552347029639995, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.805260419845581, + "num_tokens": 556199141.0, + "step": 14584 + }, + { + "epoch": 1.85536191324259, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8057059049606323, + "num_tokens": 556237104.0, + "step": 14585 + }, + { + "epoch": 1.8554891235211806, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7934452295303345, + "num_tokens": 556275343.0, + "step": 14586 + }, + { + "epoch": 1.8556163337997709, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.8152115941047668, + "num_tokens": 556316539.0, + "step": 14587 + }, + { + "epoch": 1.8557435440783614, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8101286888122559, + "num_tokens": 556351101.0, + "step": 14588 + }, + { + "epoch": 1.855870754356952, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.807268500328064, + "num_tokens": 556382458.0, + "step": 14589 + }, + { + "epoch": 1.8559979646355425, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8000108599662781, + "num_tokens": 556416980.0, + "step": 14590 + }, + { + "epoch": 1.856125174914133, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8061181306838989, + "num_tokens": 556451868.0, + "step": 14591 + }, + { + "epoch": 1.8562523851927235, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7910988926887512, + "num_tokens": 556485682.0, + "step": 14592 + }, + { + "epoch": 1.856379595471314, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8132807016372681, + "num_tokens": 556524857.0, + "step": 14593 + }, + { + "epoch": 1.8565068057499046, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8073676228523254, + "num_tokens": 556565763.0, + "step": 14594 + }, + { + "epoch": 1.856634016028495, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7941777110099792, + "num_tokens": 556606202.0, + "step": 14595 + }, + { + "epoch": 1.8567612263070856, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8129139542579651, + "num_tokens": 556646906.0, + "step": 14596 + }, + { + "epoch": 1.8568884365856761, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7997138500213623, + "num_tokens": 556684249.0, + "step": 14597 + }, + { + "epoch": 1.8570156468642667, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7901078462600708, + "num_tokens": 556726840.0, + "step": 14598 + }, + { + "epoch": 1.8571428571428572, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8020417094230652, + "num_tokens": 556764822.0, + "step": 14599 + }, + { + "epoch": 1.8572700674214477, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7944344282150269, + "num_tokens": 556800884.0, + "step": 14600 + }, + { + "epoch": 1.8573972777000383, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8064364194869995, + "num_tokens": 556841233.0, + "step": 14601 + }, + { + "epoch": 1.8575244879786288, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7937416434288025, + "num_tokens": 556880995.0, + "step": 14602 + }, + { + "epoch": 1.8576516982572193, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8050174117088318, + "num_tokens": 556914367.0, + "step": 14603 + }, + { + "epoch": 1.8577789085358098, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.780544638633728, + "num_tokens": 556955944.0, + "step": 14604 + }, + { + "epoch": 1.8579061188144002, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7830418944358826, + "num_tokens": 556989851.0, + "step": 14605 + }, + { + "epoch": 1.8580333290929907, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7953889966011047, + "num_tokens": 557033662.0, + "step": 14606 + }, + { + "epoch": 1.8581605393715812, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.794329047203064, + "num_tokens": 557074869.0, + "step": 14607 + }, + { + "epoch": 1.8582877496501717, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.784082293510437, + "num_tokens": 557110164.0, + "step": 14608 + }, + { + "epoch": 1.8584149599287623, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.8015798926353455, + "num_tokens": 557149309.0, + "step": 14609 + }, + { + "epoch": 1.8585421702073528, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7990930080413818, + "num_tokens": 557180510.0, + "step": 14610 + }, + { + "epoch": 1.858669380485943, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8061538338661194, + "num_tokens": 557218062.0, + "step": 14611 + }, + { + "epoch": 1.8587965907645336, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7936743497848511, + "num_tokens": 557258963.0, + "step": 14612 + }, + { + "epoch": 1.8589238010431242, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8038002252578735, + "num_tokens": 557296046.0, + "step": 14613 + }, + { + "epoch": 1.8590510113217147, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7975191473960876, + "num_tokens": 557328395.0, + "step": 14614 + }, + { + "epoch": 1.8591782216003052, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8073070645332336, + "num_tokens": 557363888.0, + "step": 14615 + }, + { + "epoch": 1.8593054318788957, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7913215756416321, + "num_tokens": 557402944.0, + "step": 14616 + }, + { + "epoch": 1.8594326421574863, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8011720776557922, + "num_tokens": 557442247.0, + "step": 14617 + }, + { + "epoch": 1.8595598524360768, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7926172018051147, + "num_tokens": 557483580.0, + "step": 14618 + }, + { + "epoch": 1.8596870627146673, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8055737018585205, + "num_tokens": 557520134.0, + "step": 14619 + }, + { + "epoch": 1.8598142729932579, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.8009063005447388, + "num_tokens": 557565726.0, + "step": 14620 + }, + { + "epoch": 1.8599414832718484, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7970921397209167, + "num_tokens": 557602871.0, + "step": 14621 + }, + { + "epoch": 1.860068693550439, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7966023087501526, + "num_tokens": 557640873.0, + "step": 14622 + }, + { + "epoch": 1.8601959038290294, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.79278165102005, + "num_tokens": 557684233.0, + "step": 14623 + }, + { + "epoch": 1.86032311410762, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8035486936569214, + "num_tokens": 557716895.0, + "step": 14624 + }, + { + "epoch": 1.8604503243862105, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7972441911697388, + "num_tokens": 557758197.0, + "step": 14625 + }, + { + "epoch": 1.860577534664801, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7742159366607666, + "num_tokens": 557794196.0, + "step": 14626 + }, + { + "epoch": 1.8607047449433916, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.779340386390686, + "num_tokens": 557833856.0, + "step": 14627 + }, + { + "epoch": 1.860831955221982, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7819653153419495, + "num_tokens": 557866304.0, + "step": 14628 + }, + { + "epoch": 1.8609591655005726, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8161238431930542, + "num_tokens": 557907065.0, + "step": 14629 + }, + { + "epoch": 1.861086375779163, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7983136177062988, + "num_tokens": 557947326.0, + "step": 14630 + }, + { + "epoch": 1.8612135860577534, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7971773147583008, + "num_tokens": 557990059.0, + "step": 14631 + }, + { + "epoch": 1.861340796336344, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8055628538131714, + "num_tokens": 558028406.0, + "step": 14632 + }, + { + "epoch": 1.8614680066149345, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7995375394821167, + "num_tokens": 558070239.0, + "step": 14633 + }, + { + "epoch": 1.861595216893525, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8029758930206299, + "num_tokens": 558107685.0, + "step": 14634 + }, + { + "epoch": 1.8617224271721156, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7966176271438599, + "num_tokens": 558143185.0, + "step": 14635 + }, + { + "epoch": 1.8618496374507059, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8051193952560425, + "num_tokens": 558176207.0, + "step": 14636 + }, + { + "epoch": 1.8619768477292964, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8133100271224976, + "num_tokens": 558212215.0, + "step": 14637 + }, + { + "epoch": 1.862104058007887, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.804885983467102, + "num_tokens": 558248471.0, + "step": 14638 + }, + { + "epoch": 1.8622312682864774, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7884683609008789, + "num_tokens": 558285211.0, + "step": 14639 + }, + { + "epoch": 1.862358478565068, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7931427955627441, + "num_tokens": 558314434.0, + "step": 14640 + }, + { + "epoch": 1.8624856888436585, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.811225414276123, + "num_tokens": 558353710.0, + "step": 14641 + }, + { + "epoch": 1.862612899122249, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7906970977783203, + "num_tokens": 558390476.0, + "step": 14642 + }, + { + "epoch": 1.8627401094008396, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8076090812683105, + "num_tokens": 558424917.0, + "step": 14643 + }, + { + "epoch": 1.86286731967943, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7940976023674011, + "num_tokens": 558459831.0, + "step": 14644 + }, + { + "epoch": 1.8629945299580206, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8176193237304688, + "num_tokens": 558498600.0, + "step": 14645 + }, + { + "epoch": 1.8631217402366111, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7897539734840393, + "num_tokens": 558537465.0, + "step": 14646 + }, + { + "epoch": 1.8632489505152017, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.8178058862686157, + "num_tokens": 558575357.0, + "step": 14647 + }, + { + "epoch": 1.8633761607937922, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7973552942276001, + "num_tokens": 558619422.0, + "step": 14648 + }, + { + "epoch": 1.8635033710723827, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7925825119018555, + "num_tokens": 558657677.0, + "step": 14649 + }, + { + "epoch": 1.8636305813509733, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8116463422775269, + "num_tokens": 558692912.0, + "step": 14650 + }, + { + "epoch": 1.8637577916295638, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8142794370651245, + "num_tokens": 558733470.0, + "step": 14651 + }, + { + "epoch": 1.8638850019081543, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7968036532402039, + "num_tokens": 558771899.0, + "step": 14652 + }, + { + "epoch": 1.8640122121867448, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8065581321716309, + "num_tokens": 558813002.0, + "step": 14653 + }, + { + "epoch": 1.8641394224653351, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.784113883972168, + "num_tokens": 558850827.0, + "step": 14654 + }, + { + "epoch": 1.8642666327439257, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7950373888015747, + "num_tokens": 558889003.0, + "step": 14655 + }, + { + "epoch": 1.8643938430225162, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8056037425994873, + "num_tokens": 558925849.0, + "step": 14656 + }, + { + "epoch": 1.8645210533011067, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8023208975791931, + "num_tokens": 558967471.0, + "step": 14657 + }, + { + "epoch": 1.8646482635796973, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8011865615844727, + "num_tokens": 559001938.0, + "step": 14658 + }, + { + "epoch": 1.8647754738582878, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8043313026428223, + "num_tokens": 559040515.0, + "step": 14659 + }, + { + "epoch": 1.864902684136878, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8150933980941772, + "num_tokens": 559077702.0, + "step": 14660 + }, + { + "epoch": 1.8650298944154686, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.792212963104248, + "num_tokens": 559114741.0, + "step": 14661 + }, + { + "epoch": 1.8651571046940592, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7820385098457336, + "num_tokens": 559155844.0, + "step": 14662 + }, + { + "epoch": 1.8652843149726497, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8003617525100708, + "num_tokens": 559194964.0, + "step": 14663 + }, + { + "epoch": 1.8654115252512402, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8094161748886108, + "num_tokens": 559230083.0, + "step": 14664 + }, + { + "epoch": 1.8655387355298307, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8147337436676025, + "num_tokens": 559268397.0, + "step": 14665 + }, + { + "epoch": 1.8656659458084213, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7933905124664307, + "num_tokens": 559302024.0, + "step": 14666 + }, + { + "epoch": 1.8657931560870118, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8038486242294312, + "num_tokens": 559343192.0, + "step": 14667 + }, + { + "epoch": 1.8659203663656023, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8238162994384766, + "num_tokens": 559380616.0, + "step": 14668 + }, + { + "epoch": 1.8660475766441929, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8115065097808838, + "num_tokens": 559413621.0, + "step": 14669 + }, + { + "epoch": 1.8661747869227834, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8129197359085083, + "num_tokens": 559447274.0, + "step": 14670 + }, + { + "epoch": 1.866301997201374, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.811892032623291, + "num_tokens": 559484306.0, + "step": 14671 + }, + { + "epoch": 1.8664292074799644, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8114586472511292, + "num_tokens": 559527126.0, + "step": 14672 + }, + { + "epoch": 1.866556417758555, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8043352365493774, + "num_tokens": 559568342.0, + "step": 14673 + }, + { + "epoch": 1.8666836280371455, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7998417615890503, + "num_tokens": 559609597.0, + "step": 14674 + }, + { + "epoch": 1.866810838315736, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7969852685928345, + "num_tokens": 559646392.0, + "step": 14675 + }, + { + "epoch": 1.8669380485943265, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8016374707221985, + "num_tokens": 559684836.0, + "step": 14676 + }, + { + "epoch": 1.867065258872917, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8064650297164917, + "num_tokens": 559727793.0, + "step": 14677 + }, + { + "epoch": 1.8671924691515076, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7862284183502197, + "num_tokens": 559768154.0, + "step": 14678 + }, + { + "epoch": 1.867319679430098, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.795948326587677, + "num_tokens": 559798732.0, + "step": 14679 + }, + { + "epoch": 1.8674468897086884, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8015415668487549, + "num_tokens": 559842796.0, + "step": 14680 + }, + { + "epoch": 1.867574099987279, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7983470559120178, + "num_tokens": 559884198.0, + "step": 14681 + }, + { + "epoch": 1.8677013102658695, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.780174970626831, + "num_tokens": 559916749.0, + "step": 14682 + }, + { + "epoch": 1.86782852054446, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7966240644454956, + "num_tokens": 559958430.0, + "step": 14683 + }, + { + "epoch": 1.8679557308230506, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8181617259979248, + "num_tokens": 559995720.0, + "step": 14684 + }, + { + "epoch": 1.8680829411016409, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8049296736717224, + "num_tokens": 560033304.0, + "step": 14685 + }, + { + "epoch": 1.8682101513802314, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8003543019294739, + "num_tokens": 560069879.0, + "step": 14686 + }, + { + "epoch": 1.868337361658822, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7993502616882324, + "num_tokens": 560106853.0, + "step": 14687 + }, + { + "epoch": 1.8684645719374124, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7991342544555664, + "num_tokens": 560144225.0, + "step": 14688 + }, + { + "epoch": 1.868591782216003, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.801114022731781, + "num_tokens": 560186520.0, + "step": 14689 + }, + { + "epoch": 1.8687189924945935, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8054074048995972, + "num_tokens": 560225458.0, + "step": 14690 + }, + { + "epoch": 1.868846202773184, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8087583780288696, + "num_tokens": 560263401.0, + "step": 14691 + }, + { + "epoch": 1.8689734130517746, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8008258938789368, + "num_tokens": 560305026.0, + "step": 14692 + }, + { + "epoch": 1.869100623330365, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7649057507514954, + "num_tokens": 560354823.0, + "step": 14693 + }, + { + "epoch": 1.8692278336089556, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8208578824996948, + "num_tokens": 560391888.0, + "step": 14694 + }, + { + "epoch": 1.8693550438875461, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.807153046131134, + "num_tokens": 560425350.0, + "step": 14695 + }, + { + "epoch": 1.8694822541661367, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8101627826690674, + "num_tokens": 560458521.0, + "step": 14696 + }, + { + "epoch": 1.8696094644447272, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7981253862380981, + "num_tokens": 560506378.0, + "step": 14697 + }, + { + "epoch": 1.8697366747233177, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8214601874351501, + "num_tokens": 560544055.0, + "step": 14698 + }, + { + "epoch": 1.8698638850019083, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8052841424942017, + "num_tokens": 560577240.0, + "step": 14699 + }, + { + "epoch": 1.8699910952804988, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7960599660873413, + "num_tokens": 560615662.0, + "step": 14700 + }, + { + "epoch": 1.8701183055590893, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7960691452026367, + "num_tokens": 560658386.0, + "step": 14701 + }, + { + "epoch": 1.8702455158376798, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7977123260498047, + "num_tokens": 560696366.0, + "step": 14702 + }, + { + "epoch": 1.8703727261162701, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7873456478118896, + "num_tokens": 560736419.0, + "step": 14703 + }, + { + "epoch": 1.8704999363948607, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8032310009002686, + "num_tokens": 560777839.0, + "step": 14704 + }, + { + "epoch": 1.8706271466734512, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8128979206085205, + "num_tokens": 560816167.0, + "step": 14705 + }, + { + "epoch": 1.8707543569520417, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7834041118621826, + "num_tokens": 560854524.0, + "step": 14706 + }, + { + "epoch": 1.8708815672306323, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.794586181640625, + "num_tokens": 560894393.0, + "step": 14707 + }, + { + "epoch": 1.8710087775092228, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8147056102752686, + "num_tokens": 560938640.0, + "step": 14708 + }, + { + "epoch": 1.871135987787813, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8026495575904846, + "num_tokens": 560976805.0, + "step": 14709 + }, + { + "epoch": 1.8712631980664036, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.795583963394165, + "num_tokens": 561018940.0, + "step": 14710 + }, + { + "epoch": 1.8713904083449941, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8039089441299438, + "num_tokens": 561065228.0, + "step": 14711 + }, + { + "epoch": 1.8715176186235847, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7859166860580444, + "num_tokens": 561107540.0, + "step": 14712 + }, + { + "epoch": 1.8716448289021752, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7750881910324097, + "num_tokens": 561144774.0, + "step": 14713 + }, + { + "epoch": 1.8717720391807657, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7883421182632446, + "num_tokens": 561186774.0, + "step": 14714 + }, + { + "epoch": 1.8718992494593563, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8134918212890625, + "num_tokens": 561223997.0, + "step": 14715 + }, + { + "epoch": 1.8720264597379468, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8078261017799377, + "num_tokens": 561261583.0, + "step": 14716 + }, + { + "epoch": 1.8721536700165373, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8031680583953857, + "num_tokens": 561302681.0, + "step": 14717 + }, + { + "epoch": 1.8722808802951278, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7989601492881775, + "num_tokens": 561338402.0, + "step": 14718 + }, + { + "epoch": 1.8724080905737184, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8063995838165283, + "num_tokens": 561372892.0, + "step": 14719 + }, + { + "epoch": 1.872535300852309, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.797494113445282, + "num_tokens": 561409474.0, + "step": 14720 + }, + { + "epoch": 1.8726625111308994, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7745307683944702, + "num_tokens": 561450152.0, + "step": 14721 + }, + { + "epoch": 1.87278972140949, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7960309982299805, + "num_tokens": 561489448.0, + "step": 14722 + }, + { + "epoch": 1.8729169316880805, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7737432718276978, + "num_tokens": 561529684.0, + "step": 14723 + }, + { + "epoch": 1.873044141966671, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7989276647567749, + "num_tokens": 561564259.0, + "step": 14724 + }, + { + "epoch": 1.8731713522452615, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8133074045181274, + "num_tokens": 561609650.0, + "step": 14725 + }, + { + "epoch": 1.873298562523852, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8080049753189087, + "num_tokens": 561645520.0, + "step": 14726 + }, + { + "epoch": 1.8734257728024426, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8081423044204712, + "num_tokens": 561682771.0, + "step": 14727 + }, + { + "epoch": 1.873552983081033, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7941491007804871, + "num_tokens": 561721317.0, + "step": 14728 + }, + { + "epoch": 1.8736801933596234, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8068421483039856, + "num_tokens": 561763178.0, + "step": 14729 + }, + { + "epoch": 1.873807403638214, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8077481985092163, + "num_tokens": 561802292.0, + "step": 14730 + }, + { + "epoch": 1.8739346139168045, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8105899095535278, + "num_tokens": 561838610.0, + "step": 14731 + }, + { + "epoch": 1.874061824195395, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8031677007675171, + "num_tokens": 561881992.0, + "step": 14732 + }, + { + "epoch": 1.8741890344739855, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7887132167816162, + "num_tokens": 561917661.0, + "step": 14733 + }, + { + "epoch": 1.8743162447525759, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7980872988700867, + "num_tokens": 561958218.0, + "step": 14734 + }, + { + "epoch": 1.8744434550311664, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.810438871383667, + "num_tokens": 561993882.0, + "step": 14735 + }, + { + "epoch": 1.874570665309757, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8020607233047485, + "num_tokens": 562035443.0, + "step": 14736 + }, + { + "epoch": 1.8746978755883474, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8010846972465515, + "num_tokens": 562072806.0, + "step": 14737 + }, + { + "epoch": 1.874825085866938, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7850580215454102, + "num_tokens": 562113652.0, + "step": 14738 + }, + { + "epoch": 1.8749522961455285, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.790094792842865, + "num_tokens": 562150600.0, + "step": 14739 + }, + { + "epoch": 1.875079506424119, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7996653318405151, + "num_tokens": 562187730.0, + "step": 14740 + }, + { + "epoch": 1.8752067167027096, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8067865371704102, + "num_tokens": 562228229.0, + "step": 14741 + }, + { + "epoch": 1.8753339269813, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8060405254364014, + "num_tokens": 562262293.0, + "step": 14742 + }, + { + "epoch": 1.8754611372598906, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8004027605056763, + "num_tokens": 562303389.0, + "step": 14743 + }, + { + "epoch": 1.8755883475384811, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7946354746818542, + "num_tokens": 562344461.0, + "step": 14744 + }, + { + "epoch": 1.8757155578170717, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7855813503265381, + "num_tokens": 562387068.0, + "step": 14745 + }, + { + "epoch": 1.8758427680956622, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8044120669364929, + "num_tokens": 562428225.0, + "step": 14746 + }, + { + "epoch": 1.8759699783742527, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8033959865570068, + "num_tokens": 562463879.0, + "step": 14747 + }, + { + "epoch": 1.8760971886528433, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7907627820968628, + "num_tokens": 562502327.0, + "step": 14748 + }, + { + "epoch": 1.8762243989314338, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8045613765716553, + "num_tokens": 562542162.0, + "step": 14749 + }, + { + "epoch": 1.8763516092100243, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7933177947998047, + "num_tokens": 562583364.0, + "step": 14750 + }, + { + "epoch": 1.8764788194886148, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7989883422851562, + "num_tokens": 562615870.0, + "step": 14751 + }, + { + "epoch": 1.8766060297672051, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7873260974884033, + "num_tokens": 562655086.0, + "step": 14752 + }, + { + "epoch": 1.8767332400457957, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7952197194099426, + "num_tokens": 562697532.0, + "step": 14753 + }, + { + "epoch": 1.8768604503243862, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7719149589538574, + "num_tokens": 562731057.0, + "step": 14754 + }, + { + "epoch": 1.8769876606029767, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7905004024505615, + "num_tokens": 562770402.0, + "step": 14755 + }, + { + "epoch": 1.8771148708815673, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7972720265388489, + "num_tokens": 562806191.0, + "step": 14756 + }, + { + "epoch": 1.8772420811601578, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8011400699615479, + "num_tokens": 562846864.0, + "step": 14757 + }, + { + "epoch": 1.877369291438748, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7899428606033325, + "num_tokens": 562882887.0, + "step": 14758 + }, + { + "epoch": 1.8774965017173386, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.800247848033905, + "num_tokens": 562919096.0, + "step": 14759 + }, + { + "epoch": 1.8776237119959291, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7694351673126221, + "num_tokens": 562959336.0, + "step": 14760 + }, + { + "epoch": 1.8777509222745197, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8005779385566711, + "num_tokens": 562997883.0, + "step": 14761 + }, + { + "epoch": 1.8778781325531102, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8134418725967407, + "num_tokens": 563036763.0, + "step": 14762 + }, + { + "epoch": 1.8780053428317007, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8159211874008179, + "num_tokens": 563076815.0, + "step": 14763 + }, + { + "epoch": 1.8781325531102913, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8118153810501099, + "num_tokens": 563110511.0, + "step": 14764 + }, + { + "epoch": 1.8782597633888818, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7909212708473206, + "num_tokens": 563140757.0, + "step": 14765 + }, + { + "epoch": 1.8783869736674723, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8018843531608582, + "num_tokens": 563179961.0, + "step": 14766 + }, + { + "epoch": 1.8785141839460628, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8092957139015198, + "num_tokens": 563221355.0, + "step": 14767 + }, + { + "epoch": 1.8786413942246534, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.5942, + "mean_token_accuracy": 0.8213281631469727, + "num_tokens": 563257427.0, + "step": 14768 + }, + { + "epoch": 1.878768604503244, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7987513542175293, + "num_tokens": 563294371.0, + "step": 14769 + }, + { + "epoch": 1.8788958147818344, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.815457284450531, + "num_tokens": 563338592.0, + "step": 14770 + }, + { + "epoch": 1.879023025060425, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8045252561569214, + "num_tokens": 563373853.0, + "step": 14771 + }, + { + "epoch": 1.8791502353390155, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8015479445457458, + "num_tokens": 563411137.0, + "step": 14772 + }, + { + "epoch": 1.879277445617606, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8060048222541809, + "num_tokens": 563450294.0, + "step": 14773 + }, + { + "epoch": 1.8794046558961965, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8094536066055298, + "num_tokens": 563496858.0, + "step": 14774 + }, + { + "epoch": 1.879531866174787, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7988193035125732, + "num_tokens": 563539566.0, + "step": 14775 + }, + { + "epoch": 1.8796590764533776, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8053368330001831, + "num_tokens": 563574363.0, + "step": 14776 + }, + { + "epoch": 1.879786286731968, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7988758087158203, + "num_tokens": 563605540.0, + "step": 14777 + }, + { + "epoch": 1.8799134970105584, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7900698781013489, + "num_tokens": 563646640.0, + "step": 14778 + }, + { + "epoch": 1.880040707289149, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.809074878692627, + "num_tokens": 563683576.0, + "step": 14779 + }, + { + "epoch": 1.8801679175677395, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7933201789855957, + "num_tokens": 563724088.0, + "step": 14780 + }, + { + "epoch": 1.88029512784633, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7985098361968994, + "num_tokens": 563760897.0, + "step": 14781 + }, + { + "epoch": 1.8804223381249205, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8086658716201782, + "num_tokens": 563800158.0, + "step": 14782 + }, + { + "epoch": 1.8805495484035109, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8074204921722412, + "num_tokens": 563833288.0, + "step": 14783 + }, + { + "epoch": 1.8806767586821014, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8166176676750183, + "num_tokens": 563874507.0, + "step": 14784 + }, + { + "epoch": 1.880803968960692, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7984431385993958, + "num_tokens": 563912452.0, + "step": 14785 + }, + { + "epoch": 1.8809311792392824, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7823206186294556, + "num_tokens": 563950565.0, + "step": 14786 + }, + { + "epoch": 1.881058389517873, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7871024012565613, + "num_tokens": 563984924.0, + "step": 14787 + }, + { + "epoch": 1.8811855997964635, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8088045716285706, + "num_tokens": 564022887.0, + "step": 14788 + }, + { + "epoch": 1.881312810075054, + "grad_norm": 10.125, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8041114807128906, + "num_tokens": 564064829.0, + "step": 14789 + }, + { + "epoch": 1.8814400203536445, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7760570049285889, + "num_tokens": 564100858.0, + "step": 14790 + }, + { + "epoch": 1.881567230632235, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7958204746246338, + "num_tokens": 564141477.0, + "step": 14791 + }, + { + "epoch": 1.8816944409108256, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7945206761360168, + "num_tokens": 564179271.0, + "step": 14792 + }, + { + "epoch": 1.8818216511894161, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8098008632659912, + "num_tokens": 564224336.0, + "step": 14793 + }, + { + "epoch": 1.8819488614680067, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7956693172454834, + "num_tokens": 564265157.0, + "step": 14794 + }, + { + "epoch": 1.8820760717465972, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.807582437992096, + "num_tokens": 564295023.0, + "step": 14795 + }, + { + "epoch": 1.8822032820251877, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8053462505340576, + "num_tokens": 564333700.0, + "step": 14796 + }, + { + "epoch": 1.8823304923037782, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7877333164215088, + "num_tokens": 564374342.0, + "step": 14797 + }, + { + "epoch": 1.8824577025823688, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8101218938827515, + "num_tokens": 564422877.0, + "step": 14798 + }, + { + "epoch": 1.8825849128609593, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7987685203552246, + "num_tokens": 564464713.0, + "step": 14799 + }, + { + "epoch": 1.8827121231395498, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7862728834152222, + "num_tokens": 564500667.0, + "step": 14800 + }, + { + "epoch": 1.8828393334181401, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8121117353439331, + "num_tokens": 564541460.0, + "step": 14801 + }, + { + "epoch": 1.8829665436967307, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7898260354995728, + "num_tokens": 564580983.0, + "step": 14802 + }, + { + "epoch": 1.8830937539753212, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8142713308334351, + "num_tokens": 564615967.0, + "step": 14803 + }, + { + "epoch": 1.8832209642539117, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7881979942321777, + "num_tokens": 564653255.0, + "step": 14804 + }, + { + "epoch": 1.8833481745325023, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7983558177947998, + "num_tokens": 564688420.0, + "step": 14805 + }, + { + "epoch": 1.8834753848110928, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7937010526657104, + "num_tokens": 564733123.0, + "step": 14806 + }, + { + "epoch": 1.883602595089683, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8056978583335876, + "num_tokens": 564767712.0, + "step": 14807 + }, + { + "epoch": 1.8837298053682736, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.793388843536377, + "num_tokens": 564810663.0, + "step": 14808 + }, + { + "epoch": 1.8838570156468641, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7946294546127319, + "num_tokens": 564848689.0, + "step": 14809 + }, + { + "epoch": 1.8839842259254547, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.788501501083374, + "num_tokens": 564887426.0, + "step": 14810 + }, + { + "epoch": 1.8841114362040452, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8045613765716553, + "num_tokens": 564924852.0, + "step": 14811 + }, + { + "epoch": 1.8842386464826357, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7933534383773804, + "num_tokens": 564964201.0, + "step": 14812 + }, + { + "epoch": 1.8843658567612263, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8124316334724426, + "num_tokens": 565004939.0, + "step": 14813 + }, + { + "epoch": 1.8844930670398168, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.804585874080658, + "num_tokens": 565044067.0, + "step": 14814 + }, + { + "epoch": 1.8846202773184073, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8047233819961548, + "num_tokens": 565082114.0, + "step": 14815 + }, + { + "epoch": 1.8847474875969978, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8018943071365356, + "num_tokens": 565118350.0, + "step": 14816 + }, + { + "epoch": 1.8848746978755884, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7885448336601257, + "num_tokens": 565153313.0, + "step": 14817 + }, + { + "epoch": 1.885001908154179, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7867185473442078, + "num_tokens": 565191332.0, + "step": 14818 + }, + { + "epoch": 1.8851291184327694, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7962970733642578, + "num_tokens": 565229937.0, + "step": 14819 + }, + { + "epoch": 1.88525632871136, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7957479953765869, + "num_tokens": 565266271.0, + "step": 14820 + }, + { + "epoch": 1.8853835389899505, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7967087030410767, + "num_tokens": 565300593.0, + "step": 14821 + }, + { + "epoch": 1.885510749268541, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8017361164093018, + "num_tokens": 565341801.0, + "step": 14822 + }, + { + "epoch": 1.8856379595471315, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8103035688400269, + "num_tokens": 565374380.0, + "step": 14823 + }, + { + "epoch": 1.885765169825722, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7965066432952881, + "num_tokens": 565413096.0, + "step": 14824 + }, + { + "epoch": 1.8858923801043126, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8111830353736877, + "num_tokens": 565453292.0, + "step": 14825 + }, + { + "epoch": 1.886019590382903, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7890037894248962, + "num_tokens": 565489311.0, + "step": 14826 + }, + { + "epoch": 1.8861468006614934, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7808713316917419, + "num_tokens": 565526293.0, + "step": 14827 + }, + { + "epoch": 1.886274010940084, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7914836406707764, + "num_tokens": 565569215.0, + "step": 14828 + }, + { + "epoch": 1.8864012212186745, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8138309717178345, + "num_tokens": 565610755.0, + "step": 14829 + }, + { + "epoch": 1.886528431497265, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8007382154464722, + "num_tokens": 565641964.0, + "step": 14830 + }, + { + "epoch": 1.8866556417758555, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7687873840332031, + "num_tokens": 565682852.0, + "step": 14831 + }, + { + "epoch": 1.8867828520544458, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8171558380126953, + "num_tokens": 565721352.0, + "step": 14832 + }, + { + "epoch": 1.8869100623330364, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8115661144256592, + "num_tokens": 565764045.0, + "step": 14833 + }, + { + "epoch": 1.887037272611627, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.806911826133728, + "num_tokens": 565797617.0, + "step": 14834 + }, + { + "epoch": 1.8871644828902174, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7953440546989441, + "num_tokens": 565839254.0, + "step": 14835 + }, + { + "epoch": 1.887291693168808, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.80303955078125, + "num_tokens": 565874742.0, + "step": 14836 + }, + { + "epoch": 1.8874189034473985, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7998532652854919, + "num_tokens": 565911043.0, + "step": 14837 + }, + { + "epoch": 1.887546113725989, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7929763793945312, + "num_tokens": 565950228.0, + "step": 14838 + }, + { + "epoch": 1.8876733240045795, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8126662969589233, + "num_tokens": 565982891.0, + "step": 14839 + }, + { + "epoch": 1.88780053428317, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7951050400733948, + "num_tokens": 566022601.0, + "step": 14840 + }, + { + "epoch": 1.8879277445617606, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7991970181465149, + "num_tokens": 566054928.0, + "step": 14841 + }, + { + "epoch": 1.8880549548403511, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7958012223243713, + "num_tokens": 566094900.0, + "step": 14842 + }, + { + "epoch": 1.8881821651189417, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7934796810150146, + "num_tokens": 566125230.0, + "step": 14843 + }, + { + "epoch": 1.8883093753975322, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8156900405883789, + "num_tokens": 566165320.0, + "step": 14844 + }, + { + "epoch": 1.8884365856761227, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7872412204742432, + "num_tokens": 566202769.0, + "step": 14845 + }, + { + "epoch": 1.8885637959547132, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.800839900970459, + "num_tokens": 566245874.0, + "step": 14846 + }, + { + "epoch": 1.8886910062333038, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7824693918228149, + "num_tokens": 566284855.0, + "step": 14847 + }, + { + "epoch": 1.8888182165118943, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7925089001655579, + "num_tokens": 566323422.0, + "step": 14848 + }, + { + "epoch": 1.8889454267904848, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8123118877410889, + "num_tokens": 566364357.0, + "step": 14849 + }, + { + "epoch": 1.8890726370690751, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7915974855422974, + "num_tokens": 566404498.0, + "step": 14850 + }, + { + "epoch": 1.8891998473476657, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7962912917137146, + "num_tokens": 566442960.0, + "step": 14851 + }, + { + "epoch": 1.8893270576262562, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8036357164382935, + "num_tokens": 566475087.0, + "step": 14852 + }, + { + "epoch": 1.8894542679048467, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7894698977470398, + "num_tokens": 566514337.0, + "step": 14853 + }, + { + "epoch": 1.8895814781834372, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7872880101203918, + "num_tokens": 566553204.0, + "step": 14854 + }, + { + "epoch": 1.8897086884620278, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7974777221679688, + "num_tokens": 566589697.0, + "step": 14855 + }, + { + "epoch": 1.889835898740618, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.785118818283081, + "num_tokens": 566629680.0, + "step": 14856 + }, + { + "epoch": 1.8899631090192086, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.5666, + "mean_token_accuracy": 0.8270363211631775, + "num_tokens": 566670931.0, + "step": 14857 + }, + { + "epoch": 1.8900903192977991, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7910926342010498, + "num_tokens": 566706673.0, + "step": 14858 + }, + { + "epoch": 1.8902175295763897, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.793075680732727, + "num_tokens": 566749503.0, + "step": 14859 + }, + { + "epoch": 1.8903447398549802, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.81236732006073, + "num_tokens": 566790700.0, + "step": 14860 + }, + { + "epoch": 1.8904719501335707, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7962417602539062, + "num_tokens": 566832995.0, + "step": 14861 + }, + { + "epoch": 1.8905991604121613, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7851979732513428, + "num_tokens": 566868190.0, + "step": 14862 + }, + { + "epoch": 1.8907263706907518, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7678430080413818, + "num_tokens": 566906637.0, + "step": 14863 + }, + { + "epoch": 1.8908535809693423, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7980029582977295, + "num_tokens": 566946006.0, + "step": 14864 + }, + { + "epoch": 1.8909807912479328, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7891608476638794, + "num_tokens": 566987726.0, + "step": 14865 + }, + { + "epoch": 1.8911080015265234, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8189790844917297, + "num_tokens": 567027432.0, + "step": 14866 + }, + { + "epoch": 1.891235211805114, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8008736371994019, + "num_tokens": 567060607.0, + "step": 14867 + }, + { + "epoch": 1.8913624220837044, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7923370599746704, + "num_tokens": 567100691.0, + "step": 14868 + }, + { + "epoch": 1.891489632362295, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.8191577792167664, + "num_tokens": 567139460.0, + "step": 14869 + }, + { + "epoch": 1.8916168426408855, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7922900915145874, + "num_tokens": 567185384.0, + "step": 14870 + }, + { + "epoch": 1.891744052919476, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7959458231925964, + "num_tokens": 567219898.0, + "step": 14871 + }, + { + "epoch": 1.8918712631980665, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.80889493227005, + "num_tokens": 567260322.0, + "step": 14872 + }, + { + "epoch": 1.891998473476657, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7943171262741089, + "num_tokens": 567293739.0, + "step": 14873 + }, + { + "epoch": 1.8921256837552476, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7980189323425293, + "num_tokens": 567327282.0, + "step": 14874 + }, + { + "epoch": 1.892252894033838, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7922170758247375, + "num_tokens": 567366402.0, + "step": 14875 + }, + { + "epoch": 1.8923801043124284, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7941943407058716, + "num_tokens": 567409924.0, + "step": 14876 + }, + { + "epoch": 1.892507314591019, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7993647456169128, + "num_tokens": 567450355.0, + "step": 14877 + }, + { + "epoch": 1.8926345248696095, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.8200174570083618, + "num_tokens": 567487701.0, + "step": 14878 + }, + { + "epoch": 1.8927617351482, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8040333986282349, + "num_tokens": 567527619.0, + "step": 14879 + }, + { + "epoch": 1.8928889454267905, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.798729419708252, + "num_tokens": 567567016.0, + "step": 14880 + }, + { + "epoch": 1.8930161557053808, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7897754907608032, + "num_tokens": 567605465.0, + "step": 14881 + }, + { + "epoch": 1.8931433659839714, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8045952320098877, + "num_tokens": 567646680.0, + "step": 14882 + }, + { + "epoch": 1.893270576262562, + "grad_norm": 8.8125, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.803271472454071, + "num_tokens": 567685685.0, + "step": 14883 + }, + { + "epoch": 1.8933977865411524, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7858938574790955, + "num_tokens": 567717222.0, + "step": 14884 + }, + { + "epoch": 1.893524996819743, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7918912172317505, + "num_tokens": 567761050.0, + "step": 14885 + }, + { + "epoch": 1.8936522070983335, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7924739718437195, + "num_tokens": 567801423.0, + "step": 14886 + }, + { + "epoch": 1.893779417376924, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7931811213493347, + "num_tokens": 567844614.0, + "step": 14887 + }, + { + "epoch": 1.8939066276555145, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7989730834960938, + "num_tokens": 567878803.0, + "step": 14888 + }, + { + "epoch": 1.894033837934105, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8119316101074219, + "num_tokens": 567909419.0, + "step": 14889 + }, + { + "epoch": 1.8941610482126956, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7985212206840515, + "num_tokens": 567949221.0, + "step": 14890 + }, + { + "epoch": 1.8942882584912861, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8188685178756714, + "num_tokens": 567985101.0, + "step": 14891 + }, + { + "epoch": 1.8944154687698767, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7896400690078735, + "num_tokens": 568030542.0, + "step": 14892 + }, + { + "epoch": 1.8945426790484672, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8048288822174072, + "num_tokens": 568070723.0, + "step": 14893 + }, + { + "epoch": 1.8946698893270577, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.79352867603302, + "num_tokens": 568113444.0, + "step": 14894 + }, + { + "epoch": 1.8947970996056482, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8052033185958862, + "num_tokens": 568152646.0, + "step": 14895 + }, + { + "epoch": 1.8949243098842388, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8032143115997314, + "num_tokens": 568187046.0, + "step": 14896 + }, + { + "epoch": 1.8950515201628293, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7906179428100586, + "num_tokens": 568225052.0, + "step": 14897 + }, + { + "epoch": 1.8951787304414198, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.787175714969635, + "num_tokens": 568258299.0, + "step": 14898 + }, + { + "epoch": 1.8953059407200101, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8060324192047119, + "num_tokens": 568297137.0, + "step": 14899 + }, + { + "epoch": 1.8954331509986007, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8198705315589905, + "num_tokens": 568331102.0, + "step": 14900 + }, + { + "epoch": 1.8955603612771912, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7968844175338745, + "num_tokens": 568373926.0, + "step": 14901 + }, + { + "epoch": 1.8956875715557817, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8075122237205505, + "num_tokens": 568406470.0, + "step": 14902 + }, + { + "epoch": 1.8958147818343722, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8054078221321106, + "num_tokens": 568443452.0, + "step": 14903 + }, + { + "epoch": 1.8959419921129628, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7931532859802246, + "num_tokens": 568480085.0, + "step": 14904 + }, + { + "epoch": 1.896069202391553, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7947484254837036, + "num_tokens": 568519191.0, + "step": 14905 + }, + { + "epoch": 1.8961964126701436, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8078120350837708, + "num_tokens": 568554224.0, + "step": 14906 + }, + { + "epoch": 1.8963236229487341, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8165429830551147, + "num_tokens": 568586102.0, + "step": 14907 + }, + { + "epoch": 1.8964508332273247, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7937178611755371, + "num_tokens": 568625224.0, + "step": 14908 + }, + { + "epoch": 1.8965780435059152, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7806965708732605, + "num_tokens": 568665588.0, + "step": 14909 + }, + { + "epoch": 1.8967052537845057, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8203154802322388, + "num_tokens": 568698408.0, + "step": 14910 + }, + { + "epoch": 1.8968324640630962, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.786526083946228, + "num_tokens": 568731982.0, + "step": 14911 + }, + { + "epoch": 1.8969596743416868, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7988672256469727, + "num_tokens": 568769388.0, + "step": 14912 + }, + { + "epoch": 1.8970868846202773, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.804896891117096, + "num_tokens": 568810329.0, + "step": 14913 + }, + { + "epoch": 1.8972140948988678, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8035646677017212, + "num_tokens": 568852241.0, + "step": 14914 + }, + { + "epoch": 1.8973413051774584, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.793091893196106, + "num_tokens": 568889660.0, + "step": 14915 + }, + { + "epoch": 1.8974685154560489, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.799394965171814, + "num_tokens": 568934489.0, + "step": 14916 + }, + { + "epoch": 1.8975957257346394, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.8141741156578064, + "num_tokens": 568971904.0, + "step": 14917 + }, + { + "epoch": 1.89772293601323, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7832525968551636, + "num_tokens": 569011863.0, + "step": 14918 + }, + { + "epoch": 1.8978501462918205, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.8235714435577393, + "num_tokens": 569050295.0, + "step": 14919 + }, + { + "epoch": 1.897977356570411, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7918407917022705, + "num_tokens": 569087013.0, + "step": 14920 + }, + { + "epoch": 1.8981045668490015, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7938072681427002, + "num_tokens": 569125226.0, + "step": 14921 + }, + { + "epoch": 1.898231777127592, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7965971827507019, + "num_tokens": 569161910.0, + "step": 14922 + }, + { + "epoch": 1.8983589874061826, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8000159859657288, + "num_tokens": 569196519.0, + "step": 14923 + }, + { + "epoch": 1.898486197684773, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7986135482788086, + "num_tokens": 569234404.0, + "step": 14924 + }, + { + "epoch": 1.8986134079633634, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7912417650222778, + "num_tokens": 569272728.0, + "step": 14925 + }, + { + "epoch": 1.898740618241954, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7927790284156799, + "num_tokens": 569315434.0, + "step": 14926 + }, + { + "epoch": 1.8988678285205445, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7859545946121216, + "num_tokens": 569357473.0, + "step": 14927 + }, + { + "epoch": 1.898995038799135, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7841120362281799, + "num_tokens": 569398747.0, + "step": 14928 + }, + { + "epoch": 1.8991222490777255, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8071614503860474, + "num_tokens": 569434855.0, + "step": 14929 + }, + { + "epoch": 1.8992494593563158, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.798488974571228, + "num_tokens": 569475236.0, + "step": 14930 + }, + { + "epoch": 1.8993766696349064, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7990903258323669, + "num_tokens": 569513599.0, + "step": 14931 + }, + { + "epoch": 1.899503879913497, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8117848634719849, + "num_tokens": 569559059.0, + "step": 14932 + }, + { + "epoch": 1.8996310901920874, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8215526342391968, + "num_tokens": 569595240.0, + "step": 14933 + }, + { + "epoch": 1.899758300470678, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7893153429031372, + "num_tokens": 569629336.0, + "step": 14934 + }, + { + "epoch": 1.8998855107492685, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8097974061965942, + "num_tokens": 569663613.0, + "step": 14935 + }, + { + "epoch": 1.900012721027859, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7817182540893555, + "num_tokens": 569702884.0, + "step": 14936 + }, + { + "epoch": 1.9001399313064495, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8149577379226685, + "num_tokens": 569744618.0, + "step": 14937 + }, + { + "epoch": 1.90026714158504, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8075671195983887, + "num_tokens": 569777083.0, + "step": 14938 + }, + { + "epoch": 1.9003943518636306, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8068367838859558, + "num_tokens": 569809652.0, + "step": 14939 + }, + { + "epoch": 1.9005215621422211, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8019543886184692, + "num_tokens": 569846465.0, + "step": 14940 + }, + { + "epoch": 1.9006487724208116, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7910996675491333, + "num_tokens": 569883655.0, + "step": 14941 + }, + { + "epoch": 1.9007759826994022, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7980750203132629, + "num_tokens": 569923358.0, + "step": 14942 + }, + { + "epoch": 1.9009031929779927, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7918744087219238, + "num_tokens": 569961393.0, + "step": 14943 + }, + { + "epoch": 1.9010304032565832, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7714362144470215, + "num_tokens": 570002092.0, + "step": 14944 + }, + { + "epoch": 1.9011576135351738, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7791980504989624, + "num_tokens": 570037699.0, + "step": 14945 + }, + { + "epoch": 1.9012848238137643, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7775918245315552, + "num_tokens": 570077693.0, + "step": 14946 + }, + { + "epoch": 1.9014120340923548, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8204197883605957, + "num_tokens": 570115354.0, + "step": 14947 + }, + { + "epoch": 1.9015392443709451, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8067668676376343, + "num_tokens": 570154387.0, + "step": 14948 + }, + { + "epoch": 1.9016664546495357, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7914741635322571, + "num_tokens": 570189743.0, + "step": 14949 + }, + { + "epoch": 1.9017936649281262, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8110755681991577, + "num_tokens": 570224009.0, + "step": 14950 + }, + { + "epoch": 1.9019208752067167, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8031021356582642, + "num_tokens": 570266703.0, + "step": 14951 + }, + { + "epoch": 1.9020480854853072, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8073484897613525, + "num_tokens": 570308241.0, + "step": 14952 + }, + { + "epoch": 1.9021752957638978, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7963690161705017, + "num_tokens": 570346931.0, + "step": 14953 + }, + { + "epoch": 1.902302506042488, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8111156225204468, + "num_tokens": 570380373.0, + "step": 14954 + }, + { + "epoch": 1.9024297163210786, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.8173567056655884, + "num_tokens": 570420106.0, + "step": 14955 + }, + { + "epoch": 1.9025569265996691, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8241547346115112, + "num_tokens": 570464534.0, + "step": 14956 + }, + { + "epoch": 1.9026841368782597, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7943611145019531, + "num_tokens": 570505734.0, + "step": 14957 + }, + { + "epoch": 1.9028113471568502, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.8123657703399658, + "num_tokens": 570545913.0, + "step": 14958 + }, + { + "epoch": 1.9029385574354407, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8107448816299438, + "num_tokens": 570584397.0, + "step": 14959 + }, + { + "epoch": 1.9030657677140312, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8048338294029236, + "num_tokens": 570622285.0, + "step": 14960 + }, + { + "epoch": 1.9031929779926218, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.778450071811676, + "num_tokens": 570656331.0, + "step": 14961 + }, + { + "epoch": 1.9033201882712123, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8121469020843506, + "num_tokens": 570699944.0, + "step": 14962 + }, + { + "epoch": 1.9034473985498028, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7931195497512817, + "num_tokens": 570737649.0, + "step": 14963 + }, + { + "epoch": 1.9035746088283934, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7980563640594482, + "num_tokens": 570773193.0, + "step": 14964 + }, + { + "epoch": 1.9037018191069839, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7905234098434448, + "num_tokens": 570809117.0, + "step": 14965 + }, + { + "epoch": 1.9038290293855744, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8123540282249451, + "num_tokens": 570844268.0, + "step": 14966 + }, + { + "epoch": 1.903956239664165, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7939386963844299, + "num_tokens": 570890166.0, + "step": 14967 + }, + { + "epoch": 1.9040834499427555, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7864458560943604, + "num_tokens": 570921419.0, + "step": 14968 + }, + { + "epoch": 1.904210660221346, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7976166605949402, + "num_tokens": 570966444.0, + "step": 14969 + }, + { + "epoch": 1.9043378704999365, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8171766996383667, + "num_tokens": 571011656.0, + "step": 14970 + }, + { + "epoch": 1.904465080778527, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8067813515663147, + "num_tokens": 571047828.0, + "step": 14971 + }, + { + "epoch": 1.9045922910571176, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7955006957054138, + "num_tokens": 571084541.0, + "step": 14972 + }, + { + "epoch": 1.9047195013357079, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7966324090957642, + "num_tokens": 571120978.0, + "step": 14973 + }, + { + "epoch": 1.9048467116142984, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8145397305488586, + "num_tokens": 571160181.0, + "step": 14974 + }, + { + "epoch": 1.904973921892889, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7985948324203491, + "num_tokens": 571194648.0, + "step": 14975 + }, + { + "epoch": 1.9051011321714795, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.794647216796875, + "num_tokens": 571238449.0, + "step": 14976 + }, + { + "epoch": 1.90522834245007, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8002363443374634, + "num_tokens": 571277266.0, + "step": 14977 + }, + { + "epoch": 1.9053555527286605, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.817299485206604, + "num_tokens": 571316891.0, + "step": 14978 + }, + { + "epoch": 1.9054827630072508, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8047690987586975, + "num_tokens": 571353097.0, + "step": 14979 + }, + { + "epoch": 1.9056099732858414, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.819601833820343, + "num_tokens": 571392217.0, + "step": 14980 + }, + { + "epoch": 1.905737183564432, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8134286403656006, + "num_tokens": 571430440.0, + "step": 14981 + }, + { + "epoch": 1.9058643938430224, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.798373818397522, + "num_tokens": 571472002.0, + "step": 14982 + }, + { + "epoch": 1.905991604121613, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7799916863441467, + "num_tokens": 571509368.0, + "step": 14983 + }, + { + "epoch": 1.9061188144002035, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7829633355140686, + "num_tokens": 571543898.0, + "step": 14984 + }, + { + "epoch": 1.906246024678794, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7933434844017029, + "num_tokens": 571584395.0, + "step": 14985 + }, + { + "epoch": 1.9063732349573845, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8027482628822327, + "num_tokens": 571624792.0, + "step": 14986 + }, + { + "epoch": 1.906500445235975, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8003315329551697, + "num_tokens": 571667197.0, + "step": 14987 + }, + { + "epoch": 1.9066276555145656, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7936343550682068, + "num_tokens": 571703648.0, + "step": 14988 + }, + { + "epoch": 1.9067548657931561, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7886084318161011, + "num_tokens": 571738980.0, + "step": 14989 + }, + { + "epoch": 1.9068820760717466, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8011854887008667, + "num_tokens": 571774540.0, + "step": 14990 + }, + { + "epoch": 1.9070092863503372, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8059715032577515, + "num_tokens": 571805286.0, + "step": 14991 + }, + { + "epoch": 1.9071364966289277, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7968892455101013, + "num_tokens": 571847936.0, + "step": 14992 + }, + { + "epoch": 1.9072637069075182, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7931779026985168, + "num_tokens": 571885454.0, + "step": 14993 + }, + { + "epoch": 1.9073909171861088, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8021025657653809, + "num_tokens": 571924870.0, + "step": 14994 + }, + { + "epoch": 1.9075181274646993, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7747039794921875, + "num_tokens": 571957870.0, + "step": 14995 + }, + { + "epoch": 1.9076453377432898, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7868313789367676, + "num_tokens": 571993645.0, + "step": 14996 + }, + { + "epoch": 1.9077725480218801, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8071755766868591, + "num_tokens": 572028594.0, + "step": 14997 + }, + { + "epoch": 1.9078997583004706, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7721054553985596, + "num_tokens": 572066971.0, + "step": 14998 + }, + { + "epoch": 1.9080269685790612, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.8096108436584473, + "num_tokens": 572107609.0, + "step": 14999 + }, + { + "epoch": 1.9081541788576517, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8139016628265381, + "num_tokens": 572146940.0, + "step": 15000 + }, + { + "epoch": 1.9082813891362422, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.5701, + "mean_token_accuracy": 0.8258869647979736, + "num_tokens": 572184010.0, + "step": 15001 + }, + { + "epoch": 1.9084085994148328, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7809927463531494, + "num_tokens": 572223184.0, + "step": 15002 + }, + { + "epoch": 1.908535809693423, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8008919954299927, + "num_tokens": 572261770.0, + "step": 15003 + }, + { + "epoch": 1.9086630199720136, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8083926439285278, + "num_tokens": 572303229.0, + "step": 15004 + }, + { + "epoch": 1.9087902302506041, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7828454375267029, + "num_tokens": 572337839.0, + "step": 15005 + }, + { + "epoch": 1.9089174405291947, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7942367792129517, + "num_tokens": 572373604.0, + "step": 15006 + }, + { + "epoch": 1.9090446508077852, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8072799444198608, + "num_tokens": 572410425.0, + "step": 15007 + }, + { + "epoch": 1.9091718610863757, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8133841156959534, + "num_tokens": 572443998.0, + "step": 15008 + }, + { + "epoch": 1.9092990713649662, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7863045334815979, + "num_tokens": 572485337.0, + "step": 15009 + }, + { + "epoch": 1.9094262816435568, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7916895151138306, + "num_tokens": 572528273.0, + "step": 15010 + }, + { + "epoch": 1.9095534919221473, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8025161027908325, + "num_tokens": 572569467.0, + "step": 15011 + }, + { + "epoch": 1.9096807022007378, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8067969083786011, + "num_tokens": 572609626.0, + "step": 15012 + }, + { + "epoch": 1.9098079124793284, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7849817276000977, + "num_tokens": 572652907.0, + "step": 15013 + }, + { + "epoch": 1.9099351227579189, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8049647808074951, + "num_tokens": 572687460.0, + "step": 15014 + }, + { + "epoch": 1.9100623330365094, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8005025386810303, + "num_tokens": 572723937.0, + "step": 15015 + }, + { + "epoch": 1.9101895433151, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7942724823951721, + "num_tokens": 572764929.0, + "step": 15016 + }, + { + "epoch": 1.9103167535936905, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7998092174530029, + "num_tokens": 572802545.0, + "step": 15017 + }, + { + "epoch": 1.910443963872281, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.794447124004364, + "num_tokens": 572850008.0, + "step": 15018 + }, + { + "epoch": 1.9105711741508715, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8079863786697388, + "num_tokens": 572888968.0, + "step": 15019 + }, + { + "epoch": 1.910698384429462, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.798550009727478, + "num_tokens": 572926314.0, + "step": 15020 + }, + { + "epoch": 1.9108255947080524, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8070278167724609, + "num_tokens": 572961789.0, + "step": 15021 + }, + { + "epoch": 1.9109528049866429, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7989403009414673, + "num_tokens": 572993680.0, + "step": 15022 + }, + { + "epoch": 1.9110800152652334, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7986476421356201, + "num_tokens": 573024003.0, + "step": 15023 + }, + { + "epoch": 1.911207225543824, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8027238845825195, + "num_tokens": 573055138.0, + "step": 15024 + }, + { + "epoch": 1.9113344358224145, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8043553829193115, + "num_tokens": 573092373.0, + "step": 15025 + }, + { + "epoch": 1.911461646101005, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7964608669281006, + "num_tokens": 573131486.0, + "step": 15026 + }, + { + "epoch": 1.9115888563795955, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8138965368270874, + "num_tokens": 573170114.0, + "step": 15027 + }, + { + "epoch": 1.9117160666581858, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7990831732749939, + "num_tokens": 573200906.0, + "step": 15028 + }, + { + "epoch": 1.9118432769367764, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7908793687820435, + "num_tokens": 573243731.0, + "step": 15029 + }, + { + "epoch": 1.9119704872153669, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8071817755699158, + "num_tokens": 573283335.0, + "step": 15030 + }, + { + "epoch": 1.9120976974939574, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7847764492034912, + "num_tokens": 573326628.0, + "step": 15031 + }, + { + "epoch": 1.912224907772548, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8081170320510864, + "num_tokens": 573369025.0, + "step": 15032 + }, + { + "epoch": 1.9123521180511385, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7905898094177246, + "num_tokens": 573411923.0, + "step": 15033 + }, + { + "epoch": 1.912479328329729, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7911322712898254, + "num_tokens": 573450455.0, + "step": 15034 + }, + { + "epoch": 1.9126065386083195, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7906971573829651, + "num_tokens": 573492886.0, + "step": 15035 + }, + { + "epoch": 1.91273374888691, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8091221451759338, + "num_tokens": 573532073.0, + "step": 15036 + }, + { + "epoch": 1.9128609591655006, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.5896, + "mean_token_accuracy": 0.8203783631324768, + "num_tokens": 573573632.0, + "step": 15037 + }, + { + "epoch": 1.9129881694440911, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8106849789619446, + "num_tokens": 573610194.0, + "step": 15038 + }, + { + "epoch": 1.9131153797226816, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7889643907546997, + "num_tokens": 573650237.0, + "step": 15039 + }, + { + "epoch": 1.9132425900012722, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8026776313781738, + "num_tokens": 573686859.0, + "step": 15040 + }, + { + "epoch": 1.9133698002798627, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7944628000259399, + "num_tokens": 573726770.0, + "step": 15041 + }, + { + "epoch": 1.9134970105584532, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7968496680259705, + "num_tokens": 573768668.0, + "step": 15042 + }, + { + "epoch": 1.9136242208370438, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7923531532287598, + "num_tokens": 573807880.0, + "step": 15043 + }, + { + "epoch": 1.9137514311156343, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7967371344566345, + "num_tokens": 573847018.0, + "step": 15044 + }, + { + "epoch": 1.9138786413942248, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8140068054199219, + "num_tokens": 573881193.0, + "step": 15045 + }, + { + "epoch": 1.9140058516728151, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8073879480361938, + "num_tokens": 573918457.0, + "step": 15046 + }, + { + "epoch": 1.9141330619514056, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.5927, + "mean_token_accuracy": 0.8192766904830933, + "num_tokens": 573958818.0, + "step": 15047 + }, + { + "epoch": 1.9142602722299962, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7938248515129089, + "num_tokens": 573996735.0, + "step": 15048 + }, + { + "epoch": 1.9143874825085867, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8057427406311035, + "num_tokens": 574035005.0, + "step": 15049 + }, + { + "epoch": 1.9145146927871772, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8019586205482483, + "num_tokens": 574077817.0, + "step": 15050 + }, + { + "epoch": 1.9146419030657678, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8042738437652588, + "num_tokens": 574113083.0, + "step": 15051 + }, + { + "epoch": 1.914769113344358, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.79754239320755, + "num_tokens": 574147324.0, + "step": 15052 + }, + { + "epoch": 1.9148963236229486, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.8175608515739441, + "num_tokens": 574186108.0, + "step": 15053 + }, + { + "epoch": 1.9150235339015391, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8117944598197937, + "num_tokens": 574218076.0, + "step": 15054 + }, + { + "epoch": 1.9151507441801296, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8132901787757874, + "num_tokens": 574256416.0, + "step": 15055 + }, + { + "epoch": 1.9152779544587202, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8147767782211304, + "num_tokens": 574295262.0, + "step": 15056 + }, + { + "epoch": 1.9154051647373107, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7881907820701599, + "num_tokens": 574329836.0, + "step": 15057 + }, + { + "epoch": 1.9155323750159012, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8014473915100098, + "num_tokens": 574363189.0, + "step": 15058 + }, + { + "epoch": 1.9156595852944918, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8093296885490417, + "num_tokens": 574396374.0, + "step": 15059 + }, + { + "epoch": 1.9157867955730823, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.801460862159729, + "num_tokens": 574435488.0, + "step": 15060 + }, + { + "epoch": 1.9159140058516728, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.8018308281898499, + "num_tokens": 574475261.0, + "step": 15061 + }, + { + "epoch": 1.9160412161302633, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8249857425689697, + "num_tokens": 574513531.0, + "step": 15062 + }, + { + "epoch": 1.9161684264088539, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.787356972694397, + "num_tokens": 574551736.0, + "step": 15063 + }, + { + "epoch": 1.9162956366874444, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7943655252456665, + "num_tokens": 574589978.0, + "step": 15064 + }, + { + "epoch": 1.916422846966035, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7780224084854126, + "num_tokens": 574628845.0, + "step": 15065 + }, + { + "epoch": 1.9165500572446255, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7948322892189026, + "num_tokens": 574660700.0, + "step": 15066 + }, + { + "epoch": 1.916677267523216, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7896307706832886, + "num_tokens": 574692195.0, + "step": 15067 + }, + { + "epoch": 1.9168044778018065, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7880674600601196, + "num_tokens": 574732462.0, + "step": 15068 + }, + { + "epoch": 1.916931688080397, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8106067180633545, + "num_tokens": 574769353.0, + "step": 15069 + }, + { + "epoch": 1.9170588983589874, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7987437844276428, + "num_tokens": 574804958.0, + "step": 15070 + }, + { + "epoch": 1.9171861086375779, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7996745705604553, + "num_tokens": 574843966.0, + "step": 15071 + }, + { + "epoch": 1.9173133189161684, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8057966828346252, + "num_tokens": 574882916.0, + "step": 15072 + }, + { + "epoch": 1.917440529194759, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8114521503448486, + "num_tokens": 574922154.0, + "step": 15073 + }, + { + "epoch": 1.9175677394733495, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7962279319763184, + "num_tokens": 574956894.0, + "step": 15074 + }, + { + "epoch": 1.91769494975194, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.824240505695343, + "num_tokens": 574998343.0, + "step": 15075 + }, + { + "epoch": 1.9178221600305305, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8120594620704651, + "num_tokens": 575036689.0, + "step": 15076 + }, + { + "epoch": 1.9179493703091208, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7978659272193909, + "num_tokens": 575075510.0, + "step": 15077 + }, + { + "epoch": 1.9180765805877114, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7964005470275879, + "num_tokens": 575115797.0, + "step": 15078 + }, + { + "epoch": 1.9182037908663019, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7931594848632812, + "num_tokens": 575155240.0, + "step": 15079 + }, + { + "epoch": 1.9183310011448924, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7823737859725952, + "num_tokens": 575191048.0, + "step": 15080 + }, + { + "epoch": 1.918458211423483, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7903919816017151, + "num_tokens": 575227778.0, + "step": 15081 + }, + { + "epoch": 1.9185854217020735, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.8011665940284729, + "num_tokens": 575269774.0, + "step": 15082 + }, + { + "epoch": 1.918712631980664, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7851792573928833, + "num_tokens": 575310710.0, + "step": 15083 + }, + { + "epoch": 1.9188398422592545, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.810472846031189, + "num_tokens": 575344073.0, + "step": 15084 + }, + { + "epoch": 1.918967052537845, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.795124888420105, + "num_tokens": 575386737.0, + "step": 15085 + }, + { + "epoch": 1.9190942628164356, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8029718995094299, + "num_tokens": 575424287.0, + "step": 15086 + }, + { + "epoch": 1.919221473095026, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8047928810119629, + "num_tokens": 575463859.0, + "step": 15087 + }, + { + "epoch": 1.9193486833736166, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7928321361541748, + "num_tokens": 575500838.0, + "step": 15088 + }, + { + "epoch": 1.9194758936522072, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7893658876419067, + "num_tokens": 575535164.0, + "step": 15089 + }, + { + "epoch": 1.9196031039307977, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8002663254737854, + "num_tokens": 575569493.0, + "step": 15090 + }, + { + "epoch": 1.9197303142093882, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7790415287017822, + "num_tokens": 575603396.0, + "step": 15091 + }, + { + "epoch": 1.9198575244879788, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8072351217269897, + "num_tokens": 575638255.0, + "step": 15092 + }, + { + "epoch": 1.9199847347665693, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7907599210739136, + "num_tokens": 575685281.0, + "step": 15093 + }, + { + "epoch": 1.9201119450451598, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8083691596984863, + "num_tokens": 575725030.0, + "step": 15094 + }, + { + "epoch": 1.9202391553237501, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7915453910827637, + "num_tokens": 575764237.0, + "step": 15095 + }, + { + "epoch": 1.9203663656023406, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7958703637123108, + "num_tokens": 575796732.0, + "step": 15096 + }, + { + "epoch": 1.9204935758809312, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8049771785736084, + "num_tokens": 575833133.0, + "step": 15097 + }, + { + "epoch": 1.9206207861595217, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7947526574134827, + "num_tokens": 575870920.0, + "step": 15098 + }, + { + "epoch": 1.9207479964381122, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8021865487098694, + "num_tokens": 575909132.0, + "step": 15099 + }, + { + "epoch": 1.9208752067167028, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8066874742507935, + "num_tokens": 575947979.0, + "step": 15100 + }, + { + "epoch": 1.921002416995293, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7927736639976501, + "num_tokens": 575988888.0, + "step": 15101 + }, + { + "epoch": 1.9211296272738836, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8051517605781555, + "num_tokens": 576028614.0, + "step": 15102 + }, + { + "epoch": 1.9212568375524741, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.5582, + "mean_token_accuracy": 0.8301393985748291, + "num_tokens": 576066648.0, + "step": 15103 + }, + { + "epoch": 1.9213840478310646, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7925126552581787, + "num_tokens": 576104045.0, + "step": 15104 + }, + { + "epoch": 1.9215112581096552, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.797972559928894, + "num_tokens": 576145419.0, + "step": 15105 + }, + { + "epoch": 1.9216384683882457, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7985671758651733, + "num_tokens": 576191410.0, + "step": 15106 + }, + { + "epoch": 1.9217656786668362, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7925950288772583, + "num_tokens": 576226383.0, + "step": 15107 + }, + { + "epoch": 1.9218928889454268, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7895994186401367, + "num_tokens": 576259958.0, + "step": 15108 + }, + { + "epoch": 1.9220200992240173, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8107391595840454, + "num_tokens": 576299467.0, + "step": 15109 + }, + { + "epoch": 1.9221473095026078, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7902296781539917, + "num_tokens": 576334137.0, + "step": 15110 + }, + { + "epoch": 1.9222745197811983, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7925975322723389, + "num_tokens": 576379582.0, + "step": 15111 + }, + { + "epoch": 1.9224017300597889, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8145232200622559, + "num_tokens": 576414590.0, + "step": 15112 + }, + { + "epoch": 1.9225289403383794, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7879486083984375, + "num_tokens": 576445023.0, + "step": 15113 + }, + { + "epoch": 1.92265615061697, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7957706451416016, + "num_tokens": 576477861.0, + "step": 15114 + }, + { + "epoch": 1.9227833608955605, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7920310497283936, + "num_tokens": 576513121.0, + "step": 15115 + }, + { + "epoch": 1.922910571174151, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8023929595947266, + "num_tokens": 576551039.0, + "step": 15116 + }, + { + "epoch": 1.9230377814527415, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7923582196235657, + "num_tokens": 576590651.0, + "step": 15117 + }, + { + "epoch": 1.923164991731332, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7887639999389648, + "num_tokens": 576629119.0, + "step": 15118 + }, + { + "epoch": 1.9232922020099223, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8074556589126587, + "num_tokens": 576664240.0, + "step": 15119 + }, + { + "epoch": 1.9234194122885129, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7900151610374451, + "num_tokens": 576701946.0, + "step": 15120 + }, + { + "epoch": 1.9235466225671034, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8102136850357056, + "num_tokens": 576733541.0, + "step": 15121 + }, + { + "epoch": 1.923673832845694, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7828799486160278, + "num_tokens": 576770930.0, + "step": 15122 + }, + { + "epoch": 1.9238010431242845, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8033792972564697, + "num_tokens": 576807455.0, + "step": 15123 + }, + { + "epoch": 1.923928253402875, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8034929633140564, + "num_tokens": 576849577.0, + "step": 15124 + }, + { + "epoch": 1.9240554636814655, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7935952544212341, + "num_tokens": 576883506.0, + "step": 15125 + }, + { + "epoch": 1.9241826739600558, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7803905010223389, + "num_tokens": 576923892.0, + "step": 15126 + }, + { + "epoch": 1.9243098842386464, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.5904, + "mean_token_accuracy": 0.8213505148887634, + "num_tokens": 576970832.0, + "step": 15127 + }, + { + "epoch": 1.9244370945172369, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7738041877746582, + "num_tokens": 577006990.0, + "step": 15128 + }, + { + "epoch": 1.9245643047958274, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8032115697860718, + "num_tokens": 577044932.0, + "step": 15129 + }, + { + "epoch": 1.924691515074418, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7806966304779053, + "num_tokens": 577076217.0, + "step": 15130 + }, + { + "epoch": 1.9248187253530085, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8147662878036499, + "num_tokens": 577112696.0, + "step": 15131 + }, + { + "epoch": 1.924945935631599, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8042893409729004, + "num_tokens": 577156397.0, + "step": 15132 + }, + { + "epoch": 1.9250731459101895, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8009247779846191, + "num_tokens": 577194192.0, + "step": 15133 + }, + { + "epoch": 1.92520035618878, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7934356927871704, + "num_tokens": 577232639.0, + "step": 15134 + }, + { + "epoch": 1.9253275664673706, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8062208294868469, + "num_tokens": 577267349.0, + "step": 15135 + }, + { + "epoch": 1.925454776745961, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7939106225967407, + "num_tokens": 577308329.0, + "step": 15136 + }, + { + "epoch": 1.9255819870245516, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8042259812355042, + "num_tokens": 577347825.0, + "step": 15137 + }, + { + "epoch": 1.9257091973031422, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7967949509620667, + "num_tokens": 577379981.0, + "step": 15138 + }, + { + "epoch": 1.9258364075817327, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8035438656806946, + "num_tokens": 577414134.0, + "step": 15139 + }, + { + "epoch": 1.9259636178603232, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7983437776565552, + "num_tokens": 577451170.0, + "step": 15140 + }, + { + "epoch": 1.9260908281389137, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7830791473388672, + "num_tokens": 577487447.0, + "step": 15141 + }, + { + "epoch": 1.9262180384175043, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8097392320632935, + "num_tokens": 577525732.0, + "step": 15142 + }, + { + "epoch": 1.9263452486960948, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.796090304851532, + "num_tokens": 577563303.0, + "step": 15143 + }, + { + "epoch": 1.926472458974685, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.8199139833450317, + "num_tokens": 577607592.0, + "step": 15144 + }, + { + "epoch": 1.9265996692532756, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7953515648841858, + "num_tokens": 577651646.0, + "step": 15145 + }, + { + "epoch": 1.9267268795318662, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8101422786712646, + "num_tokens": 577689530.0, + "step": 15146 + }, + { + "epoch": 1.9268540898104567, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7954198122024536, + "num_tokens": 577732524.0, + "step": 15147 + }, + { + "epoch": 1.9269813000890472, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7999449968338013, + "num_tokens": 577767164.0, + "step": 15148 + }, + { + "epoch": 1.9271085103676378, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8064099550247192, + "num_tokens": 577802347.0, + "step": 15149 + }, + { + "epoch": 1.927235720646228, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.792972207069397, + "num_tokens": 577842899.0, + "step": 15150 + }, + { + "epoch": 1.9273629309248186, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8060142397880554, + "num_tokens": 577885137.0, + "step": 15151 + }, + { + "epoch": 1.9274901412034091, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8148350715637207, + "num_tokens": 577925727.0, + "step": 15152 + }, + { + "epoch": 1.9276173514819996, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8043503165245056, + "num_tokens": 577965399.0, + "step": 15153 + }, + { + "epoch": 1.9277445617605902, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7966489791870117, + "num_tokens": 578007608.0, + "step": 15154 + }, + { + "epoch": 1.9278717720391807, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8078501224517822, + "num_tokens": 578043425.0, + "step": 15155 + }, + { + "epoch": 1.9279989823177712, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7945651412010193, + "num_tokens": 578075838.0, + "step": 15156 + }, + { + "epoch": 1.9281261925963618, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7925552129745483, + "num_tokens": 578110113.0, + "step": 15157 + }, + { + "epoch": 1.9282534028749523, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7994507551193237, + "num_tokens": 578145743.0, + "step": 15158 + }, + { + "epoch": 1.9283806131535428, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8040101528167725, + "num_tokens": 578181588.0, + "step": 15159 + }, + { + "epoch": 1.9285078234321333, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7998135089874268, + "num_tokens": 578222048.0, + "step": 15160 + }, + { + "epoch": 1.9286350337107239, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8054364323616028, + "num_tokens": 578262664.0, + "step": 15161 + }, + { + "epoch": 1.9287622439893144, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7948100566864014, + "num_tokens": 578303431.0, + "step": 15162 + }, + { + "epoch": 1.928889454267905, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.783990740776062, + "num_tokens": 578338978.0, + "step": 15163 + }, + { + "epoch": 1.9290166645464955, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7962324619293213, + "num_tokens": 578376340.0, + "step": 15164 + }, + { + "epoch": 1.929143874825086, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8039321303367615, + "num_tokens": 578414714.0, + "step": 15165 + }, + { + "epoch": 1.9292710851036765, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7896666526794434, + "num_tokens": 578453668.0, + "step": 15166 + }, + { + "epoch": 1.929398295382267, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7959840893745422, + "num_tokens": 578494148.0, + "step": 15167 + }, + { + "epoch": 1.9295255056608573, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7797759771347046, + "num_tokens": 578530743.0, + "step": 15168 + }, + { + "epoch": 1.9296527159394479, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.8235630989074707, + "num_tokens": 578567613.0, + "step": 15169 + }, + { + "epoch": 1.9297799262180384, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8010703325271606, + "num_tokens": 578602981.0, + "step": 15170 + }, + { + "epoch": 1.929907136496629, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7913590669631958, + "num_tokens": 578641027.0, + "step": 15171 + }, + { + "epoch": 1.9300343467752195, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8154424428939819, + "num_tokens": 578683006.0, + "step": 15172 + }, + { + "epoch": 1.93016155705381, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8025079965591431, + "num_tokens": 578716722.0, + "step": 15173 + }, + { + "epoch": 1.9302887673324005, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7784397602081299, + "num_tokens": 578750740.0, + "step": 15174 + }, + { + "epoch": 1.9304159776109908, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8055317401885986, + "num_tokens": 578791358.0, + "step": 15175 + }, + { + "epoch": 1.9305431878895813, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7841200232505798, + "num_tokens": 578834514.0, + "step": 15176 + }, + { + "epoch": 1.9306703981681719, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7954827547073364, + "num_tokens": 578875869.0, + "step": 15177 + }, + { + "epoch": 1.9307976084467624, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.804351806640625, + "num_tokens": 578921079.0, + "step": 15178 + }, + { + "epoch": 1.930924818725353, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7982685565948486, + "num_tokens": 578959837.0, + "step": 15179 + }, + { + "epoch": 1.9310520290039435, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7680176496505737, + "num_tokens": 579001629.0, + "step": 15180 + }, + { + "epoch": 1.931179239282534, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.5725, + "mean_token_accuracy": 0.8254643082618713, + "num_tokens": 579036762.0, + "step": 15181 + }, + { + "epoch": 1.9313064495611245, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7742857933044434, + "num_tokens": 579078455.0, + "step": 15182 + }, + { + "epoch": 1.931433659839715, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8028422594070435, + "num_tokens": 579116942.0, + "step": 15183 + }, + { + "epoch": 1.9315608701183056, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8158625364303589, + "num_tokens": 579155180.0, + "step": 15184 + }, + { + "epoch": 1.931688080396896, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7929682731628418, + "num_tokens": 579194336.0, + "step": 15185 + }, + { + "epoch": 1.9318152906754866, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7874023914337158, + "num_tokens": 579226609.0, + "step": 15186 + }, + { + "epoch": 1.9319425009540772, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8021906614303589, + "num_tokens": 579259288.0, + "step": 15187 + }, + { + "epoch": 1.9320697112326677, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8074660897254944, + "num_tokens": 579294797.0, + "step": 15188 + }, + { + "epoch": 1.9321969215112582, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8063499927520752, + "num_tokens": 579334352.0, + "step": 15189 + }, + { + "epoch": 1.9323241317898487, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7910004258155823, + "num_tokens": 579379618.0, + "step": 15190 + }, + { + "epoch": 1.9324513420684393, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7993679046630859, + "num_tokens": 579420230.0, + "step": 15191 + }, + { + "epoch": 1.9325785523470298, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8002926111221313, + "num_tokens": 579458236.0, + "step": 15192 + }, + { + "epoch": 1.93270576262562, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7974520921707153, + "num_tokens": 579495561.0, + "step": 15193 + }, + { + "epoch": 1.9328329729042106, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7916844487190247, + "num_tokens": 579530225.0, + "step": 15194 + }, + { + "epoch": 1.9329601831828012, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7935845255851746, + "num_tokens": 579570143.0, + "step": 15195 + }, + { + "epoch": 1.9330873934613917, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8116986751556396, + "num_tokens": 579607105.0, + "step": 15196 + }, + { + "epoch": 1.9332146037399822, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8104093074798584, + "num_tokens": 579640018.0, + "step": 15197 + }, + { + "epoch": 1.9333418140185727, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7953189611434937, + "num_tokens": 579681250.0, + "step": 15198 + }, + { + "epoch": 1.933469024297163, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7973803281784058, + "num_tokens": 579718535.0, + "step": 15199 + }, + { + "epoch": 1.9335962345757536, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7894912958145142, + "num_tokens": 579756959.0, + "step": 15200 + }, + { + "epoch": 1.933723444854344, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7947859764099121, + "num_tokens": 579798266.0, + "step": 15201 + }, + { + "epoch": 1.9338506551329346, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8215304613113403, + "num_tokens": 579831869.0, + "step": 15202 + }, + { + "epoch": 1.9339778654115252, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8029717206954956, + "num_tokens": 579874168.0, + "step": 15203 + }, + { + "epoch": 1.9341050756901157, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8041027784347534, + "num_tokens": 579916313.0, + "step": 15204 + }, + { + "epoch": 1.9342322859687062, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8127281665802002, + "num_tokens": 579959022.0, + "step": 15205 + }, + { + "epoch": 1.9343594962472968, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7887840270996094, + "num_tokens": 580003675.0, + "step": 15206 + }, + { + "epoch": 1.9344867065258873, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.8140353560447693, + "num_tokens": 580044384.0, + "step": 15207 + }, + { + "epoch": 1.9346139168044778, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7897070050239563, + "num_tokens": 580078955.0, + "step": 15208 + }, + { + "epoch": 1.9347411270830683, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8054967522621155, + "num_tokens": 580116793.0, + "step": 15209 + }, + { + "epoch": 1.9348683373616589, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.800485372543335, + "num_tokens": 580153622.0, + "step": 15210 + }, + { + "epoch": 1.9349955476402494, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7996363639831543, + "num_tokens": 580195287.0, + "step": 15211 + }, + { + "epoch": 1.93512275791884, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.782757580280304, + "num_tokens": 580229601.0, + "step": 15212 + }, + { + "epoch": 1.9352499681974304, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8029119372367859, + "num_tokens": 580268079.0, + "step": 15213 + }, + { + "epoch": 1.935377178476021, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7802767753601074, + "num_tokens": 580305379.0, + "step": 15214 + }, + { + "epoch": 1.9355043887546115, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7972123026847839, + "num_tokens": 580338438.0, + "step": 15215 + }, + { + "epoch": 1.935631599033202, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.8000868558883667, + "num_tokens": 580370017.0, + "step": 15216 + }, + { + "epoch": 1.9357588093117923, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.817161500453949, + "num_tokens": 580409572.0, + "step": 15217 + }, + { + "epoch": 1.9358860195903829, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8048067688941956, + "num_tokens": 580448481.0, + "step": 15218 + }, + { + "epoch": 1.9360132298689734, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8129873275756836, + "num_tokens": 580481919.0, + "step": 15219 + }, + { + "epoch": 1.936140440147564, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.793666660785675, + "num_tokens": 580520758.0, + "step": 15220 + }, + { + "epoch": 1.9362676504261545, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.790553092956543, + "num_tokens": 580559839.0, + "step": 15221 + }, + { + "epoch": 1.936394860704745, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.5734, + "mean_token_accuracy": 0.8268211483955383, + "num_tokens": 580591767.0, + "step": 15222 + }, + { + "epoch": 1.9365220709833355, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8071386814117432, + "num_tokens": 580633180.0, + "step": 15223 + }, + { + "epoch": 1.9366492812619258, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8040403127670288, + "num_tokens": 580677045.0, + "step": 15224 + }, + { + "epoch": 1.9367764915405163, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8131281137466431, + "num_tokens": 580717423.0, + "step": 15225 + }, + { + "epoch": 1.9369037018191069, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8010826110839844, + "num_tokens": 580756677.0, + "step": 15226 + }, + { + "epoch": 1.9370309120976974, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8090076446533203, + "num_tokens": 580796993.0, + "step": 15227 + }, + { + "epoch": 1.937158122376288, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7989060878753662, + "num_tokens": 580834616.0, + "step": 15228 + }, + { + "epoch": 1.9372853326548785, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8067893981933594, + "num_tokens": 580873718.0, + "step": 15229 + }, + { + "epoch": 1.937412542933469, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7975218892097473, + "num_tokens": 580911678.0, + "step": 15230 + }, + { + "epoch": 1.9375397532120595, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8048862218856812, + "num_tokens": 580953282.0, + "step": 15231 + }, + { + "epoch": 1.93766696349065, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8036134243011475, + "num_tokens": 580997954.0, + "step": 15232 + }, + { + "epoch": 1.9377941737692406, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.801466703414917, + "num_tokens": 581035623.0, + "step": 15233 + }, + { + "epoch": 1.937921384047831, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8060516119003296, + "num_tokens": 581068727.0, + "step": 15234 + }, + { + "epoch": 1.9380485943264216, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7999299764633179, + "num_tokens": 581111180.0, + "step": 15235 + }, + { + "epoch": 1.9381758046050122, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7819511890411377, + "num_tokens": 581152373.0, + "step": 15236 + }, + { + "epoch": 1.9383030148836027, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7977514266967773, + "num_tokens": 581195854.0, + "step": 15237 + }, + { + "epoch": 1.9384302251621932, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7681325674057007, + "num_tokens": 581236414.0, + "step": 15238 + }, + { + "epoch": 1.9385574354407837, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7839936017990112, + "num_tokens": 581275349.0, + "step": 15239 + }, + { + "epoch": 1.9386846457193743, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8076286315917969, + "num_tokens": 581313231.0, + "step": 15240 + }, + { + "epoch": 1.9388118559979648, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7996402978897095, + "num_tokens": 581355295.0, + "step": 15241 + }, + { + "epoch": 1.938939066276555, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8025132417678833, + "num_tokens": 581392633.0, + "step": 15242 + }, + { + "epoch": 1.9390662765551456, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8133379817008972, + "num_tokens": 581425552.0, + "step": 15243 + }, + { + "epoch": 1.9391934868337362, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.784263014793396, + "num_tokens": 581467707.0, + "step": 15244 + }, + { + "epoch": 1.9393206971123267, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8061952590942383, + "num_tokens": 581504888.0, + "step": 15245 + }, + { + "epoch": 1.9394479073909172, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7951684594154358, + "num_tokens": 581537800.0, + "step": 15246 + }, + { + "epoch": 1.9395751176695077, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8198295831680298, + "num_tokens": 581570594.0, + "step": 15247 + }, + { + "epoch": 1.939702327948098, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7920569181442261, + "num_tokens": 581610764.0, + "step": 15248 + }, + { + "epoch": 1.9398295382266886, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8075742721557617, + "num_tokens": 581652341.0, + "step": 15249 + }, + { + "epoch": 1.939956748505279, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7950685024261475, + "num_tokens": 581689978.0, + "step": 15250 + }, + { + "epoch": 1.9400839587838696, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8109116554260254, + "num_tokens": 581728811.0, + "step": 15251 + }, + { + "epoch": 1.9402111690624602, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8084421157836914, + "num_tokens": 581768016.0, + "step": 15252 + }, + { + "epoch": 1.9403383793410507, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.8041260242462158, + "num_tokens": 581805405.0, + "step": 15253 + }, + { + "epoch": 1.9404655896196412, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7790000438690186, + "num_tokens": 581845302.0, + "step": 15254 + }, + { + "epoch": 1.9405927998982317, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8066023588180542, + "num_tokens": 581875391.0, + "step": 15255 + }, + { + "epoch": 1.9407200101768223, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7988911867141724, + "num_tokens": 581916580.0, + "step": 15256 + }, + { + "epoch": 1.9408472204554128, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8011635541915894, + "num_tokens": 581958539.0, + "step": 15257 + }, + { + "epoch": 1.9409744307340033, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7974405288696289, + "num_tokens": 581996981.0, + "step": 15258 + }, + { + "epoch": 1.9411016410125939, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8079061508178711, + "num_tokens": 582036068.0, + "step": 15259 + }, + { + "epoch": 1.9412288512911844, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7978771924972534, + "num_tokens": 582073322.0, + "step": 15260 + }, + { + "epoch": 1.941356061569775, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8040997982025146, + "num_tokens": 582110736.0, + "step": 15261 + }, + { + "epoch": 1.9414832718483654, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7810398936271667, + "num_tokens": 582153431.0, + "step": 15262 + }, + { + "epoch": 1.941610482126956, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7784426212310791, + "num_tokens": 582191793.0, + "step": 15263 + }, + { + "epoch": 1.9417376924055465, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.787461519241333, + "num_tokens": 582227923.0, + "step": 15264 + }, + { + "epoch": 1.941864902684137, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7820178270339966, + "num_tokens": 582268029.0, + "step": 15265 + }, + { + "epoch": 1.9419921129627273, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7944437265396118, + "num_tokens": 582302173.0, + "step": 15266 + }, + { + "epoch": 1.9421193232413179, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.8201651573181152, + "num_tokens": 582341642.0, + "step": 15267 + }, + { + "epoch": 1.9422465335199084, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.799868106842041, + "num_tokens": 582378960.0, + "step": 15268 + }, + { + "epoch": 1.942373743798499, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.810792088508606, + "num_tokens": 582422149.0, + "step": 15269 + }, + { + "epoch": 1.9425009540770894, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7855381965637207, + "num_tokens": 582462657.0, + "step": 15270 + }, + { + "epoch": 1.94262816435568, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8118817210197449, + "num_tokens": 582499264.0, + "step": 15271 + }, + { + "epoch": 1.9427553746342705, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7994979023933411, + "num_tokens": 582534744.0, + "step": 15272 + }, + { + "epoch": 1.9428825849128608, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7951278686523438, + "num_tokens": 582577010.0, + "step": 15273 + }, + { + "epoch": 1.9430097951914513, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.790812611579895, + "num_tokens": 582617103.0, + "step": 15274 + }, + { + "epoch": 1.9431370054700419, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7937682867050171, + "num_tokens": 582661754.0, + "step": 15275 + }, + { + "epoch": 1.9432642157486324, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8036580681800842, + "num_tokens": 582701701.0, + "step": 15276 + }, + { + "epoch": 1.943391426027223, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.790023922920227, + "num_tokens": 582741977.0, + "step": 15277 + }, + { + "epoch": 1.9435186363058135, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8097529411315918, + "num_tokens": 582783420.0, + "step": 15278 + }, + { + "epoch": 1.943645846584404, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8067142367362976, + "num_tokens": 582819464.0, + "step": 15279 + }, + { + "epoch": 1.9437730568629945, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8051410913467407, + "num_tokens": 582860757.0, + "step": 15280 + }, + { + "epoch": 1.943900267141585, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.794392466545105, + "num_tokens": 582910300.0, + "step": 15281 + }, + { + "epoch": 1.9440274774201756, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7995024919509888, + "num_tokens": 582956333.0, + "step": 15282 + }, + { + "epoch": 1.944154687698766, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8083682656288147, + "num_tokens": 582994068.0, + "step": 15283 + }, + { + "epoch": 1.9442818979773566, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7956860065460205, + "num_tokens": 583027829.0, + "step": 15284 + }, + { + "epoch": 1.9444091082559471, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7962902784347534, + "num_tokens": 583068881.0, + "step": 15285 + }, + { + "epoch": 1.9445363185345377, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8008212447166443, + "num_tokens": 583100097.0, + "step": 15286 + }, + { + "epoch": 1.9446635288131282, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7968407869338989, + "num_tokens": 583135340.0, + "step": 15287 + }, + { + "epoch": 1.9447907390917187, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.799184262752533, + "num_tokens": 583172032.0, + "step": 15288 + }, + { + "epoch": 1.9449179493703093, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8036578893661499, + "num_tokens": 583205244.0, + "step": 15289 + }, + { + "epoch": 1.9450451596488998, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7946068048477173, + "num_tokens": 583247537.0, + "step": 15290 + }, + { + "epoch": 1.94517236992749, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7944943904876709, + "num_tokens": 583285349.0, + "step": 15291 + }, + { + "epoch": 1.9452995802060806, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8077381253242493, + "num_tokens": 583323810.0, + "step": 15292 + }, + { + "epoch": 1.9454267904846712, + "grad_norm": 71.0, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8066294193267822, + "num_tokens": 583354244.0, + "step": 15293 + }, + { + "epoch": 1.9455540007632617, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8192951679229736, + "num_tokens": 583391764.0, + "step": 15294 + }, + { + "epoch": 1.9456812110418522, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7892031669616699, + "num_tokens": 583426305.0, + "step": 15295 + }, + { + "epoch": 1.9458084213204427, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7932838201522827, + "num_tokens": 583465252.0, + "step": 15296 + }, + { + "epoch": 1.945935631599033, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7889528274536133, + "num_tokens": 583510001.0, + "step": 15297 + }, + { + "epoch": 1.9460628418776236, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7990228533744812, + "num_tokens": 583547380.0, + "step": 15298 + }, + { + "epoch": 1.946190052156214, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8025559186935425, + "num_tokens": 583588852.0, + "step": 15299 + }, + { + "epoch": 1.9463172624348046, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8109011054039001, + "num_tokens": 583634006.0, + "step": 15300 + }, + { + "epoch": 1.9464444727133952, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7862505912780762, + "num_tokens": 583679430.0, + "step": 15301 + }, + { + "epoch": 1.9465716829919857, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8011016845703125, + "num_tokens": 583708856.0, + "step": 15302 + }, + { + "epoch": 1.9466988932705762, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7957252860069275, + "num_tokens": 583750737.0, + "step": 15303 + }, + { + "epoch": 1.9468261035491667, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.804069995880127, + "num_tokens": 583786342.0, + "step": 15304 + }, + { + "epoch": 1.9469533138277573, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8036351799964905, + "num_tokens": 583821681.0, + "step": 15305 + }, + { + "epoch": 1.9470805241063478, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7759789228439331, + "num_tokens": 583855640.0, + "step": 15306 + }, + { + "epoch": 1.9472077343849383, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8086698651313782, + "num_tokens": 583888047.0, + "step": 15307 + }, + { + "epoch": 1.9473349446635289, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7955632209777832, + "num_tokens": 583923220.0, + "step": 15308 + }, + { + "epoch": 1.9474621549421194, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.796714723110199, + "num_tokens": 583965718.0, + "step": 15309 + }, + { + "epoch": 1.94758936522071, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8026362061500549, + "num_tokens": 584014027.0, + "step": 15310 + }, + { + "epoch": 1.9477165754993004, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8022946715354919, + "num_tokens": 584045019.0, + "step": 15311 + }, + { + "epoch": 1.947843785777891, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7813570499420166, + "num_tokens": 584086839.0, + "step": 15312 + }, + { + "epoch": 1.9479709960564815, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8192766904830933, + "num_tokens": 584123785.0, + "step": 15313 + }, + { + "epoch": 1.948098206335072, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7914935350418091, + "num_tokens": 584163768.0, + "step": 15314 + }, + { + "epoch": 1.9482254166136623, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7980806827545166, + "num_tokens": 584201668.0, + "step": 15315 + }, + { + "epoch": 1.9483526268922529, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.8097164630889893, + "num_tokens": 584236626.0, + "step": 15316 + }, + { + "epoch": 1.9484798371708434, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7849228382110596, + "num_tokens": 584279811.0, + "step": 15317 + }, + { + "epoch": 1.948607047449434, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7998926639556885, + "num_tokens": 584317314.0, + "step": 15318 + }, + { + "epoch": 1.9487342577280244, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8042218685150146, + "num_tokens": 584355768.0, + "step": 15319 + }, + { + "epoch": 1.948861468006615, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.791824996471405, + "num_tokens": 584386205.0, + "step": 15320 + }, + { + "epoch": 1.9489886782852053, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8098822832107544, + "num_tokens": 584419721.0, + "step": 15321 + }, + { + "epoch": 1.9491158885637958, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7942846417427063, + "num_tokens": 584451029.0, + "step": 15322 + }, + { + "epoch": 1.9492430988423863, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8111506700515747, + "num_tokens": 584483991.0, + "step": 15323 + }, + { + "epoch": 1.9493703091209769, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8001887202262878, + "num_tokens": 584517652.0, + "step": 15324 + }, + { + "epoch": 1.9494975193995674, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8112703561782837, + "num_tokens": 584561058.0, + "step": 15325 + }, + { + "epoch": 1.949624729678158, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8041635751724243, + "num_tokens": 584595097.0, + "step": 15326 + }, + { + "epoch": 1.9497519399567484, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.810560941696167, + "num_tokens": 584639073.0, + "step": 15327 + }, + { + "epoch": 1.949879150235339, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7872060537338257, + "num_tokens": 584680517.0, + "step": 15328 + }, + { + "epoch": 1.9500063605139295, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8005308508872986, + "num_tokens": 584717796.0, + "step": 15329 + }, + { + "epoch": 1.95013357079252, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8137571215629578, + "num_tokens": 584755595.0, + "step": 15330 + }, + { + "epoch": 1.9502607810711106, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7894439697265625, + "num_tokens": 584798681.0, + "step": 15331 + }, + { + "epoch": 1.950387991349701, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8008764982223511, + "num_tokens": 584835424.0, + "step": 15332 + }, + { + "epoch": 1.9505152016282916, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8117722272872925, + "num_tokens": 584872063.0, + "step": 15333 + }, + { + "epoch": 1.9506424119068821, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.818976640701294, + "num_tokens": 584905850.0, + "step": 15334 + }, + { + "epoch": 1.9507696221854727, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8118544816970825, + "num_tokens": 584946480.0, + "step": 15335 + }, + { + "epoch": 1.9508968324640632, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8058345317840576, + "num_tokens": 584977382.0, + "step": 15336 + }, + { + "epoch": 1.9510240427426537, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8034977912902832, + "num_tokens": 585010885.0, + "step": 15337 + }, + { + "epoch": 1.9511512530212443, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7991172075271606, + "num_tokens": 585048208.0, + "step": 15338 + }, + { + "epoch": 1.9512784632998348, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.794487476348877, + "num_tokens": 585087477.0, + "step": 15339 + }, + { + "epoch": 1.951405673578425, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7941795587539673, + "num_tokens": 585120860.0, + "step": 15340 + }, + { + "epoch": 1.9515328838570156, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8035430908203125, + "num_tokens": 585161887.0, + "step": 15341 + }, + { + "epoch": 1.9516600941356061, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8122952580451965, + "num_tokens": 585200496.0, + "step": 15342 + }, + { + "epoch": 1.9517873044141967, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.800269365310669, + "num_tokens": 585243976.0, + "step": 15343 + }, + { + "epoch": 1.9519145146927872, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8037840127944946, + "num_tokens": 585282366.0, + "step": 15344 + }, + { + "epoch": 1.9520417249713777, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8025156259536743, + "num_tokens": 585325224.0, + "step": 15345 + }, + { + "epoch": 1.952168935249968, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8075635433197021, + "num_tokens": 585361954.0, + "step": 15346 + }, + { + "epoch": 1.9522961455285586, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7967036366462708, + "num_tokens": 585406033.0, + "step": 15347 + }, + { + "epoch": 1.952423355807149, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7976495623588562, + "num_tokens": 585440457.0, + "step": 15348 + }, + { + "epoch": 1.9525505660857396, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8193516135215759, + "num_tokens": 585473451.0, + "step": 15349 + }, + { + "epoch": 1.9526777763643302, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8033562898635864, + "num_tokens": 585509652.0, + "step": 15350 + }, + { + "epoch": 1.9528049866429207, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8037306666374207, + "num_tokens": 585545591.0, + "step": 15351 + }, + { + "epoch": 1.9529321969215112, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7890152335166931, + "num_tokens": 585583757.0, + "step": 15352 + }, + { + "epoch": 1.9530594072001017, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7853459715843201, + "num_tokens": 585623897.0, + "step": 15353 + }, + { + "epoch": 1.9531866174786923, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7975282669067383, + "num_tokens": 585661043.0, + "step": 15354 + }, + { + "epoch": 1.9533138277572828, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7732865810394287, + "num_tokens": 585699201.0, + "step": 15355 + }, + { + "epoch": 1.9534410380358733, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7870590686798096, + "num_tokens": 585735325.0, + "step": 15356 + }, + { + "epoch": 1.9535682483144639, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.8069095611572266, + "num_tokens": 585769909.0, + "step": 15357 + }, + { + "epoch": 1.9536954585930544, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7943617701530457, + "num_tokens": 585804590.0, + "step": 15358 + }, + { + "epoch": 1.953822668871645, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.8000082969665527, + "num_tokens": 585843349.0, + "step": 15359 + }, + { + "epoch": 1.9539498791502354, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8166048526763916, + "num_tokens": 585881846.0, + "step": 15360 + }, + { + "epoch": 1.954077089428826, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7972412109375, + "num_tokens": 585917848.0, + "step": 15361 + }, + { + "epoch": 1.9542042997074165, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8065149188041687, + "num_tokens": 585961249.0, + "step": 15362 + }, + { + "epoch": 1.954331509986007, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.8144227266311646, + "num_tokens": 586005795.0, + "step": 15363 + }, + { + "epoch": 1.9544587202645973, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8118707537651062, + "num_tokens": 586041523.0, + "step": 15364 + }, + { + "epoch": 1.9545859305431879, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7931894063949585, + "num_tokens": 586077800.0, + "step": 15365 + }, + { + "epoch": 1.9547131408217784, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7975319623947144, + "num_tokens": 586116010.0, + "step": 15366 + }, + { + "epoch": 1.954840351100369, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.816758930683136, + "num_tokens": 586155359.0, + "step": 15367 + }, + { + "epoch": 1.9549675613789594, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8023175001144409, + "num_tokens": 586188190.0, + "step": 15368 + }, + { + "epoch": 1.95509477165755, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.8004882335662842, + "num_tokens": 586225971.0, + "step": 15369 + }, + { + "epoch": 1.9552219819361403, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7959825396537781, + "num_tokens": 586263295.0, + "step": 15370 + }, + { + "epoch": 1.9553491922147308, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.8200399279594421, + "num_tokens": 586293547.0, + "step": 15371 + }, + { + "epoch": 1.9554764024933213, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7957334518432617, + "num_tokens": 586333180.0, + "step": 15372 + }, + { + "epoch": 1.9556036127719119, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7925821542739868, + "num_tokens": 586368780.0, + "step": 15373 + }, + { + "epoch": 1.9557308230505024, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7811303734779358, + "num_tokens": 586404177.0, + "step": 15374 + }, + { + "epoch": 1.955858033329093, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7892961502075195, + "num_tokens": 586446917.0, + "step": 15375 + }, + { + "epoch": 1.9559852436076834, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7978907823562622, + "num_tokens": 586485832.0, + "step": 15376 + }, + { + "epoch": 1.956112453886274, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8106995224952698, + "num_tokens": 586524089.0, + "step": 15377 + }, + { + "epoch": 1.9562396641648645, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8094736337661743, + "num_tokens": 586564097.0, + "step": 15378 + }, + { + "epoch": 1.956366874443455, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8094466924667358, + "num_tokens": 586600653.0, + "step": 15379 + }, + { + "epoch": 1.9564940847220456, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8087870478630066, + "num_tokens": 586639527.0, + "step": 15380 + }, + { + "epoch": 1.956621295000636, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8027575612068176, + "num_tokens": 586670885.0, + "step": 15381 + }, + { + "epoch": 1.9567485052792266, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8199396133422852, + "num_tokens": 586709875.0, + "step": 15382 + }, + { + "epoch": 1.9568757155578171, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7977786064147949, + "num_tokens": 586753164.0, + "step": 15383 + }, + { + "epoch": 1.9570029258364077, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8149168491363525, + "num_tokens": 586793737.0, + "step": 15384 + }, + { + "epoch": 1.9571301361149982, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7937625050544739, + "num_tokens": 586833428.0, + "step": 15385 + }, + { + "epoch": 1.9572573463935887, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7798118591308594, + "num_tokens": 586871202.0, + "step": 15386 + }, + { + "epoch": 1.9573845566721793, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8080414533615112, + "num_tokens": 586905537.0, + "step": 15387 + }, + { + "epoch": 1.9575117669507698, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7897500991821289, + "num_tokens": 586940378.0, + "step": 15388 + }, + { + "epoch": 1.95763897722936, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8087177276611328, + "num_tokens": 586978806.0, + "step": 15389 + }, + { + "epoch": 1.9577661875079506, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7951337099075317, + "num_tokens": 587007367.0, + "step": 15390 + }, + { + "epoch": 1.9578933977865411, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.8197382092475891, + "num_tokens": 587040422.0, + "step": 15391 + }, + { + "epoch": 1.9580206080651317, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8169962763786316, + "num_tokens": 587070238.0, + "step": 15392 + }, + { + "epoch": 1.9581478183437222, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8071739673614502, + "num_tokens": 587105973.0, + "step": 15393 + }, + { + "epoch": 1.9582750286223127, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8137980699539185, + "num_tokens": 587137975.0, + "step": 15394 + }, + { + "epoch": 1.958402238900903, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.793749213218689, + "num_tokens": 587169176.0, + "step": 15395 + }, + { + "epoch": 1.9585294491794936, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.798208475112915, + "num_tokens": 587205570.0, + "step": 15396 + }, + { + "epoch": 1.958656659458084, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7948273420333862, + "num_tokens": 587246729.0, + "step": 15397 + }, + { + "epoch": 1.9587838697366746, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8088313937187195, + "num_tokens": 587284537.0, + "step": 15398 + }, + { + "epoch": 1.9589110800152651, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8077463507652283, + "num_tokens": 587324394.0, + "step": 15399 + }, + { + "epoch": 1.9590382902938557, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.81084805727005, + "num_tokens": 587359935.0, + "step": 15400 + }, + { + "epoch": 1.9591655005724462, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8144422769546509, + "num_tokens": 587402081.0, + "step": 15401 + }, + { + "epoch": 1.9592927108510367, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7999778985977173, + "num_tokens": 587442939.0, + "step": 15402 + }, + { + "epoch": 1.9594199211296273, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7917322516441345, + "num_tokens": 587479155.0, + "step": 15403 + }, + { + "epoch": 1.9595471314082178, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7901246547698975, + "num_tokens": 587519285.0, + "step": 15404 + }, + { + "epoch": 1.9596743416868083, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7853630185127258, + "num_tokens": 587563978.0, + "step": 15405 + }, + { + "epoch": 1.9598015519653988, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7928479909896851, + "num_tokens": 587599052.0, + "step": 15406 + }, + { + "epoch": 1.9599287622439894, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8013360500335693, + "num_tokens": 587639980.0, + "step": 15407 + }, + { + "epoch": 1.96005597252258, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.793130099773407, + "num_tokens": 587673836.0, + "step": 15408 + }, + { + "epoch": 1.9601831828011704, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7982720136642456, + "num_tokens": 587712896.0, + "step": 15409 + }, + { + "epoch": 1.960310393079761, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.790398895740509, + "num_tokens": 587747352.0, + "step": 15410 + }, + { + "epoch": 1.9604376033583515, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8017511367797852, + "num_tokens": 587790234.0, + "step": 15411 + }, + { + "epoch": 1.960564813636942, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.8136042356491089, + "num_tokens": 587830185.0, + "step": 15412 + }, + { + "epoch": 1.9606920239155323, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8014886975288391, + "num_tokens": 587866265.0, + "step": 15413 + }, + { + "epoch": 1.9608192341941229, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8089203834533691, + "num_tokens": 587904956.0, + "step": 15414 + }, + { + "epoch": 1.9609464444727134, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8039290308952332, + "num_tokens": 587942369.0, + "step": 15415 + }, + { + "epoch": 1.961073654751304, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8019180297851562, + "num_tokens": 587979340.0, + "step": 15416 + }, + { + "epoch": 1.9612008650298944, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7983010411262512, + "num_tokens": 588021812.0, + "step": 15417 + }, + { + "epoch": 1.961328075308485, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7997183799743652, + "num_tokens": 588064978.0, + "step": 15418 + }, + { + "epoch": 1.9614552855870753, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7974635362625122, + "num_tokens": 588098142.0, + "step": 15419 + }, + { + "epoch": 1.9615824958656658, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8132942914962769, + "num_tokens": 588141808.0, + "step": 15420 + }, + { + "epoch": 1.9617097061442563, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8019052147865295, + "num_tokens": 588178137.0, + "step": 15421 + }, + { + "epoch": 1.9618369164228469, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7944390177726746, + "num_tokens": 588215646.0, + "step": 15422 + }, + { + "epoch": 1.9619641267014374, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7943920493125916, + "num_tokens": 588251345.0, + "step": 15423 + }, + { + "epoch": 1.962091336980028, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8118712902069092, + "num_tokens": 588291222.0, + "step": 15424 + }, + { + "epoch": 1.9622185472586184, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7902379035949707, + "num_tokens": 588329480.0, + "step": 15425 + }, + { + "epoch": 1.962345757537209, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7899128198623657, + "num_tokens": 588367143.0, + "step": 15426 + }, + { + "epoch": 1.9624729678157995, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8110231161117554, + "num_tokens": 588408363.0, + "step": 15427 + }, + { + "epoch": 1.96260017809439, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.8034270405769348, + "num_tokens": 588445397.0, + "step": 15428 + }, + { + "epoch": 1.9627273883729806, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8042720556259155, + "num_tokens": 588480436.0, + "step": 15429 + }, + { + "epoch": 1.962854598651571, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8078737258911133, + "num_tokens": 588526918.0, + "step": 15430 + }, + { + "epoch": 1.9629818089301616, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.797797679901123, + "num_tokens": 588561300.0, + "step": 15431 + }, + { + "epoch": 1.9631090192087521, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7909064292907715, + "num_tokens": 588603393.0, + "step": 15432 + }, + { + "epoch": 1.9632362294873427, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8107331991195679, + "num_tokens": 588643806.0, + "step": 15433 + }, + { + "epoch": 1.9633634397659332, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8023601770401001, + "num_tokens": 588685090.0, + "step": 15434 + }, + { + "epoch": 1.9634906500445237, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7833495736122131, + "num_tokens": 588720983.0, + "step": 15435 + }, + { + "epoch": 1.9636178603231143, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7967774868011475, + "num_tokens": 588760143.0, + "step": 15436 + }, + { + "epoch": 1.9637450706017048, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8013013601303101, + "num_tokens": 588796636.0, + "step": 15437 + }, + { + "epoch": 1.963872280880295, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7968317270278931, + "num_tokens": 588831681.0, + "step": 15438 + }, + { + "epoch": 1.9639994911588856, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8197338581085205, + "num_tokens": 588875408.0, + "step": 15439 + }, + { + "epoch": 1.9641267014374761, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8071267008781433, + "num_tokens": 588913583.0, + "step": 15440 + }, + { + "epoch": 1.9642539117160667, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8128527402877808, + "num_tokens": 588949833.0, + "step": 15441 + }, + { + "epoch": 1.9643811219946572, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8015304803848267, + "num_tokens": 588985678.0, + "step": 15442 + }, + { + "epoch": 1.9645083322732477, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.8002748489379883, + "num_tokens": 589018276.0, + "step": 15443 + }, + { + "epoch": 1.964635542551838, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8023260831832886, + "num_tokens": 589061900.0, + "step": 15444 + }, + { + "epoch": 1.9647627528304286, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8018547892570496, + "num_tokens": 589102740.0, + "step": 15445 + }, + { + "epoch": 1.964889963109019, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8091381192207336, + "num_tokens": 589135276.0, + "step": 15446 + }, + { + "epoch": 1.9650171733876096, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.802200436592102, + "num_tokens": 589173390.0, + "step": 15447 + }, + { + "epoch": 1.9651443836662001, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7957563400268555, + "num_tokens": 589214592.0, + "step": 15448 + }, + { + "epoch": 1.9652715939447907, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7974716424942017, + "num_tokens": 589250682.0, + "step": 15449 + }, + { + "epoch": 1.9653988042233812, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7977524399757385, + "num_tokens": 589288326.0, + "step": 15450 + }, + { + "epoch": 1.9655260145019717, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7973508834838867, + "num_tokens": 589321632.0, + "step": 15451 + }, + { + "epoch": 1.9656532247805623, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7924109101295471, + "num_tokens": 589361134.0, + "step": 15452 + }, + { + "epoch": 1.9657804350591528, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8053684234619141, + "num_tokens": 589402338.0, + "step": 15453 + }, + { + "epoch": 1.9659076453377433, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.790225088596344, + "num_tokens": 589441080.0, + "step": 15454 + }, + { + "epoch": 1.9660348556163338, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7982000112533569, + "num_tokens": 589484019.0, + "step": 15455 + }, + { + "epoch": 1.9661620658949244, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7992636561393738, + "num_tokens": 589525890.0, + "step": 15456 + }, + { + "epoch": 1.966289276173515, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8065467476844788, + "num_tokens": 589560644.0, + "step": 15457 + }, + { + "epoch": 1.9664164864521054, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8225390315055847, + "num_tokens": 589600863.0, + "step": 15458 + }, + { + "epoch": 1.966543696730696, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7986570596694946, + "num_tokens": 589637948.0, + "step": 15459 + }, + { + "epoch": 1.9666709070092865, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.8204116821289062, + "num_tokens": 589678061.0, + "step": 15460 + }, + { + "epoch": 1.966798117287877, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.786348819732666, + "num_tokens": 589718050.0, + "step": 15461 + }, + { + "epoch": 1.9669253275664673, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7982969284057617, + "num_tokens": 589759206.0, + "step": 15462 + }, + { + "epoch": 1.9670525378450578, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7920632362365723, + "num_tokens": 589800218.0, + "step": 15463 + }, + { + "epoch": 1.9671797481236484, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8014430403709412, + "num_tokens": 589843105.0, + "step": 15464 + }, + { + "epoch": 1.967306958402239, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7907912731170654, + "num_tokens": 589881560.0, + "step": 15465 + }, + { + "epoch": 1.9674341686808294, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7894490957260132, + "num_tokens": 589922755.0, + "step": 15466 + }, + { + "epoch": 1.96756137895942, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7805367708206177, + "num_tokens": 589964982.0, + "step": 15467 + }, + { + "epoch": 1.9676885892380103, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7860407829284668, + "num_tokens": 590003371.0, + "step": 15468 + }, + { + "epoch": 1.9678157995166008, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8013186454772949, + "num_tokens": 590048361.0, + "step": 15469 + }, + { + "epoch": 1.9679430097951913, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.794948935508728, + "num_tokens": 590083141.0, + "step": 15470 + }, + { + "epoch": 1.9680702200737819, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.8002634048461914, + "num_tokens": 590128161.0, + "step": 15471 + }, + { + "epoch": 1.9681974303523724, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.8221191763877869, + "num_tokens": 590166101.0, + "step": 15472 + }, + { + "epoch": 1.968324640630963, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7998667359352112, + "num_tokens": 590207643.0, + "step": 15473 + }, + { + "epoch": 1.9684518509095534, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7931116819381714, + "num_tokens": 590247567.0, + "step": 15474 + }, + { + "epoch": 1.968579061188144, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7937939167022705, + "num_tokens": 590287940.0, + "step": 15475 + }, + { + "epoch": 1.9687062714667345, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7934080958366394, + "num_tokens": 590322558.0, + "step": 15476 + }, + { + "epoch": 1.968833481745325, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.7995891571044922, + "num_tokens": 590362913.0, + "step": 15477 + }, + { + "epoch": 1.9689606920239155, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7907840013504028, + "num_tokens": 590405877.0, + "step": 15478 + }, + { + "epoch": 1.969087902302506, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7890496850013733, + "num_tokens": 590445373.0, + "step": 15479 + }, + { + "epoch": 1.9692151125810966, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7861359715461731, + "num_tokens": 590483319.0, + "step": 15480 + }, + { + "epoch": 1.9693423228596871, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7858089208602905, + "num_tokens": 590522433.0, + "step": 15481 + }, + { + "epoch": 1.9694695331382777, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8022885322570801, + "num_tokens": 590564342.0, + "step": 15482 + }, + { + "epoch": 1.9695967434168682, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.799615204334259, + "num_tokens": 590594527.0, + "step": 15483 + }, + { + "epoch": 1.9697239536954587, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8117035031318665, + "num_tokens": 590630573.0, + "step": 15484 + }, + { + "epoch": 1.9698511639740492, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8015568256378174, + "num_tokens": 590669333.0, + "step": 15485 + }, + { + "epoch": 1.9699783742526398, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8171485662460327, + "num_tokens": 590709884.0, + "step": 15486 + }, + { + "epoch": 1.97010558453123, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8081421256065369, + "num_tokens": 590746109.0, + "step": 15487 + }, + { + "epoch": 1.9702327948098206, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.787250280380249, + "num_tokens": 590787881.0, + "step": 15488 + }, + { + "epoch": 1.9703600050884111, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7862783670425415, + "num_tokens": 590826861.0, + "step": 15489 + }, + { + "epoch": 1.9704872153670017, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8075019121170044, + "num_tokens": 590863104.0, + "step": 15490 + }, + { + "epoch": 1.9706144256455922, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8060115575790405, + "num_tokens": 590900582.0, + "step": 15491 + }, + { + "epoch": 1.9707416359241827, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8038336038589478, + "num_tokens": 590937735.0, + "step": 15492 + }, + { + "epoch": 1.970868846202773, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8154623508453369, + "num_tokens": 590973089.0, + "step": 15493 + }, + { + "epoch": 1.9709960564813636, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.786986768245697, + "num_tokens": 591012117.0, + "step": 15494 + }, + { + "epoch": 1.971123266759954, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7824052572250366, + "num_tokens": 591044978.0, + "step": 15495 + }, + { + "epoch": 1.9712504770385446, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7973508238792419, + "num_tokens": 591085562.0, + "step": 15496 + }, + { + "epoch": 1.9713776873171351, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7891349196434021, + "num_tokens": 591120140.0, + "step": 15497 + }, + { + "epoch": 1.9715048975957257, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.784351110458374, + "num_tokens": 591153952.0, + "step": 15498 + }, + { + "epoch": 1.9716321078743162, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8192143440246582, + "num_tokens": 591192983.0, + "step": 15499 + }, + { + "epoch": 1.9717593181529067, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8016300797462463, + "num_tokens": 591226260.0, + "step": 15500 + }, + { + "epoch": 1.9718865284314973, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7845940589904785, + "num_tokens": 591259361.0, + "step": 15501 + }, + { + "epoch": 1.9720137387100878, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8018832206726074, + "num_tokens": 591299753.0, + "step": 15502 + }, + { + "epoch": 1.9721409489886783, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8021255731582642, + "num_tokens": 591335008.0, + "step": 15503 + }, + { + "epoch": 1.9722681592672688, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8027991056442261, + "num_tokens": 591373001.0, + "step": 15504 + }, + { + "epoch": 1.9723953695458594, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7997080087661743, + "num_tokens": 591408637.0, + "step": 15505 + }, + { + "epoch": 1.97252257982445, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7942852973937988, + "num_tokens": 591444908.0, + "step": 15506 + }, + { + "epoch": 1.9726497901030404, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.801653265953064, + "num_tokens": 591484641.0, + "step": 15507 + }, + { + "epoch": 1.972777000381631, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.799983024597168, + "num_tokens": 591521859.0, + "step": 15508 + }, + { + "epoch": 1.9729042106602215, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.797353982925415, + "num_tokens": 591559124.0, + "step": 15509 + }, + { + "epoch": 1.973031420938812, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7999048829078674, + "num_tokens": 591597170.0, + "step": 15510 + }, + { + "epoch": 1.9731586312174023, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8023404479026794, + "num_tokens": 591634784.0, + "step": 15511 + }, + { + "epoch": 1.9732858414959928, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8079282641410828, + "num_tokens": 591671904.0, + "step": 15512 + }, + { + "epoch": 1.9734130517745834, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8057416677474976, + "num_tokens": 591712990.0, + "step": 15513 + }, + { + "epoch": 1.973540262053174, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.797730565071106, + "num_tokens": 591748579.0, + "step": 15514 + }, + { + "epoch": 1.9736674723317644, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7965478897094727, + "num_tokens": 591785865.0, + "step": 15515 + }, + { + "epoch": 1.973794682610355, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8029625415802002, + "num_tokens": 591820087.0, + "step": 15516 + }, + { + "epoch": 1.9739218928889453, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8026390075683594, + "num_tokens": 591859713.0, + "step": 15517 + }, + { + "epoch": 1.9740491031675358, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8103713393211365, + "num_tokens": 591893142.0, + "step": 15518 + }, + { + "epoch": 1.9741763134461263, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7944965362548828, + "num_tokens": 591931529.0, + "step": 15519 + }, + { + "epoch": 1.9743035237247168, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8067241311073303, + "num_tokens": 591976248.0, + "step": 15520 + }, + { + "epoch": 1.9744307340033074, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7993132472038269, + "num_tokens": 592018008.0, + "step": 15521 + }, + { + "epoch": 1.974557944281898, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7861096858978271, + "num_tokens": 592062934.0, + "step": 15522 + }, + { + "epoch": 1.9746851545604884, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7998866438865662, + "num_tokens": 592103129.0, + "step": 15523 + }, + { + "epoch": 1.974812364839079, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7976542711257935, + "num_tokens": 592142116.0, + "step": 15524 + }, + { + "epoch": 1.9749395751176695, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8059259653091431, + "num_tokens": 592182507.0, + "step": 15525 + }, + { + "epoch": 1.97506678539626, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7784691452980042, + "num_tokens": 592225593.0, + "step": 15526 + }, + { + "epoch": 1.9751939956748505, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.5729, + "mean_token_accuracy": 0.8272185325622559, + "num_tokens": 592270676.0, + "step": 15527 + }, + { + "epoch": 1.975321205953441, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.5822, + "mean_token_accuracy": 0.8265724182128906, + "num_tokens": 592305537.0, + "step": 15528 + }, + { + "epoch": 1.9754484162320316, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.8019406795501709, + "num_tokens": 592338289.0, + "step": 15529 + }, + { + "epoch": 1.9755756265106221, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8029395937919617, + "num_tokens": 592379189.0, + "step": 15530 + }, + { + "epoch": 1.9757028367892127, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8174010515213013, + "num_tokens": 592421875.0, + "step": 15531 + }, + { + "epoch": 1.9758300470678032, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7793289422988892, + "num_tokens": 592457438.0, + "step": 15532 + }, + { + "epoch": 1.9759572573463937, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8121679425239563, + "num_tokens": 592493230.0, + "step": 15533 + }, + { + "epoch": 1.9760844676249842, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7840381860733032, + "num_tokens": 592533656.0, + "step": 15534 + }, + { + "epoch": 1.9762116779035748, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8136877417564392, + "num_tokens": 592571622.0, + "step": 15535 + }, + { + "epoch": 1.976338888182165, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7919512391090393, + "num_tokens": 592613260.0, + "step": 15536 + }, + { + "epoch": 1.9764660984607556, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7926355004310608, + "num_tokens": 592648872.0, + "step": 15537 + }, + { + "epoch": 1.9765933087393461, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7973449230194092, + "num_tokens": 592682336.0, + "step": 15538 + }, + { + "epoch": 1.9767205190179367, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8161205053329468, + "num_tokens": 592716412.0, + "step": 15539 + }, + { + "epoch": 1.9768477292965272, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8030186891555786, + "num_tokens": 592753936.0, + "step": 15540 + }, + { + "epoch": 1.9769749395751177, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8015671968460083, + "num_tokens": 592792731.0, + "step": 15541 + }, + { + "epoch": 1.977102149853708, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8046398162841797, + "num_tokens": 592832035.0, + "step": 15542 + }, + { + "epoch": 1.9772293601322986, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8136458396911621, + "num_tokens": 592869693.0, + "step": 15543 + }, + { + "epoch": 1.977356570410889, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8090561032295227, + "num_tokens": 592909648.0, + "step": 15544 + }, + { + "epoch": 1.9774837806894796, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8006628751754761, + "num_tokens": 592942539.0, + "step": 15545 + }, + { + "epoch": 1.9776109909680701, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.800068199634552, + "num_tokens": 592983680.0, + "step": 15546 + }, + { + "epoch": 1.9777382012466607, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7901992797851562, + "num_tokens": 593023161.0, + "step": 15547 + }, + { + "epoch": 1.9778654115252512, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7796516418457031, + "num_tokens": 593066612.0, + "step": 15548 + }, + { + "epoch": 1.9779926218038417, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7841987609863281, + "num_tokens": 593105224.0, + "step": 15549 + }, + { + "epoch": 1.9781198320824323, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8006489276885986, + "num_tokens": 593143683.0, + "step": 15550 + }, + { + "epoch": 1.9782470423610228, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7823584675788879, + "num_tokens": 593179827.0, + "step": 15551 + }, + { + "epoch": 1.9783742526396133, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.799027144908905, + "num_tokens": 593220936.0, + "step": 15552 + }, + { + "epoch": 1.9785014629182038, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7923080921173096, + "num_tokens": 593258503.0, + "step": 15553 + }, + { + "epoch": 1.9786286731967944, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.778492271900177, + "num_tokens": 593293738.0, + "step": 15554 + }, + { + "epoch": 1.978755883475385, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7919453382492065, + "num_tokens": 593338037.0, + "step": 15555 + }, + { + "epoch": 1.9788830937539754, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7899683713912964, + "num_tokens": 593383509.0, + "step": 15556 + }, + { + "epoch": 1.979010304032566, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8028708696365356, + "num_tokens": 593417734.0, + "step": 15557 + }, + { + "epoch": 1.9791375143111565, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8110374212265015, + "num_tokens": 593454502.0, + "step": 15558 + }, + { + "epoch": 1.979264724589747, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.8003844022750854, + "num_tokens": 593498864.0, + "step": 15559 + }, + { + "epoch": 1.9793919348683373, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7986936569213867, + "num_tokens": 593540125.0, + "step": 15560 + }, + { + "epoch": 1.9795191451469278, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8033387660980225, + "num_tokens": 593566937.0, + "step": 15561 + }, + { + "epoch": 1.9796463554255184, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7911189198493958, + "num_tokens": 593600109.0, + "step": 15562 + }, + { + "epoch": 1.979773565704109, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8013015389442444, + "num_tokens": 593638982.0, + "step": 15563 + }, + { + "epoch": 1.9799007759826994, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7928290963172913, + "num_tokens": 593677270.0, + "step": 15564 + }, + { + "epoch": 1.98002798626129, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8089625835418701, + "num_tokens": 593714056.0, + "step": 15565 + }, + { + "epoch": 1.9801551965398803, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7891503572463989, + "num_tokens": 593753516.0, + "step": 15566 + }, + { + "epoch": 1.9802824068184708, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8001033067703247, + "num_tokens": 593789276.0, + "step": 15567 + }, + { + "epoch": 1.9804096170970613, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8041684031486511, + "num_tokens": 593828112.0, + "step": 15568 + }, + { + "epoch": 1.9805368273756518, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7848929166793823, + "num_tokens": 593866370.0, + "step": 15569 + }, + { + "epoch": 1.9806640376542424, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8082998991012573, + "num_tokens": 593907863.0, + "step": 15570 + }, + { + "epoch": 1.980791247932833, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8118975162506104, + "num_tokens": 593945254.0, + "step": 15571 + }, + { + "epoch": 1.9809184582114234, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8047420978546143, + "num_tokens": 593987105.0, + "step": 15572 + }, + { + "epoch": 1.981045668490014, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.810492217540741, + "num_tokens": 594024091.0, + "step": 15573 + }, + { + "epoch": 1.9811728787686045, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8128086924552917, + "num_tokens": 594056018.0, + "step": 15574 + }, + { + "epoch": 1.981300089047195, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8058943152427673, + "num_tokens": 594102297.0, + "step": 15575 + }, + { + "epoch": 1.9814272993257855, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7997887134552002, + "num_tokens": 594140278.0, + "step": 15576 + }, + { + "epoch": 1.981554509604376, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8025085926055908, + "num_tokens": 594175025.0, + "step": 15577 + }, + { + "epoch": 1.9816817198829666, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8046473264694214, + "num_tokens": 594217906.0, + "step": 15578 + }, + { + "epoch": 1.9818089301615571, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.784103512763977, + "num_tokens": 594264331.0, + "step": 15579 + }, + { + "epoch": 1.9819361404401477, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.5982, + "mean_token_accuracy": 0.8172321915626526, + "num_tokens": 594300220.0, + "step": 15580 + }, + { + "epoch": 1.9820633507187382, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.806226372718811, + "num_tokens": 594337849.0, + "step": 15581 + }, + { + "epoch": 1.9821905609973287, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7936902642250061, + "num_tokens": 594379295.0, + "step": 15582 + }, + { + "epoch": 1.9823177712759192, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7935259342193604, + "num_tokens": 594408350.0, + "step": 15583 + }, + { + "epoch": 1.9824449815545098, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.793843150138855, + "num_tokens": 594455110.0, + "step": 15584 + }, + { + "epoch": 1.9825721918331, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8098567724227905, + "num_tokens": 594499231.0, + "step": 15585 + }, + { + "epoch": 1.9826994021116906, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7784404158592224, + "num_tokens": 594541478.0, + "step": 15586 + }, + { + "epoch": 1.9828266123902811, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8075948357582092, + "num_tokens": 594579075.0, + "step": 15587 + }, + { + "epoch": 1.9829538226688717, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8170123100280762, + "num_tokens": 594616189.0, + "step": 15588 + }, + { + "epoch": 1.9830810329474622, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8047888278961182, + "num_tokens": 594656633.0, + "step": 15589 + }, + { + "epoch": 1.9832082432260527, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7805318832397461, + "num_tokens": 594696833.0, + "step": 15590 + }, + { + "epoch": 1.983335453504643, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7873011827468872, + "num_tokens": 594735138.0, + "step": 15591 + }, + { + "epoch": 1.9834626637832335, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7903270721435547, + "num_tokens": 594769587.0, + "step": 15592 + }, + { + "epoch": 1.983589874061824, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8053784370422363, + "num_tokens": 594803635.0, + "step": 15593 + }, + { + "epoch": 1.9837170843404146, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8135287761688232, + "num_tokens": 594839872.0, + "step": 15594 + }, + { + "epoch": 1.9838442946190051, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7890784740447998, + "num_tokens": 594879208.0, + "step": 15595 + }, + { + "epoch": 1.9839715048975957, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8033351898193359, + "num_tokens": 594910377.0, + "step": 15596 + }, + { + "epoch": 1.9840987151761862, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.813447892665863, + "num_tokens": 594956574.0, + "step": 15597 + }, + { + "epoch": 1.9842259254547767, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7922025918960571, + "num_tokens": 594998036.0, + "step": 15598 + }, + { + "epoch": 1.9843531357333672, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7880971431732178, + "num_tokens": 595031404.0, + "step": 15599 + }, + { + "epoch": 1.9844803460119578, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.809996485710144, + "num_tokens": 595069800.0, + "step": 15600 + }, + { + "epoch": 1.9846075562905483, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8090940713882446, + "num_tokens": 595110987.0, + "step": 15601 + }, + { + "epoch": 1.9847347665691388, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8047915101051331, + "num_tokens": 595149192.0, + "step": 15602 + }, + { + "epoch": 1.9848619768477294, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7914910316467285, + "num_tokens": 595182993.0, + "step": 15603 + }, + { + "epoch": 1.9849891871263199, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7964789271354675, + "num_tokens": 595216177.0, + "step": 15604 + }, + { + "epoch": 1.9851163974049104, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8028386235237122, + "num_tokens": 595253379.0, + "step": 15605 + }, + { + "epoch": 1.985243607683501, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8126218914985657, + "num_tokens": 595292878.0, + "step": 15606 + }, + { + "epoch": 1.9853708179620915, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8047250509262085, + "num_tokens": 595329082.0, + "step": 15607 + }, + { + "epoch": 1.985498028240682, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7946784496307373, + "num_tokens": 595367246.0, + "step": 15608 + }, + { + "epoch": 1.9856252385192723, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8042557835578918, + "num_tokens": 595404450.0, + "step": 15609 + }, + { + "epoch": 1.9857524487978628, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8102153539657593, + "num_tokens": 595439718.0, + "step": 15610 + }, + { + "epoch": 1.9858796590764534, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8008813858032227, + "num_tokens": 595473617.0, + "step": 15611 + }, + { + "epoch": 1.986006869355044, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.801348090171814, + "num_tokens": 595516432.0, + "step": 15612 + }, + { + "epoch": 1.9861340796336344, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8010238409042358, + "num_tokens": 595556278.0, + "step": 15613 + }, + { + "epoch": 1.986261289912225, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8036512136459351, + "num_tokens": 595598713.0, + "step": 15614 + }, + { + "epoch": 1.9863885001908153, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.797269344329834, + "num_tokens": 595630597.0, + "step": 15615 + }, + { + "epoch": 1.9865157104694058, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8167772889137268, + "num_tokens": 595664163.0, + "step": 15616 + }, + { + "epoch": 1.9866429207479963, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8064950704574585, + "num_tokens": 595698809.0, + "step": 15617 + }, + { + "epoch": 1.9867701310265868, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7955685257911682, + "num_tokens": 595741166.0, + "step": 15618 + }, + { + "epoch": 1.9868973413051774, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7997274994850159, + "num_tokens": 595787079.0, + "step": 15619 + }, + { + "epoch": 1.987024551583768, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7744061946868896, + "num_tokens": 595829085.0, + "step": 15620 + }, + { + "epoch": 1.9871517618623584, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8017177581787109, + "num_tokens": 595870646.0, + "step": 15621 + }, + { + "epoch": 1.987278972140949, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7911984324455261, + "num_tokens": 595918473.0, + "step": 15622 + }, + { + "epoch": 1.9874061824195395, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7859511375427246, + "num_tokens": 595954097.0, + "step": 15623 + }, + { + "epoch": 1.98753339269813, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7801295518875122, + "num_tokens": 595990191.0, + "step": 15624 + }, + { + "epoch": 1.9876606029767205, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7900953888893127, + "num_tokens": 596027109.0, + "step": 15625 + }, + { + "epoch": 1.987787813255311, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8049942255020142, + "num_tokens": 596067930.0, + "step": 15626 + }, + { + "epoch": 1.9879150235339016, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8076003789901733, + "num_tokens": 596105809.0, + "step": 15627 + }, + { + "epoch": 1.9880422338124921, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8056048154830933, + "num_tokens": 596144220.0, + "step": 15628 + }, + { + "epoch": 1.9881694440910826, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7769935131072998, + "num_tokens": 596183825.0, + "step": 15629 + }, + { + "epoch": 1.9882966543696732, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8126463890075684, + "num_tokens": 596232333.0, + "step": 15630 + }, + { + "epoch": 1.9884238646482637, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.8103877902030945, + "num_tokens": 596272138.0, + "step": 15631 + }, + { + "epoch": 1.9885510749268542, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8130711317062378, + "num_tokens": 596310254.0, + "step": 15632 + }, + { + "epoch": 1.9886782852054448, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8013051748275757, + "num_tokens": 596350396.0, + "step": 15633 + }, + { + "epoch": 1.988805495484035, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8006506562232971, + "num_tokens": 596388809.0, + "step": 15634 + }, + { + "epoch": 1.9889327057626256, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.810711145401001, + "num_tokens": 596427366.0, + "step": 15635 + }, + { + "epoch": 1.9890599160412161, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7938165068626404, + "num_tokens": 596469999.0, + "step": 15636 + }, + { + "epoch": 1.9891871263198067, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.787136435508728, + "num_tokens": 596506773.0, + "step": 15637 + }, + { + "epoch": 1.9893143365983972, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8051255941390991, + "num_tokens": 596547440.0, + "step": 15638 + }, + { + "epoch": 1.9894415468769877, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8090387582778931, + "num_tokens": 596580212.0, + "step": 15639 + }, + { + "epoch": 1.989568757155578, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.824439287185669, + "num_tokens": 596611790.0, + "step": 15640 + }, + { + "epoch": 1.9896959674341685, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5931, + "mean_token_accuracy": 0.8199693560600281, + "num_tokens": 596646984.0, + "step": 15641 + }, + { + "epoch": 1.989823177712759, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7971704602241516, + "num_tokens": 596688362.0, + "step": 15642 + }, + { + "epoch": 1.9899503879913496, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7971808910369873, + "num_tokens": 596725298.0, + "step": 15643 + }, + { + "epoch": 1.9900775982699401, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7993181943893433, + "num_tokens": 596757400.0, + "step": 15644 + }, + { + "epoch": 1.9902048085485307, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8103352189064026, + "num_tokens": 596790533.0, + "step": 15645 + }, + { + "epoch": 1.9903320188271212, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7965220808982849, + "num_tokens": 596834467.0, + "step": 15646 + }, + { + "epoch": 1.9904592291057117, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7881430387496948, + "num_tokens": 596873262.0, + "step": 15647 + }, + { + "epoch": 1.9905864393843022, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8037493228912354, + "num_tokens": 596912833.0, + "step": 15648 + }, + { + "epoch": 1.9907136496628928, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8090150952339172, + "num_tokens": 596953554.0, + "step": 15649 + }, + { + "epoch": 1.9908408599414833, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7957687377929688, + "num_tokens": 596991206.0, + "step": 15650 + }, + { + "epoch": 1.9909680702200738, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.801264226436615, + "num_tokens": 597027222.0, + "step": 15651 + }, + { + "epoch": 1.9910952804986644, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.814820408821106, + "num_tokens": 597069040.0, + "step": 15652 + }, + { + "epoch": 1.9912224907772549, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7961517572402954, + "num_tokens": 597112529.0, + "step": 15653 + }, + { + "epoch": 1.9913497010558454, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8007431030273438, + "num_tokens": 597148304.0, + "step": 15654 + }, + { + "epoch": 1.991476911334436, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8052850961685181, + "num_tokens": 597184244.0, + "step": 15655 + }, + { + "epoch": 1.9916041216130265, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7908210158348083, + "num_tokens": 597218178.0, + "step": 15656 + }, + { + "epoch": 1.991731331891617, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8027139902114868, + "num_tokens": 597259256.0, + "step": 15657 + }, + { + "epoch": 1.9918585421702073, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.807536244392395, + "num_tokens": 597296584.0, + "step": 15658 + }, + { + "epoch": 1.9919857524487978, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7910773158073425, + "num_tokens": 597341116.0, + "step": 15659 + }, + { + "epoch": 1.9921129627273884, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.79217529296875, + "num_tokens": 597375595.0, + "step": 15660 + }, + { + "epoch": 1.9922401730059789, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7848598957061768, + "num_tokens": 597419226.0, + "step": 15661 + }, + { + "epoch": 1.9923673832845694, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7822716236114502, + "num_tokens": 597461242.0, + "step": 15662 + }, + { + "epoch": 1.99249459356316, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8036766052246094, + "num_tokens": 597500030.0, + "step": 15663 + }, + { + "epoch": 1.9926218038417502, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7789269089698792, + "num_tokens": 597540428.0, + "step": 15664 + }, + { + "epoch": 1.9927490141203408, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.8145058155059814, + "num_tokens": 597575070.0, + "step": 15665 + }, + { + "epoch": 1.9928762243989313, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7935478687286377, + "num_tokens": 597617425.0, + "step": 15666 + }, + { + "epoch": 1.9930034346775218, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8087291717529297, + "num_tokens": 597660489.0, + "step": 15667 + }, + { + "epoch": 1.9931306449561124, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8072373270988464, + "num_tokens": 597703626.0, + "step": 15668 + }, + { + "epoch": 1.993257855234703, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7886886596679688, + "num_tokens": 597741176.0, + "step": 15669 + }, + { + "epoch": 1.9933850655132934, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7860079407691956, + "num_tokens": 597779763.0, + "step": 15670 + }, + { + "epoch": 1.993512275791884, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8035645484924316, + "num_tokens": 597813300.0, + "step": 15671 + }, + { + "epoch": 1.9936394860704745, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7853071689605713, + "num_tokens": 597849062.0, + "step": 15672 + }, + { + "epoch": 1.993766696349065, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7993415594100952, + "num_tokens": 597886565.0, + "step": 15673 + }, + { + "epoch": 1.9938939066276555, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8039546012878418, + "num_tokens": 597925097.0, + "step": 15674 + }, + { + "epoch": 1.994021116906246, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.807684600353241, + "num_tokens": 597960786.0, + "step": 15675 + }, + { + "epoch": 1.9941483271848366, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7858282327651978, + "num_tokens": 598000442.0, + "step": 15676 + }, + { + "epoch": 1.9942755374634271, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7992632985115051, + "num_tokens": 598042946.0, + "step": 15677 + }, + { + "epoch": 1.9944027477420176, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7828131318092346, + "num_tokens": 598081659.0, + "step": 15678 + }, + { + "epoch": 1.9945299580206082, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7955735325813293, + "num_tokens": 598120203.0, + "step": 15679 + }, + { + "epoch": 1.9946571682991987, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7822740077972412, + "num_tokens": 598154779.0, + "step": 15680 + }, + { + "epoch": 1.9947843785777892, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7982091903686523, + "num_tokens": 598197075.0, + "step": 15681 + }, + { + "epoch": 1.9949115888563798, + "grad_norm": 11.125, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7947885394096375, + "num_tokens": 598240539.0, + "step": 15682 + }, + { + "epoch": 1.99503879913497, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.802759051322937, + "num_tokens": 598281776.0, + "step": 15683 + }, + { + "epoch": 1.9951660094135606, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8109031319618225, + "num_tokens": 598322715.0, + "step": 15684 + }, + { + "epoch": 1.9952932196921511, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7931199073791504, + "num_tokens": 598364495.0, + "step": 15685 + }, + { + "epoch": 1.9954204299707416, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7977136373519897, + "num_tokens": 598407102.0, + "step": 15686 + }, + { + "epoch": 1.9955476402493322, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7878637313842773, + "num_tokens": 598448600.0, + "step": 15687 + }, + { + "epoch": 1.9956748505279227, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7907617092132568, + "num_tokens": 598491140.0, + "step": 15688 + }, + { + "epoch": 1.995802060806513, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7953153848648071, + "num_tokens": 598527831.0, + "step": 15689 + }, + { + "epoch": 1.9959292710851035, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7870622873306274, + "num_tokens": 598568680.0, + "step": 15690 + }, + { + "epoch": 1.996056481363694, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7876379489898682, + "num_tokens": 598606254.0, + "step": 15691 + }, + { + "epoch": 1.9961836916422846, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8054934740066528, + "num_tokens": 598642400.0, + "step": 15692 + }, + { + "epoch": 1.9963109019208751, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7810815572738647, + "num_tokens": 598678379.0, + "step": 15693 + }, + { + "epoch": 1.9964381121994657, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7977990508079529, + "num_tokens": 598718234.0, + "step": 15694 + }, + { + "epoch": 1.9965653224780562, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8030582666397095, + "num_tokens": 598754897.0, + "step": 15695 + }, + { + "epoch": 1.9966925327566467, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7874912023544312, + "num_tokens": 598785268.0, + "step": 15696 + }, + { + "epoch": 1.9968197430352372, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8056268692016602, + "num_tokens": 598827131.0, + "step": 15697 + }, + { + "epoch": 1.9969469533138278, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.799651026725769, + "num_tokens": 598859507.0, + "step": 15698 + }, + { + "epoch": 1.9970741635924183, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7904531955718994, + "num_tokens": 598893280.0, + "step": 15699 + }, + { + "epoch": 1.9972013738710088, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8107638359069824, + "num_tokens": 598932759.0, + "step": 15700 + }, + { + "epoch": 1.9973285841495994, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.811806857585907, + "num_tokens": 598972899.0, + "step": 15701 + }, + { + "epoch": 1.9974557944281899, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8070157766342163, + "num_tokens": 599012811.0, + "step": 15702 + }, + { + "epoch": 1.9975830047067804, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7897276878356934, + "num_tokens": 599055632.0, + "step": 15703 + }, + { + "epoch": 1.997710214985371, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7805963754653931, + "num_tokens": 599090924.0, + "step": 15704 + }, + { + "epoch": 1.9978374252639615, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8169543147087097, + "num_tokens": 599130434.0, + "step": 15705 + }, + { + "epoch": 1.997964635542552, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8010719418525696, + "num_tokens": 599171018.0, + "step": 15706 + }, + { + "epoch": 1.9980918458211423, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7922769784927368, + "num_tokens": 599212445.0, + "step": 15707 + }, + { + "epoch": 1.9982190560997328, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8010507822036743, + "num_tokens": 599250591.0, + "step": 15708 + }, + { + "epoch": 1.9983462663783234, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7939333915710449, + "num_tokens": 599284987.0, + "step": 15709 + }, + { + "epoch": 1.9984734766569139, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7943388819694519, + "num_tokens": 599324656.0, + "step": 15710 + }, + { + "epoch": 1.9986006869355044, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8095909357070923, + "num_tokens": 599364014.0, + "step": 15711 + }, + { + "epoch": 1.998727897214095, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.791580855846405, + "num_tokens": 599396781.0, + "step": 15712 + }, + { + "epoch": 1.9988551074926852, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8124960660934448, + "num_tokens": 599433240.0, + "step": 15713 + }, + { + "epoch": 1.9989823177712758, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8129668831825256, + "num_tokens": 599474883.0, + "step": 15714 + }, + { + "epoch": 1.9991095280498663, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8054172992706299, + "num_tokens": 599508492.0, + "step": 15715 + }, + { + "epoch": 1.9992367383284568, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.805020809173584, + "num_tokens": 599545946.0, + "step": 15716 + }, + { + "epoch": 1.9993639486070474, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.815043032169342, + "num_tokens": 599586217.0, + "step": 15717 + }, + { + "epoch": 1.9994911588856379, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8121805191040039, + "num_tokens": 599620431.0, + "step": 15718 + }, + { + "epoch": 1.9996183691642284, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7846509218215942, + "num_tokens": 599659959.0, + "step": 15719 + }, + { + "epoch": 1.999745579442819, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7952336072921753, + "num_tokens": 599696558.0, + "step": 15720 + }, + { + "epoch": 1.9998727897214095, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7862401008605957, + "num_tokens": 599734925.0, + "step": 15721 + }, + { + "epoch": 2.0, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8028823137283325, + "num_tokens": 599772613.0, + "step": 15722 + }, + { + "epoch": 2.0001272102785905, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.798923671245575, + "num_tokens": 599813892.0, + "step": 15723 + }, + { + "epoch": 2.000254420557181, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8000903129577637, + "num_tokens": 599852807.0, + "step": 15724 + }, + { + "epoch": 2.0003816308357716, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8057577013969421, + "num_tokens": 599893712.0, + "step": 15725 + }, + { + "epoch": 2.000508841114362, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8001405000686646, + "num_tokens": 599929954.0, + "step": 15726 + }, + { + "epoch": 2.0006360513929526, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.796377420425415, + "num_tokens": 599967763.0, + "step": 15727 + }, + { + "epoch": 2.000763261671543, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8099836111068726, + "num_tokens": 600008938.0, + "step": 15728 + }, + { + "epoch": 2.0008904719501337, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7980629801750183, + "num_tokens": 600051849.0, + "step": 15729 + }, + { + "epoch": 2.0010176822287242, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8056092262268066, + "num_tokens": 600088568.0, + "step": 15730 + }, + { + "epoch": 2.0011448925073148, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8050671219825745, + "num_tokens": 600131553.0, + "step": 15731 + }, + { + "epoch": 2.0012721027859053, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8029212951660156, + "num_tokens": 600173198.0, + "step": 15732 + }, + { + "epoch": 2.001399313064496, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8036198616027832, + "num_tokens": 600211347.0, + "step": 15733 + }, + { + "epoch": 2.0015265233430863, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7910292148590088, + "num_tokens": 600249953.0, + "step": 15734 + }, + { + "epoch": 2.0016537336216764, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7912483811378479, + "num_tokens": 600291211.0, + "step": 15735 + }, + { + "epoch": 2.001780943900267, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8004326820373535, + "num_tokens": 600332906.0, + "step": 15736 + }, + { + "epoch": 2.0019081541788575, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8017950654029846, + "num_tokens": 600370570.0, + "step": 15737 + }, + { + "epoch": 2.002035364457448, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7928786277770996, + "num_tokens": 600409368.0, + "step": 15738 + }, + { + "epoch": 2.0021625747360385, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7984079122543335, + "num_tokens": 600442019.0, + "step": 15739 + }, + { + "epoch": 2.002289785014629, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8038617968559265, + "num_tokens": 600471665.0, + "step": 15740 + }, + { + "epoch": 2.0024169952932196, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7971013188362122, + "num_tokens": 600508804.0, + "step": 15741 + }, + { + "epoch": 2.00254420557181, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7880618572235107, + "num_tokens": 600543895.0, + "step": 15742 + }, + { + "epoch": 2.0026714158504006, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8021137714385986, + "num_tokens": 600585818.0, + "step": 15743 + }, + { + "epoch": 2.002798626128991, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7987766265869141, + "num_tokens": 600623106.0, + "step": 15744 + }, + { + "epoch": 2.0029258364075817, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.5662, + "mean_token_accuracy": 0.8283097743988037, + "num_tokens": 600662662.0, + "step": 15745 + }, + { + "epoch": 2.0030530466861722, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.784685492515564, + "num_tokens": 600699416.0, + "step": 15746 + }, + { + "epoch": 2.0031802569647628, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7922683954238892, + "num_tokens": 600739351.0, + "step": 15747 + }, + { + "epoch": 2.0033074672433533, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8022490739822388, + "num_tokens": 600781198.0, + "step": 15748 + }, + { + "epoch": 2.003434677521944, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7993566989898682, + "num_tokens": 600821743.0, + "step": 15749 + }, + { + "epoch": 2.0035618878005343, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8167992830276489, + "num_tokens": 600862463.0, + "step": 15750 + }, + { + "epoch": 2.003689098079125, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7845665216445923, + "num_tokens": 600902629.0, + "step": 15751 + }, + { + "epoch": 2.0038163083577154, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.796881914138794, + "num_tokens": 600941778.0, + "step": 15752 + }, + { + "epoch": 2.003943518636306, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8085321187973022, + "num_tokens": 600979425.0, + "step": 15753 + }, + { + "epoch": 2.0040707289148965, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7955951690673828, + "num_tokens": 601017882.0, + "step": 15754 + }, + { + "epoch": 2.004197939193487, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7910854816436768, + "num_tokens": 601053394.0, + "step": 15755 + }, + { + "epoch": 2.0043251494720775, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8032839894294739, + "num_tokens": 601094940.0, + "step": 15756 + }, + { + "epoch": 2.004452359750668, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8046314716339111, + "num_tokens": 601132963.0, + "step": 15757 + }, + { + "epoch": 2.0045795700292586, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7977986335754395, + "num_tokens": 601168511.0, + "step": 15758 + }, + { + "epoch": 2.0047067803078487, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8001260757446289, + "num_tokens": 601208922.0, + "step": 15759 + }, + { + "epoch": 2.004833990586439, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7926173210144043, + "num_tokens": 601253814.0, + "step": 15760 + }, + { + "epoch": 2.0049612008650297, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.802436351776123, + "num_tokens": 601295584.0, + "step": 15761 + }, + { + "epoch": 2.0050884111436202, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7856774926185608, + "num_tokens": 601329134.0, + "step": 15762 + }, + { + "epoch": 2.0052156214222108, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.810696005821228, + "num_tokens": 601369100.0, + "step": 15763 + }, + { + "epoch": 2.0053428317008013, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8192653656005859, + "num_tokens": 601407876.0, + "step": 15764 + }, + { + "epoch": 2.005470041979392, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8152884244918823, + "num_tokens": 601446996.0, + "step": 15765 + }, + { + "epoch": 2.0055972522579824, + "grad_norm": 65.0, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7828274369239807, + "num_tokens": 601482158.0, + "step": 15766 + }, + { + "epoch": 2.005724462536573, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8091423511505127, + "num_tokens": 601517940.0, + "step": 15767 + }, + { + "epoch": 2.0058516728151634, + "grad_norm": 10.125, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8250069618225098, + "num_tokens": 601553823.0, + "step": 15768 + }, + { + "epoch": 2.005978883093754, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7861236333847046, + "num_tokens": 601595482.0, + "step": 15769 + }, + { + "epoch": 2.0061060933723445, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8117966651916504, + "num_tokens": 601632451.0, + "step": 15770 + }, + { + "epoch": 2.006233303650935, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.796796441078186, + "num_tokens": 601667573.0, + "step": 15771 + }, + { + "epoch": 2.0063605139295255, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8045541048049927, + "num_tokens": 601704494.0, + "step": 15772 + }, + { + "epoch": 2.006487724208116, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.785950243473053, + "num_tokens": 601745307.0, + "step": 15773 + }, + { + "epoch": 2.0066149344867066, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8148383498191833, + "num_tokens": 601782551.0, + "step": 15774 + }, + { + "epoch": 2.006742144765297, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8023351430892944, + "num_tokens": 601823378.0, + "step": 15775 + }, + { + "epoch": 2.0068693550438876, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8008264303207397, + "num_tokens": 601860881.0, + "step": 15776 + }, + { + "epoch": 2.006996565322478, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8005928993225098, + "num_tokens": 601903502.0, + "step": 15777 + }, + { + "epoch": 2.0071237756010687, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.8156881332397461, + "num_tokens": 601937097.0, + "step": 15778 + }, + { + "epoch": 2.007250985879659, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8027040958404541, + "num_tokens": 601978050.0, + "step": 15779 + }, + { + "epoch": 2.0073781961582498, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8002803921699524, + "num_tokens": 602018167.0, + "step": 15780 + }, + { + "epoch": 2.0075054064368403, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7807268500328064, + "num_tokens": 602054485.0, + "step": 15781 + }, + { + "epoch": 2.007632616715431, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8067879676818848, + "num_tokens": 602092881.0, + "step": 15782 + }, + { + "epoch": 2.0077598269940213, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7998899221420288, + "num_tokens": 602127424.0, + "step": 15783 + }, + { + "epoch": 2.0078870372726114, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.808228611946106, + "num_tokens": 602168340.0, + "step": 15784 + }, + { + "epoch": 2.008014247551202, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7905148863792419, + "num_tokens": 602207894.0, + "step": 15785 + }, + { + "epoch": 2.0081414578297925, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7982304096221924, + "num_tokens": 602245924.0, + "step": 15786 + }, + { + "epoch": 2.008268668108383, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7986549139022827, + "num_tokens": 602278417.0, + "step": 15787 + }, + { + "epoch": 2.0083958783869735, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8026140928268433, + "num_tokens": 602318431.0, + "step": 15788 + }, + { + "epoch": 2.008523088665564, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7880233526229858, + "num_tokens": 602356315.0, + "step": 15789 + }, + { + "epoch": 2.0086502989441546, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8185074329376221, + "num_tokens": 602398865.0, + "step": 15790 + }, + { + "epoch": 2.008777509222745, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8073816299438477, + "num_tokens": 602435835.0, + "step": 15791 + }, + { + "epoch": 2.0089047195013356, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8111509084701538, + "num_tokens": 602470359.0, + "step": 15792 + }, + { + "epoch": 2.009031929779926, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.806594729423523, + "num_tokens": 602512132.0, + "step": 15793 + }, + { + "epoch": 2.0091591400585167, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7937179803848267, + "num_tokens": 602552700.0, + "step": 15794 + }, + { + "epoch": 2.0092863503371072, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7991937398910522, + "num_tokens": 602587779.0, + "step": 15795 + }, + { + "epoch": 2.0094135606156978, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7828397154808044, + "num_tokens": 602623834.0, + "step": 15796 + }, + { + "epoch": 2.0095407708942883, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8233612775802612, + "num_tokens": 602660141.0, + "step": 15797 + }, + { + "epoch": 2.009667981172879, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8080703020095825, + "num_tokens": 602698990.0, + "step": 15798 + }, + { + "epoch": 2.0097951914514693, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7796785235404968, + "num_tokens": 602732053.0, + "step": 15799 + }, + { + "epoch": 2.00992240173006, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8049196600914001, + "num_tokens": 602770975.0, + "step": 15800 + }, + { + "epoch": 2.0100496120086504, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7913984060287476, + "num_tokens": 602813207.0, + "step": 15801 + }, + { + "epoch": 2.010176822287241, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8047434091567993, + "num_tokens": 602849951.0, + "step": 15802 + }, + { + "epoch": 2.0103040325658315, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.795039176940918, + "num_tokens": 602891418.0, + "step": 15803 + }, + { + "epoch": 2.010431242844422, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8082154989242554, + "num_tokens": 602933039.0, + "step": 15804 + }, + { + "epoch": 2.0105584531230125, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.811582088470459, + "num_tokens": 602965546.0, + "step": 15805 + }, + { + "epoch": 2.010685663401603, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8073219656944275, + "num_tokens": 603004473.0, + "step": 15806 + }, + { + "epoch": 2.0108128736801936, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7921719551086426, + "num_tokens": 603041644.0, + "step": 15807 + }, + { + "epoch": 2.0109400839587837, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7848411202430725, + "num_tokens": 603084698.0, + "step": 15808 + }, + { + "epoch": 2.011067294237374, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7797630429267883, + "num_tokens": 603121059.0, + "step": 15809 + }, + { + "epoch": 2.0111945045159647, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.8004521727561951, + "num_tokens": 603156604.0, + "step": 15810 + }, + { + "epoch": 2.0113217147945552, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7976497411727905, + "num_tokens": 603183743.0, + "step": 15811 + }, + { + "epoch": 2.0114489250731458, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7768129110336304, + "num_tokens": 603222099.0, + "step": 15812 + }, + { + "epoch": 2.0115761353517363, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7798548936843872, + "num_tokens": 603263467.0, + "step": 15813 + }, + { + "epoch": 2.011703345630327, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7755478024482727, + "num_tokens": 603296726.0, + "step": 15814 + }, + { + "epoch": 2.0118305559089174, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.804893970489502, + "num_tokens": 603329691.0, + "step": 15815 + }, + { + "epoch": 2.011957766187508, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8129428029060364, + "num_tokens": 603371619.0, + "step": 15816 + }, + { + "epoch": 2.0120849764660984, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7966570854187012, + "num_tokens": 603400018.0, + "step": 15817 + }, + { + "epoch": 2.012212186744689, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8010109066963196, + "num_tokens": 603436971.0, + "step": 15818 + }, + { + "epoch": 2.0123393970232795, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8046655058860779, + "num_tokens": 603472802.0, + "step": 15819 + }, + { + "epoch": 2.01246660730187, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8024832010269165, + "num_tokens": 603506547.0, + "step": 15820 + }, + { + "epoch": 2.0125938175804605, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7811223268508911, + "num_tokens": 603544967.0, + "step": 15821 + }, + { + "epoch": 2.012721027859051, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7952611446380615, + "num_tokens": 603582522.0, + "step": 15822 + }, + { + "epoch": 2.0128482381376416, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8127565383911133, + "num_tokens": 603619584.0, + "step": 15823 + }, + { + "epoch": 2.012975448416232, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.780677318572998, + "num_tokens": 603660542.0, + "step": 15824 + }, + { + "epoch": 2.0131026586948226, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8032437562942505, + "num_tokens": 603698900.0, + "step": 15825 + }, + { + "epoch": 2.013229868973413, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7792567014694214, + "num_tokens": 603733019.0, + "step": 15826 + }, + { + "epoch": 2.0133570792520037, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8050485849380493, + "num_tokens": 603772071.0, + "step": 15827 + }, + { + "epoch": 2.013484289530594, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7938277721405029, + "num_tokens": 603811008.0, + "step": 15828 + }, + { + "epoch": 2.0136114998091847, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7904244661331177, + "num_tokens": 603852647.0, + "step": 15829 + }, + { + "epoch": 2.0137387100877753, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8011839389801025, + "num_tokens": 603894198.0, + "step": 15830 + }, + { + "epoch": 2.013865920366366, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7992510795593262, + "num_tokens": 603930168.0, + "step": 15831 + }, + { + "epoch": 2.0139931306449563, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.812401294708252, + "num_tokens": 603966189.0, + "step": 15832 + }, + { + "epoch": 2.0141203409235464, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7978150844573975, + "num_tokens": 604011110.0, + "step": 15833 + }, + { + "epoch": 2.014247551202137, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7779856324195862, + "num_tokens": 604044256.0, + "step": 15834 + }, + { + "epoch": 2.0143747614807275, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8046690821647644, + "num_tokens": 604078652.0, + "step": 15835 + }, + { + "epoch": 2.014501971759318, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8076993227005005, + "num_tokens": 604112302.0, + "step": 15836 + }, + { + "epoch": 2.0146291820379085, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8142322301864624, + "num_tokens": 604150011.0, + "step": 15837 + }, + { + "epoch": 2.014756392316499, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8016362190246582, + "num_tokens": 604188859.0, + "step": 15838 + }, + { + "epoch": 2.0148836025950896, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7976360321044922, + "num_tokens": 604232273.0, + "step": 15839 + }, + { + "epoch": 2.01501081287368, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8059084415435791, + "num_tokens": 604265236.0, + "step": 15840 + }, + { + "epoch": 2.0151380231522706, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8133231997489929, + "num_tokens": 604304784.0, + "step": 15841 + }, + { + "epoch": 2.015265233430861, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5693, + "mean_token_accuracy": 0.8274934887886047, + "num_tokens": 604341461.0, + "step": 15842 + }, + { + "epoch": 2.0153924437094517, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8044973611831665, + "num_tokens": 604378628.0, + "step": 15843 + }, + { + "epoch": 2.0155196539880422, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8118606805801392, + "num_tokens": 604418070.0, + "step": 15844 + }, + { + "epoch": 2.0156468642666328, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8089022636413574, + "num_tokens": 604454445.0, + "step": 15845 + }, + { + "epoch": 2.0157740745452233, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8049782514572144, + "num_tokens": 604488559.0, + "step": 15846 + }, + { + "epoch": 2.015901284823814, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7861121296882629, + "num_tokens": 604522538.0, + "step": 15847 + }, + { + "epoch": 2.0160284951024043, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8216784596443176, + "num_tokens": 604561750.0, + "step": 15848 + }, + { + "epoch": 2.016155705380995, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8149237632751465, + "num_tokens": 604594961.0, + "step": 15849 + }, + { + "epoch": 2.0162829156595854, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7843586206436157, + "num_tokens": 604631301.0, + "step": 15850 + }, + { + "epoch": 2.016410125938176, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7817263603210449, + "num_tokens": 604666049.0, + "step": 15851 + }, + { + "epoch": 2.0165373362167665, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.8206169009208679, + "num_tokens": 604701984.0, + "step": 15852 + }, + { + "epoch": 2.016664546495357, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8133012056350708, + "num_tokens": 604738414.0, + "step": 15853 + }, + { + "epoch": 2.0167917567739475, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8045267462730408, + "num_tokens": 604779961.0, + "step": 15854 + }, + { + "epoch": 2.016918967052538, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7933374047279358, + "num_tokens": 604819729.0, + "step": 15855 + }, + { + "epoch": 2.0170461773311286, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.8178441524505615, + "num_tokens": 604864108.0, + "step": 15856 + }, + { + "epoch": 2.0171733876097186, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7959204316139221, + "num_tokens": 604905582.0, + "step": 15857 + }, + { + "epoch": 2.017300597888309, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8097710013389587, + "num_tokens": 604943931.0, + "step": 15858 + }, + { + "epoch": 2.0174278081668997, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7994207143783569, + "num_tokens": 604982100.0, + "step": 15859 + }, + { + "epoch": 2.0175550184454902, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7891506552696228, + "num_tokens": 605026289.0, + "step": 15860 + }, + { + "epoch": 2.0176822287240808, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8077921867370605, + "num_tokens": 605070214.0, + "step": 15861 + }, + { + "epoch": 2.0178094390026713, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7974405288696289, + "num_tokens": 605113495.0, + "step": 15862 + }, + { + "epoch": 2.017936649281262, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8048469424247742, + "num_tokens": 605145717.0, + "step": 15863 + }, + { + "epoch": 2.0180638595598523, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7957207560539246, + "num_tokens": 605180930.0, + "step": 15864 + }, + { + "epoch": 2.018191069838443, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8094395399093628, + "num_tokens": 605214004.0, + "step": 15865 + }, + { + "epoch": 2.0183182801170334, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8043699860572815, + "num_tokens": 605253244.0, + "step": 15866 + }, + { + "epoch": 2.018445490395624, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7954582571983337, + "num_tokens": 605293832.0, + "step": 15867 + }, + { + "epoch": 2.0185727006742145, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8100216388702393, + "num_tokens": 605333811.0, + "step": 15868 + }, + { + "epoch": 2.018699910952805, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8000566959381104, + "num_tokens": 605371560.0, + "step": 15869 + }, + { + "epoch": 2.0188271212313955, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8051782846450806, + "num_tokens": 605410555.0, + "step": 15870 + }, + { + "epoch": 2.018954331509986, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7977637052536011, + "num_tokens": 605451851.0, + "step": 15871 + }, + { + "epoch": 2.0190815417885766, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7872026562690735, + "num_tokens": 605490442.0, + "step": 15872 + }, + { + "epoch": 2.019208752067167, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7940762042999268, + "num_tokens": 605525237.0, + "step": 15873 + }, + { + "epoch": 2.0193359623457576, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.804095447063446, + "num_tokens": 605563624.0, + "step": 15874 + }, + { + "epoch": 2.019463172624348, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8122798204421997, + "num_tokens": 605597419.0, + "step": 15875 + }, + { + "epoch": 2.0195903829029387, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7891809940338135, + "num_tokens": 605639954.0, + "step": 15876 + }, + { + "epoch": 2.019717593181529, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7979647517204285, + "num_tokens": 605678226.0, + "step": 15877 + }, + { + "epoch": 2.0198448034601197, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8150819540023804, + "num_tokens": 605719925.0, + "step": 15878 + }, + { + "epoch": 2.0199720137387103, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7816280126571655, + "num_tokens": 605761282.0, + "step": 15879 + }, + { + "epoch": 2.020099224017301, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.814507246017456, + "num_tokens": 605800681.0, + "step": 15880 + }, + { + "epoch": 2.0202264342958913, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7779586315155029, + "num_tokens": 605836282.0, + "step": 15881 + }, + { + "epoch": 2.0203536445744814, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8043078184127808, + "num_tokens": 605880360.0, + "step": 15882 + }, + { + "epoch": 2.020480854853072, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8118274211883545, + "num_tokens": 605918363.0, + "step": 15883 + }, + { + "epoch": 2.0206080651316625, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.8163743615150452, + "num_tokens": 605953032.0, + "step": 15884 + }, + { + "epoch": 2.020735275410253, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.8040781617164612, + "num_tokens": 605992633.0, + "step": 15885 + }, + { + "epoch": 2.0208624856888435, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.800783634185791, + "num_tokens": 606031776.0, + "step": 15886 + }, + { + "epoch": 2.020989695967434, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7945033311843872, + "num_tokens": 606070614.0, + "step": 15887 + }, + { + "epoch": 2.0211169062460246, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8021386861801147, + "num_tokens": 606115898.0, + "step": 15888 + }, + { + "epoch": 2.021244116524615, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.799666166305542, + "num_tokens": 606150231.0, + "step": 15889 + }, + { + "epoch": 2.0213713268032056, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7920924425125122, + "num_tokens": 606191527.0, + "step": 15890 + }, + { + "epoch": 2.021498537081796, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7906209826469421, + "num_tokens": 606229070.0, + "step": 15891 + }, + { + "epoch": 2.0216257473603867, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7995910048484802, + "num_tokens": 606269555.0, + "step": 15892 + }, + { + "epoch": 2.021752957638977, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7932512760162354, + "num_tokens": 606302040.0, + "step": 15893 + }, + { + "epoch": 2.0218801679175677, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8114497661590576, + "num_tokens": 606341769.0, + "step": 15894 + }, + { + "epoch": 2.0220073781961583, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.783073365688324, + "num_tokens": 606379989.0, + "step": 15895 + }, + { + "epoch": 2.022134588474749, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7924556732177734, + "num_tokens": 606414396.0, + "step": 15896 + }, + { + "epoch": 2.0222617987533393, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8033959865570068, + "num_tokens": 606445949.0, + "step": 15897 + }, + { + "epoch": 2.02238900903193, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8064969778060913, + "num_tokens": 606484751.0, + "step": 15898 + }, + { + "epoch": 2.0225162193105204, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7987448573112488, + "num_tokens": 606525390.0, + "step": 15899 + }, + { + "epoch": 2.022643429589111, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7969797253608704, + "num_tokens": 606563364.0, + "step": 15900 + }, + { + "epoch": 2.0227706398677014, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8035191297531128, + "num_tokens": 606598140.0, + "step": 15901 + }, + { + "epoch": 2.022897850146292, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7944958806037903, + "num_tokens": 606636480.0, + "step": 15902 + }, + { + "epoch": 2.0230250604248825, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8072694540023804, + "num_tokens": 606676044.0, + "step": 15903 + }, + { + "epoch": 2.023152270703473, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7904106378555298, + "num_tokens": 606713262.0, + "step": 15904 + }, + { + "epoch": 2.0232794809820636, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7984885573387146, + "num_tokens": 606754844.0, + "step": 15905 + }, + { + "epoch": 2.0234066912606536, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8021916151046753, + "num_tokens": 606800910.0, + "step": 15906 + }, + { + "epoch": 2.023533901539244, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7936325669288635, + "num_tokens": 606841855.0, + "step": 15907 + }, + { + "epoch": 2.0236611118178347, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7926431894302368, + "num_tokens": 606888261.0, + "step": 15908 + }, + { + "epoch": 2.0237883220964252, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7920438051223755, + "num_tokens": 606926235.0, + "step": 15909 + }, + { + "epoch": 2.0239155323750158, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.79298996925354, + "num_tokens": 606962527.0, + "step": 15910 + }, + { + "epoch": 2.0240427426536063, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8063352108001709, + "num_tokens": 607008924.0, + "step": 15911 + }, + { + "epoch": 2.024169952932197, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8001409769058228, + "num_tokens": 607049788.0, + "step": 15912 + }, + { + "epoch": 2.0242971632107873, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7937743663787842, + "num_tokens": 607087592.0, + "step": 15913 + }, + { + "epoch": 2.024424373489378, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8124361634254456, + "num_tokens": 607119954.0, + "step": 15914 + }, + { + "epoch": 2.0245515837679684, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.784164309501648, + "num_tokens": 607156511.0, + "step": 15915 + }, + { + "epoch": 2.024678794046559, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.807702898979187, + "num_tokens": 607198292.0, + "step": 15916 + }, + { + "epoch": 2.0248060043251495, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.801768958568573, + "num_tokens": 607237572.0, + "step": 15917 + }, + { + "epoch": 2.02493321460374, + "grad_norm": 10.5, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7940855622291565, + "num_tokens": 607282321.0, + "step": 15918 + }, + { + "epoch": 2.0250604248823305, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7959425449371338, + "num_tokens": 607313355.0, + "step": 15919 + }, + { + "epoch": 2.025187635160921, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7959608435630798, + "num_tokens": 607351961.0, + "step": 15920 + }, + { + "epoch": 2.0253148454395116, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8029425144195557, + "num_tokens": 607391368.0, + "step": 15921 + }, + { + "epoch": 2.025442055718102, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8200565576553345, + "num_tokens": 607427328.0, + "step": 15922 + }, + { + "epoch": 2.0255692659966926, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7962848544120789, + "num_tokens": 607468243.0, + "step": 15923 + }, + { + "epoch": 2.025696476275283, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.774007260799408, + "num_tokens": 607502466.0, + "step": 15924 + }, + { + "epoch": 2.0258236865538737, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7889922857284546, + "num_tokens": 607544093.0, + "step": 15925 + }, + { + "epoch": 2.025950896832464, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8070874214172363, + "num_tokens": 607576054.0, + "step": 15926 + }, + { + "epoch": 2.0260781071110547, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8149824142456055, + "num_tokens": 607608462.0, + "step": 15927 + }, + { + "epoch": 2.0262053173896453, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7844862341880798, + "num_tokens": 607645272.0, + "step": 15928 + }, + { + "epoch": 2.026332527668236, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7998921871185303, + "num_tokens": 607682585.0, + "step": 15929 + }, + { + "epoch": 2.0264597379468263, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8213554620742798, + "num_tokens": 607721302.0, + "step": 15930 + }, + { + "epoch": 2.0265869482254164, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7973471879959106, + "num_tokens": 607758357.0, + "step": 15931 + }, + { + "epoch": 2.026714158504007, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8013150691986084, + "num_tokens": 607795467.0, + "step": 15932 + }, + { + "epoch": 2.0268413687825975, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7959854006767273, + "num_tokens": 607833192.0, + "step": 15933 + }, + { + "epoch": 2.026968579061188, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8102740049362183, + "num_tokens": 607866672.0, + "step": 15934 + }, + { + "epoch": 2.0270957893397785, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8054323196411133, + "num_tokens": 607907976.0, + "step": 15935 + }, + { + "epoch": 2.027222999618369, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8015785217285156, + "num_tokens": 607946009.0, + "step": 15936 + }, + { + "epoch": 2.0273502098969596, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8088967800140381, + "num_tokens": 607986800.0, + "step": 15937 + }, + { + "epoch": 2.02747742017555, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.792789876461029, + "num_tokens": 608021807.0, + "step": 15938 + }, + { + "epoch": 2.0276046304541406, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7994561195373535, + "num_tokens": 608060294.0, + "step": 15939 + }, + { + "epoch": 2.027731840732731, + "grad_norm": 78.0, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7978466153144836, + "num_tokens": 608100790.0, + "step": 15940 + }, + { + "epoch": 2.0278590510113217, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.794438362121582, + "num_tokens": 608142164.0, + "step": 15941 + }, + { + "epoch": 2.027986261289912, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8017153739929199, + "num_tokens": 608184175.0, + "step": 15942 + }, + { + "epoch": 2.0281134715685027, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7947208881378174, + "num_tokens": 608226801.0, + "step": 15943 + }, + { + "epoch": 2.0282406818470933, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8074625730514526, + "num_tokens": 608265155.0, + "step": 15944 + }, + { + "epoch": 2.028367892125684, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.810447633266449, + "num_tokens": 608303817.0, + "step": 15945 + }, + { + "epoch": 2.0284951024042743, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8094905018806458, + "num_tokens": 608338845.0, + "step": 15946 + }, + { + "epoch": 2.028622312682865, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8032998442649841, + "num_tokens": 608377661.0, + "step": 15947 + }, + { + "epoch": 2.0287495229614554, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.797272801399231, + "num_tokens": 608412897.0, + "step": 15948 + }, + { + "epoch": 2.028876733240046, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7924014925956726, + "num_tokens": 608453584.0, + "step": 15949 + }, + { + "epoch": 2.0290039435186364, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.797432541847229, + "num_tokens": 608492211.0, + "step": 15950 + }, + { + "epoch": 2.029131153797227, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7957462072372437, + "num_tokens": 608534317.0, + "step": 15951 + }, + { + "epoch": 2.0292583640758175, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.8159209489822388, + "num_tokens": 608569763.0, + "step": 15952 + }, + { + "epoch": 2.029385574354408, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8038898706436157, + "num_tokens": 608612145.0, + "step": 15953 + }, + { + "epoch": 2.0295127846329986, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7981138229370117, + "num_tokens": 608649528.0, + "step": 15954 + }, + { + "epoch": 2.0296399949115886, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7949517965316772, + "num_tokens": 608686973.0, + "step": 15955 + }, + { + "epoch": 2.029767205190179, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8040549755096436, + "num_tokens": 608729169.0, + "step": 15956 + }, + { + "epoch": 2.0298944154687697, + "grad_norm": 66.5, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7993610501289368, + "num_tokens": 608764502.0, + "step": 15957 + }, + { + "epoch": 2.0300216257473602, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8070781826972961, + "num_tokens": 608805190.0, + "step": 15958 + }, + { + "epoch": 2.0301488360259508, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8024992346763611, + "num_tokens": 608840006.0, + "step": 15959 + }, + { + "epoch": 2.0302760463045413, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8059918880462646, + "num_tokens": 608871551.0, + "step": 15960 + }, + { + "epoch": 2.030403256583132, + "grad_norm": 95.0, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.77976393699646, + "num_tokens": 608909610.0, + "step": 15961 + }, + { + "epoch": 2.0305304668617223, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.799092173576355, + "num_tokens": 608950758.0, + "step": 15962 + }, + { + "epoch": 2.030657677140313, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7824597358703613, + "num_tokens": 608990695.0, + "step": 15963 + }, + { + "epoch": 2.0307848874189034, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8181543350219727, + "num_tokens": 609030978.0, + "step": 15964 + }, + { + "epoch": 2.030912097697494, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.815015435218811, + "num_tokens": 609071974.0, + "step": 15965 + }, + { + "epoch": 2.0310393079760845, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8051403760910034, + "num_tokens": 609111268.0, + "step": 15966 + }, + { + "epoch": 2.031166518254675, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7977297306060791, + "num_tokens": 609149681.0, + "step": 15967 + }, + { + "epoch": 2.0312937285332655, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.804947018623352, + "num_tokens": 609189121.0, + "step": 15968 + }, + { + "epoch": 2.031420938811856, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8115010261535645, + "num_tokens": 609228070.0, + "step": 15969 + }, + { + "epoch": 2.0315481490904466, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8055723905563354, + "num_tokens": 609269718.0, + "step": 15970 + }, + { + "epoch": 2.031675359369037, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8049638271331787, + "num_tokens": 609307236.0, + "step": 15971 + }, + { + "epoch": 2.0318025696476276, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7794142365455627, + "num_tokens": 609342993.0, + "step": 15972 + }, + { + "epoch": 2.031929779926218, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8113396167755127, + "num_tokens": 609376817.0, + "step": 15973 + }, + { + "epoch": 2.0320569902048087, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7945122718811035, + "num_tokens": 609411209.0, + "step": 15974 + }, + { + "epoch": 2.032184200483399, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8170795440673828, + "num_tokens": 609449055.0, + "step": 15975 + }, + { + "epoch": 2.0323114107619897, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7961477637290955, + "num_tokens": 609486332.0, + "step": 15976 + }, + { + "epoch": 2.0324386210405803, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7868013381958008, + "num_tokens": 609526924.0, + "step": 15977 + }, + { + "epoch": 2.032565831319171, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8170773983001709, + "num_tokens": 609559542.0, + "step": 15978 + }, + { + "epoch": 2.032693041597761, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7884911298751831, + "num_tokens": 609598989.0, + "step": 15979 + }, + { + "epoch": 2.0328202518763514, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7839432954788208, + "num_tokens": 609635790.0, + "step": 15980 + }, + { + "epoch": 2.032947462154942, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7896409034729004, + "num_tokens": 609670120.0, + "step": 15981 + }, + { + "epoch": 2.0330746724335325, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7983725070953369, + "num_tokens": 609713260.0, + "step": 15982 + }, + { + "epoch": 2.033201882712123, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8081610798835754, + "num_tokens": 609753236.0, + "step": 15983 + }, + { + "epoch": 2.0333290929907135, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8019391298294067, + "num_tokens": 609791629.0, + "step": 15984 + }, + { + "epoch": 2.033456303269304, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.782193660736084, + "num_tokens": 609833457.0, + "step": 15985 + }, + { + "epoch": 2.0335835135478946, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.79001384973526, + "num_tokens": 609866007.0, + "step": 15986 + }, + { + "epoch": 2.033710723826485, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8050497770309448, + "num_tokens": 609903346.0, + "step": 15987 + }, + { + "epoch": 2.0338379341050756, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8061264753341675, + "num_tokens": 609935885.0, + "step": 15988 + }, + { + "epoch": 2.033965144383666, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7897326946258545, + "num_tokens": 609971127.0, + "step": 15989 + }, + { + "epoch": 2.0340923546622567, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7922998070716858, + "num_tokens": 610016719.0, + "step": 15990 + }, + { + "epoch": 2.034219564940847, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7950159311294556, + "num_tokens": 610057002.0, + "step": 15991 + }, + { + "epoch": 2.0343467752194377, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8070206046104431, + "num_tokens": 610092779.0, + "step": 15992 + }, + { + "epoch": 2.0344739854980283, + "grad_norm": 10.375, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7912783622741699, + "num_tokens": 610137092.0, + "step": 15993 + }, + { + "epoch": 2.034601195776619, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.815543532371521, + "num_tokens": 610178632.0, + "step": 15994 + }, + { + "epoch": 2.0347284060552093, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.8006588220596313, + "num_tokens": 610219789.0, + "step": 15995 + }, + { + "epoch": 2.0348556163338, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7855148315429688, + "num_tokens": 610259104.0, + "step": 15996 + }, + { + "epoch": 2.0349828266123904, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7916039824485779, + "num_tokens": 610296317.0, + "step": 15997 + }, + { + "epoch": 2.035110036890981, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8047071695327759, + "num_tokens": 610339300.0, + "step": 15998 + }, + { + "epoch": 2.0352372471695714, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7927266955375671, + "num_tokens": 610373917.0, + "step": 15999 + }, + { + "epoch": 2.035364457448162, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7989798188209534, + "num_tokens": 610419571.0, + "step": 16000 + }, + { + "epoch": 2.0354916677267525, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7954561710357666, + "num_tokens": 610458073.0, + "step": 16001 + }, + { + "epoch": 2.035618878005343, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7841978669166565, + "num_tokens": 610497223.0, + "step": 16002 + }, + { + "epoch": 2.0357460882839336, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7765927314758301, + "num_tokens": 610532563.0, + "step": 16003 + }, + { + "epoch": 2.0358732985625236, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8140301704406738, + "num_tokens": 610566644.0, + "step": 16004 + }, + { + "epoch": 2.036000508841114, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.781609296798706, + "num_tokens": 610604755.0, + "step": 16005 + }, + { + "epoch": 2.0361277191197047, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8055769801139832, + "num_tokens": 610645905.0, + "step": 16006 + }, + { + "epoch": 2.036254929398295, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8186755776405334, + "num_tokens": 610686800.0, + "step": 16007 + }, + { + "epoch": 2.0363821396768857, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7984851598739624, + "num_tokens": 610728291.0, + "step": 16008 + }, + { + "epoch": 2.0365093499554763, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7888237237930298, + "num_tokens": 610759818.0, + "step": 16009 + }, + { + "epoch": 2.036636560234067, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.783016562461853, + "num_tokens": 610800125.0, + "step": 16010 + }, + { + "epoch": 2.0367637705126573, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7937308549880981, + "num_tokens": 610839261.0, + "step": 16011 + }, + { + "epoch": 2.036890980791248, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8090255260467529, + "num_tokens": 610879092.0, + "step": 16012 + }, + { + "epoch": 2.0370181910698384, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.794455885887146, + "num_tokens": 610914494.0, + "step": 16013 + }, + { + "epoch": 2.037145401348429, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8137956261634827, + "num_tokens": 610950784.0, + "step": 16014 + }, + { + "epoch": 2.0372726116270194, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7898535132408142, + "num_tokens": 610989254.0, + "step": 16015 + }, + { + "epoch": 2.03739982190561, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7892393469810486, + "num_tokens": 611022668.0, + "step": 16016 + }, + { + "epoch": 2.0375270321842005, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7989377975463867, + "num_tokens": 611062110.0, + "step": 16017 + }, + { + "epoch": 2.037654242462791, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8034971952438354, + "num_tokens": 611098663.0, + "step": 16018 + }, + { + "epoch": 2.0377814527413816, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8030646443367004, + "num_tokens": 611134058.0, + "step": 16019 + }, + { + "epoch": 2.037908663019972, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7971811294555664, + "num_tokens": 611174748.0, + "step": 16020 + }, + { + "epoch": 2.0380358732985626, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8230609893798828, + "num_tokens": 611216792.0, + "step": 16021 + }, + { + "epoch": 2.038163083577153, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8105176687240601, + "num_tokens": 611253996.0, + "step": 16022 + }, + { + "epoch": 2.0382902938557437, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8066474199295044, + "num_tokens": 611295351.0, + "step": 16023 + }, + { + "epoch": 2.038417504134334, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.799738347530365, + "num_tokens": 611330513.0, + "step": 16024 + }, + { + "epoch": 2.0385447144129247, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8108724355697632, + "num_tokens": 611369679.0, + "step": 16025 + }, + { + "epoch": 2.0386719246915153, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.809108555316925, + "num_tokens": 611410398.0, + "step": 16026 + }, + { + "epoch": 2.038799134970106, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7926340699195862, + "num_tokens": 611447151.0, + "step": 16027 + }, + { + "epoch": 2.0389263452486963, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7894701361656189, + "num_tokens": 611486923.0, + "step": 16028 + }, + { + "epoch": 2.0390535555272864, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8076882362365723, + "num_tokens": 611519502.0, + "step": 16029 + }, + { + "epoch": 2.039180765805877, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7884940505027771, + "num_tokens": 611557647.0, + "step": 16030 + }, + { + "epoch": 2.0393079760844675, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.7990618944168091, + "num_tokens": 611597405.0, + "step": 16031 + }, + { + "epoch": 2.039435186363058, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7908119559288025, + "num_tokens": 611635996.0, + "step": 16032 + }, + { + "epoch": 2.0395623966416485, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8116490840911865, + "num_tokens": 611677900.0, + "step": 16033 + }, + { + "epoch": 2.039689606920239, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8064745664596558, + "num_tokens": 611719184.0, + "step": 16034 + }, + { + "epoch": 2.0398168171988296, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7996916770935059, + "num_tokens": 611761270.0, + "step": 16035 + }, + { + "epoch": 2.03994402747742, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7999613285064697, + "num_tokens": 611794939.0, + "step": 16036 + }, + { + "epoch": 2.0400712377560106, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8140389919281006, + "num_tokens": 611834093.0, + "step": 16037 + }, + { + "epoch": 2.040198448034601, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7954869270324707, + "num_tokens": 611871327.0, + "step": 16038 + }, + { + "epoch": 2.0403256583131917, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7931659817695618, + "num_tokens": 611907837.0, + "step": 16039 + }, + { + "epoch": 2.040452868591782, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7822935581207275, + "num_tokens": 611950964.0, + "step": 16040 + }, + { + "epoch": 2.0405800788703727, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7942512035369873, + "num_tokens": 611985867.0, + "step": 16041 + }, + { + "epoch": 2.0407072891489633, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.797392725944519, + "num_tokens": 612017819.0, + "step": 16042 + }, + { + "epoch": 2.040834499427554, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7837950587272644, + "num_tokens": 612060669.0, + "step": 16043 + }, + { + "epoch": 2.0409617097061443, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7822422981262207, + "num_tokens": 612099801.0, + "step": 16044 + }, + { + "epoch": 2.041088919984735, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7981958389282227, + "num_tokens": 612139454.0, + "step": 16045 + }, + { + "epoch": 2.0412161302633254, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7804751992225647, + "num_tokens": 612181695.0, + "step": 16046 + }, + { + "epoch": 2.041343340541916, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8191078901290894, + "num_tokens": 612223200.0, + "step": 16047 + }, + { + "epoch": 2.0414705508205064, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8021427989006042, + "num_tokens": 612265946.0, + "step": 16048 + }, + { + "epoch": 2.041597761099097, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8151555061340332, + "num_tokens": 612301246.0, + "step": 16049 + }, + { + "epoch": 2.0417249713776875, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7858096361160278, + "num_tokens": 612337789.0, + "step": 16050 + }, + { + "epoch": 2.041852181656278, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.8138054609298706, + "num_tokens": 612372391.0, + "step": 16051 + }, + { + "epoch": 2.0419793919348685, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8022333979606628, + "num_tokens": 612408806.0, + "step": 16052 + }, + { + "epoch": 2.0421066022134586, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8069536685943604, + "num_tokens": 612446349.0, + "step": 16053 + }, + { + "epoch": 2.042233812492049, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8160184621810913, + "num_tokens": 612489899.0, + "step": 16054 + }, + { + "epoch": 2.0423610227706397, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.8014175295829773, + "num_tokens": 612532314.0, + "step": 16055 + }, + { + "epoch": 2.04248823304923, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8086020946502686, + "num_tokens": 612568033.0, + "step": 16056 + }, + { + "epoch": 2.0426154433278207, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8035857081413269, + "num_tokens": 612605160.0, + "step": 16057 + }, + { + "epoch": 2.0427426536064113, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7915496826171875, + "num_tokens": 612647851.0, + "step": 16058 + }, + { + "epoch": 2.042869863885002, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7980875372886658, + "num_tokens": 612684770.0, + "step": 16059 + }, + { + "epoch": 2.0429970741635923, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7865819931030273, + "num_tokens": 612721689.0, + "step": 16060 + }, + { + "epoch": 2.043124284442183, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7956938147544861, + "num_tokens": 612757256.0, + "step": 16061 + }, + { + "epoch": 2.0432514947207734, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8047429919242859, + "num_tokens": 612797806.0, + "step": 16062 + }, + { + "epoch": 2.043378704999364, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8115838766098022, + "num_tokens": 612833507.0, + "step": 16063 + }, + { + "epoch": 2.0435059152779544, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8039119243621826, + "num_tokens": 612872920.0, + "step": 16064 + }, + { + "epoch": 2.043633125556545, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8093097805976868, + "num_tokens": 612909534.0, + "step": 16065 + }, + { + "epoch": 2.0437603358351355, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7971140146255493, + "num_tokens": 612950531.0, + "step": 16066 + }, + { + "epoch": 2.043887546113726, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7975594997406006, + "num_tokens": 612984638.0, + "step": 16067 + }, + { + "epoch": 2.0440147563923166, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8134276270866394, + "num_tokens": 613020632.0, + "step": 16068 + }, + { + "epoch": 2.044141966670907, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8021185398101807, + "num_tokens": 613057381.0, + "step": 16069 + }, + { + "epoch": 2.0442691769494976, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7987747192382812, + "num_tokens": 613094287.0, + "step": 16070 + }, + { + "epoch": 2.044396387228088, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8047155737876892, + "num_tokens": 613126315.0, + "step": 16071 + }, + { + "epoch": 2.0445235975066787, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7897751927375793, + "num_tokens": 613160970.0, + "step": 16072 + }, + { + "epoch": 2.044650807785269, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8075817227363586, + "num_tokens": 613204685.0, + "step": 16073 + }, + { + "epoch": 2.0447780180638597, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8051469326019287, + "num_tokens": 613242394.0, + "step": 16074 + }, + { + "epoch": 2.0449052283424503, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8069996237754822, + "num_tokens": 613283517.0, + "step": 16075 + }, + { + "epoch": 2.045032438621041, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8033708333969116, + "num_tokens": 613317237.0, + "step": 16076 + }, + { + "epoch": 2.045159648899631, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8075335621833801, + "num_tokens": 613355741.0, + "step": 16077 + }, + { + "epoch": 2.0452868591782214, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7879756093025208, + "num_tokens": 613392131.0, + "step": 16078 + }, + { + "epoch": 2.045414069456812, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7903885841369629, + "num_tokens": 613433687.0, + "step": 16079 + }, + { + "epoch": 2.0455412797354025, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.80694580078125, + "num_tokens": 613472882.0, + "step": 16080 + }, + { + "epoch": 2.045668490013993, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.807121992111206, + "num_tokens": 613515869.0, + "step": 16081 + }, + { + "epoch": 2.0457957002925835, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7985303997993469, + "num_tokens": 613551835.0, + "step": 16082 + }, + { + "epoch": 2.045922910571174, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8016142249107361, + "num_tokens": 613592277.0, + "step": 16083 + }, + { + "epoch": 2.0460501208497646, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7954791784286499, + "num_tokens": 613625594.0, + "step": 16084 + }, + { + "epoch": 2.046177331128355, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8039145469665527, + "num_tokens": 613659924.0, + "step": 16085 + }, + { + "epoch": 2.0463045414069456, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.782302737236023, + "num_tokens": 613697067.0, + "step": 16086 + }, + { + "epoch": 2.046431751685536, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8101645708084106, + "num_tokens": 613738211.0, + "step": 16087 + }, + { + "epoch": 2.0465589619641267, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.80935138463974, + "num_tokens": 613782590.0, + "step": 16088 + }, + { + "epoch": 2.046686172242717, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.809058666229248, + "num_tokens": 613825770.0, + "step": 16089 + }, + { + "epoch": 2.0468133825213077, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8046550154685974, + "num_tokens": 613864025.0, + "step": 16090 + }, + { + "epoch": 2.0469405927998983, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8039271831512451, + "num_tokens": 613907181.0, + "step": 16091 + }, + { + "epoch": 2.047067803078489, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.795696496963501, + "num_tokens": 613951258.0, + "step": 16092 + }, + { + "epoch": 2.0471950133570793, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8019485473632812, + "num_tokens": 613991221.0, + "step": 16093 + }, + { + "epoch": 2.04732222363567, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8119966983795166, + "num_tokens": 614026862.0, + "step": 16094 + }, + { + "epoch": 2.0474494339142604, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7947577238082886, + "num_tokens": 614067560.0, + "step": 16095 + }, + { + "epoch": 2.047576644192851, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.803220272064209, + "num_tokens": 614106974.0, + "step": 16096 + }, + { + "epoch": 2.0477038544714414, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.5569, + "mean_token_accuracy": 0.8276852369308472, + "num_tokens": 614145641.0, + "step": 16097 + }, + { + "epoch": 2.047831064750032, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.808006763458252, + "num_tokens": 614181476.0, + "step": 16098 + }, + { + "epoch": 2.0479582750286225, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7942711114883423, + "num_tokens": 614219212.0, + "step": 16099 + }, + { + "epoch": 2.048085485307213, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7756571769714355, + "num_tokens": 614264017.0, + "step": 16100 + }, + { + "epoch": 2.0482126955858035, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7832176685333252, + "num_tokens": 614299031.0, + "step": 16101 + }, + { + "epoch": 2.0483399058643936, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8082402348518372, + "num_tokens": 614334387.0, + "step": 16102 + }, + { + "epoch": 2.048467116142984, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8039748668670654, + "num_tokens": 614372438.0, + "step": 16103 + }, + { + "epoch": 2.0485943264215747, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7916706800460815, + "num_tokens": 614409976.0, + "step": 16104 + }, + { + "epoch": 2.048721536700165, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8106324076652527, + "num_tokens": 614448740.0, + "step": 16105 + }, + { + "epoch": 2.0488487469787557, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8050163388252258, + "num_tokens": 614484644.0, + "step": 16106 + }, + { + "epoch": 2.0489759572573463, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.783295214176178, + "num_tokens": 614521210.0, + "step": 16107 + }, + { + "epoch": 2.049103167535937, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.792878270149231, + "num_tokens": 614561783.0, + "step": 16108 + }, + { + "epoch": 2.0492303778145273, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8020604252815247, + "num_tokens": 614598643.0, + "step": 16109 + }, + { + "epoch": 2.049357588093118, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8007014393806458, + "num_tokens": 614633520.0, + "step": 16110 + }, + { + "epoch": 2.0494847983717084, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7869924306869507, + "num_tokens": 614673266.0, + "step": 16111 + }, + { + "epoch": 2.049612008650299, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7891595363616943, + "num_tokens": 614707201.0, + "step": 16112 + }, + { + "epoch": 2.0497392189288894, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8102436661720276, + "num_tokens": 614746542.0, + "step": 16113 + }, + { + "epoch": 2.04986642920748, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7882526516914368, + "num_tokens": 614781804.0, + "step": 16114 + }, + { + "epoch": 2.0499936394860705, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.801743745803833, + "num_tokens": 614817281.0, + "step": 16115 + }, + { + "epoch": 2.050120849764661, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7907811403274536, + "num_tokens": 614853666.0, + "step": 16116 + }, + { + "epoch": 2.0502480600432516, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7998819351196289, + "num_tokens": 614891221.0, + "step": 16117 + }, + { + "epoch": 2.050375270321842, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.80576092004776, + "num_tokens": 614925956.0, + "step": 16118 + }, + { + "epoch": 2.0505024806004326, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8121492862701416, + "num_tokens": 614959934.0, + "step": 16119 + }, + { + "epoch": 2.050629690879023, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.8001668453216553, + "num_tokens": 615000579.0, + "step": 16120 + }, + { + "epoch": 2.0507569011576137, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.800562858581543, + "num_tokens": 615037371.0, + "step": 16121 + }, + { + "epoch": 2.050884111436204, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7878214120864868, + "num_tokens": 615075948.0, + "step": 16122 + }, + { + "epoch": 2.0510113217147947, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.8168220520019531, + "num_tokens": 615114619.0, + "step": 16123 + }, + { + "epoch": 2.0511385319933853, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.800521969795227, + "num_tokens": 615149260.0, + "step": 16124 + }, + { + "epoch": 2.051265742271976, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8032753467559814, + "num_tokens": 615183284.0, + "step": 16125 + }, + { + "epoch": 2.0513929525505663, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8083024024963379, + "num_tokens": 615220015.0, + "step": 16126 + }, + { + "epoch": 2.0515201628291564, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7930771112442017, + "num_tokens": 615255614.0, + "step": 16127 + }, + { + "epoch": 2.051647373107747, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7986639142036438, + "num_tokens": 615290232.0, + "step": 16128 + }, + { + "epoch": 2.0517745833863374, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.813381552696228, + "num_tokens": 615328204.0, + "step": 16129 + }, + { + "epoch": 2.051901793664928, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7903181314468384, + "num_tokens": 615362846.0, + "step": 16130 + }, + { + "epoch": 2.0520290039435185, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.8009451627731323, + "num_tokens": 615405404.0, + "step": 16131 + }, + { + "epoch": 2.052156214222109, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7878784537315369, + "num_tokens": 615439835.0, + "step": 16132 + }, + { + "epoch": 2.0522834245006996, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8124495148658752, + "num_tokens": 615477832.0, + "step": 16133 + }, + { + "epoch": 2.05241063477929, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7888920903205872, + "num_tokens": 615518089.0, + "step": 16134 + }, + { + "epoch": 2.0525378450578806, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8053448796272278, + "num_tokens": 615557286.0, + "step": 16135 + }, + { + "epoch": 2.052665055336471, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8073745369911194, + "num_tokens": 615593284.0, + "step": 16136 + }, + { + "epoch": 2.0527922656150617, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7988216280937195, + "num_tokens": 615633579.0, + "step": 16137 + }, + { + "epoch": 2.052919475893652, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8142081499099731, + "num_tokens": 615680135.0, + "step": 16138 + }, + { + "epoch": 2.0530466861722427, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7859843969345093, + "num_tokens": 615714801.0, + "step": 16139 + }, + { + "epoch": 2.0531738964508333, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7908877730369568, + "num_tokens": 615750538.0, + "step": 16140 + }, + { + "epoch": 2.053301106729424, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8074355721473694, + "num_tokens": 615789465.0, + "step": 16141 + }, + { + "epoch": 2.0534283170080143, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7877042889595032, + "num_tokens": 615832553.0, + "step": 16142 + }, + { + "epoch": 2.053555527286605, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7957636117935181, + "num_tokens": 615868341.0, + "step": 16143 + }, + { + "epoch": 2.0536827375651954, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.810288667678833, + "num_tokens": 615904324.0, + "step": 16144 + }, + { + "epoch": 2.053809947843786, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8104891777038574, + "num_tokens": 615947656.0, + "step": 16145 + }, + { + "epoch": 2.0539371581223764, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7964441180229187, + "num_tokens": 615989050.0, + "step": 16146 + }, + { + "epoch": 2.054064368400967, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.805321216583252, + "num_tokens": 616027638.0, + "step": 16147 + }, + { + "epoch": 2.0541915786795575, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8044570088386536, + "num_tokens": 616066392.0, + "step": 16148 + }, + { + "epoch": 2.054318788958148, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8068686723709106, + "num_tokens": 616108303.0, + "step": 16149 + }, + { + "epoch": 2.0544459992367385, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7978804111480713, + "num_tokens": 616154174.0, + "step": 16150 + }, + { + "epoch": 2.0545732095153286, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.8004291653633118, + "num_tokens": 616193954.0, + "step": 16151 + }, + { + "epoch": 2.054700419793919, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8006449341773987, + "num_tokens": 616232771.0, + "step": 16152 + }, + { + "epoch": 2.0548276300725097, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8065345883369446, + "num_tokens": 616265151.0, + "step": 16153 + }, + { + "epoch": 2.0549548403511, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7929967641830444, + "num_tokens": 616303886.0, + "step": 16154 + }, + { + "epoch": 2.0550820506296907, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8082791566848755, + "num_tokens": 616340040.0, + "step": 16155 + }, + { + "epoch": 2.0552092609082813, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8056899905204773, + "num_tokens": 616371329.0, + "step": 16156 + }, + { + "epoch": 2.055336471186872, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8133190274238586, + "num_tokens": 616406383.0, + "step": 16157 + }, + { + "epoch": 2.0554636814654623, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.787886917591095, + "num_tokens": 616445600.0, + "step": 16158 + }, + { + "epoch": 2.055590891744053, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7965147495269775, + "num_tokens": 616485376.0, + "step": 16159 + }, + { + "epoch": 2.0557181020226434, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7870990037918091, + "num_tokens": 616527104.0, + "step": 16160 + }, + { + "epoch": 2.055845312301234, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7997378706932068, + "num_tokens": 616567557.0, + "step": 16161 + }, + { + "epoch": 2.0559725225798244, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.806826114654541, + "num_tokens": 616601612.0, + "step": 16162 + }, + { + "epoch": 2.056099732858415, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7982416152954102, + "num_tokens": 616638019.0, + "step": 16163 + }, + { + "epoch": 2.0562269431370055, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8050697445869446, + "num_tokens": 616675224.0, + "step": 16164 + }, + { + "epoch": 2.056354153415596, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8006811141967773, + "num_tokens": 616713628.0, + "step": 16165 + }, + { + "epoch": 2.0564813636941865, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8030275106430054, + "num_tokens": 616755985.0, + "step": 16166 + }, + { + "epoch": 2.056608573972777, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7999361157417297, + "num_tokens": 616786564.0, + "step": 16167 + }, + { + "epoch": 2.0567357842513676, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7707465291023254, + "num_tokens": 616823285.0, + "step": 16168 + }, + { + "epoch": 2.056862994529958, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8110072016716003, + "num_tokens": 616864465.0, + "step": 16169 + }, + { + "epoch": 2.0569902048085487, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7924294471740723, + "num_tokens": 616906670.0, + "step": 16170 + }, + { + "epoch": 2.057117415087139, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8055521249771118, + "num_tokens": 616947656.0, + "step": 16171 + }, + { + "epoch": 2.0572446253657297, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7910824418067932, + "num_tokens": 616980161.0, + "step": 16172 + }, + { + "epoch": 2.0573718356443202, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7962280511856079, + "num_tokens": 617015281.0, + "step": 16173 + }, + { + "epoch": 2.0574990459229108, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8013287782669067, + "num_tokens": 617054174.0, + "step": 16174 + }, + { + "epoch": 2.057626256201501, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8040469288825989, + "num_tokens": 617095160.0, + "step": 16175 + }, + { + "epoch": 2.0577534664800914, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8072041273117065, + "num_tokens": 617130781.0, + "step": 16176 + }, + { + "epoch": 2.057880676758682, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.794995129108429, + "num_tokens": 617169473.0, + "step": 16177 + }, + { + "epoch": 2.0580078870372724, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7974399924278259, + "num_tokens": 617212208.0, + "step": 16178 + }, + { + "epoch": 2.058135097315863, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8033264875411987, + "num_tokens": 617246055.0, + "step": 16179 + }, + { + "epoch": 2.0582623075944535, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7925578355789185, + "num_tokens": 617285614.0, + "step": 16180 + }, + { + "epoch": 2.058389517873044, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8105682134628296, + "num_tokens": 617323401.0, + "step": 16181 + }, + { + "epoch": 2.0585167281516346, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7908955812454224, + "num_tokens": 617372550.0, + "step": 16182 + }, + { + "epoch": 2.058643938430225, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8041882514953613, + "num_tokens": 617407894.0, + "step": 16183 + }, + { + "epoch": 2.0587711487088156, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7920045256614685, + "num_tokens": 617450117.0, + "step": 16184 + }, + { + "epoch": 2.058898358987406, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8005648851394653, + "num_tokens": 617485885.0, + "step": 16185 + }, + { + "epoch": 2.0590255692659967, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7826810479164124, + "num_tokens": 617520511.0, + "step": 16186 + }, + { + "epoch": 2.059152779544587, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7943142056465149, + "num_tokens": 617558288.0, + "step": 16187 + }, + { + "epoch": 2.0592799898231777, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8024473190307617, + "num_tokens": 617602991.0, + "step": 16188 + }, + { + "epoch": 2.0594072001017683, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.795498251914978, + "num_tokens": 617639629.0, + "step": 16189 + }, + { + "epoch": 2.059534410380359, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7956432104110718, + "num_tokens": 617676218.0, + "step": 16190 + }, + { + "epoch": 2.0596616206589493, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7950332760810852, + "num_tokens": 617716270.0, + "step": 16191 + }, + { + "epoch": 2.05978883093754, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8068313598632812, + "num_tokens": 617755559.0, + "step": 16192 + }, + { + "epoch": 2.0599160412161304, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.808382511138916, + "num_tokens": 617793522.0, + "step": 16193 + }, + { + "epoch": 2.060043251494721, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.796360969543457, + "num_tokens": 617828005.0, + "step": 16194 + }, + { + "epoch": 2.0601704617733114, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8047986030578613, + "num_tokens": 617863824.0, + "step": 16195 + }, + { + "epoch": 2.060297672051902, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7941104769706726, + "num_tokens": 617903648.0, + "step": 16196 + }, + { + "epoch": 2.0604248823304925, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7996025085449219, + "num_tokens": 617946229.0, + "step": 16197 + }, + { + "epoch": 2.060552092609083, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8016661405563354, + "num_tokens": 617990486.0, + "step": 16198 + }, + { + "epoch": 2.0606793028876735, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8163496851921082, + "num_tokens": 618031743.0, + "step": 16199 + }, + { + "epoch": 2.0608065131662636, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.798983633518219, + "num_tokens": 618070181.0, + "step": 16200 + }, + { + "epoch": 2.060933723444854, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7987592220306396, + "num_tokens": 618106396.0, + "step": 16201 + }, + { + "epoch": 2.0610609337234447, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8095564842224121, + "num_tokens": 618148281.0, + "step": 16202 + }, + { + "epoch": 2.061188144002035, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7929243445396423, + "num_tokens": 618182721.0, + "step": 16203 + }, + { + "epoch": 2.0613153542806257, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8197181820869446, + "num_tokens": 618219497.0, + "step": 16204 + }, + { + "epoch": 2.0614425645592163, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8231593370437622, + "num_tokens": 618258643.0, + "step": 16205 + }, + { + "epoch": 2.061569774837807, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8111302852630615, + "num_tokens": 618292682.0, + "step": 16206 + }, + { + "epoch": 2.0616969851163973, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7819806933403015, + "num_tokens": 618331375.0, + "step": 16207 + }, + { + "epoch": 2.061824195394988, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8034674525260925, + "num_tokens": 618367452.0, + "step": 16208 + }, + { + "epoch": 2.0619514056735784, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7911845445632935, + "num_tokens": 618410470.0, + "step": 16209 + }, + { + "epoch": 2.062078615952169, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7943919897079468, + "num_tokens": 618447524.0, + "step": 16210 + }, + { + "epoch": 2.0622058262307594, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7832472324371338, + "num_tokens": 618481793.0, + "step": 16211 + }, + { + "epoch": 2.06233303650935, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.808241605758667, + "num_tokens": 618517097.0, + "step": 16212 + }, + { + "epoch": 2.0624602467879405, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7930985689163208, + "num_tokens": 618554221.0, + "step": 16213 + }, + { + "epoch": 2.062587457066531, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8075345158576965, + "num_tokens": 618591279.0, + "step": 16214 + }, + { + "epoch": 2.0627146673451215, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7849323749542236, + "num_tokens": 618628168.0, + "step": 16215 + }, + { + "epoch": 2.062841877623712, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8055460453033447, + "num_tokens": 618664580.0, + "step": 16216 + }, + { + "epoch": 2.0629690879023026, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7885212898254395, + "num_tokens": 618699007.0, + "step": 16217 + }, + { + "epoch": 2.063096298180893, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8026698231697083, + "num_tokens": 618736630.0, + "step": 16218 + }, + { + "epoch": 2.0632235084594837, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8127431273460388, + "num_tokens": 618777868.0, + "step": 16219 + }, + { + "epoch": 2.063350718738074, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8052070140838623, + "num_tokens": 618819134.0, + "step": 16220 + }, + { + "epoch": 2.0634779290166647, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7981957793235779, + "num_tokens": 618862231.0, + "step": 16221 + }, + { + "epoch": 2.0636051392952552, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7921318411827087, + "num_tokens": 618906816.0, + "step": 16222 + }, + { + "epoch": 2.0637323495738458, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7977689504623413, + "num_tokens": 618943239.0, + "step": 16223 + }, + { + "epoch": 2.0638595598524363, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8135747909545898, + "num_tokens": 618978641.0, + "step": 16224 + }, + { + "epoch": 2.0639867701310264, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8046293258666992, + "num_tokens": 619020522.0, + "step": 16225 + }, + { + "epoch": 2.064113980409617, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8169417381286621, + "num_tokens": 619060769.0, + "step": 16226 + }, + { + "epoch": 2.0642411906882074, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.815352201461792, + "num_tokens": 619098711.0, + "step": 16227 + }, + { + "epoch": 2.064368400966798, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7949888706207275, + "num_tokens": 619133673.0, + "step": 16228 + }, + { + "epoch": 2.0644956112453885, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7965835928916931, + "num_tokens": 619174554.0, + "step": 16229 + }, + { + "epoch": 2.064622821523979, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7909282445907593, + "num_tokens": 619211095.0, + "step": 16230 + }, + { + "epoch": 2.0647500318025696, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8088474273681641, + "num_tokens": 619246346.0, + "step": 16231 + }, + { + "epoch": 2.06487724208116, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.796122670173645, + "num_tokens": 619280629.0, + "step": 16232 + }, + { + "epoch": 2.0650044523597506, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7754794359207153, + "num_tokens": 619320928.0, + "step": 16233 + }, + { + "epoch": 2.065131662638341, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8011531829833984, + "num_tokens": 619369747.0, + "step": 16234 + }, + { + "epoch": 2.0652588729169317, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8046373128890991, + "num_tokens": 619408694.0, + "step": 16235 + }, + { + "epoch": 2.065386083195522, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8114893436431885, + "num_tokens": 619446858.0, + "step": 16236 + }, + { + "epoch": 2.0655132934741127, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7982847690582275, + "num_tokens": 619482281.0, + "step": 16237 + }, + { + "epoch": 2.0656405037527032, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8061737418174744, + "num_tokens": 619520309.0, + "step": 16238 + }, + { + "epoch": 2.065767714031294, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8051966428756714, + "num_tokens": 619554662.0, + "step": 16239 + }, + { + "epoch": 2.0658949243098843, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7987791299819946, + "num_tokens": 619599632.0, + "step": 16240 + }, + { + "epoch": 2.066022134588475, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8161969184875488, + "num_tokens": 619639414.0, + "step": 16241 + }, + { + "epoch": 2.0661493448670654, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8076539039611816, + "num_tokens": 619675878.0, + "step": 16242 + }, + { + "epoch": 2.066276555145656, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.8019703030586243, + "num_tokens": 619709486.0, + "step": 16243 + }, + { + "epoch": 2.0664037654242464, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8065892457962036, + "num_tokens": 619753576.0, + "step": 16244 + }, + { + "epoch": 2.066530975702837, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8093899488449097, + "num_tokens": 619792520.0, + "step": 16245 + }, + { + "epoch": 2.0666581859814275, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7756545543670654, + "num_tokens": 619827871.0, + "step": 16246 + }, + { + "epoch": 2.066785396260018, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7953099012374878, + "num_tokens": 619864527.0, + "step": 16247 + }, + { + "epoch": 2.0669126065386085, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8031574487686157, + "num_tokens": 619904934.0, + "step": 16248 + }, + { + "epoch": 2.0670398168171986, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8114516735076904, + "num_tokens": 619945472.0, + "step": 16249 + }, + { + "epoch": 2.067167027095789, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.816191554069519, + "num_tokens": 619981738.0, + "step": 16250 + }, + { + "epoch": 2.0672942373743797, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8059185743331909, + "num_tokens": 620024593.0, + "step": 16251 + }, + { + "epoch": 2.06742144765297, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8061444759368896, + "num_tokens": 620064059.0, + "step": 16252 + }, + { + "epoch": 2.0675486579315607, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7958582043647766, + "num_tokens": 620100028.0, + "step": 16253 + }, + { + "epoch": 2.0676758682101513, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8108062744140625, + "num_tokens": 620135181.0, + "step": 16254 + }, + { + "epoch": 2.067803078488742, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8025758862495422, + "num_tokens": 620174887.0, + "step": 16255 + }, + { + "epoch": 2.0679302887673323, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8043826818466187, + "num_tokens": 620212924.0, + "step": 16256 + }, + { + "epoch": 2.068057499045923, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7853693962097168, + "num_tokens": 620251088.0, + "step": 16257 + }, + { + "epoch": 2.0681847093245134, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7951614260673523, + "num_tokens": 620287888.0, + "step": 16258 + }, + { + "epoch": 2.068311919603104, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7995563745498657, + "num_tokens": 620324937.0, + "step": 16259 + }, + { + "epoch": 2.0684391298816944, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7981029152870178, + "num_tokens": 620363390.0, + "step": 16260 + }, + { + "epoch": 2.068566340160285, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8072685599327087, + "num_tokens": 620403707.0, + "step": 16261 + }, + { + "epoch": 2.0686935504388755, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8065008521080017, + "num_tokens": 620445994.0, + "step": 16262 + }, + { + "epoch": 2.068820760717466, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.805349588394165, + "num_tokens": 620480883.0, + "step": 16263 + }, + { + "epoch": 2.0689479709960565, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8114737868309021, + "num_tokens": 620518935.0, + "step": 16264 + }, + { + "epoch": 2.069075181274647, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7934338450431824, + "num_tokens": 620551985.0, + "step": 16265 + }, + { + "epoch": 2.0692023915532376, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7960721254348755, + "num_tokens": 620592809.0, + "step": 16266 + }, + { + "epoch": 2.069329601831828, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.781015932559967, + "num_tokens": 620627951.0, + "step": 16267 + }, + { + "epoch": 2.0694568121104187, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8049724102020264, + "num_tokens": 620660892.0, + "step": 16268 + }, + { + "epoch": 2.069584022389009, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7730352878570557, + "num_tokens": 620698165.0, + "step": 16269 + }, + { + "epoch": 2.0697112326675997, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7903242111206055, + "num_tokens": 620734758.0, + "step": 16270 + }, + { + "epoch": 2.0698384429461902, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7774683237075806, + "num_tokens": 620770224.0, + "step": 16271 + }, + { + "epoch": 2.0699656532247808, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8126866817474365, + "num_tokens": 620805724.0, + "step": 16272 + }, + { + "epoch": 2.070092863503371, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7917195558547974, + "num_tokens": 620847622.0, + "step": 16273 + }, + { + "epoch": 2.0702200737819614, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.791087806224823, + "num_tokens": 620879042.0, + "step": 16274 + }, + { + "epoch": 2.070347284060552, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8042582273483276, + "num_tokens": 620922685.0, + "step": 16275 + }, + { + "epoch": 2.0704744943391424, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7934317588806152, + "num_tokens": 620954854.0, + "step": 16276 + }, + { + "epoch": 2.070601704617733, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.798572838306427, + "num_tokens": 620994581.0, + "step": 16277 + }, + { + "epoch": 2.0707289148963235, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.797360897064209, + "num_tokens": 621030052.0, + "step": 16278 + }, + { + "epoch": 2.070856125174914, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8137050867080688, + "num_tokens": 621072474.0, + "step": 16279 + }, + { + "epoch": 2.0709833354535045, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.8010554313659668, + "num_tokens": 621104365.0, + "step": 16280 + }, + { + "epoch": 2.071110545732095, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8064212799072266, + "num_tokens": 621140497.0, + "step": 16281 + }, + { + "epoch": 2.0712377560106856, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8023586273193359, + "num_tokens": 621177431.0, + "step": 16282 + }, + { + "epoch": 2.071364966289276, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8069736957550049, + "num_tokens": 621212874.0, + "step": 16283 + }, + { + "epoch": 2.0714921765678667, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8079428672790527, + "num_tokens": 621249927.0, + "step": 16284 + }, + { + "epoch": 2.071619386846457, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7692104578018188, + "num_tokens": 621280794.0, + "step": 16285 + }, + { + "epoch": 2.0717465971250477, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7840854525566101, + "num_tokens": 621321167.0, + "step": 16286 + }, + { + "epoch": 2.0718738074036382, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.813056468963623, + "num_tokens": 621357962.0, + "step": 16287 + }, + { + "epoch": 2.0720010176822288, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.8224296569824219, + "num_tokens": 621398933.0, + "step": 16288 + }, + { + "epoch": 2.0721282279608193, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8210120797157288, + "num_tokens": 621442334.0, + "step": 16289 + }, + { + "epoch": 2.07225543823941, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8039665222167969, + "num_tokens": 621478706.0, + "step": 16290 + }, + { + "epoch": 2.0723826485180004, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8080828189849854, + "num_tokens": 621513911.0, + "step": 16291 + }, + { + "epoch": 2.072509858796591, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.794620156288147, + "num_tokens": 621554080.0, + "step": 16292 + }, + { + "epoch": 2.0726370690751814, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8055710792541504, + "num_tokens": 621592059.0, + "step": 16293 + }, + { + "epoch": 2.072764279353772, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8177152276039124, + "num_tokens": 621629626.0, + "step": 16294 + }, + { + "epoch": 2.0728914896323625, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8101719617843628, + "num_tokens": 621671341.0, + "step": 16295 + }, + { + "epoch": 2.073018699910953, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.5913, + "mean_token_accuracy": 0.8205474615097046, + "num_tokens": 621708756.0, + "step": 16296 + }, + { + "epoch": 2.0731459101895435, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.807266116142273, + "num_tokens": 621748009.0, + "step": 16297 + }, + { + "epoch": 2.0732731204681336, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8151928186416626, + "num_tokens": 621788695.0, + "step": 16298 + }, + { + "epoch": 2.073400330746724, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7979553937911987, + "num_tokens": 621828099.0, + "step": 16299 + }, + { + "epoch": 2.0735275410253147, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7848889231681824, + "num_tokens": 621867033.0, + "step": 16300 + }, + { + "epoch": 2.073654751303905, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.787792980670929, + "num_tokens": 621907435.0, + "step": 16301 + }, + { + "epoch": 2.0737819615824957, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7873377203941345, + "num_tokens": 621948995.0, + "step": 16302 + }, + { + "epoch": 2.0739091718610863, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.817785382270813, + "num_tokens": 621991109.0, + "step": 16303 + }, + { + "epoch": 2.074036382139677, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8092712163925171, + "num_tokens": 622026585.0, + "step": 16304 + }, + { + "epoch": 2.0741635924182673, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8125839233398438, + "num_tokens": 622065208.0, + "step": 16305 + }, + { + "epoch": 2.074290802696858, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7948399186134338, + "num_tokens": 622104838.0, + "step": 16306 + }, + { + "epoch": 2.0744180129754484, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7936031222343445, + "num_tokens": 622136879.0, + "step": 16307 + }, + { + "epoch": 2.074545223254039, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8176329135894775, + "num_tokens": 622175938.0, + "step": 16308 + }, + { + "epoch": 2.0746724335326294, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8121556639671326, + "num_tokens": 622216663.0, + "step": 16309 + }, + { + "epoch": 2.07479964381122, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.805931031703949, + "num_tokens": 622250450.0, + "step": 16310 + }, + { + "epoch": 2.0749268540898105, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7962284088134766, + "num_tokens": 622291919.0, + "step": 16311 + }, + { + "epoch": 2.075054064368401, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8045124411582947, + "num_tokens": 622325892.0, + "step": 16312 + }, + { + "epoch": 2.0751812746469915, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7988383769989014, + "num_tokens": 622356966.0, + "step": 16313 + }, + { + "epoch": 2.075308484925582, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7991360425949097, + "num_tokens": 622389931.0, + "step": 16314 + }, + { + "epoch": 2.0754356952041726, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7997043132781982, + "num_tokens": 622427817.0, + "step": 16315 + }, + { + "epoch": 2.075562905482763, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7933043241500854, + "num_tokens": 622470353.0, + "step": 16316 + }, + { + "epoch": 2.0756901157613536, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.5832, + "mean_token_accuracy": 0.8203136920928955, + "num_tokens": 622507577.0, + "step": 16317 + }, + { + "epoch": 2.075817326039944, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7944720983505249, + "num_tokens": 622544746.0, + "step": 16318 + }, + { + "epoch": 2.0759445363185347, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7976277470588684, + "num_tokens": 622583688.0, + "step": 16319 + }, + { + "epoch": 2.0760717465971252, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8040980100631714, + "num_tokens": 622619911.0, + "step": 16320 + }, + { + "epoch": 2.0761989568757158, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.815161943435669, + "num_tokens": 622665732.0, + "step": 16321 + }, + { + "epoch": 2.0763261671543063, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8133618831634521, + "num_tokens": 622705746.0, + "step": 16322 + }, + { + "epoch": 2.0764533774328964, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8022103905677795, + "num_tokens": 622745481.0, + "step": 16323 + }, + { + "epoch": 2.076580587711487, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8050973415374756, + "num_tokens": 622786910.0, + "step": 16324 + }, + { + "epoch": 2.0767077979900774, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7953869104385376, + "num_tokens": 622826976.0, + "step": 16325 + }, + { + "epoch": 2.076835008268668, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7997899055480957, + "num_tokens": 622858945.0, + "step": 16326 + }, + { + "epoch": 2.0769622185472585, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8137563467025757, + "num_tokens": 622893229.0, + "step": 16327 + }, + { + "epoch": 2.077089428825849, + "grad_norm": 10.9375, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8155319094657898, + "num_tokens": 622933701.0, + "step": 16328 + }, + { + "epoch": 2.0772166391044395, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7911648750305176, + "num_tokens": 622971226.0, + "step": 16329 + }, + { + "epoch": 2.07734384938303, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7953235507011414, + "num_tokens": 623009232.0, + "step": 16330 + }, + { + "epoch": 2.0774710596616206, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7910860776901245, + "num_tokens": 623048005.0, + "step": 16331 + }, + { + "epoch": 2.077598269940211, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8033715486526489, + "num_tokens": 623088474.0, + "step": 16332 + }, + { + "epoch": 2.0777254802188017, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8002911806106567, + "num_tokens": 623123700.0, + "step": 16333 + }, + { + "epoch": 2.077852690497392, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7999776601791382, + "num_tokens": 623161406.0, + "step": 16334 + }, + { + "epoch": 2.0779799007759827, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7905135154724121, + "num_tokens": 623199770.0, + "step": 16335 + }, + { + "epoch": 2.0781071110545732, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8094141483306885, + "num_tokens": 623234635.0, + "step": 16336 + }, + { + "epoch": 2.0782343213331638, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8036680221557617, + "num_tokens": 623273695.0, + "step": 16337 + }, + { + "epoch": 2.0783615316117543, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7889947295188904, + "num_tokens": 623315278.0, + "step": 16338 + }, + { + "epoch": 2.078488741890345, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8077236413955688, + "num_tokens": 623351753.0, + "step": 16339 + }, + { + "epoch": 2.0786159521689354, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7801055312156677, + "num_tokens": 623387812.0, + "step": 16340 + }, + { + "epoch": 2.078743162447526, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8048278093338013, + "num_tokens": 623426784.0, + "step": 16341 + }, + { + "epoch": 2.0788703727261164, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7788031101226807, + "num_tokens": 623463362.0, + "step": 16342 + }, + { + "epoch": 2.078997583004707, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7950422763824463, + "num_tokens": 623500108.0, + "step": 16343 + }, + { + "epoch": 2.0791247932832975, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8019715547561646, + "num_tokens": 623537159.0, + "step": 16344 + }, + { + "epoch": 2.079252003561888, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.8178336024284363, + "num_tokens": 623576256.0, + "step": 16345 + }, + { + "epoch": 2.0793792138404785, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7952648997306824, + "num_tokens": 623612454.0, + "step": 16346 + }, + { + "epoch": 2.0795064241190686, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.8164188861846924, + "num_tokens": 623654392.0, + "step": 16347 + }, + { + "epoch": 2.079633634397659, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7819205522537231, + "num_tokens": 623697350.0, + "step": 16348 + }, + { + "epoch": 2.0797608446762497, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8051272034645081, + "num_tokens": 623736272.0, + "step": 16349 + }, + { + "epoch": 2.07988805495484, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7935124635696411, + "num_tokens": 623781319.0, + "step": 16350 + }, + { + "epoch": 2.0800152652334307, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.802620530128479, + "num_tokens": 623821944.0, + "step": 16351 + }, + { + "epoch": 2.0801424755120212, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.793369710445404, + "num_tokens": 623858399.0, + "step": 16352 + }, + { + "epoch": 2.0802696857906118, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8057411909103394, + "num_tokens": 623897839.0, + "step": 16353 + }, + { + "epoch": 2.0803968960692023, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7797561883926392, + "num_tokens": 623933127.0, + "step": 16354 + }, + { + "epoch": 2.080524106347793, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7994852066040039, + "num_tokens": 623972140.0, + "step": 16355 + }, + { + "epoch": 2.0806513166263834, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.793948233127594, + "num_tokens": 624008875.0, + "step": 16356 + }, + { + "epoch": 2.080778526904974, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7953546047210693, + "num_tokens": 624045162.0, + "step": 16357 + }, + { + "epoch": 2.0809057371835644, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8161370754241943, + "num_tokens": 624085275.0, + "step": 16358 + }, + { + "epoch": 2.081032947462155, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7945341467857361, + "num_tokens": 624127479.0, + "step": 16359 + }, + { + "epoch": 2.0811601577407455, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.8016279935836792, + "num_tokens": 624165231.0, + "step": 16360 + }, + { + "epoch": 2.081287368019336, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.788682222366333, + "num_tokens": 624205109.0, + "step": 16361 + }, + { + "epoch": 2.0814145782979265, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7945877313613892, + "num_tokens": 624241849.0, + "step": 16362 + }, + { + "epoch": 2.081541788576517, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7956304550170898, + "num_tokens": 624274174.0, + "step": 16363 + }, + { + "epoch": 2.0816689988551076, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.797039270401001, + "num_tokens": 624311526.0, + "step": 16364 + }, + { + "epoch": 2.081796209133698, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7876104116439819, + "num_tokens": 624346657.0, + "step": 16365 + }, + { + "epoch": 2.0819234194122886, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7823829650878906, + "num_tokens": 624386563.0, + "step": 16366 + }, + { + "epoch": 2.082050629690879, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8112325668334961, + "num_tokens": 624427299.0, + "step": 16367 + }, + { + "epoch": 2.0821778399694697, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7800818681716919, + "num_tokens": 624461253.0, + "step": 16368 + }, + { + "epoch": 2.0823050502480602, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8101668357849121, + "num_tokens": 624498699.0, + "step": 16369 + }, + { + "epoch": 2.0824322605266508, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7996526956558228, + "num_tokens": 624534655.0, + "step": 16370 + }, + { + "epoch": 2.082559470805241, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8029139041900635, + "num_tokens": 624575342.0, + "step": 16371 + }, + { + "epoch": 2.0826866810838314, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8060935735702515, + "num_tokens": 624608493.0, + "step": 16372 + }, + { + "epoch": 2.082813891362422, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7803608179092407, + "num_tokens": 624655777.0, + "step": 16373 + }, + { + "epoch": 2.0829411016410124, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8107815980911255, + "num_tokens": 624699349.0, + "step": 16374 + }, + { + "epoch": 2.083068311919603, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8110185265541077, + "num_tokens": 624730146.0, + "step": 16375 + }, + { + "epoch": 2.0831955221981935, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7909478545188904, + "num_tokens": 624773186.0, + "step": 16376 + }, + { + "epoch": 2.083322732476784, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8026630878448486, + "num_tokens": 624810256.0, + "step": 16377 + }, + { + "epoch": 2.0834499427553745, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.785835862159729, + "num_tokens": 624848166.0, + "step": 16378 + }, + { + "epoch": 2.083577153033965, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7936612963676453, + "num_tokens": 624887601.0, + "step": 16379 + }, + { + "epoch": 2.0837043633125556, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7987879514694214, + "num_tokens": 624921160.0, + "step": 16380 + }, + { + "epoch": 2.083831573591146, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.788650631904602, + "num_tokens": 624959053.0, + "step": 16381 + }, + { + "epoch": 2.0839587838697367, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8178091645240784, + "num_tokens": 624995272.0, + "step": 16382 + }, + { + "epoch": 2.084085994148327, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.818144679069519, + "num_tokens": 625033388.0, + "step": 16383 + }, + { + "epoch": 2.0842132044269177, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7922064661979675, + "num_tokens": 625071103.0, + "step": 16384 + }, + { + "epoch": 2.0843404147055082, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.811004102230072, + "num_tokens": 625109562.0, + "step": 16385 + }, + { + "epoch": 2.0844676249840988, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8053988218307495, + "num_tokens": 625147438.0, + "step": 16386 + }, + { + "epoch": 2.0845948352626893, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8112211227416992, + "num_tokens": 625184293.0, + "step": 16387 + }, + { + "epoch": 2.08472204554128, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8020054697990417, + "num_tokens": 625218201.0, + "step": 16388 + }, + { + "epoch": 2.0848492558198704, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7960732579231262, + "num_tokens": 625252072.0, + "step": 16389 + }, + { + "epoch": 2.084976466098461, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8133437037467957, + "num_tokens": 625286990.0, + "step": 16390 + }, + { + "epoch": 2.0851036763770514, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.815893828868866, + "num_tokens": 625326579.0, + "step": 16391 + }, + { + "epoch": 2.085230886655642, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8109564185142517, + "num_tokens": 625367317.0, + "step": 16392 + }, + { + "epoch": 2.0853580969342325, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7940001487731934, + "num_tokens": 625401047.0, + "step": 16393 + }, + { + "epoch": 2.085485307212823, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7877787351608276, + "num_tokens": 625441878.0, + "step": 16394 + }, + { + "epoch": 2.0856125174914135, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7995100021362305, + "num_tokens": 625477007.0, + "step": 16395 + }, + { + "epoch": 2.0857397277700036, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8106090426445007, + "num_tokens": 625519908.0, + "step": 16396 + }, + { + "epoch": 2.085866938048594, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.801252007484436, + "num_tokens": 625554946.0, + "step": 16397 + }, + { + "epoch": 2.0859941483271847, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8092226982116699, + "num_tokens": 625592121.0, + "step": 16398 + }, + { + "epoch": 2.086121358605775, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.798622727394104, + "num_tokens": 625630596.0, + "step": 16399 + }, + { + "epoch": 2.0862485688843657, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7794592380523682, + "num_tokens": 625669535.0, + "step": 16400 + }, + { + "epoch": 2.0863757791629562, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8016653060913086, + "num_tokens": 625710072.0, + "step": 16401 + }, + { + "epoch": 2.0865029894415468, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8087737560272217, + "num_tokens": 625746328.0, + "step": 16402 + }, + { + "epoch": 2.0866301997201373, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8042909502983093, + "num_tokens": 625787108.0, + "step": 16403 + }, + { + "epoch": 2.086757409998728, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8017676472663879, + "num_tokens": 625827887.0, + "step": 16404 + }, + { + "epoch": 2.0868846202773184, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8031439781188965, + "num_tokens": 625863475.0, + "step": 16405 + }, + { + "epoch": 2.087011830555909, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.8174399733543396, + "num_tokens": 625902369.0, + "step": 16406 + }, + { + "epoch": 2.0871390408344994, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7732885479927063, + "num_tokens": 625941054.0, + "step": 16407 + }, + { + "epoch": 2.08726625111309, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7788187265396118, + "num_tokens": 625981113.0, + "step": 16408 + }, + { + "epoch": 2.0873934613916805, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7915700674057007, + "num_tokens": 626014758.0, + "step": 16409 + }, + { + "epoch": 2.087520671670271, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7840549349784851, + "num_tokens": 626054442.0, + "step": 16410 + }, + { + "epoch": 2.0876478819488615, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7771390676498413, + "num_tokens": 626093993.0, + "step": 16411 + }, + { + "epoch": 2.087775092227452, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7794390320777893, + "num_tokens": 626131242.0, + "step": 16412 + }, + { + "epoch": 2.0879023025060426, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7781891226768494, + "num_tokens": 626162137.0, + "step": 16413 + }, + { + "epoch": 2.088029512784633, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8010905981063843, + "num_tokens": 626201031.0, + "step": 16414 + }, + { + "epoch": 2.0881567230632236, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8144113421440125, + "num_tokens": 626243647.0, + "step": 16415 + }, + { + "epoch": 2.088283933341814, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8039913177490234, + "num_tokens": 626281346.0, + "step": 16416 + }, + { + "epoch": 2.0884111436204047, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8073571920394897, + "num_tokens": 626315964.0, + "step": 16417 + }, + { + "epoch": 2.0885383538989952, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7951294183731079, + "num_tokens": 626367920.0, + "step": 16418 + }, + { + "epoch": 2.0886655641775858, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.798744797706604, + "num_tokens": 626412123.0, + "step": 16419 + }, + { + "epoch": 2.0887927744561763, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7668289542198181, + "num_tokens": 626451560.0, + "step": 16420 + }, + { + "epoch": 2.0889199847347664, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7975034713745117, + "num_tokens": 626489361.0, + "step": 16421 + }, + { + "epoch": 2.089047195013357, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8176577091217041, + "num_tokens": 626527150.0, + "step": 16422 + }, + { + "epoch": 2.0891744052919474, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7933464050292969, + "num_tokens": 626568110.0, + "step": 16423 + }, + { + "epoch": 2.089301615570538, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.8015612363815308, + "num_tokens": 626607215.0, + "step": 16424 + }, + { + "epoch": 2.0894288258491285, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8066413402557373, + "num_tokens": 626644424.0, + "step": 16425 + }, + { + "epoch": 2.089556036127719, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.779991090297699, + "num_tokens": 626689509.0, + "step": 16426 + }, + { + "epoch": 2.0896832464063095, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7918729186058044, + "num_tokens": 626733866.0, + "step": 16427 + }, + { + "epoch": 2.0898104566849, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.807651162147522, + "num_tokens": 626767465.0, + "step": 16428 + }, + { + "epoch": 2.0899376669634906, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7808746099472046, + "num_tokens": 626806049.0, + "step": 16429 + }, + { + "epoch": 2.090064877242081, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7925214767456055, + "num_tokens": 626848917.0, + "step": 16430 + }, + { + "epoch": 2.0901920875206716, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7851563692092896, + "num_tokens": 626884442.0, + "step": 16431 + }, + { + "epoch": 2.090319297799262, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7925418615341187, + "num_tokens": 626922517.0, + "step": 16432 + }, + { + "epoch": 2.0904465080778527, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.799453616142273, + "num_tokens": 626962880.0, + "step": 16433 + }, + { + "epoch": 2.0905737183564432, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8102673292160034, + "num_tokens": 627008115.0, + "step": 16434 + }, + { + "epoch": 2.0907009286350338, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8125115633010864, + "num_tokens": 627043935.0, + "step": 16435 + }, + { + "epoch": 2.0908281389136243, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7989130616188049, + "num_tokens": 627078495.0, + "step": 16436 + }, + { + "epoch": 2.090955349192215, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.800888180732727, + "num_tokens": 627116652.0, + "step": 16437 + }, + { + "epoch": 2.0910825594708053, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8035447597503662, + "num_tokens": 627152326.0, + "step": 16438 + }, + { + "epoch": 2.091209769749396, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8064264059066772, + "num_tokens": 627189502.0, + "step": 16439 + }, + { + "epoch": 2.0913369800279864, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8088297843933105, + "num_tokens": 627225438.0, + "step": 16440 + }, + { + "epoch": 2.091464190306577, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.799788236618042, + "num_tokens": 627258154.0, + "step": 16441 + }, + { + "epoch": 2.0915914005851675, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7925572991371155, + "num_tokens": 627301804.0, + "step": 16442 + }, + { + "epoch": 2.091718610863758, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8085342645645142, + "num_tokens": 627344301.0, + "step": 16443 + }, + { + "epoch": 2.0918458211423485, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8079965114593506, + "num_tokens": 627381068.0, + "step": 16444 + }, + { + "epoch": 2.0919730314209386, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.793620765209198, + "num_tokens": 627423410.0, + "step": 16445 + }, + { + "epoch": 2.092100241699529, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8074711561203003, + "num_tokens": 627456593.0, + "step": 16446 + }, + { + "epoch": 2.0922274519781197, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8024836778640747, + "num_tokens": 627493742.0, + "step": 16447 + }, + { + "epoch": 2.09235466225671, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.80260169506073, + "num_tokens": 627529809.0, + "step": 16448 + }, + { + "epoch": 2.0924818725353007, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8053590059280396, + "num_tokens": 627570693.0, + "step": 16449 + }, + { + "epoch": 2.0926090828138912, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8100526928901672, + "num_tokens": 627607608.0, + "step": 16450 + }, + { + "epoch": 2.0927362930924818, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7977108955383301, + "num_tokens": 627642102.0, + "step": 16451 + }, + { + "epoch": 2.0928635033710723, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7931154370307922, + "num_tokens": 627678838.0, + "step": 16452 + }, + { + "epoch": 2.092990713649663, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8017739057540894, + "num_tokens": 627714952.0, + "step": 16453 + }, + { + "epoch": 2.0931179239282534, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7919877767562866, + "num_tokens": 627754710.0, + "step": 16454 + }, + { + "epoch": 2.093245134206844, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8054946660995483, + "num_tokens": 627791504.0, + "step": 16455 + }, + { + "epoch": 2.0933723444854344, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8137519359588623, + "num_tokens": 627829683.0, + "step": 16456 + }, + { + "epoch": 2.093499554764025, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.805763304233551, + "num_tokens": 627869826.0, + "step": 16457 + }, + { + "epoch": 2.0936267650426155, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7899627685546875, + "num_tokens": 627904165.0, + "step": 16458 + }, + { + "epoch": 2.093753975321206, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8044456243515015, + "num_tokens": 627936559.0, + "step": 16459 + }, + { + "epoch": 2.0938811855997965, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8009593486785889, + "num_tokens": 627971742.0, + "step": 16460 + }, + { + "epoch": 2.094008395878387, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8113672733306885, + "num_tokens": 628013104.0, + "step": 16461 + }, + { + "epoch": 2.0941356061569776, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7999354600906372, + "num_tokens": 628049378.0, + "step": 16462 + }, + { + "epoch": 2.094262816435568, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8198020458221436, + "num_tokens": 628087213.0, + "step": 16463 + }, + { + "epoch": 2.0943900267141586, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8045483231544495, + "num_tokens": 628122060.0, + "step": 16464 + }, + { + "epoch": 2.094517236992749, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8066591024398804, + "num_tokens": 628166554.0, + "step": 16465 + }, + { + "epoch": 2.0946444472713397, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7964156866073608, + "num_tokens": 628199953.0, + "step": 16466 + }, + { + "epoch": 2.09477165754993, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.799945592880249, + "num_tokens": 628237044.0, + "step": 16467 + }, + { + "epoch": 2.0948988678285207, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8040016889572144, + "num_tokens": 628275773.0, + "step": 16468 + }, + { + "epoch": 2.095026078107111, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7842967510223389, + "num_tokens": 628314549.0, + "step": 16469 + }, + { + "epoch": 2.0951532883857014, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7916672229766846, + "num_tokens": 628349417.0, + "step": 16470 + }, + { + "epoch": 2.095280498664292, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8004684448242188, + "num_tokens": 628384435.0, + "step": 16471 + }, + { + "epoch": 2.0954077089428824, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7804642915725708, + "num_tokens": 628424124.0, + "step": 16472 + }, + { + "epoch": 2.095534919221473, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8016341328620911, + "num_tokens": 628462412.0, + "step": 16473 + }, + { + "epoch": 2.0956621295000635, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7806631326675415, + "num_tokens": 628499066.0, + "step": 16474 + }, + { + "epoch": 2.095789339778654, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.810181736946106, + "num_tokens": 628532379.0, + "step": 16475 + }, + { + "epoch": 2.0959165500572445, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8105384707450867, + "num_tokens": 628570497.0, + "step": 16476 + }, + { + "epoch": 2.096043760335835, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.801699697971344, + "num_tokens": 628608298.0, + "step": 16477 + }, + { + "epoch": 2.0961709706144256, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5912, + "mean_token_accuracy": 0.81709885597229, + "num_tokens": 628641217.0, + "step": 16478 + }, + { + "epoch": 2.096298180893016, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8058419823646545, + "num_tokens": 628682480.0, + "step": 16479 + }, + { + "epoch": 2.0964253911716066, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7907086610794067, + "num_tokens": 628718359.0, + "step": 16480 + }, + { + "epoch": 2.096552601450197, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8075476884841919, + "num_tokens": 628752214.0, + "step": 16481 + }, + { + "epoch": 2.0966798117287877, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8187333345413208, + "num_tokens": 628787882.0, + "step": 16482 + }, + { + "epoch": 2.0968070220073782, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7933107614517212, + "num_tokens": 628824910.0, + "step": 16483 + }, + { + "epoch": 2.0969342322859688, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8170629739761353, + "num_tokens": 628860657.0, + "step": 16484 + }, + { + "epoch": 2.0970614425645593, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7805536985397339, + "num_tokens": 628900619.0, + "step": 16485 + }, + { + "epoch": 2.09718865284315, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8172211050987244, + "num_tokens": 628944304.0, + "step": 16486 + }, + { + "epoch": 2.0973158631217403, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8041227459907532, + "num_tokens": 628980490.0, + "step": 16487 + }, + { + "epoch": 2.097443073400331, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8025438785552979, + "num_tokens": 629011924.0, + "step": 16488 + }, + { + "epoch": 2.0975702836789214, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7721751928329468, + "num_tokens": 629049514.0, + "step": 16489 + }, + { + "epoch": 2.097697493957512, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8140103816986084, + "num_tokens": 629085374.0, + "step": 16490 + }, + { + "epoch": 2.0978247042361025, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8010010719299316, + "num_tokens": 629126307.0, + "step": 16491 + }, + { + "epoch": 2.097951914514693, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8021289706230164, + "num_tokens": 629159693.0, + "step": 16492 + }, + { + "epoch": 2.0980791247932835, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8055410981178284, + "num_tokens": 629198544.0, + "step": 16493 + }, + { + "epoch": 2.0982063350718736, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7878777980804443, + "num_tokens": 629240956.0, + "step": 16494 + }, + { + "epoch": 2.098333545350464, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8011056184768677, + "num_tokens": 629276196.0, + "step": 16495 + }, + { + "epoch": 2.0984607556290547, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.782943069934845, + "num_tokens": 629311398.0, + "step": 16496 + }, + { + "epoch": 2.098587965907645, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7937089800834656, + "num_tokens": 629349488.0, + "step": 16497 + }, + { + "epoch": 2.0987151761862357, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8177145719528198, + "num_tokens": 629388023.0, + "step": 16498 + }, + { + "epoch": 2.0988423864648262, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8084663152694702, + "num_tokens": 629432140.0, + "step": 16499 + }, + { + "epoch": 2.0989695967434168, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7901418209075928, + "num_tokens": 629480690.0, + "step": 16500 + }, + { + "epoch": 2.0990968070220073, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8194486498832703, + "num_tokens": 629522500.0, + "step": 16501 + }, + { + "epoch": 2.099224017300598, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8065751791000366, + "num_tokens": 629563450.0, + "step": 16502 + }, + { + "epoch": 2.0993512275791884, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8196457028388977, + "num_tokens": 629603753.0, + "step": 16503 + }, + { + "epoch": 2.099478437857779, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7997764348983765, + "num_tokens": 629637828.0, + "step": 16504 + }, + { + "epoch": 2.0996056481363694, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8094958066940308, + "num_tokens": 629674636.0, + "step": 16505 + }, + { + "epoch": 2.09973285841496, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7943429350852966, + "num_tokens": 629717046.0, + "step": 16506 + }, + { + "epoch": 2.0998600686935505, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.812394380569458, + "num_tokens": 629750331.0, + "step": 16507 + }, + { + "epoch": 2.099987278972141, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7846580147743225, + "num_tokens": 629790095.0, + "step": 16508 + }, + { + "epoch": 2.1001144892507315, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8113929033279419, + "num_tokens": 629823035.0, + "step": 16509 + }, + { + "epoch": 2.100241699529322, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7935764789581299, + "num_tokens": 629857096.0, + "step": 16510 + }, + { + "epoch": 2.1003689098079126, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7973273396492004, + "num_tokens": 629893448.0, + "step": 16511 + }, + { + "epoch": 2.100496120086503, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8037332892417908, + "num_tokens": 629931301.0, + "step": 16512 + }, + { + "epoch": 2.1006233303650936, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8071269392967224, + "num_tokens": 629963481.0, + "step": 16513 + }, + { + "epoch": 2.100750540643684, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7983913421630859, + "num_tokens": 630002739.0, + "step": 16514 + }, + { + "epoch": 2.1008777509222747, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7916693687438965, + "num_tokens": 630041206.0, + "step": 16515 + }, + { + "epoch": 2.101004961200865, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7877424955368042, + "num_tokens": 630083811.0, + "step": 16516 + }, + { + "epoch": 2.1011321714794557, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8041156530380249, + "num_tokens": 630119823.0, + "step": 16517 + }, + { + "epoch": 2.1012593817580463, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8046354055404663, + "num_tokens": 630163807.0, + "step": 16518 + }, + { + "epoch": 2.1013865920366364, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7974625825881958, + "num_tokens": 630203038.0, + "step": 16519 + }, + { + "epoch": 2.101513802315227, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8037419319152832, + "num_tokens": 630238505.0, + "step": 16520 + }, + { + "epoch": 2.1016410125938174, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8026212453842163, + "num_tokens": 630275279.0, + "step": 16521 + }, + { + "epoch": 2.101768222872408, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8005542159080505, + "num_tokens": 630317288.0, + "step": 16522 + }, + { + "epoch": 2.1018954331509985, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8017497062683105, + "num_tokens": 630352786.0, + "step": 16523 + }, + { + "epoch": 2.102022643429589, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8034575581550598, + "num_tokens": 630391138.0, + "step": 16524 + }, + { + "epoch": 2.1021498537081795, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7862866520881653, + "num_tokens": 630426567.0, + "step": 16525 + }, + { + "epoch": 2.10227706398677, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7886332869529724, + "num_tokens": 630467308.0, + "step": 16526 + }, + { + "epoch": 2.1024042742653606, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8068585991859436, + "num_tokens": 630503333.0, + "step": 16527 + }, + { + "epoch": 2.102531484543951, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7945230007171631, + "num_tokens": 630536838.0, + "step": 16528 + }, + { + "epoch": 2.1026586948225416, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8175705671310425, + "num_tokens": 630570833.0, + "step": 16529 + }, + { + "epoch": 2.102785905101132, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8136868476867676, + "num_tokens": 630615043.0, + "step": 16530 + }, + { + "epoch": 2.1029131153797227, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.8232991695404053, + "num_tokens": 630651368.0, + "step": 16531 + }, + { + "epoch": 2.1030403256583132, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.792547881603241, + "num_tokens": 630690341.0, + "step": 16532 + }, + { + "epoch": 2.1031675359369038, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7977901697158813, + "num_tokens": 630727168.0, + "step": 16533 + }, + { + "epoch": 2.1032947462154943, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8027225732803345, + "num_tokens": 630766737.0, + "step": 16534 + }, + { + "epoch": 2.103421956494085, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7997779846191406, + "num_tokens": 630803721.0, + "step": 16535 + }, + { + "epoch": 2.1035491667726753, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8064148426055908, + "num_tokens": 630843697.0, + "step": 16536 + }, + { + "epoch": 2.103676377051266, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7997579574584961, + "num_tokens": 630874325.0, + "step": 16537 + }, + { + "epoch": 2.1038035873298564, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7927085161209106, + "num_tokens": 630911456.0, + "step": 16538 + }, + { + "epoch": 2.103930797608447, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8047631978988647, + "num_tokens": 630954418.0, + "step": 16539 + }, + { + "epoch": 2.1040580078870375, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7967527508735657, + "num_tokens": 630991826.0, + "step": 16540 + }, + { + "epoch": 2.104185218165628, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8038410544395447, + "num_tokens": 631025831.0, + "step": 16541 + }, + { + "epoch": 2.1043124284442185, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7868627309799194, + "num_tokens": 631062896.0, + "step": 16542 + }, + { + "epoch": 2.1044396387228086, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8126732707023621, + "num_tokens": 631099291.0, + "step": 16543 + }, + { + "epoch": 2.104566849001399, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7893335819244385, + "num_tokens": 631141570.0, + "step": 16544 + }, + { + "epoch": 2.1046940592799896, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7950887680053711, + "num_tokens": 631178882.0, + "step": 16545 + }, + { + "epoch": 2.10482126955858, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7999246120452881, + "num_tokens": 631218059.0, + "step": 16546 + }, + { + "epoch": 2.1049484798371707, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8020014762878418, + "num_tokens": 631252827.0, + "step": 16547 + }, + { + "epoch": 2.1050756901157612, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.808906078338623, + "num_tokens": 631294370.0, + "step": 16548 + }, + { + "epoch": 2.1052029003943518, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7975846529006958, + "num_tokens": 631331091.0, + "step": 16549 + }, + { + "epoch": 2.1053301106729423, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7967277765274048, + "num_tokens": 631367285.0, + "step": 16550 + }, + { + "epoch": 2.105457320951533, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7807630300521851, + "num_tokens": 631402471.0, + "step": 16551 + }, + { + "epoch": 2.1055845312301233, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7957099676132202, + "num_tokens": 631439706.0, + "step": 16552 + }, + { + "epoch": 2.105711741508714, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8056735992431641, + "num_tokens": 631471437.0, + "step": 16553 + }, + { + "epoch": 2.1058389517873044, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7982460260391235, + "num_tokens": 631515028.0, + "step": 16554 + }, + { + "epoch": 2.105966162065895, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7964898347854614, + "num_tokens": 631552946.0, + "step": 16555 + }, + { + "epoch": 2.1060933723444855, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8097286224365234, + "num_tokens": 631596045.0, + "step": 16556 + }, + { + "epoch": 2.106220582623076, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8085057735443115, + "num_tokens": 631630233.0, + "step": 16557 + }, + { + "epoch": 2.1063477929016665, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8050479888916016, + "num_tokens": 631666665.0, + "step": 16558 + }, + { + "epoch": 2.106475003180257, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8012902140617371, + "num_tokens": 631709536.0, + "step": 16559 + }, + { + "epoch": 2.1066022134588476, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8145105838775635, + "num_tokens": 631744333.0, + "step": 16560 + }, + { + "epoch": 2.106729423737438, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7904151678085327, + "num_tokens": 631783522.0, + "step": 16561 + }, + { + "epoch": 2.1068566340160286, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7833118438720703, + "num_tokens": 631822579.0, + "step": 16562 + }, + { + "epoch": 2.106983844294619, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8088516592979431, + "num_tokens": 631862642.0, + "step": 16563 + }, + { + "epoch": 2.1071110545732097, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8148857355117798, + "num_tokens": 631903161.0, + "step": 16564 + }, + { + "epoch": 2.1072382648518, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.796573281288147, + "num_tokens": 631938149.0, + "step": 16565 + }, + { + "epoch": 2.1073654751303907, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8015142679214478, + "num_tokens": 631974717.0, + "step": 16566 + }, + { + "epoch": 2.107492685408981, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8069278001785278, + "num_tokens": 632011147.0, + "step": 16567 + }, + { + "epoch": 2.1076198956875714, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8101325035095215, + "num_tokens": 632051915.0, + "step": 16568 + }, + { + "epoch": 2.107747105966162, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7867180109024048, + "num_tokens": 632083047.0, + "step": 16569 + }, + { + "epoch": 2.1078743162447524, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8110784292221069, + "num_tokens": 632119250.0, + "step": 16570 + }, + { + "epoch": 2.108001526523343, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8116569519042969, + "num_tokens": 632152696.0, + "step": 16571 + }, + { + "epoch": 2.1081287368019335, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.598, + "mean_token_accuracy": 0.8218812346458435, + "num_tokens": 632188167.0, + "step": 16572 + }, + { + "epoch": 2.108255947080524, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8030426502227783, + "num_tokens": 632229567.0, + "step": 16573 + }, + { + "epoch": 2.1083831573591145, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.8130155205726624, + "num_tokens": 632266870.0, + "step": 16574 + }, + { + "epoch": 2.108510367637705, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8027703762054443, + "num_tokens": 632303844.0, + "step": 16575 + }, + { + "epoch": 2.1086375779162956, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8008110523223877, + "num_tokens": 632339343.0, + "step": 16576 + }, + { + "epoch": 2.108764788194886, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8030099868774414, + "num_tokens": 632385516.0, + "step": 16577 + }, + { + "epoch": 2.1088919984734766, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.805317759513855, + "num_tokens": 632420664.0, + "step": 16578 + }, + { + "epoch": 2.109019208752067, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8077298998832703, + "num_tokens": 632458082.0, + "step": 16579 + }, + { + "epoch": 2.1091464190306577, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7881271243095398, + "num_tokens": 632495963.0, + "step": 16580 + }, + { + "epoch": 2.109273629309248, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8016322255134583, + "num_tokens": 632529968.0, + "step": 16581 + }, + { + "epoch": 2.1094008395878387, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.8181833028793335, + "num_tokens": 632567527.0, + "step": 16582 + }, + { + "epoch": 2.1095280498664293, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7975987195968628, + "num_tokens": 632606192.0, + "step": 16583 + }, + { + "epoch": 2.10965526014502, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8107407093048096, + "num_tokens": 632645940.0, + "step": 16584 + }, + { + "epoch": 2.1097824704236103, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.810840368270874, + "num_tokens": 632689385.0, + "step": 16585 + }, + { + "epoch": 2.109909680702201, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8057044148445129, + "num_tokens": 632720178.0, + "step": 16586 + }, + { + "epoch": 2.1100368909807914, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7967852354049683, + "num_tokens": 632763249.0, + "step": 16587 + }, + { + "epoch": 2.110164101259382, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7814263701438904, + "num_tokens": 632803293.0, + "step": 16588 + }, + { + "epoch": 2.1102913115379724, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7902989983558655, + "num_tokens": 632838731.0, + "step": 16589 + }, + { + "epoch": 2.110418521816563, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7921938896179199, + "num_tokens": 632875044.0, + "step": 16590 + }, + { + "epoch": 2.1105457320951535, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7771516442298889, + "num_tokens": 632911330.0, + "step": 16591 + }, + { + "epoch": 2.1106729423737436, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7937617301940918, + "num_tokens": 632952122.0, + "step": 16592 + }, + { + "epoch": 2.110800152652334, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7786744832992554, + "num_tokens": 632985602.0, + "step": 16593 + }, + { + "epoch": 2.1109273629309246, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8022095561027527, + "num_tokens": 633023792.0, + "step": 16594 + }, + { + "epoch": 2.111054573209515, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8037635684013367, + "num_tokens": 633065457.0, + "step": 16595 + }, + { + "epoch": 2.1111817834881057, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8031889796257019, + "num_tokens": 633101402.0, + "step": 16596 + }, + { + "epoch": 2.1113089937666962, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.795088529586792, + "num_tokens": 633147002.0, + "step": 16597 + }, + { + "epoch": 2.1114362040452868, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8025341033935547, + "num_tokens": 633184209.0, + "step": 16598 + }, + { + "epoch": 2.1115634143238773, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8068543076515198, + "num_tokens": 633224525.0, + "step": 16599 + }, + { + "epoch": 2.111690624602468, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8163636326789856, + "num_tokens": 633253171.0, + "step": 16600 + }, + { + "epoch": 2.1118178348810583, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7991268634796143, + "num_tokens": 633294326.0, + "step": 16601 + }, + { + "epoch": 2.111945045159649, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8119526505470276, + "num_tokens": 633327236.0, + "step": 16602 + }, + { + "epoch": 2.1120722554382394, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8089662790298462, + "num_tokens": 633368593.0, + "step": 16603 + }, + { + "epoch": 2.11219946571683, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7843683958053589, + "num_tokens": 633402598.0, + "step": 16604 + }, + { + "epoch": 2.1123266759954205, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7979532480239868, + "num_tokens": 633437069.0, + "step": 16605 + }, + { + "epoch": 2.112453886274011, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8055915236473083, + "num_tokens": 633479752.0, + "step": 16606 + }, + { + "epoch": 2.1125810965526015, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.798720121383667, + "num_tokens": 633516949.0, + "step": 16607 + }, + { + "epoch": 2.112708306831192, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7958559393882751, + "num_tokens": 633547511.0, + "step": 16608 + }, + { + "epoch": 2.1128355171097826, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8019452691078186, + "num_tokens": 633583940.0, + "step": 16609 + }, + { + "epoch": 2.112962727388373, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.78980553150177, + "num_tokens": 633626245.0, + "step": 16610 + }, + { + "epoch": 2.1130899376669636, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8054579496383667, + "num_tokens": 633662889.0, + "step": 16611 + }, + { + "epoch": 2.113217147945554, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.790643036365509, + "num_tokens": 633701461.0, + "step": 16612 + }, + { + "epoch": 2.1133443582241447, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7981462478637695, + "num_tokens": 633741302.0, + "step": 16613 + }, + { + "epoch": 2.113471568502735, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8058199286460876, + "num_tokens": 633780337.0, + "step": 16614 + }, + { + "epoch": 2.1135987787813257, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8083775639533997, + "num_tokens": 633814450.0, + "step": 16615 + }, + { + "epoch": 2.1137259890599163, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7931662797927856, + "num_tokens": 633856252.0, + "step": 16616 + }, + { + "epoch": 2.1138531993385064, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.785415530204773, + "num_tokens": 633895235.0, + "step": 16617 + }, + { + "epoch": 2.113980409617097, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8036237955093384, + "num_tokens": 633938948.0, + "step": 16618 + }, + { + "epoch": 2.1141076198956874, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8118950128555298, + "num_tokens": 633973393.0, + "step": 16619 + }, + { + "epoch": 2.114234830174278, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8094425201416016, + "num_tokens": 634006714.0, + "step": 16620 + }, + { + "epoch": 2.1143620404528685, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8079085946083069, + "num_tokens": 634046104.0, + "step": 16621 + }, + { + "epoch": 2.114489250731459, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7996798753738403, + "num_tokens": 634086401.0, + "step": 16622 + }, + { + "epoch": 2.1146164610100495, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.793903648853302, + "num_tokens": 634125479.0, + "step": 16623 + }, + { + "epoch": 2.11474367128864, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7819740772247314, + "num_tokens": 634163873.0, + "step": 16624 + }, + { + "epoch": 2.1148708815672306, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7993201017379761, + "num_tokens": 634200533.0, + "step": 16625 + }, + { + "epoch": 2.114998091845821, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7917670011520386, + "num_tokens": 634235891.0, + "step": 16626 + }, + { + "epoch": 2.1151253021244116, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7995945811271667, + "num_tokens": 634266748.0, + "step": 16627 + }, + { + "epoch": 2.115252512403002, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7853753566741943, + "num_tokens": 634302603.0, + "step": 16628 + }, + { + "epoch": 2.1153797226815927, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8022066354751587, + "num_tokens": 634345471.0, + "step": 16629 + }, + { + "epoch": 2.115506932960183, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.8003250360488892, + "num_tokens": 634383902.0, + "step": 16630 + }, + { + "epoch": 2.1156341432387737, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7853220701217651, + "num_tokens": 634421082.0, + "step": 16631 + }, + { + "epoch": 2.1157613535173643, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7981423139572144, + "num_tokens": 634456490.0, + "step": 16632 + }, + { + "epoch": 2.115888563795955, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7806897163391113, + "num_tokens": 634498252.0, + "step": 16633 + }, + { + "epoch": 2.1160157740745453, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8131969571113586, + "num_tokens": 634540769.0, + "step": 16634 + }, + { + "epoch": 2.116142984353136, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8109027147293091, + "num_tokens": 634573807.0, + "step": 16635 + }, + { + "epoch": 2.1162701946317264, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7950970530509949, + "num_tokens": 634609542.0, + "step": 16636 + }, + { + "epoch": 2.116397404910317, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8072024583816528, + "num_tokens": 634648641.0, + "step": 16637 + }, + { + "epoch": 2.1165246151889074, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.804462194442749, + "num_tokens": 634685531.0, + "step": 16638 + }, + { + "epoch": 2.116651825467498, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8004246950149536, + "num_tokens": 634717239.0, + "step": 16639 + }, + { + "epoch": 2.116779035746088, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7889866232872009, + "num_tokens": 634750058.0, + "step": 16640 + }, + { + "epoch": 2.1169062460246786, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7945784330368042, + "num_tokens": 634793480.0, + "step": 16641 + }, + { + "epoch": 2.117033456303269, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7829433679580688, + "num_tokens": 634829923.0, + "step": 16642 + }, + { + "epoch": 2.1171606665818596, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7972874045372009, + "num_tokens": 634868499.0, + "step": 16643 + }, + { + "epoch": 2.11728787686045, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8123387694358826, + "num_tokens": 634905922.0, + "step": 16644 + }, + { + "epoch": 2.1174150871390407, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.799159049987793, + "num_tokens": 634946190.0, + "step": 16645 + }, + { + "epoch": 2.1175422974176312, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8107671737670898, + "num_tokens": 634981840.0, + "step": 16646 + }, + { + "epoch": 2.1176695076962218, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7986592054367065, + "num_tokens": 635024755.0, + "step": 16647 + }, + { + "epoch": 2.1177967179748123, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8089699745178223, + "num_tokens": 635063076.0, + "step": 16648 + }, + { + "epoch": 2.117923928253403, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8141933083534241, + "num_tokens": 635103052.0, + "step": 16649 + }, + { + "epoch": 2.1180511385319933, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8093141317367554, + "num_tokens": 635140917.0, + "step": 16650 + }, + { + "epoch": 2.118178348810584, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8045608997344971, + "num_tokens": 635182049.0, + "step": 16651 + }, + { + "epoch": 2.1183055590891744, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7911747694015503, + "num_tokens": 635217613.0, + "step": 16652 + }, + { + "epoch": 2.118432769367765, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8102744221687317, + "num_tokens": 635252716.0, + "step": 16653 + }, + { + "epoch": 2.1185599796463555, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.801112174987793, + "num_tokens": 635288919.0, + "step": 16654 + }, + { + "epoch": 2.118687189924946, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7948126196861267, + "num_tokens": 635329781.0, + "step": 16655 + }, + { + "epoch": 2.1188144002035365, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8047329187393188, + "num_tokens": 635363844.0, + "step": 16656 + }, + { + "epoch": 2.118941610482127, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.796579122543335, + "num_tokens": 635397568.0, + "step": 16657 + }, + { + "epoch": 2.1190688207607176, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.781914472579956, + "num_tokens": 635432903.0, + "step": 16658 + }, + { + "epoch": 2.119196031039308, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8001723289489746, + "num_tokens": 635473388.0, + "step": 16659 + }, + { + "epoch": 2.1193232413178986, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8056658506393433, + "num_tokens": 635512966.0, + "step": 16660 + }, + { + "epoch": 2.119450451596489, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7785873413085938, + "num_tokens": 635547814.0, + "step": 16661 + }, + { + "epoch": 2.1195776618750797, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7854022979736328, + "num_tokens": 635583120.0, + "step": 16662 + }, + { + "epoch": 2.11970487215367, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7961536049842834, + "num_tokens": 635621531.0, + "step": 16663 + }, + { + "epoch": 2.1198320824322607, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7915543913841248, + "num_tokens": 635659641.0, + "step": 16664 + }, + { + "epoch": 2.119959292710851, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7976514101028442, + "num_tokens": 635697953.0, + "step": 16665 + }, + { + "epoch": 2.1200865029894413, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7797286510467529, + "num_tokens": 635742519.0, + "step": 16666 + }, + { + "epoch": 2.120213713268032, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8053380250930786, + "num_tokens": 635777786.0, + "step": 16667 + }, + { + "epoch": 2.1203409235466224, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8008136749267578, + "num_tokens": 635814490.0, + "step": 16668 + }, + { + "epoch": 2.120468133825213, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.789365291595459, + "num_tokens": 635858351.0, + "step": 16669 + }, + { + "epoch": 2.1205953441038035, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7937353849411011, + "num_tokens": 635895585.0, + "step": 16670 + }, + { + "epoch": 2.120722554382394, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8061671257019043, + "num_tokens": 635933591.0, + "step": 16671 + }, + { + "epoch": 2.1208497646609845, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7945163249969482, + "num_tokens": 635972040.0, + "step": 16672 + }, + { + "epoch": 2.120976974939575, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8078561425209045, + "num_tokens": 636013287.0, + "step": 16673 + }, + { + "epoch": 2.1211041852181656, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7937313318252563, + "num_tokens": 636046137.0, + "step": 16674 + }, + { + "epoch": 2.121231395496756, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7923637628555298, + "num_tokens": 636088870.0, + "step": 16675 + }, + { + "epoch": 2.1213586057753466, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8118863105773926, + "num_tokens": 636130519.0, + "step": 16676 + }, + { + "epoch": 2.121485816053937, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8023017048835754, + "num_tokens": 636165641.0, + "step": 16677 + }, + { + "epoch": 2.1216130263325277, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.784791886806488, + "num_tokens": 636203811.0, + "step": 16678 + }, + { + "epoch": 2.121740236611118, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8090701103210449, + "num_tokens": 636239503.0, + "step": 16679 + }, + { + "epoch": 2.1218674468897087, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8006771802902222, + "num_tokens": 636278994.0, + "step": 16680 + }, + { + "epoch": 2.1219946571682993, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7989622354507446, + "num_tokens": 636318025.0, + "step": 16681 + }, + { + "epoch": 2.12212186744689, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8015841245651245, + "num_tokens": 636357289.0, + "step": 16682 + }, + { + "epoch": 2.1222490777254803, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.786069393157959, + "num_tokens": 636399067.0, + "step": 16683 + }, + { + "epoch": 2.122376288004071, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8114597797393799, + "num_tokens": 636442418.0, + "step": 16684 + }, + { + "epoch": 2.1225034982826614, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7889593839645386, + "num_tokens": 636484273.0, + "step": 16685 + }, + { + "epoch": 2.122630708561252, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8101022839546204, + "num_tokens": 636525675.0, + "step": 16686 + }, + { + "epoch": 2.1227579188398424, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7987810373306274, + "num_tokens": 636559652.0, + "step": 16687 + }, + { + "epoch": 2.122885129118433, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8250675201416016, + "num_tokens": 636597900.0, + "step": 16688 + }, + { + "epoch": 2.1230123393970235, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7994420528411865, + "num_tokens": 636635822.0, + "step": 16689 + }, + { + "epoch": 2.1231395496756136, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7819945812225342, + "num_tokens": 636667275.0, + "step": 16690 + }, + { + "epoch": 2.123266759954204, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8034679889678955, + "num_tokens": 636704723.0, + "step": 16691 + }, + { + "epoch": 2.1233939702327946, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8089056611061096, + "num_tokens": 636748209.0, + "step": 16692 + }, + { + "epoch": 2.123521180511385, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8093801736831665, + "num_tokens": 636785052.0, + "step": 16693 + }, + { + "epoch": 2.1236483907899757, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8119488954544067, + "num_tokens": 636816073.0, + "step": 16694 + }, + { + "epoch": 2.123775601068566, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8017635941505432, + "num_tokens": 636848963.0, + "step": 16695 + }, + { + "epoch": 2.1239028113471567, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8129695057868958, + "num_tokens": 636880632.0, + "step": 16696 + }, + { + "epoch": 2.1240300216257473, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7883585691452026, + "num_tokens": 636923597.0, + "step": 16697 + }, + { + "epoch": 2.124157231904338, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7937033176422119, + "num_tokens": 636962250.0, + "step": 16698 + }, + { + "epoch": 2.1242844421829283, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8094562888145447, + "num_tokens": 636997973.0, + "step": 16699 + }, + { + "epoch": 2.124411652461519, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7856917381286621, + "num_tokens": 637037320.0, + "step": 16700 + }, + { + "epoch": 2.1245388627401094, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7766820788383484, + "num_tokens": 637072267.0, + "step": 16701 + }, + { + "epoch": 2.1246660730187, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.820226788520813, + "num_tokens": 637108782.0, + "step": 16702 + }, + { + "epoch": 2.1247932832972904, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7972246408462524, + "num_tokens": 637147368.0, + "step": 16703 + }, + { + "epoch": 2.124920493575881, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8153190612792969, + "num_tokens": 637181909.0, + "step": 16704 + }, + { + "epoch": 2.1250477038544715, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7977584600448608, + "num_tokens": 637218498.0, + "step": 16705 + }, + { + "epoch": 2.125174914133062, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7998669147491455, + "num_tokens": 637254825.0, + "step": 16706 + }, + { + "epoch": 2.1253021244116526, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7930012941360474, + "num_tokens": 637301593.0, + "step": 16707 + }, + { + "epoch": 2.125429334690243, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.813509464263916, + "num_tokens": 637341676.0, + "step": 16708 + }, + { + "epoch": 2.1255565449688336, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8066984415054321, + "num_tokens": 637378840.0, + "step": 16709 + }, + { + "epoch": 2.125683755247424, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7826038599014282, + "num_tokens": 637412711.0, + "step": 16710 + }, + { + "epoch": 2.1258109655260147, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8026508092880249, + "num_tokens": 637448287.0, + "step": 16711 + }, + { + "epoch": 2.125938175804605, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7951008677482605, + "num_tokens": 637487899.0, + "step": 16712 + }, + { + "epoch": 2.1260653860831957, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8089101314544678, + "num_tokens": 637528161.0, + "step": 16713 + }, + { + "epoch": 2.1261925963617863, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7812834978103638, + "num_tokens": 637565692.0, + "step": 16714 + }, + { + "epoch": 2.1263198066403763, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7888727188110352, + "num_tokens": 637600368.0, + "step": 16715 + }, + { + "epoch": 2.126447016918967, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7948892116546631, + "num_tokens": 637635732.0, + "step": 16716 + }, + { + "epoch": 2.1265742271975574, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7940263748168945, + "num_tokens": 637677127.0, + "step": 16717 + }, + { + "epoch": 2.126701437476148, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7988361716270447, + "num_tokens": 637715278.0, + "step": 16718 + }, + { + "epoch": 2.1268286477547385, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7976982593536377, + "num_tokens": 637742937.0, + "step": 16719 + }, + { + "epoch": 2.126955858033329, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8004578351974487, + "num_tokens": 637776761.0, + "step": 16720 + }, + { + "epoch": 2.1270830683119195, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7995383739471436, + "num_tokens": 637812197.0, + "step": 16721 + }, + { + "epoch": 2.12721027859051, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8066096305847168, + "num_tokens": 637850442.0, + "step": 16722 + }, + { + "epoch": 2.1273374888691006, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8032217025756836, + "num_tokens": 637889352.0, + "step": 16723 + }, + { + "epoch": 2.127464699147691, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7799856662750244, + "num_tokens": 637927195.0, + "step": 16724 + }, + { + "epoch": 2.1275919094262816, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7874284982681274, + "num_tokens": 637969431.0, + "step": 16725 + }, + { + "epoch": 2.127719119704872, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8165919780731201, + "num_tokens": 638013910.0, + "step": 16726 + }, + { + "epoch": 2.1278463299834627, + "grad_norm": 51.0, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7906163930892944, + "num_tokens": 638051848.0, + "step": 16727 + }, + { + "epoch": 2.127973540262053, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.799424946308136, + "num_tokens": 638092192.0, + "step": 16728 + }, + { + "epoch": 2.1281007505406437, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8097297549247742, + "num_tokens": 638134306.0, + "step": 16729 + }, + { + "epoch": 2.1282279608192343, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.78834068775177, + "num_tokens": 638177988.0, + "step": 16730 + }, + { + "epoch": 2.128355171097825, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8121160864830017, + "num_tokens": 638212237.0, + "step": 16731 + }, + { + "epoch": 2.1284823813764153, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7907465696334839, + "num_tokens": 638252492.0, + "step": 16732 + }, + { + "epoch": 2.128609591655006, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7986557483673096, + "num_tokens": 638294308.0, + "step": 16733 + }, + { + "epoch": 2.1287368019335964, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8002599477767944, + "num_tokens": 638330178.0, + "step": 16734 + }, + { + "epoch": 2.128864012212187, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7831058502197266, + "num_tokens": 638372731.0, + "step": 16735 + }, + { + "epoch": 2.1289912224907774, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7914891242980957, + "num_tokens": 638407522.0, + "step": 16736 + }, + { + "epoch": 2.129118432769368, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.807266116142273, + "num_tokens": 638443262.0, + "step": 16737 + }, + { + "epoch": 2.129245643047958, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8060171008110046, + "num_tokens": 638482729.0, + "step": 16738 + }, + { + "epoch": 2.129372853326549, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8067142367362976, + "num_tokens": 638520034.0, + "step": 16739 + }, + { + "epoch": 2.129500063605139, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8096622228622437, + "num_tokens": 638558631.0, + "step": 16740 + }, + { + "epoch": 2.1296272738837296, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8084795475006104, + "num_tokens": 638593565.0, + "step": 16741 + }, + { + "epoch": 2.12975448416232, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8022894859313965, + "num_tokens": 638625952.0, + "step": 16742 + }, + { + "epoch": 2.1298816944409107, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7974388599395752, + "num_tokens": 638662000.0, + "step": 16743 + }, + { + "epoch": 2.130008904719501, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7876539826393127, + "num_tokens": 638706020.0, + "step": 16744 + }, + { + "epoch": 2.1301361149980917, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8080275654792786, + "num_tokens": 638745354.0, + "step": 16745 + }, + { + "epoch": 2.1302633252766823, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7938534021377563, + "num_tokens": 638785497.0, + "step": 16746 + }, + { + "epoch": 2.130390535555273, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.782948911190033, + "num_tokens": 638821650.0, + "step": 16747 + }, + { + "epoch": 2.1305177458338633, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7813553810119629, + "num_tokens": 638855589.0, + "step": 16748 + }, + { + "epoch": 2.130644956112454, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.80130535364151, + "num_tokens": 638895766.0, + "step": 16749 + }, + { + "epoch": 2.1307721663910444, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.786572277545929, + "num_tokens": 638932622.0, + "step": 16750 + }, + { + "epoch": 2.130899376669635, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7849063873291016, + "num_tokens": 638971588.0, + "step": 16751 + }, + { + "epoch": 2.1310265869482254, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8056870698928833, + "num_tokens": 639009921.0, + "step": 16752 + }, + { + "epoch": 2.131153797226816, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7947654724121094, + "num_tokens": 639048160.0, + "step": 16753 + }, + { + "epoch": 2.1312810075054065, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8118903040885925, + "num_tokens": 639080260.0, + "step": 16754 + }, + { + "epoch": 2.131408217783997, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7929558157920837, + "num_tokens": 639118758.0, + "step": 16755 + }, + { + "epoch": 2.1315354280625876, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7986623048782349, + "num_tokens": 639159528.0, + "step": 16756 + }, + { + "epoch": 2.131662638341178, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.805448591709137, + "num_tokens": 639200394.0, + "step": 16757 + }, + { + "epoch": 2.1317898486197686, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7916150689125061, + "num_tokens": 639240938.0, + "step": 16758 + }, + { + "epoch": 2.131917058898359, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7972158193588257, + "num_tokens": 639278097.0, + "step": 16759 + }, + { + "epoch": 2.1320442691769497, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8045209646224976, + "num_tokens": 639324512.0, + "step": 16760 + }, + { + "epoch": 2.13217147945554, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8070112466812134, + "num_tokens": 639366833.0, + "step": 16761 + }, + { + "epoch": 2.1322986897341307, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7810335159301758, + "num_tokens": 639405765.0, + "step": 16762 + }, + { + "epoch": 2.132425900012721, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7964029312133789, + "num_tokens": 639439622.0, + "step": 16763 + }, + { + "epoch": 2.1325531102913113, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8124939203262329, + "num_tokens": 639479943.0, + "step": 16764 + }, + { + "epoch": 2.132680320569902, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7936185598373413, + "num_tokens": 639520487.0, + "step": 16765 + }, + { + "epoch": 2.1328075308484924, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8000494241714478, + "num_tokens": 639558997.0, + "step": 16766 + }, + { + "epoch": 2.132934741127083, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7960209846496582, + "num_tokens": 639594569.0, + "step": 16767 + }, + { + "epoch": 2.1330619514056735, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8166372179985046, + "num_tokens": 639629827.0, + "step": 16768 + }, + { + "epoch": 2.133189161684264, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.788345456123352, + "num_tokens": 639667634.0, + "step": 16769 + }, + { + "epoch": 2.1333163719628545, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7925223708152771, + "num_tokens": 639700536.0, + "step": 16770 + }, + { + "epoch": 2.133443582241445, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8081234097480774, + "num_tokens": 639744006.0, + "step": 16771 + }, + { + "epoch": 2.1335707925200356, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8097800612449646, + "num_tokens": 639784950.0, + "step": 16772 + }, + { + "epoch": 2.133698002798626, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.8177976608276367, + "num_tokens": 639823567.0, + "step": 16773 + }, + { + "epoch": 2.1338252130772166, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.8125847578048706, + "num_tokens": 639863386.0, + "step": 16774 + }, + { + "epoch": 2.133952423355807, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7841334342956543, + "num_tokens": 639901219.0, + "step": 16775 + }, + { + "epoch": 2.1340796336343977, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7810574173927307, + "num_tokens": 639935240.0, + "step": 16776 + }, + { + "epoch": 2.134206843912988, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7892266511917114, + "num_tokens": 639975370.0, + "step": 16777 + }, + { + "epoch": 2.1343340541915787, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7811838388442993, + "num_tokens": 640020463.0, + "step": 16778 + }, + { + "epoch": 2.1344612644701693, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7933350801467896, + "num_tokens": 640057142.0, + "step": 16779 + }, + { + "epoch": 2.13458847474876, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7998582720756531, + "num_tokens": 640098879.0, + "step": 16780 + }, + { + "epoch": 2.1347156850273503, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8025987148284912, + "num_tokens": 640136077.0, + "step": 16781 + }, + { + "epoch": 2.134842895305941, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7824399471282959, + "num_tokens": 640175160.0, + "step": 16782 + }, + { + "epoch": 2.1349701055845314, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.558, + "mean_token_accuracy": 0.8321835994720459, + "num_tokens": 640216670.0, + "step": 16783 + }, + { + "epoch": 2.135097315863122, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7966779470443726, + "num_tokens": 640254872.0, + "step": 16784 + }, + { + "epoch": 2.1352245261417124, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7990390062332153, + "num_tokens": 640289319.0, + "step": 16785 + }, + { + "epoch": 2.135351736420303, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8128262758255005, + "num_tokens": 640323897.0, + "step": 16786 + }, + { + "epoch": 2.1354789466988935, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.8108431696891785, + "num_tokens": 640366203.0, + "step": 16787 + }, + { + "epoch": 2.1356061569774836, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8097932934761047, + "num_tokens": 640408089.0, + "step": 16788 + }, + { + "epoch": 2.135733367256074, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8093224763870239, + "num_tokens": 640443532.0, + "step": 16789 + }, + { + "epoch": 2.1358605775346646, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7905733585357666, + "num_tokens": 640480425.0, + "step": 16790 + }, + { + "epoch": 2.135987787813255, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8079735040664673, + "num_tokens": 640525658.0, + "step": 16791 + }, + { + "epoch": 2.1361149980918457, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8105472326278687, + "num_tokens": 640563276.0, + "step": 16792 + }, + { + "epoch": 2.136242208370436, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7916710376739502, + "num_tokens": 640608150.0, + "step": 16793 + }, + { + "epoch": 2.1363694186490267, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7980157136917114, + "num_tokens": 640646127.0, + "step": 16794 + }, + { + "epoch": 2.1364966289276173, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8088480234146118, + "num_tokens": 640683195.0, + "step": 16795 + }, + { + "epoch": 2.136623839206208, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8182026147842407, + "num_tokens": 640724102.0, + "step": 16796 + }, + { + "epoch": 2.1367510494847983, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7931010723114014, + "num_tokens": 640762624.0, + "step": 16797 + }, + { + "epoch": 2.136878259763389, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7881996035575867, + "num_tokens": 640799437.0, + "step": 16798 + }, + { + "epoch": 2.1370054700419794, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8044614195823669, + "num_tokens": 640837115.0, + "step": 16799 + }, + { + "epoch": 2.13713268032057, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8050524592399597, + "num_tokens": 640882312.0, + "step": 16800 + }, + { + "epoch": 2.1372598905991604, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7981212139129639, + "num_tokens": 640921032.0, + "step": 16801 + }, + { + "epoch": 2.137387100877751, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7975757122039795, + "num_tokens": 640957424.0, + "step": 16802 + }, + { + "epoch": 2.1375143111563415, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7939169406890869, + "num_tokens": 640992985.0, + "step": 16803 + }, + { + "epoch": 2.137641521434932, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8122990131378174, + "num_tokens": 641030379.0, + "step": 16804 + }, + { + "epoch": 2.1377687317135226, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7868486642837524, + "num_tokens": 641068526.0, + "step": 16805 + }, + { + "epoch": 2.137895941992113, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7897152304649353, + "num_tokens": 641103351.0, + "step": 16806 + }, + { + "epoch": 2.1380231522707036, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8024815917015076, + "num_tokens": 641143570.0, + "step": 16807 + }, + { + "epoch": 2.138150362549294, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8057851195335388, + "num_tokens": 641180625.0, + "step": 16808 + }, + { + "epoch": 2.1382775728278847, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7792588472366333, + "num_tokens": 641226770.0, + "step": 16809 + }, + { + "epoch": 2.138404783106475, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8120743036270142, + "num_tokens": 641269682.0, + "step": 16810 + }, + { + "epoch": 2.1385319933850653, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8081586360931396, + "num_tokens": 641303803.0, + "step": 16811 + }, + { + "epoch": 2.1386592036636562, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7987489104270935, + "num_tokens": 641343141.0, + "step": 16812 + }, + { + "epoch": 2.1387864139422463, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8042646050453186, + "num_tokens": 641384006.0, + "step": 16813 + }, + { + "epoch": 2.138913624220837, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7798328399658203, + "num_tokens": 641423998.0, + "step": 16814 + }, + { + "epoch": 2.1390408344994274, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7973235249519348, + "num_tokens": 641457886.0, + "step": 16815 + }, + { + "epoch": 2.139168044778018, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8055698275566101, + "num_tokens": 641503386.0, + "step": 16816 + }, + { + "epoch": 2.1392952550566084, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8097898960113525, + "num_tokens": 641539533.0, + "step": 16817 + }, + { + "epoch": 2.139422465335199, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7875142693519592, + "num_tokens": 641574020.0, + "step": 16818 + }, + { + "epoch": 2.1395496756137895, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8033565282821655, + "num_tokens": 641605206.0, + "step": 16819 + }, + { + "epoch": 2.13967688589238, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7903124094009399, + "num_tokens": 641649857.0, + "step": 16820 + }, + { + "epoch": 2.1398040961709706, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7892906665802002, + "num_tokens": 641683360.0, + "step": 16821 + }, + { + "epoch": 2.139931306449561, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7968233823776245, + "num_tokens": 641720105.0, + "step": 16822 + }, + { + "epoch": 2.1400585167281516, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8020597100257874, + "num_tokens": 641750718.0, + "step": 16823 + }, + { + "epoch": 2.140185727006742, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.80124831199646, + "num_tokens": 641791620.0, + "step": 16824 + }, + { + "epoch": 2.1403129372853327, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7878016829490662, + "num_tokens": 641833566.0, + "step": 16825 + }, + { + "epoch": 2.140440147563923, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8056668639183044, + "num_tokens": 641869080.0, + "step": 16826 + }, + { + "epoch": 2.1405673578425137, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7949090003967285, + "num_tokens": 641910195.0, + "step": 16827 + }, + { + "epoch": 2.1406945681211043, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7871948480606079, + "num_tokens": 641947270.0, + "step": 16828 + }, + { + "epoch": 2.140821778399695, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7997714281082153, + "num_tokens": 641986472.0, + "step": 16829 + }, + { + "epoch": 2.1409489886782853, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7901254892349243, + "num_tokens": 642028130.0, + "step": 16830 + }, + { + "epoch": 2.141076198956876, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7886930108070374, + "num_tokens": 642067097.0, + "step": 16831 + }, + { + "epoch": 2.1412034092354664, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8125724196434021, + "num_tokens": 642099628.0, + "step": 16832 + }, + { + "epoch": 2.141330619514057, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.5619, + "mean_token_accuracy": 0.827622652053833, + "num_tokens": 642135941.0, + "step": 16833 + }, + { + "epoch": 2.1414578297926474, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.786894679069519, + "num_tokens": 642169465.0, + "step": 16834 + }, + { + "epoch": 2.141585040071238, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8012927174568176, + "num_tokens": 642208961.0, + "step": 16835 + }, + { + "epoch": 2.141712250349828, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7980129718780518, + "num_tokens": 642249691.0, + "step": 16836 + }, + { + "epoch": 2.141839460628419, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8033660054206848, + "num_tokens": 642292425.0, + "step": 16837 + }, + { + "epoch": 2.141966670907009, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8089547157287598, + "num_tokens": 642330982.0, + "step": 16838 + }, + { + "epoch": 2.1420938811855996, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8011476993560791, + "num_tokens": 642365414.0, + "step": 16839 + }, + { + "epoch": 2.14222109146419, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7926597595214844, + "num_tokens": 642398660.0, + "step": 16840 + }, + { + "epoch": 2.1423483017427807, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8073549270629883, + "num_tokens": 642432605.0, + "step": 16841 + }, + { + "epoch": 2.142475512021371, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.8127985000610352, + "num_tokens": 642472041.0, + "step": 16842 + }, + { + "epoch": 2.1426027222999617, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8116406798362732, + "num_tokens": 642512399.0, + "step": 16843 + }, + { + "epoch": 2.1427299325785523, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7993797063827515, + "num_tokens": 642550829.0, + "step": 16844 + }, + { + "epoch": 2.142857142857143, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8012025356292725, + "num_tokens": 642591255.0, + "step": 16845 + }, + { + "epoch": 2.1429843531357333, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7999476194381714, + "num_tokens": 642625772.0, + "step": 16846 + }, + { + "epoch": 2.143111563414324, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7955824136734009, + "num_tokens": 642667596.0, + "step": 16847 + }, + { + "epoch": 2.1432387736929144, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8015384674072266, + "num_tokens": 642702159.0, + "step": 16848 + }, + { + "epoch": 2.143365983971505, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8182010054588318, + "num_tokens": 642737307.0, + "step": 16849 + }, + { + "epoch": 2.1434931942500954, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7874847650527954, + "num_tokens": 642776660.0, + "step": 16850 + }, + { + "epoch": 2.143620404528686, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7921388149261475, + "num_tokens": 642814131.0, + "step": 16851 + }, + { + "epoch": 2.1437476148072765, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7807047367095947, + "num_tokens": 642854138.0, + "step": 16852 + }, + { + "epoch": 2.143874825085867, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8066921234130859, + "num_tokens": 642893767.0, + "step": 16853 + }, + { + "epoch": 2.1440020353644575, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.8164929151535034, + "num_tokens": 642928286.0, + "step": 16854 + }, + { + "epoch": 2.144129245643048, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8116103410720825, + "num_tokens": 642971220.0, + "step": 16855 + }, + { + "epoch": 2.1442564559216386, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.805539071559906, + "num_tokens": 643013618.0, + "step": 16856 + }, + { + "epoch": 2.144383666200229, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.8028618693351746, + "num_tokens": 643058122.0, + "step": 16857 + }, + { + "epoch": 2.1445108764788197, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7813140153884888, + "num_tokens": 643096662.0, + "step": 16858 + }, + { + "epoch": 2.14463808675741, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8169533014297485, + "num_tokens": 643137923.0, + "step": 16859 + }, + { + "epoch": 2.1447652970360007, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8068304657936096, + "num_tokens": 643176918.0, + "step": 16860 + }, + { + "epoch": 2.144892507314591, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8213869333267212, + "num_tokens": 643217525.0, + "step": 16861 + }, + { + "epoch": 2.1450197175931813, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8204569220542908, + "num_tokens": 643255131.0, + "step": 16862 + }, + { + "epoch": 2.145146927871772, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8106565475463867, + "num_tokens": 643292092.0, + "step": 16863 + }, + { + "epoch": 2.1452741381503624, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7944533824920654, + "num_tokens": 643330510.0, + "step": 16864 + }, + { + "epoch": 2.145401348428953, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8021857738494873, + "num_tokens": 643372900.0, + "step": 16865 + }, + { + "epoch": 2.1455285587075434, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8118444681167603, + "num_tokens": 643412866.0, + "step": 16866 + }, + { + "epoch": 2.145655768986134, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7765703201293945, + "num_tokens": 643452846.0, + "step": 16867 + }, + { + "epoch": 2.1457829792647245, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8109855651855469, + "num_tokens": 643488264.0, + "step": 16868 + }, + { + "epoch": 2.145910189543315, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.790027916431427, + "num_tokens": 643528131.0, + "step": 16869 + }, + { + "epoch": 2.1460373998219056, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7988182902336121, + "num_tokens": 643564041.0, + "step": 16870 + }, + { + "epoch": 2.146164610100496, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7868591547012329, + "num_tokens": 643601507.0, + "step": 16871 + }, + { + "epoch": 2.1462918203790866, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7853646278381348, + "num_tokens": 643647161.0, + "step": 16872 + }, + { + "epoch": 2.146419030657677, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.8104618787765503, + "num_tokens": 643687114.0, + "step": 16873 + }, + { + "epoch": 2.1465462409362677, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7863852381706238, + "num_tokens": 643722110.0, + "step": 16874 + }, + { + "epoch": 2.146673451214858, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7802086472511292, + "num_tokens": 643765770.0, + "step": 16875 + }, + { + "epoch": 2.1468006614934487, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7945449352264404, + "num_tokens": 643810097.0, + "step": 16876 + }, + { + "epoch": 2.1469278717720393, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7707233428955078, + "num_tokens": 643847904.0, + "step": 16877 + }, + { + "epoch": 2.14705508205063, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8077536821365356, + "num_tokens": 643889157.0, + "step": 16878 + }, + { + "epoch": 2.1471822923292203, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8017537593841553, + "num_tokens": 643931181.0, + "step": 16879 + }, + { + "epoch": 2.147309502607811, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7965721487998962, + "num_tokens": 643965415.0, + "step": 16880 + }, + { + "epoch": 2.1474367128864014, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7851552963256836, + "num_tokens": 644010647.0, + "step": 16881 + }, + { + "epoch": 2.147563923164992, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8178948163986206, + "num_tokens": 644044347.0, + "step": 16882 + }, + { + "epoch": 2.1476911334435824, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8099881410598755, + "num_tokens": 644083456.0, + "step": 16883 + }, + { + "epoch": 2.147818343722173, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7922384142875671, + "num_tokens": 644123851.0, + "step": 16884 + }, + { + "epoch": 2.1479455540007635, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8130447864532471, + "num_tokens": 644168880.0, + "step": 16885 + }, + { + "epoch": 2.1480727642793536, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8039620518684387, + "num_tokens": 644204532.0, + "step": 16886 + }, + { + "epoch": 2.148199974557944, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8241393566131592, + "num_tokens": 644246160.0, + "step": 16887 + }, + { + "epoch": 2.1483271848365346, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8163971900939941, + "num_tokens": 644284881.0, + "step": 16888 + }, + { + "epoch": 2.148454395115125, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7908360958099365, + "num_tokens": 644326661.0, + "step": 16889 + }, + { + "epoch": 2.1485816053937157, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8041720390319824, + "num_tokens": 644364420.0, + "step": 16890 + }, + { + "epoch": 2.148708815672306, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7810382843017578, + "num_tokens": 644398626.0, + "step": 16891 + }, + { + "epoch": 2.1488360259508967, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7972402572631836, + "num_tokens": 644434772.0, + "step": 16892 + }, + { + "epoch": 2.1489632362294873, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7956174612045288, + "num_tokens": 644474425.0, + "step": 16893 + }, + { + "epoch": 2.149090446508078, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7897845506668091, + "num_tokens": 644507796.0, + "step": 16894 + }, + { + "epoch": 2.1492176567866683, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7909795641899109, + "num_tokens": 644544172.0, + "step": 16895 + }, + { + "epoch": 2.149344867065259, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8018794059753418, + "num_tokens": 644582217.0, + "step": 16896 + }, + { + "epoch": 2.1494720773438494, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8082211017608643, + "num_tokens": 644620028.0, + "step": 16897 + }, + { + "epoch": 2.14959928762244, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7836074829101562, + "num_tokens": 644663743.0, + "step": 16898 + }, + { + "epoch": 2.1497264979010304, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8029327392578125, + "num_tokens": 644700914.0, + "step": 16899 + }, + { + "epoch": 2.149853708179621, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.793440043926239, + "num_tokens": 644744232.0, + "step": 16900 + }, + { + "epoch": 2.1499809184582115, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7889081835746765, + "num_tokens": 644780851.0, + "step": 16901 + }, + { + "epoch": 2.150108128736802, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8098434209823608, + "num_tokens": 644822103.0, + "step": 16902 + }, + { + "epoch": 2.1502353390153925, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7988671660423279, + "num_tokens": 644861209.0, + "step": 16903 + }, + { + "epoch": 2.150362549293983, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7864279747009277, + "num_tokens": 644899860.0, + "step": 16904 + }, + { + "epoch": 2.1504897595725736, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7957326173782349, + "num_tokens": 644943101.0, + "step": 16905 + }, + { + "epoch": 2.150616969851164, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7859315872192383, + "num_tokens": 644981786.0, + "step": 16906 + }, + { + "epoch": 2.1507441801297547, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8033968210220337, + "num_tokens": 645017949.0, + "step": 16907 + }, + { + "epoch": 2.150871390408345, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7965919971466064, + "num_tokens": 645053443.0, + "step": 16908 + }, + { + "epoch": 2.1509986006869353, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7969557046890259, + "num_tokens": 645089916.0, + "step": 16909 + }, + { + "epoch": 2.1511258109655262, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7838230133056641, + "num_tokens": 645134531.0, + "step": 16910 + }, + { + "epoch": 2.1512530212441163, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7951379418373108, + "num_tokens": 645172608.0, + "step": 16911 + }, + { + "epoch": 2.151380231522707, + "grad_norm": 10.1875, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7987008690834045, + "num_tokens": 645209895.0, + "step": 16912 + }, + { + "epoch": 2.1515074418012974, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8058722019195557, + "num_tokens": 645245259.0, + "step": 16913 + }, + { + "epoch": 2.151634652079888, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.795121967792511, + "num_tokens": 645284962.0, + "step": 16914 + }, + { + "epoch": 2.1517618623584784, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8118692636489868, + "num_tokens": 645322352.0, + "step": 16915 + }, + { + "epoch": 2.151889072637069, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8050230741500854, + "num_tokens": 645358401.0, + "step": 16916 + }, + { + "epoch": 2.1520162829156595, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8170669674873352, + "num_tokens": 645399147.0, + "step": 16917 + }, + { + "epoch": 2.15214349319425, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7810059189796448, + "num_tokens": 645437286.0, + "step": 16918 + }, + { + "epoch": 2.1522707034728406, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7938151359558105, + "num_tokens": 645477853.0, + "step": 16919 + }, + { + "epoch": 2.152397913751431, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7877193093299866, + "num_tokens": 645520476.0, + "step": 16920 + }, + { + "epoch": 2.1525251240300216, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7874688506126404, + "num_tokens": 645561539.0, + "step": 16921 + }, + { + "epoch": 2.152652334308612, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7948956489562988, + "num_tokens": 645602950.0, + "step": 16922 + }, + { + "epoch": 2.1527795445872027, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7920186519622803, + "num_tokens": 645639215.0, + "step": 16923 + }, + { + "epoch": 2.152906754865793, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7933499813079834, + "num_tokens": 645677287.0, + "step": 16924 + }, + { + "epoch": 2.1530339651443837, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8095946907997131, + "num_tokens": 645710784.0, + "step": 16925 + }, + { + "epoch": 2.1531611754229742, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7972278594970703, + "num_tokens": 645750568.0, + "step": 16926 + }, + { + "epoch": 2.1532883857015648, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8044246435165405, + "num_tokens": 645783911.0, + "step": 16927 + }, + { + "epoch": 2.1534155959801553, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8190212249755859, + "num_tokens": 645822524.0, + "step": 16928 + }, + { + "epoch": 2.153542806258746, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7915751934051514, + "num_tokens": 645856924.0, + "step": 16929 + }, + { + "epoch": 2.1536700165373364, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7962293028831482, + "num_tokens": 645896995.0, + "step": 16930 + }, + { + "epoch": 2.153797226815927, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8046388626098633, + "num_tokens": 645941768.0, + "step": 16931 + }, + { + "epoch": 2.1539244370945174, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.796920657157898, + "num_tokens": 645981909.0, + "step": 16932 + }, + { + "epoch": 2.154051647373108, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7894501686096191, + "num_tokens": 646022034.0, + "step": 16933 + }, + { + "epoch": 2.154178857651698, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7921257019042969, + "num_tokens": 646056714.0, + "step": 16934 + }, + { + "epoch": 2.1543060679302886, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7937678098678589, + "num_tokens": 646095668.0, + "step": 16935 + }, + { + "epoch": 2.154433278208879, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7823463082313538, + "num_tokens": 646136745.0, + "step": 16936 + }, + { + "epoch": 2.1545604884874696, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7958975434303284, + "num_tokens": 646180314.0, + "step": 16937 + }, + { + "epoch": 2.15468769876606, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8017160296440125, + "num_tokens": 646218191.0, + "step": 16938 + }, + { + "epoch": 2.1548149090446507, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8165228366851807, + "num_tokens": 646256350.0, + "step": 16939 + }, + { + "epoch": 2.154942119323241, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.548, + "mean_token_accuracy": 0.8292319178581238, + "num_tokens": 646290915.0, + "step": 16940 + }, + { + "epoch": 2.1550693296018317, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7942515015602112, + "num_tokens": 646325838.0, + "step": 16941 + }, + { + "epoch": 2.1551965398804223, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7803271412849426, + "num_tokens": 646362742.0, + "step": 16942 + }, + { + "epoch": 2.155323750159013, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7894989848136902, + "num_tokens": 646403958.0, + "step": 16943 + }, + { + "epoch": 2.1554509604376033, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7896881699562073, + "num_tokens": 646442872.0, + "step": 16944 + }, + { + "epoch": 2.155578170716194, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8098851442337036, + "num_tokens": 646484642.0, + "step": 16945 + }, + { + "epoch": 2.1557053809947844, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8046076893806458, + "num_tokens": 646521688.0, + "step": 16946 + }, + { + "epoch": 2.155832591273375, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8085176944732666, + "num_tokens": 646560182.0, + "step": 16947 + }, + { + "epoch": 2.1559598015519654, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8036537170410156, + "num_tokens": 646593252.0, + "step": 16948 + }, + { + "epoch": 2.156087011830556, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8010931015014648, + "num_tokens": 646635514.0, + "step": 16949 + }, + { + "epoch": 2.1562142221091465, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7853879928588867, + "num_tokens": 646674785.0, + "step": 16950 + }, + { + "epoch": 2.156341432387737, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8025460243225098, + "num_tokens": 646713832.0, + "step": 16951 + }, + { + "epoch": 2.1564686426663275, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8033924102783203, + "num_tokens": 646752218.0, + "step": 16952 + }, + { + "epoch": 2.156595852944918, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7953814268112183, + "num_tokens": 646786196.0, + "step": 16953 + }, + { + "epoch": 2.1567230632235086, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.790321946144104, + "num_tokens": 646821462.0, + "step": 16954 + }, + { + "epoch": 2.156850273502099, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.809341549873352, + "num_tokens": 646856613.0, + "step": 16955 + }, + { + "epoch": 2.1569774837806897, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8105335831642151, + "num_tokens": 646893013.0, + "step": 16956 + }, + { + "epoch": 2.15710469405928, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8045914173126221, + "num_tokens": 646936180.0, + "step": 16957 + }, + { + "epoch": 2.1572319043378707, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8058034181594849, + "num_tokens": 646979314.0, + "step": 16958 + }, + { + "epoch": 2.157359114616461, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8054232597351074, + "num_tokens": 647020136.0, + "step": 16959 + }, + { + "epoch": 2.1574863248950513, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8029770851135254, + "num_tokens": 647052970.0, + "step": 16960 + }, + { + "epoch": 2.157613535173642, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.792831301689148, + "num_tokens": 647093195.0, + "step": 16961 + }, + { + "epoch": 2.1577407454522324, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.8006410598754883, + "num_tokens": 647133800.0, + "step": 16962 + }, + { + "epoch": 2.157867955730823, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7801350951194763, + "num_tokens": 647167704.0, + "step": 16963 + }, + { + "epoch": 2.1579951660094134, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7872301340103149, + "num_tokens": 647208474.0, + "step": 16964 + }, + { + "epoch": 2.158122376288004, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8139474391937256, + "num_tokens": 647242131.0, + "step": 16965 + }, + { + "epoch": 2.1582495865665945, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7789019346237183, + "num_tokens": 647280952.0, + "step": 16966 + }, + { + "epoch": 2.158376796845185, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7930678725242615, + "num_tokens": 647318411.0, + "step": 16967 + }, + { + "epoch": 2.1585040071237755, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7827150225639343, + "num_tokens": 647361002.0, + "step": 16968 + }, + { + "epoch": 2.158631217402366, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8046660423278809, + "num_tokens": 647402995.0, + "step": 16969 + }, + { + "epoch": 2.1587584276809566, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8019497394561768, + "num_tokens": 647441220.0, + "step": 16970 + }, + { + "epoch": 2.158885637959547, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.786620557308197, + "num_tokens": 647476832.0, + "step": 16971 + }, + { + "epoch": 2.1590128482381377, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8049290180206299, + "num_tokens": 647516855.0, + "step": 16972 + }, + { + "epoch": 2.159140058516728, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7945876121520996, + "num_tokens": 647556913.0, + "step": 16973 + }, + { + "epoch": 2.1592672687953187, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.811255931854248, + "num_tokens": 647591242.0, + "step": 16974 + }, + { + "epoch": 2.1593944790739092, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8052959442138672, + "num_tokens": 647632156.0, + "step": 16975 + }, + { + "epoch": 2.1595216893524998, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8035840392112732, + "num_tokens": 647669857.0, + "step": 16976 + }, + { + "epoch": 2.1596488996310903, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7919109463691711, + "num_tokens": 647715580.0, + "step": 16977 + }, + { + "epoch": 2.159776109909681, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7875058054924011, + "num_tokens": 647755917.0, + "step": 16978 + }, + { + "epoch": 2.1599033201882714, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.819787859916687, + "num_tokens": 647802205.0, + "step": 16979 + }, + { + "epoch": 2.160030530466862, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7914853692054749, + "num_tokens": 647846627.0, + "step": 16980 + }, + { + "epoch": 2.1601577407454524, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.791013240814209, + "num_tokens": 647885033.0, + "step": 16981 + }, + { + "epoch": 2.160284951024043, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7904655933380127, + "num_tokens": 647923370.0, + "step": 16982 + }, + { + "epoch": 2.1604121613026335, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8090466260910034, + "num_tokens": 647963431.0, + "step": 16983 + }, + { + "epoch": 2.1605393715812236, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7942121028900146, + "num_tokens": 648002854.0, + "step": 16984 + }, + { + "epoch": 2.160666581859814, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7831812500953674, + "num_tokens": 648041082.0, + "step": 16985 + }, + { + "epoch": 2.1607937921384046, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7915191650390625, + "num_tokens": 648082109.0, + "step": 16986 + }, + { + "epoch": 2.160921002416995, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8099198341369629, + "num_tokens": 648117398.0, + "step": 16987 + }, + { + "epoch": 2.1610482126955857, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8096274733543396, + "num_tokens": 648158997.0, + "step": 16988 + }, + { + "epoch": 2.161175422974176, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7952567338943481, + "num_tokens": 648194334.0, + "step": 16989 + }, + { + "epoch": 2.1613026332527667, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8006826043128967, + "num_tokens": 648230601.0, + "step": 16990 + }, + { + "epoch": 2.1614298435313573, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8017920255661011, + "num_tokens": 648274979.0, + "step": 16991 + }, + { + "epoch": 2.161557053809948, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7800572514533997, + "num_tokens": 648319229.0, + "step": 16992 + }, + { + "epoch": 2.1616842640885383, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.81241375207901, + "num_tokens": 648354550.0, + "step": 16993 + }, + { + "epoch": 2.161811474367129, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8157234191894531, + "num_tokens": 648389820.0, + "step": 16994 + }, + { + "epoch": 2.1619386846457194, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8156062364578247, + "num_tokens": 648423841.0, + "step": 16995 + }, + { + "epoch": 2.16206589492431, + "grad_norm": 72.0, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7835744619369507, + "num_tokens": 648459865.0, + "step": 16996 + }, + { + "epoch": 2.1621931052029004, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8186386227607727, + "num_tokens": 648492639.0, + "step": 16997 + }, + { + "epoch": 2.162320315481491, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8067365884780884, + "num_tokens": 648533456.0, + "step": 16998 + }, + { + "epoch": 2.1624475257600815, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8106988668441772, + "num_tokens": 648571955.0, + "step": 16999 + }, + { + "epoch": 2.162574736038672, + "grad_norm": 9.75, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8153213858604431, + "num_tokens": 648614284.0, + "step": 17000 + }, + { + "epoch": 2.1627019463172625, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8042666912078857, + "num_tokens": 648654399.0, + "step": 17001 + }, + { + "epoch": 2.162829156595853, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.8061354160308838, + "num_tokens": 648694490.0, + "step": 17002 + }, + { + "epoch": 2.1629563668744436, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8062355518341064, + "num_tokens": 648730665.0, + "step": 17003 + }, + { + "epoch": 2.163083577153034, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8031835556030273, + "num_tokens": 648765289.0, + "step": 17004 + }, + { + "epoch": 2.1632107874316246, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8015702962875366, + "num_tokens": 648803982.0, + "step": 17005 + }, + { + "epoch": 2.163337997710215, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8021270632743835, + "num_tokens": 648840992.0, + "step": 17006 + }, + { + "epoch": 2.1634652079888053, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7954796552658081, + "num_tokens": 648883621.0, + "step": 17007 + }, + { + "epoch": 2.1635924182673962, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.781704306602478, + "num_tokens": 648922258.0, + "step": 17008 + }, + { + "epoch": 2.1637196285459863, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7986360192298889, + "num_tokens": 648965010.0, + "step": 17009 + }, + { + "epoch": 2.163846838824577, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7946522235870361, + "num_tokens": 649008124.0, + "step": 17010 + }, + { + "epoch": 2.1639740491031674, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8002456426620483, + "num_tokens": 649044743.0, + "step": 17011 + }, + { + "epoch": 2.164101259381758, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8137544393539429, + "num_tokens": 649078410.0, + "step": 17012 + }, + { + "epoch": 2.1642284696603484, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8117901086807251, + "num_tokens": 649115415.0, + "step": 17013 + }, + { + "epoch": 2.164355679938939, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8002779483795166, + "num_tokens": 649156266.0, + "step": 17014 + }, + { + "epoch": 2.1644828902175295, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8071412444114685, + "num_tokens": 649192552.0, + "step": 17015 + }, + { + "epoch": 2.16461010049612, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8014918565750122, + "num_tokens": 649224163.0, + "step": 17016 + }, + { + "epoch": 2.1647373107747105, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6005, + "mean_token_accuracy": 0.8150044679641724, + "num_tokens": 649259348.0, + "step": 17017 + }, + { + "epoch": 2.164864521053301, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8140325546264648, + "num_tokens": 649296875.0, + "step": 17018 + }, + { + "epoch": 2.1649917313318916, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7816449403762817, + "num_tokens": 649333455.0, + "step": 17019 + }, + { + "epoch": 2.165118941610482, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7906391620635986, + "num_tokens": 649364567.0, + "step": 17020 + }, + { + "epoch": 2.1652461518890727, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7809540033340454, + "num_tokens": 649400579.0, + "step": 17021 + }, + { + "epoch": 2.165373362167663, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8102927207946777, + "num_tokens": 649432940.0, + "step": 17022 + }, + { + "epoch": 2.1655005724462537, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8041584491729736, + "num_tokens": 649466695.0, + "step": 17023 + }, + { + "epoch": 2.1656277827248442, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7943096160888672, + "num_tokens": 649504578.0, + "step": 17024 + }, + { + "epoch": 2.1657549930034348, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7932461500167847, + "num_tokens": 649541536.0, + "step": 17025 + }, + { + "epoch": 2.1658822032820253, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7983407974243164, + "num_tokens": 649581280.0, + "step": 17026 + }, + { + "epoch": 2.166009413560616, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5798, + "mean_token_accuracy": 0.8295495510101318, + "num_tokens": 649619837.0, + "step": 17027 + }, + { + "epoch": 2.1661366238392064, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8090609312057495, + "num_tokens": 649654708.0, + "step": 17028 + }, + { + "epoch": 2.166263834117797, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7781146764755249, + "num_tokens": 649691709.0, + "step": 17029 + }, + { + "epoch": 2.1663910443963874, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.800225019454956, + "num_tokens": 649734876.0, + "step": 17030 + }, + { + "epoch": 2.166518254674978, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8101938962936401, + "num_tokens": 649775477.0, + "step": 17031 + }, + { + "epoch": 2.166645464953568, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8112971782684326, + "num_tokens": 649816136.0, + "step": 17032 + }, + { + "epoch": 2.1667726752321586, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.805451512336731, + "num_tokens": 649847942.0, + "step": 17033 + }, + { + "epoch": 2.166899885510749, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7978060841560364, + "num_tokens": 649887875.0, + "step": 17034 + }, + { + "epoch": 2.1670270957893396, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7817381620407104, + "num_tokens": 649920557.0, + "step": 17035 + }, + { + "epoch": 2.16715430606793, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8106061220169067, + "num_tokens": 649959704.0, + "step": 17036 + }, + { + "epoch": 2.1672815163465207, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8072814345359802, + "num_tokens": 650000433.0, + "step": 17037 + }, + { + "epoch": 2.167408726625111, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8145526647567749, + "num_tokens": 650038793.0, + "step": 17038 + }, + { + "epoch": 2.1675359369037017, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.81109219789505, + "num_tokens": 650072877.0, + "step": 17039 + }, + { + "epoch": 2.1676631471822922, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7840983867645264, + "num_tokens": 650109245.0, + "step": 17040 + }, + { + "epoch": 2.1677903574608828, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8154718279838562, + "num_tokens": 650150254.0, + "step": 17041 + }, + { + "epoch": 2.1679175677394733, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7781445384025574, + "num_tokens": 650189587.0, + "step": 17042 + }, + { + "epoch": 2.168044778018064, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7942123413085938, + "num_tokens": 650225443.0, + "step": 17043 + }, + { + "epoch": 2.1681719882966544, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7935901284217834, + "num_tokens": 650262319.0, + "step": 17044 + }, + { + "epoch": 2.168299198575245, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7764477133750916, + "num_tokens": 650297137.0, + "step": 17045 + }, + { + "epoch": 2.1684264088538354, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7927934527397156, + "num_tokens": 650335521.0, + "step": 17046 + }, + { + "epoch": 2.168553619132426, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7918407320976257, + "num_tokens": 650373050.0, + "step": 17047 + }, + { + "epoch": 2.1686808294110165, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8082140684127808, + "num_tokens": 650407438.0, + "step": 17048 + }, + { + "epoch": 2.168808039689607, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7983923554420471, + "num_tokens": 650447793.0, + "step": 17049 + }, + { + "epoch": 2.1689352499681975, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7999478578567505, + "num_tokens": 650487167.0, + "step": 17050 + }, + { + "epoch": 2.169062460246788, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7790529131889343, + "num_tokens": 650522960.0, + "step": 17051 + }, + { + "epoch": 2.1691896705253786, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7942179441452026, + "num_tokens": 650564161.0, + "step": 17052 + }, + { + "epoch": 2.169316880803969, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7990295886993408, + "num_tokens": 650599201.0, + "step": 17053 + }, + { + "epoch": 2.1694440910825596, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7875070571899414, + "num_tokens": 650637385.0, + "step": 17054 + }, + { + "epoch": 2.16957130136115, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.8007476329803467, + "num_tokens": 650671954.0, + "step": 17055 + }, + { + "epoch": 2.1696985116397407, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.5954, + "mean_token_accuracy": 0.8194645047187805, + "num_tokens": 650708475.0, + "step": 17056 + }, + { + "epoch": 2.169825721918331, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8052388429641724, + "num_tokens": 650748089.0, + "step": 17057 + }, + { + "epoch": 2.1699529321969213, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8047860860824585, + "num_tokens": 650790869.0, + "step": 17058 + }, + { + "epoch": 2.170080142475512, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7856655120849609, + "num_tokens": 650830697.0, + "step": 17059 + }, + { + "epoch": 2.1702073527541024, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8143157958984375, + "num_tokens": 650868755.0, + "step": 17060 + }, + { + "epoch": 2.170334563032693, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8115110397338867, + "num_tokens": 650908019.0, + "step": 17061 + }, + { + "epoch": 2.1704617733112834, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7985501289367676, + "num_tokens": 650954195.0, + "step": 17062 + }, + { + "epoch": 2.170588983589874, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8180063962936401, + "num_tokens": 650992740.0, + "step": 17063 + }, + { + "epoch": 2.1707161938684645, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.814241886138916, + "num_tokens": 651031524.0, + "step": 17064 + }, + { + "epoch": 2.170843404147055, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.799980878829956, + "num_tokens": 651070004.0, + "step": 17065 + }, + { + "epoch": 2.1709706144256455, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7835264801979065, + "num_tokens": 651104474.0, + "step": 17066 + }, + { + "epoch": 2.171097824704236, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7905250191688538, + "num_tokens": 651148913.0, + "step": 17067 + }, + { + "epoch": 2.1712250349828266, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8062256574630737, + "num_tokens": 651192072.0, + "step": 17068 + }, + { + "epoch": 2.171352245261417, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7931337356567383, + "num_tokens": 651226132.0, + "step": 17069 + }, + { + "epoch": 2.1714794555400077, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7984089851379395, + "num_tokens": 651263467.0, + "step": 17070 + }, + { + "epoch": 2.171606665818598, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7924554347991943, + "num_tokens": 651309521.0, + "step": 17071 + }, + { + "epoch": 2.1717338760971887, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7983412742614746, + "num_tokens": 651347026.0, + "step": 17072 + }, + { + "epoch": 2.1718610863757792, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8064762353897095, + "num_tokens": 651380545.0, + "step": 17073 + }, + { + "epoch": 2.1719882966543698, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.805046796798706, + "num_tokens": 651409428.0, + "step": 17074 + }, + { + "epoch": 2.1721155069329603, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7903738021850586, + "num_tokens": 651447993.0, + "step": 17075 + }, + { + "epoch": 2.172242717211551, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7945969104766846, + "num_tokens": 651488210.0, + "step": 17076 + }, + { + "epoch": 2.1723699274901414, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7965800762176514, + "num_tokens": 651525742.0, + "step": 17077 + }, + { + "epoch": 2.172497137768732, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8081917762756348, + "num_tokens": 651565569.0, + "step": 17078 + }, + { + "epoch": 2.1726243480473224, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.795707106590271, + "num_tokens": 651605291.0, + "step": 17079 + }, + { + "epoch": 2.172751558325913, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8027217984199524, + "num_tokens": 651641998.0, + "step": 17080 + }, + { + "epoch": 2.1728787686045035, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7961831092834473, + "num_tokens": 651683040.0, + "step": 17081 + }, + { + "epoch": 2.1730059788830935, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8125988245010376, + "num_tokens": 651718067.0, + "step": 17082 + }, + { + "epoch": 2.173133189161684, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8152579069137573, + "num_tokens": 651761995.0, + "step": 17083 + }, + { + "epoch": 2.1732603994402746, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8155226707458496, + "num_tokens": 651797780.0, + "step": 17084 + }, + { + "epoch": 2.173387609718865, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7883695363998413, + "num_tokens": 651832931.0, + "step": 17085 + }, + { + "epoch": 2.1735148199974557, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7933040261268616, + "num_tokens": 651868440.0, + "step": 17086 + }, + { + "epoch": 2.173642030276046, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7997725009918213, + "num_tokens": 651900029.0, + "step": 17087 + }, + { + "epoch": 2.1737692405546367, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7969638705253601, + "num_tokens": 651944892.0, + "step": 17088 + }, + { + "epoch": 2.1738964508332272, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8081686496734619, + "num_tokens": 651979330.0, + "step": 17089 + }, + { + "epoch": 2.1740236611118178, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7838248610496521, + "num_tokens": 652017389.0, + "step": 17090 + }, + { + "epoch": 2.1741508713904083, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7984442114830017, + "num_tokens": 652052260.0, + "step": 17091 + }, + { + "epoch": 2.174278081668999, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7866971492767334, + "num_tokens": 652089917.0, + "step": 17092 + }, + { + "epoch": 2.1744052919475894, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8050717115402222, + "num_tokens": 652129849.0, + "step": 17093 + }, + { + "epoch": 2.17453250222618, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7881439924240112, + "num_tokens": 652163178.0, + "step": 17094 + }, + { + "epoch": 2.1746597125047704, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8093672394752502, + "num_tokens": 652201521.0, + "step": 17095 + }, + { + "epoch": 2.174786922783361, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7991429567337036, + "num_tokens": 652234820.0, + "step": 17096 + }, + { + "epoch": 2.1749141330619515, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.801025390625, + "num_tokens": 652275686.0, + "step": 17097 + }, + { + "epoch": 2.175041343340542, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7910991907119751, + "num_tokens": 652313243.0, + "step": 17098 + }, + { + "epoch": 2.1751685536191325, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.8188276886940002, + "num_tokens": 652351550.0, + "step": 17099 + }, + { + "epoch": 2.175295763897723, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8004259467124939, + "num_tokens": 652399834.0, + "step": 17100 + }, + { + "epoch": 2.1754229741763136, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7871774435043335, + "num_tokens": 652435839.0, + "step": 17101 + }, + { + "epoch": 2.175550184454904, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7965166568756104, + "num_tokens": 652475747.0, + "step": 17102 + }, + { + "epoch": 2.1756773947334946, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8161393404006958, + "num_tokens": 652517147.0, + "step": 17103 + }, + { + "epoch": 2.175804605012085, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8106935620307922, + "num_tokens": 652552811.0, + "step": 17104 + }, + { + "epoch": 2.1759318152906753, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8048272728919983, + "num_tokens": 652587709.0, + "step": 17105 + }, + { + "epoch": 2.1760590255692662, + "grad_norm": 10.625, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7925384044647217, + "num_tokens": 652626213.0, + "step": 17106 + }, + { + "epoch": 2.1761862358478563, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7705078125, + "num_tokens": 652665692.0, + "step": 17107 + }, + { + "epoch": 2.176313446126447, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7957248687744141, + "num_tokens": 652705515.0, + "step": 17108 + }, + { + "epoch": 2.1764406564050374, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8145790100097656, + "num_tokens": 652743343.0, + "step": 17109 + }, + { + "epoch": 2.176567866683628, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7957415580749512, + "num_tokens": 652784587.0, + "step": 17110 + }, + { + "epoch": 2.1766950769622184, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7904013395309448, + "num_tokens": 652821985.0, + "step": 17111 + }, + { + "epoch": 2.176822287240809, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8081132173538208, + "num_tokens": 652855772.0, + "step": 17112 + }, + { + "epoch": 2.1769494975193995, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7957653999328613, + "num_tokens": 652890497.0, + "step": 17113 + }, + { + "epoch": 2.17707670779799, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.8089166879653931, + "num_tokens": 652931611.0, + "step": 17114 + }, + { + "epoch": 2.1772039180765805, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8008196353912354, + "num_tokens": 652972129.0, + "step": 17115 + }, + { + "epoch": 2.177331128355171, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.797859787940979, + "num_tokens": 653009848.0, + "step": 17116 + }, + { + "epoch": 2.1774583386337616, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.797764241695404, + "num_tokens": 653044614.0, + "step": 17117 + }, + { + "epoch": 2.177585548912352, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.816652774810791, + "num_tokens": 653087614.0, + "step": 17118 + }, + { + "epoch": 2.1777127591909426, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8084952235221863, + "num_tokens": 653120434.0, + "step": 17119 + }, + { + "epoch": 2.177839969469533, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8078901767730713, + "num_tokens": 653159390.0, + "step": 17120 + }, + { + "epoch": 2.1779671797481237, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7929855585098267, + "num_tokens": 653201045.0, + "step": 17121 + }, + { + "epoch": 2.1780943900267142, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8071860671043396, + "num_tokens": 653244968.0, + "step": 17122 + }, + { + "epoch": 2.1782216003053048, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7954477071762085, + "num_tokens": 653281157.0, + "step": 17123 + }, + { + "epoch": 2.1783488105838953, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8128288984298706, + "num_tokens": 653321055.0, + "step": 17124 + }, + { + "epoch": 2.178476020862486, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7840827703475952, + "num_tokens": 653359320.0, + "step": 17125 + }, + { + "epoch": 2.1786032311410763, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5794, + "mean_token_accuracy": 0.8229655027389526, + "num_tokens": 653404186.0, + "step": 17126 + }, + { + "epoch": 2.178730441419667, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.812552809715271, + "num_tokens": 653438122.0, + "step": 17127 + }, + { + "epoch": 2.1788576516982574, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7984591126441956, + "num_tokens": 653474727.0, + "step": 17128 + }, + { + "epoch": 2.178984861976848, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8046945929527283, + "num_tokens": 653513531.0, + "step": 17129 + }, + { + "epoch": 2.179112072255438, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.8150233030319214, + "num_tokens": 653555871.0, + "step": 17130 + }, + { + "epoch": 2.1792392825340285, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.804440975189209, + "num_tokens": 653598825.0, + "step": 17131 + }, + { + "epoch": 2.179366492812619, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8130469918251038, + "num_tokens": 653639838.0, + "step": 17132 + }, + { + "epoch": 2.1794937030912096, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7914690971374512, + "num_tokens": 653678958.0, + "step": 17133 + }, + { + "epoch": 2.1796209133698, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8029803037643433, + "num_tokens": 653718043.0, + "step": 17134 + }, + { + "epoch": 2.1797481236483907, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8175413012504578, + "num_tokens": 653756308.0, + "step": 17135 + }, + { + "epoch": 2.179875333926981, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8079680800437927, + "num_tokens": 653793140.0, + "step": 17136 + }, + { + "epoch": 2.1800025442055717, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8040281534194946, + "num_tokens": 653823441.0, + "step": 17137 + }, + { + "epoch": 2.1801297544841622, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7909238934516907, + "num_tokens": 653860743.0, + "step": 17138 + }, + { + "epoch": 2.1802569647627528, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8079701662063599, + "num_tokens": 653894785.0, + "step": 17139 + }, + { + "epoch": 2.1803841750413433, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7953251600265503, + "num_tokens": 653932208.0, + "step": 17140 + }, + { + "epoch": 2.180511385319934, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7866764068603516, + "num_tokens": 653968818.0, + "step": 17141 + }, + { + "epoch": 2.1806385955985244, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8023905754089355, + "num_tokens": 654010696.0, + "step": 17142 + }, + { + "epoch": 2.180765805877115, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.8083335757255554, + "num_tokens": 654047632.0, + "step": 17143 + }, + { + "epoch": 2.1808930161557054, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8061525821685791, + "num_tokens": 654088200.0, + "step": 17144 + }, + { + "epoch": 2.181020226434296, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7915982007980347, + "num_tokens": 654124225.0, + "step": 17145 + }, + { + "epoch": 2.1811474367128865, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8025558590888977, + "num_tokens": 654163317.0, + "step": 17146 + }, + { + "epoch": 2.181274646991477, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7967708110809326, + "num_tokens": 654199208.0, + "step": 17147 + }, + { + "epoch": 2.1814018572700675, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7884687185287476, + "num_tokens": 654239988.0, + "step": 17148 + }, + { + "epoch": 2.181529067548658, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7916837930679321, + "num_tokens": 654276361.0, + "step": 17149 + }, + { + "epoch": 2.1816562778272486, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7993617057800293, + "num_tokens": 654316690.0, + "step": 17150 + }, + { + "epoch": 2.181783488105839, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8124743700027466, + "num_tokens": 654354112.0, + "step": 17151 + }, + { + "epoch": 2.1819106983844296, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8138841390609741, + "num_tokens": 654398896.0, + "step": 17152 + }, + { + "epoch": 2.18203790866302, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8176473379135132, + "num_tokens": 654435180.0, + "step": 17153 + }, + { + "epoch": 2.1821651189416107, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7898220419883728, + "num_tokens": 654473958.0, + "step": 17154 + }, + { + "epoch": 2.1822923292202008, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7880339622497559, + "num_tokens": 654512918.0, + "step": 17155 + }, + { + "epoch": 2.1824195394987913, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8083610534667969, + "num_tokens": 654552296.0, + "step": 17156 + }, + { + "epoch": 2.182546749777382, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.804876983165741, + "num_tokens": 654592088.0, + "step": 17157 + }, + { + "epoch": 2.1826739600559724, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7960956692695618, + "num_tokens": 654635119.0, + "step": 17158 + }, + { + "epoch": 2.182801170334563, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7928584814071655, + "num_tokens": 654671204.0, + "step": 17159 + }, + { + "epoch": 2.1829283806131534, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7978598475456238, + "num_tokens": 654702986.0, + "step": 17160 + }, + { + "epoch": 2.183055590891744, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7750691175460815, + "num_tokens": 654745813.0, + "step": 17161 + }, + { + "epoch": 2.1831828011703345, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.8000868558883667, + "num_tokens": 654783305.0, + "step": 17162 + }, + { + "epoch": 2.183310011448925, + "grad_norm": 63.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8068716526031494, + "num_tokens": 654823427.0, + "step": 17163 + }, + { + "epoch": 2.1834372217275155, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8175761699676514, + "num_tokens": 654862400.0, + "step": 17164 + }, + { + "epoch": 2.183564432006106, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8065380454063416, + "num_tokens": 654902484.0, + "step": 17165 + }, + { + "epoch": 2.1836916422846966, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8057231903076172, + "num_tokens": 654946882.0, + "step": 17166 + }, + { + "epoch": 2.183818852563287, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7847098112106323, + "num_tokens": 654977850.0, + "step": 17167 + }, + { + "epoch": 2.1839460628418776, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7864348888397217, + "num_tokens": 655015352.0, + "step": 17168 + }, + { + "epoch": 2.184073273120468, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.8002650737762451, + "num_tokens": 655047488.0, + "step": 17169 + }, + { + "epoch": 2.1842004833990587, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8008992075920105, + "num_tokens": 655083410.0, + "step": 17170 + }, + { + "epoch": 2.1843276936776492, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8077362775802612, + "num_tokens": 655120594.0, + "step": 17171 + }, + { + "epoch": 2.1844549039562398, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7976112365722656, + "num_tokens": 655157310.0, + "step": 17172 + }, + { + "epoch": 2.1845821142348303, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8073552846908569, + "num_tokens": 655193749.0, + "step": 17173 + }, + { + "epoch": 2.184709324513421, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.817979097366333, + "num_tokens": 655235532.0, + "step": 17174 + }, + { + "epoch": 2.1848365347920113, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8094303011894226, + "num_tokens": 655282154.0, + "step": 17175 + }, + { + "epoch": 2.184963745070602, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7898542881011963, + "num_tokens": 655321342.0, + "step": 17176 + }, + { + "epoch": 2.1850909553491924, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8186579942703247, + "num_tokens": 655358107.0, + "step": 17177 + }, + { + "epoch": 2.185218165627783, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7913730144500732, + "num_tokens": 655393462.0, + "step": 17178 + }, + { + "epoch": 2.1853453759063735, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8059084415435791, + "num_tokens": 655436114.0, + "step": 17179 + }, + { + "epoch": 2.1854725861849635, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7980450987815857, + "num_tokens": 655473119.0, + "step": 17180 + }, + { + "epoch": 2.185599796463554, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7946122288703918, + "num_tokens": 655515537.0, + "step": 17181 + }, + { + "epoch": 2.1857270067421446, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7955354452133179, + "num_tokens": 655555944.0, + "step": 17182 + }, + { + "epoch": 2.185854217020735, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.772722065448761, + "num_tokens": 655589952.0, + "step": 17183 + }, + { + "epoch": 2.1859814272993257, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8082671165466309, + "num_tokens": 655628269.0, + "step": 17184 + }, + { + "epoch": 2.186108637577916, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8002426624298096, + "num_tokens": 655657315.0, + "step": 17185 + }, + { + "epoch": 2.1862358478565067, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7956300973892212, + "num_tokens": 655696117.0, + "step": 17186 + }, + { + "epoch": 2.1863630581350972, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7831078171730042, + "num_tokens": 655738823.0, + "step": 17187 + }, + { + "epoch": 2.1864902684136878, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7900516986846924, + "num_tokens": 655781457.0, + "step": 17188 + }, + { + "epoch": 2.1866174786922783, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7969841361045837, + "num_tokens": 655824704.0, + "step": 17189 + }, + { + "epoch": 2.186744688970869, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8141453266143799, + "num_tokens": 655860013.0, + "step": 17190 + }, + { + "epoch": 2.1868718992494594, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8014957904815674, + "num_tokens": 655897146.0, + "step": 17191 + }, + { + "epoch": 2.18699910952805, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8040947914123535, + "num_tokens": 655934299.0, + "step": 17192 + }, + { + "epoch": 2.1871263198066404, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8067895174026489, + "num_tokens": 655976554.0, + "step": 17193 + }, + { + "epoch": 2.187253530085231, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.809747040271759, + "num_tokens": 656011756.0, + "step": 17194 + }, + { + "epoch": 2.1873807403638215, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8029608130455017, + "num_tokens": 656054405.0, + "step": 17195 + }, + { + "epoch": 2.187507950642412, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8090323209762573, + "num_tokens": 656094668.0, + "step": 17196 + }, + { + "epoch": 2.1876351609210025, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.80250084400177, + "num_tokens": 656137664.0, + "step": 17197 + }, + { + "epoch": 2.187762371199593, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7960352897644043, + "num_tokens": 656180611.0, + "step": 17198 + }, + { + "epoch": 2.1878895814781836, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7879784107208252, + "num_tokens": 656217789.0, + "step": 17199 + }, + { + "epoch": 2.188016791756774, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7933347225189209, + "num_tokens": 656261940.0, + "step": 17200 + }, + { + "epoch": 2.1881440020353646, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8073037266731262, + "num_tokens": 656296000.0, + "step": 17201 + }, + { + "epoch": 2.188271212313955, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8128243684768677, + "num_tokens": 656340862.0, + "step": 17202 + }, + { + "epoch": 2.1883984225925452, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.5596, + "mean_token_accuracy": 0.8304809927940369, + "num_tokens": 656376923.0, + "step": 17203 + }, + { + "epoch": 2.188525632871136, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7942023277282715, + "num_tokens": 656414058.0, + "step": 17204 + }, + { + "epoch": 2.1886528431497263, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7858773469924927, + "num_tokens": 656450077.0, + "step": 17205 + }, + { + "epoch": 2.188780053428317, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7874276041984558, + "num_tokens": 656487258.0, + "step": 17206 + }, + { + "epoch": 2.1889072637069074, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8127808570861816, + "num_tokens": 656525141.0, + "step": 17207 + }, + { + "epoch": 2.189034473985498, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8056681752204895, + "num_tokens": 656561517.0, + "step": 17208 + }, + { + "epoch": 2.1891616842640884, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8096362352371216, + "num_tokens": 656595859.0, + "step": 17209 + }, + { + "epoch": 2.189288894542679, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.793576717376709, + "num_tokens": 656627652.0, + "step": 17210 + }, + { + "epoch": 2.1894161048212695, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8043892979621887, + "num_tokens": 656669951.0, + "step": 17211 + }, + { + "epoch": 2.18954331509986, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8127001523971558, + "num_tokens": 656702216.0, + "step": 17212 + }, + { + "epoch": 2.1896705253784505, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7952689528465271, + "num_tokens": 656744816.0, + "step": 17213 + }, + { + "epoch": 2.189797735657041, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7965970039367676, + "num_tokens": 656785533.0, + "step": 17214 + }, + { + "epoch": 2.1899249459356316, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7976890206336975, + "num_tokens": 656825375.0, + "step": 17215 + }, + { + "epoch": 2.190052156214222, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8127895593643188, + "num_tokens": 656857951.0, + "step": 17216 + }, + { + "epoch": 2.1901793664928126, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8128218054771423, + "num_tokens": 656896911.0, + "step": 17217 + }, + { + "epoch": 2.190306576771403, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8098254799842834, + "num_tokens": 656935723.0, + "step": 17218 + }, + { + "epoch": 2.1904337870499937, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8154808282852173, + "num_tokens": 656970881.0, + "step": 17219 + }, + { + "epoch": 2.1905609973285842, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7969870567321777, + "num_tokens": 657007721.0, + "step": 17220 + }, + { + "epoch": 2.1906882076071748, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8106213808059692, + "num_tokens": 657042359.0, + "step": 17221 + }, + { + "epoch": 2.1908154178857653, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8054734468460083, + "num_tokens": 657084662.0, + "step": 17222 + }, + { + "epoch": 2.190942628164356, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7991522550582886, + "num_tokens": 657122002.0, + "step": 17223 + }, + { + "epoch": 2.1910698384429463, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8017961978912354, + "num_tokens": 657158768.0, + "step": 17224 + }, + { + "epoch": 2.191197048721537, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7908701300621033, + "num_tokens": 657193445.0, + "step": 17225 + }, + { + "epoch": 2.1913242590001274, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.8017171025276184, + "num_tokens": 657234922.0, + "step": 17226 + }, + { + "epoch": 2.191451469278718, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7855352163314819, + "num_tokens": 657270774.0, + "step": 17227 + }, + { + "epoch": 2.191578679557308, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.8179386854171753, + "num_tokens": 657307046.0, + "step": 17228 + }, + { + "epoch": 2.1917058898358985, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7899854183197021, + "num_tokens": 657352865.0, + "step": 17229 + }, + { + "epoch": 2.191833100114489, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7836950421333313, + "num_tokens": 657393372.0, + "step": 17230 + }, + { + "epoch": 2.1919603103930796, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7824174165725708, + "num_tokens": 657432445.0, + "step": 17231 + }, + { + "epoch": 2.19208752067167, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8026108741760254, + "num_tokens": 657469145.0, + "step": 17232 + }, + { + "epoch": 2.1922147309502606, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.813153862953186, + "num_tokens": 657510369.0, + "step": 17233 + }, + { + "epoch": 2.192341941228851, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.807709276676178, + "num_tokens": 657550257.0, + "step": 17234 + }, + { + "epoch": 2.1924691515074417, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7964215278625488, + "num_tokens": 657587095.0, + "step": 17235 + }, + { + "epoch": 2.1925963617860322, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7998651266098022, + "num_tokens": 657619710.0, + "step": 17236 + }, + { + "epoch": 2.1927235720646228, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7944381833076477, + "num_tokens": 657653316.0, + "step": 17237 + }, + { + "epoch": 2.1928507823432133, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7991008758544922, + "num_tokens": 657693570.0, + "step": 17238 + }, + { + "epoch": 2.192977992621804, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8079057931900024, + "num_tokens": 657736781.0, + "step": 17239 + }, + { + "epoch": 2.1931052029003943, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7937293648719788, + "num_tokens": 657775603.0, + "step": 17240 + }, + { + "epoch": 2.193232413178985, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7955499887466431, + "num_tokens": 657809065.0, + "step": 17241 + }, + { + "epoch": 2.1933596234575754, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8027397394180298, + "num_tokens": 657851369.0, + "step": 17242 + }, + { + "epoch": 2.193486833736166, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.575, + "mean_token_accuracy": 0.83101487159729, + "num_tokens": 657889008.0, + "step": 17243 + }, + { + "epoch": 2.1936140440147565, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8027352094650269, + "num_tokens": 657925867.0, + "step": 17244 + }, + { + "epoch": 2.193741254293347, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7895864844322205, + "num_tokens": 657959502.0, + "step": 17245 + }, + { + "epoch": 2.1938684645719375, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8030453324317932, + "num_tokens": 657999063.0, + "step": 17246 + }, + { + "epoch": 2.193995674850528, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7854050993919373, + "num_tokens": 658033016.0, + "step": 17247 + }, + { + "epoch": 2.1941228851291186, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8088885545730591, + "num_tokens": 658071259.0, + "step": 17248 + }, + { + "epoch": 2.194250095407709, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8034602999687195, + "num_tokens": 658103190.0, + "step": 17249 + }, + { + "epoch": 2.1943773056862996, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7920032739639282, + "num_tokens": 658143346.0, + "step": 17250 + }, + { + "epoch": 2.19450451596489, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8113563060760498, + "num_tokens": 658181747.0, + "step": 17251 + }, + { + "epoch": 2.1946317262434807, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8034527897834778, + "num_tokens": 658221770.0, + "step": 17252 + }, + { + "epoch": 2.1947589365220708, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8053829073905945, + "num_tokens": 658260157.0, + "step": 17253 + }, + { + "epoch": 2.1948861468006613, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7846145629882812, + "num_tokens": 658301815.0, + "step": 17254 + }, + { + "epoch": 2.195013357079252, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8056696057319641, + "num_tokens": 658340732.0, + "step": 17255 + }, + { + "epoch": 2.1951405673578424, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.796721875667572, + "num_tokens": 658380569.0, + "step": 17256 + }, + { + "epoch": 2.195267777636433, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.8136542439460754, + "num_tokens": 658415842.0, + "step": 17257 + }, + { + "epoch": 2.1953949879150234, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8058822154998779, + "num_tokens": 658452861.0, + "step": 17258 + }, + { + "epoch": 2.195522198193614, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.805222749710083, + "num_tokens": 658487019.0, + "step": 17259 + }, + { + "epoch": 2.1956494084722045, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.802893877029419, + "num_tokens": 658529290.0, + "step": 17260 + }, + { + "epoch": 2.195776618750795, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7952196002006531, + "num_tokens": 658572118.0, + "step": 17261 + }, + { + "epoch": 2.1959038290293855, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7984403371810913, + "num_tokens": 658615972.0, + "step": 17262 + }, + { + "epoch": 2.196031039307976, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.810857355594635, + "num_tokens": 658653848.0, + "step": 17263 + }, + { + "epoch": 2.1961582495865666, + "grad_norm": 47.25, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7990906238555908, + "num_tokens": 658686247.0, + "step": 17264 + }, + { + "epoch": 2.196285459865157, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7935726642608643, + "num_tokens": 658725787.0, + "step": 17265 + }, + { + "epoch": 2.1964126701437476, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7946064472198486, + "num_tokens": 658765957.0, + "step": 17266 + }, + { + "epoch": 2.196539880422338, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8116955757141113, + "num_tokens": 658800996.0, + "step": 17267 + }, + { + "epoch": 2.1966670907009287, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7921204566955566, + "num_tokens": 658833779.0, + "step": 17268 + }, + { + "epoch": 2.196794300979519, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8054362535476685, + "num_tokens": 658874425.0, + "step": 17269 + }, + { + "epoch": 2.1969215112581097, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8073631525039673, + "num_tokens": 658912007.0, + "step": 17270 + }, + { + "epoch": 2.1970487215367003, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.791731059551239, + "num_tokens": 658956775.0, + "step": 17271 + }, + { + "epoch": 2.197175931815291, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8149442672729492, + "num_tokens": 658997132.0, + "step": 17272 + }, + { + "epoch": 2.1973031420938813, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8091355562210083, + "num_tokens": 659040437.0, + "step": 17273 + }, + { + "epoch": 2.197430352372472, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8178134560585022, + "num_tokens": 659074461.0, + "step": 17274 + }, + { + "epoch": 2.1975575626510624, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.797073245048523, + "num_tokens": 659113818.0, + "step": 17275 + }, + { + "epoch": 2.197684772929653, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7822914123535156, + "num_tokens": 659149297.0, + "step": 17276 + }, + { + "epoch": 2.1978119832082434, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8078016042709351, + "num_tokens": 659187500.0, + "step": 17277 + }, + { + "epoch": 2.1979391934868335, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7728458642959595, + "num_tokens": 659228624.0, + "step": 17278 + }, + { + "epoch": 2.198066403765424, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.814329981803894, + "num_tokens": 659270317.0, + "step": 17279 + }, + { + "epoch": 2.1981936140440146, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7982020378112793, + "num_tokens": 659302133.0, + "step": 17280 + }, + { + "epoch": 2.198320824322605, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8080739974975586, + "num_tokens": 659342469.0, + "step": 17281 + }, + { + "epoch": 2.1984480346011956, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8162544965744019, + "num_tokens": 659373786.0, + "step": 17282 + }, + { + "epoch": 2.198575244879786, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7923115491867065, + "num_tokens": 659412498.0, + "step": 17283 + }, + { + "epoch": 2.1987024551583767, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7837927341461182, + "num_tokens": 659446415.0, + "step": 17284 + }, + { + "epoch": 2.1988296654369672, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7951755523681641, + "num_tokens": 659488824.0, + "step": 17285 + }, + { + "epoch": 2.1989568757155578, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7996146082878113, + "num_tokens": 659523036.0, + "step": 17286 + }, + { + "epoch": 2.1990840859941483, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7944228649139404, + "num_tokens": 659562780.0, + "step": 17287 + }, + { + "epoch": 2.199211296272739, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8063990473747253, + "num_tokens": 659599364.0, + "step": 17288 + }, + { + "epoch": 2.1993385065513293, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7972615957260132, + "num_tokens": 659636275.0, + "step": 17289 + }, + { + "epoch": 2.19946571682992, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.7975406646728516, + "num_tokens": 659670315.0, + "step": 17290 + }, + { + "epoch": 2.1995929271085104, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7970794439315796, + "num_tokens": 659710968.0, + "step": 17291 + }, + { + "epoch": 2.199720137387101, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8089593648910522, + "num_tokens": 659751334.0, + "step": 17292 + }, + { + "epoch": 2.1998473476656915, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8086867928504944, + "num_tokens": 659793867.0, + "step": 17293 + }, + { + "epoch": 2.199974557944282, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8110185265541077, + "num_tokens": 659838724.0, + "step": 17294 + }, + { + "epoch": 2.2001017682228725, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.8002545237541199, + "num_tokens": 659876589.0, + "step": 17295 + }, + { + "epoch": 2.200228978501463, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7882617712020874, + "num_tokens": 659914864.0, + "step": 17296 + }, + { + "epoch": 2.2003561887800536, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8110687732696533, + "num_tokens": 659952482.0, + "step": 17297 + }, + { + "epoch": 2.200483399058644, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7959231734275818, + "num_tokens": 659985192.0, + "step": 17298 + }, + { + "epoch": 2.2006106093372346, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8082044124603271, + "num_tokens": 660019505.0, + "step": 17299 + }, + { + "epoch": 2.200737819615825, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7912849187850952, + "num_tokens": 660054942.0, + "step": 17300 + }, + { + "epoch": 2.2008650298944152, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8000637888908386, + "num_tokens": 660092836.0, + "step": 17301 + }, + { + "epoch": 2.200992240173006, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.798122227191925, + "num_tokens": 660129419.0, + "step": 17302 + }, + { + "epoch": 2.2011194504515963, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.804853081703186, + "num_tokens": 660168116.0, + "step": 17303 + }, + { + "epoch": 2.201246660730187, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7961294651031494, + "num_tokens": 660208006.0, + "step": 17304 + }, + { + "epoch": 2.2013738710087773, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8037463426589966, + "num_tokens": 660254046.0, + "step": 17305 + }, + { + "epoch": 2.201501081287368, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.8029879331588745, + "num_tokens": 660289321.0, + "step": 17306 + }, + { + "epoch": 2.2016282915659584, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7957993149757385, + "num_tokens": 660325052.0, + "step": 17307 + }, + { + "epoch": 2.201755501844549, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7910747528076172, + "num_tokens": 660357942.0, + "step": 17308 + }, + { + "epoch": 2.2018827121231395, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7714985609054565, + "num_tokens": 660390654.0, + "step": 17309 + }, + { + "epoch": 2.20200992240173, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.801112949848175, + "num_tokens": 660432020.0, + "step": 17310 + }, + { + "epoch": 2.2021371326803205, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7998181581497192, + "num_tokens": 660465051.0, + "step": 17311 + }, + { + "epoch": 2.202264342958911, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7955501079559326, + "num_tokens": 660504967.0, + "step": 17312 + }, + { + "epoch": 2.2023915532375016, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7899236679077148, + "num_tokens": 660543826.0, + "step": 17313 + }, + { + "epoch": 2.202518763516092, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7928176522254944, + "num_tokens": 660583074.0, + "step": 17314 + }, + { + "epoch": 2.2026459737946826, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8207535743713379, + "num_tokens": 660619974.0, + "step": 17315 + }, + { + "epoch": 2.202773184073273, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8041462302207947, + "num_tokens": 660654005.0, + "step": 17316 + }, + { + "epoch": 2.2029003943518637, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7886416912078857, + "num_tokens": 660689179.0, + "step": 17317 + }, + { + "epoch": 2.203027604630454, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7870409488677979, + "num_tokens": 660724923.0, + "step": 17318 + }, + { + "epoch": 2.2031548149090447, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8066501021385193, + "num_tokens": 660760763.0, + "step": 17319 + }, + { + "epoch": 2.2032820251876353, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8002671003341675, + "num_tokens": 660795520.0, + "step": 17320 + }, + { + "epoch": 2.203409235466226, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8083412647247314, + "num_tokens": 660834858.0, + "step": 17321 + }, + { + "epoch": 2.2035364457448163, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8128300905227661, + "num_tokens": 660870319.0, + "step": 17322 + }, + { + "epoch": 2.203663656023407, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7897013425827026, + "num_tokens": 660907313.0, + "step": 17323 + }, + { + "epoch": 2.2037908663019974, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.816034197807312, + "num_tokens": 660948087.0, + "step": 17324 + }, + { + "epoch": 2.203918076580588, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7973895072937012, + "num_tokens": 660984984.0, + "step": 17325 + }, + { + "epoch": 2.204045286859178, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8001146912574768, + "num_tokens": 661022898.0, + "step": 17326 + }, + { + "epoch": 2.2041724971377685, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.8099520802497864, + "num_tokens": 661058645.0, + "step": 17327 + }, + { + "epoch": 2.204299707416359, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8211468458175659, + "num_tokens": 661096954.0, + "step": 17328 + }, + { + "epoch": 2.2044269176949496, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7917002439498901, + "num_tokens": 661136947.0, + "step": 17329 + }, + { + "epoch": 2.20455412797354, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.797275185585022, + "num_tokens": 661175376.0, + "step": 17330 + }, + { + "epoch": 2.2046813382521306, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7912342548370361, + "num_tokens": 661213238.0, + "step": 17331 + }, + { + "epoch": 2.204808548530721, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.596, + "mean_token_accuracy": 0.8186734318733215, + "num_tokens": 661246512.0, + "step": 17332 + }, + { + "epoch": 2.2049357588093117, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.82014000415802, + "num_tokens": 661288272.0, + "step": 17333 + }, + { + "epoch": 2.2050629690879022, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8113521933555603, + "num_tokens": 661328705.0, + "step": 17334 + }, + { + "epoch": 2.2051901793664928, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8069900274276733, + "num_tokens": 661369877.0, + "step": 17335 + }, + { + "epoch": 2.2053173896450833, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.808517575263977, + "num_tokens": 661405209.0, + "step": 17336 + }, + { + "epoch": 2.205444599923674, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8071613311767578, + "num_tokens": 661441083.0, + "step": 17337 + }, + { + "epoch": 2.2055718102022643, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8037902116775513, + "num_tokens": 661476938.0, + "step": 17338 + }, + { + "epoch": 2.205699020480855, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7943441867828369, + "num_tokens": 661519281.0, + "step": 17339 + }, + { + "epoch": 2.2058262307594454, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.5904, + "mean_token_accuracy": 0.8201422691345215, + "num_tokens": 661555880.0, + "step": 17340 + }, + { + "epoch": 2.205953441038036, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7968812584877014, + "num_tokens": 661595937.0, + "step": 17341 + }, + { + "epoch": 2.2060806513166265, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7851017713546753, + "num_tokens": 661636868.0, + "step": 17342 + }, + { + "epoch": 2.206207861595217, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7908893823623657, + "num_tokens": 661681503.0, + "step": 17343 + }, + { + "epoch": 2.2063350718738075, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8061375021934509, + "num_tokens": 661718438.0, + "step": 17344 + }, + { + "epoch": 2.206462282152398, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7789973020553589, + "num_tokens": 661759048.0, + "step": 17345 + }, + { + "epoch": 2.2065894924309886, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.8128873109817505, + "num_tokens": 661803131.0, + "step": 17346 + }, + { + "epoch": 2.206716702709579, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7986787557601929, + "num_tokens": 661838139.0, + "step": 17347 + }, + { + "epoch": 2.2068439129881696, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7956643104553223, + "num_tokens": 661878568.0, + "step": 17348 + }, + { + "epoch": 2.20697112326676, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7917413711547852, + "num_tokens": 661909110.0, + "step": 17349 + }, + { + "epoch": 2.2070983335453507, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8163340091705322, + "num_tokens": 661944472.0, + "step": 17350 + }, + { + "epoch": 2.2072255438239408, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8167079091072083, + "num_tokens": 661985865.0, + "step": 17351 + }, + { + "epoch": 2.2073527541025313, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7962493300437927, + "num_tokens": 662019460.0, + "step": 17352 + }, + { + "epoch": 2.207479964381122, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.8052780628204346, + "num_tokens": 662059109.0, + "step": 17353 + }, + { + "epoch": 2.2076071746597123, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7919329404830933, + "num_tokens": 662096040.0, + "step": 17354 + }, + { + "epoch": 2.207734384938303, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8112317323684692, + "num_tokens": 662139351.0, + "step": 17355 + }, + { + "epoch": 2.2078615952168934, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8048238158226013, + "num_tokens": 662177353.0, + "step": 17356 + }, + { + "epoch": 2.207988805495484, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7830405235290527, + "num_tokens": 662209519.0, + "step": 17357 + }, + { + "epoch": 2.2081160157740745, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7820131778717041, + "num_tokens": 662251311.0, + "step": 17358 + }, + { + "epoch": 2.208243226052665, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8036149740219116, + "num_tokens": 662282770.0, + "step": 17359 + }, + { + "epoch": 2.2083704363312555, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.796178936958313, + "num_tokens": 662326770.0, + "step": 17360 + }, + { + "epoch": 2.208497646609846, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8074789047241211, + "num_tokens": 662366811.0, + "step": 17361 + }, + { + "epoch": 2.2086248568884366, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7964608669281006, + "num_tokens": 662404517.0, + "step": 17362 + }, + { + "epoch": 2.208752067167027, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8058212399482727, + "num_tokens": 662442091.0, + "step": 17363 + }, + { + "epoch": 2.2088792774456176, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7751892805099487, + "num_tokens": 662483343.0, + "step": 17364 + }, + { + "epoch": 2.209006487724208, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7948143482208252, + "num_tokens": 662521994.0, + "step": 17365 + }, + { + "epoch": 2.2091336980027987, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7974931597709656, + "num_tokens": 662562737.0, + "step": 17366 + }, + { + "epoch": 2.209260908281389, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7950171232223511, + "num_tokens": 662596927.0, + "step": 17367 + }, + { + "epoch": 2.2093881185599797, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8106440305709839, + "num_tokens": 662632831.0, + "step": 17368 + }, + { + "epoch": 2.2095153288385703, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7946027517318726, + "num_tokens": 662675554.0, + "step": 17369 + }, + { + "epoch": 2.209642539117161, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.8216280341148376, + "num_tokens": 662720248.0, + "step": 17370 + }, + { + "epoch": 2.2097697493957513, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.806023359298706, + "num_tokens": 662761388.0, + "step": 17371 + }, + { + "epoch": 2.209896959674342, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7945154905319214, + "num_tokens": 662798007.0, + "step": 17372 + }, + { + "epoch": 2.2100241699529324, + "grad_norm": 83.0, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7965162992477417, + "num_tokens": 662836042.0, + "step": 17373 + }, + { + "epoch": 2.210151380231523, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7837918400764465, + "num_tokens": 662869465.0, + "step": 17374 + }, + { + "epoch": 2.2102785905101134, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7937769889831543, + "num_tokens": 662909440.0, + "step": 17375 + }, + { + "epoch": 2.2104058007887035, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7989768981933594, + "num_tokens": 662949637.0, + "step": 17376 + }, + { + "epoch": 2.210533011067294, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.797932505607605, + "num_tokens": 662987679.0, + "step": 17377 + }, + { + "epoch": 2.2106602213458846, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7830916047096252, + "num_tokens": 663027651.0, + "step": 17378 + }, + { + "epoch": 2.210787431624475, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.8215665221214294, + "num_tokens": 663063708.0, + "step": 17379 + }, + { + "epoch": 2.2109146419030656, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7824847102165222, + "num_tokens": 663097255.0, + "step": 17380 + }, + { + "epoch": 2.211041852181656, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7962117195129395, + "num_tokens": 663134496.0, + "step": 17381 + }, + { + "epoch": 2.2111690624602467, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8084173798561096, + "num_tokens": 663170278.0, + "step": 17382 + }, + { + "epoch": 2.211296272738837, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7972941398620605, + "num_tokens": 663205407.0, + "step": 17383 + }, + { + "epoch": 2.2114234830174277, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8044160604476929, + "num_tokens": 663243166.0, + "step": 17384 + }, + { + "epoch": 2.2115506932960183, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.5804, + "mean_token_accuracy": 0.8254821300506592, + "num_tokens": 663282267.0, + "step": 17385 + }, + { + "epoch": 2.211677903574609, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7938867211341858, + "num_tokens": 663318620.0, + "step": 17386 + }, + { + "epoch": 2.2118051138531993, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8141507506370544, + "num_tokens": 663353256.0, + "step": 17387 + }, + { + "epoch": 2.21193232413179, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8020249605178833, + "num_tokens": 663386691.0, + "step": 17388 + }, + { + "epoch": 2.2120595344103804, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7960894107818604, + "num_tokens": 663416550.0, + "step": 17389 + }, + { + "epoch": 2.212186744688971, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8065797090530396, + "num_tokens": 663464912.0, + "step": 17390 + }, + { + "epoch": 2.2123139549675614, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7852482795715332, + "num_tokens": 663503786.0, + "step": 17391 + }, + { + "epoch": 2.212441165246152, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8093103170394897, + "num_tokens": 663540579.0, + "step": 17392 + }, + { + "epoch": 2.2125683755247425, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8013861775398254, + "num_tokens": 663577284.0, + "step": 17393 + }, + { + "epoch": 2.212695585803333, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7906880974769592, + "num_tokens": 663613118.0, + "step": 17394 + }, + { + "epoch": 2.2128227960819236, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8020138740539551, + "num_tokens": 663651548.0, + "step": 17395 + }, + { + "epoch": 2.212950006360514, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7870121002197266, + "num_tokens": 663694381.0, + "step": 17396 + }, + { + "epoch": 2.2130772166391046, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8053513765335083, + "num_tokens": 663734686.0, + "step": 17397 + }, + { + "epoch": 2.213204426917695, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7899863719940186, + "num_tokens": 663768764.0, + "step": 17398 + }, + { + "epoch": 2.2133316371962852, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.82237708568573, + "num_tokens": 663807650.0, + "step": 17399 + }, + { + "epoch": 2.213458847474876, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7996532917022705, + "num_tokens": 663849308.0, + "step": 17400 + }, + { + "epoch": 2.2135860577534663, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7802240252494812, + "num_tokens": 663888917.0, + "step": 17401 + }, + { + "epoch": 2.213713268032057, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7956481575965881, + "num_tokens": 663921048.0, + "step": 17402 + }, + { + "epoch": 2.2138404783106473, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8049869537353516, + "num_tokens": 663963433.0, + "step": 17403 + }, + { + "epoch": 2.213967688589238, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7856234312057495, + "num_tokens": 664003294.0, + "step": 17404 + }, + { + "epoch": 2.2140948988678284, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7933778166770935, + "num_tokens": 664043161.0, + "step": 17405 + }, + { + "epoch": 2.214222109146419, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8131792545318604, + "num_tokens": 664082457.0, + "step": 17406 + }, + { + "epoch": 2.2143493194250095, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.8187757730484009, + "num_tokens": 664125999.0, + "step": 17407 + }, + { + "epoch": 2.2144765297036, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8046001195907593, + "num_tokens": 664167256.0, + "step": 17408 + }, + { + "epoch": 2.2146037399821905, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7955825328826904, + "num_tokens": 664201887.0, + "step": 17409 + }, + { + "epoch": 2.214730950260781, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8041568398475647, + "num_tokens": 664238986.0, + "step": 17410 + }, + { + "epoch": 2.2148581605393716, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7982077598571777, + "num_tokens": 664283436.0, + "step": 17411 + }, + { + "epoch": 2.214985370817962, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7957763075828552, + "num_tokens": 664325196.0, + "step": 17412 + }, + { + "epoch": 2.2151125810965526, + "grad_norm": 81.5, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7922536134719849, + "num_tokens": 664365722.0, + "step": 17413 + }, + { + "epoch": 2.215239791375143, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8001664876937866, + "num_tokens": 664411570.0, + "step": 17414 + }, + { + "epoch": 2.2153670016537337, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8062573075294495, + "num_tokens": 664453545.0, + "step": 17415 + }, + { + "epoch": 2.215494211932324, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8041089773178101, + "num_tokens": 664489882.0, + "step": 17416 + }, + { + "epoch": 2.2156214222109147, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8018062114715576, + "num_tokens": 664528720.0, + "step": 17417 + }, + { + "epoch": 2.2157486324895053, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7901990413665771, + "num_tokens": 664569645.0, + "step": 17418 + }, + { + "epoch": 2.215875842768096, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7988340854644775, + "num_tokens": 664612033.0, + "step": 17419 + }, + { + "epoch": 2.2160030530466863, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8019554615020752, + "num_tokens": 664651408.0, + "step": 17420 + }, + { + "epoch": 2.216130263325277, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7906451225280762, + "num_tokens": 664695764.0, + "step": 17421 + }, + { + "epoch": 2.2162574736038674, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8028855323791504, + "num_tokens": 664735528.0, + "step": 17422 + }, + { + "epoch": 2.216384683882458, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7902160286903381, + "num_tokens": 664775612.0, + "step": 17423 + }, + { + "epoch": 2.216511894161048, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7962162494659424, + "num_tokens": 664812485.0, + "step": 17424 + }, + { + "epoch": 2.2166391044396385, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7856950163841248, + "num_tokens": 664853166.0, + "step": 17425 + }, + { + "epoch": 2.216766314718229, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7854305505752563, + "num_tokens": 664888778.0, + "step": 17426 + }, + { + "epoch": 2.2168935249968196, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.8012003898620605, + "num_tokens": 664933783.0, + "step": 17427 + }, + { + "epoch": 2.21702073527541, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7769995927810669, + "num_tokens": 664965228.0, + "step": 17428 + }, + { + "epoch": 2.2171479455540006, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7937694787979126, + "num_tokens": 664996691.0, + "step": 17429 + }, + { + "epoch": 2.217275155832591, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8111797571182251, + "num_tokens": 665032006.0, + "step": 17430 + }, + { + "epoch": 2.2174023661111817, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7862704992294312, + "num_tokens": 665071819.0, + "step": 17431 + }, + { + "epoch": 2.217529576389772, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7862032651901245, + "num_tokens": 665118883.0, + "step": 17432 + }, + { + "epoch": 2.2176567866683627, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7718599438667297, + "num_tokens": 665154926.0, + "step": 17433 + }, + { + "epoch": 2.2177839969469533, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8052279949188232, + "num_tokens": 665189910.0, + "step": 17434 + }, + { + "epoch": 2.217911207225544, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.807090163230896, + "num_tokens": 665235849.0, + "step": 17435 + }, + { + "epoch": 2.2180384175041343, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7898677587509155, + "num_tokens": 665274231.0, + "step": 17436 + }, + { + "epoch": 2.218165627782725, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.5918, + "mean_token_accuracy": 0.8198285698890686, + "num_tokens": 665309057.0, + "step": 17437 + }, + { + "epoch": 2.2182928380613154, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7858610153198242, + "num_tokens": 665337031.0, + "step": 17438 + }, + { + "epoch": 2.218420048339906, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7981793880462646, + "num_tokens": 665376557.0, + "step": 17439 + }, + { + "epoch": 2.2185472586184964, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8016570210456848, + "num_tokens": 665419928.0, + "step": 17440 + }, + { + "epoch": 2.218674468897087, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8018945455551147, + "num_tokens": 665458827.0, + "step": 17441 + }, + { + "epoch": 2.2188016791756775, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8030050992965698, + "num_tokens": 665498623.0, + "step": 17442 + }, + { + "epoch": 2.218928889454268, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7905835509300232, + "num_tokens": 665536103.0, + "step": 17443 + }, + { + "epoch": 2.2190560997328586, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7933809161186218, + "num_tokens": 665572089.0, + "step": 17444 + }, + { + "epoch": 2.219183310011449, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7871724367141724, + "num_tokens": 665608088.0, + "step": 17445 + }, + { + "epoch": 2.2193105202900396, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.80710369348526, + "num_tokens": 665641883.0, + "step": 17446 + }, + { + "epoch": 2.21943773056863, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8071298003196716, + "num_tokens": 665682211.0, + "step": 17447 + }, + { + "epoch": 2.2195649408472207, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8012126684188843, + "num_tokens": 665726182.0, + "step": 17448 + }, + { + "epoch": 2.2196921511258108, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8070495128631592, + "num_tokens": 665760404.0, + "step": 17449 + }, + { + "epoch": 2.2198193614044013, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7952156066894531, + "num_tokens": 665798235.0, + "step": 17450 + }, + { + "epoch": 2.219946571682992, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8045780062675476, + "num_tokens": 665840118.0, + "step": 17451 + }, + { + "epoch": 2.2200737819615823, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.798051118850708, + "num_tokens": 665878498.0, + "step": 17452 + }, + { + "epoch": 2.220200992240173, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7952494025230408, + "num_tokens": 665924179.0, + "step": 17453 + }, + { + "epoch": 2.2203282025187634, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.8160803914070129, + "num_tokens": 665963160.0, + "step": 17454 + }, + { + "epoch": 2.220455412797354, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8100229501724243, + "num_tokens": 665998854.0, + "step": 17455 + }, + { + "epoch": 2.2205826230759445, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8092598915100098, + "num_tokens": 666036420.0, + "step": 17456 + }, + { + "epoch": 2.220709833354535, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7797355055809021, + "num_tokens": 666076692.0, + "step": 17457 + }, + { + "epoch": 2.2208370436331255, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7857458591461182, + "num_tokens": 666118009.0, + "step": 17458 + }, + { + "epoch": 2.220964253911716, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7885866761207581, + "num_tokens": 666158132.0, + "step": 17459 + }, + { + "epoch": 2.2210914641903066, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7921236157417297, + "num_tokens": 666187195.0, + "step": 17460 + }, + { + "epoch": 2.221218674468897, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8068071603775024, + "num_tokens": 666227790.0, + "step": 17461 + }, + { + "epoch": 2.2213458847474876, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7932450175285339, + "num_tokens": 666268517.0, + "step": 17462 + }, + { + "epoch": 2.221473095026078, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.5869, + "mean_token_accuracy": 0.8211824893951416, + "num_tokens": 666313727.0, + "step": 17463 + }, + { + "epoch": 2.2216003053046687, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8020098209381104, + "num_tokens": 666352713.0, + "step": 17464 + }, + { + "epoch": 2.221727515583259, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.811834990978241, + "num_tokens": 666395065.0, + "step": 17465 + }, + { + "epoch": 2.2218547258618497, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8000687956809998, + "num_tokens": 666428019.0, + "step": 17466 + }, + { + "epoch": 2.2219819361404403, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.8016842603683472, + "num_tokens": 666466830.0, + "step": 17467 + }, + { + "epoch": 2.222109146419031, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8157974481582642, + "num_tokens": 666501028.0, + "step": 17468 + }, + { + "epoch": 2.2222363566976213, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8116825819015503, + "num_tokens": 666538554.0, + "step": 17469 + }, + { + "epoch": 2.222363566976212, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7947711944580078, + "num_tokens": 666577019.0, + "step": 17470 + }, + { + "epoch": 2.2224907772548024, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8192793130874634, + "num_tokens": 666616597.0, + "step": 17471 + }, + { + "epoch": 2.222617987533393, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8019486665725708, + "num_tokens": 666651523.0, + "step": 17472 + }, + { + "epoch": 2.2227451978119834, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7892460823059082, + "num_tokens": 666689270.0, + "step": 17473 + }, + { + "epoch": 2.2228724080905735, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7973679304122925, + "num_tokens": 666726836.0, + "step": 17474 + }, + { + "epoch": 2.222999618369164, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7937006950378418, + "num_tokens": 666761207.0, + "step": 17475 + }, + { + "epoch": 2.2231268286477546, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.8094814419746399, + "num_tokens": 666800293.0, + "step": 17476 + }, + { + "epoch": 2.223254038926345, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.8147521018981934, + "num_tokens": 666843046.0, + "step": 17477 + }, + { + "epoch": 2.2233812492049356, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8144904375076294, + "num_tokens": 666885840.0, + "step": 17478 + }, + { + "epoch": 2.223508459483526, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7995582222938538, + "num_tokens": 666925654.0, + "step": 17479 + }, + { + "epoch": 2.2236356697621167, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8009607791900635, + "num_tokens": 666968098.0, + "step": 17480 + }, + { + "epoch": 2.223762880040707, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8127003908157349, + "num_tokens": 667010349.0, + "step": 17481 + }, + { + "epoch": 2.2238900903192977, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7874929904937744, + "num_tokens": 667047086.0, + "step": 17482 + }, + { + "epoch": 2.2240173005978883, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7827798128128052, + "num_tokens": 667083955.0, + "step": 17483 + }, + { + "epoch": 2.224144510876479, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8037638664245605, + "num_tokens": 667119751.0, + "step": 17484 + }, + { + "epoch": 2.2242717211550693, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7895581722259521, + "num_tokens": 667150980.0, + "step": 17485 + }, + { + "epoch": 2.22439893143366, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7949706315994263, + "num_tokens": 667185198.0, + "step": 17486 + }, + { + "epoch": 2.2245261417122504, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7993526458740234, + "num_tokens": 667225286.0, + "step": 17487 + }, + { + "epoch": 2.224653351990841, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7937416434288025, + "num_tokens": 667266294.0, + "step": 17488 + }, + { + "epoch": 2.2247805622694314, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8023195266723633, + "num_tokens": 667302430.0, + "step": 17489 + }, + { + "epoch": 2.224907772548022, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7914667129516602, + "num_tokens": 667342881.0, + "step": 17490 + }, + { + "epoch": 2.2250349828266125, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7965521216392517, + "num_tokens": 667376122.0, + "step": 17491 + }, + { + "epoch": 2.225162193105203, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7934734225273132, + "num_tokens": 667415120.0, + "step": 17492 + }, + { + "epoch": 2.2252894033837936, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7978581190109253, + "num_tokens": 667451014.0, + "step": 17493 + }, + { + "epoch": 2.225416613662384, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7852921485900879, + "num_tokens": 667485600.0, + "step": 17494 + }, + { + "epoch": 2.2255438239409746, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7991775870323181, + "num_tokens": 667527561.0, + "step": 17495 + }, + { + "epoch": 2.225671034219565, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8086912631988525, + "num_tokens": 667569677.0, + "step": 17496 + }, + { + "epoch": 2.225798244498155, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7947045564651489, + "num_tokens": 667605954.0, + "step": 17497 + }, + { + "epoch": 2.225925454776746, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8013107776641846, + "num_tokens": 667645082.0, + "step": 17498 + }, + { + "epoch": 2.2260526650553363, + "grad_norm": 67.5, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7993082404136658, + "num_tokens": 667681390.0, + "step": 17499 + }, + { + "epoch": 2.226179875333927, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7914246916770935, + "num_tokens": 667722772.0, + "step": 17500 + }, + { + "epoch": 2.2263070856125173, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7709167003631592, + "num_tokens": 667758647.0, + "step": 17501 + }, + { + "epoch": 2.226434295891108, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7968361377716064, + "num_tokens": 667790625.0, + "step": 17502 + }, + { + "epoch": 2.2265615061696984, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7862867116928101, + "num_tokens": 667823724.0, + "step": 17503 + }, + { + "epoch": 2.226688716448289, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8143460750579834, + "num_tokens": 667862548.0, + "step": 17504 + }, + { + "epoch": 2.2268159267268794, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7907939553260803, + "num_tokens": 667905994.0, + "step": 17505 + }, + { + "epoch": 2.22694313700547, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.797385573387146, + "num_tokens": 667945087.0, + "step": 17506 + }, + { + "epoch": 2.2270703472840605, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8152920603752136, + "num_tokens": 667984736.0, + "step": 17507 + }, + { + "epoch": 2.227197557562651, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7940089702606201, + "num_tokens": 668024020.0, + "step": 17508 + }, + { + "epoch": 2.2273247678412416, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7919466495513916, + "num_tokens": 668059567.0, + "step": 17509 + }, + { + "epoch": 2.227451978119832, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8077511191368103, + "num_tokens": 668099901.0, + "step": 17510 + }, + { + "epoch": 2.2275791883984226, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.8138952851295471, + "num_tokens": 668135974.0, + "step": 17511 + }, + { + "epoch": 2.227706398677013, + "grad_norm": 72.5, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8014878630638123, + "num_tokens": 668179200.0, + "step": 17512 + }, + { + "epoch": 2.2278336089556037, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7776209115982056, + "num_tokens": 668219683.0, + "step": 17513 + }, + { + "epoch": 2.227960819234194, + "grad_norm": 54.5, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7968770861625671, + "num_tokens": 668253899.0, + "step": 17514 + }, + { + "epoch": 2.2280880295127847, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7916560769081116, + "num_tokens": 668294140.0, + "step": 17515 + }, + { + "epoch": 2.2282152397913753, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8043333292007446, + "num_tokens": 668328442.0, + "step": 17516 + }, + { + "epoch": 2.228342450069966, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7983646392822266, + "num_tokens": 668367903.0, + "step": 17517 + }, + { + "epoch": 2.2284696603485563, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.8149743676185608, + "num_tokens": 668409889.0, + "step": 17518 + }, + { + "epoch": 2.228596870627147, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7995094656944275, + "num_tokens": 668451061.0, + "step": 17519 + }, + { + "epoch": 2.2287240809057374, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7795020341873169, + "num_tokens": 668493038.0, + "step": 17520 + }, + { + "epoch": 2.228851291184328, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.806901216506958, + "num_tokens": 668528087.0, + "step": 17521 + }, + { + "epoch": 2.228978501462918, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8010043501853943, + "num_tokens": 668562325.0, + "step": 17522 + }, + { + "epoch": 2.2291057117415085, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.80402672290802, + "num_tokens": 668601336.0, + "step": 17523 + }, + { + "epoch": 2.229232922020099, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8171560764312744, + "num_tokens": 668637113.0, + "step": 17524 + }, + { + "epoch": 2.2293601322986896, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7945053577423096, + "num_tokens": 668675645.0, + "step": 17525 + }, + { + "epoch": 2.22948734257728, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7956350445747375, + "num_tokens": 668706480.0, + "step": 17526 + }, + { + "epoch": 2.2296145528558706, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7962628602981567, + "num_tokens": 668743956.0, + "step": 17527 + }, + { + "epoch": 2.229741763134461, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8052036762237549, + "num_tokens": 668779134.0, + "step": 17528 + }, + { + "epoch": 2.2298689734130517, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8052241802215576, + "num_tokens": 668814978.0, + "step": 17529 + }, + { + "epoch": 2.229996183691642, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7815431356430054, + "num_tokens": 668854194.0, + "step": 17530 + }, + { + "epoch": 2.2301233939702327, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7963590025901794, + "num_tokens": 668893095.0, + "step": 17531 + }, + { + "epoch": 2.2302506042488233, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8083606958389282, + "num_tokens": 668930626.0, + "step": 17532 + }, + { + "epoch": 2.230377814527414, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8078234791755676, + "num_tokens": 668960941.0, + "step": 17533 + }, + { + "epoch": 2.2305050248060043, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7983895540237427, + "num_tokens": 668997771.0, + "step": 17534 + }, + { + "epoch": 2.230632235084595, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8159351348876953, + "num_tokens": 669038714.0, + "step": 17535 + }, + { + "epoch": 2.2307594453631854, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8040796518325806, + "num_tokens": 669072336.0, + "step": 17536 + }, + { + "epoch": 2.230886655641776, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8106144666671753, + "num_tokens": 669110303.0, + "step": 17537 + }, + { + "epoch": 2.2310138659203664, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7839971780776978, + "num_tokens": 669146874.0, + "step": 17538 + }, + { + "epoch": 2.231141076198957, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8098845481872559, + "num_tokens": 669181682.0, + "step": 17539 + }, + { + "epoch": 2.2312682864775475, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.8028431534767151, + "num_tokens": 669227201.0, + "step": 17540 + }, + { + "epoch": 2.231395496756138, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8076807856559753, + "num_tokens": 669265915.0, + "step": 17541 + }, + { + "epoch": 2.2315227070347285, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.806596040725708, + "num_tokens": 669304130.0, + "step": 17542 + }, + { + "epoch": 2.231649917313319, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7936084270477295, + "num_tokens": 669344225.0, + "step": 17543 + }, + { + "epoch": 2.2317771275919096, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7807111144065857, + "num_tokens": 669385126.0, + "step": 17544 + }, + { + "epoch": 2.2319043378705, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7887781858444214, + "num_tokens": 669425774.0, + "step": 17545 + }, + { + "epoch": 2.2320315481490907, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7895075678825378, + "num_tokens": 669459789.0, + "step": 17546 + }, + { + "epoch": 2.2321587584276807, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.8189303278923035, + "num_tokens": 669498709.0, + "step": 17547 + }, + { + "epoch": 2.2322859687062713, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8033555150032043, + "num_tokens": 669536021.0, + "step": 17548 + }, + { + "epoch": 2.232413178984862, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7831524014472961, + "num_tokens": 669574316.0, + "step": 17549 + }, + { + "epoch": 2.2325403892634523, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7889550924301147, + "num_tokens": 669611826.0, + "step": 17550 + }, + { + "epoch": 2.232667599542043, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8004468679428101, + "num_tokens": 669655277.0, + "step": 17551 + }, + { + "epoch": 2.2327948098206334, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7931808233261108, + "num_tokens": 669693481.0, + "step": 17552 + }, + { + "epoch": 2.232922020099224, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7905315160751343, + "num_tokens": 669724104.0, + "step": 17553 + }, + { + "epoch": 2.2330492303778144, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.783205509185791, + "num_tokens": 669768725.0, + "step": 17554 + }, + { + "epoch": 2.233176440656405, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8068736791610718, + "num_tokens": 669809508.0, + "step": 17555 + }, + { + "epoch": 2.2333036509349955, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7994140982627869, + "num_tokens": 669848047.0, + "step": 17556 + }, + { + "epoch": 2.233430861213586, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8155457973480225, + "num_tokens": 669881274.0, + "step": 17557 + }, + { + "epoch": 2.2335580714921766, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7931384444236755, + "num_tokens": 669918517.0, + "step": 17558 + }, + { + "epoch": 2.233685281770767, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7954723834991455, + "num_tokens": 669960990.0, + "step": 17559 + }, + { + "epoch": 2.2338124920493576, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7986379265785217, + "num_tokens": 669995542.0, + "step": 17560 + }, + { + "epoch": 2.233939702327948, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8023910522460938, + "num_tokens": 670036798.0, + "step": 17561 + }, + { + "epoch": 2.2340669126065387, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7971596717834473, + "num_tokens": 670071763.0, + "step": 17562 + }, + { + "epoch": 2.234194122885129, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7984577417373657, + "num_tokens": 670100864.0, + "step": 17563 + }, + { + "epoch": 2.2343213331637197, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7900297045707703, + "num_tokens": 670144114.0, + "step": 17564 + }, + { + "epoch": 2.2344485434423103, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8047716617584229, + "num_tokens": 670182719.0, + "step": 17565 + }, + { + "epoch": 2.234575753720901, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8025325536727905, + "num_tokens": 670218059.0, + "step": 17566 + }, + { + "epoch": 2.2347029639994913, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7989768981933594, + "num_tokens": 670259589.0, + "step": 17567 + }, + { + "epoch": 2.234830174278082, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.793358564376831, + "num_tokens": 670301755.0, + "step": 17568 + }, + { + "epoch": 2.2349573845566724, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8025214672088623, + "num_tokens": 670339205.0, + "step": 17569 + }, + { + "epoch": 2.235084594835263, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7990064024925232, + "num_tokens": 670382961.0, + "step": 17570 + }, + { + "epoch": 2.2352118051138534, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7852931022644043, + "num_tokens": 670420747.0, + "step": 17571 + }, + { + "epoch": 2.2353390153924435, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.5813, + "mean_token_accuracy": 0.8219913244247437, + "num_tokens": 670457906.0, + "step": 17572 + }, + { + "epoch": 2.235466225671034, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7865323424339294, + "num_tokens": 670493187.0, + "step": 17573 + }, + { + "epoch": 2.2355934359496246, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8105762004852295, + "num_tokens": 670529750.0, + "step": 17574 + }, + { + "epoch": 2.235720646228215, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7943600416183472, + "num_tokens": 670570147.0, + "step": 17575 + }, + { + "epoch": 2.2358478565068056, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.799307107925415, + "num_tokens": 670604113.0, + "step": 17576 + }, + { + "epoch": 2.235975066785396, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7991597652435303, + "num_tokens": 670645451.0, + "step": 17577 + }, + { + "epoch": 2.2361022770639867, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8085001707077026, + "num_tokens": 670684055.0, + "step": 17578 + }, + { + "epoch": 2.236229487342577, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.785897970199585, + "num_tokens": 670724397.0, + "step": 17579 + }, + { + "epoch": 2.2363566976211677, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8161945343017578, + "num_tokens": 670761282.0, + "step": 17580 + }, + { + "epoch": 2.2364839078997583, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8036990761756897, + "num_tokens": 670800007.0, + "step": 17581 + }, + { + "epoch": 2.236611118178349, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7970107793807983, + "num_tokens": 670840040.0, + "step": 17582 + }, + { + "epoch": 2.2367383284569393, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7925119996070862, + "num_tokens": 670881193.0, + "step": 17583 + }, + { + "epoch": 2.23686553873553, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8110358715057373, + "num_tokens": 670927275.0, + "step": 17584 + }, + { + "epoch": 2.2369927490141204, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.801276683807373, + "num_tokens": 670967327.0, + "step": 17585 + }, + { + "epoch": 2.237119959292711, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8055806159973145, + "num_tokens": 671009250.0, + "step": 17586 + }, + { + "epoch": 2.2372471695713014, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8056178092956543, + "num_tokens": 671045068.0, + "step": 17587 + }, + { + "epoch": 2.237374379849892, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7872545719146729, + "num_tokens": 671084786.0, + "step": 17588 + }, + { + "epoch": 2.2375015901284825, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7932047843933105, + "num_tokens": 671123235.0, + "step": 17589 + }, + { + "epoch": 2.237628800407073, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8170125484466553, + "num_tokens": 671161702.0, + "step": 17590 + }, + { + "epoch": 2.2377560106856635, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8079158067703247, + "num_tokens": 671195487.0, + "step": 17591 + }, + { + "epoch": 2.237883220964254, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8006735444068909, + "num_tokens": 671232288.0, + "step": 17592 + }, + { + "epoch": 2.2380104312428446, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7851573824882507, + "num_tokens": 671275154.0, + "step": 17593 + }, + { + "epoch": 2.238137641521435, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7922382950782776, + "num_tokens": 671312266.0, + "step": 17594 + }, + { + "epoch": 2.238264851800025, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8047078847885132, + "num_tokens": 671349690.0, + "step": 17595 + }, + { + "epoch": 2.238392062078616, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8023638725280762, + "num_tokens": 671387898.0, + "step": 17596 + }, + { + "epoch": 2.2385192723572063, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7744684219360352, + "num_tokens": 671428054.0, + "step": 17597 + }, + { + "epoch": 2.238646482635797, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8058739900588989, + "num_tokens": 671467096.0, + "step": 17598 + }, + { + "epoch": 2.2387736929143873, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8022409677505493, + "num_tokens": 671506752.0, + "step": 17599 + }, + { + "epoch": 2.238900903192978, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7954294681549072, + "num_tokens": 671542403.0, + "step": 17600 + }, + { + "epoch": 2.2390281134715684, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8097438216209412, + "num_tokens": 671578115.0, + "step": 17601 + }, + { + "epoch": 2.239155323750159, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8137136697769165, + "num_tokens": 671619854.0, + "step": 17602 + }, + { + "epoch": 2.2392825340287494, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8122260570526123, + "num_tokens": 671661617.0, + "step": 17603 + }, + { + "epoch": 2.23940974430734, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7914432883262634, + "num_tokens": 671702231.0, + "step": 17604 + }, + { + "epoch": 2.2395369545859305, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7959674000740051, + "num_tokens": 671741483.0, + "step": 17605 + }, + { + "epoch": 2.239664164864521, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.796520471572876, + "num_tokens": 671780168.0, + "step": 17606 + }, + { + "epoch": 2.2397913751431116, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7728146314620972, + "num_tokens": 671818980.0, + "step": 17607 + }, + { + "epoch": 2.239918585421702, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7922700643539429, + "num_tokens": 671850010.0, + "step": 17608 + }, + { + "epoch": 2.2400457957002926, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7945473790168762, + "num_tokens": 671888506.0, + "step": 17609 + }, + { + "epoch": 2.240173005978883, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8064312934875488, + "num_tokens": 671930965.0, + "step": 17610 + }, + { + "epoch": 2.2403002162574737, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8094106316566467, + "num_tokens": 671967472.0, + "step": 17611 + }, + { + "epoch": 2.240427426536064, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8234944939613342, + "num_tokens": 672005986.0, + "step": 17612 + }, + { + "epoch": 2.2405546368146547, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7979933023452759, + "num_tokens": 672044729.0, + "step": 17613 + }, + { + "epoch": 2.2406818470932452, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8002352714538574, + "num_tokens": 672080141.0, + "step": 17614 + }, + { + "epoch": 2.2408090573718358, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8054056167602539, + "num_tokens": 672119277.0, + "step": 17615 + }, + { + "epoch": 2.2409362676504263, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7967393398284912, + "num_tokens": 672154205.0, + "step": 17616 + }, + { + "epoch": 2.241063477929017, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8004137873649597, + "num_tokens": 672193667.0, + "step": 17617 + }, + { + "epoch": 2.2411906882076074, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7611608505249023, + "num_tokens": 672232360.0, + "step": 17618 + }, + { + "epoch": 2.241317898486198, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7891436219215393, + "num_tokens": 672264127.0, + "step": 17619 + }, + { + "epoch": 2.241445108764788, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8085886240005493, + "num_tokens": 672300719.0, + "step": 17620 + }, + { + "epoch": 2.2415723190433785, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.811691164970398, + "num_tokens": 672342796.0, + "step": 17621 + }, + { + "epoch": 2.241699529321969, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8054600954055786, + "num_tokens": 672382597.0, + "step": 17622 + }, + { + "epoch": 2.2418267396005596, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7875116467475891, + "num_tokens": 672427350.0, + "step": 17623 + }, + { + "epoch": 2.24195394987915, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7932178974151611, + "num_tokens": 672468997.0, + "step": 17624 + }, + { + "epoch": 2.2420811601577406, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8064694404602051, + "num_tokens": 672505061.0, + "step": 17625 + }, + { + "epoch": 2.242208370436331, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8014339208602905, + "num_tokens": 672544241.0, + "step": 17626 + }, + { + "epoch": 2.2423355807149217, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8027389049530029, + "num_tokens": 672586334.0, + "step": 17627 + }, + { + "epoch": 2.242462790993512, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7960684299468994, + "num_tokens": 672628627.0, + "step": 17628 + }, + { + "epoch": 2.2425900012721027, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7945072054862976, + "num_tokens": 672663046.0, + "step": 17629 + }, + { + "epoch": 2.2427172115506933, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8079469203948975, + "num_tokens": 672703394.0, + "step": 17630 + }, + { + "epoch": 2.242844421829284, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7886632680892944, + "num_tokens": 672743262.0, + "step": 17631 + }, + { + "epoch": 2.2429716321078743, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7633293867111206, + "num_tokens": 672782662.0, + "step": 17632 + }, + { + "epoch": 2.243098842386465, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7983974814414978, + "num_tokens": 672818400.0, + "step": 17633 + }, + { + "epoch": 2.2432260526650554, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8095229864120483, + "num_tokens": 672858843.0, + "step": 17634 + }, + { + "epoch": 2.243353262943646, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8066762089729309, + "num_tokens": 672900397.0, + "step": 17635 + }, + { + "epoch": 2.2434804732222364, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7952684164047241, + "num_tokens": 672936025.0, + "step": 17636 + }, + { + "epoch": 2.243607683500827, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.781914234161377, + "num_tokens": 672979180.0, + "step": 17637 + }, + { + "epoch": 2.2437348937794175, + "grad_norm": 10.625, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.8132075071334839, + "num_tokens": 673018053.0, + "step": 17638 + }, + { + "epoch": 2.243862104058008, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7756021022796631, + "num_tokens": 673057596.0, + "step": 17639 + }, + { + "epoch": 2.2439893143365985, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8085880279541016, + "num_tokens": 673097740.0, + "step": 17640 + }, + { + "epoch": 2.244116524615189, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.8081839680671692, + "num_tokens": 673141821.0, + "step": 17641 + }, + { + "epoch": 2.2442437348937796, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8116965293884277, + "num_tokens": 673184279.0, + "step": 17642 + }, + { + "epoch": 2.24437094517237, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8098406791687012, + "num_tokens": 673224328.0, + "step": 17643 + }, + { + "epoch": 2.2444981554509607, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.784357488155365, + "num_tokens": 673263328.0, + "step": 17644 + }, + { + "epoch": 2.2446253657295507, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8039810657501221, + "num_tokens": 673301663.0, + "step": 17645 + }, + { + "epoch": 2.2447525760081413, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7983713746070862, + "num_tokens": 673340606.0, + "step": 17646 + }, + { + "epoch": 2.244879786286732, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8013861179351807, + "num_tokens": 673382201.0, + "step": 17647 + }, + { + "epoch": 2.2450069965653223, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.8157087564468384, + "num_tokens": 673423330.0, + "step": 17648 + }, + { + "epoch": 2.245134206843913, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8141985535621643, + "num_tokens": 673459690.0, + "step": 17649 + }, + { + "epoch": 2.2452614171225034, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7984042167663574, + "num_tokens": 673494827.0, + "step": 17650 + }, + { + "epoch": 2.245388627401094, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7893663644790649, + "num_tokens": 673540568.0, + "step": 17651 + }, + { + "epoch": 2.2455158376796844, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8147541284561157, + "num_tokens": 673574022.0, + "step": 17652 + }, + { + "epoch": 2.245643047958275, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7969231605529785, + "num_tokens": 673612493.0, + "step": 17653 + }, + { + "epoch": 2.2457702582368655, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7979258894920349, + "num_tokens": 673653729.0, + "step": 17654 + }, + { + "epoch": 2.245897468515456, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8122667074203491, + "num_tokens": 673695626.0, + "step": 17655 + }, + { + "epoch": 2.2460246787940465, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7994603514671326, + "num_tokens": 673732188.0, + "step": 17656 + }, + { + "epoch": 2.246151889072637, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7928229570388794, + "num_tokens": 673770259.0, + "step": 17657 + }, + { + "epoch": 2.2462790993512276, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7934914231300354, + "num_tokens": 673805018.0, + "step": 17658 + }, + { + "epoch": 2.246406309629818, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7942041754722595, + "num_tokens": 673846058.0, + "step": 17659 + }, + { + "epoch": 2.2465335199084087, + "grad_norm": 58.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7911632061004639, + "num_tokens": 673886437.0, + "step": 17660 + }, + { + "epoch": 2.246660730186999, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8084614872932434, + "num_tokens": 673925701.0, + "step": 17661 + }, + { + "epoch": 2.2467879404655897, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8079754710197449, + "num_tokens": 673969854.0, + "step": 17662 + }, + { + "epoch": 2.2469151507441802, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8007441759109497, + "num_tokens": 674005048.0, + "step": 17663 + }, + { + "epoch": 2.2470423610227708, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7936476469039917, + "num_tokens": 674046548.0, + "step": 17664 + }, + { + "epoch": 2.2471695713013613, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7923463582992554, + "num_tokens": 674086905.0, + "step": 17665 + }, + { + "epoch": 2.247296781579952, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8082091212272644, + "num_tokens": 674124589.0, + "step": 17666 + }, + { + "epoch": 2.2474239918585424, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.815004825592041, + "num_tokens": 674165135.0, + "step": 17667 + }, + { + "epoch": 2.247551202137133, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.796006441116333, + "num_tokens": 674207269.0, + "step": 17668 + }, + { + "epoch": 2.2476784124157234, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.8220592737197876, + "num_tokens": 674240806.0, + "step": 17669 + }, + { + "epoch": 2.2478056226943135, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7841473817825317, + "num_tokens": 674274697.0, + "step": 17670 + }, + { + "epoch": 2.247932832972904, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7933120727539062, + "num_tokens": 674311362.0, + "step": 17671 + }, + { + "epoch": 2.2480600432514946, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8049837946891785, + "num_tokens": 674352505.0, + "step": 17672 + }, + { + "epoch": 2.248187253530085, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8029274940490723, + "num_tokens": 674383938.0, + "step": 17673 + }, + { + "epoch": 2.2483144638086756, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7935967445373535, + "num_tokens": 674421665.0, + "step": 17674 + }, + { + "epoch": 2.248441674087266, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8007889986038208, + "num_tokens": 674466065.0, + "step": 17675 + }, + { + "epoch": 2.2485688843658567, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8051121234893799, + "num_tokens": 674506868.0, + "step": 17676 + }, + { + "epoch": 2.248696094644447, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7938571572303772, + "num_tokens": 674545756.0, + "step": 17677 + }, + { + "epoch": 2.2488233049230377, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7982560992240906, + "num_tokens": 674590680.0, + "step": 17678 + }, + { + "epoch": 2.2489505152016283, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8103448152542114, + "num_tokens": 674631624.0, + "step": 17679 + }, + { + "epoch": 2.249077725480219, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7888550758361816, + "num_tokens": 674667365.0, + "step": 17680 + }, + { + "epoch": 2.2492049357588093, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7969269752502441, + "num_tokens": 674706587.0, + "step": 17681 + }, + { + "epoch": 2.2493321460374, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7956333160400391, + "num_tokens": 674744480.0, + "step": 17682 + }, + { + "epoch": 2.2494593563159904, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7981338500976562, + "num_tokens": 674783104.0, + "step": 17683 + }, + { + "epoch": 2.249586566594581, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7898898720741272, + "num_tokens": 674823488.0, + "step": 17684 + }, + { + "epoch": 2.2497137768731714, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8046445846557617, + "num_tokens": 674857322.0, + "step": 17685 + }, + { + "epoch": 2.249840987151762, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7906911373138428, + "num_tokens": 674901566.0, + "step": 17686 + }, + { + "epoch": 2.2499681974303525, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8060951828956604, + "num_tokens": 674942803.0, + "step": 17687 + }, + { + "epoch": 2.250095407708943, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7910012006759644, + "num_tokens": 674983160.0, + "step": 17688 + }, + { + "epoch": 2.2502226179875335, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8025258779525757, + "num_tokens": 675021243.0, + "step": 17689 + }, + { + "epoch": 2.250349828266124, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7940873503684998, + "num_tokens": 675059435.0, + "step": 17690 + }, + { + "epoch": 2.2504770385447146, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8034424781799316, + "num_tokens": 675091689.0, + "step": 17691 + }, + { + "epoch": 2.250604248823305, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7954380512237549, + "num_tokens": 675130368.0, + "step": 17692 + }, + { + "epoch": 2.250731459101895, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8062057495117188, + "num_tokens": 675171745.0, + "step": 17693 + }, + { + "epoch": 2.250858669380486, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7949131727218628, + "num_tokens": 675206052.0, + "step": 17694 + }, + { + "epoch": 2.2509858796590763, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7939646244049072, + "num_tokens": 675243231.0, + "step": 17695 + }, + { + "epoch": 2.251113089937667, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7863286733627319, + "num_tokens": 675273582.0, + "step": 17696 + }, + { + "epoch": 2.2512403002162573, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8006081581115723, + "num_tokens": 675314247.0, + "step": 17697 + }, + { + "epoch": 2.251367510494848, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.799915075302124, + "num_tokens": 675355675.0, + "step": 17698 + }, + { + "epoch": 2.2514947207734384, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7956854104995728, + "num_tokens": 675394290.0, + "step": 17699 + }, + { + "epoch": 2.251621931052029, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8019418716430664, + "num_tokens": 675434251.0, + "step": 17700 + }, + { + "epoch": 2.2517491413306194, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8101800084114075, + "num_tokens": 675474731.0, + "step": 17701 + }, + { + "epoch": 2.25187635160921, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7937142848968506, + "num_tokens": 675515180.0, + "step": 17702 + }, + { + "epoch": 2.2520035618878005, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7847120761871338, + "num_tokens": 675553429.0, + "step": 17703 + }, + { + "epoch": 2.252130772166391, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8045796155929565, + "num_tokens": 675592919.0, + "step": 17704 + }, + { + "epoch": 2.2522579824449815, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8081251978874207, + "num_tokens": 675627528.0, + "step": 17705 + }, + { + "epoch": 2.252385192723572, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.8081987500190735, + "num_tokens": 675663721.0, + "step": 17706 + }, + { + "epoch": 2.2525124030021626, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7905204892158508, + "num_tokens": 675705187.0, + "step": 17707 + }, + { + "epoch": 2.252639613280753, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8123878836631775, + "num_tokens": 675752713.0, + "step": 17708 + }, + { + "epoch": 2.2527668235593437, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7969812154769897, + "num_tokens": 675791009.0, + "step": 17709 + }, + { + "epoch": 2.252894033837934, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8012058734893799, + "num_tokens": 675828890.0, + "step": 17710 + }, + { + "epoch": 2.2530212441165247, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7939811944961548, + "num_tokens": 675862855.0, + "step": 17711 + }, + { + "epoch": 2.2531484543951152, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7973999381065369, + "num_tokens": 675903032.0, + "step": 17712 + }, + { + "epoch": 2.2532756646737058, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7922292947769165, + "num_tokens": 675943035.0, + "step": 17713 + }, + { + "epoch": 2.2534028749522963, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.803136944770813, + "num_tokens": 675983073.0, + "step": 17714 + }, + { + "epoch": 2.253530085230887, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8067159652709961, + "num_tokens": 676022341.0, + "step": 17715 + }, + { + "epoch": 2.2536572955094774, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.798498809337616, + "num_tokens": 676060907.0, + "step": 17716 + }, + { + "epoch": 2.253784505788068, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7968037128448486, + "num_tokens": 676100693.0, + "step": 17717 + }, + { + "epoch": 2.253911716066658, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8172612190246582, + "num_tokens": 676137526.0, + "step": 17718 + }, + { + "epoch": 2.254038926345249, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.806427001953125, + "num_tokens": 676177245.0, + "step": 17719 + }, + { + "epoch": 2.254166136623839, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8088696599006653, + "num_tokens": 676214568.0, + "step": 17720 + }, + { + "epoch": 2.2542933469024296, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7993464469909668, + "num_tokens": 676252552.0, + "step": 17721 + }, + { + "epoch": 2.25442055718102, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7846702337265015, + "num_tokens": 676290480.0, + "step": 17722 + }, + { + "epoch": 2.2545477674596106, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.807428777217865, + "num_tokens": 676324950.0, + "step": 17723 + }, + { + "epoch": 2.254674977738201, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7931111454963684, + "num_tokens": 676363851.0, + "step": 17724 + }, + { + "epoch": 2.2548021880167917, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8109692335128784, + "num_tokens": 676399920.0, + "step": 17725 + }, + { + "epoch": 2.254929398295382, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.8021272420883179, + "num_tokens": 676443243.0, + "step": 17726 + }, + { + "epoch": 2.2550566085739727, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7846391201019287, + "num_tokens": 676480272.0, + "step": 17727 + }, + { + "epoch": 2.2551838188525632, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7941846251487732, + "num_tokens": 676516630.0, + "step": 17728 + }, + { + "epoch": 2.2553110291311538, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8061878681182861, + "num_tokens": 676557842.0, + "step": 17729 + }, + { + "epoch": 2.2554382394097443, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8073099851608276, + "num_tokens": 676599413.0, + "step": 17730 + }, + { + "epoch": 2.255565449688335, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7940424680709839, + "num_tokens": 676637791.0, + "step": 17731 + }, + { + "epoch": 2.2556926599669254, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7918176054954529, + "num_tokens": 676677108.0, + "step": 17732 + }, + { + "epoch": 2.255819870245516, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8096909523010254, + "num_tokens": 676720698.0, + "step": 17733 + }, + { + "epoch": 2.2559470805241064, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7865217328071594, + "num_tokens": 676763719.0, + "step": 17734 + }, + { + "epoch": 2.256074290802697, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8049517869949341, + "num_tokens": 676801083.0, + "step": 17735 + }, + { + "epoch": 2.2562015010812875, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7915765047073364, + "num_tokens": 676840861.0, + "step": 17736 + }, + { + "epoch": 2.256328711359878, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8130682706832886, + "num_tokens": 676876864.0, + "step": 17737 + }, + { + "epoch": 2.2564559216384685, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.8214898109436035, + "num_tokens": 676920983.0, + "step": 17738 + }, + { + "epoch": 2.256583131917059, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8097517490386963, + "num_tokens": 676961294.0, + "step": 17739 + }, + { + "epoch": 2.2567103421956496, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7901507616043091, + "num_tokens": 676996318.0, + "step": 17740 + }, + { + "epoch": 2.2568375524742397, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8020342588424683, + "num_tokens": 677034083.0, + "step": 17741 + }, + { + "epoch": 2.2569647627528306, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7944157719612122, + "num_tokens": 677071206.0, + "step": 17742 + }, + { + "epoch": 2.2570919730314207, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.8021227121353149, + "num_tokens": 677103573.0, + "step": 17743 + }, + { + "epoch": 2.2572191833100113, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8057730197906494, + "num_tokens": 677141406.0, + "step": 17744 + }, + { + "epoch": 2.257346393588602, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7895505428314209, + "num_tokens": 677185968.0, + "step": 17745 + }, + { + "epoch": 2.2574736038671923, + "grad_norm": 64.5, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7906134128570557, + "num_tokens": 677229004.0, + "step": 17746 + }, + { + "epoch": 2.257600814145783, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8133788108825684, + "num_tokens": 677266317.0, + "step": 17747 + }, + { + "epoch": 2.2577280244243734, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8075367212295532, + "num_tokens": 677302344.0, + "step": 17748 + }, + { + "epoch": 2.257855234702964, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7947205305099487, + "num_tokens": 677342103.0, + "step": 17749 + }, + { + "epoch": 2.2579824449815544, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.782382607460022, + "num_tokens": 677378602.0, + "step": 17750 + }, + { + "epoch": 2.258109655260145, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7986527681350708, + "num_tokens": 677415005.0, + "step": 17751 + }, + { + "epoch": 2.2582368655387355, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7811599969863892, + "num_tokens": 677457908.0, + "step": 17752 + }, + { + "epoch": 2.258364075817326, + "grad_norm": 69.0, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7912006378173828, + "num_tokens": 677492759.0, + "step": 17753 + }, + { + "epoch": 2.2584912860959165, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7963894605636597, + "num_tokens": 677535771.0, + "step": 17754 + }, + { + "epoch": 2.258618496374507, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8110079765319824, + "num_tokens": 677575325.0, + "step": 17755 + }, + { + "epoch": 2.2587457066530976, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.797211229801178, + "num_tokens": 677614132.0, + "step": 17756 + }, + { + "epoch": 2.258872916931688, + "grad_norm": 51.75, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7900298833847046, + "num_tokens": 677645657.0, + "step": 17757 + }, + { + "epoch": 2.2590001272102787, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8104950189590454, + "num_tokens": 677686238.0, + "step": 17758 + }, + { + "epoch": 2.259127337488869, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8042153716087341, + "num_tokens": 677724080.0, + "step": 17759 + }, + { + "epoch": 2.2592545477674597, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8008387684822083, + "num_tokens": 677765768.0, + "step": 17760 + }, + { + "epoch": 2.2593817580460502, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8000521063804626, + "num_tokens": 677803937.0, + "step": 17761 + }, + { + "epoch": 2.2595089683246408, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.813130259513855, + "num_tokens": 677839685.0, + "step": 17762 + }, + { + "epoch": 2.2596361786032313, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7872651219367981, + "num_tokens": 677880411.0, + "step": 17763 + }, + { + "epoch": 2.259763388881822, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8103761076927185, + "num_tokens": 677921177.0, + "step": 17764 + }, + { + "epoch": 2.2598905991604123, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8111234903335571, + "num_tokens": 677956393.0, + "step": 17765 + }, + { + "epoch": 2.2600178094390024, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7957938313484192, + "num_tokens": 677988364.0, + "step": 17766 + }, + { + "epoch": 2.2601450197175934, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8103782534599304, + "num_tokens": 678028093.0, + "step": 17767 + }, + { + "epoch": 2.2602722299961835, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7976937294006348, + "num_tokens": 678068174.0, + "step": 17768 + }, + { + "epoch": 2.260399440274774, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7987140417098999, + "num_tokens": 678105784.0, + "step": 17769 + }, + { + "epoch": 2.2605266505533645, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8124953508377075, + "num_tokens": 678142873.0, + "step": 17770 + }, + { + "epoch": 2.260653860831955, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7967603802680969, + "num_tokens": 678181527.0, + "step": 17771 + }, + { + "epoch": 2.2607810711105456, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8058363199234009, + "num_tokens": 678221225.0, + "step": 17772 + }, + { + "epoch": 2.260908281389136, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8074485659599304, + "num_tokens": 678254767.0, + "step": 17773 + }, + { + "epoch": 2.2610354916677267, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8036826848983765, + "num_tokens": 678290830.0, + "step": 17774 + }, + { + "epoch": 2.261162701946317, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8102887868881226, + "num_tokens": 678327102.0, + "step": 17775 + }, + { + "epoch": 2.2612899122249077, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8131020665168762, + "num_tokens": 678364872.0, + "step": 17776 + }, + { + "epoch": 2.2614171225034982, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7891420722007751, + "num_tokens": 678399233.0, + "step": 17777 + }, + { + "epoch": 2.2615443327820888, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8125033974647522, + "num_tokens": 678434577.0, + "step": 17778 + }, + { + "epoch": 2.2616715430606793, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8047248125076294, + "num_tokens": 678466835.0, + "step": 17779 + }, + { + "epoch": 2.26179875333927, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.5871, + "mean_token_accuracy": 0.8201716542243958, + "num_tokens": 678506817.0, + "step": 17780 + }, + { + "epoch": 2.2619259636178604, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8003413677215576, + "num_tokens": 678543734.0, + "step": 17781 + }, + { + "epoch": 2.262053173896451, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7937207818031311, + "num_tokens": 678585066.0, + "step": 17782 + }, + { + "epoch": 2.2621803841750414, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8163753747940063, + "num_tokens": 678614951.0, + "step": 17783 + }, + { + "epoch": 2.262307594453632, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8041243553161621, + "num_tokens": 678649671.0, + "step": 17784 + }, + { + "epoch": 2.2624348047322225, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8190228939056396, + "num_tokens": 678693418.0, + "step": 17785 + }, + { + "epoch": 2.262562015010813, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8137984275817871, + "num_tokens": 678730891.0, + "step": 17786 + }, + { + "epoch": 2.2626892252894035, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8038997054100037, + "num_tokens": 678766369.0, + "step": 17787 + }, + { + "epoch": 2.262816435567994, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7920519113540649, + "num_tokens": 678809118.0, + "step": 17788 + }, + { + "epoch": 2.2629436458465846, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7999567985534668, + "num_tokens": 678850057.0, + "step": 17789 + }, + { + "epoch": 2.263070856125175, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8051708936691284, + "num_tokens": 678890253.0, + "step": 17790 + }, + { + "epoch": 2.263198066403765, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8147677183151245, + "num_tokens": 678922535.0, + "step": 17791 + }, + { + "epoch": 2.263325276682356, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7958647608757019, + "num_tokens": 678956136.0, + "step": 17792 + }, + { + "epoch": 2.2634524869609463, + "grad_norm": 62.75, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8013923168182373, + "num_tokens": 678990864.0, + "step": 17793 + }, + { + "epoch": 2.263579697239537, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8150752782821655, + "num_tokens": 679028345.0, + "step": 17794 + }, + { + "epoch": 2.2637069075181273, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7938247919082642, + "num_tokens": 679073679.0, + "step": 17795 + }, + { + "epoch": 2.263834117796718, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7944169044494629, + "num_tokens": 679109274.0, + "step": 17796 + }, + { + "epoch": 2.2639613280753084, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8034263849258423, + "num_tokens": 679141644.0, + "step": 17797 + }, + { + "epoch": 2.264088538353899, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7842175364494324, + "num_tokens": 679184351.0, + "step": 17798 + }, + { + "epoch": 2.2642157486324894, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7960891127586365, + "num_tokens": 679222509.0, + "step": 17799 + }, + { + "epoch": 2.26434295891108, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7851258516311646, + "num_tokens": 679263362.0, + "step": 17800 + }, + { + "epoch": 2.2644701691896705, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7812882661819458, + "num_tokens": 679307859.0, + "step": 17801 + }, + { + "epoch": 2.264597379468261, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8109295964241028, + "num_tokens": 679344801.0, + "step": 17802 + }, + { + "epoch": 2.2647245897468515, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7920957803726196, + "num_tokens": 679382795.0, + "step": 17803 + }, + { + "epoch": 2.264851800025442, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.8008474111557007, + "num_tokens": 679425282.0, + "step": 17804 + }, + { + "epoch": 2.2649790103040326, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7875993251800537, + "num_tokens": 679466881.0, + "step": 17805 + }, + { + "epoch": 2.265106220582623, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8153442144393921, + "num_tokens": 679509268.0, + "step": 17806 + }, + { + "epoch": 2.2652334308612136, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7960140109062195, + "num_tokens": 679545082.0, + "step": 17807 + }, + { + "epoch": 2.265360641139804, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.8013697266578674, + "num_tokens": 679581795.0, + "step": 17808 + }, + { + "epoch": 2.2654878514183947, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.8007122278213501, + "num_tokens": 679618978.0, + "step": 17809 + }, + { + "epoch": 2.2656150616969852, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8023585081100464, + "num_tokens": 679655002.0, + "step": 17810 + }, + { + "epoch": 2.2657422719755758, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8155450820922852, + "num_tokens": 679691143.0, + "step": 17811 + }, + { + "epoch": 2.2658694822541663, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8038902282714844, + "num_tokens": 679729168.0, + "step": 17812 + }, + { + "epoch": 2.265996692532757, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8024834394454956, + "num_tokens": 679773385.0, + "step": 17813 + }, + { + "epoch": 2.2661239028113473, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8151862621307373, + "num_tokens": 679813884.0, + "step": 17814 + }, + { + "epoch": 2.266251113089938, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8021537065505981, + "num_tokens": 679850834.0, + "step": 17815 + }, + { + "epoch": 2.266378323368528, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8050986528396606, + "num_tokens": 679890210.0, + "step": 17816 + }, + { + "epoch": 2.266505533647119, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8150089979171753, + "num_tokens": 679926252.0, + "step": 17817 + }, + { + "epoch": 2.266632743925709, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7947239875793457, + "num_tokens": 679959626.0, + "step": 17818 + }, + { + "epoch": 2.2667599542042995, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8026418089866638, + "num_tokens": 679999973.0, + "step": 17819 + }, + { + "epoch": 2.26688716448289, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7933911085128784, + "num_tokens": 680041081.0, + "step": 17820 + }, + { + "epoch": 2.2670143747614806, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8077259063720703, + "num_tokens": 680078953.0, + "step": 17821 + }, + { + "epoch": 2.267141585040071, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8058735132217407, + "num_tokens": 680112309.0, + "step": 17822 + }, + { + "epoch": 2.2672687953186617, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7931269407272339, + "num_tokens": 680148607.0, + "step": 17823 + }, + { + "epoch": 2.267396005597252, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8214579224586487, + "num_tokens": 680189617.0, + "step": 17824 + }, + { + "epoch": 2.2675232158758427, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8189144134521484, + "num_tokens": 680227940.0, + "step": 17825 + }, + { + "epoch": 2.2676504261544332, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8130868673324585, + "num_tokens": 680263473.0, + "step": 17826 + }, + { + "epoch": 2.2677776364330238, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7916069030761719, + "num_tokens": 680297895.0, + "step": 17827 + }, + { + "epoch": 2.2679048467116143, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8062738180160522, + "num_tokens": 680337658.0, + "step": 17828 + }, + { + "epoch": 2.268032056990205, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7983040809631348, + "num_tokens": 680375912.0, + "step": 17829 + }, + { + "epoch": 2.2681592672687954, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8103972673416138, + "num_tokens": 680415231.0, + "step": 17830 + }, + { + "epoch": 2.268286477547386, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7965115308761597, + "num_tokens": 680460037.0, + "step": 17831 + }, + { + "epoch": 2.2684136878259764, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.776889443397522, + "num_tokens": 680505521.0, + "step": 17832 + }, + { + "epoch": 2.268540898104567, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.8149301409721375, + "num_tokens": 680547682.0, + "step": 17833 + }, + { + "epoch": 2.2686681083831575, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8075587749481201, + "num_tokens": 680587510.0, + "step": 17834 + }, + { + "epoch": 2.268795318661748, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8058533072471619, + "num_tokens": 680624233.0, + "step": 17835 + }, + { + "epoch": 2.2689225289403385, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7984209656715393, + "num_tokens": 680666091.0, + "step": 17836 + }, + { + "epoch": 2.269049739218929, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8082253932952881, + "num_tokens": 680703062.0, + "step": 17837 + }, + { + "epoch": 2.2691769494975196, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.816713809967041, + "num_tokens": 680735966.0, + "step": 17838 + }, + { + "epoch": 2.2693041597761097, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.802548885345459, + "num_tokens": 680780010.0, + "step": 17839 + }, + { + "epoch": 2.2694313700547006, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8041965961456299, + "num_tokens": 680819516.0, + "step": 17840 + }, + { + "epoch": 2.2695585803332907, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8012218475341797, + "num_tokens": 680861374.0, + "step": 17841 + }, + { + "epoch": 2.2696857906118812, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7914366722106934, + "num_tokens": 680902447.0, + "step": 17842 + }, + { + "epoch": 2.2698130008904718, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8185113668441772, + "num_tokens": 680942376.0, + "step": 17843 + }, + { + "epoch": 2.2699402111690623, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7976691126823425, + "num_tokens": 680977060.0, + "step": 17844 + }, + { + "epoch": 2.270067421447653, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7827061414718628, + "num_tokens": 681014504.0, + "step": 17845 + }, + { + "epoch": 2.2701946317262434, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7835795879364014, + "num_tokens": 681053919.0, + "step": 17846 + }, + { + "epoch": 2.270321842004834, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8039597272872925, + "num_tokens": 681090883.0, + "step": 17847 + }, + { + "epoch": 2.2704490522834244, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8136030435562134, + "num_tokens": 681128819.0, + "step": 17848 + }, + { + "epoch": 2.270576262562015, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7929621934890747, + "num_tokens": 681171434.0, + "step": 17849 + }, + { + "epoch": 2.2707034728406055, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8179705142974854, + "num_tokens": 681206297.0, + "step": 17850 + }, + { + "epoch": 2.270830683119196, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7886688709259033, + "num_tokens": 681242521.0, + "step": 17851 + }, + { + "epoch": 2.2709578933977865, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8091828227043152, + "num_tokens": 681282474.0, + "step": 17852 + }, + { + "epoch": 2.271085103676377, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7993453741073608, + "num_tokens": 681317280.0, + "step": 17853 + }, + { + "epoch": 2.2712123139549676, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7956670522689819, + "num_tokens": 681349155.0, + "step": 17854 + }, + { + "epoch": 2.271339524233558, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7961220741271973, + "num_tokens": 681380863.0, + "step": 17855 + }, + { + "epoch": 2.2714667345121486, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8034639358520508, + "num_tokens": 681421970.0, + "step": 17856 + }, + { + "epoch": 2.271593944790739, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8164268732070923, + "num_tokens": 681453511.0, + "step": 17857 + }, + { + "epoch": 2.2717211550693297, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7885667085647583, + "num_tokens": 681496293.0, + "step": 17858 + }, + { + "epoch": 2.2718483653479202, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7917060256004333, + "num_tokens": 681537074.0, + "step": 17859 + }, + { + "epoch": 2.2719755756265108, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7962220907211304, + "num_tokens": 681578911.0, + "step": 17860 + }, + { + "epoch": 2.2721027859051013, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8019077777862549, + "num_tokens": 681622134.0, + "step": 17861 + }, + { + "epoch": 2.272229996183692, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.802374005317688, + "num_tokens": 681655919.0, + "step": 17862 + }, + { + "epoch": 2.2723572064622823, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.794181227684021, + "num_tokens": 681696539.0, + "step": 17863 + }, + { + "epoch": 2.2724844167408724, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.795164167881012, + "num_tokens": 681734630.0, + "step": 17864 + }, + { + "epoch": 2.2726116270194634, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7894570827484131, + "num_tokens": 681779669.0, + "step": 17865 + }, + { + "epoch": 2.2727388372980535, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.798890233039856, + "num_tokens": 681821920.0, + "step": 17866 + }, + { + "epoch": 2.272866047576644, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7981275320053101, + "num_tokens": 681859239.0, + "step": 17867 + }, + { + "epoch": 2.2729932578552345, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7822556495666504, + "num_tokens": 681896569.0, + "step": 17868 + }, + { + "epoch": 2.273120468133825, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8204869031906128, + "num_tokens": 681932549.0, + "step": 17869 + }, + { + "epoch": 2.2732476784124156, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7907670736312866, + "num_tokens": 681973981.0, + "step": 17870 + }, + { + "epoch": 2.273374888691006, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7779604196548462, + "num_tokens": 682009512.0, + "step": 17871 + }, + { + "epoch": 2.2735020989695967, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7966147661209106, + "num_tokens": 682048058.0, + "step": 17872 + }, + { + "epoch": 2.273629309248187, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7960742712020874, + "num_tokens": 682084181.0, + "step": 17873 + }, + { + "epoch": 2.2737565195267777, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.8216961622238159, + "num_tokens": 682119613.0, + "step": 17874 + }, + { + "epoch": 2.2738837298053682, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8004986047744751, + "num_tokens": 682159501.0, + "step": 17875 + }, + { + "epoch": 2.2740109400839588, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8131531476974487, + "num_tokens": 682197323.0, + "step": 17876 + }, + { + "epoch": 2.2741381503625493, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8024206757545471, + "num_tokens": 682231150.0, + "step": 17877 + }, + { + "epoch": 2.27426536064114, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8038221001625061, + "num_tokens": 682269351.0, + "step": 17878 + }, + { + "epoch": 2.2743925709197303, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8083710670471191, + "num_tokens": 682303633.0, + "step": 17879 + }, + { + "epoch": 2.274519781198321, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8020501136779785, + "num_tokens": 682340890.0, + "step": 17880 + }, + { + "epoch": 2.2746469914769114, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8186914920806885, + "num_tokens": 682374253.0, + "step": 17881 + }, + { + "epoch": 2.274774201755502, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8032055497169495, + "num_tokens": 682412675.0, + "step": 17882 + }, + { + "epoch": 2.2749014120340925, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.8055261969566345, + "num_tokens": 682447864.0, + "step": 17883 + }, + { + "epoch": 2.275028622312683, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.8051367402076721, + "num_tokens": 682486445.0, + "step": 17884 + }, + { + "epoch": 2.2751558325912735, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7962795495986938, + "num_tokens": 682518173.0, + "step": 17885 + }, + { + "epoch": 2.275283042869864, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7985703945159912, + "num_tokens": 682558627.0, + "step": 17886 + }, + { + "epoch": 2.2754102531484546, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8065295219421387, + "num_tokens": 682600821.0, + "step": 17887 + }, + { + "epoch": 2.275537463427045, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7929739952087402, + "num_tokens": 682643947.0, + "step": 17888 + }, + { + "epoch": 2.275664673705635, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7932938933372498, + "num_tokens": 682683045.0, + "step": 17889 + }, + { + "epoch": 2.275791883984226, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7992751598358154, + "num_tokens": 682721617.0, + "step": 17890 + }, + { + "epoch": 2.2759190942628162, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7870326042175293, + "num_tokens": 682758339.0, + "step": 17891 + }, + { + "epoch": 2.2760463045414068, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8079788684844971, + "num_tokens": 682800501.0, + "step": 17892 + }, + { + "epoch": 2.2761735148199973, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7978443503379822, + "num_tokens": 682833231.0, + "step": 17893 + }, + { + "epoch": 2.276300725098588, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8063600063323975, + "num_tokens": 682873945.0, + "step": 17894 + }, + { + "epoch": 2.2764279353771784, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8032360076904297, + "num_tokens": 682918329.0, + "step": 17895 + }, + { + "epoch": 2.276555145655769, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8045375943183899, + "num_tokens": 682956277.0, + "step": 17896 + }, + { + "epoch": 2.2766823559343594, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.803179919719696, + "num_tokens": 682990648.0, + "step": 17897 + }, + { + "epoch": 2.27680956621295, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8143109083175659, + "num_tokens": 683029046.0, + "step": 17898 + }, + { + "epoch": 2.2769367764915405, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7954120635986328, + "num_tokens": 683064541.0, + "step": 17899 + }, + { + "epoch": 2.277063986770131, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8028000593185425, + "num_tokens": 683105015.0, + "step": 17900 + }, + { + "epoch": 2.2771911970487215, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7920066714286804, + "num_tokens": 683142938.0, + "step": 17901 + }, + { + "epoch": 2.277318407327312, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8131363391876221, + "num_tokens": 683181593.0, + "step": 17902 + }, + { + "epoch": 2.2774456176059026, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.788837730884552, + "num_tokens": 683218641.0, + "step": 17903 + }, + { + "epoch": 2.277572827884493, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7857396602630615, + "num_tokens": 683261479.0, + "step": 17904 + }, + { + "epoch": 2.2777000381630836, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7753183841705322, + "num_tokens": 683300602.0, + "step": 17905 + }, + { + "epoch": 2.277827248441674, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7985371947288513, + "num_tokens": 683337012.0, + "step": 17906 + }, + { + "epoch": 2.2779544587202647, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8018883466720581, + "num_tokens": 683368949.0, + "step": 17907 + }, + { + "epoch": 2.2780816689988552, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7937413454055786, + "num_tokens": 683408741.0, + "step": 17908 + }, + { + "epoch": 2.2782088792774458, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7925078868865967, + "num_tokens": 683447843.0, + "step": 17909 + }, + { + "epoch": 2.2783360895560363, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7946630716323853, + "num_tokens": 683489695.0, + "step": 17910 + }, + { + "epoch": 2.278463299834627, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.799950897693634, + "num_tokens": 683528381.0, + "step": 17911 + }, + { + "epoch": 2.2785905101132173, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8005411624908447, + "num_tokens": 683567261.0, + "step": 17912 + }, + { + "epoch": 2.278717720391808, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7810359001159668, + "num_tokens": 683601083.0, + "step": 17913 + }, + { + "epoch": 2.278844930670398, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7947081923484802, + "num_tokens": 683643494.0, + "step": 17914 + }, + { + "epoch": 2.278972140948989, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7959733009338379, + "num_tokens": 683684267.0, + "step": 17915 + }, + { + "epoch": 2.279099351227579, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7855443954467773, + "num_tokens": 683723292.0, + "step": 17916 + }, + { + "epoch": 2.2792265615061695, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8127642869949341, + "num_tokens": 683761378.0, + "step": 17917 + }, + { + "epoch": 2.27935377178476, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7863389253616333, + "num_tokens": 683794838.0, + "step": 17918 + }, + { + "epoch": 2.2794809820633506, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7782540321350098, + "num_tokens": 683832307.0, + "step": 17919 + }, + { + "epoch": 2.279608192341941, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.789181113243103, + "num_tokens": 683871848.0, + "step": 17920 + }, + { + "epoch": 2.2797354026205316, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8079787492752075, + "num_tokens": 683915086.0, + "step": 17921 + }, + { + "epoch": 2.279862612899122, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7888020277023315, + "num_tokens": 683945411.0, + "step": 17922 + }, + { + "epoch": 2.2799898231777127, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7988125085830688, + "num_tokens": 683979664.0, + "step": 17923 + }, + { + "epoch": 2.2801170334563032, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7926875352859497, + "num_tokens": 684025526.0, + "step": 17924 + }, + { + "epoch": 2.2802442437348938, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7866112589836121, + "num_tokens": 684056992.0, + "step": 17925 + }, + { + "epoch": 2.2803714540134843, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8039658069610596, + "num_tokens": 684094515.0, + "step": 17926 + }, + { + "epoch": 2.280498664292075, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8068241477012634, + "num_tokens": 684132294.0, + "step": 17927 + }, + { + "epoch": 2.2806258745706653, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8011583685874939, + "num_tokens": 684164010.0, + "step": 17928 + }, + { + "epoch": 2.280753084849256, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8080434799194336, + "num_tokens": 684202077.0, + "step": 17929 + }, + { + "epoch": 2.2808802951278464, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.776044487953186, + "num_tokens": 684240799.0, + "step": 17930 + }, + { + "epoch": 2.281007505406437, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8014903664588928, + "num_tokens": 684284080.0, + "step": 17931 + }, + { + "epoch": 2.2811347156850275, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7801259756088257, + "num_tokens": 684317915.0, + "step": 17932 + }, + { + "epoch": 2.281261925963618, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.5191, + "mean_token_accuracy": 0.8403053879737854, + "num_tokens": 684354665.0, + "step": 17933 + }, + { + "epoch": 2.2813891362422085, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7992529273033142, + "num_tokens": 684392866.0, + "step": 17934 + }, + { + "epoch": 2.281516346520799, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8070857524871826, + "num_tokens": 684428422.0, + "step": 17935 + }, + { + "epoch": 2.2816435567993896, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7888003587722778, + "num_tokens": 684467129.0, + "step": 17936 + }, + { + "epoch": 2.2817707670779797, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7904837727546692, + "num_tokens": 684500093.0, + "step": 17937 + }, + { + "epoch": 2.2818979773565706, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7874531149864197, + "num_tokens": 684541422.0, + "step": 17938 + }, + { + "epoch": 2.2820251876351607, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7931752800941467, + "num_tokens": 684579164.0, + "step": 17939 + }, + { + "epoch": 2.2821523979137512, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7844353914260864, + "num_tokens": 684618617.0, + "step": 17940 + }, + { + "epoch": 2.2822796081923418, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7778948545455933, + "num_tokens": 684655525.0, + "step": 17941 + }, + { + "epoch": 2.2824068184709323, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7990283966064453, + "num_tokens": 684694525.0, + "step": 17942 + }, + { + "epoch": 2.282534028749523, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8030413389205933, + "num_tokens": 684733247.0, + "step": 17943 + }, + { + "epoch": 2.2826612390281134, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8086237907409668, + "num_tokens": 684763642.0, + "step": 17944 + }, + { + "epoch": 2.282788449306704, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.804115355014801, + "num_tokens": 684806164.0, + "step": 17945 + }, + { + "epoch": 2.2829156595852944, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7981661558151245, + "num_tokens": 684844193.0, + "step": 17946 + }, + { + "epoch": 2.283042869863885, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8039728999137878, + "num_tokens": 684882103.0, + "step": 17947 + }, + { + "epoch": 2.2831700801424755, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8137948513031006, + "num_tokens": 684917667.0, + "step": 17948 + }, + { + "epoch": 2.283297290421066, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8057711124420166, + "num_tokens": 684954007.0, + "step": 17949 + }, + { + "epoch": 2.2834245006996565, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7947782278060913, + "num_tokens": 684994693.0, + "step": 17950 + }, + { + "epoch": 2.283551710978247, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7885856032371521, + "num_tokens": 685032713.0, + "step": 17951 + }, + { + "epoch": 2.2836789212568376, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7956684231758118, + "num_tokens": 685066597.0, + "step": 17952 + }, + { + "epoch": 2.283806131535428, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.793912410736084, + "num_tokens": 685106243.0, + "step": 17953 + }, + { + "epoch": 2.2839333418140186, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.796625554561615, + "num_tokens": 685145177.0, + "step": 17954 + }, + { + "epoch": 2.284060552092609, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8057945966720581, + "num_tokens": 685189130.0, + "step": 17955 + }, + { + "epoch": 2.2841877623711997, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8024940490722656, + "num_tokens": 685227536.0, + "step": 17956 + }, + { + "epoch": 2.28431497264979, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8111499547958374, + "num_tokens": 685267213.0, + "step": 17957 + }, + { + "epoch": 2.2844421829283807, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7977861166000366, + "num_tokens": 685301115.0, + "step": 17958 + }, + { + "epoch": 2.2845693932069713, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.8117995262145996, + "num_tokens": 685338497.0, + "step": 17959 + }, + { + "epoch": 2.284696603485562, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7877744436264038, + "num_tokens": 685374983.0, + "step": 17960 + }, + { + "epoch": 2.2848238137641523, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8044819831848145, + "num_tokens": 685408859.0, + "step": 17961 + }, + { + "epoch": 2.2849510240427424, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7953687906265259, + "num_tokens": 685445103.0, + "step": 17962 + }, + { + "epoch": 2.2850782343213334, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8001269102096558, + "num_tokens": 685484657.0, + "step": 17963 + }, + { + "epoch": 2.2852054445999235, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8066637516021729, + "num_tokens": 685524034.0, + "step": 17964 + }, + { + "epoch": 2.285332654878514, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.805725634098053, + "num_tokens": 685560368.0, + "step": 17965 + }, + { + "epoch": 2.2854598651571045, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7965859174728394, + "num_tokens": 685598192.0, + "step": 17966 + }, + { + "epoch": 2.285587075435695, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8052672743797302, + "num_tokens": 685636300.0, + "step": 17967 + }, + { + "epoch": 2.2857142857142856, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8023993372917175, + "num_tokens": 685671457.0, + "step": 17968 + }, + { + "epoch": 2.285841495992876, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7864458560943604, + "num_tokens": 685709687.0, + "step": 17969 + }, + { + "epoch": 2.2859687062714666, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7805209159851074, + "num_tokens": 685747854.0, + "step": 17970 + }, + { + "epoch": 2.286095916550057, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8183543682098389, + "num_tokens": 685785215.0, + "step": 17971 + }, + { + "epoch": 2.2862231268286477, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8016115427017212, + "num_tokens": 685818871.0, + "step": 17972 + }, + { + "epoch": 2.2863503371072382, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7864537239074707, + "num_tokens": 685859089.0, + "step": 17973 + }, + { + "epoch": 2.2864775473858288, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7924403548240662, + "num_tokens": 685900346.0, + "step": 17974 + }, + { + "epoch": 2.2866047576644193, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8026931285858154, + "num_tokens": 685938186.0, + "step": 17975 + }, + { + "epoch": 2.28673196794301, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8045830726623535, + "num_tokens": 685974911.0, + "step": 17976 + }, + { + "epoch": 2.2868591782216003, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8036137819290161, + "num_tokens": 686010430.0, + "step": 17977 + }, + { + "epoch": 2.286986388500191, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8115078210830688, + "num_tokens": 686042727.0, + "step": 17978 + }, + { + "epoch": 2.2871135987787814, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7969182729721069, + "num_tokens": 686078819.0, + "step": 17979 + }, + { + "epoch": 2.287240809057372, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8031919598579407, + "num_tokens": 686117760.0, + "step": 17980 + }, + { + "epoch": 2.2873680193359625, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7937110662460327, + "num_tokens": 686152992.0, + "step": 17981 + }, + { + "epoch": 2.287495229614553, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8036558628082275, + "num_tokens": 686195353.0, + "step": 17982 + }, + { + "epoch": 2.2876224398931435, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7925338745117188, + "num_tokens": 686232990.0, + "step": 17983 + }, + { + "epoch": 2.287749650171734, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8071790933609009, + "num_tokens": 686265933.0, + "step": 17984 + }, + { + "epoch": 2.2878768604503246, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8053423762321472, + "num_tokens": 686307111.0, + "step": 17985 + }, + { + "epoch": 2.288004070728915, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7981042861938477, + "num_tokens": 686343363.0, + "step": 17986 + }, + { + "epoch": 2.288131281007505, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7953619956970215, + "num_tokens": 686380888.0, + "step": 17987 + }, + { + "epoch": 2.288258491286096, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7940359115600586, + "num_tokens": 686417350.0, + "step": 17988 + }, + { + "epoch": 2.2883857015646862, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8118994235992432, + "num_tokens": 686458658.0, + "step": 17989 + }, + { + "epoch": 2.2885129118432768, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8019759654998779, + "num_tokens": 686493945.0, + "step": 17990 + }, + { + "epoch": 2.2886401221218673, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7992905974388123, + "num_tokens": 686527608.0, + "step": 17991 + }, + { + "epoch": 2.288767332400458, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8116183280944824, + "num_tokens": 686562272.0, + "step": 17992 + }, + { + "epoch": 2.2888945426790483, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8185534477233887, + "num_tokens": 686599937.0, + "step": 17993 + }, + { + "epoch": 2.289021752957639, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8017445206642151, + "num_tokens": 686632103.0, + "step": 17994 + }, + { + "epoch": 2.2891489632362294, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8099255561828613, + "num_tokens": 686662802.0, + "step": 17995 + }, + { + "epoch": 2.28927617351482, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.573, + "mean_token_accuracy": 0.8243898153305054, + "num_tokens": 686699386.0, + "step": 17996 + }, + { + "epoch": 2.2894033837934105, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8033977746963501, + "num_tokens": 686738369.0, + "step": 17997 + }, + { + "epoch": 2.289530594072001, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7894835472106934, + "num_tokens": 686774543.0, + "step": 17998 + }, + { + "epoch": 2.2896578043505915, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.791085958480835, + "num_tokens": 686807048.0, + "step": 17999 + }, + { + "epoch": 2.289785014629182, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8026001453399658, + "num_tokens": 686844678.0, + "step": 18000 + }, + { + "epoch": 2.2899122249077726, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7960007190704346, + "num_tokens": 686877721.0, + "step": 18001 + }, + { + "epoch": 2.290039435186363, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7867598533630371, + "num_tokens": 686918329.0, + "step": 18002 + }, + { + "epoch": 2.2901666454649536, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.794000506401062, + "num_tokens": 686960050.0, + "step": 18003 + }, + { + "epoch": 2.290293855743544, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8106274604797363, + "num_tokens": 686997849.0, + "step": 18004 + }, + { + "epoch": 2.2904210660221347, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8016813397407532, + "num_tokens": 687032363.0, + "step": 18005 + }, + { + "epoch": 2.290548276300725, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7823994159698486, + "num_tokens": 687069606.0, + "step": 18006 + }, + { + "epoch": 2.2906754865793157, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7954885363578796, + "num_tokens": 687112857.0, + "step": 18007 + }, + { + "epoch": 2.2908026968579063, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7990589141845703, + "num_tokens": 687152245.0, + "step": 18008 + }, + { + "epoch": 2.290929907136497, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7969906330108643, + "num_tokens": 687193988.0, + "step": 18009 + }, + { + "epoch": 2.2910571174150873, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.794449508190155, + "num_tokens": 687232375.0, + "step": 18010 + }, + { + "epoch": 2.291184327693678, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8057147264480591, + "num_tokens": 687274566.0, + "step": 18011 + }, + { + "epoch": 2.291311537972268, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7992198467254639, + "num_tokens": 687313231.0, + "step": 18012 + }, + { + "epoch": 2.291438748250859, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8037787675857544, + "num_tokens": 687354502.0, + "step": 18013 + }, + { + "epoch": 2.291565958529449, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7961609959602356, + "num_tokens": 687399160.0, + "step": 18014 + }, + { + "epoch": 2.2916931688080395, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7902703285217285, + "num_tokens": 687438775.0, + "step": 18015 + }, + { + "epoch": 2.29182037908663, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8051888346672058, + "num_tokens": 687471435.0, + "step": 18016 + }, + { + "epoch": 2.2919475893652206, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7992815971374512, + "num_tokens": 687508657.0, + "step": 18017 + }, + { + "epoch": 2.292074799643811, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8100436925888062, + "num_tokens": 687545024.0, + "step": 18018 + }, + { + "epoch": 2.2922020099224016, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8048155307769775, + "num_tokens": 687579432.0, + "step": 18019 + }, + { + "epoch": 2.292329220200992, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7901301980018616, + "num_tokens": 687618378.0, + "step": 18020 + }, + { + "epoch": 2.2924564304795827, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7939498424530029, + "num_tokens": 687657119.0, + "step": 18021 + }, + { + "epoch": 2.2925836407581732, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7915120124816895, + "num_tokens": 687691924.0, + "step": 18022 + }, + { + "epoch": 2.2927108510367638, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7744569778442383, + "num_tokens": 687724221.0, + "step": 18023 + }, + { + "epoch": 2.2928380613153543, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8070275783538818, + "num_tokens": 687765049.0, + "step": 18024 + }, + { + "epoch": 2.292965271593945, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7901960611343384, + "num_tokens": 687810222.0, + "step": 18025 + }, + { + "epoch": 2.2930924818725353, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7954453229904175, + "num_tokens": 687851961.0, + "step": 18026 + }, + { + "epoch": 2.293219692151126, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7862015962600708, + "num_tokens": 687890448.0, + "step": 18027 + }, + { + "epoch": 2.2933469024297164, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8083803057670593, + "num_tokens": 687929509.0, + "step": 18028 + }, + { + "epoch": 2.293474112708307, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7882936000823975, + "num_tokens": 687966814.0, + "step": 18029 + }, + { + "epoch": 2.2936013229868975, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8078234195709229, + "num_tokens": 688006398.0, + "step": 18030 + }, + { + "epoch": 2.293728533265488, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7947673797607422, + "num_tokens": 688044604.0, + "step": 18031 + }, + { + "epoch": 2.2938557435440785, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7931658029556274, + "num_tokens": 688082969.0, + "step": 18032 + }, + { + "epoch": 2.293982953822669, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8078964948654175, + "num_tokens": 688121442.0, + "step": 18033 + }, + { + "epoch": 2.2941101641012596, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8003374338150024, + "num_tokens": 688161096.0, + "step": 18034 + }, + { + "epoch": 2.2942373743798496, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.80291748046875, + "num_tokens": 688200782.0, + "step": 18035 + }, + { + "epoch": 2.2943645846584406, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8096527457237244, + "num_tokens": 688241333.0, + "step": 18036 + }, + { + "epoch": 2.2944917949370307, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7989997863769531, + "num_tokens": 688278297.0, + "step": 18037 + }, + { + "epoch": 2.2946190052156212, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7790345549583435, + "num_tokens": 688318835.0, + "step": 18038 + }, + { + "epoch": 2.2947462154942118, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7993754744529724, + "num_tokens": 688354814.0, + "step": 18039 + }, + { + "epoch": 2.2948734257728023, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7980066537857056, + "num_tokens": 688390008.0, + "step": 18040 + }, + { + "epoch": 2.295000636051393, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.803056001663208, + "num_tokens": 688430938.0, + "step": 18041 + }, + { + "epoch": 2.2951278463299833, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7638314962387085, + "num_tokens": 688466730.0, + "step": 18042 + }, + { + "epoch": 2.295255056608574, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7978662252426147, + "num_tokens": 688506666.0, + "step": 18043 + }, + { + "epoch": 2.2953822668871644, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7842944264411926, + "num_tokens": 688547726.0, + "step": 18044 + }, + { + "epoch": 2.295509477165755, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7964204549789429, + "num_tokens": 688587487.0, + "step": 18045 + }, + { + "epoch": 2.2956366874443455, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7964953780174255, + "num_tokens": 688632454.0, + "step": 18046 + }, + { + "epoch": 2.295763897722936, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8105812072753906, + "num_tokens": 688668489.0, + "step": 18047 + }, + { + "epoch": 2.2958911080015265, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8043630719184875, + "num_tokens": 688707899.0, + "step": 18048 + }, + { + "epoch": 2.296018318280117, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8053827881813049, + "num_tokens": 688747171.0, + "step": 18049 + }, + { + "epoch": 2.2961455285587076, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8012550473213196, + "num_tokens": 688789553.0, + "step": 18050 + }, + { + "epoch": 2.296272738837298, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7984936237335205, + "num_tokens": 688827632.0, + "step": 18051 + }, + { + "epoch": 2.2963999491158886, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8082616329193115, + "num_tokens": 688861505.0, + "step": 18052 + }, + { + "epoch": 2.296527159394479, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7936257719993591, + "num_tokens": 688898270.0, + "step": 18053 + }, + { + "epoch": 2.2966543696730697, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8023229837417603, + "num_tokens": 688937420.0, + "step": 18054 + }, + { + "epoch": 2.29678157995166, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7999227046966553, + "num_tokens": 688972426.0, + "step": 18055 + }, + { + "epoch": 2.2969087902302507, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7938854694366455, + "num_tokens": 689005801.0, + "step": 18056 + }, + { + "epoch": 2.2970360005088413, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7923462986946106, + "num_tokens": 689039999.0, + "step": 18057 + }, + { + "epoch": 2.297163210787432, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7780901193618774, + "num_tokens": 689079404.0, + "step": 18058 + }, + { + "epoch": 2.2972904210660223, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7726789712905884, + "num_tokens": 689116197.0, + "step": 18059 + }, + { + "epoch": 2.2974176313446124, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8055123090744019, + "num_tokens": 689153446.0, + "step": 18060 + }, + { + "epoch": 2.2975448416232034, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7760049104690552, + "num_tokens": 689185310.0, + "step": 18061 + }, + { + "epoch": 2.2976720519017935, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8116823434829712, + "num_tokens": 689224150.0, + "step": 18062 + }, + { + "epoch": 2.297799262180384, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7895867228507996, + "num_tokens": 689261206.0, + "step": 18063 + }, + { + "epoch": 2.2979264724589745, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8044170141220093, + "num_tokens": 689299129.0, + "step": 18064 + }, + { + "epoch": 2.298053682737565, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8118394017219543, + "num_tokens": 689330554.0, + "step": 18065 + }, + { + "epoch": 2.2981808930161556, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8051321506500244, + "num_tokens": 689364084.0, + "step": 18066 + }, + { + "epoch": 2.298308103294746, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7915499210357666, + "num_tokens": 689405254.0, + "step": 18067 + }, + { + "epoch": 2.2984353135733366, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7854292392730713, + "num_tokens": 689441109.0, + "step": 18068 + }, + { + "epoch": 2.298562523851927, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7903224229812622, + "num_tokens": 689474937.0, + "step": 18069 + }, + { + "epoch": 2.2986897341305177, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7959071397781372, + "num_tokens": 689515971.0, + "step": 18070 + }, + { + "epoch": 2.298816944409108, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7738571166992188, + "num_tokens": 689553620.0, + "step": 18071 + }, + { + "epoch": 2.2989441546876987, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.8010774850845337, + "num_tokens": 689592157.0, + "step": 18072 + }, + { + "epoch": 2.2990713649662893, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8133864402770996, + "num_tokens": 689626889.0, + "step": 18073 + }, + { + "epoch": 2.29919857524488, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8124592900276184, + "num_tokens": 689666161.0, + "step": 18074 + }, + { + "epoch": 2.2993257855234703, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7956169247627258, + "num_tokens": 689713713.0, + "step": 18075 + }, + { + "epoch": 2.299452995802061, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7936722040176392, + "num_tokens": 689755146.0, + "step": 18076 + }, + { + "epoch": 2.2995802060806514, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8181383609771729, + "num_tokens": 689795925.0, + "step": 18077 + }, + { + "epoch": 2.299707416359242, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8115365505218506, + "num_tokens": 689830673.0, + "step": 18078 + }, + { + "epoch": 2.2998346266378324, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7946577668190002, + "num_tokens": 689863649.0, + "step": 18079 + }, + { + "epoch": 2.299961836916423, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7975864410400391, + "num_tokens": 689901946.0, + "step": 18080 + }, + { + "epoch": 2.3000890471950135, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7966864705085754, + "num_tokens": 689939968.0, + "step": 18081 + }, + { + "epoch": 2.300216257473604, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.811010479927063, + "num_tokens": 689979792.0, + "step": 18082 + }, + { + "epoch": 2.3003434677521946, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8113791346549988, + "num_tokens": 690013974.0, + "step": 18083 + }, + { + "epoch": 2.300470678030785, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8088980913162231, + "num_tokens": 690056540.0, + "step": 18084 + }, + { + "epoch": 2.300597888309375, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7959882616996765, + "num_tokens": 690101032.0, + "step": 18085 + }, + { + "epoch": 2.300725098587966, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8111228942871094, + "num_tokens": 690135417.0, + "step": 18086 + }, + { + "epoch": 2.3008523088665562, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8011776208877563, + "num_tokens": 690178552.0, + "step": 18087 + }, + { + "epoch": 2.3009795191451468, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7963843941688538, + "num_tokens": 690221928.0, + "step": 18088 + }, + { + "epoch": 2.3011067294237373, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8041858673095703, + "num_tokens": 690257040.0, + "step": 18089 + }, + { + "epoch": 2.301233939702328, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7984634041786194, + "num_tokens": 690302720.0, + "step": 18090 + }, + { + "epoch": 2.3013611499809183, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7863031029701233, + "num_tokens": 690337042.0, + "step": 18091 + }, + { + "epoch": 2.301488360259509, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8119683861732483, + "num_tokens": 690380247.0, + "step": 18092 + }, + { + "epoch": 2.3016155705380994, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7992159128189087, + "num_tokens": 690425361.0, + "step": 18093 + }, + { + "epoch": 2.30174278081669, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7990714311599731, + "num_tokens": 690465137.0, + "step": 18094 + }, + { + "epoch": 2.3018699910952805, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7952990531921387, + "num_tokens": 690501437.0, + "step": 18095 + }, + { + "epoch": 2.301997201373871, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7958948612213135, + "num_tokens": 690544134.0, + "step": 18096 + }, + { + "epoch": 2.3021244116524615, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.806693434715271, + "num_tokens": 690581062.0, + "step": 18097 + }, + { + "epoch": 2.302251621931052, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7992526292800903, + "num_tokens": 690615145.0, + "step": 18098 + }, + { + "epoch": 2.3023788322096426, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8034584522247314, + "num_tokens": 690655483.0, + "step": 18099 + }, + { + "epoch": 2.302506042488233, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7984590530395508, + "num_tokens": 690694133.0, + "step": 18100 + }, + { + "epoch": 2.3026332527668236, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.779967188835144, + "num_tokens": 690729939.0, + "step": 18101 + }, + { + "epoch": 2.302760463045414, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8118918538093567, + "num_tokens": 690770139.0, + "step": 18102 + }, + { + "epoch": 2.3028876733240047, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8028707504272461, + "num_tokens": 690811379.0, + "step": 18103 + }, + { + "epoch": 2.303014883602595, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.792048454284668, + "num_tokens": 690844194.0, + "step": 18104 + }, + { + "epoch": 2.3031420938811857, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.791267991065979, + "num_tokens": 690882129.0, + "step": 18105 + }, + { + "epoch": 2.3032693041597763, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8080365061759949, + "num_tokens": 690915568.0, + "step": 18106 + }, + { + "epoch": 2.303396514438367, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7959178686141968, + "num_tokens": 690953412.0, + "step": 18107 + }, + { + "epoch": 2.3035237247169573, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7957189083099365, + "num_tokens": 690990047.0, + "step": 18108 + }, + { + "epoch": 2.303650934995548, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8096221685409546, + "num_tokens": 691031010.0, + "step": 18109 + }, + { + "epoch": 2.303778145274138, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8033653497695923, + "num_tokens": 691065830.0, + "step": 18110 + }, + { + "epoch": 2.303905355552729, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8096903562545776, + "num_tokens": 691101418.0, + "step": 18111 + }, + { + "epoch": 2.304032565831319, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7871103882789612, + "num_tokens": 691139447.0, + "step": 18112 + }, + { + "epoch": 2.3041597761099095, + "grad_norm": 77.5, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7711061239242554, + "num_tokens": 691168809.0, + "step": 18113 + }, + { + "epoch": 2.3042869863885, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8016209602355957, + "num_tokens": 691200420.0, + "step": 18114 + }, + { + "epoch": 2.3044141966670906, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7970389723777771, + "num_tokens": 691238587.0, + "step": 18115 + }, + { + "epoch": 2.304541406945681, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7931407690048218, + "num_tokens": 691277137.0, + "step": 18116 + }, + { + "epoch": 2.3046686172242716, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.801855742931366, + "num_tokens": 691319564.0, + "step": 18117 + }, + { + "epoch": 2.304795827502862, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7996363043785095, + "num_tokens": 691359652.0, + "step": 18118 + }, + { + "epoch": 2.3049230377814527, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7998515367507935, + "num_tokens": 691399705.0, + "step": 18119 + }, + { + "epoch": 2.305050248060043, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7942323088645935, + "num_tokens": 691439447.0, + "step": 18120 + }, + { + "epoch": 2.3051774583386337, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7862081527709961, + "num_tokens": 691477453.0, + "step": 18121 + }, + { + "epoch": 2.3053046686172243, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8004090785980225, + "num_tokens": 691517766.0, + "step": 18122 + }, + { + "epoch": 2.305431878895815, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8017809391021729, + "num_tokens": 691553956.0, + "step": 18123 + }, + { + "epoch": 2.3055590891744053, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.8016530871391296, + "num_tokens": 691592522.0, + "step": 18124 + }, + { + "epoch": 2.305686299452996, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8018873333930969, + "num_tokens": 691629202.0, + "step": 18125 + }, + { + "epoch": 2.3058135097315864, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8038730621337891, + "num_tokens": 691665324.0, + "step": 18126 + }, + { + "epoch": 2.305940720010177, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8175345659255981, + "num_tokens": 691704223.0, + "step": 18127 + }, + { + "epoch": 2.3060679302887674, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7947226762771606, + "num_tokens": 691745061.0, + "step": 18128 + }, + { + "epoch": 2.306195140567358, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8165437579154968, + "num_tokens": 691783630.0, + "step": 18129 + }, + { + "epoch": 2.3063223508459485, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.8001003861427307, + "num_tokens": 691818283.0, + "step": 18130 + }, + { + "epoch": 2.306449561124539, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7864688634872437, + "num_tokens": 691854585.0, + "step": 18131 + }, + { + "epoch": 2.3065767714031296, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7761341333389282, + "num_tokens": 691891784.0, + "step": 18132 + }, + { + "epoch": 2.3067039816817196, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7983396053314209, + "num_tokens": 691927389.0, + "step": 18133 + }, + { + "epoch": 2.3068311919603106, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7879406213760376, + "num_tokens": 691962928.0, + "step": 18134 + }, + { + "epoch": 2.3069584022389007, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8028669357299805, + "num_tokens": 692007064.0, + "step": 18135 + }, + { + "epoch": 2.3070856125174912, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7998422384262085, + "num_tokens": 692040668.0, + "step": 18136 + }, + { + "epoch": 2.3072128227960818, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7967292070388794, + "num_tokens": 692077710.0, + "step": 18137 + }, + { + "epoch": 2.3073400330746723, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.807784914970398, + "num_tokens": 692112316.0, + "step": 18138 + }, + { + "epoch": 2.307467243353263, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8001763224601746, + "num_tokens": 692153201.0, + "step": 18139 + }, + { + "epoch": 2.3075944536318533, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7934086322784424, + "num_tokens": 692192942.0, + "step": 18140 + }, + { + "epoch": 2.307721663910444, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.819519579410553, + "num_tokens": 692235515.0, + "step": 18141 + }, + { + "epoch": 2.3078488741890344, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.786376953125, + "num_tokens": 692271338.0, + "step": 18142 + }, + { + "epoch": 2.307976084467625, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7800610065460205, + "num_tokens": 692311469.0, + "step": 18143 + }, + { + "epoch": 2.3081032947462155, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7781265377998352, + "num_tokens": 692352052.0, + "step": 18144 + }, + { + "epoch": 2.308230505024806, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.778236985206604, + "num_tokens": 692382978.0, + "step": 18145 + }, + { + "epoch": 2.3083577153033965, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7981561422348022, + "num_tokens": 692415651.0, + "step": 18146 + }, + { + "epoch": 2.308484925581987, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7814673185348511, + "num_tokens": 692455937.0, + "step": 18147 + }, + { + "epoch": 2.3086121358605776, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7944438457489014, + "num_tokens": 692496277.0, + "step": 18148 + }, + { + "epoch": 2.308739346139168, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7974885106086731, + "num_tokens": 692530485.0, + "step": 18149 + }, + { + "epoch": 2.3088665564177586, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7905232310295105, + "num_tokens": 692564794.0, + "step": 18150 + }, + { + "epoch": 2.308993766696349, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7942189574241638, + "num_tokens": 692610359.0, + "step": 18151 + }, + { + "epoch": 2.3091209769749397, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8051947355270386, + "num_tokens": 692641560.0, + "step": 18152 + }, + { + "epoch": 2.30924818725353, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8008339405059814, + "num_tokens": 692680114.0, + "step": 18153 + }, + { + "epoch": 2.3093753975321207, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7780716419219971, + "num_tokens": 692714199.0, + "step": 18154 + }, + { + "epoch": 2.3095026078107113, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8021783232688904, + "num_tokens": 692752331.0, + "step": 18155 + }, + { + "epoch": 2.309629818089302, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8085169792175293, + "num_tokens": 692789167.0, + "step": 18156 + }, + { + "epoch": 2.3097570283678923, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7985576391220093, + "num_tokens": 692829234.0, + "step": 18157 + }, + { + "epoch": 2.3098842386464824, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7740855813026428, + "num_tokens": 692863749.0, + "step": 18158 + }, + { + "epoch": 2.3100114489250734, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7940900325775146, + "num_tokens": 692900972.0, + "step": 18159 + }, + { + "epoch": 2.3101386592036635, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.789096474647522, + "num_tokens": 692938522.0, + "step": 18160 + }, + { + "epoch": 2.310265869482254, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.808638870716095, + "num_tokens": 692973653.0, + "step": 18161 + }, + { + "epoch": 2.3103930797608445, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.793965220451355, + "num_tokens": 693016325.0, + "step": 18162 + }, + { + "epoch": 2.310520290039435, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.812523603439331, + "num_tokens": 693052022.0, + "step": 18163 + }, + { + "epoch": 2.3106475003180256, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8038907051086426, + "num_tokens": 693093936.0, + "step": 18164 + }, + { + "epoch": 2.310774710596616, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.8134641647338867, + "num_tokens": 693130160.0, + "step": 18165 + }, + { + "epoch": 2.3109019208752066, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8048815727233887, + "num_tokens": 693172405.0, + "step": 18166 + }, + { + "epoch": 2.311029131153797, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7974377274513245, + "num_tokens": 693206499.0, + "step": 18167 + }, + { + "epoch": 2.3111563414323877, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.802741527557373, + "num_tokens": 693244287.0, + "step": 18168 + }, + { + "epoch": 2.311283551710978, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7906731963157654, + "num_tokens": 693286164.0, + "step": 18169 + }, + { + "epoch": 2.3114107619895687, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.812296450138092, + "num_tokens": 693330961.0, + "step": 18170 + }, + { + "epoch": 2.3115379722681593, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8105449676513672, + "num_tokens": 693375169.0, + "step": 18171 + }, + { + "epoch": 2.31166518254675, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.795055627822876, + "num_tokens": 693414127.0, + "step": 18172 + }, + { + "epoch": 2.3117923928253403, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7991703152656555, + "num_tokens": 693450446.0, + "step": 18173 + }, + { + "epoch": 2.311919603103931, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7924249172210693, + "num_tokens": 693490973.0, + "step": 18174 + }, + { + "epoch": 2.3120468133825214, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.803849458694458, + "num_tokens": 693526815.0, + "step": 18175 + }, + { + "epoch": 2.312174023661112, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.788453221321106, + "num_tokens": 693566949.0, + "step": 18176 + }, + { + "epoch": 2.3123012339397024, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7853682041168213, + "num_tokens": 693601116.0, + "step": 18177 + }, + { + "epoch": 2.312428444218293, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.5486, + "mean_token_accuracy": 0.8366556167602539, + "num_tokens": 693640588.0, + "step": 18178 + }, + { + "epoch": 2.3125556544968835, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7989050149917603, + "num_tokens": 693683352.0, + "step": 18179 + }, + { + "epoch": 2.312682864775474, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8069350123405457, + "num_tokens": 693722579.0, + "step": 18180 + }, + { + "epoch": 2.3128100750540646, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.805791974067688, + "num_tokens": 693766743.0, + "step": 18181 + }, + { + "epoch": 2.312937285332655, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7995663285255432, + "num_tokens": 693809699.0, + "step": 18182 + }, + { + "epoch": 2.313064495611245, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8027682900428772, + "num_tokens": 693845508.0, + "step": 18183 + }, + { + "epoch": 2.313191705889836, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7976240515708923, + "num_tokens": 693883763.0, + "step": 18184 + }, + { + "epoch": 2.313318916168426, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8023573160171509, + "num_tokens": 693917093.0, + "step": 18185 + }, + { + "epoch": 2.3134461264470167, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7964637875556946, + "num_tokens": 693949219.0, + "step": 18186 + }, + { + "epoch": 2.3135733367256073, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7959487438201904, + "num_tokens": 693992755.0, + "step": 18187 + }, + { + "epoch": 2.313700547004198, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7850773334503174, + "num_tokens": 694030386.0, + "step": 18188 + }, + { + "epoch": 2.3138277572827883, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8121231198310852, + "num_tokens": 694065019.0, + "step": 18189 + }, + { + "epoch": 2.313954967561379, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8027395606040955, + "num_tokens": 694106645.0, + "step": 18190 + }, + { + "epoch": 2.3140821778399694, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8107098937034607, + "num_tokens": 694149146.0, + "step": 18191 + }, + { + "epoch": 2.31420938811856, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7858131527900696, + "num_tokens": 694193567.0, + "step": 18192 + }, + { + "epoch": 2.3143365983971504, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8112053871154785, + "num_tokens": 694232160.0, + "step": 18193 + }, + { + "epoch": 2.314463808675741, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.8004947900772095, + "num_tokens": 694275153.0, + "step": 18194 + }, + { + "epoch": 2.3145910189543315, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.807310938835144, + "num_tokens": 694313722.0, + "step": 18195 + }, + { + "epoch": 2.314718229232922, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8052708506584167, + "num_tokens": 694350544.0, + "step": 18196 + }, + { + "epoch": 2.3148454395115126, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8004277348518372, + "num_tokens": 694386552.0, + "step": 18197 + }, + { + "epoch": 2.314972649790103, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7970194816589355, + "num_tokens": 694434368.0, + "step": 18198 + }, + { + "epoch": 2.3150998600686936, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8025531768798828, + "num_tokens": 694479856.0, + "step": 18199 + }, + { + "epoch": 2.315227070347284, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8085222840309143, + "num_tokens": 694521107.0, + "step": 18200 + }, + { + "epoch": 2.3153542806258747, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.8006114363670349, + "num_tokens": 694560886.0, + "step": 18201 + }, + { + "epoch": 2.315481490904465, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7981236577033997, + "num_tokens": 694598404.0, + "step": 18202 + }, + { + "epoch": 2.3156087011830557, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7940630912780762, + "num_tokens": 694637162.0, + "step": 18203 + }, + { + "epoch": 2.3157359114616463, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8084414601325989, + "num_tokens": 694677435.0, + "step": 18204 + }, + { + "epoch": 2.315863121740237, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8096685409545898, + "num_tokens": 694713980.0, + "step": 18205 + }, + { + "epoch": 2.3159903320188273, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.794688880443573, + "num_tokens": 694753884.0, + "step": 18206 + }, + { + "epoch": 2.316117542297418, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8086552619934082, + "num_tokens": 694785739.0, + "step": 18207 + }, + { + "epoch": 2.316244752576008, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7988693118095398, + "num_tokens": 694827629.0, + "step": 18208 + }, + { + "epoch": 2.316371962854599, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.80023592710495, + "num_tokens": 694868241.0, + "step": 18209 + }, + { + "epoch": 2.316499173133189, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7812026143074036, + "num_tokens": 694905298.0, + "step": 18210 + }, + { + "epoch": 2.3166263834117795, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7903481721878052, + "num_tokens": 694938054.0, + "step": 18211 + }, + { + "epoch": 2.31675359369037, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7921192049980164, + "num_tokens": 694978761.0, + "step": 18212 + }, + { + "epoch": 2.3168808039689606, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7949486374855042, + "num_tokens": 695015073.0, + "step": 18213 + }, + { + "epoch": 2.317008014247551, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7960059642791748, + "num_tokens": 695052288.0, + "step": 18214 + }, + { + "epoch": 2.3171352245261416, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7839784622192383, + "num_tokens": 695088976.0, + "step": 18215 + }, + { + "epoch": 2.317262434804732, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8024312257766724, + "num_tokens": 695128482.0, + "step": 18216 + }, + { + "epoch": 2.3173896450833227, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.795374870300293, + "num_tokens": 695164963.0, + "step": 18217 + }, + { + "epoch": 2.317516855361913, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.8211103677749634, + "num_tokens": 695201515.0, + "step": 18218 + }, + { + "epoch": 2.3176440656405037, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8001855611801147, + "num_tokens": 695242057.0, + "step": 18219 + }, + { + "epoch": 2.3177712759190943, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.800998866558075, + "num_tokens": 695277128.0, + "step": 18220 + }, + { + "epoch": 2.317898486197685, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8152855634689331, + "num_tokens": 695311984.0, + "step": 18221 + }, + { + "epoch": 2.3180256964762753, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7864006757736206, + "num_tokens": 695348457.0, + "step": 18222 + }, + { + "epoch": 2.318152906754866, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8015972375869751, + "num_tokens": 695386518.0, + "step": 18223 + }, + { + "epoch": 2.3182801170334564, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7971161007881165, + "num_tokens": 695425365.0, + "step": 18224 + }, + { + "epoch": 2.318407327312047, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7974180579185486, + "num_tokens": 695459559.0, + "step": 18225 + }, + { + "epoch": 2.3185345375906374, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7901923656463623, + "num_tokens": 695491386.0, + "step": 18226 + }, + { + "epoch": 2.318661747869228, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7854762077331543, + "num_tokens": 695534660.0, + "step": 18227 + }, + { + "epoch": 2.3187889581478185, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7919158935546875, + "num_tokens": 695578127.0, + "step": 18228 + }, + { + "epoch": 2.318916168426409, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7842332124710083, + "num_tokens": 695614181.0, + "step": 18229 + }, + { + "epoch": 2.3190433787049995, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.8009721040725708, + "num_tokens": 695652753.0, + "step": 18230 + }, + { + "epoch": 2.3191705889835896, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.806676983833313, + "num_tokens": 695692449.0, + "step": 18231 + }, + { + "epoch": 2.3192977992621806, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8043183088302612, + "num_tokens": 695730812.0, + "step": 18232 + }, + { + "epoch": 2.3194250095407707, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8173055648803711, + "num_tokens": 695767577.0, + "step": 18233 + }, + { + "epoch": 2.319552219819361, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8036447763442993, + "num_tokens": 695812372.0, + "step": 18234 + }, + { + "epoch": 2.3196794300979517, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.8229450583457947, + "num_tokens": 695847470.0, + "step": 18235 + }, + { + "epoch": 2.3198066403765423, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8188210129737854, + "num_tokens": 695888082.0, + "step": 18236 + }, + { + "epoch": 2.319933850655133, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.8051894903182983, + "num_tokens": 695928904.0, + "step": 18237 + }, + { + "epoch": 2.3200610609337233, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7865167856216431, + "num_tokens": 695967549.0, + "step": 18238 + }, + { + "epoch": 2.320188271212314, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7941924333572388, + "num_tokens": 696005611.0, + "step": 18239 + }, + { + "epoch": 2.3203154814909044, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8023827075958252, + "num_tokens": 696034864.0, + "step": 18240 + }, + { + "epoch": 2.320442691769495, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.798271656036377, + "num_tokens": 696072137.0, + "step": 18241 + }, + { + "epoch": 2.3205699020480854, + "grad_norm": 10.3125, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.806638777256012, + "num_tokens": 696115744.0, + "step": 18242 + }, + { + "epoch": 2.320697112326676, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.79557204246521, + "num_tokens": 696151293.0, + "step": 18243 + }, + { + "epoch": 2.3208243226052665, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.819233238697052, + "num_tokens": 696190742.0, + "step": 18244 + }, + { + "epoch": 2.320951532883857, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7925457954406738, + "num_tokens": 696232807.0, + "step": 18245 + }, + { + "epoch": 2.3210787431624476, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7947125434875488, + "num_tokens": 696269416.0, + "step": 18246 + }, + { + "epoch": 2.321205953441038, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7976455688476562, + "num_tokens": 696305144.0, + "step": 18247 + }, + { + "epoch": 2.3213331637196286, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7920751571655273, + "num_tokens": 696344854.0, + "step": 18248 + }, + { + "epoch": 2.321460373998219, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.819364070892334, + "num_tokens": 696380154.0, + "step": 18249 + }, + { + "epoch": 2.3215875842768097, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.784753680229187, + "num_tokens": 696416458.0, + "step": 18250 + }, + { + "epoch": 2.3217147945554, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7981769442558289, + "num_tokens": 696453232.0, + "step": 18251 + }, + { + "epoch": 2.3218420048339907, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.776698648929596, + "num_tokens": 696487930.0, + "step": 18252 + }, + { + "epoch": 2.3219692151125813, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7881103754043579, + "num_tokens": 696523824.0, + "step": 18253 + }, + { + "epoch": 2.322096425391172, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8159527778625488, + "num_tokens": 696567007.0, + "step": 18254 + }, + { + "epoch": 2.3222236356697623, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.584, + "mean_token_accuracy": 0.8239197134971619, + "num_tokens": 696610055.0, + "step": 18255 + }, + { + "epoch": 2.3223508459483524, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7839779853820801, + "num_tokens": 696644058.0, + "step": 18256 + }, + { + "epoch": 2.3224780562269434, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.810976505279541, + "num_tokens": 696686381.0, + "step": 18257 + }, + { + "epoch": 2.3226052665055335, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.800428032875061, + "num_tokens": 696727567.0, + "step": 18258 + }, + { + "epoch": 2.322732476784124, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7893505096435547, + "num_tokens": 696767266.0, + "step": 18259 + }, + { + "epoch": 2.3228596870627145, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.8183851838111877, + "num_tokens": 696806059.0, + "step": 18260 + }, + { + "epoch": 2.322986897341305, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7955225706100464, + "num_tokens": 696844232.0, + "step": 18261 + }, + { + "epoch": 2.3231141076198956, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.8000836968421936, + "num_tokens": 696886223.0, + "step": 18262 + }, + { + "epoch": 2.323241317898486, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7870779037475586, + "num_tokens": 696922625.0, + "step": 18263 + }, + { + "epoch": 2.3233685281770766, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7883023023605347, + "num_tokens": 696965160.0, + "step": 18264 + }, + { + "epoch": 2.323495738455667, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8033137917518616, + "num_tokens": 697000854.0, + "step": 18265 + }, + { + "epoch": 2.3236229487342577, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8182178735733032, + "num_tokens": 697041264.0, + "step": 18266 + }, + { + "epoch": 2.323750159012848, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7952369451522827, + "num_tokens": 697080079.0, + "step": 18267 + }, + { + "epoch": 2.3238773692914387, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.8008027076721191, + "num_tokens": 697123755.0, + "step": 18268 + }, + { + "epoch": 2.3240045795700293, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7706560492515564, + "num_tokens": 697162012.0, + "step": 18269 + }, + { + "epoch": 2.32413178984862, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8089919090270996, + "num_tokens": 697200188.0, + "step": 18270 + }, + { + "epoch": 2.3242590001272103, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.808659553527832, + "num_tokens": 697240536.0, + "step": 18271 + }, + { + "epoch": 2.324386210405801, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8095682859420776, + "num_tokens": 697279459.0, + "step": 18272 + }, + { + "epoch": 2.3245134206843914, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.8235284090042114, + "num_tokens": 697312594.0, + "step": 18273 + }, + { + "epoch": 2.324640630962982, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7792438268661499, + "num_tokens": 697357834.0, + "step": 18274 + }, + { + "epoch": 2.3247678412415724, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8026680946350098, + "num_tokens": 697401522.0, + "step": 18275 + }, + { + "epoch": 2.324895051520163, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7979099750518799, + "num_tokens": 697440415.0, + "step": 18276 + }, + { + "epoch": 2.3250222617987535, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.77503502368927, + "num_tokens": 697477609.0, + "step": 18277 + }, + { + "epoch": 2.325149472077344, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7891625761985779, + "num_tokens": 697515566.0, + "step": 18278 + }, + { + "epoch": 2.3252766823559345, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8005053997039795, + "num_tokens": 697549537.0, + "step": 18279 + }, + { + "epoch": 2.325403892634525, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.801344633102417, + "num_tokens": 697582618.0, + "step": 18280 + }, + { + "epoch": 2.325531102913115, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8157267570495605, + "num_tokens": 697622234.0, + "step": 18281 + }, + { + "epoch": 2.325658313191706, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7913750410079956, + "num_tokens": 697657504.0, + "step": 18282 + }, + { + "epoch": 2.325785523470296, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.787468671798706, + "num_tokens": 697699496.0, + "step": 18283 + }, + { + "epoch": 2.3259127337488867, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8127158880233765, + "num_tokens": 697739969.0, + "step": 18284 + }, + { + "epoch": 2.3260399440274773, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.78426593542099, + "num_tokens": 697774346.0, + "step": 18285 + }, + { + "epoch": 2.326167154306068, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7787978053092957, + "num_tokens": 697813013.0, + "step": 18286 + }, + { + "epoch": 2.3262943645846583, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.789950966835022, + "num_tokens": 697854163.0, + "step": 18287 + }, + { + "epoch": 2.326421574863249, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8072969317436218, + "num_tokens": 697898049.0, + "step": 18288 + }, + { + "epoch": 2.3265487851418394, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.799362301826477, + "num_tokens": 697936086.0, + "step": 18289 + }, + { + "epoch": 2.32667599542043, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7730305194854736, + "num_tokens": 697981529.0, + "step": 18290 + }, + { + "epoch": 2.3268032056990204, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8098165988922119, + "num_tokens": 698016047.0, + "step": 18291 + }, + { + "epoch": 2.326930415977611, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8143068552017212, + "num_tokens": 698052408.0, + "step": 18292 + }, + { + "epoch": 2.3270576262562015, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7904914617538452, + "num_tokens": 698089751.0, + "step": 18293 + }, + { + "epoch": 2.327184836534792, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7916731834411621, + "num_tokens": 698121274.0, + "step": 18294 + }, + { + "epoch": 2.3273120468133826, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8070186972618103, + "num_tokens": 698155880.0, + "step": 18295 + }, + { + "epoch": 2.327439257091973, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8036699891090393, + "num_tokens": 698191448.0, + "step": 18296 + }, + { + "epoch": 2.3275664673705636, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7900887131690979, + "num_tokens": 698228437.0, + "step": 18297 + }, + { + "epoch": 2.327693677649154, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7881482839584351, + "num_tokens": 698268714.0, + "step": 18298 + }, + { + "epoch": 2.3278208879277447, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7961462140083313, + "num_tokens": 698305575.0, + "step": 18299 + }, + { + "epoch": 2.327948098206335, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7971864938735962, + "num_tokens": 698347701.0, + "step": 18300 + }, + { + "epoch": 2.3280753084849257, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8150718212127686, + "num_tokens": 698385906.0, + "step": 18301 + }, + { + "epoch": 2.3282025187635162, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7978794574737549, + "num_tokens": 698415511.0, + "step": 18302 + }, + { + "epoch": 2.3283297290421068, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7924365401268005, + "num_tokens": 698460528.0, + "step": 18303 + }, + { + "epoch": 2.3284569393206973, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7834982872009277, + "num_tokens": 698500800.0, + "step": 18304 + }, + { + "epoch": 2.328584149599288, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7976580858230591, + "num_tokens": 698539373.0, + "step": 18305 + }, + { + "epoch": 2.328711359877878, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8144490122795105, + "num_tokens": 698577067.0, + "step": 18306 + }, + { + "epoch": 2.328838570156469, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7953197360038757, + "num_tokens": 698613515.0, + "step": 18307 + }, + { + "epoch": 2.328965780435059, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7949265241622925, + "num_tokens": 698647278.0, + "step": 18308 + }, + { + "epoch": 2.3290929907136495, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8064224123954773, + "num_tokens": 698688738.0, + "step": 18309 + }, + { + "epoch": 2.32922020099224, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8059258460998535, + "num_tokens": 698725003.0, + "step": 18310 + }, + { + "epoch": 2.3293474112708306, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.783034086227417, + "num_tokens": 698768205.0, + "step": 18311 + }, + { + "epoch": 2.329474621549421, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7774984836578369, + "num_tokens": 698805352.0, + "step": 18312 + }, + { + "epoch": 2.3296018318280116, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8127524852752686, + "num_tokens": 698836610.0, + "step": 18313 + }, + { + "epoch": 2.329729042106602, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8081268072128296, + "num_tokens": 698877514.0, + "step": 18314 + }, + { + "epoch": 2.3298562523851927, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8062148690223694, + "num_tokens": 698916699.0, + "step": 18315 + }, + { + "epoch": 2.329983462663783, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8003464341163635, + "num_tokens": 698953568.0, + "step": 18316 + }, + { + "epoch": 2.3301106729423737, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7906122207641602, + "num_tokens": 698990562.0, + "step": 18317 + }, + { + "epoch": 2.3302378832209643, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.803370475769043, + "num_tokens": 699024095.0, + "step": 18318 + }, + { + "epoch": 2.330365093499555, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7792091965675354, + "num_tokens": 699062327.0, + "step": 18319 + }, + { + "epoch": 2.3304923037781453, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8005751967430115, + "num_tokens": 699101591.0, + "step": 18320 + }, + { + "epoch": 2.330619514056736, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7873667478561401, + "num_tokens": 699134836.0, + "step": 18321 + }, + { + "epoch": 2.3307467243353264, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7863909006118774, + "num_tokens": 699171218.0, + "step": 18322 + }, + { + "epoch": 2.330873934613917, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.8195763826370239, + "num_tokens": 699200891.0, + "step": 18323 + }, + { + "epoch": 2.3310011448925074, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.8166199326515198, + "num_tokens": 699238080.0, + "step": 18324 + }, + { + "epoch": 2.331128355171098, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7882195711135864, + "num_tokens": 699276051.0, + "step": 18325 + }, + { + "epoch": 2.3312555654496885, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8086829781532288, + "num_tokens": 699315721.0, + "step": 18326 + }, + { + "epoch": 2.331382775728279, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8113446831703186, + "num_tokens": 699350342.0, + "step": 18327 + }, + { + "epoch": 2.3315099860068695, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8027936220169067, + "num_tokens": 699390301.0, + "step": 18328 + }, + { + "epoch": 2.3316371962854596, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7829650640487671, + "num_tokens": 699427135.0, + "step": 18329 + }, + { + "epoch": 2.3317644065640506, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8065508604049683, + "num_tokens": 699466461.0, + "step": 18330 + }, + { + "epoch": 2.3318916168426407, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7945328950881958, + "num_tokens": 699502446.0, + "step": 18331 + }, + { + "epoch": 2.332018827121231, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7982409596443176, + "num_tokens": 699546357.0, + "step": 18332 + }, + { + "epoch": 2.3321460373998217, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7993856072425842, + "num_tokens": 699581879.0, + "step": 18333 + }, + { + "epoch": 2.3322732476784123, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8098164796829224, + "num_tokens": 699616287.0, + "step": 18334 + }, + { + "epoch": 2.332400457957003, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7972496747970581, + "num_tokens": 699655409.0, + "step": 18335 + }, + { + "epoch": 2.3325276682355933, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7771397829055786, + "num_tokens": 699701420.0, + "step": 18336 + }, + { + "epoch": 2.332654878514184, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8025131225585938, + "num_tokens": 699736151.0, + "step": 18337 + }, + { + "epoch": 2.3327820887927744, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7868315577507019, + "num_tokens": 699770111.0, + "step": 18338 + }, + { + "epoch": 2.332909299071365, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7957919239997864, + "num_tokens": 699807442.0, + "step": 18339 + }, + { + "epoch": 2.3330365093499554, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8005112409591675, + "num_tokens": 699849364.0, + "step": 18340 + }, + { + "epoch": 2.333163719628546, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8162631988525391, + "num_tokens": 699886234.0, + "step": 18341 + }, + { + "epoch": 2.3332909299071365, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7823131680488586, + "num_tokens": 699925660.0, + "step": 18342 + }, + { + "epoch": 2.333418140185727, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7728080153465271, + "num_tokens": 699967894.0, + "step": 18343 + }, + { + "epoch": 2.3335453504643175, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7936328053474426, + "num_tokens": 699999211.0, + "step": 18344 + }, + { + "epoch": 2.333672560742908, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.80633544921875, + "num_tokens": 700037639.0, + "step": 18345 + }, + { + "epoch": 2.3337997710214986, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.792665421962738, + "num_tokens": 700075677.0, + "step": 18346 + }, + { + "epoch": 2.333926981300089, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8070307970046997, + "num_tokens": 700115697.0, + "step": 18347 + }, + { + "epoch": 2.3340541915786797, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7815284729003906, + "num_tokens": 700152553.0, + "step": 18348 + }, + { + "epoch": 2.33418140185727, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8078595399856567, + "num_tokens": 700190958.0, + "step": 18349 + }, + { + "epoch": 2.3343086121358607, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7943161725997925, + "num_tokens": 700224244.0, + "step": 18350 + }, + { + "epoch": 2.3344358224144512, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.5442, + "mean_token_accuracy": 0.8325815200805664, + "num_tokens": 700259736.0, + "step": 18351 + }, + { + "epoch": 2.3345630326930418, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7977958917617798, + "num_tokens": 700293624.0, + "step": 18352 + }, + { + "epoch": 2.3346902429716323, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8003588914871216, + "num_tokens": 700328052.0, + "step": 18353 + }, + { + "epoch": 2.3348174532502224, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7894279956817627, + "num_tokens": 700364652.0, + "step": 18354 + }, + { + "epoch": 2.3349446635288134, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.81385338306427, + "num_tokens": 700403698.0, + "step": 18355 + }, + { + "epoch": 2.3350718738074034, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8142175674438477, + "num_tokens": 700439391.0, + "step": 18356 + }, + { + "epoch": 2.335199084085994, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7994132041931152, + "num_tokens": 700475095.0, + "step": 18357 + }, + { + "epoch": 2.3353262943645845, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8049870729446411, + "num_tokens": 700514167.0, + "step": 18358 + }, + { + "epoch": 2.335453504643175, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8067228198051453, + "num_tokens": 700548333.0, + "step": 18359 + }, + { + "epoch": 2.3355807149217656, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8027085065841675, + "num_tokens": 700594261.0, + "step": 18360 + }, + { + "epoch": 2.335707925200356, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8078234195709229, + "num_tokens": 700632513.0, + "step": 18361 + }, + { + "epoch": 2.3358351354789466, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8027772903442383, + "num_tokens": 700672430.0, + "step": 18362 + }, + { + "epoch": 2.335962345757537, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8122270107269287, + "num_tokens": 700714321.0, + "step": 18363 + }, + { + "epoch": 2.3360895560361277, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.80365389585495, + "num_tokens": 700751714.0, + "step": 18364 + }, + { + "epoch": 2.336216766314718, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7961156964302063, + "num_tokens": 700794087.0, + "step": 18365 + }, + { + "epoch": 2.3363439765933087, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7900835275650024, + "num_tokens": 700832031.0, + "step": 18366 + }, + { + "epoch": 2.3364711868718993, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8008691072463989, + "num_tokens": 700867175.0, + "step": 18367 + }, + { + "epoch": 2.33659839715049, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8075167536735535, + "num_tokens": 700906762.0, + "step": 18368 + }, + { + "epoch": 2.3367256074290803, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8012480139732361, + "num_tokens": 700944316.0, + "step": 18369 + }, + { + "epoch": 2.336852817707671, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7988535165786743, + "num_tokens": 700979369.0, + "step": 18370 + }, + { + "epoch": 2.3369800279862614, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7936860918998718, + "num_tokens": 701016213.0, + "step": 18371 + }, + { + "epoch": 2.337107238264852, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7956104278564453, + "num_tokens": 701055270.0, + "step": 18372 + }, + { + "epoch": 2.3372344485434424, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7995114326477051, + "num_tokens": 701091058.0, + "step": 18373 + }, + { + "epoch": 2.337361658822033, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.8027747869491577, + "num_tokens": 701129446.0, + "step": 18374 + }, + { + "epoch": 2.3374888691006235, + "grad_norm": 10.625, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8071459531784058, + "num_tokens": 701168726.0, + "step": 18375 + }, + { + "epoch": 2.337616079379214, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7938333749771118, + "num_tokens": 701204041.0, + "step": 18376 + }, + { + "epoch": 2.3377432896578045, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7785241007804871, + "num_tokens": 701244660.0, + "step": 18377 + }, + { + "epoch": 2.337870499936395, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.785645604133606, + "num_tokens": 701284830.0, + "step": 18378 + }, + { + "epoch": 2.337997710214985, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8030790686607361, + "num_tokens": 701319720.0, + "step": 18379 + }, + { + "epoch": 2.338124920493576, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7794848680496216, + "num_tokens": 701361770.0, + "step": 18380 + }, + { + "epoch": 2.338252130772166, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.799027681350708, + "num_tokens": 701399777.0, + "step": 18381 + }, + { + "epoch": 2.3383793410507567, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7939126491546631, + "num_tokens": 701432680.0, + "step": 18382 + }, + { + "epoch": 2.3385065513293473, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8064066171646118, + "num_tokens": 701474101.0, + "step": 18383 + }, + { + "epoch": 2.338633761607938, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8178622126579285, + "num_tokens": 701517568.0, + "step": 18384 + }, + { + "epoch": 2.3387609718865283, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7936780452728271, + "num_tokens": 701558466.0, + "step": 18385 + }, + { + "epoch": 2.338888182165119, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7910059690475464, + "num_tokens": 701591104.0, + "step": 18386 + }, + { + "epoch": 2.3390153924437094, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7961785793304443, + "num_tokens": 701633242.0, + "step": 18387 + }, + { + "epoch": 2.3391426027223, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8003351092338562, + "num_tokens": 701668864.0, + "step": 18388 + }, + { + "epoch": 2.3392698130008904, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8000025749206543, + "num_tokens": 701703648.0, + "step": 18389 + }, + { + "epoch": 2.339397023279481, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7961554527282715, + "num_tokens": 701744417.0, + "step": 18390 + }, + { + "epoch": 2.3395242335580715, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7653869390487671, + "num_tokens": 701783541.0, + "step": 18391 + }, + { + "epoch": 2.339651443836662, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7936379909515381, + "num_tokens": 701825054.0, + "step": 18392 + }, + { + "epoch": 2.3397786541152525, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8097976446151733, + "num_tokens": 701866173.0, + "step": 18393 + }, + { + "epoch": 2.339905864393843, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7917640209197998, + "num_tokens": 701904812.0, + "step": 18394 + }, + { + "epoch": 2.3400330746724336, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7833577394485474, + "num_tokens": 701942590.0, + "step": 18395 + }, + { + "epoch": 2.340160284951024, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8055354356765747, + "num_tokens": 701980773.0, + "step": 18396 + }, + { + "epoch": 2.3402874952296147, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8081591725349426, + "num_tokens": 702014508.0, + "step": 18397 + }, + { + "epoch": 2.340414705508205, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8119105100631714, + "num_tokens": 702055754.0, + "step": 18398 + }, + { + "epoch": 2.3405419157867957, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8026807308197021, + "num_tokens": 702100022.0, + "step": 18399 + }, + { + "epoch": 2.3406691260653862, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7842295169830322, + "num_tokens": 702133217.0, + "step": 18400 + }, + { + "epoch": 2.3407963363439768, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8157954216003418, + "num_tokens": 702176998.0, + "step": 18401 + }, + { + "epoch": 2.3409235466225673, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8047775030136108, + "num_tokens": 702216219.0, + "step": 18402 + }, + { + "epoch": 2.341050756901158, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7893218994140625, + "num_tokens": 702246502.0, + "step": 18403 + }, + { + "epoch": 2.341177967179748, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8095446825027466, + "num_tokens": 702277990.0, + "step": 18404 + }, + { + "epoch": 2.341305177458339, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7986081838607788, + "num_tokens": 702312742.0, + "step": 18405 + }, + { + "epoch": 2.341432387736929, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.820679783821106, + "num_tokens": 702348575.0, + "step": 18406 + }, + { + "epoch": 2.3415595980155195, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7852269411087036, + "num_tokens": 702385956.0, + "step": 18407 + }, + { + "epoch": 2.34168680829411, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.801543653011322, + "num_tokens": 702423973.0, + "step": 18408 + }, + { + "epoch": 2.3418140185727006, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7958394885063171, + "num_tokens": 702455508.0, + "step": 18409 + }, + { + "epoch": 2.341941228851291, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8138238191604614, + "num_tokens": 702493365.0, + "step": 18410 + }, + { + "epoch": 2.3420684391298816, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8051596879959106, + "num_tokens": 702528036.0, + "step": 18411 + }, + { + "epoch": 2.342195649408472, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7928714752197266, + "num_tokens": 702571238.0, + "step": 18412 + }, + { + "epoch": 2.3423228596870627, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7910649180412292, + "num_tokens": 702611247.0, + "step": 18413 + }, + { + "epoch": 2.342450069965653, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7973393201828003, + "num_tokens": 702649661.0, + "step": 18414 + }, + { + "epoch": 2.3425772802442437, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7940394878387451, + "num_tokens": 702693573.0, + "step": 18415 + }, + { + "epoch": 2.3427044905228342, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.8140044212341309, + "num_tokens": 702735351.0, + "step": 18416 + }, + { + "epoch": 2.3428317008014248, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8064351677894592, + "num_tokens": 702773406.0, + "step": 18417 + }, + { + "epoch": 2.3429589110800153, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8063819408416748, + "num_tokens": 702809939.0, + "step": 18418 + }, + { + "epoch": 2.343086121358606, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7952624559402466, + "num_tokens": 702851120.0, + "step": 18419 + }, + { + "epoch": 2.3432133316371964, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8058385252952576, + "num_tokens": 702889492.0, + "step": 18420 + }, + { + "epoch": 2.343340541915787, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7947827577590942, + "num_tokens": 702930486.0, + "step": 18421 + }, + { + "epoch": 2.3434677521943774, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7925664186477661, + "num_tokens": 702969866.0, + "step": 18422 + }, + { + "epoch": 2.343594962472968, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7942070960998535, + "num_tokens": 703013703.0, + "step": 18423 + }, + { + "epoch": 2.3437221727515585, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7930066585540771, + "num_tokens": 703054849.0, + "step": 18424 + }, + { + "epoch": 2.343849383030149, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7901577353477478, + "num_tokens": 703088250.0, + "step": 18425 + }, + { + "epoch": 2.3439765933087395, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.794413685798645, + "num_tokens": 703124072.0, + "step": 18426 + }, + { + "epoch": 2.3441038035873296, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7807667255401611, + "num_tokens": 703158869.0, + "step": 18427 + }, + { + "epoch": 2.3442310138659206, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7993287444114685, + "num_tokens": 703193342.0, + "step": 18428 + }, + { + "epoch": 2.3443582241445107, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8157424926757812, + "num_tokens": 703229899.0, + "step": 18429 + }, + { + "epoch": 2.344485434423101, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.7995455265045166, + "num_tokens": 703266735.0, + "step": 18430 + }, + { + "epoch": 2.3446126447016917, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7926222085952759, + "num_tokens": 703307515.0, + "step": 18431 + }, + { + "epoch": 2.3447398549802823, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.8117986917495728, + "num_tokens": 703341237.0, + "step": 18432 + }, + { + "epoch": 2.344867065258873, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8076897859573364, + "num_tokens": 703378900.0, + "step": 18433 + }, + { + "epoch": 2.3449942755374633, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8124706745147705, + "num_tokens": 703415260.0, + "step": 18434 + }, + { + "epoch": 2.345121485816054, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8088200092315674, + "num_tokens": 703452399.0, + "step": 18435 + }, + { + "epoch": 2.3452486960946444, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8032676577568054, + "num_tokens": 703492700.0, + "step": 18436 + }, + { + "epoch": 2.345375906373235, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8164074420928955, + "num_tokens": 703530805.0, + "step": 18437 + }, + { + "epoch": 2.3455031166518254, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8085795640945435, + "num_tokens": 703571894.0, + "step": 18438 + }, + { + "epoch": 2.345630326930416, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8104035258293152, + "num_tokens": 703612583.0, + "step": 18439 + }, + { + "epoch": 2.3457575372090065, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.792218804359436, + "num_tokens": 703649100.0, + "step": 18440 + }, + { + "epoch": 2.345884747487597, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8103659152984619, + "num_tokens": 703680425.0, + "step": 18441 + }, + { + "epoch": 2.3460119577661875, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8046440482139587, + "num_tokens": 703721819.0, + "step": 18442 + }, + { + "epoch": 2.346139168044778, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7907055616378784, + "num_tokens": 703755432.0, + "step": 18443 + }, + { + "epoch": 2.3462663783233686, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8033881187438965, + "num_tokens": 703795903.0, + "step": 18444 + }, + { + "epoch": 2.346393588601959, + "grad_norm": 10.5625, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.8152663707733154, + "num_tokens": 703836074.0, + "step": 18445 + }, + { + "epoch": 2.3465207988805497, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8051204681396484, + "num_tokens": 703870609.0, + "step": 18446 + }, + { + "epoch": 2.34664800915914, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8055915832519531, + "num_tokens": 703909034.0, + "step": 18447 + }, + { + "epoch": 2.3467752194377307, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7893115282058716, + "num_tokens": 703946976.0, + "step": 18448 + }, + { + "epoch": 2.3469024297163212, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8049858808517456, + "num_tokens": 703986379.0, + "step": 18449 + }, + { + "epoch": 2.3470296399949118, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8149628639221191, + "num_tokens": 704019461.0, + "step": 18450 + }, + { + "epoch": 2.3471568502735023, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7855557203292847, + "num_tokens": 704053411.0, + "step": 18451 + }, + { + "epoch": 2.3472840605520924, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8045686483383179, + "num_tokens": 704092633.0, + "step": 18452 + }, + { + "epoch": 2.3474112708306833, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8031749725341797, + "num_tokens": 704126955.0, + "step": 18453 + }, + { + "epoch": 2.3475384811092734, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8178252577781677, + "num_tokens": 704162222.0, + "step": 18454 + }, + { + "epoch": 2.347665691387864, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8015397787094116, + "num_tokens": 704198949.0, + "step": 18455 + }, + { + "epoch": 2.3477929016664545, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7929984331130981, + "num_tokens": 704235327.0, + "step": 18456 + }, + { + "epoch": 2.347920111945045, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8066929578781128, + "num_tokens": 704272783.0, + "step": 18457 + }, + { + "epoch": 2.3480473222236355, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7802842855453491, + "num_tokens": 704314396.0, + "step": 18458 + }, + { + "epoch": 2.348174532502226, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8152209520339966, + "num_tokens": 704346903.0, + "step": 18459 + }, + { + "epoch": 2.3483017427808166, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8070786595344543, + "num_tokens": 704380456.0, + "step": 18460 + }, + { + "epoch": 2.348428953059407, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.813788652420044, + "num_tokens": 704416403.0, + "step": 18461 + }, + { + "epoch": 2.3485561633379977, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.803410530090332, + "num_tokens": 704460431.0, + "step": 18462 + }, + { + "epoch": 2.348683373616588, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8128374814987183, + "num_tokens": 704497872.0, + "step": 18463 + }, + { + "epoch": 2.3488105838951787, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8016522526741028, + "num_tokens": 704538180.0, + "step": 18464 + }, + { + "epoch": 2.3489377941737692, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8161531686782837, + "num_tokens": 704578342.0, + "step": 18465 + }, + { + "epoch": 2.3490650044523598, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8099682331085205, + "num_tokens": 704611792.0, + "step": 18466 + }, + { + "epoch": 2.3491922147309503, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.804394543170929, + "num_tokens": 704649898.0, + "step": 18467 + }, + { + "epoch": 2.349319425009541, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8130000829696655, + "num_tokens": 704690632.0, + "step": 18468 + }, + { + "epoch": 2.3494466352881314, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7898674607276917, + "num_tokens": 704722955.0, + "step": 18469 + }, + { + "epoch": 2.349573845566722, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8018428087234497, + "num_tokens": 704762678.0, + "step": 18470 + }, + { + "epoch": 2.3497010558453124, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.5668, + "mean_token_accuracy": 0.8287010192871094, + "num_tokens": 704805440.0, + "step": 18471 + }, + { + "epoch": 2.349828266123903, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.781194269657135, + "num_tokens": 704848153.0, + "step": 18472 + }, + { + "epoch": 2.3499554764024935, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7982544898986816, + "num_tokens": 704886739.0, + "step": 18473 + }, + { + "epoch": 2.350082686681084, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8044242262840271, + "num_tokens": 704926145.0, + "step": 18474 + }, + { + "epoch": 2.3502098969596745, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7844946384429932, + "num_tokens": 704964135.0, + "step": 18475 + }, + { + "epoch": 2.350337107238265, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.787749171257019, + "num_tokens": 705002171.0, + "step": 18476 + }, + { + "epoch": 2.350464317516855, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.8175490498542786, + "num_tokens": 705040660.0, + "step": 18477 + }, + { + "epoch": 2.350591527795446, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8066262006759644, + "num_tokens": 705073655.0, + "step": 18478 + }, + { + "epoch": 2.350718738074036, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7968525886535645, + "num_tokens": 705115497.0, + "step": 18479 + }, + { + "epoch": 2.3508459483526267, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.821171224117279, + "num_tokens": 705152642.0, + "step": 18480 + }, + { + "epoch": 2.3509731586312173, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7816749811172485, + "num_tokens": 705187282.0, + "step": 18481 + }, + { + "epoch": 2.351100368909808, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8053652048110962, + "num_tokens": 705225635.0, + "step": 18482 + }, + { + "epoch": 2.3512275791883983, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8060111999511719, + "num_tokens": 705260376.0, + "step": 18483 + }, + { + "epoch": 2.351354789466989, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7961027026176453, + "num_tokens": 705298799.0, + "step": 18484 + }, + { + "epoch": 2.3514819997455794, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7781929969787598, + "num_tokens": 705332570.0, + "step": 18485 + }, + { + "epoch": 2.35160921002417, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.8028914332389832, + "num_tokens": 705371214.0, + "step": 18486 + }, + { + "epoch": 2.3517364203027604, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7956594824790955, + "num_tokens": 705405051.0, + "step": 18487 + }, + { + "epoch": 2.351863630581351, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7952044010162354, + "num_tokens": 705442460.0, + "step": 18488 + }, + { + "epoch": 2.3519908408599415, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7984838485717773, + "num_tokens": 705482929.0, + "step": 18489 + }, + { + "epoch": 2.352118051138532, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7895423173904419, + "num_tokens": 705525313.0, + "step": 18490 + }, + { + "epoch": 2.3522452614171225, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7938444018363953, + "num_tokens": 705563730.0, + "step": 18491 + }, + { + "epoch": 2.352372471695713, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8009486794471741, + "num_tokens": 705602459.0, + "step": 18492 + }, + { + "epoch": 2.3524996819743036, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7988067865371704, + "num_tokens": 705636223.0, + "step": 18493 + }, + { + "epoch": 2.352626892252894, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.8028828501701355, + "num_tokens": 705676178.0, + "step": 18494 + }, + { + "epoch": 2.3527541025314846, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8120988011360168, + "num_tokens": 705716687.0, + "step": 18495 + }, + { + "epoch": 2.352881312810075, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8151768445968628, + "num_tokens": 705759163.0, + "step": 18496 + }, + { + "epoch": 2.3530085230886657, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7993268966674805, + "num_tokens": 705795982.0, + "step": 18497 + }, + { + "epoch": 2.3531357333672562, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8042098879814148, + "num_tokens": 705835601.0, + "step": 18498 + }, + { + "epoch": 2.3532629436458468, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7985180616378784, + "num_tokens": 705876858.0, + "step": 18499 + }, + { + "epoch": 2.353390153924437, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.792576014995575, + "num_tokens": 705921944.0, + "step": 18500 + }, + { + "epoch": 2.353517364203028, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8023496270179749, + "num_tokens": 705967999.0, + "step": 18501 + }, + { + "epoch": 2.353644574481618, + "grad_norm": 61.0, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7888186573982239, + "num_tokens": 706003370.0, + "step": 18502 + }, + { + "epoch": 2.353771784760209, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7863861918449402, + "num_tokens": 706036551.0, + "step": 18503 + }, + { + "epoch": 2.353898995038799, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.8206323385238647, + "num_tokens": 706075965.0, + "step": 18504 + }, + { + "epoch": 2.3540262053173895, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7925020456314087, + "num_tokens": 706117943.0, + "step": 18505 + }, + { + "epoch": 2.35415341559598, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8059309720993042, + "num_tokens": 706157183.0, + "step": 18506 + }, + { + "epoch": 2.3542806258745705, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8069848418235779, + "num_tokens": 706196878.0, + "step": 18507 + }, + { + "epoch": 2.354407836153161, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7803734540939331, + "num_tokens": 706231102.0, + "step": 18508 + }, + { + "epoch": 2.3545350464317516, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8110494613647461, + "num_tokens": 706273372.0, + "step": 18509 + }, + { + "epoch": 2.354662256710342, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7836872935295105, + "num_tokens": 706308106.0, + "step": 18510 + }, + { + "epoch": 2.3547894669889327, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8069570064544678, + "num_tokens": 706344267.0, + "step": 18511 + }, + { + "epoch": 2.354916677267523, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8211216330528259, + "num_tokens": 706385019.0, + "step": 18512 + }, + { + "epoch": 2.3550438875461137, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7969563007354736, + "num_tokens": 706424913.0, + "step": 18513 + }, + { + "epoch": 2.3551710978247042, + "grad_norm": 66.0, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8070343136787415, + "num_tokens": 706466064.0, + "step": 18514 + }, + { + "epoch": 2.3552983081032948, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8075678944587708, + "num_tokens": 706504572.0, + "step": 18515 + }, + { + "epoch": 2.3554255183818853, + "grad_norm": 87.5, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7928110361099243, + "num_tokens": 706543901.0, + "step": 18516 + }, + { + "epoch": 2.355552728660476, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7916839718818665, + "num_tokens": 706579095.0, + "step": 18517 + }, + { + "epoch": 2.3556799389390664, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.790971040725708, + "num_tokens": 706614078.0, + "step": 18518 + }, + { + "epoch": 2.355807149217657, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8041509985923767, + "num_tokens": 706652596.0, + "step": 18519 + }, + { + "epoch": 2.3559343594962474, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.5641, + "mean_token_accuracy": 0.8250061273574829, + "num_tokens": 706691297.0, + "step": 18520 + }, + { + "epoch": 2.356061569774838, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7864111661911011, + "num_tokens": 706727344.0, + "step": 18521 + }, + { + "epoch": 2.3561887800534285, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8111754655838013, + "num_tokens": 706770411.0, + "step": 18522 + }, + { + "epoch": 2.356315990332019, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8062945604324341, + "num_tokens": 706816215.0, + "step": 18523 + }, + { + "epoch": 2.3564432006106095, + "grad_norm": 70.0, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.79777991771698, + "num_tokens": 706857400.0, + "step": 18524 + }, + { + "epoch": 2.3565704108891996, + "grad_norm": 86.5, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7913538217544556, + "num_tokens": 706892715.0, + "step": 18525 + }, + { + "epoch": 2.3566976211677906, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8001185655593872, + "num_tokens": 706934373.0, + "step": 18526 + }, + { + "epoch": 2.3568248314463807, + "grad_norm": 56.75, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8054687976837158, + "num_tokens": 706976885.0, + "step": 18527 + }, + { + "epoch": 2.356952041724971, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8122236728668213, + "num_tokens": 707016957.0, + "step": 18528 + }, + { + "epoch": 2.3570792520035617, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8112664222717285, + "num_tokens": 707053687.0, + "step": 18529 + }, + { + "epoch": 2.3572064622821522, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.791003942489624, + "num_tokens": 707087327.0, + "step": 18530 + }, + { + "epoch": 2.3573336725607428, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8129298686981201, + "num_tokens": 707124479.0, + "step": 18531 + }, + { + "epoch": 2.3574608828393333, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8054671287536621, + "num_tokens": 707162452.0, + "step": 18532 + }, + { + "epoch": 2.357588093117924, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8089542388916016, + "num_tokens": 707196765.0, + "step": 18533 + }, + { + "epoch": 2.3577153033965144, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8042306303977966, + "num_tokens": 707230151.0, + "step": 18534 + }, + { + "epoch": 2.357842513675105, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7802899479866028, + "num_tokens": 707264373.0, + "step": 18535 + }, + { + "epoch": 2.3579697239536954, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7993114590644836, + "num_tokens": 707304844.0, + "step": 18536 + }, + { + "epoch": 2.358096934232286, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8014443516731262, + "num_tokens": 707347081.0, + "step": 18537 + }, + { + "epoch": 2.3582241445108765, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7928029298782349, + "num_tokens": 707388243.0, + "step": 18538 + }, + { + "epoch": 2.358351354789467, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7918245792388916, + "num_tokens": 707427109.0, + "step": 18539 + }, + { + "epoch": 2.3584785650680575, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.8011740446090698, + "num_tokens": 707462502.0, + "step": 18540 + }, + { + "epoch": 2.358605775346648, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7942625284194946, + "num_tokens": 707502861.0, + "step": 18541 + }, + { + "epoch": 2.3587329856252386, + "grad_norm": 70.5, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.775310218334198, + "num_tokens": 707535868.0, + "step": 18542 + }, + { + "epoch": 2.358860195903829, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.8011922836303711, + "num_tokens": 707566128.0, + "step": 18543 + }, + { + "epoch": 2.3589874061824196, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8023239970207214, + "num_tokens": 707605113.0, + "step": 18544 + }, + { + "epoch": 2.35911461646101, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8014143109321594, + "num_tokens": 707647968.0, + "step": 18545 + }, + { + "epoch": 2.3592418267396007, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8111705780029297, + "num_tokens": 707682183.0, + "step": 18546 + }, + { + "epoch": 2.3593690370181912, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8038661479949951, + "num_tokens": 707720938.0, + "step": 18547 + }, + { + "epoch": 2.3594962472967818, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7875652313232422, + "num_tokens": 707762624.0, + "step": 18548 + }, + { + "epoch": 2.3596234575753723, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.78668212890625, + "num_tokens": 707798406.0, + "step": 18549 + }, + { + "epoch": 2.3597506678539624, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7954654693603516, + "num_tokens": 707835935.0, + "step": 18550 + }, + { + "epoch": 2.3598778781325533, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8107857704162598, + "num_tokens": 707872971.0, + "step": 18551 + }, + { + "epoch": 2.3600050884111434, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8115969896316528, + "num_tokens": 707909798.0, + "step": 18552 + }, + { + "epoch": 2.360132298689734, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8066633343696594, + "num_tokens": 707951130.0, + "step": 18553 + }, + { + "epoch": 2.3602595089683245, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7892652750015259, + "num_tokens": 707989502.0, + "step": 18554 + }, + { + "epoch": 2.360386719246915, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7966517210006714, + "num_tokens": 708023521.0, + "step": 18555 + }, + { + "epoch": 2.3605139295255055, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.800054669380188, + "num_tokens": 708057594.0, + "step": 18556 + }, + { + "epoch": 2.360641139804096, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8029428124427795, + "num_tokens": 708101393.0, + "step": 18557 + }, + { + "epoch": 2.3607683500826866, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8076683282852173, + "num_tokens": 708137141.0, + "step": 18558 + }, + { + "epoch": 2.360895560361277, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8039184808731079, + "num_tokens": 708173497.0, + "step": 18559 + }, + { + "epoch": 2.3610227706398677, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8047376871109009, + "num_tokens": 708209412.0, + "step": 18560 + }, + { + "epoch": 2.361149980918458, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8076026439666748, + "num_tokens": 708239675.0, + "step": 18561 + }, + { + "epoch": 2.3612771911970487, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7942400574684143, + "num_tokens": 708274918.0, + "step": 18562 + }, + { + "epoch": 2.3614044014756392, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7870471477508545, + "num_tokens": 708313645.0, + "step": 18563 + }, + { + "epoch": 2.3615316117542298, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8072531223297119, + "num_tokens": 708357985.0, + "step": 18564 + }, + { + "epoch": 2.3616588220328203, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7759201526641846, + "num_tokens": 708393606.0, + "step": 18565 + }, + { + "epoch": 2.361786032311411, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.8099992275238037, + "num_tokens": 708428941.0, + "step": 18566 + }, + { + "epoch": 2.3619132425900013, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8054660558700562, + "num_tokens": 708467847.0, + "step": 18567 + }, + { + "epoch": 2.362040452868592, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.8164460062980652, + "num_tokens": 708503212.0, + "step": 18568 + }, + { + "epoch": 2.3621676631471824, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.800132155418396, + "num_tokens": 708542897.0, + "step": 18569 + }, + { + "epoch": 2.362294873425773, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7819376587867737, + "num_tokens": 708574725.0, + "step": 18570 + }, + { + "epoch": 2.3624220837043635, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6018, + "mean_token_accuracy": 0.8176091909408569, + "num_tokens": 708613557.0, + "step": 18571 + }, + { + "epoch": 2.362549293982954, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7986758947372437, + "num_tokens": 708651872.0, + "step": 18572 + }, + { + "epoch": 2.3626765042615445, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7870421409606934, + "num_tokens": 708688654.0, + "step": 18573 + }, + { + "epoch": 2.362803714540135, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.803589940071106, + "num_tokens": 708723285.0, + "step": 18574 + }, + { + "epoch": 2.362930924818725, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8137375116348267, + "num_tokens": 708760617.0, + "step": 18575 + }, + { + "epoch": 2.363058135097316, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8024976253509521, + "num_tokens": 708790792.0, + "step": 18576 + }, + { + "epoch": 2.363185345375906, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7916901707649231, + "num_tokens": 708829935.0, + "step": 18577 + }, + { + "epoch": 2.3633125556544967, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7895498275756836, + "num_tokens": 708870578.0, + "step": 18578 + }, + { + "epoch": 2.3634397659330872, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8002554178237915, + "num_tokens": 708907044.0, + "step": 18579 + }, + { + "epoch": 2.3635669762116778, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7821896076202393, + "num_tokens": 708947443.0, + "step": 18580 + }, + { + "epoch": 2.3636941864902683, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8003638386726379, + "num_tokens": 708983174.0, + "step": 18581 + }, + { + "epoch": 2.363821396768859, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8033011555671692, + "num_tokens": 709021950.0, + "step": 18582 + }, + { + "epoch": 2.3639486070474494, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8098899126052856, + "num_tokens": 709057236.0, + "step": 18583 + }, + { + "epoch": 2.36407581732604, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8051798343658447, + "num_tokens": 709093376.0, + "step": 18584 + }, + { + "epoch": 2.3642030276046304, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7956605553627014, + "num_tokens": 709129502.0, + "step": 18585 + }, + { + "epoch": 2.364330237883221, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7979830503463745, + "num_tokens": 709162538.0, + "step": 18586 + }, + { + "epoch": 2.3644574481618115, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7919440865516663, + "num_tokens": 709201690.0, + "step": 18587 + }, + { + "epoch": 2.364584658440402, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7935649752616882, + "num_tokens": 709240593.0, + "step": 18588 + }, + { + "epoch": 2.3647118687189925, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.8145778179168701, + "num_tokens": 709278192.0, + "step": 18589 + }, + { + "epoch": 2.364839078997583, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.8226840496063232, + "num_tokens": 709311797.0, + "step": 18590 + }, + { + "epoch": 2.3649662892761736, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8007537126541138, + "num_tokens": 709350379.0, + "step": 18591 + }, + { + "epoch": 2.365093499554764, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8279306292533875, + "num_tokens": 709389030.0, + "step": 18592 + }, + { + "epoch": 2.3652207098333546, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7993307113647461, + "num_tokens": 709433692.0, + "step": 18593 + }, + { + "epoch": 2.365347920111945, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8004413843154907, + "num_tokens": 709470773.0, + "step": 18594 + }, + { + "epoch": 2.3654751303905357, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7968273162841797, + "num_tokens": 709506764.0, + "step": 18595 + }, + { + "epoch": 2.3656023406691262, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.798779308795929, + "num_tokens": 709546931.0, + "step": 18596 + }, + { + "epoch": 2.3657295509477168, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8025108575820923, + "num_tokens": 709582842.0, + "step": 18597 + }, + { + "epoch": 2.365856761226307, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8060811758041382, + "num_tokens": 709621574.0, + "step": 18598 + }, + { + "epoch": 2.365983971504898, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8233398199081421, + "num_tokens": 709662437.0, + "step": 18599 + }, + { + "epoch": 2.366111181783488, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8129482269287109, + "num_tokens": 709701318.0, + "step": 18600 + }, + { + "epoch": 2.3662383920620784, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8016170859336853, + "num_tokens": 709745156.0, + "step": 18601 + }, + { + "epoch": 2.366365602340669, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.793839693069458, + "num_tokens": 709785849.0, + "step": 18602 + }, + { + "epoch": 2.3664928126192595, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8091833591461182, + "num_tokens": 709830038.0, + "step": 18603 + }, + { + "epoch": 2.36662002289785, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.8011626601219177, + "num_tokens": 709871174.0, + "step": 18604 + }, + { + "epoch": 2.3667472331764405, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.822872519493103, + "num_tokens": 709905921.0, + "step": 18605 + }, + { + "epoch": 2.366874443455031, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.8129110932350159, + "num_tokens": 709939856.0, + "step": 18606 + }, + { + "epoch": 2.3670016537336216, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8166754245758057, + "num_tokens": 709978805.0, + "step": 18607 + }, + { + "epoch": 2.367128864012212, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8169485330581665, + "num_tokens": 710015015.0, + "step": 18608 + }, + { + "epoch": 2.3672560742908026, + "grad_norm": 58.25, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7887253761291504, + "num_tokens": 710054887.0, + "step": 18609 + }, + { + "epoch": 2.367383284569393, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8004984855651855, + "num_tokens": 710097012.0, + "step": 18610 + }, + { + "epoch": 2.3675104948479837, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7799218893051147, + "num_tokens": 710130912.0, + "step": 18611 + }, + { + "epoch": 2.3676377051265742, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.801331102848053, + "num_tokens": 710168880.0, + "step": 18612 + }, + { + "epoch": 2.3677649154051648, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8125317692756653, + "num_tokens": 710205619.0, + "step": 18613 + }, + { + "epoch": 2.3678921256837553, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8074616193771362, + "num_tokens": 710239475.0, + "step": 18614 + }, + { + "epoch": 2.368019335962346, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8059428930282593, + "num_tokens": 710276533.0, + "step": 18615 + }, + { + "epoch": 2.3681465462409363, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8216793537139893, + "num_tokens": 710316731.0, + "step": 18616 + }, + { + "epoch": 2.368273756519527, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7930643558502197, + "num_tokens": 710352310.0, + "step": 18617 + }, + { + "epoch": 2.3684009667981174, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7938442230224609, + "num_tokens": 710400203.0, + "step": 18618 + }, + { + "epoch": 2.368528177076708, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8013889789581299, + "num_tokens": 710433054.0, + "step": 18619 + }, + { + "epoch": 2.3686553873552985, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.805784285068512, + "num_tokens": 710466879.0, + "step": 18620 + }, + { + "epoch": 2.368782597633889, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7882174253463745, + "num_tokens": 710502808.0, + "step": 18621 + }, + { + "epoch": 2.3689098079124795, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7865613698959351, + "num_tokens": 710546190.0, + "step": 18622 + }, + { + "epoch": 2.3690370181910696, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.807402491569519, + "num_tokens": 710585556.0, + "step": 18623 + }, + { + "epoch": 2.3691642284696606, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8047788143157959, + "num_tokens": 710621366.0, + "step": 18624 + }, + { + "epoch": 2.3692914387482507, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8063552975654602, + "num_tokens": 710660217.0, + "step": 18625 + }, + { + "epoch": 2.369418649026841, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8013671040534973, + "num_tokens": 710695798.0, + "step": 18626 + }, + { + "epoch": 2.3695458593054317, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7862677574157715, + "num_tokens": 710732739.0, + "step": 18627 + }, + { + "epoch": 2.3696730695840222, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7964587211608887, + "num_tokens": 710769651.0, + "step": 18628 + }, + { + "epoch": 2.3698002798626128, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7969282269477844, + "num_tokens": 710809487.0, + "step": 18629 + }, + { + "epoch": 2.3699274901412033, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.806302547454834, + "num_tokens": 710849152.0, + "step": 18630 + }, + { + "epoch": 2.370054700419794, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7861005067825317, + "num_tokens": 710890933.0, + "step": 18631 + }, + { + "epoch": 2.3701819106983844, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8101770877838135, + "num_tokens": 710928262.0, + "step": 18632 + }, + { + "epoch": 2.370309120976975, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8060126304626465, + "num_tokens": 710962969.0, + "step": 18633 + }, + { + "epoch": 2.3704363312555654, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7996691465377808, + "num_tokens": 710998717.0, + "step": 18634 + }, + { + "epoch": 2.370563541534156, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7792419195175171, + "num_tokens": 711032819.0, + "step": 18635 + }, + { + "epoch": 2.3706907518127465, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7903357744216919, + "num_tokens": 711069395.0, + "step": 18636 + }, + { + "epoch": 2.370817962091337, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7994316220283508, + "num_tokens": 711105754.0, + "step": 18637 + }, + { + "epoch": 2.3709451723699275, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7978609800338745, + "num_tokens": 711145291.0, + "step": 18638 + }, + { + "epoch": 2.371072382648518, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8052624464035034, + "num_tokens": 711183431.0, + "step": 18639 + }, + { + "epoch": 2.3711995929271086, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7873985767364502, + "num_tokens": 711229754.0, + "step": 18640 + }, + { + "epoch": 2.371326803205699, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8036532402038574, + "num_tokens": 711269071.0, + "step": 18641 + }, + { + "epoch": 2.3714540134842896, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7974367141723633, + "num_tokens": 711312036.0, + "step": 18642 + }, + { + "epoch": 2.37158122376288, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8117740750312805, + "num_tokens": 711347945.0, + "step": 18643 + }, + { + "epoch": 2.3717084340414707, + "grad_norm": 74.5, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7770337462425232, + "num_tokens": 711388433.0, + "step": 18644 + }, + { + "epoch": 2.371835644320061, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8060529828071594, + "num_tokens": 711425821.0, + "step": 18645 + }, + { + "epoch": 2.3719628545986517, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7982594966888428, + "num_tokens": 711462725.0, + "step": 18646 + }, + { + "epoch": 2.3720900648772423, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8186190128326416, + "num_tokens": 711499987.0, + "step": 18647 + }, + { + "epoch": 2.3722172751558324, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7836480736732483, + "num_tokens": 711543658.0, + "step": 18648 + }, + { + "epoch": 2.3723444854344233, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8100779056549072, + "num_tokens": 711583875.0, + "step": 18649 + }, + { + "epoch": 2.3724716957130134, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7822951078414917, + "num_tokens": 711625543.0, + "step": 18650 + }, + { + "epoch": 2.372598905991604, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.79942387342453, + "num_tokens": 711660528.0, + "step": 18651 + }, + { + "epoch": 2.3727261162701945, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8118995428085327, + "num_tokens": 711695646.0, + "step": 18652 + }, + { + "epoch": 2.372853326548785, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7801103591918945, + "num_tokens": 711734691.0, + "step": 18653 + }, + { + "epoch": 2.3729805368273755, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8021901845932007, + "num_tokens": 711771518.0, + "step": 18654 + }, + { + "epoch": 2.373107747105966, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7910138368606567, + "num_tokens": 711808968.0, + "step": 18655 + }, + { + "epoch": 2.3732349573845566, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8085883259773254, + "num_tokens": 711842944.0, + "step": 18656 + }, + { + "epoch": 2.373362167663147, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7987270951271057, + "num_tokens": 711882740.0, + "step": 18657 + }, + { + "epoch": 2.3734893779417376, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8004924058914185, + "num_tokens": 711925362.0, + "step": 18658 + }, + { + "epoch": 2.373616588220328, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8121508955955505, + "num_tokens": 711971285.0, + "step": 18659 + }, + { + "epoch": 2.3737437984989187, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7989922165870667, + "num_tokens": 712013306.0, + "step": 18660 + }, + { + "epoch": 2.3738710087775092, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8041886687278748, + "num_tokens": 712050837.0, + "step": 18661 + }, + { + "epoch": 2.3739982190560998, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.817772388458252, + "num_tokens": 712085140.0, + "step": 18662 + }, + { + "epoch": 2.3741254293346903, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.5946, + "mean_token_accuracy": 0.8199481964111328, + "num_tokens": 712122495.0, + "step": 18663 + }, + { + "epoch": 2.374252639613281, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7982542514801025, + "num_tokens": 712161312.0, + "step": 18664 + }, + { + "epoch": 2.3743798498918713, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.800031304359436, + "num_tokens": 712197275.0, + "step": 18665 + }, + { + "epoch": 2.374507060170462, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7835538387298584, + "num_tokens": 712231309.0, + "step": 18666 + }, + { + "epoch": 2.3746342704490524, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7970633506774902, + "num_tokens": 712264864.0, + "step": 18667 + }, + { + "epoch": 2.374761480727643, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8012347221374512, + "num_tokens": 712303564.0, + "step": 18668 + }, + { + "epoch": 2.3748886910062335, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8114508390426636, + "num_tokens": 712341872.0, + "step": 18669 + }, + { + "epoch": 2.375015901284824, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7918198108673096, + "num_tokens": 712379835.0, + "step": 18670 + }, + { + "epoch": 2.3751431115634145, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7987059950828552, + "num_tokens": 712417462.0, + "step": 18671 + }, + { + "epoch": 2.375270321842005, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7749674916267395, + "num_tokens": 712461581.0, + "step": 18672 + }, + { + "epoch": 2.375397532120595, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7992448806762695, + "num_tokens": 712499615.0, + "step": 18673 + }, + { + "epoch": 2.375524742399186, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8094333410263062, + "num_tokens": 712533264.0, + "step": 18674 + }, + { + "epoch": 2.375651952677776, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.794654130935669, + "num_tokens": 712571496.0, + "step": 18675 + }, + { + "epoch": 2.3757791629563667, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8075897693634033, + "num_tokens": 712606510.0, + "step": 18676 + }, + { + "epoch": 2.3759063732349572, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8112049698829651, + "num_tokens": 712646218.0, + "step": 18677 + }, + { + "epoch": 2.3760335835135478, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7706071138381958, + "num_tokens": 712681905.0, + "step": 18678 + }, + { + "epoch": 2.3761607937921383, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.8157716393470764, + "num_tokens": 712719916.0, + "step": 18679 + }, + { + "epoch": 2.376288004070729, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8097702264785767, + "num_tokens": 712760849.0, + "step": 18680 + }, + { + "epoch": 2.3764152143493193, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8112210631370544, + "num_tokens": 712801199.0, + "step": 18681 + }, + { + "epoch": 2.37654242462791, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7949581742286682, + "num_tokens": 712843471.0, + "step": 18682 + }, + { + "epoch": 2.3766696349065004, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8027603030204773, + "num_tokens": 712880649.0, + "step": 18683 + }, + { + "epoch": 2.376796845185091, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7766205072402954, + "num_tokens": 712920574.0, + "step": 18684 + }, + { + "epoch": 2.3769240554636815, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8151558637619019, + "num_tokens": 712962544.0, + "step": 18685 + }, + { + "epoch": 2.377051265742272, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8026911020278931, + "num_tokens": 712999138.0, + "step": 18686 + }, + { + "epoch": 2.3771784760208625, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.802611231803894, + "num_tokens": 713041799.0, + "step": 18687 + }, + { + "epoch": 2.377305686299453, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7928647398948669, + "num_tokens": 713088248.0, + "step": 18688 + }, + { + "epoch": 2.3774328965780436, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8071980476379395, + "num_tokens": 713127239.0, + "step": 18689 + }, + { + "epoch": 2.377560106856634, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8072835206985474, + "num_tokens": 713168389.0, + "step": 18690 + }, + { + "epoch": 2.3776873171352246, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7897157073020935, + "num_tokens": 713208137.0, + "step": 18691 + }, + { + "epoch": 2.377814527413815, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7877563834190369, + "num_tokens": 713249344.0, + "step": 18692 + }, + { + "epoch": 2.3779417376924057, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7965163588523865, + "num_tokens": 713291376.0, + "step": 18693 + }, + { + "epoch": 2.378068947970996, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7959577441215515, + "num_tokens": 713329925.0, + "step": 18694 + }, + { + "epoch": 2.3781961582495867, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8044310808181763, + "num_tokens": 713372509.0, + "step": 18695 + }, + { + "epoch": 2.378323368528177, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.79444819688797, + "num_tokens": 713410055.0, + "step": 18696 + }, + { + "epoch": 2.378450578806768, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7968478202819824, + "num_tokens": 713453958.0, + "step": 18697 + }, + { + "epoch": 2.378577789085358, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7948276996612549, + "num_tokens": 713485683.0, + "step": 18698 + }, + { + "epoch": 2.3787049993639484, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8057706356048584, + "num_tokens": 713528565.0, + "step": 18699 + }, + { + "epoch": 2.378832209642539, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7921963930130005, + "num_tokens": 713560932.0, + "step": 18700 + }, + { + "epoch": 2.3789594199211295, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7905887961387634, + "num_tokens": 713592891.0, + "step": 18701 + }, + { + "epoch": 2.37908663019972, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.798256516456604, + "num_tokens": 713631285.0, + "step": 18702 + }, + { + "epoch": 2.3792138404783105, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7880064249038696, + "num_tokens": 713675143.0, + "step": 18703 + }, + { + "epoch": 2.379341050756901, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8150472044944763, + "num_tokens": 713717556.0, + "step": 18704 + }, + { + "epoch": 2.3794682610354916, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.793978750705719, + "num_tokens": 713763515.0, + "step": 18705 + }, + { + "epoch": 2.379595471314082, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8033103942871094, + "num_tokens": 713801064.0, + "step": 18706 + }, + { + "epoch": 2.3797226815926726, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8029956817626953, + "num_tokens": 713841662.0, + "step": 18707 + }, + { + "epoch": 2.379849891871263, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8078651428222656, + "num_tokens": 713883646.0, + "step": 18708 + }, + { + "epoch": 2.3799771021498537, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8225757479667664, + "num_tokens": 713925365.0, + "step": 18709 + }, + { + "epoch": 2.3801043124284442, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8001436591148376, + "num_tokens": 713961142.0, + "step": 18710 + }, + { + "epoch": 2.3802315227070348, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8122318983078003, + "num_tokens": 714001652.0, + "step": 18711 + }, + { + "epoch": 2.3803587329856253, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8037445545196533, + "num_tokens": 714043023.0, + "step": 18712 + }, + { + "epoch": 2.380485943264216, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7947264909744263, + "num_tokens": 714080389.0, + "step": 18713 + }, + { + "epoch": 2.3806131535428063, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8023310899734497, + "num_tokens": 714117695.0, + "step": 18714 + }, + { + "epoch": 2.380740363821397, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8139870166778564, + "num_tokens": 714154204.0, + "step": 18715 + }, + { + "epoch": 2.3808675740999874, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7890317440032959, + "num_tokens": 714194857.0, + "step": 18716 + }, + { + "epoch": 2.380994784378578, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8120349049568176, + "num_tokens": 714228034.0, + "step": 18717 + }, + { + "epoch": 2.3811219946571685, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8086875081062317, + "num_tokens": 714269688.0, + "step": 18718 + }, + { + "epoch": 2.381249204935759, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7953093647956848, + "num_tokens": 714309039.0, + "step": 18719 + }, + { + "epoch": 2.3813764152143495, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8057364225387573, + "num_tokens": 714347453.0, + "step": 18720 + }, + { + "epoch": 2.3815036254929396, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7930351495742798, + "num_tokens": 714385850.0, + "step": 18721 + }, + { + "epoch": 2.3816308357715306, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.8006190061569214, + "num_tokens": 714426735.0, + "step": 18722 + }, + { + "epoch": 2.3817580460501206, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8084680438041687, + "num_tokens": 714465279.0, + "step": 18723 + }, + { + "epoch": 2.381885256328711, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7908856272697449, + "num_tokens": 714505703.0, + "step": 18724 + }, + { + "epoch": 2.3820124666073017, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7901329398155212, + "num_tokens": 714543642.0, + "step": 18725 + }, + { + "epoch": 2.3821396768858922, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.819357693195343, + "num_tokens": 714583533.0, + "step": 18726 + }, + { + "epoch": 2.3822668871644828, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.808643639087677, + "num_tokens": 714625577.0, + "step": 18727 + }, + { + "epoch": 2.3823940974430733, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.79486083984375, + "num_tokens": 714664987.0, + "step": 18728 + }, + { + "epoch": 2.382521307721664, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7951068878173828, + "num_tokens": 714705881.0, + "step": 18729 + }, + { + "epoch": 2.3826485180002543, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7892919778823853, + "num_tokens": 714746012.0, + "step": 18730 + }, + { + "epoch": 2.382775728278845, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8005809783935547, + "num_tokens": 714790001.0, + "step": 18731 + }, + { + "epoch": 2.3829029385574354, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.809181272983551, + "num_tokens": 714824515.0, + "step": 18732 + }, + { + "epoch": 2.383030148836026, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7970229387283325, + "num_tokens": 714860841.0, + "step": 18733 + }, + { + "epoch": 2.3831573591146165, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8035801649093628, + "num_tokens": 714896373.0, + "step": 18734 + }, + { + "epoch": 2.383284569393207, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7976458072662354, + "num_tokens": 714926684.0, + "step": 18735 + }, + { + "epoch": 2.3834117796717975, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.8167665004730225, + "num_tokens": 714961057.0, + "step": 18736 + }, + { + "epoch": 2.383538989950388, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8084575533866882, + "num_tokens": 715001215.0, + "step": 18737 + }, + { + "epoch": 2.3836662002289786, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7791767120361328, + "num_tokens": 715038252.0, + "step": 18738 + }, + { + "epoch": 2.383793410507569, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8007436990737915, + "num_tokens": 715078320.0, + "step": 18739 + }, + { + "epoch": 2.3839206207861596, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7960897088050842, + "num_tokens": 715115628.0, + "step": 18740 + }, + { + "epoch": 2.38404783106475, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8073683977127075, + "num_tokens": 715153294.0, + "step": 18741 + }, + { + "epoch": 2.3841750413433407, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8009252548217773, + "num_tokens": 715193736.0, + "step": 18742 + }, + { + "epoch": 2.384302251621931, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8010662794113159, + "num_tokens": 715234539.0, + "step": 18743 + }, + { + "epoch": 2.3844294619005217, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.8128201961517334, + "num_tokens": 715271567.0, + "step": 18744 + }, + { + "epoch": 2.3845566721791123, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.8179121017456055, + "num_tokens": 715314850.0, + "step": 18745 + }, + { + "epoch": 2.3846838824577024, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.80613112449646, + "num_tokens": 715353353.0, + "step": 18746 + }, + { + "epoch": 2.3848110927362933, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.81169593334198, + "num_tokens": 715387843.0, + "step": 18747 + }, + { + "epoch": 2.3849383030148834, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7867823839187622, + "num_tokens": 715428037.0, + "step": 18748 + }, + { + "epoch": 2.385065513293474, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8185938596725464, + "num_tokens": 715464403.0, + "step": 18749 + }, + { + "epoch": 2.3851927235720645, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8142676949501038, + "num_tokens": 715501820.0, + "step": 18750 + }, + { + "epoch": 2.385319933850655, + "grad_norm": 57.75, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8166254162788391, + "num_tokens": 715543287.0, + "step": 18751 + }, + { + "epoch": 2.3854471441292455, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8119187951087952, + "num_tokens": 715578189.0, + "step": 18752 + }, + { + "epoch": 2.385574354407836, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7926411628723145, + "num_tokens": 715616447.0, + "step": 18753 + }, + { + "epoch": 2.3857015646864266, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8092807531356812, + "num_tokens": 715651692.0, + "step": 18754 + }, + { + "epoch": 2.385828774965017, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7848000526428223, + "num_tokens": 715687733.0, + "step": 18755 + }, + { + "epoch": 2.3859559852436076, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7721927165985107, + "num_tokens": 715727481.0, + "step": 18756 + }, + { + "epoch": 2.386083195522198, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7808681726455688, + "num_tokens": 715767549.0, + "step": 18757 + }, + { + "epoch": 2.3862104058007887, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7906067371368408, + "num_tokens": 715807251.0, + "step": 18758 + }, + { + "epoch": 2.386337616079379, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7850390672683716, + "num_tokens": 715846893.0, + "step": 18759 + }, + { + "epoch": 2.3864648263579697, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7829009294509888, + "num_tokens": 715883611.0, + "step": 18760 + }, + { + "epoch": 2.3865920366365603, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7969318628311157, + "num_tokens": 715915598.0, + "step": 18761 + }, + { + "epoch": 2.386719246915151, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8094295263290405, + "num_tokens": 715956647.0, + "step": 18762 + }, + { + "epoch": 2.3868464571937413, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8050583004951477, + "num_tokens": 715996627.0, + "step": 18763 + }, + { + "epoch": 2.386973667472332, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.5825, + "mean_token_accuracy": 0.8233335018157959, + "num_tokens": 716027823.0, + "step": 18764 + }, + { + "epoch": 2.3871008777509224, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.786074697971344, + "num_tokens": 716067430.0, + "step": 18765 + }, + { + "epoch": 2.387228088029513, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7982723712921143, + "num_tokens": 716109436.0, + "step": 18766 + }, + { + "epoch": 2.3873552983081034, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.795829176902771, + "num_tokens": 716148157.0, + "step": 18767 + }, + { + "epoch": 2.387482508586694, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8048655390739441, + "num_tokens": 716184662.0, + "step": 18768 + }, + { + "epoch": 2.3876097188652845, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7984588146209717, + "num_tokens": 716227382.0, + "step": 18769 + }, + { + "epoch": 2.387736929143875, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8072127103805542, + "num_tokens": 716262236.0, + "step": 18770 + }, + { + "epoch": 2.387864139422465, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8079049587249756, + "num_tokens": 716295541.0, + "step": 18771 + }, + { + "epoch": 2.387991349701056, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8079269528388977, + "num_tokens": 716331796.0, + "step": 18772 + }, + { + "epoch": 2.388118559979646, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7903460264205933, + "num_tokens": 716369539.0, + "step": 18773 + }, + { + "epoch": 2.3882457702582367, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5872, + "mean_token_accuracy": 0.8234527707099915, + "num_tokens": 716409806.0, + "step": 18774 + }, + { + "epoch": 2.3883729805368272, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7900036573410034, + "num_tokens": 716448191.0, + "step": 18775 + }, + { + "epoch": 2.3885001908154178, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8042410612106323, + "num_tokens": 716486522.0, + "step": 18776 + }, + { + "epoch": 2.3886274010940083, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7946490049362183, + "num_tokens": 716519383.0, + "step": 18777 + }, + { + "epoch": 2.388754611372599, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8059221506118774, + "num_tokens": 716556136.0, + "step": 18778 + }, + { + "epoch": 2.3888818216511893, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7930971384048462, + "num_tokens": 716592346.0, + "step": 18779 + }, + { + "epoch": 2.38900903192978, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7948676347732544, + "num_tokens": 716633996.0, + "step": 18780 + }, + { + "epoch": 2.3891362422083704, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.799267053604126, + "num_tokens": 716669638.0, + "step": 18781 + }, + { + "epoch": 2.389263452486961, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7913789749145508, + "num_tokens": 716711194.0, + "step": 18782 + }, + { + "epoch": 2.3893906627655515, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8092247843742371, + "num_tokens": 716753402.0, + "step": 18783 + }, + { + "epoch": 2.389517873044142, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8058299422264099, + "num_tokens": 716794638.0, + "step": 18784 + }, + { + "epoch": 2.3896450833227325, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7722269296646118, + "num_tokens": 716834003.0, + "step": 18785 + }, + { + "epoch": 2.389772293601323, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8042856454849243, + "num_tokens": 716867349.0, + "step": 18786 + }, + { + "epoch": 2.3898995038799136, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8044430017471313, + "num_tokens": 716904271.0, + "step": 18787 + }, + { + "epoch": 2.390026714158504, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.801252543926239, + "num_tokens": 716936165.0, + "step": 18788 + }, + { + "epoch": 2.3901539244370946, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7992770075798035, + "num_tokens": 716974011.0, + "step": 18789 + }, + { + "epoch": 2.390281134715685, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.8000837564468384, + "num_tokens": 717012438.0, + "step": 18790 + }, + { + "epoch": 2.3904083449942757, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7912884950637817, + "num_tokens": 717051755.0, + "step": 18791 + }, + { + "epoch": 2.390535555272866, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8066670894622803, + "num_tokens": 717090740.0, + "step": 18792 + }, + { + "epoch": 2.3906627655514567, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7750929594039917, + "num_tokens": 717132033.0, + "step": 18793 + }, + { + "epoch": 2.390789975830047, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8071463108062744, + "num_tokens": 717168255.0, + "step": 18794 + }, + { + "epoch": 2.390917186108638, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7975131273269653, + "num_tokens": 717206863.0, + "step": 18795 + }, + { + "epoch": 2.391044396387228, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7811764478683472, + "num_tokens": 717245374.0, + "step": 18796 + }, + { + "epoch": 2.3911716066658184, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.8213168978691101, + "num_tokens": 717279821.0, + "step": 18797 + }, + { + "epoch": 2.391298816944409, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7969461679458618, + "num_tokens": 717318896.0, + "step": 18798 + }, + { + "epoch": 2.3914260272229995, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7894779443740845, + "num_tokens": 717361493.0, + "step": 18799 + }, + { + "epoch": 2.39155323750159, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7907553911209106, + "num_tokens": 717402801.0, + "step": 18800 + }, + { + "epoch": 2.3916804477801805, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8166011571884155, + "num_tokens": 717441275.0, + "step": 18801 + }, + { + "epoch": 2.391807658058771, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.784183144569397, + "num_tokens": 717477423.0, + "step": 18802 + }, + { + "epoch": 2.3919348683373616, + "grad_norm": 71.5, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7916092276573181, + "num_tokens": 717512978.0, + "step": 18803 + }, + { + "epoch": 2.392062078615952, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8206155300140381, + "num_tokens": 717549710.0, + "step": 18804 + }, + { + "epoch": 2.3921892888945426, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.793269157409668, + "num_tokens": 717586588.0, + "step": 18805 + }, + { + "epoch": 2.392316499173133, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.8166937828063965, + "num_tokens": 717621507.0, + "step": 18806 + }, + { + "epoch": 2.3924437094517237, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7978028059005737, + "num_tokens": 717655915.0, + "step": 18807 + }, + { + "epoch": 2.392570919730314, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7989940643310547, + "num_tokens": 717696357.0, + "step": 18808 + }, + { + "epoch": 2.3926981300089047, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8102295398712158, + "num_tokens": 717735397.0, + "step": 18809 + }, + { + "epoch": 2.3928253402874953, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.800083577632904, + "num_tokens": 717770081.0, + "step": 18810 + }, + { + "epoch": 2.392952550566086, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.8162058591842651, + "num_tokens": 717805262.0, + "step": 18811 + }, + { + "epoch": 2.3930797608446763, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.794730544090271, + "num_tokens": 717841877.0, + "step": 18812 + }, + { + "epoch": 2.393206971123267, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7853580713272095, + "num_tokens": 717881435.0, + "step": 18813 + }, + { + "epoch": 2.3933341814018574, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8203456997871399, + "num_tokens": 717922240.0, + "step": 18814 + }, + { + "epoch": 2.393461391680448, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7944815754890442, + "num_tokens": 717958438.0, + "step": 18815 + }, + { + "epoch": 2.3935886019590384, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7916207313537598, + "num_tokens": 717994451.0, + "step": 18816 + }, + { + "epoch": 2.393715812237629, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8052326440811157, + "num_tokens": 718034318.0, + "step": 18817 + }, + { + "epoch": 2.3938430225162195, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8016799688339233, + "num_tokens": 718074994.0, + "step": 18818 + }, + { + "epoch": 2.3939702327948096, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7973047494888306, + "num_tokens": 718118849.0, + "step": 18819 + }, + { + "epoch": 2.3940974430734006, + "grad_norm": 96.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8070382475852966, + "num_tokens": 718152874.0, + "step": 18820 + }, + { + "epoch": 2.3942246533519906, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7922760248184204, + "num_tokens": 718194206.0, + "step": 18821 + }, + { + "epoch": 2.394351863630581, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8034191727638245, + "num_tokens": 718227408.0, + "step": 18822 + }, + { + "epoch": 2.3944790739091717, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7990590333938599, + "num_tokens": 718266962.0, + "step": 18823 + }, + { + "epoch": 2.3946062841877622, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8089110255241394, + "num_tokens": 718307873.0, + "step": 18824 + }, + { + "epoch": 2.3947334944663528, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8097635507583618, + "num_tokens": 718350296.0, + "step": 18825 + }, + { + "epoch": 2.3948607047449433, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8021690249443054, + "num_tokens": 718390734.0, + "step": 18826 + }, + { + "epoch": 2.394987915023534, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.5827, + "mean_token_accuracy": 0.8224663734436035, + "num_tokens": 718435630.0, + "step": 18827 + }, + { + "epoch": 2.3951151253021243, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.5919, + "mean_token_accuracy": 0.8154038786888123, + "num_tokens": 718474791.0, + "step": 18828 + }, + { + "epoch": 2.395242335580715, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7788976430892944, + "num_tokens": 718514957.0, + "step": 18829 + }, + { + "epoch": 2.3953695458593054, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8082926273345947, + "num_tokens": 718551761.0, + "step": 18830 + }, + { + "epoch": 2.395496756137896, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7964180707931519, + "num_tokens": 718586445.0, + "step": 18831 + }, + { + "epoch": 2.3956239664164865, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7961190342903137, + "num_tokens": 718631105.0, + "step": 18832 + }, + { + "epoch": 2.395751176695077, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7977945804595947, + "num_tokens": 718672278.0, + "step": 18833 + }, + { + "epoch": 2.3958783869736675, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.806923508644104, + "num_tokens": 718713191.0, + "step": 18834 + }, + { + "epoch": 2.396005597252258, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7926996350288391, + "num_tokens": 718745012.0, + "step": 18835 + }, + { + "epoch": 2.3961328075308486, + "grad_norm": 56.25, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8016670942306519, + "num_tokens": 718780149.0, + "step": 18836 + }, + { + "epoch": 2.396260017809439, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7902207970619202, + "num_tokens": 718812831.0, + "step": 18837 + }, + { + "epoch": 2.3963872280880296, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8079067468643188, + "num_tokens": 718847650.0, + "step": 18838 + }, + { + "epoch": 2.39651443836662, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8053582906723022, + "num_tokens": 718883417.0, + "step": 18839 + }, + { + "epoch": 2.3966416486452107, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.8238610029220581, + "num_tokens": 718915501.0, + "step": 18840 + }, + { + "epoch": 2.396768858923801, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8154153823852539, + "num_tokens": 718954485.0, + "step": 18841 + }, + { + "epoch": 2.3968960692023917, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7868972420692444, + "num_tokens": 718995815.0, + "step": 18842 + }, + { + "epoch": 2.3970232794809823, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.5771, + "mean_token_accuracy": 0.8193889260292053, + "num_tokens": 719032820.0, + "step": 18843 + }, + { + "epoch": 2.3971504897595723, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.8002716302871704, + "num_tokens": 719066966.0, + "step": 18844 + }, + { + "epoch": 2.3972777000381633, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8013657927513123, + "num_tokens": 719108768.0, + "step": 18845 + }, + { + "epoch": 2.3974049103167534, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7966516017913818, + "num_tokens": 719151451.0, + "step": 18846 + }, + { + "epoch": 2.397532120595344, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7957143187522888, + "num_tokens": 719192103.0, + "step": 18847 + }, + { + "epoch": 2.3976593308739345, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7920924425125122, + "num_tokens": 719226514.0, + "step": 18848 + }, + { + "epoch": 2.397786541152525, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.5431, + "mean_token_accuracy": 0.831899881362915, + "num_tokens": 719262944.0, + "step": 18849 + }, + { + "epoch": 2.3979137514311155, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8104928731918335, + "num_tokens": 719306707.0, + "step": 18850 + }, + { + "epoch": 2.398040961709706, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8108777403831482, + "num_tokens": 719347523.0, + "step": 18851 + }, + { + "epoch": 2.3981681719882966, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8186739683151245, + "num_tokens": 719379474.0, + "step": 18852 + }, + { + "epoch": 2.398295382266887, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7891262769699097, + "num_tokens": 719421624.0, + "step": 18853 + }, + { + "epoch": 2.3984225925454776, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7917274832725525, + "num_tokens": 719455570.0, + "step": 18854 + }, + { + "epoch": 2.398549802824068, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.5918, + "mean_token_accuracy": 0.8206174373626709, + "num_tokens": 719500300.0, + "step": 18855 + }, + { + "epoch": 2.3986770131026587, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7866723537445068, + "num_tokens": 719538119.0, + "step": 18856 + }, + { + "epoch": 2.398804223381249, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7858636379241943, + "num_tokens": 719574012.0, + "step": 18857 + }, + { + "epoch": 2.3989314336598397, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8139191269874573, + "num_tokens": 719609548.0, + "step": 18858 + }, + { + "epoch": 2.3990586439384303, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.795174241065979, + "num_tokens": 719647280.0, + "step": 18859 + }, + { + "epoch": 2.399185854217021, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.5889, + "mean_token_accuracy": 0.820747971534729, + "num_tokens": 719686987.0, + "step": 18860 + }, + { + "epoch": 2.3993130644956113, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.82051682472229, + "num_tokens": 719727974.0, + "step": 18861 + }, + { + "epoch": 2.399440274774202, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7978389263153076, + "num_tokens": 719764308.0, + "step": 18862 + }, + { + "epoch": 2.3995674850527924, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7917484045028687, + "num_tokens": 719801872.0, + "step": 18863 + }, + { + "epoch": 2.399694695331383, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8054147362709045, + "num_tokens": 719842504.0, + "step": 18864 + }, + { + "epoch": 2.3998219056099734, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.794069766998291, + "num_tokens": 719878885.0, + "step": 18865 + }, + { + "epoch": 2.399949115888564, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7878744602203369, + "num_tokens": 719922891.0, + "step": 18866 + }, + { + "epoch": 2.4000763261671545, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8049089908599854, + "num_tokens": 719960113.0, + "step": 18867 + }, + { + "epoch": 2.400203536445745, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8006086945533752, + "num_tokens": 720006079.0, + "step": 18868 + }, + { + "epoch": 2.400330746724335, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7886867523193359, + "num_tokens": 720042219.0, + "step": 18869 + }, + { + "epoch": 2.400457957002926, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8028677701950073, + "num_tokens": 720082574.0, + "step": 18870 + }, + { + "epoch": 2.400585167281516, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7982978820800781, + "num_tokens": 720115583.0, + "step": 18871 + }, + { + "epoch": 2.4007123775601067, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7917511463165283, + "num_tokens": 720151350.0, + "step": 18872 + }, + { + "epoch": 2.400839587838697, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7963367700576782, + "num_tokens": 720187072.0, + "step": 18873 + }, + { + "epoch": 2.4009667981172877, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8018201589584351, + "num_tokens": 720222249.0, + "step": 18874 + }, + { + "epoch": 2.4010940083958783, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7996108531951904, + "num_tokens": 720264682.0, + "step": 18875 + }, + { + "epoch": 2.401221218674469, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7984343767166138, + "num_tokens": 720309935.0, + "step": 18876 + }, + { + "epoch": 2.4013484289530593, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7919026613235474, + "num_tokens": 720343992.0, + "step": 18877 + }, + { + "epoch": 2.40147563923165, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.799174964427948, + "num_tokens": 720379707.0, + "step": 18878 + }, + { + "epoch": 2.4016028495102404, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8009535670280457, + "num_tokens": 720419796.0, + "step": 18879 + }, + { + "epoch": 2.401730059788831, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8063012957572937, + "num_tokens": 720460919.0, + "step": 18880 + }, + { + "epoch": 2.4018572700674214, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7986193895339966, + "num_tokens": 720498403.0, + "step": 18881 + }, + { + "epoch": 2.401984480346012, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8011822700500488, + "num_tokens": 720534931.0, + "step": 18882 + }, + { + "epoch": 2.4021116906246025, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8092384934425354, + "num_tokens": 720574062.0, + "step": 18883 + }, + { + "epoch": 2.402238900903193, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7895045280456543, + "num_tokens": 720611552.0, + "step": 18884 + }, + { + "epoch": 2.4023661111817836, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.803780198097229, + "num_tokens": 720647886.0, + "step": 18885 + }, + { + "epoch": 2.402493321460374, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8093340992927551, + "num_tokens": 720688034.0, + "step": 18886 + }, + { + "epoch": 2.4026205317389646, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8075616955757141, + "num_tokens": 720727805.0, + "step": 18887 + }, + { + "epoch": 2.402747742017555, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7988258600234985, + "num_tokens": 720759717.0, + "step": 18888 + }, + { + "epoch": 2.4028749522961457, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7912711501121521, + "num_tokens": 720790435.0, + "step": 18889 + }, + { + "epoch": 2.403002162574736, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7987502813339233, + "num_tokens": 720831715.0, + "step": 18890 + }, + { + "epoch": 2.4031293728533267, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8049573302268982, + "num_tokens": 720869598.0, + "step": 18891 + }, + { + "epoch": 2.403256583131917, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7898427248001099, + "num_tokens": 720911122.0, + "step": 18892 + }, + { + "epoch": 2.403383793410508, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.809390127658844, + "num_tokens": 720952265.0, + "step": 18893 + }, + { + "epoch": 2.403511003689098, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.80377596616745, + "num_tokens": 720991191.0, + "step": 18894 + }, + { + "epoch": 2.4036382139676884, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8102490305900574, + "num_tokens": 721032524.0, + "step": 18895 + }, + { + "epoch": 2.403765424246279, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7992856502532959, + "num_tokens": 721072699.0, + "step": 18896 + }, + { + "epoch": 2.4038926345248695, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8116391897201538, + "num_tokens": 721109467.0, + "step": 18897 + }, + { + "epoch": 2.40401984480346, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7672381401062012, + "num_tokens": 721142108.0, + "step": 18898 + }, + { + "epoch": 2.4041470550820505, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7894737124443054, + "num_tokens": 721179875.0, + "step": 18899 + }, + { + "epoch": 2.404274265360641, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7967474460601807, + "num_tokens": 721217141.0, + "step": 18900 + }, + { + "epoch": 2.4044014756392316, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7928898334503174, + "num_tokens": 721256357.0, + "step": 18901 + }, + { + "epoch": 2.404528685917822, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8006784319877625, + "num_tokens": 721292610.0, + "step": 18902 + }, + { + "epoch": 2.4046558961964126, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.8050010800361633, + "num_tokens": 721329048.0, + "step": 18903 + }, + { + "epoch": 2.404783106475003, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8124676942825317, + "num_tokens": 721364249.0, + "step": 18904 + }, + { + "epoch": 2.4049103167535937, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8011099100112915, + "num_tokens": 721391154.0, + "step": 18905 + }, + { + "epoch": 2.405037527032184, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8106603622436523, + "num_tokens": 721428443.0, + "step": 18906 + }, + { + "epoch": 2.4051647373107747, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7903610467910767, + "num_tokens": 721474545.0, + "step": 18907 + }, + { + "epoch": 2.4052919475893653, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7951589822769165, + "num_tokens": 721513153.0, + "step": 18908 + }, + { + "epoch": 2.405419157867956, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8051056861877441, + "num_tokens": 721556287.0, + "step": 18909 + }, + { + "epoch": 2.4055463681465463, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.807440996170044, + "num_tokens": 721590219.0, + "step": 18910 + }, + { + "epoch": 2.405673578425137, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.7998636364936829, + "num_tokens": 721627308.0, + "step": 18911 + }, + { + "epoch": 2.4058007887037274, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7987996339797974, + "num_tokens": 721664218.0, + "step": 18912 + }, + { + "epoch": 2.405927998982318, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8104875087738037, + "num_tokens": 721703511.0, + "step": 18913 + }, + { + "epoch": 2.4060552092609084, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8211055994033813, + "num_tokens": 721741512.0, + "step": 18914 + }, + { + "epoch": 2.406182419539499, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8043098449707031, + "num_tokens": 721778443.0, + "step": 18915 + }, + { + "epoch": 2.4063096298180895, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7990503311157227, + "num_tokens": 721817226.0, + "step": 18916 + }, + { + "epoch": 2.4064368400966796, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7870383262634277, + "num_tokens": 721859419.0, + "step": 18917 + }, + { + "epoch": 2.4065640503752705, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.814169704914093, + "num_tokens": 721895183.0, + "step": 18918 + }, + { + "epoch": 2.4066912606538606, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7813037037849426, + "num_tokens": 721933566.0, + "step": 18919 + }, + { + "epoch": 2.406818470932451, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8020323514938354, + "num_tokens": 721968827.0, + "step": 18920 + }, + { + "epoch": 2.4069456812110417, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8041558861732483, + "num_tokens": 722004989.0, + "step": 18921 + }, + { + "epoch": 2.407072891489632, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8091062307357788, + "num_tokens": 722043681.0, + "step": 18922 + }, + { + "epoch": 2.4072001017682227, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8022177219390869, + "num_tokens": 722086541.0, + "step": 18923 + }, + { + "epoch": 2.4073273120468133, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.801540732383728, + "num_tokens": 722125519.0, + "step": 18924 + }, + { + "epoch": 2.407454522325404, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7938429713249207, + "num_tokens": 722157665.0, + "step": 18925 + }, + { + "epoch": 2.4075817326039943, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8128741979598999, + "num_tokens": 722191819.0, + "step": 18926 + }, + { + "epoch": 2.407708942882585, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7890418767929077, + "num_tokens": 722229755.0, + "step": 18927 + }, + { + "epoch": 2.4078361531611754, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.791555643081665, + "num_tokens": 722267440.0, + "step": 18928 + }, + { + "epoch": 2.407963363439766, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8006401062011719, + "num_tokens": 722310102.0, + "step": 18929 + }, + { + "epoch": 2.4080905737183564, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7911642789840698, + "num_tokens": 722346713.0, + "step": 18930 + }, + { + "epoch": 2.408217783996947, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7969481945037842, + "num_tokens": 722387803.0, + "step": 18931 + }, + { + "epoch": 2.4083449942755375, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7890777587890625, + "num_tokens": 722428943.0, + "step": 18932 + }, + { + "epoch": 2.408472204554128, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7892788648605347, + "num_tokens": 722467860.0, + "step": 18933 + }, + { + "epoch": 2.4085994148327186, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7983932495117188, + "num_tokens": 722503067.0, + "step": 18934 + }, + { + "epoch": 2.408726625111309, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7982599139213562, + "num_tokens": 722537842.0, + "step": 18935 + }, + { + "epoch": 2.4088538353898996, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8006597757339478, + "num_tokens": 722574036.0, + "step": 18936 + }, + { + "epoch": 2.40898104566849, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8014003038406372, + "num_tokens": 722606087.0, + "step": 18937 + }, + { + "epoch": 2.4091082559470807, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7991859912872314, + "num_tokens": 722648745.0, + "step": 18938 + }, + { + "epoch": 2.409235466225671, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7952438592910767, + "num_tokens": 722688589.0, + "step": 18939 + }, + { + "epoch": 2.4093626765042617, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.5871, + "mean_token_accuracy": 0.822784423828125, + "num_tokens": 722723941.0, + "step": 18940 + }, + { + "epoch": 2.4094898867828523, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.784193754196167, + "num_tokens": 722768242.0, + "step": 18941 + }, + { + "epoch": 2.4096170970614423, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8001647591590881, + "num_tokens": 722800970.0, + "step": 18942 + }, + { + "epoch": 2.4097443073400333, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7946499586105347, + "num_tokens": 722839168.0, + "step": 18943 + }, + { + "epoch": 2.4098715176186234, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8134743571281433, + "num_tokens": 722883451.0, + "step": 18944 + }, + { + "epoch": 2.409998727897214, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8045364618301392, + "num_tokens": 722928581.0, + "step": 18945 + }, + { + "epoch": 2.4101259381758044, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8132420778274536, + "num_tokens": 722967141.0, + "step": 18946 + }, + { + "epoch": 2.410253148454395, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7927043437957764, + "num_tokens": 723006263.0, + "step": 18947 + }, + { + "epoch": 2.4103803587329855, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8069353699684143, + "num_tokens": 723050843.0, + "step": 18948 + }, + { + "epoch": 2.410507569011576, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7971261739730835, + "num_tokens": 723092124.0, + "step": 18949 + }, + { + "epoch": 2.4106347792901666, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8066360354423523, + "num_tokens": 723122478.0, + "step": 18950 + }, + { + "epoch": 2.410761989568757, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7908742427825928, + "num_tokens": 723163054.0, + "step": 18951 + }, + { + "epoch": 2.4108891998473476, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8031972646713257, + "num_tokens": 723201518.0, + "step": 18952 + }, + { + "epoch": 2.411016410125938, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8003929853439331, + "num_tokens": 723235685.0, + "step": 18953 + }, + { + "epoch": 2.4111436204045287, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7958824634552002, + "num_tokens": 723266510.0, + "step": 18954 + }, + { + "epoch": 2.411270830683119, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8028566241264343, + "num_tokens": 723302800.0, + "step": 18955 + }, + { + "epoch": 2.4113980409617097, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.8017832040786743, + "num_tokens": 723340669.0, + "step": 18956 + }, + { + "epoch": 2.4115252512403003, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.80081707239151, + "num_tokens": 723377556.0, + "step": 18957 + }, + { + "epoch": 2.411652461518891, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8050355911254883, + "num_tokens": 723410942.0, + "step": 18958 + }, + { + "epoch": 2.4117796717974813, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.8149068355560303, + "num_tokens": 723451131.0, + "step": 18959 + }, + { + "epoch": 2.411906882076072, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7995285391807556, + "num_tokens": 723485876.0, + "step": 18960 + }, + { + "epoch": 2.4120340923546624, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8062136769294739, + "num_tokens": 723521252.0, + "step": 18961 + }, + { + "epoch": 2.412161302633253, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7804995775222778, + "num_tokens": 723561094.0, + "step": 18962 + }, + { + "epoch": 2.4122885129118434, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7999721169471741, + "num_tokens": 723597281.0, + "step": 18963 + }, + { + "epoch": 2.412415723190434, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8241732716560364, + "num_tokens": 723637690.0, + "step": 18964 + }, + { + "epoch": 2.4125429334690245, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7976226210594177, + "num_tokens": 723676250.0, + "step": 18965 + }, + { + "epoch": 2.412670143747615, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8100932836532593, + "num_tokens": 723713157.0, + "step": 18966 + }, + { + "epoch": 2.412797354026205, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8060905337333679, + "num_tokens": 723745877.0, + "step": 18967 + }, + { + "epoch": 2.412924564304796, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8043719530105591, + "num_tokens": 723784285.0, + "step": 18968 + }, + { + "epoch": 2.413051774583386, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7876348495483398, + "num_tokens": 723817085.0, + "step": 18969 + }, + { + "epoch": 2.4131789848619767, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7937048673629761, + "num_tokens": 723858466.0, + "step": 18970 + }, + { + "epoch": 2.413306195140567, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8048425912857056, + "num_tokens": 723896627.0, + "step": 18971 + }, + { + "epoch": 2.4134334054191577, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7939600348472595, + "num_tokens": 723930727.0, + "step": 18972 + }, + { + "epoch": 2.4135606156977483, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7841503620147705, + "num_tokens": 723965996.0, + "step": 18973 + }, + { + "epoch": 2.413687825976339, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7948446869850159, + "num_tokens": 724002441.0, + "step": 18974 + }, + { + "epoch": 2.4138150362549293, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8026138544082642, + "num_tokens": 724039238.0, + "step": 18975 + }, + { + "epoch": 2.41394224653352, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8060184121131897, + "num_tokens": 724077430.0, + "step": 18976 + }, + { + "epoch": 2.4140694568121104, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7991304397583008, + "num_tokens": 724113175.0, + "step": 18977 + }, + { + "epoch": 2.414196667090701, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7946693897247314, + "num_tokens": 724156409.0, + "step": 18978 + }, + { + "epoch": 2.4143238773692914, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8020737171173096, + "num_tokens": 724199879.0, + "step": 18979 + }, + { + "epoch": 2.414451087647882, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8169176578521729, + "num_tokens": 724238065.0, + "step": 18980 + }, + { + "epoch": 2.4145782979264725, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7922592163085938, + "num_tokens": 724272336.0, + "step": 18981 + }, + { + "epoch": 2.414705508205063, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7852955460548401, + "num_tokens": 724305200.0, + "step": 18982 + }, + { + "epoch": 2.4148327184836536, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8077748417854309, + "num_tokens": 724341295.0, + "step": 18983 + }, + { + "epoch": 2.414959928762244, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8076955080032349, + "num_tokens": 724384336.0, + "step": 18984 + }, + { + "epoch": 2.4150871390408346, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7942073941230774, + "num_tokens": 724417279.0, + "step": 18985 + }, + { + "epoch": 2.415214349319425, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7981618046760559, + "num_tokens": 724456509.0, + "step": 18986 + }, + { + "epoch": 2.4153415595980157, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8161634206771851, + "num_tokens": 724498812.0, + "step": 18987 + }, + { + "epoch": 2.415468769876606, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7917482852935791, + "num_tokens": 724538174.0, + "step": 18988 + }, + { + "epoch": 2.4155959801551967, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8033703565597534, + "num_tokens": 724577171.0, + "step": 18989 + }, + { + "epoch": 2.415723190433787, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8258973360061646, + "num_tokens": 724610626.0, + "step": 18990 + }, + { + "epoch": 2.4158504007123778, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8148376941680908, + "num_tokens": 724647613.0, + "step": 18991 + }, + { + "epoch": 2.415977610990968, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7997725605964661, + "num_tokens": 724684898.0, + "step": 18992 + }, + { + "epoch": 2.4161048212695584, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7961015701293945, + "num_tokens": 724725566.0, + "step": 18993 + }, + { + "epoch": 2.416232031548149, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7863638997077942, + "num_tokens": 724763701.0, + "step": 18994 + }, + { + "epoch": 2.4163592418267394, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7996953129768372, + "num_tokens": 724798108.0, + "step": 18995 + }, + { + "epoch": 2.41648645210533, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7750332355499268, + "num_tokens": 724842430.0, + "step": 18996 + }, + { + "epoch": 2.4166136623839205, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7974218130111694, + "num_tokens": 724884585.0, + "step": 18997 + }, + { + "epoch": 2.416740872662511, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8174445629119873, + "num_tokens": 724919338.0, + "step": 18998 + }, + { + "epoch": 2.4168680829411016, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.5797, + "mean_token_accuracy": 0.8269727826118469, + "num_tokens": 724956007.0, + "step": 18999 + }, + { + "epoch": 2.416995293219692, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8180115222930908, + "num_tokens": 724996994.0, + "step": 19000 + }, + { + "epoch": 2.4171225034982826, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7969608902931213, + "num_tokens": 725035607.0, + "step": 19001 + }, + { + "epoch": 2.417249713776873, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8066853284835815, + "num_tokens": 725075370.0, + "step": 19002 + }, + { + "epoch": 2.4173769240554637, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8048344850540161, + "num_tokens": 725119536.0, + "step": 19003 + }, + { + "epoch": 2.417504134334054, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8010997772216797, + "num_tokens": 725158048.0, + "step": 19004 + }, + { + "epoch": 2.4176313446126447, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8132352828979492, + "num_tokens": 725196343.0, + "step": 19005 + }, + { + "epoch": 2.4177585548912353, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8019552230834961, + "num_tokens": 725234613.0, + "step": 19006 + }, + { + "epoch": 2.417885765169826, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7951947450637817, + "num_tokens": 725272129.0, + "step": 19007 + }, + { + "epoch": 2.4180129754484163, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.792117714881897, + "num_tokens": 725305740.0, + "step": 19008 + }, + { + "epoch": 2.418140185727007, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7953488826751709, + "num_tokens": 725340579.0, + "step": 19009 + }, + { + "epoch": 2.4182673960055974, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8011407256126404, + "num_tokens": 725380492.0, + "step": 19010 + }, + { + "epoch": 2.418394606284188, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8022709488868713, + "num_tokens": 725418043.0, + "step": 19011 + }, + { + "epoch": 2.4185218165627784, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7891527414321899, + "num_tokens": 725456260.0, + "step": 19012 + }, + { + "epoch": 2.418649026841369, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.798293948173523, + "num_tokens": 725496605.0, + "step": 19013 + }, + { + "epoch": 2.4187762371199595, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8009066581726074, + "num_tokens": 725532395.0, + "step": 19014 + }, + { + "epoch": 2.4189034473985496, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7983046770095825, + "num_tokens": 725575335.0, + "step": 19015 + }, + { + "epoch": 2.4190306576771405, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8125640153884888, + "num_tokens": 725612345.0, + "step": 19016 + }, + { + "epoch": 2.4191578679557306, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8047981262207031, + "num_tokens": 725645566.0, + "step": 19017 + }, + { + "epoch": 2.419285078234321, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7938254475593567, + "num_tokens": 725684218.0, + "step": 19018 + }, + { + "epoch": 2.4194122885129117, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8071710467338562, + "num_tokens": 725718478.0, + "step": 19019 + }, + { + "epoch": 2.419539498791502, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7943843603134155, + "num_tokens": 725756208.0, + "step": 19020 + }, + { + "epoch": 2.4196667090700927, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7950964570045471, + "num_tokens": 725791884.0, + "step": 19021 + }, + { + "epoch": 2.4197939193486833, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.8047712445259094, + "num_tokens": 725829138.0, + "step": 19022 + }, + { + "epoch": 2.419921129627274, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8056316375732422, + "num_tokens": 725868301.0, + "step": 19023 + }, + { + "epoch": 2.4200483399058643, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7823236584663391, + "num_tokens": 725906186.0, + "step": 19024 + }, + { + "epoch": 2.420175550184455, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7894147634506226, + "num_tokens": 725947006.0, + "step": 19025 + }, + { + "epoch": 2.4203027604630454, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8048962354660034, + "num_tokens": 725987363.0, + "step": 19026 + }, + { + "epoch": 2.420429970741636, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7928313612937927, + "num_tokens": 726026176.0, + "step": 19027 + }, + { + "epoch": 2.4205571810202264, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.801932156085968, + "num_tokens": 726068124.0, + "step": 19028 + }, + { + "epoch": 2.420684391298817, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7948586344718933, + "num_tokens": 726110270.0, + "step": 19029 + }, + { + "epoch": 2.4208116015774075, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.8174537420272827, + "num_tokens": 726143908.0, + "step": 19030 + }, + { + "epoch": 2.420938811855998, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7979847192764282, + "num_tokens": 726179082.0, + "step": 19031 + }, + { + "epoch": 2.4210660221345885, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7958716154098511, + "num_tokens": 726216387.0, + "step": 19032 + }, + { + "epoch": 2.421193232413179, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8110738396644592, + "num_tokens": 726258831.0, + "step": 19033 + }, + { + "epoch": 2.4213204426917696, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8029849529266357, + "num_tokens": 726301929.0, + "step": 19034 + }, + { + "epoch": 2.42144765297036, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8027064204216003, + "num_tokens": 726339938.0, + "step": 19035 + }, + { + "epoch": 2.4215748632489507, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8026225566864014, + "num_tokens": 726377197.0, + "step": 19036 + }, + { + "epoch": 2.421702073527541, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8056895136833191, + "num_tokens": 726417591.0, + "step": 19037 + }, + { + "epoch": 2.4218292838061317, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7910565137863159, + "num_tokens": 726456094.0, + "step": 19038 + }, + { + "epoch": 2.4219564940847222, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.813431978225708, + "num_tokens": 726491158.0, + "step": 19039 + }, + { + "epoch": 2.4220837043633123, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7901171445846558, + "num_tokens": 726531334.0, + "step": 19040 + }, + { + "epoch": 2.4222109146419033, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.79570472240448, + "num_tokens": 726571999.0, + "step": 19041 + }, + { + "epoch": 2.4223381249204934, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7907789945602417, + "num_tokens": 726604945.0, + "step": 19042 + }, + { + "epoch": 2.422465335199084, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7905762195587158, + "num_tokens": 726645023.0, + "step": 19043 + }, + { + "epoch": 2.4225925454776744, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8062084913253784, + "num_tokens": 726684297.0, + "step": 19044 + }, + { + "epoch": 2.422719755756265, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7891122102737427, + "num_tokens": 726726950.0, + "step": 19045 + }, + { + "epoch": 2.4228469660348555, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7800289392471313, + "num_tokens": 726761944.0, + "step": 19046 + }, + { + "epoch": 2.422974176313446, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8098786473274231, + "num_tokens": 726799041.0, + "step": 19047 + }, + { + "epoch": 2.4231013865920366, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7966381311416626, + "num_tokens": 726838160.0, + "step": 19048 + }, + { + "epoch": 2.423228596870627, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8051850199699402, + "num_tokens": 726873150.0, + "step": 19049 + }, + { + "epoch": 2.4233558071492176, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8043522238731384, + "num_tokens": 726905253.0, + "step": 19050 + }, + { + "epoch": 2.423483017427808, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8127015829086304, + "num_tokens": 726942842.0, + "step": 19051 + }, + { + "epoch": 2.4236102277063987, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8084384202957153, + "num_tokens": 726987974.0, + "step": 19052 + }, + { + "epoch": 2.423737437984989, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8175588250160217, + "num_tokens": 727021614.0, + "step": 19053 + }, + { + "epoch": 2.4238646482635797, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8029935956001282, + "num_tokens": 727061773.0, + "step": 19054 + }, + { + "epoch": 2.4239918585421703, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7963708639144897, + "num_tokens": 727095877.0, + "step": 19055 + }, + { + "epoch": 2.424119068820761, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8102331161499023, + "num_tokens": 727128676.0, + "step": 19056 + }, + { + "epoch": 2.4242462790993513, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.8106116652488708, + "num_tokens": 727164528.0, + "step": 19057 + }, + { + "epoch": 2.424373489377942, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8108165264129639, + "num_tokens": 727202403.0, + "step": 19058 + }, + { + "epoch": 2.4245006996565324, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8094724416732788, + "num_tokens": 727242231.0, + "step": 19059 + }, + { + "epoch": 2.424627909935123, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7800965905189514, + "num_tokens": 727276371.0, + "step": 19060 + }, + { + "epoch": 2.4247551202137134, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8050440549850464, + "num_tokens": 727320101.0, + "step": 19061 + }, + { + "epoch": 2.424882330492304, + "grad_norm": 11.0625, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7937079668045044, + "num_tokens": 727363687.0, + "step": 19062 + }, + { + "epoch": 2.4250095407708945, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8024822473526001, + "num_tokens": 727399772.0, + "step": 19063 + }, + { + "epoch": 2.425136751049485, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8158429861068726, + "num_tokens": 727442944.0, + "step": 19064 + }, + { + "epoch": 2.425263961328075, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8092283010482788, + "num_tokens": 727483283.0, + "step": 19065 + }, + { + "epoch": 2.425391171606666, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7986553907394409, + "num_tokens": 727526316.0, + "step": 19066 + }, + { + "epoch": 2.425518381885256, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8095518350601196, + "num_tokens": 727563218.0, + "step": 19067 + }, + { + "epoch": 2.4256455921638467, + "grad_norm": 64.0, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.80986487865448, + "num_tokens": 727594877.0, + "step": 19068 + }, + { + "epoch": 2.425772802442437, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.809350848197937, + "num_tokens": 727627561.0, + "step": 19069 + }, + { + "epoch": 2.4259000127210277, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7890124320983887, + "num_tokens": 727664384.0, + "step": 19070 + }, + { + "epoch": 2.4260272229996183, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7983741164207458, + "num_tokens": 727700866.0, + "step": 19071 + }, + { + "epoch": 2.426154433278209, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8227242231369019, + "num_tokens": 727741539.0, + "step": 19072 + }, + { + "epoch": 2.4262816435567993, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8043226003646851, + "num_tokens": 727777828.0, + "step": 19073 + }, + { + "epoch": 2.42640885383539, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8094639778137207, + "num_tokens": 727818979.0, + "step": 19074 + }, + { + "epoch": 2.4265360641139804, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7967643737792969, + "num_tokens": 727856440.0, + "step": 19075 + }, + { + "epoch": 2.426663274392571, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8117227554321289, + "num_tokens": 727895036.0, + "step": 19076 + }, + { + "epoch": 2.4267904846711614, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.785564124584198, + "num_tokens": 727932292.0, + "step": 19077 + }, + { + "epoch": 2.426917694949752, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7986728549003601, + "num_tokens": 727967005.0, + "step": 19078 + }, + { + "epoch": 2.4270449052283425, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7887872457504272, + "num_tokens": 728004262.0, + "step": 19079 + }, + { + "epoch": 2.427172115506933, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.8153585195541382, + "num_tokens": 728041451.0, + "step": 19080 + }, + { + "epoch": 2.4272993257855235, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7998867630958557, + "num_tokens": 728079217.0, + "step": 19081 + }, + { + "epoch": 2.427426536064114, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.8233497142791748, + "num_tokens": 728116221.0, + "step": 19082 + }, + { + "epoch": 2.4275537463427046, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7851271629333496, + "num_tokens": 728154861.0, + "step": 19083 + }, + { + "epoch": 2.427680956621295, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7928146719932556, + "num_tokens": 728188276.0, + "step": 19084 + }, + { + "epoch": 2.4278081668998857, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8153469562530518, + "num_tokens": 728229313.0, + "step": 19085 + }, + { + "epoch": 2.427935377178476, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7956421375274658, + "num_tokens": 728268706.0, + "step": 19086 + }, + { + "epoch": 2.4280625874570667, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8154034614562988, + "num_tokens": 728297490.0, + "step": 19087 + }, + { + "epoch": 2.428189797735657, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7919344902038574, + "num_tokens": 728336041.0, + "step": 19088 + }, + { + "epoch": 2.4283170080142478, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8133254051208496, + "num_tokens": 728372701.0, + "step": 19089 + }, + { + "epoch": 2.428444218292838, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8066211938858032, + "num_tokens": 728408138.0, + "step": 19090 + }, + { + "epoch": 2.4285714285714284, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.8074741363525391, + "num_tokens": 728452649.0, + "step": 19091 + }, + { + "epoch": 2.428698638850019, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.786404013633728, + "num_tokens": 728490551.0, + "step": 19092 + }, + { + "epoch": 2.4288258491286094, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.806096613407135, + "num_tokens": 728529558.0, + "step": 19093 + }, + { + "epoch": 2.4289530594072, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7768023014068604, + "num_tokens": 728573495.0, + "step": 19094 + }, + { + "epoch": 2.4290802696857905, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.8000607490539551, + "num_tokens": 728609166.0, + "step": 19095 + }, + { + "epoch": 2.429207479964381, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8105280995368958, + "num_tokens": 728653070.0, + "step": 19096 + }, + { + "epoch": 2.4293346902429716, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.8027942180633545, + "num_tokens": 728689142.0, + "step": 19097 + }, + { + "epoch": 2.429461900521562, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.80705726146698, + "num_tokens": 728725615.0, + "step": 19098 + }, + { + "epoch": 2.4295891108001526, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7881948947906494, + "num_tokens": 728762366.0, + "step": 19099 + }, + { + "epoch": 2.429716321078743, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7867540717124939, + "num_tokens": 728801951.0, + "step": 19100 + }, + { + "epoch": 2.4298435313573337, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8058854341506958, + "num_tokens": 728836403.0, + "step": 19101 + }, + { + "epoch": 2.429970741635924, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8092431426048279, + "num_tokens": 728878485.0, + "step": 19102 + }, + { + "epoch": 2.4300979519145147, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8124065399169922, + "num_tokens": 728915935.0, + "step": 19103 + }, + { + "epoch": 2.4302251621931052, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7855859994888306, + "num_tokens": 728956847.0, + "step": 19104 + }, + { + "epoch": 2.4303523724716958, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8051774501800537, + "num_tokens": 728994865.0, + "step": 19105 + }, + { + "epoch": 2.4304795827502863, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7929040193557739, + "num_tokens": 729033030.0, + "step": 19106 + }, + { + "epoch": 2.430606793028877, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8160128593444824, + "num_tokens": 729070430.0, + "step": 19107 + }, + { + "epoch": 2.4307340033074674, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8161023855209351, + "num_tokens": 729102127.0, + "step": 19108 + }, + { + "epoch": 2.430861213586058, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8060553073883057, + "num_tokens": 729138128.0, + "step": 19109 + }, + { + "epoch": 2.4309884238646484, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.797783374786377, + "num_tokens": 729179756.0, + "step": 19110 + }, + { + "epoch": 2.431115634143239, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7899628281593323, + "num_tokens": 729221356.0, + "step": 19111 + }, + { + "epoch": 2.4312428444218295, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.8049682974815369, + "num_tokens": 729260456.0, + "step": 19112 + }, + { + "epoch": 2.4313700547004196, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.788844645023346, + "num_tokens": 729299654.0, + "step": 19113 + }, + { + "epoch": 2.4314972649790105, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8037011623382568, + "num_tokens": 729341296.0, + "step": 19114 + }, + { + "epoch": 2.4316244752576006, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8030961751937866, + "num_tokens": 729379632.0, + "step": 19115 + }, + { + "epoch": 2.431751685536191, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8010127544403076, + "num_tokens": 729423202.0, + "step": 19116 + }, + { + "epoch": 2.4318788958147817, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.5674, + "mean_token_accuracy": 0.822787344455719, + "num_tokens": 729457108.0, + "step": 19117 + }, + { + "epoch": 2.432006106093372, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7990403175354004, + "num_tokens": 729494065.0, + "step": 19118 + }, + { + "epoch": 2.4321333163719627, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8069737553596497, + "num_tokens": 729536621.0, + "step": 19119 + }, + { + "epoch": 2.4322605266505533, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8002384901046753, + "num_tokens": 729574423.0, + "step": 19120 + }, + { + "epoch": 2.432387736929144, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.79756760597229, + "num_tokens": 729614673.0, + "step": 19121 + }, + { + "epoch": 2.4325149472077343, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.8015623688697815, + "num_tokens": 729653049.0, + "step": 19122 + }, + { + "epoch": 2.432642157486325, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.794725775718689, + "num_tokens": 729689667.0, + "step": 19123 + }, + { + "epoch": 2.4327693677649154, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8035570979118347, + "num_tokens": 729734747.0, + "step": 19124 + }, + { + "epoch": 2.432896578043506, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7974619269371033, + "num_tokens": 729770135.0, + "step": 19125 + }, + { + "epoch": 2.4330237883220964, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8016590476036072, + "num_tokens": 729809203.0, + "step": 19126 + }, + { + "epoch": 2.433150998600687, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8069158792495728, + "num_tokens": 729850164.0, + "step": 19127 + }, + { + "epoch": 2.4332782088792775, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.78802490234375, + "num_tokens": 729884337.0, + "step": 19128 + }, + { + "epoch": 2.433405419157868, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.8148539066314697, + "num_tokens": 729924098.0, + "step": 19129 + }, + { + "epoch": 2.4335326294364585, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.5738, + "mean_token_accuracy": 0.828490138053894, + "num_tokens": 729960671.0, + "step": 19130 + }, + { + "epoch": 2.433659839715049, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8051127791404724, + "num_tokens": 730000780.0, + "step": 19131 + }, + { + "epoch": 2.4337870499936396, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.799257755279541, + "num_tokens": 730045198.0, + "step": 19132 + }, + { + "epoch": 2.43391426027223, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7963265776634216, + "num_tokens": 730085010.0, + "step": 19133 + }, + { + "epoch": 2.4340414705508207, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7973678112030029, + "num_tokens": 730121512.0, + "step": 19134 + }, + { + "epoch": 2.434168680829411, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8028982877731323, + "num_tokens": 730164859.0, + "step": 19135 + }, + { + "epoch": 2.4342958911080017, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8004143238067627, + "num_tokens": 730202273.0, + "step": 19136 + }, + { + "epoch": 2.4344231013865922, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7903530597686768, + "num_tokens": 730248208.0, + "step": 19137 + }, + { + "epoch": 2.4345503116651823, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.81023108959198, + "num_tokens": 730291084.0, + "step": 19138 + }, + { + "epoch": 2.4346775219437733, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8122093677520752, + "num_tokens": 730334504.0, + "step": 19139 + }, + { + "epoch": 2.4348047322223634, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.806219756603241, + "num_tokens": 730373356.0, + "step": 19140 + }, + { + "epoch": 2.434931942500954, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7853174209594727, + "num_tokens": 730416229.0, + "step": 19141 + }, + { + "epoch": 2.4350591527795444, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8149732351303101, + "num_tokens": 730458922.0, + "step": 19142 + }, + { + "epoch": 2.435186363058135, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7996144890785217, + "num_tokens": 730494583.0, + "step": 19143 + }, + { + "epoch": 2.4353135733367255, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7971237897872925, + "num_tokens": 730529526.0, + "step": 19144 + }, + { + "epoch": 2.435440783615316, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7978233695030212, + "num_tokens": 730558584.0, + "step": 19145 + }, + { + "epoch": 2.4355679938939065, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7992255687713623, + "num_tokens": 730598677.0, + "step": 19146 + }, + { + "epoch": 2.435695204172497, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8132121562957764, + "num_tokens": 730636985.0, + "step": 19147 + }, + { + "epoch": 2.4358224144510876, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7824401259422302, + "num_tokens": 730675409.0, + "step": 19148 + }, + { + "epoch": 2.435949624729678, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7899999022483826, + "num_tokens": 730714739.0, + "step": 19149 + }, + { + "epoch": 2.4360768350082687, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7956216335296631, + "num_tokens": 730757059.0, + "step": 19150 + }, + { + "epoch": 2.436204045286859, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7954100370407104, + "num_tokens": 730793469.0, + "step": 19151 + }, + { + "epoch": 2.4363312555654497, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8082801699638367, + "num_tokens": 730829336.0, + "step": 19152 + }, + { + "epoch": 2.4364584658440402, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8044775724411011, + "num_tokens": 730869304.0, + "step": 19153 + }, + { + "epoch": 2.4365856761226308, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7865744829177856, + "num_tokens": 730905315.0, + "step": 19154 + }, + { + "epoch": 2.4367128864012213, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7958616614341736, + "num_tokens": 730947346.0, + "step": 19155 + }, + { + "epoch": 2.436840096679812, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8000684976577759, + "num_tokens": 730976907.0, + "step": 19156 + }, + { + "epoch": 2.4369673069584024, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7978044152259827, + "num_tokens": 731015940.0, + "step": 19157 + }, + { + "epoch": 2.437094517236993, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.5851, + "mean_token_accuracy": 0.8201018571853638, + "num_tokens": 731051345.0, + "step": 19158 + }, + { + "epoch": 2.4372217275155834, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7691200375556946, + "num_tokens": 731089642.0, + "step": 19159 + }, + { + "epoch": 2.437348937794174, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.8161517381668091, + "num_tokens": 731133196.0, + "step": 19160 + }, + { + "epoch": 2.4374761480727645, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7923485636711121, + "num_tokens": 731175510.0, + "step": 19161 + }, + { + "epoch": 2.437603358351355, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8123478889465332, + "num_tokens": 731217526.0, + "step": 19162 + }, + { + "epoch": 2.437730568629945, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7927371263504028, + "num_tokens": 731259525.0, + "step": 19163 + }, + { + "epoch": 2.437857778908536, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8092111349105835, + "num_tokens": 731301480.0, + "step": 19164 + }, + { + "epoch": 2.437984989187126, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8087472915649414, + "num_tokens": 731347385.0, + "step": 19165 + }, + { + "epoch": 2.4381121994657167, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.804299533367157, + "num_tokens": 731383666.0, + "step": 19166 + }, + { + "epoch": 2.438239409744307, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7988444566726685, + "num_tokens": 731420297.0, + "step": 19167 + }, + { + "epoch": 2.4383666200228977, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7956622838973999, + "num_tokens": 731455083.0, + "step": 19168 + }, + { + "epoch": 2.4384938303014883, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8048169612884521, + "num_tokens": 731482555.0, + "step": 19169 + }, + { + "epoch": 2.438621040580079, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8161081075668335, + "num_tokens": 731519459.0, + "step": 19170 + }, + { + "epoch": 2.4387482508586693, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8054172992706299, + "num_tokens": 731554577.0, + "step": 19171 + }, + { + "epoch": 2.43887546113726, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7931841611862183, + "num_tokens": 731594993.0, + "step": 19172 + }, + { + "epoch": 2.4390026714158504, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8048701286315918, + "num_tokens": 731629182.0, + "step": 19173 + }, + { + "epoch": 2.439129881694441, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7977347373962402, + "num_tokens": 731667661.0, + "step": 19174 + }, + { + "epoch": 2.4392570919730314, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8058477640151978, + "num_tokens": 731704911.0, + "step": 19175 + }, + { + "epoch": 2.439384302251622, + "grad_norm": 80.5, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7794603109359741, + "num_tokens": 731744990.0, + "step": 19176 + }, + { + "epoch": 2.4395115125302125, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.809135377407074, + "num_tokens": 731783335.0, + "step": 19177 + }, + { + "epoch": 2.439638722808803, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8036579489707947, + "num_tokens": 731819206.0, + "step": 19178 + }, + { + "epoch": 2.4397659330873935, + "grad_norm": 10.4375, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7983250617980957, + "num_tokens": 731855291.0, + "step": 19179 + }, + { + "epoch": 2.439893143365984, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8050335049629211, + "num_tokens": 731899392.0, + "step": 19180 + }, + { + "epoch": 2.4400203536445746, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8006399869918823, + "num_tokens": 731937502.0, + "step": 19181 + }, + { + "epoch": 2.440147563923165, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.801985502243042, + "num_tokens": 731974946.0, + "step": 19182 + }, + { + "epoch": 2.4402747742017556, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7889741659164429, + "num_tokens": 732018732.0, + "step": 19183 + }, + { + "epoch": 2.440401984480346, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8001958727836609, + "num_tokens": 732052036.0, + "step": 19184 + }, + { + "epoch": 2.4405291947589367, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7981705665588379, + "num_tokens": 732090944.0, + "step": 19185 + }, + { + "epoch": 2.440656405037527, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.804000973701477, + "num_tokens": 732131406.0, + "step": 19186 + }, + { + "epoch": 2.4407836153161178, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.8008582592010498, + "num_tokens": 732171701.0, + "step": 19187 + }, + { + "epoch": 2.440910825594708, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.8117831945419312, + "num_tokens": 732209658.0, + "step": 19188 + }, + { + "epoch": 2.4410380358732984, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7929450869560242, + "num_tokens": 732251395.0, + "step": 19189 + }, + { + "epoch": 2.441165246151889, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7881948947906494, + "num_tokens": 732289306.0, + "step": 19190 + }, + { + "epoch": 2.4412924564304794, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7879116535186768, + "num_tokens": 732337916.0, + "step": 19191 + }, + { + "epoch": 2.44141966670907, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7913328409194946, + "num_tokens": 732380358.0, + "step": 19192 + }, + { + "epoch": 2.4415468769876605, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8120679259300232, + "num_tokens": 732416534.0, + "step": 19193 + }, + { + "epoch": 2.441674087266251, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7756541967391968, + "num_tokens": 732458927.0, + "step": 19194 + }, + { + "epoch": 2.4418012975448415, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7941107749938965, + "num_tokens": 732501261.0, + "step": 19195 + }, + { + "epoch": 2.441928507823432, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8084393739700317, + "num_tokens": 732541162.0, + "step": 19196 + }, + { + "epoch": 2.4420557181020226, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7975279092788696, + "num_tokens": 732578237.0, + "step": 19197 + }, + { + "epoch": 2.442182928380613, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7874219417572021, + "num_tokens": 732618624.0, + "step": 19198 + }, + { + "epoch": 2.4423101386592037, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8026587963104248, + "num_tokens": 732661747.0, + "step": 19199 + }, + { + "epoch": 2.442437348937794, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7791405916213989, + "num_tokens": 732703511.0, + "step": 19200 + }, + { + "epoch": 2.4425645592163847, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7813751697540283, + "num_tokens": 732746749.0, + "step": 19201 + }, + { + "epoch": 2.4426917694949752, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8062059283256531, + "num_tokens": 732783184.0, + "step": 19202 + }, + { + "epoch": 2.4428189797735658, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7868015766143799, + "num_tokens": 732818664.0, + "step": 19203 + }, + { + "epoch": 2.4429461900521563, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7880077958106995, + "num_tokens": 732854286.0, + "step": 19204 + }, + { + "epoch": 2.443073400330747, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8093836307525635, + "num_tokens": 732893032.0, + "step": 19205 + }, + { + "epoch": 2.4432006106093374, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8030338883399963, + "num_tokens": 732929942.0, + "step": 19206 + }, + { + "epoch": 2.443327820887928, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7963546514511108, + "num_tokens": 732968218.0, + "step": 19207 + }, + { + "epoch": 2.4434550311665184, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.5725, + "mean_token_accuracy": 0.8272311687469482, + "num_tokens": 733001464.0, + "step": 19208 + }, + { + "epoch": 2.443582241445109, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7905246615409851, + "num_tokens": 733036311.0, + "step": 19209 + }, + { + "epoch": 2.4437094517236995, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8045303821563721, + "num_tokens": 733070487.0, + "step": 19210 + }, + { + "epoch": 2.4438366620022896, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7978297472000122, + "num_tokens": 733110357.0, + "step": 19211 + }, + { + "epoch": 2.4439638722808805, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7899119853973389, + "num_tokens": 733149070.0, + "step": 19212 + }, + { + "epoch": 2.4440910825594706, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.820576548576355, + "num_tokens": 733180848.0, + "step": 19213 + }, + { + "epoch": 2.444218292838061, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7943868041038513, + "num_tokens": 733219204.0, + "step": 19214 + }, + { + "epoch": 2.4443455031166517, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7883094549179077, + "num_tokens": 733255237.0, + "step": 19215 + }, + { + "epoch": 2.444472713395242, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8140978813171387, + "num_tokens": 733293243.0, + "step": 19216 + }, + { + "epoch": 2.4445999236738327, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.80908203125, + "num_tokens": 733333530.0, + "step": 19217 + }, + { + "epoch": 2.4447271339524232, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7835170030593872, + "num_tokens": 733373790.0, + "step": 19218 + }, + { + "epoch": 2.4448543442310138, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7987536191940308, + "num_tokens": 733416037.0, + "step": 19219 + }, + { + "epoch": 2.4449815545096043, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7950155138969421, + "num_tokens": 733453014.0, + "step": 19220 + }, + { + "epoch": 2.445108764788195, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7907391786575317, + "num_tokens": 733491197.0, + "step": 19221 + }, + { + "epoch": 2.4452359750667854, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.8169530630111694, + "num_tokens": 733531142.0, + "step": 19222 + }, + { + "epoch": 2.445363185345376, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7929372787475586, + "num_tokens": 733573916.0, + "step": 19223 + }, + { + "epoch": 2.4454903956239664, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7968055009841919, + "num_tokens": 733613204.0, + "step": 19224 + }, + { + "epoch": 2.445617605902557, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7923212647438049, + "num_tokens": 733653618.0, + "step": 19225 + }, + { + "epoch": 2.4457448161811475, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.8191094398498535, + "num_tokens": 733687795.0, + "step": 19226 + }, + { + "epoch": 2.445872026459738, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.795954167842865, + "num_tokens": 733723574.0, + "step": 19227 + }, + { + "epoch": 2.4459992367383285, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.810478925704956, + "num_tokens": 733762051.0, + "step": 19228 + }, + { + "epoch": 2.446126447016919, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.822629451751709, + "num_tokens": 733800640.0, + "step": 19229 + }, + { + "epoch": 2.4462536572955096, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7879482507705688, + "num_tokens": 733841321.0, + "step": 19230 + }, + { + "epoch": 2.4463808675741, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7999691963195801, + "num_tokens": 733874104.0, + "step": 19231 + }, + { + "epoch": 2.4465080778526906, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.802987813949585, + "num_tokens": 733913127.0, + "step": 19232 + }, + { + "epoch": 2.446635288131281, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.8191792964935303, + "num_tokens": 733951900.0, + "step": 19233 + }, + { + "epoch": 2.4467624984098717, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8034077882766724, + "num_tokens": 733992866.0, + "step": 19234 + }, + { + "epoch": 2.4468897086884622, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7958575487136841, + "num_tokens": 734029895.0, + "step": 19235 + }, + { + "epoch": 2.4470169189670523, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8178673982620239, + "num_tokens": 734066954.0, + "step": 19236 + }, + { + "epoch": 2.4471441292456433, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.5683, + "mean_token_accuracy": 0.8268370628356934, + "num_tokens": 734102412.0, + "step": 19237 + }, + { + "epoch": 2.4472713395242334, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8185139894485474, + "num_tokens": 734141400.0, + "step": 19238 + }, + { + "epoch": 2.447398549802824, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7813134789466858, + "num_tokens": 734176628.0, + "step": 19239 + }, + { + "epoch": 2.4475257600814144, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8080437779426575, + "num_tokens": 734219756.0, + "step": 19240 + }, + { + "epoch": 2.447652970360005, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7927898168563843, + "num_tokens": 734250509.0, + "step": 19241 + }, + { + "epoch": 2.4477801806385955, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7998621463775635, + "num_tokens": 734287887.0, + "step": 19242 + }, + { + "epoch": 2.447907390917186, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8099806904792786, + "num_tokens": 734330446.0, + "step": 19243 + }, + { + "epoch": 2.4480346011957765, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8108673095703125, + "num_tokens": 734368427.0, + "step": 19244 + }, + { + "epoch": 2.448161811474367, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.8192564249038696, + "num_tokens": 734405103.0, + "step": 19245 + }, + { + "epoch": 2.4482890217529576, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.8023083806037903, + "num_tokens": 734448069.0, + "step": 19246 + }, + { + "epoch": 2.448416232031548, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7829761505126953, + "num_tokens": 734486016.0, + "step": 19247 + }, + { + "epoch": 2.4485434423101387, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7841521501541138, + "num_tokens": 734526943.0, + "step": 19248 + }, + { + "epoch": 2.448670652588729, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8132195472717285, + "num_tokens": 734566137.0, + "step": 19249 + }, + { + "epoch": 2.4487978628673197, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.8000984191894531, + "num_tokens": 734602366.0, + "step": 19250 + }, + { + "epoch": 2.4489250731459102, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8174166679382324, + "num_tokens": 734644251.0, + "step": 19251 + }, + { + "epoch": 2.4490522834245008, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7951050996780396, + "num_tokens": 734682938.0, + "step": 19252 + }, + { + "epoch": 2.4491794937030913, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8019533157348633, + "num_tokens": 734722975.0, + "step": 19253 + }, + { + "epoch": 2.449306703981682, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8054617643356323, + "num_tokens": 734761361.0, + "step": 19254 + }, + { + "epoch": 2.4494339142602723, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7930623888969421, + "num_tokens": 734800597.0, + "step": 19255 + }, + { + "epoch": 2.449561124538863, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7912641763687134, + "num_tokens": 734840575.0, + "step": 19256 + }, + { + "epoch": 2.4496883348174534, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7922006845474243, + "num_tokens": 734876943.0, + "step": 19257 + }, + { + "epoch": 2.449815545096044, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.8192324638366699, + "num_tokens": 734916436.0, + "step": 19258 + }, + { + "epoch": 2.4499427553746345, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7994711995124817, + "num_tokens": 734956336.0, + "step": 19259 + }, + { + "epoch": 2.450069965653225, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7977211475372314, + "num_tokens": 734997132.0, + "step": 19260 + }, + { + "epoch": 2.450197175931815, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.788631796836853, + "num_tokens": 735036192.0, + "step": 19261 + }, + { + "epoch": 2.450324386210406, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8221821784973145, + "num_tokens": 735079368.0, + "step": 19262 + }, + { + "epoch": 2.450451596488996, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.806435227394104, + "num_tokens": 735109714.0, + "step": 19263 + }, + { + "epoch": 2.4505788067675867, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8018485903739929, + "num_tokens": 735152712.0, + "step": 19264 + }, + { + "epoch": 2.450706017046177, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7983318567276001, + "num_tokens": 735193313.0, + "step": 19265 + }, + { + "epoch": 2.4508332273247677, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8188958168029785, + "num_tokens": 735235413.0, + "step": 19266 + }, + { + "epoch": 2.4509604376033582, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.805837869644165, + "num_tokens": 735270809.0, + "step": 19267 + }, + { + "epoch": 2.4510876478819488, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7947035431861877, + "num_tokens": 735309790.0, + "step": 19268 + }, + { + "epoch": 2.4512148581605393, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.798856258392334, + "num_tokens": 735341084.0, + "step": 19269 + }, + { + "epoch": 2.45134206843913, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8184722661972046, + "num_tokens": 735376964.0, + "step": 19270 + }, + { + "epoch": 2.4514692787177204, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8152702450752258, + "num_tokens": 735413296.0, + "step": 19271 + }, + { + "epoch": 2.451596488996311, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.8083902597427368, + "num_tokens": 735454992.0, + "step": 19272 + }, + { + "epoch": 2.4517236992749014, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8022251129150391, + "num_tokens": 735493315.0, + "step": 19273 + }, + { + "epoch": 2.451850909553492, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8144654035568237, + "num_tokens": 735532666.0, + "step": 19274 + }, + { + "epoch": 2.4519781198320825, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.797120988368988, + "num_tokens": 735571987.0, + "step": 19275 + }, + { + "epoch": 2.452105330110673, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.796043872833252, + "num_tokens": 735609409.0, + "step": 19276 + }, + { + "epoch": 2.4522325403892635, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7704834938049316, + "num_tokens": 735647299.0, + "step": 19277 + }, + { + "epoch": 2.452359750667854, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8044959306716919, + "num_tokens": 735677757.0, + "step": 19278 + }, + { + "epoch": 2.4524869609464446, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.811711311340332, + "num_tokens": 735723924.0, + "step": 19279 + }, + { + "epoch": 2.452614171225035, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7827574014663696, + "num_tokens": 735758105.0, + "step": 19280 + }, + { + "epoch": 2.4527413815036256, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.8005035519599915, + "num_tokens": 735797603.0, + "step": 19281 + }, + { + "epoch": 2.452868591782216, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7733021974563599, + "num_tokens": 735838958.0, + "step": 19282 + }, + { + "epoch": 2.4529958020608067, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7969773411750793, + "num_tokens": 735878894.0, + "step": 19283 + }, + { + "epoch": 2.453123012339397, + "grad_norm": 50.25, + "learning_rate": 1e-06, + "loss": 0.5505, + "mean_token_accuracy": 0.8329346179962158, + "num_tokens": 735913222.0, + "step": 19284 + }, + { + "epoch": 2.4532502226179878, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8085935115814209, + "num_tokens": 735951884.0, + "step": 19285 + }, + { + "epoch": 2.453377432896578, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7991905212402344, + "num_tokens": 735998955.0, + "step": 19286 + }, + { + "epoch": 2.4535046431751684, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7961641550064087, + "num_tokens": 736039631.0, + "step": 19287 + }, + { + "epoch": 2.453631853453759, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8040031790733337, + "num_tokens": 736085100.0, + "step": 19288 + }, + { + "epoch": 2.4537590637323494, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8180590867996216, + "num_tokens": 736119680.0, + "step": 19289 + }, + { + "epoch": 2.45388627401094, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8151573538780212, + "num_tokens": 736159769.0, + "step": 19290 + }, + { + "epoch": 2.4540134842895305, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8127761483192444, + "num_tokens": 736199566.0, + "step": 19291 + }, + { + "epoch": 2.454140694568121, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8042629957199097, + "num_tokens": 736239238.0, + "step": 19292 + }, + { + "epoch": 2.4542679048467115, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8048214912414551, + "num_tokens": 736274576.0, + "step": 19293 + }, + { + "epoch": 2.454395115125302, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7889610528945923, + "num_tokens": 736307509.0, + "step": 19294 + }, + { + "epoch": 2.4545223254038926, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8013623952865601, + "num_tokens": 736348313.0, + "step": 19295 + }, + { + "epoch": 2.454649535682483, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7980915307998657, + "num_tokens": 736380716.0, + "step": 19296 + }, + { + "epoch": 2.4547767459610736, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.782438337802887, + "num_tokens": 736422669.0, + "step": 19297 + }, + { + "epoch": 2.454903956239664, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8036396503448486, + "num_tokens": 736452929.0, + "step": 19298 + }, + { + "epoch": 2.4550311665182547, + "grad_norm": 50.5, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8012612462043762, + "num_tokens": 736486755.0, + "step": 19299 + }, + { + "epoch": 2.4551583767968452, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8103810548782349, + "num_tokens": 736521542.0, + "step": 19300 + }, + { + "epoch": 2.4552855870754358, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8015033006668091, + "num_tokens": 736557312.0, + "step": 19301 + }, + { + "epoch": 2.4554127973540263, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8002082109451294, + "num_tokens": 736596469.0, + "step": 19302 + }, + { + "epoch": 2.455540007632617, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7868742942810059, + "num_tokens": 736636078.0, + "step": 19303 + }, + { + "epoch": 2.4556672179112073, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8151382207870483, + "num_tokens": 736674117.0, + "step": 19304 + }, + { + "epoch": 2.455794428189798, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7978496551513672, + "num_tokens": 736709116.0, + "step": 19305 + }, + { + "epoch": 2.4559216384683884, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7941336631774902, + "num_tokens": 736750383.0, + "step": 19306 + }, + { + "epoch": 2.456048848746979, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7905383110046387, + "num_tokens": 736790836.0, + "step": 19307 + }, + { + "epoch": 2.4561760590255695, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7912377715110779, + "num_tokens": 736826085.0, + "step": 19308 + }, + { + "epoch": 2.4563032693041595, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7999280095100403, + "num_tokens": 736860599.0, + "step": 19309 + }, + { + "epoch": 2.4564304795827505, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5786, + "mean_token_accuracy": 0.8215692043304443, + "num_tokens": 736893286.0, + "step": 19310 + }, + { + "epoch": 2.4565576898613406, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8108410835266113, + "num_tokens": 736931597.0, + "step": 19311 + }, + { + "epoch": 2.456684900139931, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8087148070335388, + "num_tokens": 736972183.0, + "step": 19312 + }, + { + "epoch": 2.4568121104185217, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8056552410125732, + "num_tokens": 737004184.0, + "step": 19313 + }, + { + "epoch": 2.456939320697112, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8012134432792664, + "num_tokens": 737042644.0, + "step": 19314 + }, + { + "epoch": 2.4570665309757027, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7928407192230225, + "num_tokens": 737074830.0, + "step": 19315 + }, + { + "epoch": 2.4571937412542932, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8059546947479248, + "num_tokens": 737107756.0, + "step": 19316 + }, + { + "epoch": 2.4573209515328838, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8082543611526489, + "num_tokens": 737143457.0, + "step": 19317 + }, + { + "epoch": 2.4574481618114743, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7961801886558533, + "num_tokens": 737177005.0, + "step": 19318 + }, + { + "epoch": 2.457575372090065, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7955625057220459, + "num_tokens": 737218265.0, + "step": 19319 + }, + { + "epoch": 2.4577025823686554, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7922145128250122, + "num_tokens": 737259495.0, + "step": 19320 + }, + { + "epoch": 2.457829792647246, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7981938123703003, + "num_tokens": 737300073.0, + "step": 19321 + }, + { + "epoch": 2.4579570029258364, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.801002025604248, + "num_tokens": 737334716.0, + "step": 19322 + }, + { + "epoch": 2.458084213204427, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8059564828872681, + "num_tokens": 737377086.0, + "step": 19323 + }, + { + "epoch": 2.4582114234830175, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7939841747283936, + "num_tokens": 737418259.0, + "step": 19324 + }, + { + "epoch": 2.458338633761608, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7872968316078186, + "num_tokens": 737453150.0, + "step": 19325 + }, + { + "epoch": 2.4584658440401985, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8204259872436523, + "num_tokens": 737495552.0, + "step": 19326 + }, + { + "epoch": 2.458593054318789, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.8108553290367126, + "num_tokens": 737537518.0, + "step": 19327 + }, + { + "epoch": 2.4587202645973796, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.800334632396698, + "num_tokens": 737577472.0, + "step": 19328 + }, + { + "epoch": 2.45884747487597, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7922103404998779, + "num_tokens": 737617250.0, + "step": 19329 + }, + { + "epoch": 2.4589746851545606, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8057020902633667, + "num_tokens": 737655544.0, + "step": 19330 + }, + { + "epoch": 2.459101895433151, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8002179265022278, + "num_tokens": 737688644.0, + "step": 19331 + }, + { + "epoch": 2.4592291057117417, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8034137487411499, + "num_tokens": 737730264.0, + "step": 19332 + }, + { + "epoch": 2.459356315990332, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7964715957641602, + "num_tokens": 737763578.0, + "step": 19333 + }, + { + "epoch": 2.4594835262689223, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8030405044555664, + "num_tokens": 737804062.0, + "step": 19334 + }, + { + "epoch": 2.4596107365475133, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.817815899848938, + "num_tokens": 737845608.0, + "step": 19335 + }, + { + "epoch": 2.4597379468261034, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7763269543647766, + "num_tokens": 737881366.0, + "step": 19336 + }, + { + "epoch": 2.459865157104694, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7968482375144958, + "num_tokens": 737922595.0, + "step": 19337 + }, + { + "epoch": 2.4599923673832844, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.807118833065033, + "num_tokens": 737956368.0, + "step": 19338 + }, + { + "epoch": 2.460119577661875, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8101649284362793, + "num_tokens": 738000167.0, + "step": 19339 + }, + { + "epoch": 2.4602467879404655, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7913596034049988, + "num_tokens": 738040850.0, + "step": 19340 + }, + { + "epoch": 2.460373998219056, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7996898889541626, + "num_tokens": 738080370.0, + "step": 19341 + }, + { + "epoch": 2.4605012084976465, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8065845370292664, + "num_tokens": 738116751.0, + "step": 19342 + }, + { + "epoch": 2.460628418776237, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.5732, + "mean_token_accuracy": 0.8225609660148621, + "num_tokens": 738154597.0, + "step": 19343 + }, + { + "epoch": 2.4607556290548276, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8091741800308228, + "num_tokens": 738186484.0, + "step": 19344 + }, + { + "epoch": 2.460882839333418, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8024958372116089, + "num_tokens": 738217722.0, + "step": 19345 + }, + { + "epoch": 2.4610100496120086, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.815078616142273, + "num_tokens": 738255304.0, + "step": 19346 + }, + { + "epoch": 2.461137259890599, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8037139177322388, + "num_tokens": 738292162.0, + "step": 19347 + }, + { + "epoch": 2.4612644701691897, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8144814968109131, + "num_tokens": 738332851.0, + "step": 19348 + }, + { + "epoch": 2.4613916804477802, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8128447532653809, + "num_tokens": 738375120.0, + "step": 19349 + }, + { + "epoch": 2.4615188907263708, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7901952266693115, + "num_tokens": 738413857.0, + "step": 19350 + }, + { + "epoch": 2.4616461010049613, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7917256951332092, + "num_tokens": 738450097.0, + "step": 19351 + }, + { + "epoch": 2.461773311283552, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7861148715019226, + "num_tokens": 738483784.0, + "step": 19352 + }, + { + "epoch": 2.4619005215621423, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7809074521064758, + "num_tokens": 738526820.0, + "step": 19353 + }, + { + "epoch": 2.462027731840733, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7887383699417114, + "num_tokens": 738569049.0, + "step": 19354 + }, + { + "epoch": 2.4621549421193234, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.798443078994751, + "num_tokens": 738608320.0, + "step": 19355 + }, + { + "epoch": 2.462282152397914, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8158378601074219, + "num_tokens": 738645566.0, + "step": 19356 + }, + { + "epoch": 2.4624093626765045, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7775344848632812, + "num_tokens": 738686184.0, + "step": 19357 + }, + { + "epoch": 2.462536572955095, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7867108583450317, + "num_tokens": 738728820.0, + "step": 19358 + }, + { + "epoch": 2.462663783233685, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7658130526542664, + "num_tokens": 738760180.0, + "step": 19359 + }, + { + "epoch": 2.462790993512276, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8090335130691528, + "num_tokens": 738795501.0, + "step": 19360 + }, + { + "epoch": 2.462918203790866, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7913326621055603, + "num_tokens": 738831027.0, + "step": 19361 + }, + { + "epoch": 2.4630454140694567, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.781402587890625, + "num_tokens": 738870383.0, + "step": 19362 + }, + { + "epoch": 2.463172624348047, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.809296727180481, + "num_tokens": 738908231.0, + "step": 19363 + }, + { + "epoch": 2.4632998346266377, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7912315130233765, + "num_tokens": 738950927.0, + "step": 19364 + }, + { + "epoch": 2.4634270449052282, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.811121940612793, + "num_tokens": 738989857.0, + "step": 19365 + }, + { + "epoch": 2.4635542551838188, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.8119824528694153, + "num_tokens": 739021281.0, + "step": 19366 + }, + { + "epoch": 2.4636814654624093, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.8108301162719727, + "num_tokens": 739063475.0, + "step": 19367 + }, + { + "epoch": 2.463808675741, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7978023886680603, + "num_tokens": 739107083.0, + "step": 19368 + }, + { + "epoch": 2.4639358860195903, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7803287506103516, + "num_tokens": 739152002.0, + "step": 19369 + }, + { + "epoch": 2.464063096298181, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8001877069473267, + "num_tokens": 739191781.0, + "step": 19370 + }, + { + "epoch": 2.4641903065767714, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.8264236450195312, + "num_tokens": 739230374.0, + "step": 19371 + }, + { + "epoch": 2.464317516855362, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7926124334335327, + "num_tokens": 739267310.0, + "step": 19372 + }, + { + "epoch": 2.4644447271339525, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8021160960197449, + "num_tokens": 739305074.0, + "step": 19373 + }, + { + "epoch": 2.464571937412543, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7920923233032227, + "num_tokens": 739343028.0, + "step": 19374 + }, + { + "epoch": 2.4646991476911335, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8037987947463989, + "num_tokens": 739384364.0, + "step": 19375 + }, + { + "epoch": 2.464826357969724, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8221613168716431, + "num_tokens": 739423564.0, + "step": 19376 + }, + { + "epoch": 2.4649535682483146, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8095811605453491, + "num_tokens": 739457609.0, + "step": 19377 + }, + { + "epoch": 2.465080778526905, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8050087690353394, + "num_tokens": 739497204.0, + "step": 19378 + }, + { + "epoch": 2.4652079888054956, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8006010055541992, + "num_tokens": 739543688.0, + "step": 19379 + }, + { + "epoch": 2.465335199084086, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8092000484466553, + "num_tokens": 739578221.0, + "step": 19380 + }, + { + "epoch": 2.4654624093626767, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.806309700012207, + "num_tokens": 739615892.0, + "step": 19381 + }, + { + "epoch": 2.4655896196412668, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7866554856300354, + "num_tokens": 739657697.0, + "step": 19382 + }, + { + "epoch": 2.4657168299198577, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8123027086257935, + "num_tokens": 739694173.0, + "step": 19383 + }, + { + "epoch": 2.465844040198448, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8024173974990845, + "num_tokens": 739734111.0, + "step": 19384 + }, + { + "epoch": 2.4659712504770384, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7968519330024719, + "num_tokens": 739771409.0, + "step": 19385 + }, + { + "epoch": 2.466098460755629, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7849942445755005, + "num_tokens": 739807185.0, + "step": 19386 + }, + { + "epoch": 2.4662256710342194, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8107306957244873, + "num_tokens": 739837750.0, + "step": 19387 + }, + { + "epoch": 2.46635288131281, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8044561147689819, + "num_tokens": 739879516.0, + "step": 19388 + }, + { + "epoch": 2.4664800915914005, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8083842992782593, + "num_tokens": 739918688.0, + "step": 19389 + }, + { + "epoch": 2.466607301869991, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8059694170951843, + "num_tokens": 739958978.0, + "step": 19390 + }, + { + "epoch": 2.4667345121485815, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.8049138784408569, + "num_tokens": 740002428.0, + "step": 19391 + }, + { + "epoch": 2.466861722427172, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.8224201798439026, + "num_tokens": 740045856.0, + "step": 19392 + }, + { + "epoch": 2.4669889327057626, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8014406561851501, + "num_tokens": 740084512.0, + "step": 19393 + }, + { + "epoch": 2.467116142984353, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8151832818984985, + "num_tokens": 740129624.0, + "step": 19394 + }, + { + "epoch": 2.4672433532629436, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7819857597351074, + "num_tokens": 740168368.0, + "step": 19395 + }, + { + "epoch": 2.467370563541534, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7960756421089172, + "num_tokens": 740206670.0, + "step": 19396 + }, + { + "epoch": 2.4674977738201247, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7759096026420593, + "num_tokens": 740246422.0, + "step": 19397 + }, + { + "epoch": 2.4676249840987152, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7988554835319519, + "num_tokens": 740284450.0, + "step": 19398 + }, + { + "epoch": 2.4677521943773058, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7757828235626221, + "num_tokens": 740319062.0, + "step": 19399 + }, + { + "epoch": 2.4678794046558963, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7988293170928955, + "num_tokens": 740353996.0, + "step": 19400 + }, + { + "epoch": 2.468006614934487, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8012832403182983, + "num_tokens": 740394694.0, + "step": 19401 + }, + { + "epoch": 2.4681338252130773, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.5765, + "mean_token_accuracy": 0.8292629718780518, + "num_tokens": 740436567.0, + "step": 19402 + }, + { + "epoch": 2.468261035491668, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7889511585235596, + "num_tokens": 740476458.0, + "step": 19403 + }, + { + "epoch": 2.4683882457702584, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8068084716796875, + "num_tokens": 740514703.0, + "step": 19404 + }, + { + "epoch": 2.468515456048849, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8114882111549377, + "num_tokens": 740552328.0, + "step": 19405 + }, + { + "epoch": 2.4686426663274394, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7799457311630249, + "num_tokens": 740593141.0, + "step": 19406 + }, + { + "epoch": 2.4687698766060295, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8101781606674194, + "num_tokens": 740634097.0, + "step": 19407 + }, + { + "epoch": 2.4688970868846205, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8062296509742737, + "num_tokens": 740672917.0, + "step": 19408 + }, + { + "epoch": 2.4690242971632106, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7972824573516846, + "num_tokens": 740709351.0, + "step": 19409 + }, + { + "epoch": 2.469151507441801, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7835909724235535, + "num_tokens": 740744049.0, + "step": 19410 + }, + { + "epoch": 2.4692787177203916, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8048840761184692, + "num_tokens": 740784765.0, + "step": 19411 + }, + { + "epoch": 2.469405927998982, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7888488173484802, + "num_tokens": 740815859.0, + "step": 19412 + }, + { + "epoch": 2.4695331382775727, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.5757, + "mean_token_accuracy": 0.8255369663238525, + "num_tokens": 740854300.0, + "step": 19413 + }, + { + "epoch": 2.4696603485561632, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7850330471992493, + "num_tokens": 740891836.0, + "step": 19414 + }, + { + "epoch": 2.4697875588347538, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8102141618728638, + "num_tokens": 740927434.0, + "step": 19415 + }, + { + "epoch": 2.4699147691133443, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8070623874664307, + "num_tokens": 740973707.0, + "step": 19416 + }, + { + "epoch": 2.470041979391935, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8122683763504028, + "num_tokens": 741015953.0, + "step": 19417 + }, + { + "epoch": 2.4701691896705253, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8038626909255981, + "num_tokens": 741049634.0, + "step": 19418 + }, + { + "epoch": 2.470296399949116, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8099411725997925, + "num_tokens": 741084267.0, + "step": 19419 + }, + { + "epoch": 2.4704236102277064, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8095991611480713, + "num_tokens": 741122237.0, + "step": 19420 + }, + { + "epoch": 2.470550820506297, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7876294255256653, + "num_tokens": 741163593.0, + "step": 19421 + }, + { + "epoch": 2.4706780307848875, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7914800643920898, + "num_tokens": 741202328.0, + "step": 19422 + }, + { + "epoch": 2.470805241063478, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8190551996231079, + "num_tokens": 741240095.0, + "step": 19423 + }, + { + "epoch": 2.4709324513420685, + "grad_norm": 10.4375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8027907609939575, + "num_tokens": 741278509.0, + "step": 19424 + }, + { + "epoch": 2.471059661620659, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7932724952697754, + "num_tokens": 741321446.0, + "step": 19425 + }, + { + "epoch": 2.4711868718992496, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7973268628120422, + "num_tokens": 741360469.0, + "step": 19426 + }, + { + "epoch": 2.47131408217784, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7856947183609009, + "num_tokens": 741393833.0, + "step": 19427 + }, + { + "epoch": 2.4714412924564306, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8049513101577759, + "num_tokens": 741426956.0, + "step": 19428 + }, + { + "epoch": 2.471568502735021, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8062903881072998, + "num_tokens": 741464820.0, + "step": 19429 + }, + { + "epoch": 2.4716957130136117, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8090915679931641, + "num_tokens": 741499519.0, + "step": 19430 + }, + { + "epoch": 2.471822923292202, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7989007830619812, + "num_tokens": 741533443.0, + "step": 19431 + }, + { + "epoch": 2.4719501335707923, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.798631489276886, + "num_tokens": 741576849.0, + "step": 19432 + }, + { + "epoch": 2.4720773438493833, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8082798719406128, + "num_tokens": 741614653.0, + "step": 19433 + }, + { + "epoch": 2.4722045541279734, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7915552854537964, + "num_tokens": 741649354.0, + "step": 19434 + }, + { + "epoch": 2.472331764406564, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8028455972671509, + "num_tokens": 741692257.0, + "step": 19435 + }, + { + "epoch": 2.4724589746851544, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7997055053710938, + "num_tokens": 741729739.0, + "step": 19436 + }, + { + "epoch": 2.472586184963745, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7929985523223877, + "num_tokens": 741763409.0, + "step": 19437 + }, + { + "epoch": 2.4727133952423355, + "grad_norm": 56.0, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.795192539691925, + "num_tokens": 741798746.0, + "step": 19438 + }, + { + "epoch": 2.472840605520926, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8126296401023865, + "num_tokens": 741840488.0, + "step": 19439 + }, + { + "epoch": 2.4729678157995165, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8195999264717102, + "num_tokens": 741880839.0, + "step": 19440 + }, + { + "epoch": 2.473095026078107, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7962038516998291, + "num_tokens": 741919317.0, + "step": 19441 + }, + { + "epoch": 2.4732222363566976, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7784377932548523, + "num_tokens": 741955782.0, + "step": 19442 + }, + { + "epoch": 2.473349446635288, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7914414405822754, + "num_tokens": 741990410.0, + "step": 19443 + }, + { + "epoch": 2.4734766569138786, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7978387475013733, + "num_tokens": 742031536.0, + "step": 19444 + }, + { + "epoch": 2.473603867192469, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7913639545440674, + "num_tokens": 742067222.0, + "step": 19445 + }, + { + "epoch": 2.4737310774710597, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8174980878829956, + "num_tokens": 742114740.0, + "step": 19446 + }, + { + "epoch": 2.47385828774965, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7973185777664185, + "num_tokens": 742149344.0, + "step": 19447 + }, + { + "epoch": 2.4739854980282407, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8004839420318604, + "num_tokens": 742191089.0, + "step": 19448 + }, + { + "epoch": 2.4741127083068313, + "grad_norm": 63.5, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8011283278465271, + "num_tokens": 742224944.0, + "step": 19449 + }, + { + "epoch": 2.474239918585422, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7873405814170837, + "num_tokens": 742263674.0, + "step": 19450 + }, + { + "epoch": 2.4743671288640123, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7979168891906738, + "num_tokens": 742300424.0, + "step": 19451 + }, + { + "epoch": 2.474494339142603, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.808413028717041, + "num_tokens": 742333854.0, + "step": 19452 + }, + { + "epoch": 2.4746215494211934, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8169879913330078, + "num_tokens": 742377126.0, + "step": 19453 + }, + { + "epoch": 2.474748759699784, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8027373552322388, + "num_tokens": 742421284.0, + "step": 19454 + }, + { + "epoch": 2.4748759699783744, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7821035981178284, + "num_tokens": 742459124.0, + "step": 19455 + }, + { + "epoch": 2.475003180256965, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7873817682266235, + "num_tokens": 742493137.0, + "step": 19456 + }, + { + "epoch": 2.475130390535555, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8033044934272766, + "num_tokens": 742529805.0, + "step": 19457 + }, + { + "epoch": 2.475257600814146, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8119444847106934, + "num_tokens": 742560859.0, + "step": 19458 + }, + { + "epoch": 2.475384811092736, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8164107799530029, + "num_tokens": 742601465.0, + "step": 19459 + }, + { + "epoch": 2.4755120213713266, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7959680557250977, + "num_tokens": 742639242.0, + "step": 19460 + }, + { + "epoch": 2.475639231649917, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8030446767807007, + "num_tokens": 742676574.0, + "step": 19461 + }, + { + "epoch": 2.4757664419285077, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7878507375717163, + "num_tokens": 742713326.0, + "step": 19462 + }, + { + "epoch": 2.4758936522070982, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7807358503341675, + "num_tokens": 742756883.0, + "step": 19463 + }, + { + "epoch": 2.4760208624856888, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7948368787765503, + "num_tokens": 742794301.0, + "step": 19464 + }, + { + "epoch": 2.4761480727642793, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8022979497909546, + "num_tokens": 742832966.0, + "step": 19465 + }, + { + "epoch": 2.47627528304287, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.798008382320404, + "num_tokens": 742872926.0, + "step": 19466 + }, + { + "epoch": 2.4764024933214603, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8135266304016113, + "num_tokens": 742910838.0, + "step": 19467 + }, + { + "epoch": 2.476529703600051, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8108714818954468, + "num_tokens": 742950098.0, + "step": 19468 + }, + { + "epoch": 2.4766569138786414, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8122369050979614, + "num_tokens": 742983872.0, + "step": 19469 + }, + { + "epoch": 2.476784124157232, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7911102771759033, + "num_tokens": 743029816.0, + "step": 19470 + }, + { + "epoch": 2.4769113344358225, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7915213108062744, + "num_tokens": 743068565.0, + "step": 19471 + }, + { + "epoch": 2.477038544714413, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8047060370445251, + "num_tokens": 743112137.0, + "step": 19472 + }, + { + "epoch": 2.4771657549930035, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8139054775238037, + "num_tokens": 743145067.0, + "step": 19473 + }, + { + "epoch": 2.477292965271594, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8035023808479309, + "num_tokens": 743177017.0, + "step": 19474 + }, + { + "epoch": 2.4774201755501846, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8031618595123291, + "num_tokens": 743218121.0, + "step": 19475 + }, + { + "epoch": 2.477547385828775, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8086303472518921, + "num_tokens": 743256635.0, + "step": 19476 + }, + { + "epoch": 2.4776745961073656, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8026119470596313, + "num_tokens": 743293801.0, + "step": 19477 + }, + { + "epoch": 2.477801806385956, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8137093782424927, + "num_tokens": 743331582.0, + "step": 19478 + }, + { + "epoch": 2.4779290166645467, + "grad_norm": 10.4375, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8035221099853516, + "num_tokens": 743369901.0, + "step": 19479 + }, + { + "epoch": 2.4780562269431368, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7957479953765869, + "num_tokens": 743405808.0, + "step": 19480 + }, + { + "epoch": 2.4781834372217277, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8022710680961609, + "num_tokens": 743447236.0, + "step": 19481 + }, + { + "epoch": 2.478310647500318, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8069319725036621, + "num_tokens": 743486494.0, + "step": 19482 + }, + { + "epoch": 2.4784378577789083, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8016160726547241, + "num_tokens": 743523941.0, + "step": 19483 + }, + { + "epoch": 2.478565068057499, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8008863925933838, + "num_tokens": 743554169.0, + "step": 19484 + }, + { + "epoch": 2.4786922783360894, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7850203514099121, + "num_tokens": 743594656.0, + "step": 19485 + }, + { + "epoch": 2.47881948861468, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8099538683891296, + "num_tokens": 743628638.0, + "step": 19486 + }, + { + "epoch": 2.4789466988932705, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7896491885185242, + "num_tokens": 743665382.0, + "step": 19487 + }, + { + "epoch": 2.479073909171861, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8131197094917297, + "num_tokens": 743701712.0, + "step": 19488 + }, + { + "epoch": 2.4792011194504515, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.803179144859314, + "num_tokens": 743740347.0, + "step": 19489 + }, + { + "epoch": 2.479328329729042, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8030775785446167, + "num_tokens": 743775434.0, + "step": 19490 + }, + { + "epoch": 2.4794555400076326, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7926584482192993, + "num_tokens": 743812475.0, + "step": 19491 + }, + { + "epoch": 2.479582750286223, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.792534589767456, + "num_tokens": 743851350.0, + "step": 19492 + }, + { + "epoch": 2.4797099605648136, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7942670583724976, + "num_tokens": 743885086.0, + "step": 19493 + }, + { + "epoch": 2.479837170843404, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8085945844650269, + "num_tokens": 743931747.0, + "step": 19494 + }, + { + "epoch": 2.4799643811219947, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8221762180328369, + "num_tokens": 743971806.0, + "step": 19495 + }, + { + "epoch": 2.480091591400585, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8089147806167603, + "num_tokens": 744005304.0, + "step": 19496 + }, + { + "epoch": 2.4802188016791757, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7908666729927063, + "num_tokens": 744051159.0, + "step": 19497 + }, + { + "epoch": 2.4803460119577663, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7826143503189087, + "num_tokens": 744089856.0, + "step": 19498 + }, + { + "epoch": 2.480473222236357, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7794688940048218, + "num_tokens": 744136355.0, + "step": 19499 + }, + { + "epoch": 2.4806004325149473, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.789760947227478, + "num_tokens": 744177682.0, + "step": 19500 + }, + { + "epoch": 2.480727642793538, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8050662279129028, + "num_tokens": 744214209.0, + "step": 19501 + }, + { + "epoch": 2.4808548530721284, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7827528715133667, + "num_tokens": 744257923.0, + "step": 19502 + }, + { + "epoch": 2.480982063350719, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.810287594795227, + "num_tokens": 744293370.0, + "step": 19503 + }, + { + "epoch": 2.4811092736293094, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8119416236877441, + "num_tokens": 744328572.0, + "step": 19504 + }, + { + "epoch": 2.4812364839078995, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7803828716278076, + "num_tokens": 744364124.0, + "step": 19505 + }, + { + "epoch": 2.4813636941864905, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8053982257843018, + "num_tokens": 744403679.0, + "step": 19506 + }, + { + "epoch": 2.4814909044650806, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8046404719352722, + "num_tokens": 744440922.0, + "step": 19507 + }, + { + "epoch": 2.481618114743671, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8005518317222595, + "num_tokens": 744479025.0, + "step": 19508 + }, + { + "epoch": 2.4817453250222616, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.801511287689209, + "num_tokens": 744514165.0, + "step": 19509 + }, + { + "epoch": 2.481872535300852, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.795712947845459, + "num_tokens": 744547210.0, + "step": 19510 + }, + { + "epoch": 2.4819997455794427, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8019082546234131, + "num_tokens": 744585132.0, + "step": 19511 + }, + { + "epoch": 2.4821269558580332, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8126721978187561, + "num_tokens": 744617921.0, + "step": 19512 + }, + { + "epoch": 2.4822541661366238, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7951434254646301, + "num_tokens": 744653095.0, + "step": 19513 + }, + { + "epoch": 2.4823813764152143, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8003792762756348, + "num_tokens": 744688258.0, + "step": 19514 + }, + { + "epoch": 2.482508586693805, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7949424386024475, + "num_tokens": 744723500.0, + "step": 19515 + }, + { + "epoch": 2.4826357969723953, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.8002126216888428, + "num_tokens": 744757133.0, + "step": 19516 + }, + { + "epoch": 2.482763007250986, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8131371736526489, + "num_tokens": 744793814.0, + "step": 19517 + }, + { + "epoch": 2.4828902175295764, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8046563267707825, + "num_tokens": 744828680.0, + "step": 19518 + }, + { + "epoch": 2.483017427808167, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7904343008995056, + "num_tokens": 744861642.0, + "step": 19519 + }, + { + "epoch": 2.4831446380867574, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.8146131634712219, + "num_tokens": 744900345.0, + "step": 19520 + }, + { + "epoch": 2.483271848365348, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8098613023757935, + "num_tokens": 744938589.0, + "step": 19521 + }, + { + "epoch": 2.4833990586439385, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7877819538116455, + "num_tokens": 744975385.0, + "step": 19522 + }, + { + "epoch": 2.483526268922529, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.791534423828125, + "num_tokens": 745010860.0, + "step": 19523 + }, + { + "epoch": 2.4836534792011196, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7921663522720337, + "num_tokens": 745047183.0, + "step": 19524 + }, + { + "epoch": 2.48378068947971, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7919067144393921, + "num_tokens": 745081906.0, + "step": 19525 + }, + { + "epoch": 2.4839078997583006, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8059184551239014, + "num_tokens": 745127285.0, + "step": 19526 + }, + { + "epoch": 2.484035110036891, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.8206006288528442, + "num_tokens": 745171564.0, + "step": 19527 + }, + { + "epoch": 2.4841623203154817, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8160527348518372, + "num_tokens": 745214252.0, + "step": 19528 + }, + { + "epoch": 2.484289530594072, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.802661657333374, + "num_tokens": 745251983.0, + "step": 19529 + }, + { + "epoch": 2.4844167408726623, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7992090582847595, + "num_tokens": 745287037.0, + "step": 19530 + }, + { + "epoch": 2.4845439511512533, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8012304902076721, + "num_tokens": 745327854.0, + "step": 19531 + }, + { + "epoch": 2.4846711614298433, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7871004343032837, + "num_tokens": 745370227.0, + "step": 19532 + }, + { + "epoch": 2.484798371708434, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8003549575805664, + "num_tokens": 745409416.0, + "step": 19533 + }, + { + "epoch": 2.4849255819870244, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7834247350692749, + "num_tokens": 745449908.0, + "step": 19534 + }, + { + "epoch": 2.485052792265615, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7920083403587341, + "num_tokens": 745486384.0, + "step": 19535 + }, + { + "epoch": 2.4851800025442055, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7900160551071167, + "num_tokens": 745523058.0, + "step": 19536 + }, + { + "epoch": 2.485307212822796, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7808020114898682, + "num_tokens": 745559276.0, + "step": 19537 + }, + { + "epoch": 2.4854344231013865, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.798425555229187, + "num_tokens": 745592726.0, + "step": 19538 + }, + { + "epoch": 2.485561633379977, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8027496337890625, + "num_tokens": 745630117.0, + "step": 19539 + }, + { + "epoch": 2.4856888436585676, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7890900373458862, + "num_tokens": 745664968.0, + "step": 19540 + }, + { + "epoch": 2.485816053937158, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8014757037162781, + "num_tokens": 745701893.0, + "step": 19541 + }, + { + "epoch": 2.4859432642157486, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8009733557701111, + "num_tokens": 745736884.0, + "step": 19542 + }, + { + "epoch": 2.486070474494339, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.814882755279541, + "num_tokens": 745782381.0, + "step": 19543 + }, + { + "epoch": 2.4861976847729297, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.5893, + "mean_token_accuracy": 0.8229390382766724, + "num_tokens": 745827630.0, + "step": 19544 + }, + { + "epoch": 2.48632489505152, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8053227663040161, + "num_tokens": 745862223.0, + "step": 19545 + }, + { + "epoch": 2.4864521053301107, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8188698887825012, + "num_tokens": 745903077.0, + "step": 19546 + }, + { + "epoch": 2.4865793156087013, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8016758561134338, + "num_tokens": 745945534.0, + "step": 19547 + }, + { + "epoch": 2.486706525887292, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7848368287086487, + "num_tokens": 745986308.0, + "step": 19548 + }, + { + "epoch": 2.4868337361658823, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.789776623249054, + "num_tokens": 746022206.0, + "step": 19549 + }, + { + "epoch": 2.486960946444473, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8090834617614746, + "num_tokens": 746059854.0, + "step": 19550 + }, + { + "epoch": 2.4870881567230634, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8074723482131958, + "num_tokens": 746103994.0, + "step": 19551 + }, + { + "epoch": 2.487215367001654, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7886139750480652, + "num_tokens": 746138491.0, + "step": 19552 + }, + { + "epoch": 2.4873425772802444, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7966547012329102, + "num_tokens": 746172598.0, + "step": 19553 + }, + { + "epoch": 2.487469787558835, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.8114255666732788, + "num_tokens": 746210385.0, + "step": 19554 + }, + { + "epoch": 2.487596997837425, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8058891296386719, + "num_tokens": 746252563.0, + "step": 19555 + }, + { + "epoch": 2.487724208116016, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8091420531272888, + "num_tokens": 746292356.0, + "step": 19556 + }, + { + "epoch": 2.487851418394606, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7891888618469238, + "num_tokens": 746325467.0, + "step": 19557 + }, + { + "epoch": 2.4879786286731966, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7943540215492249, + "num_tokens": 746362285.0, + "step": 19558 + }, + { + "epoch": 2.488105838951787, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8135450482368469, + "num_tokens": 746401540.0, + "step": 19559 + }, + { + "epoch": 2.4882330492303777, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8078683614730835, + "num_tokens": 746439282.0, + "step": 19560 + }, + { + "epoch": 2.488360259508968, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8168847560882568, + "num_tokens": 746480993.0, + "step": 19561 + }, + { + "epoch": 2.4884874697875587, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7930225133895874, + "num_tokens": 746519318.0, + "step": 19562 + }, + { + "epoch": 2.4886146800661493, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7869573831558228, + "num_tokens": 746559761.0, + "step": 19563 + }, + { + "epoch": 2.48874189034474, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8052306771278381, + "num_tokens": 746591622.0, + "step": 19564 + }, + { + "epoch": 2.4888691006233303, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7953511476516724, + "num_tokens": 746627709.0, + "step": 19565 + }, + { + "epoch": 2.488996310901921, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7970333099365234, + "num_tokens": 746663445.0, + "step": 19566 + }, + { + "epoch": 2.4891235211805114, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7884529829025269, + "num_tokens": 746706869.0, + "step": 19567 + }, + { + "epoch": 2.489250731459102, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8146586418151855, + "num_tokens": 746749502.0, + "step": 19568 + }, + { + "epoch": 2.4893779417376924, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8089726567268372, + "num_tokens": 746786510.0, + "step": 19569 + }, + { + "epoch": 2.489505152016283, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7947736382484436, + "num_tokens": 746819430.0, + "step": 19570 + }, + { + "epoch": 2.4896323622948735, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8065845966339111, + "num_tokens": 746856018.0, + "step": 19571 + }, + { + "epoch": 2.489759572573464, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8002675771713257, + "num_tokens": 746897350.0, + "step": 19572 + }, + { + "epoch": 2.4898867828520546, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.790015697479248, + "num_tokens": 746937163.0, + "step": 19573 + }, + { + "epoch": 2.490013993130645, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.8103086948394775, + "num_tokens": 746976140.0, + "step": 19574 + }, + { + "epoch": 2.4901412034092356, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.800561249256134, + "num_tokens": 747012576.0, + "step": 19575 + }, + { + "epoch": 2.490268413687826, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7971634864807129, + "num_tokens": 747048179.0, + "step": 19576 + }, + { + "epoch": 2.4903956239664167, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7811065912246704, + "num_tokens": 747085291.0, + "step": 19577 + }, + { + "epoch": 2.4905228342450068, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8043873310089111, + "num_tokens": 747119085.0, + "step": 19578 + }, + { + "epoch": 2.4906500445235977, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8019648790359497, + "num_tokens": 747153463.0, + "step": 19579 + }, + { + "epoch": 2.490777254802188, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7724204063415527, + "num_tokens": 747194781.0, + "step": 19580 + }, + { + "epoch": 2.4909044650807783, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8038333654403687, + "num_tokens": 747235328.0, + "step": 19581 + }, + { + "epoch": 2.491031675359369, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7997788190841675, + "num_tokens": 747278481.0, + "step": 19582 + }, + { + "epoch": 2.4911588856379594, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8167464733123779, + "num_tokens": 747313077.0, + "step": 19583 + }, + { + "epoch": 2.49128609591655, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8108581304550171, + "num_tokens": 747352693.0, + "step": 19584 + }, + { + "epoch": 2.4914133061951405, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7882201671600342, + "num_tokens": 747397717.0, + "step": 19585 + }, + { + "epoch": 2.491540516473731, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7987776398658752, + "num_tokens": 747435457.0, + "step": 19586 + }, + { + "epoch": 2.4916677267523215, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8071171045303345, + "num_tokens": 747479017.0, + "step": 19587 + }, + { + "epoch": 2.491794937030912, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8017922639846802, + "num_tokens": 747520247.0, + "step": 19588 + }, + { + "epoch": 2.4919221473095026, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7856254577636719, + "num_tokens": 747559983.0, + "step": 19589 + }, + { + "epoch": 2.492049357588093, + "grad_norm": 59.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7966780662536621, + "num_tokens": 747595746.0, + "step": 19590 + }, + { + "epoch": 2.4921765678666836, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.79864501953125, + "num_tokens": 747639261.0, + "step": 19591 + }, + { + "epoch": 2.492303778145274, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7879600524902344, + "num_tokens": 747680665.0, + "step": 19592 + }, + { + "epoch": 2.4924309884238647, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.804037868976593, + "num_tokens": 747713320.0, + "step": 19593 + }, + { + "epoch": 2.492558198702455, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7776095867156982, + "num_tokens": 747752167.0, + "step": 19594 + }, + { + "epoch": 2.4926854089810457, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7971327900886536, + "num_tokens": 747787274.0, + "step": 19595 + }, + { + "epoch": 2.4928126192596363, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7956691980361938, + "num_tokens": 747826647.0, + "step": 19596 + }, + { + "epoch": 2.492939829538227, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7953860759735107, + "num_tokens": 747866797.0, + "step": 19597 + }, + { + "epoch": 2.4930670398168173, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8022764325141907, + "num_tokens": 747907589.0, + "step": 19598 + }, + { + "epoch": 2.493194250095408, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7865216732025146, + "num_tokens": 747939865.0, + "step": 19599 + }, + { + "epoch": 2.4933214603739984, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7961642742156982, + "num_tokens": 747976238.0, + "step": 19600 + }, + { + "epoch": 2.493448670652589, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.798907995223999, + "num_tokens": 748012076.0, + "step": 19601 + }, + { + "epoch": 2.4935758809311794, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7848835587501526, + "num_tokens": 748042518.0, + "step": 19602 + }, + { + "epoch": 2.4937030912097695, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.814098596572876, + "num_tokens": 748077509.0, + "step": 19603 + }, + { + "epoch": 2.4938303014883605, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8112940192222595, + "num_tokens": 748118164.0, + "step": 19604 + }, + { + "epoch": 2.4939575117669506, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7921282649040222, + "num_tokens": 748153609.0, + "step": 19605 + }, + { + "epoch": 2.494084722045541, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8005796670913696, + "num_tokens": 748189573.0, + "step": 19606 + }, + { + "epoch": 2.4942119323241316, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7962470054626465, + "num_tokens": 748231121.0, + "step": 19607 + }, + { + "epoch": 2.494339142602722, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8117949366569519, + "num_tokens": 748268744.0, + "step": 19608 + }, + { + "epoch": 2.4944663528813127, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8047646880149841, + "num_tokens": 748307441.0, + "step": 19609 + }, + { + "epoch": 2.494593563159903, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.801932692527771, + "num_tokens": 748343506.0, + "step": 19610 + }, + { + "epoch": 2.4947207734384937, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8025500774383545, + "num_tokens": 748382209.0, + "step": 19611 + }, + { + "epoch": 2.4948479837170843, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.5804, + "mean_token_accuracy": 0.8235056400299072, + "num_tokens": 748413660.0, + "step": 19612 + }, + { + "epoch": 2.494975193995675, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7979204654693604, + "num_tokens": 748457037.0, + "step": 19613 + }, + { + "epoch": 2.4951024042742653, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.773374617099762, + "num_tokens": 748497227.0, + "step": 19614 + }, + { + "epoch": 2.495229614552856, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7907933592796326, + "num_tokens": 748529901.0, + "step": 19615 + }, + { + "epoch": 2.4953568248314464, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7843947410583496, + "num_tokens": 748568238.0, + "step": 19616 + }, + { + "epoch": 2.495484035110037, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7867580652236938, + "num_tokens": 748608153.0, + "step": 19617 + }, + { + "epoch": 2.4956112453886274, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7931693196296692, + "num_tokens": 748646977.0, + "step": 19618 + }, + { + "epoch": 2.495738455667218, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7891749739646912, + "num_tokens": 748689586.0, + "step": 19619 + }, + { + "epoch": 2.4958656659458085, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.799651026725769, + "num_tokens": 748730915.0, + "step": 19620 + }, + { + "epoch": 2.495992876224399, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7926524877548218, + "num_tokens": 748766996.0, + "step": 19621 + }, + { + "epoch": 2.4961200865029896, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7912452220916748, + "num_tokens": 748801225.0, + "step": 19622 + }, + { + "epoch": 2.49624729678158, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7922655344009399, + "num_tokens": 748841355.0, + "step": 19623 + }, + { + "epoch": 2.4963745070601706, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.799875020980835, + "num_tokens": 748879636.0, + "step": 19624 + }, + { + "epoch": 2.496501717338761, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7948986291885376, + "num_tokens": 748917795.0, + "step": 19625 + }, + { + "epoch": 2.4966289276173517, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7720170617103577, + "num_tokens": 748956512.0, + "step": 19626 + }, + { + "epoch": 2.496756137895942, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8068318367004395, + "num_tokens": 748995577.0, + "step": 19627 + }, + { + "epoch": 2.4968833481745323, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7930716276168823, + "num_tokens": 749027672.0, + "step": 19628 + }, + { + "epoch": 2.4970105584531233, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8038884997367859, + "num_tokens": 749059101.0, + "step": 19629 + }, + { + "epoch": 2.4971377687317133, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7817500233650208, + "num_tokens": 749097034.0, + "step": 19630 + }, + { + "epoch": 2.497264979010304, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7970845699310303, + "num_tokens": 749137536.0, + "step": 19631 + }, + { + "epoch": 2.4973921892888944, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7903146743774414, + "num_tokens": 749171487.0, + "step": 19632 + }, + { + "epoch": 2.497519399567485, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7925236225128174, + "num_tokens": 749213234.0, + "step": 19633 + }, + { + "epoch": 2.4976466098460754, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7848222255706787, + "num_tokens": 749256184.0, + "step": 19634 + }, + { + "epoch": 2.497773820124666, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7946805953979492, + "num_tokens": 749292909.0, + "step": 19635 + }, + { + "epoch": 2.4979010304032565, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7837657332420349, + "num_tokens": 749327947.0, + "step": 19636 + }, + { + "epoch": 2.498028240681847, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8077954053878784, + "num_tokens": 749364107.0, + "step": 19637 + }, + { + "epoch": 2.4981554509604376, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8146709203720093, + "num_tokens": 749398055.0, + "step": 19638 + }, + { + "epoch": 2.498282661239028, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7983607053756714, + "num_tokens": 749434626.0, + "step": 19639 + }, + { + "epoch": 2.4984098715176186, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7945660352706909, + "num_tokens": 749470616.0, + "step": 19640 + }, + { + "epoch": 2.498537081796209, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8017056584358215, + "num_tokens": 749513072.0, + "step": 19641 + }, + { + "epoch": 2.4986642920747997, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8091829419136047, + "num_tokens": 749546988.0, + "step": 19642 + }, + { + "epoch": 2.49879150235339, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7938077449798584, + "num_tokens": 749585851.0, + "step": 19643 + }, + { + "epoch": 2.4989187126319807, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7917508482933044, + "num_tokens": 749619510.0, + "step": 19644 + }, + { + "epoch": 2.4990459229105713, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8055175542831421, + "num_tokens": 749653660.0, + "step": 19645 + }, + { + "epoch": 2.499173133189162, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7971323728561401, + "num_tokens": 749691321.0, + "step": 19646 + }, + { + "epoch": 2.4993003434677523, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7998881340026855, + "num_tokens": 749729305.0, + "step": 19647 + }, + { + "epoch": 2.499427553746343, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8027939200401306, + "num_tokens": 749766134.0, + "step": 19648 + }, + { + "epoch": 2.4995547640249334, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8022869229316711, + "num_tokens": 749807715.0, + "step": 19649 + }, + { + "epoch": 2.499681974303524, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7917720079421997, + "num_tokens": 749838799.0, + "step": 19650 + }, + { + "epoch": 2.499809184582114, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8147159814834595, + "num_tokens": 749874283.0, + "step": 19651 + }, + { + "epoch": 2.499936394860705, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7957758903503418, + "num_tokens": 749915912.0, + "step": 19652 + }, + { + "epoch": 2.500063605139295, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8090363144874573, + "num_tokens": 749959133.0, + "step": 19653 + }, + { + "epoch": 2.500190815417886, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7928094267845154, + "num_tokens": 749997846.0, + "step": 19654 + }, + { + "epoch": 2.500318025696476, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8077290058135986, + "num_tokens": 750040973.0, + "step": 19655 + }, + { + "epoch": 2.5004452359750666, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7959535121917725, + "num_tokens": 750082171.0, + "step": 19656 + }, + { + "epoch": 2.500572446253657, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8148269653320312, + "num_tokens": 750123083.0, + "step": 19657 + }, + { + "epoch": 2.5006996565322477, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7871582508087158, + "num_tokens": 750162235.0, + "step": 19658 + }, + { + "epoch": 2.500826866810838, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8114392757415771, + "num_tokens": 750196501.0, + "step": 19659 + }, + { + "epoch": 2.5009540770894287, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7747055292129517, + "num_tokens": 750241178.0, + "step": 19660 + }, + { + "epoch": 2.5010812873680193, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8055906295776367, + "num_tokens": 750284104.0, + "step": 19661 + }, + { + "epoch": 2.50120849764661, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8061180710792542, + "num_tokens": 750318525.0, + "step": 19662 + }, + { + "epoch": 2.5013357079252003, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8102951049804688, + "num_tokens": 750358892.0, + "step": 19663 + }, + { + "epoch": 2.501462918203791, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7895326614379883, + "num_tokens": 750399325.0, + "step": 19664 + }, + { + "epoch": 2.5015901284823814, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7950776815414429, + "num_tokens": 750437493.0, + "step": 19665 + }, + { + "epoch": 2.501717338760972, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8071721792221069, + "num_tokens": 750472381.0, + "step": 19666 + }, + { + "epoch": 2.5018445490395624, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7770702838897705, + "num_tokens": 750511786.0, + "step": 19667 + }, + { + "epoch": 2.501971759318153, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.802912175655365, + "num_tokens": 750550596.0, + "step": 19668 + }, + { + "epoch": 2.5020989695967435, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8066922426223755, + "num_tokens": 750594790.0, + "step": 19669 + }, + { + "epoch": 2.502226179875334, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8111093044281006, + "num_tokens": 750636738.0, + "step": 19670 + }, + { + "epoch": 2.5023533901539246, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8123951554298401, + "num_tokens": 750672338.0, + "step": 19671 + }, + { + "epoch": 2.502480600432515, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7892399430274963, + "num_tokens": 750709149.0, + "step": 19672 + }, + { + "epoch": 2.5026078107111056, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8051196336746216, + "num_tokens": 750741727.0, + "step": 19673 + }, + { + "epoch": 2.5027350209896957, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7824684381484985, + "num_tokens": 750775903.0, + "step": 19674 + }, + { + "epoch": 2.5028622312682867, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7938514947891235, + "num_tokens": 750810502.0, + "step": 19675 + }, + { + "epoch": 2.5029894415468767, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8065351843833923, + "num_tokens": 750847762.0, + "step": 19676 + }, + { + "epoch": 2.5031166518254677, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.8048429489135742, + "num_tokens": 750887201.0, + "step": 19677 + }, + { + "epoch": 2.503243862104058, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.8004796504974365, + "num_tokens": 750927323.0, + "step": 19678 + }, + { + "epoch": 2.5033710723826488, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7956563234329224, + "num_tokens": 750972501.0, + "step": 19679 + }, + { + "epoch": 2.503498282661239, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8122814893722534, + "num_tokens": 751010743.0, + "step": 19680 + }, + { + "epoch": 2.5036254929398294, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7957311868667603, + "num_tokens": 751050440.0, + "step": 19681 + }, + { + "epoch": 2.50375270321842, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8038364052772522, + "num_tokens": 751088319.0, + "step": 19682 + }, + { + "epoch": 2.5038799134970104, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7960147261619568, + "num_tokens": 751123766.0, + "step": 19683 + }, + { + "epoch": 2.504007123775601, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8178204298019409, + "num_tokens": 751163375.0, + "step": 19684 + }, + { + "epoch": 2.5041343340541915, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7965994477272034, + "num_tokens": 751204526.0, + "step": 19685 + }, + { + "epoch": 2.504261544332782, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.8085861206054688, + "num_tokens": 751245133.0, + "step": 19686 + }, + { + "epoch": 2.5043887546113726, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7909919619560242, + "num_tokens": 751284578.0, + "step": 19687 + }, + { + "epoch": 2.504515964889963, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7948814630508423, + "num_tokens": 751325794.0, + "step": 19688 + }, + { + "epoch": 2.5046431751685536, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.801042914390564, + "num_tokens": 751363777.0, + "step": 19689 + }, + { + "epoch": 2.504770385447144, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7858395576477051, + "num_tokens": 751400499.0, + "step": 19690 + }, + { + "epoch": 2.5048975957257347, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7898941040039062, + "num_tokens": 751434895.0, + "step": 19691 + }, + { + "epoch": 2.505024806004325, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8133189678192139, + "num_tokens": 751473984.0, + "step": 19692 + }, + { + "epoch": 2.5051520162829157, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7953391075134277, + "num_tokens": 751515030.0, + "step": 19693 + }, + { + "epoch": 2.5052792265615063, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7970626354217529, + "num_tokens": 751550023.0, + "step": 19694 + }, + { + "epoch": 2.505406436840097, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7985652685165405, + "num_tokens": 751584793.0, + "step": 19695 + }, + { + "epoch": 2.5055336471186873, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7775360345840454, + "num_tokens": 751619817.0, + "step": 19696 + }, + { + "epoch": 2.505660857397278, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8060142993927002, + "num_tokens": 751657097.0, + "step": 19697 + }, + { + "epoch": 2.5057880676758684, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7845004796981812, + "num_tokens": 751687042.0, + "step": 19698 + }, + { + "epoch": 2.5059152779544585, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8104104995727539, + "num_tokens": 751724281.0, + "step": 19699 + }, + { + "epoch": 2.5060424882330494, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7750203609466553, + "num_tokens": 751765144.0, + "step": 19700 + }, + { + "epoch": 2.5061696985116395, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8056998252868652, + "num_tokens": 751799602.0, + "step": 19701 + }, + { + "epoch": 2.5062969087902305, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8127025961875916, + "num_tokens": 751836972.0, + "step": 19702 + }, + { + "epoch": 2.5064241190688206, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7944247722625732, + "num_tokens": 751869627.0, + "step": 19703 + }, + { + "epoch": 2.5065513293474115, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7945125102996826, + "num_tokens": 751911103.0, + "step": 19704 + }, + { + "epoch": 2.5066785396260016, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7903633117675781, + "num_tokens": 751948956.0, + "step": 19705 + }, + { + "epoch": 2.506805749904592, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7999494075775146, + "num_tokens": 751992185.0, + "step": 19706 + }, + { + "epoch": 2.5069329601831827, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7936060428619385, + "num_tokens": 752032661.0, + "step": 19707 + }, + { + "epoch": 2.507060170461773, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8061395287513733, + "num_tokens": 752073946.0, + "step": 19708 + }, + { + "epoch": 2.5071873807403637, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7865245938301086, + "num_tokens": 752118766.0, + "step": 19709 + }, + { + "epoch": 2.5073145910189543, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8017281293869019, + "num_tokens": 752159669.0, + "step": 19710 + }, + { + "epoch": 2.507441801297545, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.8011850118637085, + "num_tokens": 752198139.0, + "step": 19711 + }, + { + "epoch": 2.5075690115761353, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8049414157867432, + "num_tokens": 752239686.0, + "step": 19712 + }, + { + "epoch": 2.507696221854726, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7809274196624756, + "num_tokens": 752275740.0, + "step": 19713 + }, + { + "epoch": 2.5078234321333164, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7837882041931152, + "num_tokens": 752310554.0, + "step": 19714 + }, + { + "epoch": 2.507950642411907, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8061918020248413, + "num_tokens": 752347761.0, + "step": 19715 + }, + { + "epoch": 2.5080778526904974, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7849242687225342, + "num_tokens": 752387128.0, + "step": 19716 + }, + { + "epoch": 2.508205062969088, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7969820499420166, + "num_tokens": 752428722.0, + "step": 19717 + }, + { + "epoch": 2.5083322732476785, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7944294214248657, + "num_tokens": 752468920.0, + "step": 19718 + }, + { + "epoch": 2.508459483526269, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7885620594024658, + "num_tokens": 752505226.0, + "step": 19719 + }, + { + "epoch": 2.5085866938048595, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.791218638420105, + "num_tokens": 752543693.0, + "step": 19720 + }, + { + "epoch": 2.50871390408345, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.783980131149292, + "num_tokens": 752578758.0, + "step": 19721 + }, + { + "epoch": 2.5088411143620406, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8104937076568604, + "num_tokens": 752612115.0, + "step": 19722 + }, + { + "epoch": 2.508968324640631, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8011496067047119, + "num_tokens": 752644601.0, + "step": 19723 + }, + { + "epoch": 2.509095534919221, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8103072643280029, + "num_tokens": 752682333.0, + "step": 19724 + }, + { + "epoch": 2.509222745197812, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7973640561103821, + "num_tokens": 752721802.0, + "step": 19725 + }, + { + "epoch": 2.5093499554764023, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.8123077154159546, + "num_tokens": 752755673.0, + "step": 19726 + }, + { + "epoch": 2.5094771657549932, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8097602128982544, + "num_tokens": 752795575.0, + "step": 19727 + }, + { + "epoch": 2.5096043760335833, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8196775913238525, + "num_tokens": 752833172.0, + "step": 19728 + }, + { + "epoch": 2.5097315863121743, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.786466121673584, + "num_tokens": 752872857.0, + "step": 19729 + }, + { + "epoch": 2.5098587965907644, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8095545768737793, + "num_tokens": 752908553.0, + "step": 19730 + }, + { + "epoch": 2.509986006869355, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8006331920623779, + "num_tokens": 752937762.0, + "step": 19731 + }, + { + "epoch": 2.5101132171479454, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7839878797531128, + "num_tokens": 752977184.0, + "step": 19732 + }, + { + "epoch": 2.510240427426536, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7983461618423462, + "num_tokens": 753021071.0, + "step": 19733 + }, + { + "epoch": 2.5103676377051265, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.770416259765625, + "num_tokens": 753056597.0, + "step": 19734 + }, + { + "epoch": 2.510494847983717, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8023339509963989, + "num_tokens": 753096178.0, + "step": 19735 + }, + { + "epoch": 2.5106220582623076, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8012646436691284, + "num_tokens": 753130388.0, + "step": 19736 + }, + { + "epoch": 2.510749268540898, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8059991002082825, + "num_tokens": 753166585.0, + "step": 19737 + }, + { + "epoch": 2.5108764788194886, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8118137121200562, + "num_tokens": 753203403.0, + "step": 19738 + }, + { + "epoch": 2.511003689098079, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8046687841415405, + "num_tokens": 753241632.0, + "step": 19739 + }, + { + "epoch": 2.5111308993766697, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8124243021011353, + "num_tokens": 753279292.0, + "step": 19740 + }, + { + "epoch": 2.51125810965526, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7921196818351746, + "num_tokens": 753319121.0, + "step": 19741 + }, + { + "epoch": 2.5113853199338507, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7818396687507629, + "num_tokens": 753353074.0, + "step": 19742 + }, + { + "epoch": 2.5115125302124413, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8236720561981201, + "num_tokens": 753388994.0, + "step": 19743 + }, + { + "epoch": 2.511639740491032, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8087353706359863, + "num_tokens": 753430271.0, + "step": 19744 + }, + { + "epoch": 2.5117669507696223, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7908203601837158, + "num_tokens": 753473820.0, + "step": 19745 + }, + { + "epoch": 2.511894161048213, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7967420816421509, + "num_tokens": 753516397.0, + "step": 19746 + }, + { + "epoch": 2.5120213713268034, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8082940578460693, + "num_tokens": 753557589.0, + "step": 19747 + }, + { + "epoch": 2.512148581605394, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8064153790473938, + "num_tokens": 753593764.0, + "step": 19748 + }, + { + "epoch": 2.512275791883984, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7878321409225464, + "num_tokens": 753635967.0, + "step": 19749 + }, + { + "epoch": 2.512403002162575, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.805330753326416, + "num_tokens": 753674595.0, + "step": 19750 + }, + { + "epoch": 2.512530212441165, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7845219373703003, + "num_tokens": 753714118.0, + "step": 19751 + }, + { + "epoch": 2.512657422719756, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7863451242446899, + "num_tokens": 753758139.0, + "step": 19752 + }, + { + "epoch": 2.512784632998346, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7898101806640625, + "num_tokens": 753800373.0, + "step": 19753 + }, + { + "epoch": 2.5129118432769366, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.793290376663208, + "num_tokens": 753832366.0, + "step": 19754 + }, + { + "epoch": 2.513039053555527, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8216067552566528, + "num_tokens": 753866176.0, + "step": 19755 + }, + { + "epoch": 2.5131662638341177, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7990637421607971, + "num_tokens": 753911470.0, + "step": 19756 + }, + { + "epoch": 2.513293474112708, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8006437420845032, + "num_tokens": 753953027.0, + "step": 19757 + }, + { + "epoch": 2.5134206843912987, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7887064218521118, + "num_tokens": 753987992.0, + "step": 19758 + }, + { + "epoch": 2.5135478946698893, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8080000281333923, + "num_tokens": 754026671.0, + "step": 19759 + }, + { + "epoch": 2.51367510494848, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7974299192428589, + "num_tokens": 754063526.0, + "step": 19760 + }, + { + "epoch": 2.5138023152270703, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7847045660018921, + "num_tokens": 754103361.0, + "step": 19761 + }, + { + "epoch": 2.513929525505661, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8011362552642822, + "num_tokens": 754141969.0, + "step": 19762 + }, + { + "epoch": 2.5140567357842514, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8157811164855957, + "num_tokens": 754182691.0, + "step": 19763 + }, + { + "epoch": 2.514183946062842, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8003480434417725, + "num_tokens": 754219956.0, + "step": 19764 + }, + { + "epoch": 2.5143111563414324, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7920457720756531, + "num_tokens": 754259467.0, + "step": 19765 + }, + { + "epoch": 2.514438366620023, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8060949444770813, + "num_tokens": 754300344.0, + "step": 19766 + }, + { + "epoch": 2.5145655768986135, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8051358461380005, + "num_tokens": 754342606.0, + "step": 19767 + }, + { + "epoch": 2.514692787177204, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7957990765571594, + "num_tokens": 754379946.0, + "step": 19768 + }, + { + "epoch": 2.5148199974557945, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8161881566047668, + "num_tokens": 754414353.0, + "step": 19769 + }, + { + "epoch": 2.514947207734385, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.784864604473114, + "num_tokens": 754455506.0, + "step": 19770 + }, + { + "epoch": 2.5150744180129756, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7852195501327515, + "num_tokens": 754498316.0, + "step": 19771 + }, + { + "epoch": 2.5152016282915657, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7897897958755493, + "num_tokens": 754534943.0, + "step": 19772 + }, + { + "epoch": 2.5153288385701567, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7984617352485657, + "num_tokens": 754572775.0, + "step": 19773 + }, + { + "epoch": 2.5154560488487467, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.8117773532867432, + "num_tokens": 754610473.0, + "step": 19774 + }, + { + "epoch": 2.5155832591273377, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8027319312095642, + "num_tokens": 754649527.0, + "step": 19775 + }, + { + "epoch": 2.515710469405928, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8007327914237976, + "num_tokens": 754691513.0, + "step": 19776 + }, + { + "epoch": 2.5158376796845188, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7823178768157959, + "num_tokens": 754727932.0, + "step": 19777 + }, + { + "epoch": 2.515964889963109, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8055769801139832, + "num_tokens": 754764571.0, + "step": 19778 + }, + { + "epoch": 2.5160921002416994, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.5982, + "mean_token_accuracy": 0.8152841329574585, + "num_tokens": 754805846.0, + "step": 19779 + }, + { + "epoch": 2.51621931052029, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8101704120635986, + "num_tokens": 754844016.0, + "step": 19780 + }, + { + "epoch": 2.5163465207988804, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7974244356155396, + "num_tokens": 754886493.0, + "step": 19781 + }, + { + "epoch": 2.516473731077471, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8061832189559937, + "num_tokens": 754928195.0, + "step": 19782 + }, + { + "epoch": 2.5166009413560615, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8055229187011719, + "num_tokens": 754958399.0, + "step": 19783 + }, + { + "epoch": 2.516728151634652, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8057218790054321, + "num_tokens": 754997926.0, + "step": 19784 + }, + { + "epoch": 2.5168553619132426, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8106653690338135, + "num_tokens": 755040577.0, + "step": 19785 + }, + { + "epoch": 2.516982572191833, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7917490005493164, + "num_tokens": 755070458.0, + "step": 19786 + }, + { + "epoch": 2.5171097824704236, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7816433906555176, + "num_tokens": 755117205.0, + "step": 19787 + }, + { + "epoch": 2.517236992749014, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8032141923904419, + "num_tokens": 755162098.0, + "step": 19788 + }, + { + "epoch": 2.5173642030276047, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7945942878723145, + "num_tokens": 755199655.0, + "step": 19789 + }, + { + "epoch": 2.517491413306195, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.8202001452445984, + "num_tokens": 755234165.0, + "step": 19790 + }, + { + "epoch": 2.5176186235847857, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.800561249256134, + "num_tokens": 755271933.0, + "step": 19791 + }, + { + "epoch": 2.5177458338633762, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8102929592132568, + "num_tokens": 755306903.0, + "step": 19792 + }, + { + "epoch": 2.5178730441419668, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8140933513641357, + "num_tokens": 755347613.0, + "step": 19793 + }, + { + "epoch": 2.5180002544205573, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.8188481330871582, + "num_tokens": 755380541.0, + "step": 19794 + }, + { + "epoch": 2.518127464699148, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.797086775302887, + "num_tokens": 755422502.0, + "step": 19795 + }, + { + "epoch": 2.5182546749777384, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7957302927970886, + "num_tokens": 755463975.0, + "step": 19796 + }, + { + "epoch": 2.5183818852563284, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8016469478607178, + "num_tokens": 755504873.0, + "step": 19797 + }, + { + "epoch": 2.5185090955349194, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7728840708732605, + "num_tokens": 755545646.0, + "step": 19798 + }, + { + "epoch": 2.5186363058135095, + "grad_norm": 49.25, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7832357287406921, + "num_tokens": 755577228.0, + "step": 19799 + }, + { + "epoch": 2.5187635160921005, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8156018853187561, + "num_tokens": 755621437.0, + "step": 19800 + }, + { + "epoch": 2.5188907263706906, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7963402271270752, + "num_tokens": 755658266.0, + "step": 19801 + }, + { + "epoch": 2.5190179366492815, + "grad_norm": 57.25, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7851572036743164, + "num_tokens": 755696755.0, + "step": 19802 + }, + { + "epoch": 2.5191451469278716, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8136485815048218, + "num_tokens": 755732434.0, + "step": 19803 + }, + { + "epoch": 2.519272357206462, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8022400140762329, + "num_tokens": 755768488.0, + "step": 19804 + }, + { + "epoch": 2.5193995674850527, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.804867684841156, + "num_tokens": 755799637.0, + "step": 19805 + }, + { + "epoch": 2.519526777763643, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7810017466545105, + "num_tokens": 755834229.0, + "step": 19806 + }, + { + "epoch": 2.5196539880422337, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8056187629699707, + "num_tokens": 755871776.0, + "step": 19807 + }, + { + "epoch": 2.5197811983208243, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8084449768066406, + "num_tokens": 755911125.0, + "step": 19808 + }, + { + "epoch": 2.519908408599415, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8150496482849121, + "num_tokens": 755952204.0, + "step": 19809 + }, + { + "epoch": 2.5200356188780053, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.787406325340271, + "num_tokens": 755996702.0, + "step": 19810 + }, + { + "epoch": 2.520162829156596, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.5846, + "mean_token_accuracy": 0.8218687772750854, + "num_tokens": 756033285.0, + "step": 19811 + }, + { + "epoch": 2.5202900394351864, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.800804853439331, + "num_tokens": 756068851.0, + "step": 19812 + }, + { + "epoch": 2.520417249713777, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7931486368179321, + "num_tokens": 756111676.0, + "step": 19813 + }, + { + "epoch": 2.5205444599923674, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7767930030822754, + "num_tokens": 756146575.0, + "step": 19814 + }, + { + "epoch": 2.520671670270958, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7901736497879028, + "num_tokens": 756190006.0, + "step": 19815 + }, + { + "epoch": 2.5207988805495485, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.8135969638824463, + "num_tokens": 756232828.0, + "step": 19816 + }, + { + "epoch": 2.520926090828139, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.803382933139801, + "num_tokens": 756273077.0, + "step": 19817 + }, + { + "epoch": 2.5210533011067295, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7800155282020569, + "num_tokens": 756313421.0, + "step": 19818 + }, + { + "epoch": 2.52118051138532, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7947887182235718, + "num_tokens": 756356461.0, + "step": 19819 + }, + { + "epoch": 2.5213077216639106, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7894700765609741, + "num_tokens": 756388309.0, + "step": 19820 + }, + { + "epoch": 2.521434931942501, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.8133848905563354, + "num_tokens": 756426655.0, + "step": 19821 + }, + { + "epoch": 2.521562142221091, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7939358949661255, + "num_tokens": 756464961.0, + "step": 19822 + }, + { + "epoch": 2.521689352499682, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8142862319946289, + "num_tokens": 756497338.0, + "step": 19823 + }, + { + "epoch": 2.5218165627782723, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.80732262134552, + "num_tokens": 756535324.0, + "step": 19824 + }, + { + "epoch": 2.5219437730568632, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8085782527923584, + "num_tokens": 756574138.0, + "step": 19825 + }, + { + "epoch": 2.5220709833354533, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8096576929092407, + "num_tokens": 756608967.0, + "step": 19826 + }, + { + "epoch": 2.522198193614044, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7796095609664917, + "num_tokens": 756653882.0, + "step": 19827 + }, + { + "epoch": 2.5223254038926344, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7897922992706299, + "num_tokens": 756695638.0, + "step": 19828 + }, + { + "epoch": 2.522452614171225, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7862465381622314, + "num_tokens": 756734964.0, + "step": 19829 + }, + { + "epoch": 2.5225798244498154, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8026307821273804, + "num_tokens": 756768454.0, + "step": 19830 + }, + { + "epoch": 2.522707034728406, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7894028425216675, + "num_tokens": 756801696.0, + "step": 19831 + }, + { + "epoch": 2.5228342450069965, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7985583543777466, + "num_tokens": 756839874.0, + "step": 19832 + }, + { + "epoch": 2.522961455285587, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7924398183822632, + "num_tokens": 756884179.0, + "step": 19833 + }, + { + "epoch": 2.5230886655641775, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8037939667701721, + "num_tokens": 756922725.0, + "step": 19834 + }, + { + "epoch": 2.523215875842768, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7950428128242493, + "num_tokens": 756956080.0, + "step": 19835 + }, + { + "epoch": 2.5233430861213586, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7902262210845947, + "num_tokens": 756995137.0, + "step": 19836 + }, + { + "epoch": 2.523470296399949, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8091854453086853, + "num_tokens": 757035300.0, + "step": 19837 + }, + { + "epoch": 2.5235975066785397, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7697094082832336, + "num_tokens": 757068789.0, + "step": 19838 + }, + { + "epoch": 2.52372471695713, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.787065863609314, + "num_tokens": 757102913.0, + "step": 19839 + }, + { + "epoch": 2.5238519272357207, + "grad_norm": 12.1875, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8036462068557739, + "num_tokens": 757138932.0, + "step": 19840 + }, + { + "epoch": 2.5239791375143112, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8046941161155701, + "num_tokens": 757179181.0, + "step": 19841 + }, + { + "epoch": 2.5241063477929018, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7641825079917908, + "num_tokens": 757224209.0, + "step": 19842 + }, + { + "epoch": 2.5242335580714923, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8085720539093018, + "num_tokens": 757262398.0, + "step": 19843 + }, + { + "epoch": 2.524360768350083, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7890584468841553, + "num_tokens": 757301744.0, + "step": 19844 + }, + { + "epoch": 2.5244879786286734, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8008410930633545, + "num_tokens": 757338273.0, + "step": 19845 + }, + { + "epoch": 2.524615188907264, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7940402030944824, + "num_tokens": 757383169.0, + "step": 19846 + }, + { + "epoch": 2.524742399185854, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7915257215499878, + "num_tokens": 757420998.0, + "step": 19847 + }, + { + "epoch": 2.524869609464445, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8040335774421692, + "num_tokens": 757463281.0, + "step": 19848 + }, + { + "epoch": 2.524996819743035, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7926208972930908, + "num_tokens": 757495022.0, + "step": 19849 + }, + { + "epoch": 2.525124030021626, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8070358037948608, + "num_tokens": 757536801.0, + "step": 19850 + }, + { + "epoch": 2.525251240300216, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.8147287964820862, + "num_tokens": 757573419.0, + "step": 19851 + }, + { + "epoch": 2.5253784505788066, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.790802001953125, + "num_tokens": 757610773.0, + "step": 19852 + }, + { + "epoch": 2.525505660857397, + "grad_norm": 9.875, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.809054970741272, + "num_tokens": 757648067.0, + "step": 19853 + }, + { + "epoch": 2.5256328711359877, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7998900413513184, + "num_tokens": 757681598.0, + "step": 19854 + }, + { + "epoch": 2.525760081414578, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7784708738327026, + "num_tokens": 757724752.0, + "step": 19855 + }, + { + "epoch": 2.5258872916931687, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.81380295753479, + "num_tokens": 757763801.0, + "step": 19856 + }, + { + "epoch": 2.5260145019717593, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7982773780822754, + "num_tokens": 757799235.0, + "step": 19857 + }, + { + "epoch": 2.52614171225035, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8094230890274048, + "num_tokens": 757839695.0, + "step": 19858 + }, + { + "epoch": 2.5262689225289403, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7863397598266602, + "num_tokens": 757877152.0, + "step": 19859 + }, + { + "epoch": 2.526396132807531, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7980425953865051, + "num_tokens": 757916173.0, + "step": 19860 + }, + { + "epoch": 2.5265233430861214, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8083397746086121, + "num_tokens": 757957271.0, + "step": 19861 + }, + { + "epoch": 2.526650553364712, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7901928424835205, + "num_tokens": 757995553.0, + "step": 19862 + }, + { + "epoch": 2.5267777636433024, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7872895002365112, + "num_tokens": 758035095.0, + "step": 19863 + }, + { + "epoch": 2.526904973921893, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7988389730453491, + "num_tokens": 758071185.0, + "step": 19864 + }, + { + "epoch": 2.5270321842004835, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.799461305141449, + "num_tokens": 758110251.0, + "step": 19865 + }, + { + "epoch": 2.527159394479074, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7819138765335083, + "num_tokens": 758144385.0, + "step": 19866 + }, + { + "epoch": 2.5272866047576645, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.8126580715179443, + "num_tokens": 758185935.0, + "step": 19867 + }, + { + "epoch": 2.527413815036255, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8043677806854248, + "num_tokens": 758226688.0, + "step": 19868 + }, + { + "epoch": 2.5275410253148456, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8020208477973938, + "num_tokens": 758263866.0, + "step": 19869 + }, + { + "epoch": 2.5276682355934357, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8152168989181519, + "num_tokens": 758303607.0, + "step": 19870 + }, + { + "epoch": 2.5277954458720266, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8016902208328247, + "num_tokens": 758340120.0, + "step": 19871 + }, + { + "epoch": 2.5279226561506167, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7908692955970764, + "num_tokens": 758373792.0, + "step": 19872 + }, + { + "epoch": 2.5280498664292077, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7883399724960327, + "num_tokens": 758418157.0, + "step": 19873 + }, + { + "epoch": 2.528177076707798, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7928876876831055, + "num_tokens": 758454042.0, + "step": 19874 + }, + { + "epoch": 2.5283042869863888, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7998054623603821, + "num_tokens": 758495305.0, + "step": 19875 + }, + { + "epoch": 2.528431497264979, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8078314065933228, + "num_tokens": 758532646.0, + "step": 19876 + }, + { + "epoch": 2.5285587075435694, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.792238712310791, + "num_tokens": 758568329.0, + "step": 19877 + }, + { + "epoch": 2.52868591782216, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.809496283531189, + "num_tokens": 758608246.0, + "step": 19878 + }, + { + "epoch": 2.5288131281007504, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8111099004745483, + "num_tokens": 758645356.0, + "step": 19879 + }, + { + "epoch": 2.528940338379341, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8165809512138367, + "num_tokens": 758685657.0, + "step": 19880 + }, + { + "epoch": 2.5290675486579315, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.8032586574554443, + "num_tokens": 758718843.0, + "step": 19881 + }, + { + "epoch": 2.529194758936522, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8111836910247803, + "num_tokens": 758757946.0, + "step": 19882 + }, + { + "epoch": 2.5293219692151125, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7753921747207642, + "num_tokens": 758792832.0, + "step": 19883 + }, + { + "epoch": 2.529449179493703, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7891988754272461, + "num_tokens": 758826062.0, + "step": 19884 + }, + { + "epoch": 2.5295763897722936, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.80587238073349, + "num_tokens": 758860940.0, + "step": 19885 + }, + { + "epoch": 2.529703600050884, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.802018940448761, + "num_tokens": 758900740.0, + "step": 19886 + }, + { + "epoch": 2.5298308103294747, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7969452142715454, + "num_tokens": 758942786.0, + "step": 19887 + }, + { + "epoch": 2.529958020608065, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8077184557914734, + "num_tokens": 758983376.0, + "step": 19888 + }, + { + "epoch": 2.5300852308866557, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7881261110305786, + "num_tokens": 759024007.0, + "step": 19889 + }, + { + "epoch": 2.5302124411652462, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8083192110061646, + "num_tokens": 759061611.0, + "step": 19890 + }, + { + "epoch": 2.5303396514438368, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.783764660358429, + "num_tokens": 759101942.0, + "step": 19891 + }, + { + "epoch": 2.5304668617224273, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8182811737060547, + "num_tokens": 759140281.0, + "step": 19892 + }, + { + "epoch": 2.530594072001018, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7972232699394226, + "num_tokens": 759174319.0, + "step": 19893 + }, + { + "epoch": 2.5307212822796084, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.8018624186515808, + "num_tokens": 759208981.0, + "step": 19894 + }, + { + "epoch": 2.5308484925581984, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7978745698928833, + "num_tokens": 759240559.0, + "step": 19895 + }, + { + "epoch": 2.5309757028367894, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8115105628967285, + "num_tokens": 759280550.0, + "step": 19896 + }, + { + "epoch": 2.5311029131153795, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7924783229827881, + "num_tokens": 759321298.0, + "step": 19897 + }, + { + "epoch": 2.5312301233939705, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8121462464332581, + "num_tokens": 759354627.0, + "step": 19898 + }, + { + "epoch": 2.5313573336725606, + "grad_norm": 83.5, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7970843315124512, + "num_tokens": 759388129.0, + "step": 19899 + }, + { + "epoch": 2.5314845439511515, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8069483041763306, + "num_tokens": 759423667.0, + "step": 19900 + }, + { + "epoch": 2.5316117542297416, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7955858707427979, + "num_tokens": 759466586.0, + "step": 19901 + }, + { + "epoch": 2.531738964508332, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7914694547653198, + "num_tokens": 759504214.0, + "step": 19902 + }, + { + "epoch": 2.5318661747869227, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7856295704841614, + "num_tokens": 759543865.0, + "step": 19903 + }, + { + "epoch": 2.531993385065513, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.793763279914856, + "num_tokens": 759581917.0, + "step": 19904 + }, + { + "epoch": 2.5321205953441037, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8028964996337891, + "num_tokens": 759617641.0, + "step": 19905 + }, + { + "epoch": 2.5322478056226942, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8103136420249939, + "num_tokens": 759649097.0, + "step": 19906 + }, + { + "epoch": 2.5323750159012848, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7919778823852539, + "num_tokens": 759688802.0, + "step": 19907 + }, + { + "epoch": 2.5325022261798753, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8047457337379456, + "num_tokens": 759724243.0, + "step": 19908 + }, + { + "epoch": 2.532629436458466, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7942966818809509, + "num_tokens": 759764646.0, + "step": 19909 + }, + { + "epoch": 2.5327566467370564, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.804489016532898, + "num_tokens": 759806939.0, + "step": 19910 + }, + { + "epoch": 2.532883857015647, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8054677844047546, + "num_tokens": 759848138.0, + "step": 19911 + }, + { + "epoch": 2.5330110672942374, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7957234978675842, + "num_tokens": 759881877.0, + "step": 19912 + }, + { + "epoch": 2.533138277572828, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8121116161346436, + "num_tokens": 759922625.0, + "step": 19913 + }, + { + "epoch": 2.5332654878514185, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8119094371795654, + "num_tokens": 759952429.0, + "step": 19914 + }, + { + "epoch": 2.533392698130009, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8015140891075134, + "num_tokens": 759987678.0, + "step": 19915 + }, + { + "epoch": 2.5335199084085995, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7813883423805237, + "num_tokens": 760023943.0, + "step": 19916 + }, + { + "epoch": 2.53364711868719, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8066097497940063, + "num_tokens": 760068229.0, + "step": 19917 + }, + { + "epoch": 2.5337743289657806, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7872412204742432, + "num_tokens": 760109440.0, + "step": 19918 + }, + { + "epoch": 2.533901539244371, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7980678081512451, + "num_tokens": 760146616.0, + "step": 19919 + }, + { + "epoch": 2.534028749522961, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7762966156005859, + "num_tokens": 760188708.0, + "step": 19920 + }, + { + "epoch": 2.534155959801552, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8128548860549927, + "num_tokens": 760231355.0, + "step": 19921 + }, + { + "epoch": 2.5342831700801423, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7857015132904053, + "num_tokens": 760268316.0, + "step": 19922 + }, + { + "epoch": 2.5344103803587332, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7921802401542664, + "num_tokens": 760305420.0, + "step": 19923 + }, + { + "epoch": 2.5345375906373233, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.826378583908081, + "num_tokens": 760342195.0, + "step": 19924 + }, + { + "epoch": 2.534664800915914, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.815040111541748, + "num_tokens": 760375394.0, + "step": 19925 + }, + { + "epoch": 2.5347920111945044, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7846706509590149, + "num_tokens": 760411629.0, + "step": 19926 + }, + { + "epoch": 2.534919221473095, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8069689869880676, + "num_tokens": 760456212.0, + "step": 19927 + }, + { + "epoch": 2.5350464317516854, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7913455963134766, + "num_tokens": 760496385.0, + "step": 19928 + }, + { + "epoch": 2.535173642030276, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7919648885726929, + "num_tokens": 760530426.0, + "step": 19929 + }, + { + "epoch": 2.5353008523088665, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8124873638153076, + "num_tokens": 760568174.0, + "step": 19930 + }, + { + "epoch": 2.535428062587457, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8190411329269409, + "num_tokens": 760608097.0, + "step": 19931 + }, + { + "epoch": 2.5355552728660475, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8001816868782043, + "num_tokens": 760643018.0, + "step": 19932 + }, + { + "epoch": 2.535682483144638, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8057717084884644, + "num_tokens": 760676553.0, + "step": 19933 + }, + { + "epoch": 2.5358096934232286, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.8155041933059692, + "num_tokens": 760708850.0, + "step": 19934 + }, + { + "epoch": 2.535936903701819, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7938268184661865, + "num_tokens": 760746088.0, + "step": 19935 + }, + { + "epoch": 2.5360641139804097, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8047991991043091, + "num_tokens": 760785716.0, + "step": 19936 + }, + { + "epoch": 2.536191324259, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.807201623916626, + "num_tokens": 760821301.0, + "step": 19937 + }, + { + "epoch": 2.5363185345375907, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7922550439834595, + "num_tokens": 760861200.0, + "step": 19938 + }, + { + "epoch": 2.5364457448161812, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.795438289642334, + "num_tokens": 760896398.0, + "step": 19939 + }, + { + "epoch": 2.5365729550947718, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7780437469482422, + "num_tokens": 760937300.0, + "step": 19940 + }, + { + "epoch": 2.5367001653733623, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7905937433242798, + "num_tokens": 760973726.0, + "step": 19941 + }, + { + "epoch": 2.536827375651953, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7965112924575806, + "num_tokens": 761009048.0, + "step": 19942 + }, + { + "epoch": 2.5369545859305433, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8024014830589294, + "num_tokens": 761050836.0, + "step": 19943 + }, + { + "epoch": 2.537081796209134, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7914890050888062, + "num_tokens": 761087517.0, + "step": 19944 + }, + { + "epoch": 2.537209006487724, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8128700256347656, + "num_tokens": 761121401.0, + "step": 19945 + }, + { + "epoch": 2.537336216766315, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7838706970214844, + "num_tokens": 761154973.0, + "step": 19946 + }, + { + "epoch": 2.537463427044905, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8072586059570312, + "num_tokens": 761194211.0, + "step": 19947 + }, + { + "epoch": 2.537590637323496, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7940000295639038, + "num_tokens": 761233486.0, + "step": 19948 + }, + { + "epoch": 2.537717847602086, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.801959753036499, + "num_tokens": 761275846.0, + "step": 19949 + }, + { + "epoch": 2.5378450578806766, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7867994904518127, + "num_tokens": 761311210.0, + "step": 19950 + }, + { + "epoch": 2.537972268159267, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7970580458641052, + "num_tokens": 761349961.0, + "step": 19951 + }, + { + "epoch": 2.5380994784378577, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.807340145111084, + "num_tokens": 761386993.0, + "step": 19952 + }, + { + "epoch": 2.538226688716448, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7907483577728271, + "num_tokens": 761426910.0, + "step": 19953 + }, + { + "epoch": 2.5383538989950387, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8032747507095337, + "num_tokens": 761466633.0, + "step": 19954 + }, + { + "epoch": 2.5384811092736292, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7829177975654602, + "num_tokens": 761497670.0, + "step": 19955 + }, + { + "epoch": 2.5386083195522198, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7952730655670166, + "num_tokens": 761534626.0, + "step": 19956 + }, + { + "epoch": 2.5387355298308103, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8076974749565125, + "num_tokens": 761581125.0, + "step": 19957 + }, + { + "epoch": 2.538862740109401, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.8030828237533569, + "num_tokens": 761616533.0, + "step": 19958 + }, + { + "epoch": 2.5389899503879914, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7886624932289124, + "num_tokens": 761650615.0, + "step": 19959 + }, + { + "epoch": 2.539117160666582, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8097723126411438, + "num_tokens": 761687020.0, + "step": 19960 + }, + { + "epoch": 2.5392443709451724, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8121263980865479, + "num_tokens": 761721653.0, + "step": 19961 + }, + { + "epoch": 2.539371581223763, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8067789077758789, + "num_tokens": 761758377.0, + "step": 19962 + }, + { + "epoch": 2.5394987915023535, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7988539338111877, + "num_tokens": 761795263.0, + "step": 19963 + }, + { + "epoch": 2.539626001780944, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8101529479026794, + "num_tokens": 761834758.0, + "step": 19964 + }, + { + "epoch": 2.5397532120595345, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7955930233001709, + "num_tokens": 761880482.0, + "step": 19965 + }, + { + "epoch": 2.539880422338125, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8012776374816895, + "num_tokens": 761922494.0, + "step": 19966 + }, + { + "epoch": 2.5400076326167156, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8164390325546265, + "num_tokens": 761959462.0, + "step": 19967 + }, + { + "epoch": 2.5401348428953057, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7843437194824219, + "num_tokens": 761990316.0, + "step": 19968 + }, + { + "epoch": 2.5402620531738966, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.792248010635376, + "num_tokens": 762023671.0, + "step": 19969 + }, + { + "epoch": 2.5403892634524867, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8077014088630676, + "num_tokens": 762056719.0, + "step": 19970 + }, + { + "epoch": 2.5405164737310777, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8041448593139648, + "num_tokens": 762098802.0, + "step": 19971 + }, + { + "epoch": 2.540643684009668, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7979656457901001, + "num_tokens": 762134470.0, + "step": 19972 + }, + { + "epoch": 2.5407708942882588, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8011247515678406, + "num_tokens": 762177163.0, + "step": 19973 + }, + { + "epoch": 2.540898104566849, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.812107264995575, + "num_tokens": 762217323.0, + "step": 19974 + }, + { + "epoch": 2.5410253148454394, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7930601835250854, + "num_tokens": 762252144.0, + "step": 19975 + }, + { + "epoch": 2.54115252512403, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7937235832214355, + "num_tokens": 762282108.0, + "step": 19976 + }, + { + "epoch": 2.5412797354026204, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8070639371871948, + "num_tokens": 762318154.0, + "step": 19977 + }, + { + "epoch": 2.541406945681211, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7966865301132202, + "num_tokens": 762354704.0, + "step": 19978 + }, + { + "epoch": 2.5415341559598015, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8038344383239746, + "num_tokens": 762389943.0, + "step": 19979 + }, + { + "epoch": 2.541661366238392, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.794519305229187, + "num_tokens": 762421560.0, + "step": 19980 + }, + { + "epoch": 2.5417885765169825, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8056244850158691, + "num_tokens": 762458509.0, + "step": 19981 + }, + { + "epoch": 2.541915786795573, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.796912431716919, + "num_tokens": 762502382.0, + "step": 19982 + }, + { + "epoch": 2.5420429970741636, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7917249798774719, + "num_tokens": 762536611.0, + "step": 19983 + }, + { + "epoch": 2.542170207352754, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8042309284210205, + "num_tokens": 762577801.0, + "step": 19984 + }, + { + "epoch": 2.5422974176313446, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8001710176467896, + "num_tokens": 762617413.0, + "step": 19985 + }, + { + "epoch": 2.542424627909935, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8138709664344788, + "num_tokens": 762654811.0, + "step": 19986 + }, + { + "epoch": 2.5425518381885257, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8083663582801819, + "num_tokens": 762692902.0, + "step": 19987 + }, + { + "epoch": 2.5426790484671162, + "grad_norm": 74.0, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.800794243812561, + "num_tokens": 762731984.0, + "step": 19988 + }, + { + "epoch": 2.5428062587457068, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8111071586608887, + "num_tokens": 762767104.0, + "step": 19989 + }, + { + "epoch": 2.5429334690242973, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.787359356880188, + "num_tokens": 762804982.0, + "step": 19990 + }, + { + "epoch": 2.543060679302888, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7909301519393921, + "num_tokens": 762842956.0, + "step": 19991 + }, + { + "epoch": 2.5431878895814783, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.809744119644165, + "num_tokens": 762875445.0, + "step": 19992 + }, + { + "epoch": 2.5433150998600684, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8085267543792725, + "num_tokens": 762906061.0, + "step": 19993 + }, + { + "epoch": 2.5434423101386594, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8092005252838135, + "num_tokens": 762945550.0, + "step": 19994 + }, + { + "epoch": 2.5435695204172495, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7986704111099243, + "num_tokens": 762982586.0, + "step": 19995 + }, + { + "epoch": 2.5436967306958405, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7980881929397583, + "num_tokens": 763013527.0, + "step": 19996 + }, + { + "epoch": 2.5438239409744305, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7954870462417603, + "num_tokens": 763054363.0, + "step": 19997 + }, + { + "epoch": 2.5439511512530215, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.8018866181373596, + "num_tokens": 763096834.0, + "step": 19998 + }, + { + "epoch": 2.5440783615316116, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8073331117630005, + "num_tokens": 763139611.0, + "step": 19999 + }, + { + "epoch": 2.544205571810202, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8059384822845459, + "num_tokens": 763181085.0, + "step": 20000 + }, + { + "epoch": 2.5443327820887927, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.797382116317749, + "num_tokens": 763219811.0, + "step": 20001 + }, + { + "epoch": 2.544459992367383, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7899975776672363, + "num_tokens": 763257745.0, + "step": 20002 + }, + { + "epoch": 2.5445872026459737, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7868055701255798, + "num_tokens": 763299475.0, + "step": 20003 + }, + { + "epoch": 2.5447144129245642, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7794433236122131, + "num_tokens": 763333960.0, + "step": 20004 + }, + { + "epoch": 2.5448416232031548, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7901387214660645, + "num_tokens": 763378646.0, + "step": 20005 + }, + { + "epoch": 2.5449688334817453, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.8028677701950073, + "num_tokens": 763412441.0, + "step": 20006 + }, + { + "epoch": 2.545096043760336, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8025091886520386, + "num_tokens": 763446713.0, + "step": 20007 + }, + { + "epoch": 2.5452232540389264, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.8043098449707031, + "num_tokens": 763478686.0, + "step": 20008 + }, + { + "epoch": 2.545350464317517, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8161722421646118, + "num_tokens": 763517258.0, + "step": 20009 + }, + { + "epoch": 2.5454776745961074, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7937097549438477, + "num_tokens": 763553229.0, + "step": 20010 + }, + { + "epoch": 2.545604884874698, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8095248937606812, + "num_tokens": 763594274.0, + "step": 20011 + }, + { + "epoch": 2.5457320951532885, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7958459854125977, + "num_tokens": 763632162.0, + "step": 20012 + }, + { + "epoch": 2.545859305431879, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8052346706390381, + "num_tokens": 763666797.0, + "step": 20013 + }, + { + "epoch": 2.5459865157104695, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7937886118888855, + "num_tokens": 763705475.0, + "step": 20014 + }, + { + "epoch": 2.54611372598906, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8169570565223694, + "num_tokens": 763739434.0, + "step": 20015 + }, + { + "epoch": 2.5462409362676506, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7830212116241455, + "num_tokens": 763773262.0, + "step": 20016 + }, + { + "epoch": 2.546368146546241, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7903479337692261, + "num_tokens": 763812573.0, + "step": 20017 + }, + { + "epoch": 2.546495356824831, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8023513555526733, + "num_tokens": 763848555.0, + "step": 20018 + }, + { + "epoch": 2.546622567103422, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7926087975502014, + "num_tokens": 763884608.0, + "step": 20019 + }, + { + "epoch": 2.5467497773820122, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7951880097389221, + "num_tokens": 763922038.0, + "step": 20020 + }, + { + "epoch": 2.546876987660603, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.796341061592102, + "num_tokens": 763958089.0, + "step": 20021 + }, + { + "epoch": 2.5470041979391933, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8021398186683655, + "num_tokens": 764000802.0, + "step": 20022 + }, + { + "epoch": 2.547131408217784, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8034857511520386, + "num_tokens": 764036628.0, + "step": 20023 + }, + { + "epoch": 2.5472586184963744, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8038243055343628, + "num_tokens": 764073643.0, + "step": 20024 + }, + { + "epoch": 2.547385828774965, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8046920299530029, + "num_tokens": 764110473.0, + "step": 20025 + }, + { + "epoch": 2.5475130390535554, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7980935573577881, + "num_tokens": 764142941.0, + "step": 20026 + }, + { + "epoch": 2.547640249332146, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8031604290008545, + "num_tokens": 764178752.0, + "step": 20027 + }, + { + "epoch": 2.5477674596107365, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7982041835784912, + "num_tokens": 764217440.0, + "step": 20028 + }, + { + "epoch": 2.547894669889327, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7882477045059204, + "num_tokens": 764257524.0, + "step": 20029 + }, + { + "epoch": 2.5480218801679175, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.796101450920105, + "num_tokens": 764293645.0, + "step": 20030 + }, + { + "epoch": 2.548149090446508, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8072248697280884, + "num_tokens": 764331660.0, + "step": 20031 + }, + { + "epoch": 2.5482763007250986, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.5737, + "mean_token_accuracy": 0.8268385529518127, + "num_tokens": 764376577.0, + "step": 20032 + }, + { + "epoch": 2.548403511003689, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.8003984689712524, + "num_tokens": 764415492.0, + "step": 20033 + }, + { + "epoch": 2.5485307212822796, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7885259389877319, + "num_tokens": 764453948.0, + "step": 20034 + }, + { + "epoch": 2.54865793156087, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8092558979988098, + "num_tokens": 764493906.0, + "step": 20035 + }, + { + "epoch": 2.5487851418394607, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7957137227058411, + "num_tokens": 764532379.0, + "step": 20036 + }, + { + "epoch": 2.5489123521180512, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7737594842910767, + "num_tokens": 764567726.0, + "step": 20037 + }, + { + "epoch": 2.5490395623966418, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8026754260063171, + "num_tokens": 764606229.0, + "step": 20038 + }, + { + "epoch": 2.5491667726752323, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8153457641601562, + "num_tokens": 764641346.0, + "step": 20039 + }, + { + "epoch": 2.549293982953823, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8051896691322327, + "num_tokens": 764687286.0, + "step": 20040 + }, + { + "epoch": 2.5494211932324133, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8083757758140564, + "num_tokens": 764729235.0, + "step": 20041 + }, + { + "epoch": 2.549548403511004, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.795494556427002, + "num_tokens": 764771834.0, + "step": 20042 + }, + { + "epoch": 2.549675613789594, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7953780889511108, + "num_tokens": 764808307.0, + "step": 20043 + }, + { + "epoch": 2.549802824068185, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7950990200042725, + "num_tokens": 764848373.0, + "step": 20044 + }, + { + "epoch": 2.549930034346775, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7958670258522034, + "num_tokens": 764886468.0, + "step": 20045 + }, + { + "epoch": 2.550057244625366, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8070344924926758, + "num_tokens": 764928690.0, + "step": 20046 + }, + { + "epoch": 2.550184454903956, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7636356353759766, + "num_tokens": 764969148.0, + "step": 20047 + }, + { + "epoch": 2.5503116651825466, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7967172861099243, + "num_tokens": 765010754.0, + "step": 20048 + }, + { + "epoch": 2.550438875461137, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.8013436794281006, + "num_tokens": 765043562.0, + "step": 20049 + }, + { + "epoch": 2.5505660857397277, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7828454971313477, + "num_tokens": 765085638.0, + "step": 20050 + }, + { + "epoch": 2.550693296018318, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7995179891586304, + "num_tokens": 765115847.0, + "step": 20051 + }, + { + "epoch": 2.5508205062969087, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7980928421020508, + "num_tokens": 765146839.0, + "step": 20052 + }, + { + "epoch": 2.5509477165754992, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7997820377349854, + "num_tokens": 765185921.0, + "step": 20053 + }, + { + "epoch": 2.5510749268540898, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7943931221961975, + "num_tokens": 765224534.0, + "step": 20054 + }, + { + "epoch": 2.5512021371326803, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8122514486312866, + "num_tokens": 765267160.0, + "step": 20055 + }, + { + "epoch": 2.551329347411271, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.802981972694397, + "num_tokens": 765304770.0, + "step": 20056 + }, + { + "epoch": 2.5514565576898613, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7958999276161194, + "num_tokens": 765341966.0, + "step": 20057 + }, + { + "epoch": 2.551583767968452, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.8178722858428955, + "num_tokens": 765382653.0, + "step": 20058 + }, + { + "epoch": 2.5517109782470424, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7951602339744568, + "num_tokens": 765414940.0, + "step": 20059 + }, + { + "epoch": 2.551838188525633, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8045651316642761, + "num_tokens": 765445549.0, + "step": 20060 + }, + { + "epoch": 2.5519653988042235, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.807668149471283, + "num_tokens": 765484474.0, + "step": 20061 + }, + { + "epoch": 2.552092609082814, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8009740114212036, + "num_tokens": 765522651.0, + "step": 20062 + }, + { + "epoch": 2.5522198193614045, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8006161451339722, + "num_tokens": 765565983.0, + "step": 20063 + }, + { + "epoch": 2.552347029639995, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.8002147674560547, + "num_tokens": 765604056.0, + "step": 20064 + }, + { + "epoch": 2.5524742399185856, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8153360486030579, + "num_tokens": 765646062.0, + "step": 20065 + }, + { + "epoch": 2.5526014501971757, + "grad_norm": 10.3125, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7980161905288696, + "num_tokens": 765687284.0, + "step": 20066 + }, + { + "epoch": 2.5527286604757666, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.786763072013855, + "num_tokens": 765723743.0, + "step": 20067 + }, + { + "epoch": 2.5528558707543567, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8007661700248718, + "num_tokens": 765768472.0, + "step": 20068 + }, + { + "epoch": 2.5529830810329477, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8169100284576416, + "num_tokens": 765809040.0, + "step": 20069 + }, + { + "epoch": 2.5531102913115378, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.8033032417297363, + "num_tokens": 765845968.0, + "step": 20070 + }, + { + "epoch": 2.5532375015901287, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8032040596008301, + "num_tokens": 765881037.0, + "step": 20071 + }, + { + "epoch": 2.553364711868719, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8058542013168335, + "num_tokens": 765918945.0, + "step": 20072 + }, + { + "epoch": 2.5534919221473094, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8137935400009155, + "num_tokens": 765956128.0, + "step": 20073 + }, + { + "epoch": 2.5536191324259, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.8157497048377991, + "num_tokens": 765997109.0, + "step": 20074 + }, + { + "epoch": 2.5537463427044904, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.800376296043396, + "num_tokens": 766032557.0, + "step": 20075 + }, + { + "epoch": 2.553873552983081, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7971588373184204, + "num_tokens": 766075285.0, + "step": 20076 + }, + { + "epoch": 2.5540007632616715, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7968155741691589, + "num_tokens": 766110600.0, + "step": 20077 + }, + { + "epoch": 2.554127973540262, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7856010794639587, + "num_tokens": 766146877.0, + "step": 20078 + }, + { + "epoch": 2.5542551838188525, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7963021993637085, + "num_tokens": 766185164.0, + "step": 20079 + }, + { + "epoch": 2.554382394097443, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.5764, + "mean_token_accuracy": 0.8245422840118408, + "num_tokens": 766221442.0, + "step": 20080 + }, + { + "epoch": 2.5545096043760336, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7724546194076538, + "num_tokens": 766255258.0, + "step": 20081 + }, + { + "epoch": 2.554636814654624, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7979221343994141, + "num_tokens": 766292392.0, + "step": 20082 + }, + { + "epoch": 2.5547640249332146, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.793647289276123, + "num_tokens": 766335312.0, + "step": 20083 + }, + { + "epoch": 2.554891235211805, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8031019568443298, + "num_tokens": 766372189.0, + "step": 20084 + }, + { + "epoch": 2.5550184454903957, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7870610356330872, + "num_tokens": 766414021.0, + "step": 20085 + }, + { + "epoch": 2.5551456557689862, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.8139349818229675, + "num_tokens": 766452005.0, + "step": 20086 + }, + { + "epoch": 2.5552728660475768, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7953230142593384, + "num_tokens": 766492150.0, + "step": 20087 + }, + { + "epoch": 2.5554000763261673, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8144624829292297, + "num_tokens": 766533889.0, + "step": 20088 + }, + { + "epoch": 2.555527286604758, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7989289164543152, + "num_tokens": 766565008.0, + "step": 20089 + }, + { + "epoch": 2.5556544968833483, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8014620542526245, + "num_tokens": 766605744.0, + "step": 20090 + }, + { + "epoch": 2.5557817071619384, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8120578527450562, + "num_tokens": 766645782.0, + "step": 20091 + }, + { + "epoch": 2.5559089174405294, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7852372527122498, + "num_tokens": 766683135.0, + "step": 20092 + }, + { + "epoch": 2.5560361277191195, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8011858463287354, + "num_tokens": 766720691.0, + "step": 20093 + }, + { + "epoch": 2.5561633379977104, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7929261922836304, + "num_tokens": 766755426.0, + "step": 20094 + }, + { + "epoch": 2.5562905482763005, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8009309768676758, + "num_tokens": 766788315.0, + "step": 20095 + }, + { + "epoch": 2.5564177585548915, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8168730139732361, + "num_tokens": 766825664.0, + "step": 20096 + }, + { + "epoch": 2.5565449688334816, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8098466396331787, + "num_tokens": 766868145.0, + "step": 20097 + }, + { + "epoch": 2.556672179112072, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7887583374977112, + "num_tokens": 766907278.0, + "step": 20098 + }, + { + "epoch": 2.5567993893906626, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8081984519958496, + "num_tokens": 766941662.0, + "step": 20099 + }, + { + "epoch": 2.556926599669253, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7835574150085449, + "num_tokens": 766977725.0, + "step": 20100 + }, + { + "epoch": 2.5570538099478437, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.788394033908844, + "num_tokens": 767011011.0, + "step": 20101 + }, + { + "epoch": 2.5571810202264342, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7976443767547607, + "num_tokens": 767047078.0, + "step": 20102 + }, + { + "epoch": 2.5573082305050248, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.8005266785621643, + "num_tokens": 767089208.0, + "step": 20103 + }, + { + "epoch": 2.5574354407836153, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7836676836013794, + "num_tokens": 767131225.0, + "step": 20104 + }, + { + "epoch": 2.557562651062206, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7814717292785645, + "num_tokens": 767165222.0, + "step": 20105 + }, + { + "epoch": 2.5576898613407963, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.802742600440979, + "num_tokens": 767200870.0, + "step": 20106 + }, + { + "epoch": 2.557817071619387, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.800958514213562, + "num_tokens": 767236881.0, + "step": 20107 + }, + { + "epoch": 2.5579442818979774, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8020258545875549, + "num_tokens": 767269716.0, + "step": 20108 + }, + { + "epoch": 2.558071492176568, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7876933217048645, + "num_tokens": 767306840.0, + "step": 20109 + }, + { + "epoch": 2.5581987024551585, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8119153380393982, + "num_tokens": 767346966.0, + "step": 20110 + }, + { + "epoch": 2.558325912733749, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8005126118659973, + "num_tokens": 767387084.0, + "step": 20111 + }, + { + "epoch": 2.5584531230123395, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7760549783706665, + "num_tokens": 767417857.0, + "step": 20112 + }, + { + "epoch": 2.55858033329093, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7985166311264038, + "num_tokens": 767455771.0, + "step": 20113 + }, + { + "epoch": 2.5587075435695206, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7945277094841003, + "num_tokens": 767493187.0, + "step": 20114 + }, + { + "epoch": 2.558834753848111, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7919106483459473, + "num_tokens": 767534526.0, + "step": 20115 + }, + { + "epoch": 2.558961964126701, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7918431162834167, + "num_tokens": 767569297.0, + "step": 20116 + }, + { + "epoch": 2.559089174405292, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7931365370750427, + "num_tokens": 767605860.0, + "step": 20117 + }, + { + "epoch": 2.5592163846838822, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7987662553787231, + "num_tokens": 767645252.0, + "step": 20118 + }, + { + "epoch": 2.559343594962473, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7935129404067993, + "num_tokens": 767681942.0, + "step": 20119 + }, + { + "epoch": 2.5594708052410633, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.803257942199707, + "num_tokens": 767720631.0, + "step": 20120 + }, + { + "epoch": 2.559598015519654, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8050867319107056, + "num_tokens": 767757018.0, + "step": 20121 + }, + { + "epoch": 2.5597252257982444, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8061888217926025, + "num_tokens": 767796046.0, + "step": 20122 + }, + { + "epoch": 2.559852436076835, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8142974376678467, + "num_tokens": 767835678.0, + "step": 20123 + }, + { + "epoch": 2.5599796463554254, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8131647109985352, + "num_tokens": 767870515.0, + "step": 20124 + }, + { + "epoch": 2.560106856634016, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7832727432250977, + "num_tokens": 767911573.0, + "step": 20125 + }, + { + "epoch": 2.5602340669126065, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7969628572463989, + "num_tokens": 767950161.0, + "step": 20126 + }, + { + "epoch": 2.560361277191197, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7995668649673462, + "num_tokens": 767989431.0, + "step": 20127 + }, + { + "epoch": 2.5604884874697875, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8052709102630615, + "num_tokens": 768023551.0, + "step": 20128 + }, + { + "epoch": 2.560615697748378, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8056542277336121, + "num_tokens": 768055317.0, + "step": 20129 + }, + { + "epoch": 2.5607429080269686, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7879589796066284, + "num_tokens": 768089575.0, + "step": 20130 + }, + { + "epoch": 2.560870118305559, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7925042510032654, + "num_tokens": 768131795.0, + "step": 20131 + }, + { + "epoch": 2.5609973285841496, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8014265894889832, + "num_tokens": 768165714.0, + "step": 20132 + }, + { + "epoch": 2.56112453886274, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.8091387748718262, + "num_tokens": 768203475.0, + "step": 20133 + }, + { + "epoch": 2.5612517491413307, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.5823, + "mean_token_accuracy": 0.8253754377365112, + "num_tokens": 768244841.0, + "step": 20134 + }, + { + "epoch": 2.561378959419921, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7958616614341736, + "num_tokens": 768286502.0, + "step": 20135 + }, + { + "epoch": 2.5615061696985117, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8047168254852295, + "num_tokens": 768320628.0, + "step": 20136 + }, + { + "epoch": 2.5616333799771023, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7913713455200195, + "num_tokens": 768353630.0, + "step": 20137 + }, + { + "epoch": 2.561760590255693, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7944962978363037, + "num_tokens": 768388170.0, + "step": 20138 + }, + { + "epoch": 2.5618878005342833, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.815231442451477, + "num_tokens": 768428238.0, + "step": 20139 + }, + { + "epoch": 2.562015010812874, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.812645435333252, + "num_tokens": 768460075.0, + "step": 20140 + }, + { + "epoch": 2.562142221091464, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8195371627807617, + "num_tokens": 768493982.0, + "step": 20141 + }, + { + "epoch": 2.562269431370055, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8024023771286011, + "num_tokens": 768529121.0, + "step": 20142 + }, + { + "epoch": 2.562396641648645, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8001360893249512, + "num_tokens": 768560730.0, + "step": 20143 + }, + { + "epoch": 2.562523851927236, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8153025507926941, + "num_tokens": 768603015.0, + "step": 20144 + }, + { + "epoch": 2.562651062205826, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8005406856536865, + "num_tokens": 768639847.0, + "step": 20145 + }, + { + "epoch": 2.5627782724844166, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7981711030006409, + "num_tokens": 768681614.0, + "step": 20146 + }, + { + "epoch": 2.562905482763007, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8048446774482727, + "num_tokens": 768722052.0, + "step": 20147 + }, + { + "epoch": 2.5630326930415976, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7891390919685364, + "num_tokens": 768754191.0, + "step": 20148 + }, + { + "epoch": 2.563159903320188, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7907441854476929, + "num_tokens": 768792981.0, + "step": 20149 + }, + { + "epoch": 2.5632871135987787, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7864859700202942, + "num_tokens": 768827040.0, + "step": 20150 + }, + { + "epoch": 2.5634143238773692, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.800786018371582, + "num_tokens": 768865669.0, + "step": 20151 + }, + { + "epoch": 2.5635415341559598, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8008607625961304, + "num_tokens": 768903306.0, + "step": 20152 + }, + { + "epoch": 2.5636687444345503, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8010923266410828, + "num_tokens": 768946322.0, + "step": 20153 + }, + { + "epoch": 2.563795954713141, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.5735, + "mean_token_accuracy": 0.8212952613830566, + "num_tokens": 768987350.0, + "step": 20154 + }, + { + "epoch": 2.5639231649917313, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.81296706199646, + "num_tokens": 769030762.0, + "step": 20155 + }, + { + "epoch": 2.564050375270322, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.811526894569397, + "num_tokens": 769069827.0, + "step": 20156 + }, + { + "epoch": 2.5641775855489124, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.794579267501831, + "num_tokens": 769107742.0, + "step": 20157 + }, + { + "epoch": 2.564304795827503, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.8005313873291016, + "num_tokens": 769153286.0, + "step": 20158 + }, + { + "epoch": 2.5644320061060935, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.8039161562919617, + "num_tokens": 769192646.0, + "step": 20159 + }, + { + "epoch": 2.564559216384684, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7789385318756104, + "num_tokens": 769230360.0, + "step": 20160 + }, + { + "epoch": 2.5646864266632745, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8159303069114685, + "num_tokens": 769270754.0, + "step": 20161 + }, + { + "epoch": 2.564813636941865, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7794008255004883, + "num_tokens": 769310109.0, + "step": 20162 + }, + { + "epoch": 2.5649408472204556, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7980776429176331, + "num_tokens": 769352101.0, + "step": 20163 + }, + { + "epoch": 2.5650680574990457, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8147685527801514, + "num_tokens": 769386570.0, + "step": 20164 + }, + { + "epoch": 2.5651952677776366, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8165416717529297, + "num_tokens": 769427167.0, + "step": 20165 + }, + { + "epoch": 2.5653224780562267, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7865756750106812, + "num_tokens": 769465456.0, + "step": 20166 + }, + { + "epoch": 2.5654496883348177, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8063651323318481, + "num_tokens": 769497135.0, + "step": 20167 + }, + { + "epoch": 2.5655768986134078, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7941315174102783, + "num_tokens": 769533706.0, + "step": 20168 + }, + { + "epoch": 2.5657041088919987, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7962366342544556, + "num_tokens": 769578698.0, + "step": 20169 + }, + { + "epoch": 2.565831319170589, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8177506327629089, + "num_tokens": 769619799.0, + "step": 20170 + }, + { + "epoch": 2.5659585294491793, + "grad_norm": 46.25, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8004904985427856, + "num_tokens": 769655973.0, + "step": 20171 + }, + { + "epoch": 2.56608573972777, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7930251359939575, + "num_tokens": 769692333.0, + "step": 20172 + }, + { + "epoch": 2.5662129500063604, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8042696714401245, + "num_tokens": 769731804.0, + "step": 20173 + }, + { + "epoch": 2.566340160284951, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8011235594749451, + "num_tokens": 769767409.0, + "step": 20174 + }, + { + "epoch": 2.5664673705635415, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.792503297328949, + "num_tokens": 769806468.0, + "step": 20175 + }, + { + "epoch": 2.566594580842132, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8066321015357971, + "num_tokens": 769842025.0, + "step": 20176 + }, + { + "epoch": 2.5667217911207225, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.799663782119751, + "num_tokens": 769884125.0, + "step": 20177 + }, + { + "epoch": 2.566849001399313, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7838610410690308, + "num_tokens": 769920523.0, + "step": 20178 + }, + { + "epoch": 2.5669762116779036, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.80100417137146, + "num_tokens": 769960141.0, + "step": 20179 + }, + { + "epoch": 2.567103421956494, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8028320074081421, + "num_tokens": 769995281.0, + "step": 20180 + }, + { + "epoch": 2.5672306322350846, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.80018150806427, + "num_tokens": 770028230.0, + "step": 20181 + }, + { + "epoch": 2.567357842513675, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8089931011199951, + "num_tokens": 770068996.0, + "step": 20182 + }, + { + "epoch": 2.5674850527922657, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7839736938476562, + "num_tokens": 770107018.0, + "step": 20183 + }, + { + "epoch": 2.567612263070856, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7981942296028137, + "num_tokens": 770140251.0, + "step": 20184 + }, + { + "epoch": 2.5677394733494467, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7959476709365845, + "num_tokens": 770182241.0, + "step": 20185 + }, + { + "epoch": 2.5678666836280373, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8197052478790283, + "num_tokens": 770223267.0, + "step": 20186 + }, + { + "epoch": 2.567993893906628, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7909326553344727, + "num_tokens": 770261856.0, + "step": 20187 + }, + { + "epoch": 2.5681211041852183, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7891031503677368, + "num_tokens": 770299762.0, + "step": 20188 + }, + { + "epoch": 2.5682483144638084, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7853585481643677, + "num_tokens": 770333690.0, + "step": 20189 + }, + { + "epoch": 2.5683755247423994, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8043817281723022, + "num_tokens": 770375755.0, + "step": 20190 + }, + { + "epoch": 2.5685027350209895, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8151648044586182, + "num_tokens": 770417805.0, + "step": 20191 + }, + { + "epoch": 2.5686299452995804, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7971316576004028, + "num_tokens": 770458183.0, + "step": 20192 + }, + { + "epoch": 2.5687571555781705, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8023554086685181, + "num_tokens": 770488973.0, + "step": 20193 + }, + { + "epoch": 2.5688843658567615, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7924779653549194, + "num_tokens": 770529030.0, + "step": 20194 + }, + { + "epoch": 2.5690115761353516, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7904040217399597, + "num_tokens": 770574697.0, + "step": 20195 + }, + { + "epoch": 2.569138786413942, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8093368411064148, + "num_tokens": 770615417.0, + "step": 20196 + }, + { + "epoch": 2.5692659966925326, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7825593948364258, + "num_tokens": 770653228.0, + "step": 20197 + }, + { + "epoch": 2.569393206971123, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8209472894668579, + "num_tokens": 770693876.0, + "step": 20198 + }, + { + "epoch": 2.5695204172497137, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7975924015045166, + "num_tokens": 770728754.0, + "step": 20199 + }, + { + "epoch": 2.5696476275283042, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8010131120681763, + "num_tokens": 770767961.0, + "step": 20200 + }, + { + "epoch": 2.5697748378068948, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.77008056640625, + "num_tokens": 770808880.0, + "step": 20201 + }, + { + "epoch": 2.5699020480854853, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7927678227424622, + "num_tokens": 770846483.0, + "step": 20202 + }, + { + "epoch": 2.570029258364076, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7758076190948486, + "num_tokens": 770884234.0, + "step": 20203 + }, + { + "epoch": 2.5701564686426663, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7798421382904053, + "num_tokens": 770918785.0, + "step": 20204 + }, + { + "epoch": 2.570283678921257, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8222870826721191, + "num_tokens": 770956889.0, + "step": 20205 + }, + { + "epoch": 2.5704108891998474, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7859923839569092, + "num_tokens": 770994937.0, + "step": 20206 + }, + { + "epoch": 2.570538099478438, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8061614036560059, + "num_tokens": 771031451.0, + "step": 20207 + }, + { + "epoch": 2.5706653097570284, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8046519756317139, + "num_tokens": 771061575.0, + "step": 20208 + }, + { + "epoch": 2.570792520035619, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8030376434326172, + "num_tokens": 771103366.0, + "step": 20209 + }, + { + "epoch": 2.5709197303142095, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8083876967430115, + "num_tokens": 771142178.0, + "step": 20210 + }, + { + "epoch": 2.5710469405928, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.797091007232666, + "num_tokens": 771179296.0, + "step": 20211 + }, + { + "epoch": 2.5711741508713906, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7954905033111572, + "num_tokens": 771219084.0, + "step": 20212 + }, + { + "epoch": 2.571301361149981, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.8092926740646362, + "num_tokens": 771250609.0, + "step": 20213 + }, + { + "epoch": 2.571428571428571, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8021681308746338, + "num_tokens": 771286191.0, + "step": 20214 + }, + { + "epoch": 2.571555781707162, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7957549691200256, + "num_tokens": 771320898.0, + "step": 20215 + }, + { + "epoch": 2.5716829919857522, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7868898510932922, + "num_tokens": 771353977.0, + "step": 20216 + }, + { + "epoch": 2.571810202264343, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8036831617355347, + "num_tokens": 771391960.0, + "step": 20217 + }, + { + "epoch": 2.5719374125429333, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7919119596481323, + "num_tokens": 771427764.0, + "step": 20218 + }, + { + "epoch": 2.572064622821524, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7966216802597046, + "num_tokens": 771469706.0, + "step": 20219 + }, + { + "epoch": 2.5721918331001143, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.8241666555404663, + "num_tokens": 771504296.0, + "step": 20220 + }, + { + "epoch": 2.572319043378705, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8007678389549255, + "num_tokens": 771544496.0, + "step": 20221 + }, + { + "epoch": 2.5724462536572954, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.809402346611023, + "num_tokens": 771584322.0, + "step": 20222 + }, + { + "epoch": 2.572573463935886, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8124153017997742, + "num_tokens": 771621424.0, + "step": 20223 + }, + { + "epoch": 2.5727006742144765, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.8009235262870789, + "num_tokens": 771658173.0, + "step": 20224 + }, + { + "epoch": 2.572827884493067, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7939456701278687, + "num_tokens": 771698182.0, + "step": 20225 + }, + { + "epoch": 2.5729550947716575, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7818847894668579, + "num_tokens": 771731559.0, + "step": 20226 + }, + { + "epoch": 2.573082305050248, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8018798828125, + "num_tokens": 771774362.0, + "step": 20227 + }, + { + "epoch": 2.5732095153288386, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7893222570419312, + "num_tokens": 771810540.0, + "step": 20228 + }, + { + "epoch": 2.573336725607429, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7986169457435608, + "num_tokens": 771846221.0, + "step": 20229 + }, + { + "epoch": 2.5734639358860196, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8086002469062805, + "num_tokens": 771882268.0, + "step": 20230 + }, + { + "epoch": 2.57359114616461, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.796766996383667, + "num_tokens": 771913176.0, + "step": 20231 + }, + { + "epoch": 2.5737183564432007, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7885643243789673, + "num_tokens": 771948163.0, + "step": 20232 + }, + { + "epoch": 2.573845566721791, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8118394017219543, + "num_tokens": 771988296.0, + "step": 20233 + }, + { + "epoch": 2.5739727770003817, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.805608332157135, + "num_tokens": 772024330.0, + "step": 20234 + }, + { + "epoch": 2.5740999872789723, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7877404093742371, + "num_tokens": 772062291.0, + "step": 20235 + }, + { + "epoch": 2.574227197557563, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.803507387638092, + "num_tokens": 772110239.0, + "step": 20236 + }, + { + "epoch": 2.5743544078361533, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.8011073470115662, + "num_tokens": 772153201.0, + "step": 20237 + }, + { + "epoch": 2.574481618114744, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7966775894165039, + "num_tokens": 772186816.0, + "step": 20238 + }, + { + "epoch": 2.574608828393334, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.810939610004425, + "num_tokens": 772224238.0, + "step": 20239 + }, + { + "epoch": 2.574736038671925, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8084408044815063, + "num_tokens": 772263505.0, + "step": 20240 + }, + { + "epoch": 2.574863248950515, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8155759572982788, + "num_tokens": 772303205.0, + "step": 20241 + }, + { + "epoch": 2.574990459229106, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8064886331558228, + "num_tokens": 772344495.0, + "step": 20242 + }, + { + "epoch": 2.575117669507696, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7975595593452454, + "num_tokens": 772380296.0, + "step": 20243 + }, + { + "epoch": 2.5752448797862866, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8158133029937744, + "num_tokens": 772423933.0, + "step": 20244 + }, + { + "epoch": 2.575372090064877, + "grad_norm": 45.75, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.785594642162323, + "num_tokens": 772464514.0, + "step": 20245 + }, + { + "epoch": 2.5754993003434676, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8184019327163696, + "num_tokens": 772498501.0, + "step": 20246 + }, + { + "epoch": 2.575626510622058, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7945053577423096, + "num_tokens": 772538072.0, + "step": 20247 + }, + { + "epoch": 2.5757537209006487, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.8014323115348816, + "num_tokens": 772577434.0, + "step": 20248 + }, + { + "epoch": 2.575880931179239, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8154086470603943, + "num_tokens": 772612917.0, + "step": 20249 + }, + { + "epoch": 2.5760081414578297, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8138654828071594, + "num_tokens": 772652939.0, + "step": 20250 + }, + { + "epoch": 2.5761353517364203, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.806645929813385, + "num_tokens": 772688018.0, + "step": 20251 + }, + { + "epoch": 2.576262562015011, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7908685207366943, + "num_tokens": 772728071.0, + "step": 20252 + }, + { + "epoch": 2.5763897722936013, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8024380803108215, + "num_tokens": 772762730.0, + "step": 20253 + }, + { + "epoch": 2.576516982572192, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.802393913269043, + "num_tokens": 772799125.0, + "step": 20254 + }, + { + "epoch": 2.5766441928507824, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8132274150848389, + "num_tokens": 772838657.0, + "step": 20255 + }, + { + "epoch": 2.576771403129373, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8040881156921387, + "num_tokens": 772876470.0, + "step": 20256 + }, + { + "epoch": 2.5768986134079634, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7921700477600098, + "num_tokens": 772910369.0, + "step": 20257 + }, + { + "epoch": 2.577025823686554, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8065987825393677, + "num_tokens": 772948129.0, + "step": 20258 + }, + { + "epoch": 2.5771530339651445, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8254024982452393, + "num_tokens": 772985359.0, + "step": 20259 + }, + { + "epoch": 2.577280244243735, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.800307035446167, + "num_tokens": 773019402.0, + "step": 20260 + }, + { + "epoch": 2.5774074545223256, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7997061014175415, + "num_tokens": 773059176.0, + "step": 20261 + }, + { + "epoch": 2.5775346648009156, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7982200384140015, + "num_tokens": 773094274.0, + "step": 20262 + }, + { + "epoch": 2.5776618750795066, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7998383045196533, + "num_tokens": 773135679.0, + "step": 20263 + }, + { + "epoch": 2.5777890853580967, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7980777025222778, + "num_tokens": 773174270.0, + "step": 20264 + }, + { + "epoch": 2.5779162956366877, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8080947995185852, + "num_tokens": 773213582.0, + "step": 20265 + }, + { + "epoch": 2.5780435059152778, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7979874014854431, + "num_tokens": 773244980.0, + "step": 20266 + }, + { + "epoch": 2.5781707161938687, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8152535557746887, + "num_tokens": 773287395.0, + "step": 20267 + }, + { + "epoch": 2.578297926472459, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7930608987808228, + "num_tokens": 773324593.0, + "step": 20268 + }, + { + "epoch": 2.5784251367510493, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7984605431556702, + "num_tokens": 773361802.0, + "step": 20269 + }, + { + "epoch": 2.57855234702964, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8063626885414124, + "num_tokens": 773401348.0, + "step": 20270 + }, + { + "epoch": 2.5786795573082304, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.82146817445755, + "num_tokens": 773439040.0, + "step": 20271 + }, + { + "epoch": 2.578806767586821, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8082767724990845, + "num_tokens": 773471382.0, + "step": 20272 + }, + { + "epoch": 2.5789339778654115, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7864947319030762, + "num_tokens": 773508924.0, + "step": 20273 + }, + { + "epoch": 2.579061188144002, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.5974, + "mean_token_accuracy": 0.8182922005653381, + "num_tokens": 773550219.0, + "step": 20274 + }, + { + "epoch": 2.5791883984225925, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.8039582371711731, + "num_tokens": 773589363.0, + "step": 20275 + }, + { + "epoch": 2.579315608701183, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.8007113933563232, + "num_tokens": 773630289.0, + "step": 20276 + }, + { + "epoch": 2.5794428189797736, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7982245683670044, + "num_tokens": 773668660.0, + "step": 20277 + }, + { + "epoch": 2.579570029258364, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7983149290084839, + "num_tokens": 773704318.0, + "step": 20278 + }, + { + "epoch": 2.5796972395369546, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7861864566802979, + "num_tokens": 773745821.0, + "step": 20279 + }, + { + "epoch": 2.579824449815545, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8137936592102051, + "num_tokens": 773782880.0, + "step": 20280 + }, + { + "epoch": 2.5799516600941357, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.783760666847229, + "num_tokens": 773824880.0, + "step": 20281 + }, + { + "epoch": 2.580078870372726, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8071436882019043, + "num_tokens": 773863210.0, + "step": 20282 + }, + { + "epoch": 2.5802060806513167, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8102505803108215, + "num_tokens": 773903432.0, + "step": 20283 + }, + { + "epoch": 2.5803332909299073, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8077424764633179, + "num_tokens": 773942424.0, + "step": 20284 + }, + { + "epoch": 2.580460501208498, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7954689264297485, + "num_tokens": 773976598.0, + "step": 20285 + }, + { + "epoch": 2.5805877114870883, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7987390756607056, + "num_tokens": 774016696.0, + "step": 20286 + }, + { + "epoch": 2.5807149217656784, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7916616797447205, + "num_tokens": 774059046.0, + "step": 20287 + }, + { + "epoch": 2.5808421320442694, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7974884510040283, + "num_tokens": 774095322.0, + "step": 20288 + }, + { + "epoch": 2.5809693423228595, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.792081356048584, + "num_tokens": 774134506.0, + "step": 20289 + }, + { + "epoch": 2.5810965526014504, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8221917748451233, + "num_tokens": 774176997.0, + "step": 20290 + }, + { + "epoch": 2.5812237628800405, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7968581914901733, + "num_tokens": 774213900.0, + "step": 20291 + }, + { + "epoch": 2.5813509731586315, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7950302362442017, + "num_tokens": 774259687.0, + "step": 20292 + }, + { + "epoch": 2.5814781834372216, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8096941709518433, + "num_tokens": 774297200.0, + "step": 20293 + }, + { + "epoch": 2.581605393715812, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8147550821304321, + "num_tokens": 774336642.0, + "step": 20294 + }, + { + "epoch": 2.5817326039944026, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.786457896232605, + "num_tokens": 774376762.0, + "step": 20295 + }, + { + "epoch": 2.581859814272993, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8140593767166138, + "num_tokens": 774421360.0, + "step": 20296 + }, + { + "epoch": 2.5819870245515837, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8097715973854065, + "num_tokens": 774460850.0, + "step": 20297 + }, + { + "epoch": 2.582114234830174, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7988253831863403, + "num_tokens": 774497542.0, + "step": 20298 + }, + { + "epoch": 2.5822414451087647, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7887285947799683, + "num_tokens": 774538164.0, + "step": 20299 + }, + { + "epoch": 2.5823686553873553, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7937240600585938, + "num_tokens": 774576118.0, + "step": 20300 + }, + { + "epoch": 2.582495865665946, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.825355052947998, + "num_tokens": 774617506.0, + "step": 20301 + }, + { + "epoch": 2.5826230759445363, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7766351699829102, + "num_tokens": 774653720.0, + "step": 20302 + }, + { + "epoch": 2.582750286223127, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8211770057678223, + "num_tokens": 774691056.0, + "step": 20303 + }, + { + "epoch": 2.5828774965017174, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7908563017845154, + "num_tokens": 774726514.0, + "step": 20304 + }, + { + "epoch": 2.583004706780308, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7884219884872437, + "num_tokens": 774760831.0, + "step": 20305 + }, + { + "epoch": 2.5831319170588984, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8101147413253784, + "num_tokens": 774799399.0, + "step": 20306 + }, + { + "epoch": 2.583259127337489, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7981171607971191, + "num_tokens": 774839910.0, + "step": 20307 + }, + { + "epoch": 2.5833863376160795, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8052363991737366, + "num_tokens": 774875694.0, + "step": 20308 + }, + { + "epoch": 2.58351354789467, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8004627227783203, + "num_tokens": 774908020.0, + "step": 20309 + }, + { + "epoch": 2.5836407581732606, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7942753434181213, + "num_tokens": 774946716.0, + "step": 20310 + }, + { + "epoch": 2.583767968451851, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7973902225494385, + "num_tokens": 774981732.0, + "step": 20311 + }, + { + "epoch": 2.583895178730441, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8071770668029785, + "num_tokens": 775015280.0, + "step": 20312 + }, + { + "epoch": 2.584022389009032, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8132218718528748, + "num_tokens": 775054518.0, + "step": 20313 + }, + { + "epoch": 2.5841495992876222, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.8075329065322876, + "num_tokens": 775091548.0, + "step": 20314 + }, + { + "epoch": 2.584276809566213, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7927393913269043, + "num_tokens": 775130916.0, + "step": 20315 + }, + { + "epoch": 2.5844040198448033, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8108022809028625, + "num_tokens": 775165875.0, + "step": 20316 + }, + { + "epoch": 2.584531230123394, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7901891469955444, + "num_tokens": 775211463.0, + "step": 20317 + }, + { + "epoch": 2.5846584404019843, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.814068078994751, + "num_tokens": 775248338.0, + "step": 20318 + }, + { + "epoch": 2.584785650680575, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7894094586372375, + "num_tokens": 775285707.0, + "step": 20319 + }, + { + "epoch": 2.5849128609591654, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8140735626220703, + "num_tokens": 775322682.0, + "step": 20320 + }, + { + "epoch": 2.585040071237756, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8125549554824829, + "num_tokens": 775358274.0, + "step": 20321 + }, + { + "epoch": 2.5851672815163464, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7922642230987549, + "num_tokens": 775394472.0, + "step": 20322 + }, + { + "epoch": 2.585294491794937, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8127901554107666, + "num_tokens": 775433647.0, + "step": 20323 + }, + { + "epoch": 2.5854217020735275, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7876278162002563, + "num_tokens": 775469159.0, + "step": 20324 + }, + { + "epoch": 2.585548912352118, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8106438517570496, + "num_tokens": 775507329.0, + "step": 20325 + }, + { + "epoch": 2.5856761226307086, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8021159768104553, + "num_tokens": 775544638.0, + "step": 20326 + }, + { + "epoch": 2.585803332909299, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8036537170410156, + "num_tokens": 775584918.0, + "step": 20327 + }, + { + "epoch": 2.5859305431878896, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7947149872779846, + "num_tokens": 775624214.0, + "step": 20328 + }, + { + "epoch": 2.58605775346648, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7905582189559937, + "num_tokens": 775668372.0, + "step": 20329 + }, + { + "epoch": 2.5861849637450707, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7709289789199829, + "num_tokens": 775700578.0, + "step": 20330 + }, + { + "epoch": 2.586312174023661, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8158963918685913, + "num_tokens": 775746980.0, + "step": 20331 + }, + { + "epoch": 2.5864393843022517, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.791654109954834, + "num_tokens": 775780762.0, + "step": 20332 + }, + { + "epoch": 2.5865665945808423, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8027210235595703, + "num_tokens": 775815591.0, + "step": 20333 + }, + { + "epoch": 2.586693804859433, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7943317890167236, + "num_tokens": 775856414.0, + "step": 20334 + }, + { + "epoch": 2.5868210151380233, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.8061672449111938, + "num_tokens": 775896939.0, + "step": 20335 + }, + { + "epoch": 2.586948225416614, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7952595949172974, + "num_tokens": 775941778.0, + "step": 20336 + }, + { + "epoch": 2.587075435695204, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8051246404647827, + "num_tokens": 775979346.0, + "step": 20337 + }, + { + "epoch": 2.587202645973795, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7970434427261353, + "num_tokens": 776013793.0, + "step": 20338 + }, + { + "epoch": 2.587329856252385, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8184177279472351, + "num_tokens": 776058513.0, + "step": 20339 + }, + { + "epoch": 2.587457066530976, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8047846555709839, + "num_tokens": 776102790.0, + "step": 20340 + }, + { + "epoch": 2.587584276809566, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8128246665000916, + "num_tokens": 776133898.0, + "step": 20341 + }, + { + "epoch": 2.5877114870881566, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8001748323440552, + "num_tokens": 776167590.0, + "step": 20342 + }, + { + "epoch": 2.587838697366747, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7980518341064453, + "num_tokens": 776207966.0, + "step": 20343 + }, + { + "epoch": 2.5879659076453376, + "grad_norm": 54.75, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8026522397994995, + "num_tokens": 776245601.0, + "step": 20344 + }, + { + "epoch": 2.588093117923928, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.819670557975769, + "num_tokens": 776285377.0, + "step": 20345 + }, + { + "epoch": 2.5882203282025187, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7976728081703186, + "num_tokens": 776328116.0, + "step": 20346 + }, + { + "epoch": 2.588347538481109, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7833553552627563, + "num_tokens": 776369008.0, + "step": 20347 + }, + { + "epoch": 2.5884747487596997, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7950262427330017, + "num_tokens": 776410151.0, + "step": 20348 + }, + { + "epoch": 2.5886019590382903, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.8255051970481873, + "num_tokens": 776445944.0, + "step": 20349 + }, + { + "epoch": 2.588729169316881, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.802631139755249, + "num_tokens": 776485388.0, + "step": 20350 + }, + { + "epoch": 2.5888563795954713, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7754150629043579, + "num_tokens": 776529135.0, + "step": 20351 + }, + { + "epoch": 2.588983589874062, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7910075187683105, + "num_tokens": 776564310.0, + "step": 20352 + }, + { + "epoch": 2.5891108001526524, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8200926780700684, + "num_tokens": 776609004.0, + "step": 20353 + }, + { + "epoch": 2.589238010431243, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8047769069671631, + "num_tokens": 776645796.0, + "step": 20354 + }, + { + "epoch": 2.5893652207098334, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7910444736480713, + "num_tokens": 776684248.0, + "step": 20355 + }, + { + "epoch": 2.589492430988424, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8058409690856934, + "num_tokens": 776722684.0, + "step": 20356 + }, + { + "epoch": 2.5896196412670145, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8048583269119263, + "num_tokens": 776764635.0, + "step": 20357 + }, + { + "epoch": 2.589746851545605, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7838248014450073, + "num_tokens": 776805683.0, + "step": 20358 + }, + { + "epoch": 2.5898740618241956, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8032191395759583, + "num_tokens": 776841229.0, + "step": 20359 + }, + { + "epoch": 2.5900012721027856, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8077586889266968, + "num_tokens": 776881319.0, + "step": 20360 + }, + { + "epoch": 2.5901284823813766, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8056808710098267, + "num_tokens": 776919436.0, + "step": 20361 + }, + { + "epoch": 2.5902556926599667, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7858259677886963, + "num_tokens": 776958644.0, + "step": 20362 + }, + { + "epoch": 2.5903829029385577, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.5954, + "mean_token_accuracy": 0.8191509246826172, + "num_tokens": 776994509.0, + "step": 20363 + }, + { + "epoch": 2.5905101132171477, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.806829571723938, + "num_tokens": 777037557.0, + "step": 20364 + }, + { + "epoch": 2.5906373234957387, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7826087474822998, + "num_tokens": 777077237.0, + "step": 20365 + }, + { + "epoch": 2.590764533774329, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8100680708885193, + "num_tokens": 777120813.0, + "step": 20366 + }, + { + "epoch": 2.5908917440529193, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7816324234008789, + "num_tokens": 777155809.0, + "step": 20367 + }, + { + "epoch": 2.59101895433151, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.804551899433136, + "num_tokens": 777195868.0, + "step": 20368 + }, + { + "epoch": 2.5911461646101004, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8034089803695679, + "num_tokens": 777233406.0, + "step": 20369 + }, + { + "epoch": 2.591273374888691, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.805365800857544, + "num_tokens": 777272817.0, + "step": 20370 + }, + { + "epoch": 2.5914005851672814, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8004260063171387, + "num_tokens": 777312398.0, + "step": 20371 + }, + { + "epoch": 2.591527795445872, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.8152869939804077, + "num_tokens": 777349375.0, + "step": 20372 + }, + { + "epoch": 2.5916550057244625, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8102728724479675, + "num_tokens": 777394806.0, + "step": 20373 + }, + { + "epoch": 2.591782216003053, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8111073970794678, + "num_tokens": 777434350.0, + "step": 20374 + }, + { + "epoch": 2.5919094262816436, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7735915780067444, + "num_tokens": 777469690.0, + "step": 20375 + }, + { + "epoch": 2.592036636560234, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7930800914764404, + "num_tokens": 777502095.0, + "step": 20376 + }, + { + "epoch": 2.5921638468388246, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8157387971878052, + "num_tokens": 777540387.0, + "step": 20377 + }, + { + "epoch": 2.592291057117415, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8133494853973389, + "num_tokens": 777581855.0, + "step": 20378 + }, + { + "epoch": 2.5924182673960057, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8017107248306274, + "num_tokens": 777619852.0, + "step": 20379 + }, + { + "epoch": 2.592545477674596, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7981916666030884, + "num_tokens": 777662772.0, + "step": 20380 + }, + { + "epoch": 2.5926726879531867, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7896122932434082, + "num_tokens": 777701171.0, + "step": 20381 + }, + { + "epoch": 2.5927998982317773, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7821639776229858, + "num_tokens": 777738245.0, + "step": 20382 + }, + { + "epoch": 2.592927108510368, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7988629341125488, + "num_tokens": 777775732.0, + "step": 20383 + }, + { + "epoch": 2.5930543187889583, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7979979515075684, + "num_tokens": 777811077.0, + "step": 20384 + }, + { + "epoch": 2.5931815290675484, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.80112224817276, + "num_tokens": 777854107.0, + "step": 20385 + }, + { + "epoch": 2.5933087393461394, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8024981021881104, + "num_tokens": 777891950.0, + "step": 20386 + }, + { + "epoch": 2.5934359496247295, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8199735879898071, + "num_tokens": 777932638.0, + "step": 20387 + }, + { + "epoch": 2.5935631599033204, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.803716778755188, + "num_tokens": 777977733.0, + "step": 20388 + }, + { + "epoch": 2.5936903701819105, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8056744337081909, + "num_tokens": 778020451.0, + "step": 20389 + }, + { + "epoch": 2.5938175804605015, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8085561990737915, + "num_tokens": 778059594.0, + "step": 20390 + }, + { + "epoch": 2.5939447907390916, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8077278137207031, + "num_tokens": 778099235.0, + "step": 20391 + }, + { + "epoch": 2.594072001017682, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8105812668800354, + "num_tokens": 778135430.0, + "step": 20392 + }, + { + "epoch": 2.5941992112962726, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7931763529777527, + "num_tokens": 778175111.0, + "step": 20393 + }, + { + "epoch": 2.594326421574863, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7913870811462402, + "num_tokens": 778211954.0, + "step": 20394 + }, + { + "epoch": 2.5944536318534537, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7870565056800842, + "num_tokens": 778246580.0, + "step": 20395 + }, + { + "epoch": 2.594580842132044, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.8054968118667603, + "num_tokens": 778288659.0, + "step": 20396 + }, + { + "epoch": 2.5947080524106347, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8005485534667969, + "num_tokens": 778325365.0, + "step": 20397 + }, + { + "epoch": 2.5948352626892253, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8065074682235718, + "num_tokens": 778362982.0, + "step": 20398 + }, + { + "epoch": 2.594962472967816, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7800776958465576, + "num_tokens": 778394230.0, + "step": 20399 + }, + { + "epoch": 2.5950896832464063, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7871866226196289, + "num_tokens": 778434345.0, + "step": 20400 + }, + { + "epoch": 2.595216893524997, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7955283522605896, + "num_tokens": 778478303.0, + "step": 20401 + }, + { + "epoch": 2.5953441038035874, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8094756603240967, + "num_tokens": 778513915.0, + "step": 20402 + }, + { + "epoch": 2.595471314082178, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5837, + "mean_token_accuracy": 0.8235946893692017, + "num_tokens": 778554294.0, + "step": 20403 + }, + { + "epoch": 2.5955985243607684, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7990860939025879, + "num_tokens": 778592836.0, + "step": 20404 + }, + { + "epoch": 2.595725734639359, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7940291166305542, + "num_tokens": 778629316.0, + "step": 20405 + }, + { + "epoch": 2.5958529449179495, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7890588045120239, + "num_tokens": 778659065.0, + "step": 20406 + }, + { + "epoch": 2.59598015519654, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8169260025024414, + "num_tokens": 778693949.0, + "step": 20407 + }, + { + "epoch": 2.5961073654751305, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8102483749389648, + "num_tokens": 778738045.0, + "step": 20408 + }, + { + "epoch": 2.596234575753721, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7485449910163879, + "num_tokens": 778769578.0, + "step": 20409 + }, + { + "epoch": 2.596361786032311, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7971941232681274, + "num_tokens": 778810052.0, + "step": 20410 + }, + { + "epoch": 2.596488996310902, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7954205274581909, + "num_tokens": 778850868.0, + "step": 20411 + }, + { + "epoch": 2.596616206589492, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8162083625793457, + "num_tokens": 778890129.0, + "step": 20412 + }, + { + "epoch": 2.596743416868083, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8057690858840942, + "num_tokens": 778923181.0, + "step": 20413 + }, + { + "epoch": 2.5968706271466733, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7949247360229492, + "num_tokens": 778963328.0, + "step": 20414 + }, + { + "epoch": 2.596997837425264, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8151887059211731, + "num_tokens": 779003932.0, + "step": 20415 + }, + { + "epoch": 2.5971250477038543, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7929161787033081, + "num_tokens": 779043913.0, + "step": 20416 + }, + { + "epoch": 2.597252257982445, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.791901707649231, + "num_tokens": 779077126.0, + "step": 20417 + }, + { + "epoch": 2.5973794682610354, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8051730394363403, + "num_tokens": 779111544.0, + "step": 20418 + }, + { + "epoch": 2.597506678539626, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8010912537574768, + "num_tokens": 779145493.0, + "step": 20419 + }, + { + "epoch": 2.5976338888182164, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.796210527420044, + "num_tokens": 779189216.0, + "step": 20420 + }, + { + "epoch": 2.597761099096807, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.806740403175354, + "num_tokens": 779225914.0, + "step": 20421 + }, + { + "epoch": 2.5978883093753975, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8098465204238892, + "num_tokens": 779264118.0, + "step": 20422 + }, + { + "epoch": 2.598015519653988, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8088878393173218, + "num_tokens": 779301694.0, + "step": 20423 + }, + { + "epoch": 2.5981427299325786, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.8029075860977173, + "num_tokens": 779339012.0, + "step": 20424 + }, + { + "epoch": 2.598269940211169, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8069378137588501, + "num_tokens": 779376358.0, + "step": 20425 + }, + { + "epoch": 2.5983971504897596, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.795476496219635, + "num_tokens": 779416344.0, + "step": 20426 + }, + { + "epoch": 2.59852436076835, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.785469651222229, + "num_tokens": 779452189.0, + "step": 20427 + }, + { + "epoch": 2.5986515710469407, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7996195554733276, + "num_tokens": 779493513.0, + "step": 20428 + }, + { + "epoch": 2.598778781325531, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8052636384963989, + "num_tokens": 779533061.0, + "step": 20429 + }, + { + "epoch": 2.5989059916041217, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7948634028434753, + "num_tokens": 779570159.0, + "step": 20430 + }, + { + "epoch": 2.5990332018827123, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7983009815216064, + "num_tokens": 779609066.0, + "step": 20431 + }, + { + "epoch": 2.599160412161303, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8036106824874878, + "num_tokens": 779651087.0, + "step": 20432 + }, + { + "epoch": 2.5992876224398933, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8146727085113525, + "num_tokens": 779689531.0, + "step": 20433 + }, + { + "epoch": 2.599414832718484, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8005170822143555, + "num_tokens": 779724062.0, + "step": 20434 + }, + { + "epoch": 2.599542042997074, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7850766181945801, + "num_tokens": 779766029.0, + "step": 20435 + }, + { + "epoch": 2.599669253275665, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7894591093063354, + "num_tokens": 779811065.0, + "step": 20436 + }, + { + "epoch": 2.599796463554255, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7850253582000732, + "num_tokens": 779852442.0, + "step": 20437 + }, + { + "epoch": 2.599923673832846, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8058280348777771, + "num_tokens": 779885993.0, + "step": 20438 + }, + { + "epoch": 2.600050884111436, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.804470419883728, + "num_tokens": 779927791.0, + "step": 20439 + }, + { + "epoch": 2.6001780943900266, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8050552010536194, + "num_tokens": 779961517.0, + "step": 20440 + }, + { + "epoch": 2.600305304668617, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7846872806549072, + "num_tokens": 779992902.0, + "step": 20441 + }, + { + "epoch": 2.6004325149472076, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.80301833152771, + "num_tokens": 780032849.0, + "step": 20442 + }, + { + "epoch": 2.600559725225798, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.811715841293335, + "num_tokens": 780072563.0, + "step": 20443 + }, + { + "epoch": 2.6006869355043887, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8068839311599731, + "num_tokens": 780115299.0, + "step": 20444 + }, + { + "epoch": 2.600814145782979, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8004575967788696, + "num_tokens": 780155792.0, + "step": 20445 + }, + { + "epoch": 2.6009413560615697, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7939678430557251, + "num_tokens": 780189040.0, + "step": 20446 + }, + { + "epoch": 2.6010685663401603, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.80290687084198, + "num_tokens": 780229463.0, + "step": 20447 + }, + { + "epoch": 2.601195776618751, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.8133137226104736, + "num_tokens": 780265961.0, + "step": 20448 + }, + { + "epoch": 2.6013229868973413, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7922013998031616, + "num_tokens": 780303220.0, + "step": 20449 + }, + { + "epoch": 2.601450197175932, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7963907718658447, + "num_tokens": 780340456.0, + "step": 20450 + }, + { + "epoch": 2.6015774074545224, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7975345849990845, + "num_tokens": 780381402.0, + "step": 20451 + }, + { + "epoch": 2.601704617733113, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8107113838195801, + "num_tokens": 780422571.0, + "step": 20452 + }, + { + "epoch": 2.6018318280117034, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8009864091873169, + "num_tokens": 780462447.0, + "step": 20453 + }, + { + "epoch": 2.601959038290294, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8026353120803833, + "num_tokens": 780499183.0, + "step": 20454 + }, + { + "epoch": 2.6020862485688845, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7859296798706055, + "num_tokens": 780539216.0, + "step": 20455 + }, + { + "epoch": 2.602213458847475, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8098705410957336, + "num_tokens": 780578725.0, + "step": 20456 + }, + { + "epoch": 2.6023406691260655, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7977830767631531, + "num_tokens": 780615175.0, + "step": 20457 + }, + { + "epoch": 2.6024678794046556, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8026890754699707, + "num_tokens": 780651186.0, + "step": 20458 + }, + { + "epoch": 2.6025950896832466, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7963268160820007, + "num_tokens": 780685550.0, + "step": 20459 + }, + { + "epoch": 2.6027222999618367, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.5597, + "mean_token_accuracy": 0.8267942667007446, + "num_tokens": 780725477.0, + "step": 20460 + }, + { + "epoch": 2.6028495102404277, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8027921319007874, + "num_tokens": 780762481.0, + "step": 20461 + }, + { + "epoch": 2.6029767205190177, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.785619854927063, + "num_tokens": 780800305.0, + "step": 20462 + }, + { + "epoch": 2.6031039307976087, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7996392250061035, + "num_tokens": 780836271.0, + "step": 20463 + }, + { + "epoch": 2.603231141076199, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.8101233243942261, + "num_tokens": 780877363.0, + "step": 20464 + }, + { + "epoch": 2.6033583513547893, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7833585739135742, + "num_tokens": 780922982.0, + "step": 20465 + }, + { + "epoch": 2.60348556163338, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.791743278503418, + "num_tokens": 780957925.0, + "step": 20466 + }, + { + "epoch": 2.6036127719119704, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.779320240020752, + "num_tokens": 780996922.0, + "step": 20467 + }, + { + "epoch": 2.603739982190561, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7812484502792358, + "num_tokens": 781037209.0, + "step": 20468 + }, + { + "epoch": 2.6038671924691514, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7924717664718628, + "num_tokens": 781078077.0, + "step": 20469 + }, + { + "epoch": 2.603994402747742, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8022915720939636, + "num_tokens": 781122516.0, + "step": 20470 + }, + { + "epoch": 2.6041216130263325, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7877687215805054, + "num_tokens": 781156290.0, + "step": 20471 + }, + { + "epoch": 2.604248823304923, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.803234875202179, + "num_tokens": 781192769.0, + "step": 20472 + }, + { + "epoch": 2.6043760335835135, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8060660362243652, + "num_tokens": 781227594.0, + "step": 20473 + }, + { + "epoch": 2.604503243862104, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8096652030944824, + "num_tokens": 781259369.0, + "step": 20474 + }, + { + "epoch": 2.6046304541406946, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.797385036945343, + "num_tokens": 781296788.0, + "step": 20475 + }, + { + "epoch": 2.604757664419285, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.803066611289978, + "num_tokens": 781333103.0, + "step": 20476 + }, + { + "epoch": 2.6048848746978757, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8119332790374756, + "num_tokens": 781372691.0, + "step": 20477 + }, + { + "epoch": 2.605012084976466, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7971397042274475, + "num_tokens": 781410480.0, + "step": 20478 + }, + { + "epoch": 2.6051392952550567, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7931901216506958, + "num_tokens": 781454110.0, + "step": 20479 + }, + { + "epoch": 2.6052665055336472, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7982900142669678, + "num_tokens": 781494034.0, + "step": 20480 + }, + { + "epoch": 2.6053937158122378, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.804380476474762, + "num_tokens": 781536225.0, + "step": 20481 + }, + { + "epoch": 2.6055209260908283, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8067520260810852, + "num_tokens": 781573987.0, + "step": 20482 + }, + { + "epoch": 2.6056481363694184, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8036384582519531, + "num_tokens": 781608987.0, + "step": 20483 + }, + { + "epoch": 2.6057753466480094, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8024190068244934, + "num_tokens": 781646185.0, + "step": 20484 + }, + { + "epoch": 2.6059025569265994, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7864656448364258, + "num_tokens": 781682999.0, + "step": 20485 + }, + { + "epoch": 2.6060297672051904, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.798312783241272, + "num_tokens": 781726052.0, + "step": 20486 + }, + { + "epoch": 2.6061569774837805, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8107718825340271, + "num_tokens": 781760562.0, + "step": 20487 + }, + { + "epoch": 2.6062841877623715, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7893911600112915, + "num_tokens": 781803765.0, + "step": 20488 + }, + { + "epoch": 2.6064113980409616, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8032079339027405, + "num_tokens": 781845419.0, + "step": 20489 + }, + { + "epoch": 2.606538608319552, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8116012811660767, + "num_tokens": 781883371.0, + "step": 20490 + }, + { + "epoch": 2.6066658185981426, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7970030307769775, + "num_tokens": 781919669.0, + "step": 20491 + }, + { + "epoch": 2.606793028876733, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8047695755958557, + "num_tokens": 781953212.0, + "step": 20492 + }, + { + "epoch": 2.6069202391553237, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8073580861091614, + "num_tokens": 781996824.0, + "step": 20493 + }, + { + "epoch": 2.607047449433914, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.5771, + "mean_token_accuracy": 0.8230067491531372, + "num_tokens": 782035350.0, + "step": 20494 + }, + { + "epoch": 2.6071746597125047, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7930982708930969, + "num_tokens": 782073217.0, + "step": 20495 + }, + { + "epoch": 2.6073018699910953, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7885005474090576, + "num_tokens": 782110777.0, + "step": 20496 + }, + { + "epoch": 2.607429080269686, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7964297533035278, + "num_tokens": 782153300.0, + "step": 20497 + }, + { + "epoch": 2.6075562905482763, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8104616403579712, + "num_tokens": 782191945.0, + "step": 20498 + }, + { + "epoch": 2.607683500826867, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8100314140319824, + "num_tokens": 782230472.0, + "step": 20499 + }, + { + "epoch": 2.6078107111054574, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7876120805740356, + "num_tokens": 782263868.0, + "step": 20500 + }, + { + "epoch": 2.607937921384048, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8052304983139038, + "num_tokens": 782299781.0, + "step": 20501 + }, + { + "epoch": 2.6080651316626384, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7978149652481079, + "num_tokens": 782343376.0, + "step": 20502 + }, + { + "epoch": 2.608192341941229, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8121649622917175, + "num_tokens": 782389894.0, + "step": 20503 + }, + { + "epoch": 2.6083195522198195, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8007098436355591, + "num_tokens": 782430055.0, + "step": 20504 + }, + { + "epoch": 2.60844676249841, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.8011526465415955, + "num_tokens": 782467211.0, + "step": 20505 + }, + { + "epoch": 2.6085739727770005, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7939700484275818, + "num_tokens": 782505913.0, + "step": 20506 + }, + { + "epoch": 2.608701183055591, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8030768036842346, + "num_tokens": 782541858.0, + "step": 20507 + }, + { + "epoch": 2.608828393334181, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7956882119178772, + "num_tokens": 782575854.0, + "step": 20508 + }, + { + "epoch": 2.608955603612772, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8035894632339478, + "num_tokens": 782614005.0, + "step": 20509 + }, + { + "epoch": 2.609082813891362, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8083890676498413, + "num_tokens": 782650147.0, + "step": 20510 + }, + { + "epoch": 2.609210024169953, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.8006551265716553, + "num_tokens": 782683215.0, + "step": 20511 + }, + { + "epoch": 2.6093372344485433, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.8064700365066528, + "num_tokens": 782721184.0, + "step": 20512 + }, + { + "epoch": 2.609464444727134, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.796883225440979, + "num_tokens": 782759891.0, + "step": 20513 + }, + { + "epoch": 2.6095916550057243, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7898849248886108, + "num_tokens": 782796398.0, + "step": 20514 + }, + { + "epoch": 2.609718865284315, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.810395359992981, + "num_tokens": 782835968.0, + "step": 20515 + }, + { + "epoch": 2.6098460755629054, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.811930775642395, + "num_tokens": 782871367.0, + "step": 20516 + }, + { + "epoch": 2.609973285841496, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.8237957954406738, + "num_tokens": 782905766.0, + "step": 20517 + }, + { + "epoch": 2.6101004961200864, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8083419799804688, + "num_tokens": 782942256.0, + "step": 20518 + }, + { + "epoch": 2.610227706398677, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8106968402862549, + "num_tokens": 782976156.0, + "step": 20519 + }, + { + "epoch": 2.6103549166772675, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.8134890198707581, + "num_tokens": 783011450.0, + "step": 20520 + }, + { + "epoch": 2.610482126955858, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8042163252830505, + "num_tokens": 783050355.0, + "step": 20521 + }, + { + "epoch": 2.6106093372344485, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8086363673210144, + "num_tokens": 783091621.0, + "step": 20522 + }, + { + "epoch": 2.610736547513039, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7924588918685913, + "num_tokens": 783130471.0, + "step": 20523 + }, + { + "epoch": 2.6108637577916296, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8041698932647705, + "num_tokens": 783168340.0, + "step": 20524 + }, + { + "epoch": 2.61099096807022, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7790964245796204, + "num_tokens": 783207279.0, + "step": 20525 + }, + { + "epoch": 2.6111181783488107, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7943360805511475, + "num_tokens": 783243010.0, + "step": 20526 + }, + { + "epoch": 2.611245388627401, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7922482490539551, + "num_tokens": 783287548.0, + "step": 20527 + }, + { + "epoch": 2.6113725989059917, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8022228479385376, + "num_tokens": 783331549.0, + "step": 20528 + }, + { + "epoch": 2.6114998091845822, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8155592679977417, + "num_tokens": 783369445.0, + "step": 20529 + }, + { + "epoch": 2.6116270194631728, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7946747541427612, + "num_tokens": 783413218.0, + "step": 20530 + }, + { + "epoch": 2.6117542297417633, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8003719449043274, + "num_tokens": 783449944.0, + "step": 20531 + }, + { + "epoch": 2.611881440020354, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8047904968261719, + "num_tokens": 783486142.0, + "step": 20532 + }, + { + "epoch": 2.612008650298944, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8056827187538147, + "num_tokens": 783521170.0, + "step": 20533 + }, + { + "epoch": 2.612135860577535, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8024164438247681, + "num_tokens": 783563902.0, + "step": 20534 + }, + { + "epoch": 2.612263070856125, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8064473867416382, + "num_tokens": 783597655.0, + "step": 20535 + }, + { + "epoch": 2.612390281134716, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.789443850517273, + "num_tokens": 783635233.0, + "step": 20536 + }, + { + "epoch": 2.612517491413306, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8000006675720215, + "num_tokens": 783679677.0, + "step": 20537 + }, + { + "epoch": 2.6126447016918966, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8079322576522827, + "num_tokens": 783720608.0, + "step": 20538 + }, + { + "epoch": 2.612771911970487, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8078752160072327, + "num_tokens": 783753198.0, + "step": 20539 + }, + { + "epoch": 2.6128991222490776, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8004558086395264, + "num_tokens": 783791068.0, + "step": 20540 + }, + { + "epoch": 2.613026332527668, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8043426871299744, + "num_tokens": 783831319.0, + "step": 20541 + }, + { + "epoch": 2.6131535428062587, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8084065318107605, + "num_tokens": 783867537.0, + "step": 20542 + }, + { + "epoch": 2.613280753084849, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.795907735824585, + "num_tokens": 783907787.0, + "step": 20543 + }, + { + "epoch": 2.6134079633634397, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7935006618499756, + "num_tokens": 783948558.0, + "step": 20544 + }, + { + "epoch": 2.6135351736420303, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8230528831481934, + "num_tokens": 783986726.0, + "step": 20545 + }, + { + "epoch": 2.613662383920621, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7814087867736816, + "num_tokens": 784023449.0, + "step": 20546 + }, + { + "epoch": 2.6137895941992113, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.8200551867485046, + "num_tokens": 784059529.0, + "step": 20547 + }, + { + "epoch": 2.613916804477802, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8066024780273438, + "num_tokens": 784103530.0, + "step": 20548 + }, + { + "epoch": 2.6140440147563924, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7863538265228271, + "num_tokens": 784141721.0, + "step": 20549 + }, + { + "epoch": 2.614171225034983, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8013991713523865, + "num_tokens": 784175868.0, + "step": 20550 + }, + { + "epoch": 2.6142984353135734, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.809510350227356, + "num_tokens": 784210646.0, + "step": 20551 + }, + { + "epoch": 2.614425645592164, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8096796274185181, + "num_tokens": 784255588.0, + "step": 20552 + }, + { + "epoch": 2.6145528558707545, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7990634441375732, + "num_tokens": 784292877.0, + "step": 20553 + }, + { + "epoch": 2.614680066149345, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8145356178283691, + "num_tokens": 784336150.0, + "step": 20554 + }, + { + "epoch": 2.6148072764279355, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8137884140014648, + "num_tokens": 784371547.0, + "step": 20555 + }, + { + "epoch": 2.6149344867065256, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8060594797134399, + "num_tokens": 784411473.0, + "step": 20556 + }, + { + "epoch": 2.6150616969851166, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7971134185791016, + "num_tokens": 784456713.0, + "step": 20557 + }, + { + "epoch": 2.6151889072637067, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8196106553077698, + "num_tokens": 784493685.0, + "step": 20558 + }, + { + "epoch": 2.6153161175422976, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.8162331581115723, + "num_tokens": 784534641.0, + "step": 20559 + }, + { + "epoch": 2.6154433278208877, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7770833969116211, + "num_tokens": 784576184.0, + "step": 20560 + }, + { + "epoch": 2.6155705380994787, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8038370013237, + "num_tokens": 784613851.0, + "step": 20561 + }, + { + "epoch": 2.615697748378069, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8072658777236938, + "num_tokens": 784654133.0, + "step": 20562 + }, + { + "epoch": 2.6158249586566593, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8049088716506958, + "num_tokens": 784690529.0, + "step": 20563 + }, + { + "epoch": 2.61595216893525, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8023812174797058, + "num_tokens": 784731972.0, + "step": 20564 + }, + { + "epoch": 2.6160793792138404, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8024477958679199, + "num_tokens": 784773275.0, + "step": 20565 + }, + { + "epoch": 2.616206589492431, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7834310531616211, + "num_tokens": 784811468.0, + "step": 20566 + }, + { + "epoch": 2.6163337997710214, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8001748323440552, + "num_tokens": 784850611.0, + "step": 20567 + }, + { + "epoch": 2.616461010049612, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8071393966674805, + "num_tokens": 784886502.0, + "step": 20568 + }, + { + "epoch": 2.6165882203282025, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7919042110443115, + "num_tokens": 784924489.0, + "step": 20569 + }, + { + "epoch": 2.616715430606793, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8026464581489563, + "num_tokens": 784957224.0, + "step": 20570 + }, + { + "epoch": 2.6168426408853835, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7744337320327759, + "num_tokens": 784997754.0, + "step": 20571 + }, + { + "epoch": 2.616969851163974, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7846328020095825, + "num_tokens": 785033932.0, + "step": 20572 + }, + { + "epoch": 2.6170970614425646, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8025195598602295, + "num_tokens": 785073190.0, + "step": 20573 + }, + { + "epoch": 2.617224271721155, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8166804909706116, + "num_tokens": 785108397.0, + "step": 20574 + }, + { + "epoch": 2.6173514819997457, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.8274669647216797, + "num_tokens": 785142671.0, + "step": 20575 + }, + { + "epoch": 2.617478692278336, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7978146076202393, + "num_tokens": 785183467.0, + "step": 20576 + }, + { + "epoch": 2.6176059025569267, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7935121059417725, + "num_tokens": 785222072.0, + "step": 20577 + }, + { + "epoch": 2.6177331128355172, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7926474809646606, + "num_tokens": 785260458.0, + "step": 20578 + }, + { + "epoch": 2.6178603231141078, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.791242241859436, + "num_tokens": 785302434.0, + "step": 20579 + }, + { + "epoch": 2.6179875333926983, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7829841375350952, + "num_tokens": 785338251.0, + "step": 20580 + }, + { + "epoch": 2.6181147436712884, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8020426034927368, + "num_tokens": 785384789.0, + "step": 20581 + }, + { + "epoch": 2.6182419539498794, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8004885911941528, + "num_tokens": 785424978.0, + "step": 20582 + }, + { + "epoch": 2.6183691642284694, + "grad_norm": 63.25, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7976456880569458, + "num_tokens": 785464421.0, + "step": 20583 + }, + { + "epoch": 2.6184963745070604, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8209817409515381, + "num_tokens": 785496166.0, + "step": 20584 + }, + { + "epoch": 2.6186235847856505, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.782057523727417, + "num_tokens": 785528504.0, + "step": 20585 + }, + { + "epoch": 2.6187507950642415, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.8028982877731323, + "num_tokens": 785570896.0, + "step": 20586 + }, + { + "epoch": 2.6188780053428315, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.797201931476593, + "num_tokens": 785612464.0, + "step": 20587 + }, + { + "epoch": 2.619005215621422, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.800919771194458, + "num_tokens": 785655500.0, + "step": 20588 + }, + { + "epoch": 2.6191324259000126, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7957611680030823, + "num_tokens": 785691186.0, + "step": 20589 + }, + { + "epoch": 2.619259636178603, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8186116218566895, + "num_tokens": 785728575.0, + "step": 20590 + }, + { + "epoch": 2.6193868464571937, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7959465980529785, + "num_tokens": 785765782.0, + "step": 20591 + }, + { + "epoch": 2.619514056735784, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.80662602186203, + "num_tokens": 785800130.0, + "step": 20592 + }, + { + "epoch": 2.6196412670143747, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.5454, + "mean_token_accuracy": 0.8328471183776855, + "num_tokens": 785836647.0, + "step": 20593 + }, + { + "epoch": 2.6197684772929652, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7910163402557373, + "num_tokens": 785873897.0, + "step": 20594 + }, + { + "epoch": 2.6198956875715558, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7980023622512817, + "num_tokens": 785914296.0, + "step": 20595 + }, + { + "epoch": 2.6200228978501463, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7840958833694458, + "num_tokens": 785951426.0, + "step": 20596 + }, + { + "epoch": 2.620150108128737, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8041733503341675, + "num_tokens": 785989906.0, + "step": 20597 + }, + { + "epoch": 2.6202773184073274, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7924381494522095, + "num_tokens": 786024763.0, + "step": 20598 + }, + { + "epoch": 2.620404528685918, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7909510731697083, + "num_tokens": 786063055.0, + "step": 20599 + }, + { + "epoch": 2.6205317389645084, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7737482786178589, + "num_tokens": 786097675.0, + "step": 20600 + }, + { + "epoch": 2.620658949243099, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8121384382247925, + "num_tokens": 786133721.0, + "step": 20601 + }, + { + "epoch": 2.6207861595216895, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7829010486602783, + "num_tokens": 786179615.0, + "step": 20602 + }, + { + "epoch": 2.62091336980028, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8213012218475342, + "num_tokens": 786214945.0, + "step": 20603 + }, + { + "epoch": 2.6210405800788705, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.808974027633667, + "num_tokens": 786250880.0, + "step": 20604 + }, + { + "epoch": 2.621167790357461, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8049896955490112, + "num_tokens": 786286095.0, + "step": 20605 + }, + { + "epoch": 2.621295000636051, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8072277307510376, + "num_tokens": 786319547.0, + "step": 20606 + }, + { + "epoch": 2.621422210914642, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8002788424491882, + "num_tokens": 786358699.0, + "step": 20607 + }, + { + "epoch": 2.621549421193232, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8004499077796936, + "num_tokens": 786399732.0, + "step": 20608 + }, + { + "epoch": 2.621676631471823, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7820825576782227, + "num_tokens": 786437375.0, + "step": 20609 + }, + { + "epoch": 2.6218038417504133, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7869740724563599, + "num_tokens": 786475538.0, + "step": 20610 + }, + { + "epoch": 2.621931052029004, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8072012662887573, + "num_tokens": 786516802.0, + "step": 20611 + }, + { + "epoch": 2.6220582623075943, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8060914278030396, + "num_tokens": 786558470.0, + "step": 20612 + }, + { + "epoch": 2.622185472586185, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7986729145050049, + "num_tokens": 786601163.0, + "step": 20613 + }, + { + "epoch": 2.6223126828647754, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.8070787191390991, + "num_tokens": 786636741.0, + "step": 20614 + }, + { + "epoch": 2.622439893143366, + "grad_norm": 55.25, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7860453724861145, + "num_tokens": 786672931.0, + "step": 20615 + }, + { + "epoch": 2.6225671034219564, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.8148653507232666, + "num_tokens": 786709310.0, + "step": 20616 + }, + { + "epoch": 2.622694313700547, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8103370666503906, + "num_tokens": 786747113.0, + "step": 20617 + }, + { + "epoch": 2.6228215239791375, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.80133056640625, + "num_tokens": 786782033.0, + "step": 20618 + }, + { + "epoch": 2.622948734257728, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8033708930015564, + "num_tokens": 786827229.0, + "step": 20619 + }, + { + "epoch": 2.6230759445363185, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8103634715080261, + "num_tokens": 786861722.0, + "step": 20620 + }, + { + "epoch": 2.623203154814909, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8056572675704956, + "num_tokens": 786896202.0, + "step": 20621 + }, + { + "epoch": 2.6233303650934996, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8028440475463867, + "num_tokens": 786932154.0, + "step": 20622 + }, + { + "epoch": 2.62345757537209, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7934046983718872, + "num_tokens": 786972184.0, + "step": 20623 + }, + { + "epoch": 2.6235847856506807, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.8000984191894531, + "num_tokens": 787015514.0, + "step": 20624 + }, + { + "epoch": 2.623711995929271, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.797366738319397, + "num_tokens": 787054023.0, + "step": 20625 + }, + { + "epoch": 2.6238392062078617, + "grad_norm": 58.0, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7999067902565002, + "num_tokens": 787088858.0, + "step": 20626 + }, + { + "epoch": 2.6239664164864522, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8038375973701477, + "num_tokens": 787124805.0, + "step": 20627 + }, + { + "epoch": 2.6240936267650428, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8198378086090088, + "num_tokens": 787164922.0, + "step": 20628 + }, + { + "epoch": 2.6242208370436333, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8141095638275146, + "num_tokens": 787204936.0, + "step": 20629 + }, + { + "epoch": 2.624348047322224, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.811831533908844, + "num_tokens": 787238292.0, + "step": 20630 + }, + { + "epoch": 2.624475257600814, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5831, + "mean_token_accuracy": 0.8194130063056946, + "num_tokens": 787269054.0, + "step": 20631 + }, + { + "epoch": 2.624602467879405, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.8005073666572571, + "num_tokens": 787310022.0, + "step": 20632 + }, + { + "epoch": 2.624729678157995, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7899492383003235, + "num_tokens": 787350917.0, + "step": 20633 + }, + { + "epoch": 2.624856888436586, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8089607954025269, + "num_tokens": 787388187.0, + "step": 20634 + }, + { + "epoch": 2.624984098715176, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8073892593383789, + "num_tokens": 787424998.0, + "step": 20635 + }, + { + "epoch": 2.6251113089937665, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8117966651916504, + "num_tokens": 787465456.0, + "step": 20636 + }, + { + "epoch": 2.625238519272357, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7973561882972717, + "num_tokens": 787504797.0, + "step": 20637 + }, + { + "epoch": 2.6253657295509476, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8169378042221069, + "num_tokens": 787542911.0, + "step": 20638 + }, + { + "epoch": 2.625492939829538, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7881215810775757, + "num_tokens": 787576143.0, + "step": 20639 + }, + { + "epoch": 2.6256201501081287, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8074003458023071, + "num_tokens": 787619912.0, + "step": 20640 + }, + { + "epoch": 2.625747360386719, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.787948727607727, + "num_tokens": 787658181.0, + "step": 20641 + }, + { + "epoch": 2.6258745706653097, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7962923049926758, + "num_tokens": 787694889.0, + "step": 20642 + }, + { + "epoch": 2.6260017809439002, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.8224121332168579, + "num_tokens": 787737633.0, + "step": 20643 + }, + { + "epoch": 2.6261289912224908, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7908104658126831, + "num_tokens": 787779069.0, + "step": 20644 + }, + { + "epoch": 2.6262562015010813, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7849462032318115, + "num_tokens": 787815544.0, + "step": 20645 + }, + { + "epoch": 2.626383411779672, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7936514019966125, + "num_tokens": 787852684.0, + "step": 20646 + }, + { + "epoch": 2.6265106220582624, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.797020673751831, + "num_tokens": 787893452.0, + "step": 20647 + }, + { + "epoch": 2.626637832336853, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7987964153289795, + "num_tokens": 787929430.0, + "step": 20648 + }, + { + "epoch": 2.6267650426154434, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7989863157272339, + "num_tokens": 787969768.0, + "step": 20649 + }, + { + "epoch": 2.626892252894034, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8021431565284729, + "num_tokens": 788009605.0, + "step": 20650 + }, + { + "epoch": 2.6270194631726245, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.816162645816803, + "num_tokens": 788044652.0, + "step": 20651 + }, + { + "epoch": 2.627146673451215, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7687499523162842, + "num_tokens": 788085981.0, + "step": 20652 + }, + { + "epoch": 2.6272738837298055, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7997626066207886, + "num_tokens": 788132196.0, + "step": 20653 + }, + { + "epoch": 2.6274010940083956, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.8095709085464478, + "num_tokens": 788160099.0, + "step": 20654 + }, + { + "epoch": 2.6275283042869866, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7971255779266357, + "num_tokens": 788197264.0, + "step": 20655 + }, + { + "epoch": 2.6276555145655767, + "grad_norm": 52.5, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8058185577392578, + "num_tokens": 788232900.0, + "step": 20656 + }, + { + "epoch": 2.6277827248441676, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7901172041893005, + "num_tokens": 788273720.0, + "step": 20657 + }, + { + "epoch": 2.6279099351227577, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8061313629150391, + "num_tokens": 788308425.0, + "step": 20658 + }, + { + "epoch": 2.6280371454013487, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7932088971138, + "num_tokens": 788343701.0, + "step": 20659 + }, + { + "epoch": 2.628164355679939, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7859306335449219, + "num_tokens": 788372789.0, + "step": 20660 + }, + { + "epoch": 2.6282915659585293, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.8072682619094849, + "num_tokens": 788408422.0, + "step": 20661 + }, + { + "epoch": 2.62841877623712, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.8037451505661011, + "num_tokens": 788446113.0, + "step": 20662 + }, + { + "epoch": 2.6285459865157104, + "grad_norm": 76.5, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7971608638763428, + "num_tokens": 788485009.0, + "step": 20663 + }, + { + "epoch": 2.628673196794301, + "grad_norm": 80.0, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8046900033950806, + "num_tokens": 788523446.0, + "step": 20664 + }, + { + "epoch": 2.6288004070728914, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8097391724586487, + "num_tokens": 788557423.0, + "step": 20665 + }, + { + "epoch": 2.628927617351482, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7902898788452148, + "num_tokens": 788588177.0, + "step": 20666 + }, + { + "epoch": 2.6290548276300725, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8033162355422974, + "num_tokens": 788624877.0, + "step": 20667 + }, + { + "epoch": 2.629182037908663, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8091040253639221, + "num_tokens": 788662532.0, + "step": 20668 + }, + { + "epoch": 2.6293092481872535, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8042491674423218, + "num_tokens": 788702734.0, + "step": 20669 + }, + { + "epoch": 2.629436458465844, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7947131395339966, + "num_tokens": 788738041.0, + "step": 20670 + }, + { + "epoch": 2.6295636687444346, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8001636266708374, + "num_tokens": 788772998.0, + "step": 20671 + }, + { + "epoch": 2.629690879023025, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7852811813354492, + "num_tokens": 788814310.0, + "step": 20672 + }, + { + "epoch": 2.6298180893016156, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8094691038131714, + "num_tokens": 788852811.0, + "step": 20673 + }, + { + "epoch": 2.629945299580206, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.801970362663269, + "num_tokens": 788888756.0, + "step": 20674 + }, + { + "epoch": 2.6300725098587967, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8088944554328918, + "num_tokens": 788926292.0, + "step": 20675 + }, + { + "epoch": 2.6301997201373872, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8050640821456909, + "num_tokens": 788965625.0, + "step": 20676 + }, + { + "epoch": 2.6303269304159778, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8112903833389282, + "num_tokens": 789007390.0, + "step": 20677 + }, + { + "epoch": 2.6304541406945683, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8089580535888672, + "num_tokens": 789047940.0, + "step": 20678 + }, + { + "epoch": 2.6305813509731584, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8035824298858643, + "num_tokens": 789086321.0, + "step": 20679 + }, + { + "epoch": 2.6307085612517493, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.817020058631897, + "num_tokens": 789121757.0, + "step": 20680 + }, + { + "epoch": 2.6308357715303394, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.8163906335830688, + "num_tokens": 789160076.0, + "step": 20681 + }, + { + "epoch": 2.6309629818089304, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7935709953308105, + "num_tokens": 789197080.0, + "step": 20682 + }, + { + "epoch": 2.6310901920875205, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.798087477684021, + "num_tokens": 789249876.0, + "step": 20683 + }, + { + "epoch": 2.6312174023661115, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7964262366294861, + "num_tokens": 789290146.0, + "step": 20684 + }, + { + "epoch": 2.6313446126447015, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.807315468788147, + "num_tokens": 789335610.0, + "step": 20685 + }, + { + "epoch": 2.631471822923292, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7984174489974976, + "num_tokens": 789371956.0, + "step": 20686 + }, + { + "epoch": 2.6315990332018826, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8010943531990051, + "num_tokens": 789415257.0, + "step": 20687 + }, + { + "epoch": 2.631726243480473, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8022709488868713, + "num_tokens": 789452533.0, + "step": 20688 + }, + { + "epoch": 2.6318534537590637, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7892572283744812, + "num_tokens": 789488780.0, + "step": 20689 + }, + { + "epoch": 2.631980664037654, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7982321381568909, + "num_tokens": 789528169.0, + "step": 20690 + }, + { + "epoch": 2.6321078743162447, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7942264676094055, + "num_tokens": 789568382.0, + "step": 20691 + }, + { + "epoch": 2.6322350845948352, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7983452081680298, + "num_tokens": 789606895.0, + "step": 20692 + }, + { + "epoch": 2.6323622948734258, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7859500646591187, + "num_tokens": 789649106.0, + "step": 20693 + }, + { + "epoch": 2.6324895051520163, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8073728084564209, + "num_tokens": 789689449.0, + "step": 20694 + }, + { + "epoch": 2.632616715430607, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.812499463558197, + "num_tokens": 789726179.0, + "step": 20695 + }, + { + "epoch": 2.6327439257091974, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7827936410903931, + "num_tokens": 789764858.0, + "step": 20696 + }, + { + "epoch": 2.632871135987788, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7868008613586426, + "num_tokens": 789799196.0, + "step": 20697 + }, + { + "epoch": 2.6329983462663784, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.801665723323822, + "num_tokens": 789840498.0, + "step": 20698 + }, + { + "epoch": 2.633125556544969, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8052800297737122, + "num_tokens": 789874269.0, + "step": 20699 + }, + { + "epoch": 2.6332527668235595, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8067531585693359, + "num_tokens": 789914938.0, + "step": 20700 + }, + { + "epoch": 2.63337997710215, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.8014882802963257, + "num_tokens": 789959219.0, + "step": 20701 + }, + { + "epoch": 2.6335071873807405, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7947924137115479, + "num_tokens": 789993707.0, + "step": 20702 + }, + { + "epoch": 2.633634397659331, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7821383476257324, + "num_tokens": 790034285.0, + "step": 20703 + }, + { + "epoch": 2.633761607937921, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.80253005027771, + "num_tokens": 790069714.0, + "step": 20704 + }, + { + "epoch": 2.633888818216512, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8027794361114502, + "num_tokens": 790110733.0, + "step": 20705 + }, + { + "epoch": 2.634016028495102, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8077784776687622, + "num_tokens": 790144468.0, + "step": 20706 + }, + { + "epoch": 2.634143238773693, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7977524399757385, + "num_tokens": 790186881.0, + "step": 20707 + }, + { + "epoch": 2.6342704490522832, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8031214475631714, + "num_tokens": 790225083.0, + "step": 20708 + }, + { + "epoch": 2.6343976593308738, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8000479340553284, + "num_tokens": 790266822.0, + "step": 20709 + }, + { + "epoch": 2.6345248696094643, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8038273453712463, + "num_tokens": 790305948.0, + "step": 20710 + }, + { + "epoch": 2.634652079888055, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8043686151504517, + "num_tokens": 790341639.0, + "step": 20711 + }, + { + "epoch": 2.6347792901666454, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8118288516998291, + "num_tokens": 790377442.0, + "step": 20712 + }, + { + "epoch": 2.634906500445236, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.799838662147522, + "num_tokens": 790414911.0, + "step": 20713 + }, + { + "epoch": 2.6350337107238264, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.8005452752113342, + "num_tokens": 790453968.0, + "step": 20714 + }, + { + "epoch": 2.635160921002417, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7975360751152039, + "num_tokens": 790493216.0, + "step": 20715 + }, + { + "epoch": 2.6352881312810075, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7968209385871887, + "num_tokens": 790529380.0, + "step": 20716 + }, + { + "epoch": 2.635415341559598, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7942892909049988, + "num_tokens": 790569568.0, + "step": 20717 + }, + { + "epoch": 2.6355425518381885, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8040530681610107, + "num_tokens": 790602269.0, + "step": 20718 + }, + { + "epoch": 2.635669762116779, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.787885308265686, + "num_tokens": 790639723.0, + "step": 20719 + }, + { + "epoch": 2.6357969723953696, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7877527475357056, + "num_tokens": 790676103.0, + "step": 20720 + }, + { + "epoch": 2.63592418267396, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8050336837768555, + "num_tokens": 790712118.0, + "step": 20721 + }, + { + "epoch": 2.6360513929525506, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7964770793914795, + "num_tokens": 790747079.0, + "step": 20722 + }, + { + "epoch": 2.636178603231141, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7864425182342529, + "num_tokens": 790781741.0, + "step": 20723 + }, + { + "epoch": 2.6363058135097317, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8016257286071777, + "num_tokens": 790820243.0, + "step": 20724 + }, + { + "epoch": 2.6364330237883222, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8145855665206909, + "num_tokens": 790851492.0, + "step": 20725 + }, + { + "epoch": 2.6365602340669128, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8115732669830322, + "num_tokens": 790882293.0, + "step": 20726 + }, + { + "epoch": 2.636687444345503, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.794795036315918, + "num_tokens": 790922188.0, + "step": 20727 + }, + { + "epoch": 2.636814654624094, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8101584911346436, + "num_tokens": 790960675.0, + "step": 20728 + }, + { + "epoch": 2.636941864902684, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8091030120849609, + "num_tokens": 790999352.0, + "step": 20729 + }, + { + "epoch": 2.637069075181275, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8046847581863403, + "num_tokens": 791035898.0, + "step": 20730 + }, + { + "epoch": 2.637196285459865, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7774571180343628, + "num_tokens": 791079731.0, + "step": 20731 + }, + { + "epoch": 2.637323495738456, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.8147574067115784, + "num_tokens": 791118565.0, + "step": 20732 + }, + { + "epoch": 2.637450706017046, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.8001800179481506, + "num_tokens": 791154876.0, + "step": 20733 + }, + { + "epoch": 2.6375779162956365, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.796633780002594, + "num_tokens": 791196003.0, + "step": 20734 + }, + { + "epoch": 2.637705126574227, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7977868914604187, + "num_tokens": 791236462.0, + "step": 20735 + }, + { + "epoch": 2.6378323368528176, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7842739820480347, + "num_tokens": 791271918.0, + "step": 20736 + }, + { + "epoch": 2.637959547131408, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8100261688232422, + "num_tokens": 791309813.0, + "step": 20737 + }, + { + "epoch": 2.6380867574099987, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8093314170837402, + "num_tokens": 791338854.0, + "step": 20738 + }, + { + "epoch": 2.638213967688589, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.811124324798584, + "num_tokens": 791376580.0, + "step": 20739 + }, + { + "epoch": 2.6383411779671797, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7962021827697754, + "num_tokens": 791415931.0, + "step": 20740 + }, + { + "epoch": 2.6384683882457702, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7923603057861328, + "num_tokens": 791448201.0, + "step": 20741 + }, + { + "epoch": 2.6385955985243608, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7889268398284912, + "num_tokens": 791489183.0, + "step": 20742 + }, + { + "epoch": 2.6387228088029513, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7978376150131226, + "num_tokens": 791528414.0, + "step": 20743 + }, + { + "epoch": 2.638850019081542, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8211902976036072, + "num_tokens": 791566379.0, + "step": 20744 + }, + { + "epoch": 2.6389772293601323, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.797289252281189, + "num_tokens": 791599789.0, + "step": 20745 + }, + { + "epoch": 2.639104439638723, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7859481573104858, + "num_tokens": 791642333.0, + "step": 20746 + }, + { + "epoch": 2.6392316499173134, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8050581216812134, + "num_tokens": 791687190.0, + "step": 20747 + }, + { + "epoch": 2.639358860195904, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7838036417961121, + "num_tokens": 791722806.0, + "step": 20748 + }, + { + "epoch": 2.6394860704744945, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7865499258041382, + "num_tokens": 791757726.0, + "step": 20749 + }, + { + "epoch": 2.639613280753085, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.8070050477981567, + "num_tokens": 791795160.0, + "step": 20750 + }, + { + "epoch": 2.6397404910316755, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8147268295288086, + "num_tokens": 791831682.0, + "step": 20751 + }, + { + "epoch": 2.6398677013102656, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.8065405488014221, + "num_tokens": 791863230.0, + "step": 20752 + }, + { + "epoch": 2.6399949115888566, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8168842792510986, + "num_tokens": 791904323.0, + "step": 20753 + }, + { + "epoch": 2.6401221218674467, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7958318591117859, + "num_tokens": 791951836.0, + "step": 20754 + }, + { + "epoch": 2.6402493321460376, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8035415410995483, + "num_tokens": 791989782.0, + "step": 20755 + }, + { + "epoch": 2.6403765424246277, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.8012547492980957, + "num_tokens": 792024038.0, + "step": 20756 + }, + { + "epoch": 2.6405037527032187, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8031347393989563, + "num_tokens": 792067893.0, + "step": 20757 + }, + { + "epoch": 2.6406309629818088, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8106203079223633, + "num_tokens": 792108593.0, + "step": 20758 + }, + { + "epoch": 2.6407581732603993, + "grad_norm": 55.0, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8115271925926208, + "num_tokens": 792141201.0, + "step": 20759 + }, + { + "epoch": 2.64088538353899, + "grad_norm": 10.8125, + "learning_rate": 1e-06, + "loss": 0.5777, + "mean_token_accuracy": 0.8254810571670532, + "num_tokens": 792182908.0, + "step": 20760 + }, + { + "epoch": 2.6410125938175804, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.804192066192627, + "num_tokens": 792225223.0, + "step": 20761 + }, + { + "epoch": 2.641139804096171, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7811026573181152, + "num_tokens": 792258131.0, + "step": 20762 + }, + { + "epoch": 2.6412670143747614, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8011029362678528, + "num_tokens": 792291429.0, + "step": 20763 + }, + { + "epoch": 2.641394224653352, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7947633266448975, + "num_tokens": 792327777.0, + "step": 20764 + }, + { + "epoch": 2.6415214349319425, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8084703087806702, + "num_tokens": 792367918.0, + "step": 20765 + }, + { + "epoch": 2.641648645210533, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8043026328086853, + "num_tokens": 792405550.0, + "step": 20766 + }, + { + "epoch": 2.6417758554891235, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8015175461769104, + "num_tokens": 792440948.0, + "step": 20767 + }, + { + "epoch": 2.641903065767714, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8111132383346558, + "num_tokens": 792488065.0, + "step": 20768 + }, + { + "epoch": 2.6420302760463046, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.8060678243637085, + "num_tokens": 792522845.0, + "step": 20769 + }, + { + "epoch": 2.642157486324895, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8121804594993591, + "num_tokens": 792561118.0, + "step": 20770 + }, + { + "epoch": 2.6422846966034856, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.8041219115257263, + "num_tokens": 792600125.0, + "step": 20771 + }, + { + "epoch": 2.642411906882076, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8025157451629639, + "num_tokens": 792633856.0, + "step": 20772 + }, + { + "epoch": 2.6425391171606667, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.795384407043457, + "num_tokens": 792672427.0, + "step": 20773 + }, + { + "epoch": 2.6426663274392572, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7758083343505859, + "num_tokens": 792713856.0, + "step": 20774 + }, + { + "epoch": 2.6427935377178478, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7942031025886536, + "num_tokens": 792757602.0, + "step": 20775 + }, + { + "epoch": 2.6429207479964383, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7904175519943237, + "num_tokens": 792797067.0, + "step": 20776 + }, + { + "epoch": 2.6430479582750284, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8064929246902466, + "num_tokens": 792835748.0, + "step": 20777 + }, + { + "epoch": 2.6431751685536193, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.8040212392807007, + "num_tokens": 792875039.0, + "step": 20778 + }, + { + "epoch": 2.6433023788322094, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8079984188079834, + "num_tokens": 792916580.0, + "step": 20779 + }, + { + "epoch": 2.6434295891108004, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.8046860694885254, + "num_tokens": 792952507.0, + "step": 20780 + }, + { + "epoch": 2.6435567993893905, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8022298812866211, + "num_tokens": 792994046.0, + "step": 20781 + }, + { + "epoch": 2.6436840096679814, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.8231073021888733, + "num_tokens": 793033855.0, + "step": 20782 + }, + { + "epoch": 2.6438112199465715, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7995160818099976, + "num_tokens": 793072226.0, + "step": 20783 + }, + { + "epoch": 2.643938430225162, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8081142902374268, + "num_tokens": 793110884.0, + "step": 20784 + }, + { + "epoch": 2.6440656405037526, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7845135927200317, + "num_tokens": 793151885.0, + "step": 20785 + }, + { + "epoch": 2.644192850782343, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7745251655578613, + "num_tokens": 793187253.0, + "step": 20786 + }, + { + "epoch": 2.6443200610609336, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.805522084236145, + "num_tokens": 793231249.0, + "step": 20787 + }, + { + "epoch": 2.644447271339524, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8168987035751343, + "num_tokens": 793269918.0, + "step": 20788 + }, + { + "epoch": 2.6445744816181147, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.816856861114502, + "num_tokens": 793306856.0, + "step": 20789 + }, + { + "epoch": 2.6447016918967052, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7915207743644714, + "num_tokens": 793347866.0, + "step": 20790 + }, + { + "epoch": 2.6448289021752958, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8036561608314514, + "num_tokens": 793391037.0, + "step": 20791 + }, + { + "epoch": 2.6449561124538863, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7841473817825317, + "num_tokens": 793423560.0, + "step": 20792 + }, + { + "epoch": 2.645083322732477, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.8016549348831177, + "num_tokens": 793460215.0, + "step": 20793 + }, + { + "epoch": 2.6452105330110673, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7828415632247925, + "num_tokens": 793492671.0, + "step": 20794 + }, + { + "epoch": 2.645337743289658, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.801391065120697, + "num_tokens": 793529526.0, + "step": 20795 + }, + { + "epoch": 2.6454649535682484, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.8104634881019592, + "num_tokens": 793566748.0, + "step": 20796 + }, + { + "epoch": 2.645592163846839, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8163673281669617, + "num_tokens": 793600703.0, + "step": 20797 + }, + { + "epoch": 2.6457193741254295, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7988730072975159, + "num_tokens": 793639226.0, + "step": 20798 + }, + { + "epoch": 2.64584658440402, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8130121231079102, + "num_tokens": 793676147.0, + "step": 20799 + }, + { + "epoch": 2.6459737946826105, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8137711882591248, + "num_tokens": 793715778.0, + "step": 20800 + }, + { + "epoch": 2.646101004961201, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8173791170120239, + "num_tokens": 793752096.0, + "step": 20801 + }, + { + "epoch": 2.646228215239791, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7953770160675049, + "num_tokens": 793789809.0, + "step": 20802 + }, + { + "epoch": 2.646355425518382, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7990880012512207, + "num_tokens": 793826505.0, + "step": 20803 + }, + { + "epoch": 2.646482635796972, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.801888108253479, + "num_tokens": 793868360.0, + "step": 20804 + }, + { + "epoch": 2.646609846075563, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7961656451225281, + "num_tokens": 793906566.0, + "step": 20805 + }, + { + "epoch": 2.6467370563541532, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7932188510894775, + "num_tokens": 793947814.0, + "step": 20806 + }, + { + "epoch": 2.6468642666327438, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.806762158870697, + "num_tokens": 793987391.0, + "step": 20807 + }, + { + "epoch": 2.6469914769113343, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8148353099822998, + "num_tokens": 794024045.0, + "step": 20808 + }, + { + "epoch": 2.647118687189925, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8098008632659912, + "num_tokens": 794063405.0, + "step": 20809 + }, + { + "epoch": 2.6472458974685154, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7750899791717529, + "num_tokens": 794099729.0, + "step": 20810 + }, + { + "epoch": 2.647373107747106, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.791056215763092, + "num_tokens": 794139669.0, + "step": 20811 + }, + { + "epoch": 2.6475003180256964, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8046566247940063, + "num_tokens": 794176906.0, + "step": 20812 + }, + { + "epoch": 2.647627528304287, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8060330152511597, + "num_tokens": 794216594.0, + "step": 20813 + }, + { + "epoch": 2.6477547385828775, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7978748083114624, + "num_tokens": 794248379.0, + "step": 20814 + }, + { + "epoch": 2.647881948861468, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7758833765983582, + "num_tokens": 794293604.0, + "step": 20815 + }, + { + "epoch": 2.6480091591400585, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7972098588943481, + "num_tokens": 794332084.0, + "step": 20816 + }, + { + "epoch": 2.648136369418649, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.8222061991691589, + "num_tokens": 794369158.0, + "step": 20817 + }, + { + "epoch": 2.6482635796972396, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8035199046134949, + "num_tokens": 794410531.0, + "step": 20818 + }, + { + "epoch": 2.64839078997583, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8053194880485535, + "num_tokens": 794457556.0, + "step": 20819 + }, + { + "epoch": 2.6485180002544206, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7939960360527039, + "num_tokens": 794492737.0, + "step": 20820 + }, + { + "epoch": 2.648645210533011, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8141369819641113, + "num_tokens": 794532589.0, + "step": 20821 + }, + { + "epoch": 2.6487724208116017, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7859785556793213, + "num_tokens": 794566982.0, + "step": 20822 + }, + { + "epoch": 2.648899631090192, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7871416807174683, + "num_tokens": 794604278.0, + "step": 20823 + }, + { + "epoch": 2.6490268413687827, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8132665753364563, + "num_tokens": 794643785.0, + "step": 20824 + }, + { + "epoch": 2.649154051647373, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8068629503250122, + "num_tokens": 794678302.0, + "step": 20825 + }, + { + "epoch": 2.649281261925964, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7958147525787354, + "num_tokens": 794715732.0, + "step": 20826 + }, + { + "epoch": 2.649408472204554, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.8054094910621643, + "num_tokens": 794754359.0, + "step": 20827 + }, + { + "epoch": 2.649535682483145, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.8002622127532959, + "num_tokens": 794795770.0, + "step": 20828 + }, + { + "epoch": 2.649662892761735, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.801895797252655, + "num_tokens": 794835654.0, + "step": 20829 + }, + { + "epoch": 2.649790103040326, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.801113486289978, + "num_tokens": 794872527.0, + "step": 20830 + }, + { + "epoch": 2.649917313318916, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8097208738327026, + "num_tokens": 794918577.0, + "step": 20831 + }, + { + "epoch": 2.6500445235975065, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.8082135915756226, + "num_tokens": 794953566.0, + "step": 20832 + }, + { + "epoch": 2.650171733876097, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8097069263458252, + "num_tokens": 794990547.0, + "step": 20833 + }, + { + "epoch": 2.6502989441546876, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8172675371170044, + "num_tokens": 795028419.0, + "step": 20834 + }, + { + "epoch": 2.650426154433278, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.5837, + "mean_token_accuracy": 0.824309229850769, + "num_tokens": 795064303.0, + "step": 20835 + }, + { + "epoch": 2.6505533647118686, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7762268781661987, + "num_tokens": 795104280.0, + "step": 20836 + }, + { + "epoch": 2.650680574990459, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7916160225868225, + "num_tokens": 795142547.0, + "step": 20837 + }, + { + "epoch": 2.6508077852690497, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.802753210067749, + "num_tokens": 795186193.0, + "step": 20838 + }, + { + "epoch": 2.6509349955476402, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8173477053642273, + "num_tokens": 795215453.0, + "step": 20839 + }, + { + "epoch": 2.6510622058262308, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7993846535682678, + "num_tokens": 795252820.0, + "step": 20840 + }, + { + "epoch": 2.6511894161048213, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7968955636024475, + "num_tokens": 795290000.0, + "step": 20841 + }, + { + "epoch": 2.651316626383412, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.8238358497619629, + "num_tokens": 795330389.0, + "step": 20842 + }, + { + "epoch": 2.6514438366620023, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7978097200393677, + "num_tokens": 795366693.0, + "step": 20843 + }, + { + "epoch": 2.651571046940593, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.802710771560669, + "num_tokens": 795405748.0, + "step": 20844 + }, + { + "epoch": 2.6516982572191834, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8133745193481445, + "num_tokens": 795446179.0, + "step": 20845 + }, + { + "epoch": 2.651825467497774, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7951749563217163, + "num_tokens": 795488234.0, + "step": 20846 + }, + { + "epoch": 2.6519526777763645, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7933359146118164, + "num_tokens": 795526508.0, + "step": 20847 + }, + { + "epoch": 2.652079888054955, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8148301839828491, + "num_tokens": 795566299.0, + "step": 20848 + }, + { + "epoch": 2.6522070983335455, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8005473017692566, + "num_tokens": 795603892.0, + "step": 20849 + }, + { + "epoch": 2.6523343086121356, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8088630437850952, + "num_tokens": 795639622.0, + "step": 20850 + }, + { + "epoch": 2.6524615188907266, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.5556, + "mean_token_accuracy": 0.8282580375671387, + "num_tokens": 795677299.0, + "step": 20851 + }, + { + "epoch": 2.6525887291693167, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8013992309570312, + "num_tokens": 795714304.0, + "step": 20852 + }, + { + "epoch": 2.6527159394479076, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7977663278579712, + "num_tokens": 795752330.0, + "step": 20853 + }, + { + "epoch": 2.6528431497264977, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7774935960769653, + "num_tokens": 795789503.0, + "step": 20854 + }, + { + "epoch": 2.6529703600050887, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7989078760147095, + "num_tokens": 795828449.0, + "step": 20855 + }, + { + "epoch": 2.6530975702836788, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.803041934967041, + "num_tokens": 795868454.0, + "step": 20856 + }, + { + "epoch": 2.6532247805622693, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7854459285736084, + "num_tokens": 795904629.0, + "step": 20857 + }, + { + "epoch": 2.65335199084086, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7926986813545227, + "num_tokens": 795943197.0, + "step": 20858 + }, + { + "epoch": 2.6534792011194503, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.8002579212188721, + "num_tokens": 795984672.0, + "step": 20859 + }, + { + "epoch": 2.653606411398041, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7827706336975098, + "num_tokens": 796022599.0, + "step": 20860 + }, + { + "epoch": 2.6537336216766314, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7969391345977783, + "num_tokens": 796057798.0, + "step": 20861 + }, + { + "epoch": 2.653860831955222, + "grad_norm": 11.5625, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8039443492889404, + "num_tokens": 796094928.0, + "step": 20862 + }, + { + "epoch": 2.6539880422338125, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8027932047843933, + "num_tokens": 796132614.0, + "step": 20863 + }, + { + "epoch": 2.654115252512403, + "grad_norm": 10.25, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7980958819389343, + "num_tokens": 796171653.0, + "step": 20864 + }, + { + "epoch": 2.6542424627909935, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7972577810287476, + "num_tokens": 796215218.0, + "step": 20865 + }, + { + "epoch": 2.654369673069584, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8128840923309326, + "num_tokens": 796249165.0, + "step": 20866 + }, + { + "epoch": 2.6544968833481746, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7987408638000488, + "num_tokens": 796283878.0, + "step": 20867 + }, + { + "epoch": 2.654624093626765, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7945868968963623, + "num_tokens": 796330760.0, + "step": 20868 + }, + { + "epoch": 2.6547513039053556, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.786720335483551, + "num_tokens": 796371547.0, + "step": 20869 + }, + { + "epoch": 2.654878514183946, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7885768413543701, + "num_tokens": 796409893.0, + "step": 20870 + }, + { + "epoch": 2.6550057244625367, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7827247381210327, + "num_tokens": 796442493.0, + "step": 20871 + }, + { + "epoch": 2.655132934741127, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.804397463798523, + "num_tokens": 796480853.0, + "step": 20872 + }, + { + "epoch": 2.6552601450197177, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8080283999443054, + "num_tokens": 796520057.0, + "step": 20873 + }, + { + "epoch": 2.6553873552983083, + "grad_norm": 91.5, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7959878444671631, + "num_tokens": 796549779.0, + "step": 20874 + }, + { + "epoch": 2.6555145655768984, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8031425476074219, + "num_tokens": 796590923.0, + "step": 20875 + }, + { + "epoch": 2.6556417758554893, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8005799055099487, + "num_tokens": 796629455.0, + "step": 20876 + }, + { + "epoch": 2.6557689861340794, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7940582633018494, + "num_tokens": 796669950.0, + "step": 20877 + }, + { + "epoch": 2.6558961964126704, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7879059314727783, + "num_tokens": 796712263.0, + "step": 20878 + }, + { + "epoch": 2.6560234066912605, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8041333556175232, + "num_tokens": 796753025.0, + "step": 20879 + }, + { + "epoch": 2.6561506169698514, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8141958713531494, + "num_tokens": 796788810.0, + "step": 20880 + }, + { + "epoch": 2.6562778272484415, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.8095345497131348, + "num_tokens": 796826275.0, + "step": 20881 + }, + { + "epoch": 2.656405037527032, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7948988080024719, + "num_tokens": 796859587.0, + "step": 20882 + }, + { + "epoch": 2.6565322478056226, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8100639581680298, + "num_tokens": 796899440.0, + "step": 20883 + }, + { + "epoch": 2.656659458084213, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7929288744926453, + "num_tokens": 796938703.0, + "step": 20884 + }, + { + "epoch": 2.6567866683628036, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7905837297439575, + "num_tokens": 796982287.0, + "step": 20885 + }, + { + "epoch": 2.656913878641394, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7995875477790833, + "num_tokens": 797017640.0, + "step": 20886 + }, + { + "epoch": 2.6570410889199847, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7779388427734375, + "num_tokens": 797054050.0, + "step": 20887 + }, + { + "epoch": 2.6571682991985752, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.804497241973877, + "num_tokens": 797092429.0, + "step": 20888 + }, + { + "epoch": 2.6572955094771658, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7975572347640991, + "num_tokens": 797125646.0, + "step": 20889 + }, + { + "epoch": 2.6574227197557563, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8083510398864746, + "num_tokens": 797164560.0, + "step": 20890 + }, + { + "epoch": 2.657549930034347, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.79390949010849, + "num_tokens": 797194168.0, + "step": 20891 + }, + { + "epoch": 2.6576771403129373, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.576, + "mean_token_accuracy": 0.8253787755966187, + "num_tokens": 797227045.0, + "step": 20892 + }, + { + "epoch": 2.657804350591528, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8166302442550659, + "num_tokens": 797267607.0, + "step": 20893 + }, + { + "epoch": 2.6579315608701184, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7938361167907715, + "num_tokens": 797309320.0, + "step": 20894 + }, + { + "epoch": 2.658058771148709, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7960227727890015, + "num_tokens": 797348258.0, + "step": 20895 + }, + { + "epoch": 2.6581859814272994, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.8043219447135925, + "num_tokens": 797391563.0, + "step": 20896 + }, + { + "epoch": 2.65831319170589, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8145028948783875, + "num_tokens": 797428314.0, + "step": 20897 + }, + { + "epoch": 2.6584404019844805, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7758664488792419, + "num_tokens": 797464480.0, + "step": 20898 + }, + { + "epoch": 2.658567612263071, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.802602231502533, + "num_tokens": 797504135.0, + "step": 20899 + }, + { + "epoch": 2.658694822541661, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8071624040603638, + "num_tokens": 797544222.0, + "step": 20900 + }, + { + "epoch": 2.658822032820252, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.808110237121582, + "num_tokens": 797583695.0, + "step": 20901 + }, + { + "epoch": 2.658949243098842, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7914652228355408, + "num_tokens": 797625660.0, + "step": 20902 + }, + { + "epoch": 2.659076453377433, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8019609451293945, + "num_tokens": 797665807.0, + "step": 20903 + }, + { + "epoch": 2.6592036636560232, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7819437980651855, + "num_tokens": 797707935.0, + "step": 20904 + }, + { + "epoch": 2.6593308739346138, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7925489544868469, + "num_tokens": 797745502.0, + "step": 20905 + }, + { + "epoch": 2.6594580842132043, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8018835783004761, + "num_tokens": 797779412.0, + "step": 20906 + }, + { + "epoch": 2.659585294491795, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.8027595281600952, + "num_tokens": 797816677.0, + "step": 20907 + }, + { + "epoch": 2.6597125047703853, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7802263498306274, + "num_tokens": 797854260.0, + "step": 20908 + }, + { + "epoch": 2.659839715048976, + "grad_norm": 168.0, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7623202800750732, + "num_tokens": 797892349.0, + "step": 20909 + }, + { + "epoch": 2.6599669253275664, + "grad_norm": 60.5, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8046984672546387, + "num_tokens": 797928028.0, + "step": 20910 + }, + { + "epoch": 2.660094135606157, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7794954180717468, + "num_tokens": 797963885.0, + "step": 20911 + }, + { + "epoch": 2.6602213458847475, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7863214015960693, + "num_tokens": 798009842.0, + "step": 20912 + }, + { + "epoch": 2.660348556163338, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8101742267608643, + "num_tokens": 798043207.0, + "step": 20913 + }, + { + "epoch": 2.6604757664419285, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8166602849960327, + "num_tokens": 798080887.0, + "step": 20914 + }, + { + "epoch": 2.660602976720519, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8094364404678345, + "num_tokens": 798126393.0, + "step": 20915 + }, + { + "epoch": 2.6607301869991096, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8151779174804688, + "num_tokens": 798166099.0, + "step": 20916 + }, + { + "epoch": 2.6608573972777, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.791002631187439, + "num_tokens": 798202285.0, + "step": 20917 + }, + { + "epoch": 2.6609846075562906, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8098548054695129, + "num_tokens": 798243463.0, + "step": 20918 + }, + { + "epoch": 2.661111817834881, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8027827143669128, + "num_tokens": 798275507.0, + "step": 20919 + }, + { + "epoch": 2.6612390281134717, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.819551944732666, + "num_tokens": 798310536.0, + "step": 20920 + }, + { + "epoch": 2.661366238392062, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7973077297210693, + "num_tokens": 798347874.0, + "step": 20921 + }, + { + "epoch": 2.6614934486706527, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.5808, + "mean_token_accuracy": 0.8213964700698853, + "num_tokens": 798381865.0, + "step": 20922 + }, + { + "epoch": 2.661620658949243, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7999860048294067, + "num_tokens": 798423378.0, + "step": 20923 + }, + { + "epoch": 2.661747869227834, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8114485740661621, + "num_tokens": 798460278.0, + "step": 20924 + }, + { + "epoch": 2.661875079506424, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7982388734817505, + "num_tokens": 798496622.0, + "step": 20925 + }, + { + "epoch": 2.662002289785015, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7956899404525757, + "num_tokens": 798533061.0, + "step": 20926 + }, + { + "epoch": 2.662129500063605, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7982497811317444, + "num_tokens": 798573689.0, + "step": 20927 + }, + { + "epoch": 2.662256710342196, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8091191053390503, + "num_tokens": 798614058.0, + "step": 20928 + }, + { + "epoch": 2.662383920620786, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7888239622116089, + "num_tokens": 798657769.0, + "step": 20929 + }, + { + "epoch": 2.6625111308993765, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8017356395721436, + "num_tokens": 798701084.0, + "step": 20930 + }, + { + "epoch": 2.662638341177967, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8062105178833008, + "num_tokens": 798738029.0, + "step": 20931 + }, + { + "epoch": 2.6627655514565576, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.799547553062439, + "num_tokens": 798773537.0, + "step": 20932 + }, + { + "epoch": 2.662892761735148, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7954626679420471, + "num_tokens": 798810397.0, + "step": 20933 + }, + { + "epoch": 2.6630199720137386, + "grad_norm": 10.6875, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7973180413246155, + "num_tokens": 798850796.0, + "step": 20934 + }, + { + "epoch": 2.663147182292329, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8017845749855042, + "num_tokens": 798892379.0, + "step": 20935 + }, + { + "epoch": 2.6632743925709197, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7855205535888672, + "num_tokens": 798929444.0, + "step": 20936 + }, + { + "epoch": 2.66340160284951, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.8032903075218201, + "num_tokens": 798964086.0, + "step": 20937 + }, + { + "epoch": 2.6635288131281007, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8066542148590088, + "num_tokens": 799007408.0, + "step": 20938 + }, + { + "epoch": 2.6636560234066913, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8069676160812378, + "num_tokens": 799045930.0, + "step": 20939 + }, + { + "epoch": 2.663783233685282, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.8130403161048889, + "num_tokens": 799083568.0, + "step": 20940 + }, + { + "epoch": 2.6639104439638723, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.8011100888252258, + "num_tokens": 799123681.0, + "step": 20941 + }, + { + "epoch": 2.664037654242463, + "grad_norm": 53.75, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.798289954662323, + "num_tokens": 799164623.0, + "step": 20942 + }, + { + "epoch": 2.6641648645210534, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7982223033905029, + "num_tokens": 799200408.0, + "step": 20943 + }, + { + "epoch": 2.664292074799644, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7916720509529114, + "num_tokens": 799245106.0, + "step": 20944 + }, + { + "epoch": 2.6644192850782344, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.8106169700622559, + "num_tokens": 799290429.0, + "step": 20945 + }, + { + "epoch": 2.664546495356825, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7975221872329712, + "num_tokens": 799322951.0, + "step": 20946 + }, + { + "epoch": 2.6646737056354155, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7951302528381348, + "num_tokens": 799365498.0, + "step": 20947 + }, + { + "epoch": 2.6648009159140056, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7981072664260864, + "num_tokens": 799405894.0, + "step": 20948 + }, + { + "epoch": 2.6649281261925966, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.791905403137207, + "num_tokens": 799439750.0, + "step": 20949 + }, + { + "epoch": 2.6650553364711866, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7861595153808594, + "num_tokens": 799475170.0, + "step": 20950 + }, + { + "epoch": 2.6651825467497776, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7862570881843567, + "num_tokens": 799514231.0, + "step": 20951 + }, + { + "epoch": 2.6653097570283677, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.799609899520874, + "num_tokens": 799549512.0, + "step": 20952 + }, + { + "epoch": 2.6654369673069587, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.797688364982605, + "num_tokens": 799592144.0, + "step": 20953 + }, + { + "epoch": 2.6655641775855488, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7871767282485962, + "num_tokens": 799634779.0, + "step": 20954 + }, + { + "epoch": 2.6656913878641393, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8150225877761841, + "num_tokens": 799677548.0, + "step": 20955 + }, + { + "epoch": 2.66581859814273, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7800095081329346, + "num_tokens": 799715679.0, + "step": 20956 + }, + { + "epoch": 2.6659458084213203, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7957291603088379, + "num_tokens": 799755701.0, + "step": 20957 + }, + { + "epoch": 2.666073018699911, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7967872023582458, + "num_tokens": 799795151.0, + "step": 20958 + }, + { + "epoch": 2.6662002289785014, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8035357594490051, + "num_tokens": 799832249.0, + "step": 20959 + }, + { + "epoch": 2.666327439257092, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7730190753936768, + "num_tokens": 799867762.0, + "step": 20960 + }, + { + "epoch": 2.6664546495356825, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.8057941198348999, + "num_tokens": 799909012.0, + "step": 20961 + }, + { + "epoch": 2.666581859814273, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7990415692329407, + "num_tokens": 799950316.0, + "step": 20962 + }, + { + "epoch": 2.6667090700928635, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8164417743682861, + "num_tokens": 799990037.0, + "step": 20963 + }, + { + "epoch": 2.666836280371454, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8079705238342285, + "num_tokens": 800023935.0, + "step": 20964 + }, + { + "epoch": 2.6669634906500446, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8087437152862549, + "num_tokens": 800066357.0, + "step": 20965 + }, + { + "epoch": 2.667090700928635, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7776616215705872, + "num_tokens": 800100667.0, + "step": 20966 + }, + { + "epoch": 2.6672179112072256, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8040196895599365, + "num_tokens": 800135101.0, + "step": 20967 + }, + { + "epoch": 2.667345121485816, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.8171511292457581, + "num_tokens": 800176308.0, + "step": 20968 + }, + { + "epoch": 2.6674723317644067, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8084999918937683, + "num_tokens": 800205585.0, + "step": 20969 + }, + { + "epoch": 2.667599542042997, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8011998534202576, + "num_tokens": 800241400.0, + "step": 20970 + }, + { + "epoch": 2.6677267523215877, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8096150159835815, + "num_tokens": 800279060.0, + "step": 20971 + }, + { + "epoch": 2.6678539626001783, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8103888034820557, + "num_tokens": 800318606.0, + "step": 20972 + }, + { + "epoch": 2.6679811728787683, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8005156517028809, + "num_tokens": 800356795.0, + "step": 20973 + }, + { + "epoch": 2.6681083831573593, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8147773742675781, + "num_tokens": 800393786.0, + "step": 20974 + }, + { + "epoch": 2.6682355934359494, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.8000332117080688, + "num_tokens": 800440153.0, + "step": 20975 + }, + { + "epoch": 2.6683628037145404, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7796044945716858, + "num_tokens": 800476343.0, + "step": 20976 + }, + { + "epoch": 2.6684900139931305, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8163203597068787, + "num_tokens": 800514227.0, + "step": 20977 + }, + { + "epoch": 2.668617224271721, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8005790710449219, + "num_tokens": 800552357.0, + "step": 20978 + }, + { + "epoch": 2.6687444345503115, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8095874190330505, + "num_tokens": 800587624.0, + "step": 20979 + }, + { + "epoch": 2.668871644828902, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.795528769493103, + "num_tokens": 800631183.0, + "step": 20980 + }, + { + "epoch": 2.6689988551074926, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8033640384674072, + "num_tokens": 800670334.0, + "step": 20981 + }, + { + "epoch": 2.669126065386083, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7952595949172974, + "num_tokens": 800703500.0, + "step": 20982 + }, + { + "epoch": 2.6692532756646736, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7947539687156677, + "num_tokens": 800742904.0, + "step": 20983 + }, + { + "epoch": 2.669380485943264, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7972182035446167, + "num_tokens": 800777556.0, + "step": 20984 + }, + { + "epoch": 2.6695076962218547, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.789974570274353, + "num_tokens": 800817532.0, + "step": 20985 + }, + { + "epoch": 2.669634906500445, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8096832036972046, + "num_tokens": 800856567.0, + "step": 20986 + }, + { + "epoch": 2.6697621167790357, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.775331974029541, + "num_tokens": 800901234.0, + "step": 20987 + }, + { + "epoch": 2.6698893270576263, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.8008919954299927, + "num_tokens": 800942647.0, + "step": 20988 + }, + { + "epoch": 2.670016537336217, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8112649917602539, + "num_tokens": 800980194.0, + "step": 20989 + }, + { + "epoch": 2.6701437476148073, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8067363500595093, + "num_tokens": 801016978.0, + "step": 20990 + }, + { + "epoch": 2.670270957893398, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7825555801391602, + "num_tokens": 801051921.0, + "step": 20991 + }, + { + "epoch": 2.6703981681719884, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7718174457550049, + "num_tokens": 801087189.0, + "step": 20992 + }, + { + "epoch": 2.670525378450579, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8053469657897949, + "num_tokens": 801121574.0, + "step": 20993 + }, + { + "epoch": 2.6706525887291694, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8100442886352539, + "num_tokens": 801160497.0, + "step": 20994 + }, + { + "epoch": 2.67077979900776, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8020533323287964, + "num_tokens": 801198556.0, + "step": 20995 + }, + { + "epoch": 2.6709070092863505, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5561, + "mean_token_accuracy": 0.8264566659927368, + "num_tokens": 801230616.0, + "step": 20996 + }, + { + "epoch": 2.671034219564941, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7991490960121155, + "num_tokens": 801266582.0, + "step": 20997 + }, + { + "epoch": 2.671161429843531, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.812446117401123, + "num_tokens": 801305127.0, + "step": 20998 + }, + { + "epoch": 2.671288640122122, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.8059716820716858, + "num_tokens": 801348284.0, + "step": 20999 + }, + { + "epoch": 2.671415850400712, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8189937472343445, + "num_tokens": 801385298.0, + "step": 21000 + }, + { + "epoch": 2.671543060679303, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7855638265609741, + "num_tokens": 801417020.0, + "step": 21001 + }, + { + "epoch": 2.6716702709578932, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.8001740574836731, + "num_tokens": 801455328.0, + "step": 21002 + }, + { + "epoch": 2.6717974812364838, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.8154919147491455, + "num_tokens": 801494342.0, + "step": 21003 + }, + { + "epoch": 2.6719246915150743, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8011190295219421, + "num_tokens": 801528182.0, + "step": 21004 + }, + { + "epoch": 2.672051901793665, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8026541471481323, + "num_tokens": 801569424.0, + "step": 21005 + }, + { + "epoch": 2.6721791120722553, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8008133172988892, + "num_tokens": 801608809.0, + "step": 21006 + }, + { + "epoch": 2.672306322350846, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.794039785861969, + "num_tokens": 801651910.0, + "step": 21007 + }, + { + "epoch": 2.6724335326294364, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.8008585572242737, + "num_tokens": 801692961.0, + "step": 21008 + }, + { + "epoch": 2.672560742908027, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7853454947471619, + "num_tokens": 801731278.0, + "step": 21009 + }, + { + "epoch": 2.6726879531866174, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7967188358306885, + "num_tokens": 801769562.0, + "step": 21010 + }, + { + "epoch": 2.672815163465208, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7939304709434509, + "num_tokens": 801811046.0, + "step": 21011 + }, + { + "epoch": 2.6729423737437985, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7786032557487488, + "num_tokens": 801847833.0, + "step": 21012 + }, + { + "epoch": 2.673069584022389, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7938855886459351, + "num_tokens": 801887819.0, + "step": 21013 + }, + { + "epoch": 2.6731967943009796, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8085260987281799, + "num_tokens": 801922535.0, + "step": 21014 + }, + { + "epoch": 2.67332400457957, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8074239492416382, + "num_tokens": 801958025.0, + "step": 21015 + }, + { + "epoch": 2.6734512148581606, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8032160997390747, + "num_tokens": 801991291.0, + "step": 21016 + }, + { + "epoch": 2.673578425136751, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.807665228843689, + "num_tokens": 802024952.0, + "step": 21017 + }, + { + "epoch": 2.6737056354153417, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7986046075820923, + "num_tokens": 802065741.0, + "step": 21018 + }, + { + "epoch": 2.673832845693932, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8020426034927368, + "num_tokens": 802107153.0, + "step": 21019 + }, + { + "epoch": 2.6739600559725227, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8076860904693604, + "num_tokens": 802148177.0, + "step": 21020 + }, + { + "epoch": 2.674087266251113, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8022793531417847, + "num_tokens": 802186144.0, + "step": 21021 + }, + { + "epoch": 2.674214476529704, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.814176082611084, + "num_tokens": 802224054.0, + "step": 21022 + }, + { + "epoch": 2.674341686808294, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7851742506027222, + "num_tokens": 802264483.0, + "step": 21023 + }, + { + "epoch": 2.674468897086885, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8025595545768738, + "num_tokens": 802302933.0, + "step": 21024 + }, + { + "epoch": 2.674596107365475, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7948511838912964, + "num_tokens": 802347354.0, + "step": 21025 + }, + { + "epoch": 2.674723317644066, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8159155249595642, + "num_tokens": 802392159.0, + "step": 21026 + }, + { + "epoch": 2.674850527922656, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.800049364566803, + "num_tokens": 802432629.0, + "step": 21027 + }, + { + "epoch": 2.6749777382012465, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8158280849456787, + "num_tokens": 802471776.0, + "step": 21028 + }, + { + "epoch": 2.675104948479837, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7934392690658569, + "num_tokens": 802507314.0, + "step": 21029 + }, + { + "epoch": 2.6752321587584276, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8097521662712097, + "num_tokens": 802550587.0, + "step": 21030 + }, + { + "epoch": 2.675359369037018, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.8028491735458374, + "num_tokens": 802591544.0, + "step": 21031 + }, + { + "epoch": 2.6754865793156086, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8095669746398926, + "num_tokens": 802627336.0, + "step": 21032 + }, + { + "epoch": 2.675613789594199, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7992068529129028, + "num_tokens": 802668358.0, + "step": 21033 + }, + { + "epoch": 2.6757409998727897, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8099740147590637, + "num_tokens": 802708023.0, + "step": 21034 + }, + { + "epoch": 2.67586821015138, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.5987, + "mean_token_accuracy": 0.8206943869590759, + "num_tokens": 802745856.0, + "step": 21035 + }, + { + "epoch": 2.6759954204299707, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8150473833084106, + "num_tokens": 802779032.0, + "step": 21036 + }, + { + "epoch": 2.6761226307085613, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7969274520874023, + "num_tokens": 802815736.0, + "step": 21037 + }, + { + "epoch": 2.676249840987152, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.796347439289093, + "num_tokens": 802854133.0, + "step": 21038 + }, + { + "epoch": 2.6763770512657423, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8092795610427856, + "num_tokens": 802892969.0, + "step": 21039 + }, + { + "epoch": 2.676504261544333, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8149023652076721, + "num_tokens": 802936998.0, + "step": 21040 + }, + { + "epoch": 2.6766314718229234, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8098486661911011, + "num_tokens": 802979040.0, + "step": 21041 + }, + { + "epoch": 2.676758682101514, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8091563582420349, + "num_tokens": 803016516.0, + "step": 21042 + }, + { + "epoch": 2.6768858923801044, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7998446822166443, + "num_tokens": 803055566.0, + "step": 21043 + }, + { + "epoch": 2.677013102658695, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.8019017577171326, + "num_tokens": 803093683.0, + "step": 21044 + }, + { + "epoch": 2.6771403129372855, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.813065230846405, + "num_tokens": 803133691.0, + "step": 21045 + }, + { + "epoch": 2.6772675232158756, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8064777851104736, + "num_tokens": 803169870.0, + "step": 21046 + }, + { + "epoch": 2.6773947334944665, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7950935363769531, + "num_tokens": 803209038.0, + "step": 21047 + }, + { + "epoch": 2.6775219437730566, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7855175137519836, + "num_tokens": 803239725.0, + "step": 21048 + }, + { + "epoch": 2.6776491540516476, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8005611300468445, + "num_tokens": 803273580.0, + "step": 21049 + }, + { + "epoch": 2.6777763643302377, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8099046945571899, + "num_tokens": 803313583.0, + "step": 21050 + }, + { + "epoch": 2.6779035746088287, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.8036563396453857, + "num_tokens": 803347517.0, + "step": 21051 + }, + { + "epoch": 2.6780307848874187, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.787175714969635, + "num_tokens": 803388183.0, + "step": 21052 + }, + { + "epoch": 2.6781579951660093, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.807120680809021, + "num_tokens": 803428377.0, + "step": 21053 + }, + { + "epoch": 2.6782852054446, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8121306896209717, + "num_tokens": 803460930.0, + "step": 21054 + }, + { + "epoch": 2.6784124157231903, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8094391226768494, + "num_tokens": 803497971.0, + "step": 21055 + }, + { + "epoch": 2.678539626001781, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8203978538513184, + "num_tokens": 803535514.0, + "step": 21056 + }, + { + "epoch": 2.6786668362803714, + "grad_norm": 45.25, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7976175546646118, + "num_tokens": 803580767.0, + "step": 21057 + }, + { + "epoch": 2.678794046558962, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.8039299249649048, + "num_tokens": 803619168.0, + "step": 21058 + }, + { + "epoch": 2.6789212568375524, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7841712832450867, + "num_tokens": 803654785.0, + "step": 21059 + }, + { + "epoch": 2.679048467116143, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.81471848487854, + "num_tokens": 803692505.0, + "step": 21060 + }, + { + "epoch": 2.6791756773947335, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8142982125282288, + "num_tokens": 803732592.0, + "step": 21061 + }, + { + "epoch": 2.679302887673324, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7844909429550171, + "num_tokens": 803767601.0, + "step": 21062 + }, + { + "epoch": 2.6794300979519146, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7960400581359863, + "num_tokens": 803804329.0, + "step": 21063 + }, + { + "epoch": 2.679557308230505, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7999695539474487, + "num_tokens": 803846182.0, + "step": 21064 + }, + { + "epoch": 2.6796845185090956, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7982228994369507, + "num_tokens": 803883561.0, + "step": 21065 + }, + { + "epoch": 2.679811728787686, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.5869, + "mean_token_accuracy": 0.8222908973693848, + "num_tokens": 803916860.0, + "step": 21066 + }, + { + "epoch": 2.6799389390662767, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8015398383140564, + "num_tokens": 803951949.0, + "step": 21067 + }, + { + "epoch": 2.680066149344867, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.796924889087677, + "num_tokens": 803992094.0, + "step": 21068 + }, + { + "epoch": 2.6801933596234577, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7988418340682983, + "num_tokens": 804034058.0, + "step": 21069 + }, + { + "epoch": 2.6803205699020483, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7958579659461975, + "num_tokens": 804072722.0, + "step": 21070 + }, + { + "epoch": 2.6804477801806383, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8080406785011292, + "num_tokens": 804114253.0, + "step": 21071 + }, + { + "epoch": 2.6805749904592293, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8053259253501892, + "num_tokens": 804154877.0, + "step": 21072 + }, + { + "epoch": 2.6807022007378194, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7968329191207886, + "num_tokens": 804191060.0, + "step": 21073 + }, + { + "epoch": 2.6808294110164104, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7933002710342407, + "num_tokens": 804231574.0, + "step": 21074 + }, + { + "epoch": 2.6809566212950005, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7786160111427307, + "num_tokens": 804264027.0, + "step": 21075 + }, + { + "epoch": 2.681083831573591, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8004758954048157, + "num_tokens": 804302906.0, + "step": 21076 + }, + { + "epoch": 2.6812110418521815, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7958781719207764, + "num_tokens": 804339107.0, + "step": 21077 + }, + { + "epoch": 2.681338252130772, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8071575164794922, + "num_tokens": 804374292.0, + "step": 21078 + }, + { + "epoch": 2.6814654624093626, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8038415312767029, + "num_tokens": 804420287.0, + "step": 21079 + }, + { + "epoch": 2.681592672687953, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8032474517822266, + "num_tokens": 804453977.0, + "step": 21080 + }, + { + "epoch": 2.6817198829665436, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7845768928527832, + "num_tokens": 804494958.0, + "step": 21081 + }, + { + "epoch": 2.681847093245134, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7745885848999023, + "num_tokens": 804532209.0, + "step": 21082 + }, + { + "epoch": 2.6819743035237247, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8055417537689209, + "num_tokens": 804572937.0, + "step": 21083 + }, + { + "epoch": 2.682101513802315, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.798611044883728, + "num_tokens": 804610263.0, + "step": 21084 + }, + { + "epoch": 2.6822287240809057, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8029939532279968, + "num_tokens": 804645516.0, + "step": 21085 + }, + { + "epoch": 2.6823559343594963, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.8099442720413208, + "num_tokens": 804684540.0, + "step": 21086 + }, + { + "epoch": 2.682483144638087, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8012466430664062, + "num_tokens": 804723030.0, + "step": 21087 + }, + { + "epoch": 2.6826103549166773, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8081886768341064, + "num_tokens": 804759202.0, + "step": 21088 + }, + { + "epoch": 2.682737565195268, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8032615780830383, + "num_tokens": 804803199.0, + "step": 21089 + }, + { + "epoch": 2.6828647754738584, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7985010743141174, + "num_tokens": 804835675.0, + "step": 21090 + }, + { + "epoch": 2.682991985752449, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7864778637886047, + "num_tokens": 804876583.0, + "step": 21091 + }, + { + "epoch": 2.6831191960310394, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7945058345794678, + "num_tokens": 804906110.0, + "step": 21092 + }, + { + "epoch": 2.68324640630963, + "grad_norm": 47.0, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7918512225151062, + "num_tokens": 804942196.0, + "step": 21093 + }, + { + "epoch": 2.6833736165882205, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7992281913757324, + "num_tokens": 804978760.0, + "step": 21094 + }, + { + "epoch": 2.683500826866811, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7932065725326538, + "num_tokens": 805014212.0, + "step": 21095 + }, + { + "epoch": 2.683628037145401, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8301396369934082, + "num_tokens": 805050528.0, + "step": 21096 + }, + { + "epoch": 2.683755247423992, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8057518005371094, + "num_tokens": 805087355.0, + "step": 21097 + }, + { + "epoch": 2.683882457702582, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7828680276870728, + "num_tokens": 805117837.0, + "step": 21098 + }, + { + "epoch": 2.684009667981173, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8056550025939941, + "num_tokens": 805157734.0, + "step": 21099 + }, + { + "epoch": 2.684136878259763, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.805482029914856, + "num_tokens": 805190657.0, + "step": 21100 + }, + { + "epoch": 2.6842640885383537, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7885383367538452, + "num_tokens": 805229654.0, + "step": 21101 + }, + { + "epoch": 2.6843912988169443, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7999765872955322, + "num_tokens": 805269199.0, + "step": 21102 + }, + { + "epoch": 2.684518509095535, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7937057018280029, + "num_tokens": 805308918.0, + "step": 21103 + }, + { + "epoch": 2.6846457193741253, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7981666922569275, + "num_tokens": 805345014.0, + "step": 21104 + }, + { + "epoch": 2.684772929652716, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8170688152313232, + "num_tokens": 805383674.0, + "step": 21105 + }, + { + "epoch": 2.6849001399313064, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7994406223297119, + "num_tokens": 805419765.0, + "step": 21106 + }, + { + "epoch": 2.685027350209897, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8068766593933105, + "num_tokens": 805461346.0, + "step": 21107 + }, + { + "epoch": 2.6851545604884874, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8047343492507935, + "num_tokens": 805494519.0, + "step": 21108 + }, + { + "epoch": 2.685281770767078, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7878689169883728, + "num_tokens": 805530542.0, + "step": 21109 + }, + { + "epoch": 2.6854089810456685, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7918500304222107, + "num_tokens": 805564531.0, + "step": 21110 + }, + { + "epoch": 2.685536191324259, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8009151220321655, + "num_tokens": 805606011.0, + "step": 21111 + }, + { + "epoch": 2.6856634016028496, + "grad_norm": 85.0, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7948490381240845, + "num_tokens": 805643078.0, + "step": 21112 + }, + { + "epoch": 2.68579061188144, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8073991537094116, + "num_tokens": 805678554.0, + "step": 21113 + }, + { + "epoch": 2.6859178221600306, + "grad_norm": 56.5, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8127337694168091, + "num_tokens": 805713736.0, + "step": 21114 + }, + { + "epoch": 2.686045032438621, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7949456572532654, + "num_tokens": 805751798.0, + "step": 21115 + }, + { + "epoch": 2.6861722427172117, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8106598854064941, + "num_tokens": 805789205.0, + "step": 21116 + }, + { + "epoch": 2.686299452995802, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8060030937194824, + "num_tokens": 805824910.0, + "step": 21117 + }, + { + "epoch": 2.6864266632743927, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.796465277671814, + "num_tokens": 805861335.0, + "step": 21118 + }, + { + "epoch": 2.686553873552983, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7964977622032166, + "num_tokens": 805901488.0, + "step": 21119 + }, + { + "epoch": 2.686681083831574, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7638121843338013, + "num_tokens": 805941336.0, + "step": 21120 + }, + { + "epoch": 2.686808294110164, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.808091938495636, + "num_tokens": 805974222.0, + "step": 21121 + }, + { + "epoch": 2.686935504388755, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7970603704452515, + "num_tokens": 806014942.0, + "step": 21122 + }, + { + "epoch": 2.687062714667345, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8107906579971313, + "num_tokens": 806048894.0, + "step": 21123 + }, + { + "epoch": 2.687189924945936, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7989894151687622, + "num_tokens": 806089714.0, + "step": 21124 + }, + { + "epoch": 2.687317135224526, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7845202088356018, + "num_tokens": 806128974.0, + "step": 21125 + }, + { + "epoch": 2.6874443455031165, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8013975024223328, + "num_tokens": 806169857.0, + "step": 21126 + }, + { + "epoch": 2.687571555781707, + "grad_norm": 47.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8018206357955933, + "num_tokens": 806202826.0, + "step": 21127 + }, + { + "epoch": 2.6876987660602976, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.8014715313911438, + "num_tokens": 806242850.0, + "step": 21128 + }, + { + "epoch": 2.687825976338888, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8057653307914734, + "num_tokens": 806279024.0, + "step": 21129 + }, + { + "epoch": 2.6879531866174786, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7808187007904053, + "num_tokens": 806313669.0, + "step": 21130 + }, + { + "epoch": 2.688080396896069, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8035953044891357, + "num_tokens": 806353128.0, + "step": 21131 + }, + { + "epoch": 2.6882076071746597, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8007545471191406, + "num_tokens": 806385044.0, + "step": 21132 + }, + { + "epoch": 2.68833481745325, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7938421964645386, + "num_tokens": 806422157.0, + "step": 21133 + }, + { + "epoch": 2.6884620277318407, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8013761639595032, + "num_tokens": 806463180.0, + "step": 21134 + }, + { + "epoch": 2.6885892380104313, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7933717966079712, + "num_tokens": 806502031.0, + "step": 21135 + }, + { + "epoch": 2.688716448289022, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.79807448387146, + "num_tokens": 806539585.0, + "step": 21136 + }, + { + "epoch": 2.6888436585676123, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8215227127075195, + "num_tokens": 806573147.0, + "step": 21137 + }, + { + "epoch": 2.688970868846203, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8180046081542969, + "num_tokens": 806609785.0, + "step": 21138 + }, + { + "epoch": 2.6890980791247934, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8092842102050781, + "num_tokens": 806647709.0, + "step": 21139 + }, + { + "epoch": 2.689225289403384, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8091164231300354, + "num_tokens": 806684448.0, + "step": 21140 + }, + { + "epoch": 2.6893524996819744, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8076151609420776, + "num_tokens": 806725326.0, + "step": 21141 + }, + { + "epoch": 2.689479709960565, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.802539587020874, + "num_tokens": 806759797.0, + "step": 21142 + }, + { + "epoch": 2.6896069202391555, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.807852029800415, + "num_tokens": 806799986.0, + "step": 21143 + }, + { + "epoch": 2.6897341305177456, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.794786274433136, + "num_tokens": 806843055.0, + "step": 21144 + }, + { + "epoch": 2.6898613407963365, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.8169922232627869, + "num_tokens": 806875897.0, + "step": 21145 + }, + { + "epoch": 2.6899885510749266, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8084893226623535, + "num_tokens": 806915583.0, + "step": 21146 + }, + { + "epoch": 2.6901157613535176, + "grad_norm": 57.5, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8036158084869385, + "num_tokens": 806953851.0, + "step": 21147 + }, + { + "epoch": 2.6902429716321077, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7928298711776733, + "num_tokens": 806988519.0, + "step": 21148 + }, + { + "epoch": 2.6903701819106987, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.808003306388855, + "num_tokens": 807029612.0, + "step": 21149 + }, + { + "epoch": 2.6904973921892887, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7814489603042603, + "num_tokens": 807064009.0, + "step": 21150 + }, + { + "epoch": 2.6906246024678793, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7980884313583374, + "num_tokens": 807104694.0, + "step": 21151 + }, + { + "epoch": 2.69075181274647, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7769128084182739, + "num_tokens": 807142500.0, + "step": 21152 + }, + { + "epoch": 2.6908790230250603, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.79709392786026, + "num_tokens": 807172123.0, + "step": 21153 + }, + { + "epoch": 2.691006233303651, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.8022247552871704, + "num_tokens": 807206422.0, + "step": 21154 + }, + { + "epoch": 2.6911334435822414, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7808228135108948, + "num_tokens": 807240867.0, + "step": 21155 + }, + { + "epoch": 2.691260653860832, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8125348091125488, + "num_tokens": 807275838.0, + "step": 21156 + }, + { + "epoch": 2.6913878641394224, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8101128339767456, + "num_tokens": 807315164.0, + "step": 21157 + }, + { + "epoch": 2.691515074418013, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7955949306488037, + "num_tokens": 807358016.0, + "step": 21158 + }, + { + "epoch": 2.6916422846966035, + "grad_norm": 11.3125, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8033488392829895, + "num_tokens": 807402751.0, + "step": 21159 + }, + { + "epoch": 2.691769494975194, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8068860769271851, + "num_tokens": 807442761.0, + "step": 21160 + }, + { + "epoch": 2.6918967052537845, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.8075511455535889, + "num_tokens": 807478550.0, + "step": 21161 + }, + { + "epoch": 2.692023915532375, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8109472990036011, + "num_tokens": 807522756.0, + "step": 21162 + }, + { + "epoch": 2.6921511258109656, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8115970492362976, + "num_tokens": 807561701.0, + "step": 21163 + }, + { + "epoch": 2.692278336089556, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7981463670730591, + "num_tokens": 807604058.0, + "step": 21164 + }, + { + "epoch": 2.6924055463681467, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8053507804870605, + "num_tokens": 807637344.0, + "step": 21165 + }, + { + "epoch": 2.692532756646737, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7904275059700012, + "num_tokens": 807676867.0, + "step": 21166 + }, + { + "epoch": 2.6926599669253277, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.7999736070632935, + "num_tokens": 807707183.0, + "step": 21167 + }, + { + "epoch": 2.6927871772039182, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7877122163772583, + "num_tokens": 807737617.0, + "step": 21168 + }, + { + "epoch": 2.6929143874825083, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7874370813369751, + "num_tokens": 807774141.0, + "step": 21169 + }, + { + "epoch": 2.6930415977610993, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8099002838134766, + "num_tokens": 807813375.0, + "step": 21170 + }, + { + "epoch": 2.6931688080396894, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7843145132064819, + "num_tokens": 807852317.0, + "step": 21171 + }, + { + "epoch": 2.6932960183182804, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.799128532409668, + "num_tokens": 807895266.0, + "step": 21172 + }, + { + "epoch": 2.6934232285968704, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7977100610733032, + "num_tokens": 807938600.0, + "step": 21173 + }, + { + "epoch": 2.693550438875461, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.8174206614494324, + "num_tokens": 807974329.0, + "step": 21174 + }, + { + "epoch": 2.6936776491540515, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.8127750158309937, + "num_tokens": 808015666.0, + "step": 21175 + }, + { + "epoch": 2.693804859432642, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8112905025482178, + "num_tokens": 808050176.0, + "step": 21176 + }, + { + "epoch": 2.6939320697112326, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7934699058532715, + "num_tokens": 808091910.0, + "step": 21177 + }, + { + "epoch": 2.694059279989823, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7971271276473999, + "num_tokens": 808128646.0, + "step": 21178 + }, + { + "epoch": 2.6941864902684136, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8006443977355957, + "num_tokens": 808166544.0, + "step": 21179 + }, + { + "epoch": 2.694313700547004, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7891032099723816, + "num_tokens": 808200051.0, + "step": 21180 + }, + { + "epoch": 2.6944409108255947, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8053568005561829, + "num_tokens": 808239251.0, + "step": 21181 + }, + { + "epoch": 2.694568121104185, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7990127801895142, + "num_tokens": 808280639.0, + "step": 21182 + }, + { + "epoch": 2.6946953313827757, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8155366778373718, + "num_tokens": 808316141.0, + "step": 21183 + }, + { + "epoch": 2.6948225416613663, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8035192489624023, + "num_tokens": 808359172.0, + "step": 21184 + }, + { + "epoch": 2.694949751939957, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7673794031143188, + "num_tokens": 808397823.0, + "step": 21185 + }, + { + "epoch": 2.6950769622185473, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8057107925415039, + "num_tokens": 808435355.0, + "step": 21186 + }, + { + "epoch": 2.695204172497138, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8086872100830078, + "num_tokens": 808470792.0, + "step": 21187 + }, + { + "epoch": 2.6953313827757284, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8192100524902344, + "num_tokens": 808504591.0, + "step": 21188 + }, + { + "epoch": 2.695458593054319, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8054155707359314, + "num_tokens": 808537167.0, + "step": 21189 + }, + { + "epoch": 2.6955858033329094, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8131171464920044, + "num_tokens": 808572055.0, + "step": 21190 + }, + { + "epoch": 2.6957130136115, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.8110125064849854, + "num_tokens": 808605239.0, + "step": 21191 + }, + { + "epoch": 2.6958402238900905, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7764291763305664, + "num_tokens": 808643227.0, + "step": 21192 + }, + { + "epoch": 2.695967434168681, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7941403388977051, + "num_tokens": 808683855.0, + "step": 21193 + }, + { + "epoch": 2.696094644447271, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8103976249694824, + "num_tokens": 808723467.0, + "step": 21194 + }, + { + "epoch": 2.696221854725862, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.784194827079773, + "num_tokens": 808759805.0, + "step": 21195 + }, + { + "epoch": 2.696349065004452, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7853856086730957, + "num_tokens": 808802024.0, + "step": 21196 + }, + { + "epoch": 2.696476275283043, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7932404279708862, + "num_tokens": 808837086.0, + "step": 21197 + }, + { + "epoch": 2.696603485561633, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8034620881080627, + "num_tokens": 808875261.0, + "step": 21198 + }, + { + "epoch": 2.6967306958402237, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8076507449150085, + "num_tokens": 808913044.0, + "step": 21199 + }, + { + "epoch": 2.6968579061188143, + "grad_norm": 59.75, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.8148496150970459, + "num_tokens": 808952374.0, + "step": 21200 + }, + { + "epoch": 2.696985116397405, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8017486333847046, + "num_tokens": 808994397.0, + "step": 21201 + }, + { + "epoch": 2.6971123266759953, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8077768087387085, + "num_tokens": 809036401.0, + "step": 21202 + }, + { + "epoch": 2.697239536954586, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8056862354278564, + "num_tokens": 809073073.0, + "step": 21203 + }, + { + "epoch": 2.6973667472331764, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8014208078384399, + "num_tokens": 809117348.0, + "step": 21204 + }, + { + "epoch": 2.697493957511767, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8115038871765137, + "num_tokens": 809151357.0, + "step": 21205 + }, + { + "epoch": 2.6976211677903574, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.796528697013855, + "num_tokens": 809189000.0, + "step": 21206 + }, + { + "epoch": 2.697748378068948, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.8021400570869446, + "num_tokens": 809227380.0, + "step": 21207 + }, + { + "epoch": 2.6978755883475385, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8048081994056702, + "num_tokens": 809265131.0, + "step": 21208 + }, + { + "epoch": 2.698002798626129, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.8032095432281494, + "num_tokens": 809297190.0, + "step": 21209 + }, + { + "epoch": 2.6981300089047195, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8086656332015991, + "num_tokens": 809336774.0, + "step": 21210 + }, + { + "epoch": 2.69825721918331, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7853686809539795, + "num_tokens": 809376612.0, + "step": 21211 + }, + { + "epoch": 2.6983844294619006, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.5606, + "mean_token_accuracy": 0.8270092606544495, + "num_tokens": 809416058.0, + "step": 21212 + }, + { + "epoch": 2.698511639740491, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8115696310997009, + "num_tokens": 809449580.0, + "step": 21213 + }, + { + "epoch": 2.6986388500190817, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8057868480682373, + "num_tokens": 809484983.0, + "step": 21214 + }, + { + "epoch": 2.698766060297672, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8045704364776611, + "num_tokens": 809521086.0, + "step": 21215 + }, + { + "epoch": 2.6988932705762627, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7987891435623169, + "num_tokens": 809562475.0, + "step": 21216 + }, + { + "epoch": 2.699020480854853, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7935234308242798, + "num_tokens": 809596686.0, + "step": 21217 + }, + { + "epoch": 2.6991476911334438, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8089594841003418, + "num_tokens": 809636614.0, + "step": 21218 + }, + { + "epoch": 2.699274901412034, + "grad_norm": 45.0, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7862213850021362, + "num_tokens": 809674542.0, + "step": 21219 + }, + { + "epoch": 2.699402111690625, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8194316625595093, + "num_tokens": 809716726.0, + "step": 21220 + }, + { + "epoch": 2.699529321969215, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8072338104248047, + "num_tokens": 809756790.0, + "step": 21221 + }, + { + "epoch": 2.699656532247806, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8036447167396545, + "num_tokens": 809794616.0, + "step": 21222 + }, + { + "epoch": 2.699783742526396, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8142592310905457, + "num_tokens": 809836662.0, + "step": 21223 + }, + { + "epoch": 2.6999109528049865, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7972013354301453, + "num_tokens": 809873812.0, + "step": 21224 + }, + { + "epoch": 2.700038163083577, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7985738515853882, + "num_tokens": 809913139.0, + "step": 21225 + }, + { + "epoch": 2.7001653733621676, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7982631921768188, + "num_tokens": 809947741.0, + "step": 21226 + }, + { + "epoch": 2.700292583640758, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7942171692848206, + "num_tokens": 809987120.0, + "step": 21227 + }, + { + "epoch": 2.7004197939193486, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7932456731796265, + "num_tokens": 810026972.0, + "step": 21228 + }, + { + "epoch": 2.700547004197939, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8025172352790833, + "num_tokens": 810062767.0, + "step": 21229 + }, + { + "epoch": 2.7006742144765297, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8103965520858765, + "num_tokens": 810096930.0, + "step": 21230 + }, + { + "epoch": 2.70080142475512, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7831604480743408, + "num_tokens": 810134600.0, + "step": 21231 + }, + { + "epoch": 2.7009286350337107, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8029170036315918, + "num_tokens": 810169535.0, + "step": 21232 + }, + { + "epoch": 2.7010558453123013, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7942525148391724, + "num_tokens": 810205288.0, + "step": 21233 + }, + { + "epoch": 2.701183055590892, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7987344861030579, + "num_tokens": 810242868.0, + "step": 21234 + }, + { + "epoch": 2.7013102658694823, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7912639379501343, + "num_tokens": 810286539.0, + "step": 21235 + }, + { + "epoch": 2.701437476148073, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8059580326080322, + "num_tokens": 810331598.0, + "step": 21236 + }, + { + "epoch": 2.7015646864266634, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8161855340003967, + "num_tokens": 810369501.0, + "step": 21237 + }, + { + "epoch": 2.701691896705254, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7970470786094666, + "num_tokens": 810410727.0, + "step": 21238 + }, + { + "epoch": 2.7018191069838444, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7998219728469849, + "num_tokens": 810450315.0, + "step": 21239 + }, + { + "epoch": 2.701946317262435, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7938817739486694, + "num_tokens": 810493972.0, + "step": 21240 + }, + { + "epoch": 2.7020735275410255, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.812309980392456, + "num_tokens": 810526038.0, + "step": 21241 + }, + { + "epoch": 2.7022007378196156, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8029367923736572, + "num_tokens": 810560772.0, + "step": 21242 + }, + { + "epoch": 2.7023279480982065, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8109238147735596, + "num_tokens": 810598322.0, + "step": 21243 + }, + { + "epoch": 2.7024551583767966, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.8229489326477051, + "num_tokens": 810638652.0, + "step": 21244 + }, + { + "epoch": 2.7025823686553876, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8001582622528076, + "num_tokens": 810680318.0, + "step": 21245 + }, + { + "epoch": 2.7027095789339777, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8076746463775635, + "num_tokens": 810715001.0, + "step": 21246 + }, + { + "epoch": 2.7028367892125686, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7888362407684326, + "num_tokens": 810748897.0, + "step": 21247 + }, + { + "epoch": 2.7029639994911587, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.797285795211792, + "num_tokens": 810788954.0, + "step": 21248 + }, + { + "epoch": 2.7030912097697493, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8014034032821655, + "num_tokens": 810828897.0, + "step": 21249 + }, + { + "epoch": 2.70321842004834, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.818719208240509, + "num_tokens": 810866698.0, + "step": 21250 + }, + { + "epoch": 2.7033456303269303, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7944951057434082, + "num_tokens": 810903140.0, + "step": 21251 + }, + { + "epoch": 2.703472840605521, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8033614158630371, + "num_tokens": 810940664.0, + "step": 21252 + }, + { + "epoch": 2.7036000508841114, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8008288145065308, + "num_tokens": 810981818.0, + "step": 21253 + }, + { + "epoch": 2.703727261162702, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7972439527511597, + "num_tokens": 811019897.0, + "step": 21254 + }, + { + "epoch": 2.7038544714412924, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8051894903182983, + "num_tokens": 811056310.0, + "step": 21255 + }, + { + "epoch": 2.703981681719883, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7922236323356628, + "num_tokens": 811091505.0, + "step": 21256 + }, + { + "epoch": 2.7041088919984735, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8178555965423584, + "num_tokens": 811134861.0, + "step": 21257 + }, + { + "epoch": 2.704236102277064, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7913656234741211, + "num_tokens": 811170001.0, + "step": 21258 + }, + { + "epoch": 2.7043633125556545, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8034403324127197, + "num_tokens": 811204203.0, + "step": 21259 + }, + { + "epoch": 2.704490522834245, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8072608113288879, + "num_tokens": 811244598.0, + "step": 21260 + }, + { + "epoch": 2.7046177331128356, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.8125340938568115, + "num_tokens": 811283352.0, + "step": 21261 + }, + { + "epoch": 2.704744943391426, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7862392663955688, + "num_tokens": 811317605.0, + "step": 21262 + }, + { + "epoch": 2.7048721536700167, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8154832124710083, + "num_tokens": 811352117.0, + "step": 21263 + }, + { + "epoch": 2.704999363948607, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7936168313026428, + "num_tokens": 811386118.0, + "step": 21264 + }, + { + "epoch": 2.7051265742271977, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8120954632759094, + "num_tokens": 811424479.0, + "step": 21265 + }, + { + "epoch": 2.7052537845057882, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7902885675430298, + "num_tokens": 811464683.0, + "step": 21266 + }, + { + "epoch": 2.7053809947843783, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.5848, + "mean_token_accuracy": 0.8254094123840332, + "num_tokens": 811504801.0, + "step": 21267 + }, + { + "epoch": 2.7055082050629693, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7928231358528137, + "num_tokens": 811546278.0, + "step": 21268 + }, + { + "epoch": 2.7056354153415594, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7891688346862793, + "num_tokens": 811587019.0, + "step": 21269 + }, + { + "epoch": 2.7057626256201504, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7781230211257935, + "num_tokens": 811619843.0, + "step": 21270 + }, + { + "epoch": 2.7058898358987404, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8042941093444824, + "num_tokens": 811655797.0, + "step": 21271 + }, + { + "epoch": 2.706017046177331, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.792719841003418, + "num_tokens": 811691626.0, + "step": 21272 + }, + { + "epoch": 2.7061442564559215, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.8003251552581787, + "num_tokens": 811726046.0, + "step": 21273 + }, + { + "epoch": 2.706271466734512, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.794573187828064, + "num_tokens": 811765686.0, + "step": 21274 + }, + { + "epoch": 2.7063986770131025, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8073902130126953, + "num_tokens": 811801143.0, + "step": 21275 + }, + { + "epoch": 2.706525887291693, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.8093540668487549, + "num_tokens": 811834807.0, + "step": 21276 + }, + { + "epoch": 2.7066530975702836, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6022, + "mean_token_accuracy": 0.8184694647789001, + "num_tokens": 811865676.0, + "step": 21277 + }, + { + "epoch": 2.706780307848874, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7986021041870117, + "num_tokens": 811901259.0, + "step": 21278 + }, + { + "epoch": 2.7069075181274647, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7870858907699585, + "num_tokens": 811937768.0, + "step": 21279 + }, + { + "epoch": 2.707034728406055, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7962840795516968, + "num_tokens": 811967805.0, + "step": 21280 + }, + { + "epoch": 2.7071619386846457, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.79930579662323, + "num_tokens": 812007468.0, + "step": 21281 + }, + { + "epoch": 2.7072891489632362, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8009990453720093, + "num_tokens": 812048127.0, + "step": 21282 + }, + { + "epoch": 2.7074163592418268, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8005892038345337, + "num_tokens": 812083904.0, + "step": 21283 + }, + { + "epoch": 2.7075435695204173, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8023139238357544, + "num_tokens": 812119643.0, + "step": 21284 + }, + { + "epoch": 2.707670779799008, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7953897714614868, + "num_tokens": 812155338.0, + "step": 21285 + }, + { + "epoch": 2.7077979900775984, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8014963865280151, + "num_tokens": 812193075.0, + "step": 21286 + }, + { + "epoch": 2.707925200356189, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8071491122245789, + "num_tokens": 812228163.0, + "step": 21287 + }, + { + "epoch": 2.7080524106347794, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7994273900985718, + "num_tokens": 812268565.0, + "step": 21288 + }, + { + "epoch": 2.70817962091337, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8063100576400757, + "num_tokens": 812307292.0, + "step": 21289 + }, + { + "epoch": 2.7083068311919605, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8128437995910645, + "num_tokens": 812346180.0, + "step": 21290 + }, + { + "epoch": 2.708434041470551, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7908433675765991, + "num_tokens": 812376033.0, + "step": 21291 + }, + { + "epoch": 2.708561251749141, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8209203481674194, + "num_tokens": 812407925.0, + "step": 21292 + }, + { + "epoch": 2.708688462027732, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.811884880065918, + "num_tokens": 812452354.0, + "step": 21293 + }, + { + "epoch": 2.708815672306322, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7943333387374878, + "num_tokens": 812490015.0, + "step": 21294 + }, + { + "epoch": 2.708942882584913, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8008130192756653, + "num_tokens": 812529919.0, + "step": 21295 + }, + { + "epoch": 2.709070092863503, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8130866289138794, + "num_tokens": 812568541.0, + "step": 21296 + }, + { + "epoch": 2.7091973031420937, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7932231426239014, + "num_tokens": 812612098.0, + "step": 21297 + }, + { + "epoch": 2.7093245134206843, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.796651303768158, + "num_tokens": 812648843.0, + "step": 21298 + }, + { + "epoch": 2.709451723699275, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8038526773452759, + "num_tokens": 812692080.0, + "step": 21299 + }, + { + "epoch": 2.7095789339778653, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8141638040542603, + "num_tokens": 812729441.0, + "step": 21300 + }, + { + "epoch": 2.709706144256456, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7981821894645691, + "num_tokens": 812768908.0, + "step": 21301 + }, + { + "epoch": 2.7098333545350464, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7993170022964478, + "num_tokens": 812808384.0, + "step": 21302 + }, + { + "epoch": 2.709960564813637, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.788830578327179, + "num_tokens": 812847658.0, + "step": 21303 + }, + { + "epoch": 2.7100877750922274, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8035200834274292, + "num_tokens": 812889394.0, + "step": 21304 + }, + { + "epoch": 2.710214985370818, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7832701206207275, + "num_tokens": 812924202.0, + "step": 21305 + }, + { + "epoch": 2.7103421956494085, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.8191063404083252, + "num_tokens": 812954391.0, + "step": 21306 + }, + { + "epoch": 2.710469405927999, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7889795899391174, + "num_tokens": 812989498.0, + "step": 21307 + }, + { + "epoch": 2.7105966162065895, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8136738538742065, + "num_tokens": 813030968.0, + "step": 21308 + }, + { + "epoch": 2.71072382648518, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8217744827270508, + "num_tokens": 813065843.0, + "step": 21309 + }, + { + "epoch": 2.7108510367637706, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8060147166252136, + "num_tokens": 813107410.0, + "step": 21310 + }, + { + "epoch": 2.710978247042361, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8098336458206177, + "num_tokens": 813155959.0, + "step": 21311 + }, + { + "epoch": 2.7111054573209517, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8104151487350464, + "num_tokens": 813194133.0, + "step": 21312 + }, + { + "epoch": 2.711232667599542, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7898707389831543, + "num_tokens": 813231015.0, + "step": 21313 + }, + { + "epoch": 2.7113598778781327, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.8186259269714355, + "num_tokens": 813271497.0, + "step": 21314 + }, + { + "epoch": 2.711487088156723, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7986416816711426, + "num_tokens": 813305758.0, + "step": 21315 + }, + { + "epoch": 2.7116142984353138, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8005515336990356, + "num_tokens": 813345958.0, + "step": 21316 + }, + { + "epoch": 2.711741508713904, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8038516044616699, + "num_tokens": 813381630.0, + "step": 21317 + }, + { + "epoch": 2.711868718992495, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8038792014122009, + "num_tokens": 813421320.0, + "step": 21318 + }, + { + "epoch": 2.711995929271085, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7988622784614563, + "num_tokens": 813458323.0, + "step": 21319 + }, + { + "epoch": 2.712123139549676, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7803733348846436, + "num_tokens": 813501436.0, + "step": 21320 + }, + { + "epoch": 2.712250349828266, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8065007328987122, + "num_tokens": 813545942.0, + "step": 21321 + }, + { + "epoch": 2.7123775601068565, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7966682314872742, + "num_tokens": 813582202.0, + "step": 21322 + }, + { + "epoch": 2.712504770385447, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7844656705856323, + "num_tokens": 813621711.0, + "step": 21323 + }, + { + "epoch": 2.7126319806640375, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7969481945037842, + "num_tokens": 813665404.0, + "step": 21324 + }, + { + "epoch": 2.712759190942628, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7981271743774414, + "num_tokens": 813697470.0, + "step": 21325 + }, + { + "epoch": 2.7128864012212186, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8072148561477661, + "num_tokens": 813733308.0, + "step": 21326 + }, + { + "epoch": 2.713013611499809, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7910847067832947, + "num_tokens": 813770881.0, + "step": 21327 + }, + { + "epoch": 2.7131408217783997, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8034408092498779, + "num_tokens": 813806495.0, + "step": 21328 + }, + { + "epoch": 2.71326803205699, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8054031133651733, + "num_tokens": 813842408.0, + "step": 21329 + }, + { + "epoch": 2.7133952423355807, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7981052398681641, + "num_tokens": 813883041.0, + "step": 21330 + }, + { + "epoch": 2.7135224526141712, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7970297932624817, + "num_tokens": 813922901.0, + "step": 21331 + }, + { + "epoch": 2.7136496628927618, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.8042489290237427, + "num_tokens": 813961948.0, + "step": 21332 + }, + { + "epoch": 2.7137768731713523, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.596, + "mean_token_accuracy": 0.8226277828216553, + "num_tokens": 813999194.0, + "step": 21333 + }, + { + "epoch": 2.713904083449943, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7916970252990723, + "num_tokens": 814038986.0, + "step": 21334 + }, + { + "epoch": 2.7140312937285334, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6005, + "mean_token_accuracy": 0.8188486099243164, + "num_tokens": 814074351.0, + "step": 21335 + }, + { + "epoch": 2.714158504007124, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7982897758483887, + "num_tokens": 814109255.0, + "step": 21336 + }, + { + "epoch": 2.7142857142857144, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7954441905021667, + "num_tokens": 814146601.0, + "step": 21337 + }, + { + "epoch": 2.714412924564305, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8099517822265625, + "num_tokens": 814181299.0, + "step": 21338 + }, + { + "epoch": 2.7145401348428955, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.814441978931427, + "num_tokens": 814218178.0, + "step": 21339 + }, + { + "epoch": 2.7146673451214856, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8041831254959106, + "num_tokens": 814254161.0, + "step": 21340 + }, + { + "epoch": 2.7147945554000765, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8019580245018005, + "num_tokens": 814290164.0, + "step": 21341 + }, + { + "epoch": 2.7149217656786666, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7822052240371704, + "num_tokens": 814334459.0, + "step": 21342 + }, + { + "epoch": 2.7150489759572576, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7943189144134521, + "num_tokens": 814374291.0, + "step": 21343 + }, + { + "epoch": 2.7151761862358477, + "grad_norm": 11.6875, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.801603376865387, + "num_tokens": 814413196.0, + "step": 21344 + }, + { + "epoch": 2.7153033965144386, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7979679107666016, + "num_tokens": 814455312.0, + "step": 21345 + }, + { + "epoch": 2.7154306067930287, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8030014038085938, + "num_tokens": 814496586.0, + "step": 21346 + }, + { + "epoch": 2.7155578170716193, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.8075390458106995, + "num_tokens": 814531566.0, + "step": 21347 + }, + { + "epoch": 2.71568502735021, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8099379539489746, + "num_tokens": 814566168.0, + "step": 21348 + }, + { + "epoch": 2.7158122376288003, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8035821318626404, + "num_tokens": 814604927.0, + "step": 21349 + }, + { + "epoch": 2.715939447907391, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5987, + "mean_token_accuracy": 0.8147432208061218, + "num_tokens": 814641401.0, + "step": 21350 + }, + { + "epoch": 2.7160666581859814, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8013098239898682, + "num_tokens": 814674252.0, + "step": 21351 + }, + { + "epoch": 2.716193868464572, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.80384761095047, + "num_tokens": 814704965.0, + "step": 21352 + }, + { + "epoch": 2.7163210787431624, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7693837285041809, + "num_tokens": 814743740.0, + "step": 21353 + }, + { + "epoch": 2.716448289021753, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.811354398727417, + "num_tokens": 814784598.0, + "step": 21354 + }, + { + "epoch": 2.7165754993003435, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7963021397590637, + "num_tokens": 814821437.0, + "step": 21355 + }, + { + "epoch": 2.716702709578934, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.8045017719268799, + "num_tokens": 814861638.0, + "step": 21356 + }, + { + "epoch": 2.7168299198575245, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.803631067276001, + "num_tokens": 814893799.0, + "step": 21357 + }, + { + "epoch": 2.716957130136115, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8100826740264893, + "num_tokens": 814930726.0, + "step": 21358 + }, + { + "epoch": 2.7170843404147056, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7980449199676514, + "num_tokens": 814970041.0, + "step": 21359 + }, + { + "epoch": 2.717211550693296, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8055725693702698, + "num_tokens": 815008587.0, + "step": 21360 + }, + { + "epoch": 2.7173387609718866, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7870262861251831, + "num_tokens": 815047659.0, + "step": 21361 + }, + { + "epoch": 2.717465971250477, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7967411279678345, + "num_tokens": 815082631.0, + "step": 21362 + }, + { + "epoch": 2.7175931815290677, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.8203967213630676, + "num_tokens": 815117568.0, + "step": 21363 + }, + { + "epoch": 2.7177203918076582, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8026385307312012, + "num_tokens": 815155019.0, + "step": 21364 + }, + { + "epoch": 2.7178476020862483, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8062042593955994, + "num_tokens": 815195386.0, + "step": 21365 + }, + { + "epoch": 2.7179748123648393, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.804462194442749, + "num_tokens": 815228304.0, + "step": 21366 + }, + { + "epoch": 2.7181020226434294, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.816329836845398, + "num_tokens": 815260111.0, + "step": 21367 + }, + { + "epoch": 2.7182292329220203, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8048439025878906, + "num_tokens": 815297272.0, + "step": 21368 + }, + { + "epoch": 2.7183564432006104, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7994769811630249, + "num_tokens": 815336016.0, + "step": 21369 + }, + { + "epoch": 2.718483653479201, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7680113315582275, + "num_tokens": 815373940.0, + "step": 21370 + }, + { + "epoch": 2.7186108637577915, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.787134051322937, + "num_tokens": 815410836.0, + "step": 21371 + }, + { + "epoch": 2.718738074036382, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8068476319313049, + "num_tokens": 815449577.0, + "step": 21372 + }, + { + "epoch": 2.7188652843149725, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8041998147964478, + "num_tokens": 815485869.0, + "step": 21373 + }, + { + "epoch": 2.718992494593563, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7828400135040283, + "num_tokens": 815523614.0, + "step": 21374 + }, + { + "epoch": 2.7191197048721536, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7766960859298706, + "num_tokens": 815561835.0, + "step": 21375 + }, + { + "epoch": 2.719246915150744, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7897899150848389, + "num_tokens": 815605596.0, + "step": 21376 + }, + { + "epoch": 2.7193741254293347, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7968366146087646, + "num_tokens": 815638720.0, + "step": 21377 + }, + { + "epoch": 2.719501335707925, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8172909021377563, + "num_tokens": 815673996.0, + "step": 21378 + }, + { + "epoch": 2.7196285459865157, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7908245325088501, + "num_tokens": 815716571.0, + "step": 21379 + }, + { + "epoch": 2.7197557562651062, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.8004391193389893, + "num_tokens": 815748973.0, + "step": 21380 + }, + { + "epoch": 2.7198829665436968, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8096681833267212, + "num_tokens": 815794754.0, + "step": 21381 + }, + { + "epoch": 2.7200101768222873, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8005611896514893, + "num_tokens": 815831051.0, + "step": 21382 + }, + { + "epoch": 2.720137387100878, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8047996759414673, + "num_tokens": 815874291.0, + "step": 21383 + }, + { + "epoch": 2.7202645973794684, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7955805063247681, + "num_tokens": 815912484.0, + "step": 21384 + }, + { + "epoch": 2.720391807658059, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7833118438720703, + "num_tokens": 815948626.0, + "step": 21385 + }, + { + "epoch": 2.7205190179366494, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.8121477365493774, + "num_tokens": 815983154.0, + "step": 21386 + }, + { + "epoch": 2.72064622821524, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8074048757553101, + "num_tokens": 816017524.0, + "step": 21387 + }, + { + "epoch": 2.7207734384938305, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7996696829795837, + "num_tokens": 816055204.0, + "step": 21388 + }, + { + "epoch": 2.720900648772421, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8086696863174438, + "num_tokens": 816086671.0, + "step": 21389 + }, + { + "epoch": 2.721027859051011, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7752131223678589, + "num_tokens": 816114173.0, + "step": 21390 + }, + { + "epoch": 2.721155069329602, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8188579082489014, + "num_tokens": 816156434.0, + "step": 21391 + }, + { + "epoch": 2.721282279608192, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.796064019203186, + "num_tokens": 816191037.0, + "step": 21392 + }, + { + "epoch": 2.721409489886783, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8201344609260559, + "num_tokens": 816227031.0, + "step": 21393 + }, + { + "epoch": 2.721536700165373, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8078885078430176, + "num_tokens": 816264308.0, + "step": 21394 + }, + { + "epoch": 2.7216639104439637, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7963674068450928, + "num_tokens": 816300955.0, + "step": 21395 + }, + { + "epoch": 2.7217911207225542, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8011070489883423, + "num_tokens": 816332881.0, + "step": 21396 + }, + { + "epoch": 2.7219183310011448, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.8214889764785767, + "num_tokens": 816367446.0, + "step": 21397 + }, + { + "epoch": 2.7220455412797353, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7870419025421143, + "num_tokens": 816407286.0, + "step": 21398 + }, + { + "epoch": 2.722172751558326, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8144474029541016, + "num_tokens": 816445277.0, + "step": 21399 + }, + { + "epoch": 2.7222999618369164, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.8144078850746155, + "num_tokens": 816481638.0, + "step": 21400 + }, + { + "epoch": 2.722427172115507, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8020948171615601, + "num_tokens": 816521386.0, + "step": 21401 + }, + { + "epoch": 2.7225543823940974, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.8092836141586304, + "num_tokens": 816554760.0, + "step": 21402 + }, + { + "epoch": 2.722681592672688, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.807599663734436, + "num_tokens": 816595195.0, + "step": 21403 + }, + { + "epoch": 2.7228088029512785, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8057451248168945, + "num_tokens": 816633046.0, + "step": 21404 + }, + { + "epoch": 2.722936013229869, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8199570178985596, + "num_tokens": 816675283.0, + "step": 21405 + }, + { + "epoch": 2.7230632235084595, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7968783378601074, + "num_tokens": 816716499.0, + "step": 21406 + }, + { + "epoch": 2.72319043378705, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7944051027297974, + "num_tokens": 816761825.0, + "step": 21407 + }, + { + "epoch": 2.7233176440656406, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8051462173461914, + "num_tokens": 816792080.0, + "step": 21408 + }, + { + "epoch": 2.723444854344231, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.795709490776062, + "num_tokens": 816832121.0, + "step": 21409 + }, + { + "epoch": 2.7235720646228216, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.793785572052002, + "num_tokens": 816874032.0, + "step": 21410 + }, + { + "epoch": 2.723699274901412, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7841342687606812, + "num_tokens": 816920471.0, + "step": 21411 + }, + { + "epoch": 2.7238264851800027, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8031326532363892, + "num_tokens": 816958042.0, + "step": 21412 + }, + { + "epoch": 2.723953695458593, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7922077178955078, + "num_tokens": 816992158.0, + "step": 21413 + }, + { + "epoch": 2.7240809057371838, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7871716022491455, + "num_tokens": 817032552.0, + "step": 21414 + }, + { + "epoch": 2.724208116015774, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.8049865961074829, + "num_tokens": 817072240.0, + "step": 21415 + }, + { + "epoch": 2.724335326294365, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7868465185165405, + "num_tokens": 817104741.0, + "step": 21416 + }, + { + "epoch": 2.724462536572955, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7829068303108215, + "num_tokens": 817143876.0, + "step": 21417 + }, + { + "epoch": 2.724589746851546, + "grad_norm": 11.1875, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8255296945571899, + "num_tokens": 817183647.0, + "step": 21418 + }, + { + "epoch": 2.724716957130136, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7916346788406372, + "num_tokens": 817214520.0, + "step": 21419 + }, + { + "epoch": 2.7248441674087265, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7947983145713806, + "num_tokens": 817256936.0, + "step": 21420 + }, + { + "epoch": 2.724971377687317, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8023693561553955, + "num_tokens": 817292840.0, + "step": 21421 + }, + { + "epoch": 2.7250985879659075, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7965089082717896, + "num_tokens": 817332872.0, + "step": 21422 + }, + { + "epoch": 2.725225798244498, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8069765567779541, + "num_tokens": 817367668.0, + "step": 21423 + }, + { + "epoch": 2.7253530085230886, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8250539302825928, + "num_tokens": 817407726.0, + "step": 21424 + }, + { + "epoch": 2.725480218801679, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.8106961250305176, + "num_tokens": 817441003.0, + "step": 21425 + }, + { + "epoch": 2.7256074290802697, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8116759657859802, + "num_tokens": 817482587.0, + "step": 21426 + }, + { + "epoch": 2.72573463935886, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8073635697364807, + "num_tokens": 817519727.0, + "step": 21427 + }, + { + "epoch": 2.7258618496374507, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7984612584114075, + "num_tokens": 817553809.0, + "step": 21428 + }, + { + "epoch": 2.7259890599160412, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7794598937034607, + "num_tokens": 817588893.0, + "step": 21429 + }, + { + "epoch": 2.7261162701946318, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.810736894607544, + "num_tokens": 817627789.0, + "step": 21430 + }, + { + "epoch": 2.7262434804732223, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7916756272315979, + "num_tokens": 817670048.0, + "step": 21431 + }, + { + "epoch": 2.726370690751813, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.80796879529953, + "num_tokens": 817707108.0, + "step": 21432 + }, + { + "epoch": 2.7264979010304033, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.799924910068512, + "num_tokens": 817748126.0, + "step": 21433 + }, + { + "epoch": 2.726625111308994, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8002036809921265, + "num_tokens": 817781578.0, + "step": 21434 + }, + { + "epoch": 2.7267523215875844, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8074840903282166, + "num_tokens": 817822143.0, + "step": 21435 + }, + { + "epoch": 2.726879531866175, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8094632029533386, + "num_tokens": 817858456.0, + "step": 21436 + }, + { + "epoch": 2.7270067421447655, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8118823170661926, + "num_tokens": 817892863.0, + "step": 21437 + }, + { + "epoch": 2.7271339524233555, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8087283372879028, + "num_tokens": 817931120.0, + "step": 21438 + }, + { + "epoch": 2.7272611627019465, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.8005778789520264, + "num_tokens": 817963076.0, + "step": 21439 + }, + { + "epoch": 2.7273883729805366, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8048357963562012, + "num_tokens": 818003844.0, + "step": 21440 + }, + { + "epoch": 2.7275155832591276, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8066047430038452, + "num_tokens": 818048533.0, + "step": 21441 + }, + { + "epoch": 2.7276427935377177, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7957836389541626, + "num_tokens": 818083290.0, + "step": 21442 + }, + { + "epoch": 2.7277700038163086, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8106322288513184, + "num_tokens": 818120037.0, + "step": 21443 + }, + { + "epoch": 2.7278972140948987, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8041943311691284, + "num_tokens": 818164122.0, + "step": 21444 + }, + { + "epoch": 2.7280244243734892, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8025640845298767, + "num_tokens": 818198065.0, + "step": 21445 + }, + { + "epoch": 2.7281516346520798, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8037807941436768, + "num_tokens": 818236843.0, + "step": 21446 + }, + { + "epoch": 2.7282788449306703, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8011003732681274, + "num_tokens": 818271300.0, + "step": 21447 + }, + { + "epoch": 2.728406055209261, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7814036011695862, + "num_tokens": 818311799.0, + "step": 21448 + }, + { + "epoch": 2.7285332654878514, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8066858053207397, + "num_tokens": 818346662.0, + "step": 21449 + }, + { + "epoch": 2.728660475766442, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8050121068954468, + "num_tokens": 818384465.0, + "step": 21450 + }, + { + "epoch": 2.7287876860450324, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.8056613206863403, + "num_tokens": 818425625.0, + "step": 21451 + }, + { + "epoch": 2.728914896323623, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8091297149658203, + "num_tokens": 818458544.0, + "step": 21452 + }, + { + "epoch": 2.7290421066022135, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7972879409790039, + "num_tokens": 818503507.0, + "step": 21453 + }, + { + "epoch": 2.729169316880804, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.816489040851593, + "num_tokens": 818545411.0, + "step": 21454 + }, + { + "epoch": 2.7292965271593945, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7984364032745361, + "num_tokens": 818582100.0, + "step": 21455 + }, + { + "epoch": 2.729423737437985, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7983846068382263, + "num_tokens": 818615549.0, + "step": 21456 + }, + { + "epoch": 2.7295509477165756, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.8079880475997925, + "num_tokens": 818652476.0, + "step": 21457 + }, + { + "epoch": 2.729678157995166, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.8040297031402588, + "num_tokens": 818692435.0, + "step": 21458 + }, + { + "epoch": 2.7298053682737566, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8056487441062927, + "num_tokens": 818733206.0, + "step": 21459 + }, + { + "epoch": 2.729932578552347, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8041451573371887, + "num_tokens": 818771717.0, + "step": 21460 + }, + { + "epoch": 2.7300597888309377, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8030559420585632, + "num_tokens": 818806999.0, + "step": 21461 + }, + { + "epoch": 2.7301869991095282, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8115686178207397, + "num_tokens": 818846226.0, + "step": 21462 + }, + { + "epoch": 2.7303142093881183, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8048779964447021, + "num_tokens": 818885667.0, + "step": 21463 + }, + { + "epoch": 2.7304414196667093, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8118566274642944, + "num_tokens": 818924009.0, + "step": 21464 + }, + { + "epoch": 2.7305686299452994, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.8063420057296753, + "num_tokens": 818962614.0, + "step": 21465 + }, + { + "epoch": 2.7306958402238903, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.803683876991272, + "num_tokens": 819004486.0, + "step": 21466 + }, + { + "epoch": 2.7308230505024804, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7968548536300659, + "num_tokens": 819048364.0, + "step": 21467 + }, + { + "epoch": 2.730950260781071, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7974928617477417, + "num_tokens": 819086250.0, + "step": 21468 + }, + { + "epoch": 2.7310774710596615, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8135356903076172, + "num_tokens": 819126705.0, + "step": 21469 + }, + { + "epoch": 2.731204681338252, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8031013011932373, + "num_tokens": 819165967.0, + "step": 21470 + }, + { + "epoch": 2.7313318916168425, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.8038320541381836, + "num_tokens": 819208066.0, + "step": 21471 + }, + { + "epoch": 2.731459101895433, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8007786870002747, + "num_tokens": 819246269.0, + "step": 21472 + }, + { + "epoch": 2.7315863121740236, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8022099733352661, + "num_tokens": 819286027.0, + "step": 21473 + }, + { + "epoch": 2.731713522452614, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8138551712036133, + "num_tokens": 819319994.0, + "step": 21474 + }, + { + "epoch": 2.7318407327312046, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7990984916687012, + "num_tokens": 819360113.0, + "step": 21475 + }, + { + "epoch": 2.731967943009795, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.8102754950523376, + "num_tokens": 819394575.0, + "step": 21476 + }, + { + "epoch": 2.7320951532883857, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7974820733070374, + "num_tokens": 819434545.0, + "step": 21477 + }, + { + "epoch": 2.7322223635669762, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8121548295021057, + "num_tokens": 819470699.0, + "step": 21478 + }, + { + "epoch": 2.7323495738455668, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7976394295692444, + "num_tokens": 819514447.0, + "step": 21479 + }, + { + "epoch": 2.7324767841241573, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7920223474502563, + "num_tokens": 819553434.0, + "step": 21480 + }, + { + "epoch": 2.732603994402748, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7954814434051514, + "num_tokens": 819591594.0, + "step": 21481 + }, + { + "epoch": 2.7327312046813383, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8013225197792053, + "num_tokens": 819634530.0, + "step": 21482 + }, + { + "epoch": 2.732858414959929, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7914289236068726, + "num_tokens": 819669358.0, + "step": 21483 + }, + { + "epoch": 2.7329856252385194, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8024770021438599, + "num_tokens": 819711336.0, + "step": 21484 + }, + { + "epoch": 2.73311283551711, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7910865545272827, + "num_tokens": 819746973.0, + "step": 21485 + }, + { + "epoch": 2.7332400457957005, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.8068256974220276, + "num_tokens": 819780305.0, + "step": 21486 + }, + { + "epoch": 2.733367256074291, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8141719102859497, + "num_tokens": 819817326.0, + "step": 21487 + }, + { + "epoch": 2.733494466352881, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7947947382926941, + "num_tokens": 819857597.0, + "step": 21488 + }, + { + "epoch": 2.733621676631472, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7976130843162537, + "num_tokens": 819893471.0, + "step": 21489 + }, + { + "epoch": 2.733748886910062, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7861174941062927, + "num_tokens": 819934175.0, + "step": 21490 + }, + { + "epoch": 2.733876097188653, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8080229759216309, + "num_tokens": 819971451.0, + "step": 21491 + }, + { + "epoch": 2.734003307467243, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.799243152141571, + "num_tokens": 820011566.0, + "step": 21492 + }, + { + "epoch": 2.7341305177458337, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7713900804519653, + "num_tokens": 820047944.0, + "step": 21493 + }, + { + "epoch": 2.7342577280244242, + "grad_norm": 50.0, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.78628009557724, + "num_tokens": 820083416.0, + "step": 21494 + }, + { + "epoch": 2.7343849383030148, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7962650060653687, + "num_tokens": 820122476.0, + "step": 21495 + }, + { + "epoch": 2.7345121485816053, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8077078461647034, + "num_tokens": 820159866.0, + "step": 21496 + }, + { + "epoch": 2.734639358860196, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7996830940246582, + "num_tokens": 820202710.0, + "step": 21497 + }, + { + "epoch": 2.7347665691387864, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.8009344935417175, + "num_tokens": 820239156.0, + "step": 21498 + }, + { + "epoch": 2.734893779417377, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.5853, + "mean_token_accuracy": 0.8212864398956299, + "num_tokens": 820274010.0, + "step": 21499 + }, + { + "epoch": 2.7350209896959674, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8038551211357117, + "num_tokens": 820319171.0, + "step": 21500 + }, + { + "epoch": 2.735148199974558, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7805602550506592, + "num_tokens": 820355038.0, + "step": 21501 + }, + { + "epoch": 2.7352754102531485, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7976794242858887, + "num_tokens": 820390973.0, + "step": 21502 + }, + { + "epoch": 2.735402620531739, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7867295742034912, + "num_tokens": 820424543.0, + "step": 21503 + }, + { + "epoch": 2.7355298308103295, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8018434047698975, + "num_tokens": 820462702.0, + "step": 21504 + }, + { + "epoch": 2.73565704108892, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8119261264801025, + "num_tokens": 820498840.0, + "step": 21505 + }, + { + "epoch": 2.7357842513675106, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7993311882019043, + "num_tokens": 820534083.0, + "step": 21506 + }, + { + "epoch": 2.735911461646101, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8041291236877441, + "num_tokens": 820572479.0, + "step": 21507 + }, + { + "epoch": 2.7360386719246916, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7860534191131592, + "num_tokens": 820610283.0, + "step": 21508 + }, + { + "epoch": 2.736165882203282, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.8217577934265137, + "num_tokens": 820647580.0, + "step": 21509 + }, + { + "epoch": 2.7362930924818727, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.8088729977607727, + "num_tokens": 820686774.0, + "step": 21510 + }, + { + "epoch": 2.7364203027604628, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.813065230846405, + "num_tokens": 820724635.0, + "step": 21511 + }, + { + "epoch": 2.7365475130390537, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8055607080459595, + "num_tokens": 820761467.0, + "step": 21512 + }, + { + "epoch": 2.736674723317644, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.786575436592102, + "num_tokens": 820801001.0, + "step": 21513 + }, + { + "epoch": 2.736801933596235, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7985424995422363, + "num_tokens": 820841028.0, + "step": 21514 + }, + { + "epoch": 2.736929143874825, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8113195896148682, + "num_tokens": 820881613.0, + "step": 21515 + }, + { + "epoch": 2.737056354153416, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.813815176486969, + "num_tokens": 820918205.0, + "step": 21516 + }, + { + "epoch": 2.737183564432006, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.792100191116333, + "num_tokens": 820950751.0, + "step": 21517 + }, + { + "epoch": 2.7373107747105965, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8030191659927368, + "num_tokens": 820987385.0, + "step": 21518 + }, + { + "epoch": 2.737437984989187, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7955693006515503, + "num_tokens": 821026273.0, + "step": 21519 + }, + { + "epoch": 2.7375651952677775, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7995432615280151, + "num_tokens": 821067978.0, + "step": 21520 + }, + { + "epoch": 2.737692405546368, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8106766939163208, + "num_tokens": 821111925.0, + "step": 21521 + }, + { + "epoch": 2.7378196158249586, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8071380257606506, + "num_tokens": 821151019.0, + "step": 21522 + }, + { + "epoch": 2.737946826103549, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8084852695465088, + "num_tokens": 821185520.0, + "step": 21523 + }, + { + "epoch": 2.7380740363821396, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8073461055755615, + "num_tokens": 821224273.0, + "step": 21524 + }, + { + "epoch": 2.73820124666073, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8136798143386841, + "num_tokens": 821263541.0, + "step": 21525 + }, + { + "epoch": 2.7383284569393207, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7870789766311646, + "num_tokens": 821304053.0, + "step": 21526 + }, + { + "epoch": 2.7384556672179112, + "grad_norm": 61.25, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.801632821559906, + "num_tokens": 821339710.0, + "step": 21527 + }, + { + "epoch": 2.7385828774965018, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7935284376144409, + "num_tokens": 821378377.0, + "step": 21528 + }, + { + "epoch": 2.7387100877750923, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.813686192035675, + "num_tokens": 821417434.0, + "step": 21529 + }, + { + "epoch": 2.738837298053683, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8021488785743713, + "num_tokens": 821464323.0, + "step": 21530 + }, + { + "epoch": 2.7389645083322733, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8048123121261597, + "num_tokens": 821509891.0, + "step": 21531 + }, + { + "epoch": 2.739091718610864, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.793673574924469, + "num_tokens": 821557309.0, + "step": 21532 + }, + { + "epoch": 2.7392189288894544, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8017293214797974, + "num_tokens": 821597751.0, + "step": 21533 + }, + { + "epoch": 2.739346139168045, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8174595236778259, + "num_tokens": 821637961.0, + "step": 21534 + }, + { + "epoch": 2.7394733494466355, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7848982810974121, + "num_tokens": 821684728.0, + "step": 21535 + }, + { + "epoch": 2.7396005597252255, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8016815185546875, + "num_tokens": 821715359.0, + "step": 21536 + }, + { + "epoch": 2.7397277700038165, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7962883114814758, + "num_tokens": 821755540.0, + "step": 21537 + }, + { + "epoch": 2.7398549802824066, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7991276383399963, + "num_tokens": 821788506.0, + "step": 21538 + }, + { + "epoch": 2.7399821905609976, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.79830002784729, + "num_tokens": 821826555.0, + "step": 21539 + }, + { + "epoch": 2.7401094008395877, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7922497987747192, + "num_tokens": 821866160.0, + "step": 21540 + }, + { + "epoch": 2.7402366111181786, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8073745369911194, + "num_tokens": 821905595.0, + "step": 21541 + }, + { + "epoch": 2.7403638213967687, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7942200899124146, + "num_tokens": 821944769.0, + "step": 21542 + }, + { + "epoch": 2.7404910316753592, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.814126193523407, + "num_tokens": 821988889.0, + "step": 21543 + }, + { + "epoch": 2.7406182419539498, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8109427690505981, + "num_tokens": 822025343.0, + "step": 21544 + }, + { + "epoch": 2.7407454522325403, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7941071391105652, + "num_tokens": 822067651.0, + "step": 21545 + }, + { + "epoch": 2.740872662511131, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7952014207839966, + "num_tokens": 822102885.0, + "step": 21546 + }, + { + "epoch": 2.7409998727897213, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7927764058113098, + "num_tokens": 822142515.0, + "step": 21547 + }, + { + "epoch": 2.741127083068312, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.8027209639549255, + "num_tokens": 822180745.0, + "step": 21548 + }, + { + "epoch": 2.7412542933469024, + "grad_norm": 41.5, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8111346364021301, + "num_tokens": 822215117.0, + "step": 21549 + }, + { + "epoch": 2.741381503625493, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7906959652900696, + "num_tokens": 822258883.0, + "step": 21550 + }, + { + "epoch": 2.7415087139040835, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7971438765525818, + "num_tokens": 822295254.0, + "step": 21551 + }, + { + "epoch": 2.741635924182674, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8007744550704956, + "num_tokens": 822335769.0, + "step": 21552 + }, + { + "epoch": 2.7417631344612645, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.809766948223114, + "num_tokens": 822376873.0, + "step": 21553 + }, + { + "epoch": 2.741890344739855, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7943542003631592, + "num_tokens": 822419868.0, + "step": 21554 + }, + { + "epoch": 2.7420175550184456, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.800685703754425, + "num_tokens": 822456757.0, + "step": 21555 + }, + { + "epoch": 2.742144765297036, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8151282072067261, + "num_tokens": 822492772.0, + "step": 21556 + }, + { + "epoch": 2.7422719755756266, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7849979996681213, + "num_tokens": 822535400.0, + "step": 21557 + }, + { + "epoch": 2.742399185854217, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8088895678520203, + "num_tokens": 822568313.0, + "step": 21558 + }, + { + "epoch": 2.7425263961328077, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8112267255783081, + "num_tokens": 822608487.0, + "step": 21559 + }, + { + "epoch": 2.742653606411398, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8018205761909485, + "num_tokens": 822647938.0, + "step": 21560 + }, + { + "epoch": 2.7427808166899883, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7962208986282349, + "num_tokens": 822685974.0, + "step": 21561 + }, + { + "epoch": 2.7429080269685793, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8065146803855896, + "num_tokens": 822719166.0, + "step": 21562 + }, + { + "epoch": 2.7430352372471694, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7880212068557739, + "num_tokens": 822758364.0, + "step": 21563 + }, + { + "epoch": 2.7431624475257603, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8094594478607178, + "num_tokens": 822798002.0, + "step": 21564 + }, + { + "epoch": 2.7432896578043504, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8052948117256165, + "num_tokens": 822837958.0, + "step": 21565 + }, + { + "epoch": 2.743416868082941, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7979836463928223, + "num_tokens": 822873076.0, + "step": 21566 + }, + { + "epoch": 2.7435440783615315, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7987726330757141, + "num_tokens": 822904693.0, + "step": 21567 + }, + { + "epoch": 2.743671288640122, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7958264350891113, + "num_tokens": 822941192.0, + "step": 21568 + }, + { + "epoch": 2.7437984989187125, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7889943718910217, + "num_tokens": 822982166.0, + "step": 21569 + }, + { + "epoch": 2.743925709197303, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.795052707195282, + "num_tokens": 823021630.0, + "step": 21570 + }, + { + "epoch": 2.7440529194758936, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.8017046451568604, + "num_tokens": 823062133.0, + "step": 21571 + }, + { + "epoch": 2.744180129754484, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7974234819412231, + "num_tokens": 823105163.0, + "step": 21572 + }, + { + "epoch": 2.7443073400330746, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7780778408050537, + "num_tokens": 823146878.0, + "step": 21573 + }, + { + "epoch": 2.744434550311665, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.8183636665344238, + "num_tokens": 823182216.0, + "step": 21574 + }, + { + "epoch": 2.7445617605902557, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.809281587600708, + "num_tokens": 823227838.0, + "step": 21575 + }, + { + "epoch": 2.7446889708688462, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7910414934158325, + "num_tokens": 823268609.0, + "step": 21576 + }, + { + "epoch": 2.7448161811474368, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.8107496500015259, + "num_tokens": 823309426.0, + "step": 21577 + }, + { + "epoch": 2.7449433914260273, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7957596182823181, + "num_tokens": 823346049.0, + "step": 21578 + }, + { + "epoch": 2.745070601704618, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7927823662757874, + "num_tokens": 823378915.0, + "step": 21579 + }, + { + "epoch": 2.7451978119832083, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7950006723403931, + "num_tokens": 823422985.0, + "step": 21580 + }, + { + "epoch": 2.745325022261799, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7836965322494507, + "num_tokens": 823460703.0, + "step": 21581 + }, + { + "epoch": 2.7454522325403894, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8106967210769653, + "num_tokens": 823499651.0, + "step": 21582 + }, + { + "epoch": 2.74557944281898, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7993128299713135, + "num_tokens": 823537044.0, + "step": 21583 + }, + { + "epoch": 2.7457066530975704, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8163384795188904, + "num_tokens": 823571832.0, + "step": 21584 + }, + { + "epoch": 2.745833863376161, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8085480332374573, + "num_tokens": 823609852.0, + "step": 21585 + }, + { + "epoch": 2.745961073654751, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8040050864219666, + "num_tokens": 823644419.0, + "step": 21586 + }, + { + "epoch": 2.746088283933342, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.804216742515564, + "num_tokens": 823682451.0, + "step": 21587 + }, + { + "epoch": 2.746215494211932, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8149890899658203, + "num_tokens": 823723051.0, + "step": 21588 + }, + { + "epoch": 2.746342704490523, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8007737398147583, + "num_tokens": 823761353.0, + "step": 21589 + }, + { + "epoch": 2.746469914769113, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.782335638999939, + "num_tokens": 823800281.0, + "step": 21590 + }, + { + "epoch": 2.7465971250477037, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.598, + "mean_token_accuracy": 0.8212023377418518, + "num_tokens": 823843255.0, + "step": 21591 + }, + { + "epoch": 2.7467243353262942, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7927707433700562, + "num_tokens": 823880736.0, + "step": 21592 + }, + { + "epoch": 2.7468515456048848, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7983437776565552, + "num_tokens": 823923232.0, + "step": 21593 + }, + { + "epoch": 2.7469787558834753, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.808417022228241, + "num_tokens": 823957488.0, + "step": 21594 + }, + { + "epoch": 2.747105966162066, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7964764833450317, + "num_tokens": 823999568.0, + "step": 21595 + }, + { + "epoch": 2.7472331764406563, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8077099919319153, + "num_tokens": 824036947.0, + "step": 21596 + }, + { + "epoch": 2.747360386719247, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8147997260093689, + "num_tokens": 824067966.0, + "step": 21597 + }, + { + "epoch": 2.7474875969978374, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7992456555366516, + "num_tokens": 824112216.0, + "step": 21598 + }, + { + "epoch": 2.747614807276428, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8140971064567566, + "num_tokens": 824148749.0, + "step": 21599 + }, + { + "epoch": 2.7477420175550185, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7996397018432617, + "num_tokens": 824186685.0, + "step": 21600 + }, + { + "epoch": 2.747869227833609, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7959994077682495, + "num_tokens": 824223716.0, + "step": 21601 + }, + { + "epoch": 2.7479964381121995, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8023884892463684, + "num_tokens": 824264627.0, + "step": 21602 + }, + { + "epoch": 2.74812364839079, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.8017762899398804, + "num_tokens": 824301901.0, + "step": 21603 + }, + { + "epoch": 2.7482508586693806, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.8125172853469849, + "num_tokens": 824344621.0, + "step": 21604 + }, + { + "epoch": 2.748378068947971, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.8098545670509338, + "num_tokens": 824385850.0, + "step": 21605 + }, + { + "epoch": 2.7485052792265616, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.546, + "mean_token_accuracy": 0.8319495320320129, + "num_tokens": 824425545.0, + "step": 21606 + }, + { + "epoch": 2.748632489505152, + "grad_norm": 122.0, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8177781701087952, + "num_tokens": 824468611.0, + "step": 21607 + }, + { + "epoch": 2.7487596997837427, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8085354566574097, + "num_tokens": 824512161.0, + "step": 21608 + }, + { + "epoch": 2.7488869100623328, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8149774074554443, + "num_tokens": 824555502.0, + "step": 21609 + }, + { + "epoch": 2.7490141203409237, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7968466281890869, + "num_tokens": 824596277.0, + "step": 21610 + }, + { + "epoch": 2.749141330619514, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.785285472869873, + "num_tokens": 824633704.0, + "step": 21611 + }, + { + "epoch": 2.749268540898105, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8051722049713135, + "num_tokens": 824673584.0, + "step": 21612 + }, + { + "epoch": 2.749395751176695, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7902910709381104, + "num_tokens": 824712508.0, + "step": 21613 + }, + { + "epoch": 2.749522961455286, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8174225091934204, + "num_tokens": 824754353.0, + "step": 21614 + }, + { + "epoch": 2.749650171733876, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7831135988235474, + "num_tokens": 824796333.0, + "step": 21615 + }, + { + "epoch": 2.7497773820124665, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7950439453125, + "num_tokens": 824832048.0, + "step": 21616 + }, + { + "epoch": 2.749904592291057, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7917181253433228, + "num_tokens": 824871889.0, + "step": 21617 + }, + { + "epoch": 2.7500318025696475, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8188728094100952, + "num_tokens": 824912547.0, + "step": 21618 + }, + { + "epoch": 2.750159012848238, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7806358933448792, + "num_tokens": 824955626.0, + "step": 21619 + }, + { + "epoch": 2.7502862231268286, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8040039539337158, + "num_tokens": 824991813.0, + "step": 21620 + }, + { + "epoch": 2.750413433405419, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.811779260635376, + "num_tokens": 825024356.0, + "step": 21621 + }, + { + "epoch": 2.7505406436840096, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7938776612281799, + "num_tokens": 825060316.0, + "step": 21622 + }, + { + "epoch": 2.7506678539626, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.792778730392456, + "num_tokens": 825096044.0, + "step": 21623 + }, + { + "epoch": 2.7507950642411907, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7900076508522034, + "num_tokens": 825123999.0, + "step": 21624 + }, + { + "epoch": 2.750922274519781, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7926470637321472, + "num_tokens": 825170155.0, + "step": 21625 + }, + { + "epoch": 2.7510494847983717, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7979801893234253, + "num_tokens": 825213388.0, + "step": 21626 + }, + { + "epoch": 2.7511766950769623, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8072237968444824, + "num_tokens": 825247171.0, + "step": 21627 + }, + { + "epoch": 2.751303905355553, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7980840802192688, + "num_tokens": 825283271.0, + "step": 21628 + }, + { + "epoch": 2.7514311156341433, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7990326285362244, + "num_tokens": 825318976.0, + "step": 21629 + }, + { + "epoch": 2.751558325912734, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8050117492675781, + "num_tokens": 825358019.0, + "step": 21630 + }, + { + "epoch": 2.7516855361913244, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.818267822265625, + "num_tokens": 825394025.0, + "step": 21631 + }, + { + "epoch": 2.751812746469915, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.8133165836334229, + "num_tokens": 825433021.0, + "step": 21632 + }, + { + "epoch": 2.7519399567485054, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8038589954376221, + "num_tokens": 825477026.0, + "step": 21633 + }, + { + "epoch": 2.7520671670270955, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.808137834072113, + "num_tokens": 825511054.0, + "step": 21634 + }, + { + "epoch": 2.7521943773056865, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7834482789039612, + "num_tokens": 825549654.0, + "step": 21635 + }, + { + "epoch": 2.7523215875842766, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8070145845413208, + "num_tokens": 825587269.0, + "step": 21636 + }, + { + "epoch": 2.7524487978628676, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.5742, + "mean_token_accuracy": 0.8260415196418762, + "num_tokens": 825624864.0, + "step": 21637 + }, + { + "epoch": 2.7525760081414576, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7987401485443115, + "num_tokens": 825667025.0, + "step": 21638 + }, + { + "epoch": 2.7527032184200486, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7860397100448608, + "num_tokens": 825709505.0, + "step": 21639 + }, + { + "epoch": 2.7528304286986387, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.8063560724258423, + "num_tokens": 825747376.0, + "step": 21640 + }, + { + "epoch": 2.7529576389772292, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.80907142162323, + "num_tokens": 825784817.0, + "step": 21641 + }, + { + "epoch": 2.7530848492558198, + "grad_norm": 53.0, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7913766503334045, + "num_tokens": 825815631.0, + "step": 21642 + }, + { + "epoch": 2.7532120595344103, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.8004347085952759, + "num_tokens": 825854202.0, + "step": 21643 + }, + { + "epoch": 2.753339269813001, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8150358200073242, + "num_tokens": 825890636.0, + "step": 21644 + }, + { + "epoch": 2.7534664800915913, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7794157266616821, + "num_tokens": 825931054.0, + "step": 21645 + }, + { + "epoch": 2.753593690370182, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.804029643535614, + "num_tokens": 825972519.0, + "step": 21646 + }, + { + "epoch": 2.7537209006487724, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8059930801391602, + "num_tokens": 826010010.0, + "step": 21647 + }, + { + "epoch": 2.753848110927363, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8067823052406311, + "num_tokens": 826048348.0, + "step": 21648 + }, + { + "epoch": 2.7539753212059535, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7998862862586975, + "num_tokens": 826087807.0, + "step": 21649 + }, + { + "epoch": 2.754102531484544, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7968400120735168, + "num_tokens": 826124657.0, + "step": 21650 + }, + { + "epoch": 2.7542297417631345, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8036530017852783, + "num_tokens": 826164190.0, + "step": 21651 + }, + { + "epoch": 2.754356952041725, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.8025010824203491, + "num_tokens": 826202876.0, + "step": 21652 + }, + { + "epoch": 2.7544841623203156, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7994696497917175, + "num_tokens": 826242955.0, + "step": 21653 + }, + { + "epoch": 2.754611372598906, + "grad_norm": 11.4375, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.8139292001724243, + "num_tokens": 826279925.0, + "step": 21654 + }, + { + "epoch": 2.7547385828774966, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8137102127075195, + "num_tokens": 826314076.0, + "step": 21655 + }, + { + "epoch": 2.754865793156087, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8016824126243591, + "num_tokens": 826354138.0, + "step": 21656 + }, + { + "epoch": 2.7549930034346777, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8011436462402344, + "num_tokens": 826395393.0, + "step": 21657 + }, + { + "epoch": 2.755120213713268, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7919260263442993, + "num_tokens": 826437539.0, + "step": 21658 + }, + { + "epoch": 2.7552474239918583, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7943429350852966, + "num_tokens": 826475857.0, + "step": 21659 + }, + { + "epoch": 2.7553746342704493, + "grad_norm": 9.8125, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8105539083480835, + "num_tokens": 826513956.0, + "step": 21660 + }, + { + "epoch": 2.7555018445490393, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7997739315032959, + "num_tokens": 826552747.0, + "step": 21661 + }, + { + "epoch": 2.7556290548276303, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.8058425188064575, + "num_tokens": 826590830.0, + "step": 21662 + }, + { + "epoch": 2.7557562651062204, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8068298101425171, + "num_tokens": 826633414.0, + "step": 21663 + }, + { + "epoch": 2.755883475384811, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.8097890019416809, + "num_tokens": 826668344.0, + "step": 21664 + }, + { + "epoch": 2.7560106856634015, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7902088761329651, + "num_tokens": 826710587.0, + "step": 21665 + }, + { + "epoch": 2.756137895941992, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.79001784324646, + "num_tokens": 826748854.0, + "step": 21666 + }, + { + "epoch": 2.7562651062205825, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7891119718551636, + "num_tokens": 826789591.0, + "step": 21667 + }, + { + "epoch": 2.756392316499173, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8011926412582397, + "num_tokens": 826831890.0, + "step": 21668 + }, + { + "epoch": 2.7565195267777636, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7823306918144226, + "num_tokens": 826870265.0, + "step": 21669 + }, + { + "epoch": 2.756646737056354, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8024693131446838, + "num_tokens": 826910993.0, + "step": 21670 + }, + { + "epoch": 2.7567739473349446, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7980648279190063, + "num_tokens": 826947492.0, + "step": 21671 + }, + { + "epoch": 2.756901157613535, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8069579005241394, + "num_tokens": 826987215.0, + "step": 21672 + }, + { + "epoch": 2.7570283678921257, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7952485084533691, + "num_tokens": 827024092.0, + "step": 21673 + }, + { + "epoch": 2.757155578170716, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.784881055355072, + "num_tokens": 827060541.0, + "step": 21674 + }, + { + "epoch": 2.7572827884493067, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8027616143226624, + "num_tokens": 827101205.0, + "step": 21675 + }, + { + "epoch": 2.7574099987278973, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7877224087715149, + "num_tokens": 827132441.0, + "step": 21676 + }, + { + "epoch": 2.757537209006488, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7953631281852722, + "num_tokens": 827169530.0, + "step": 21677 + }, + { + "epoch": 2.7576644192850783, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.8038572072982788, + "num_tokens": 827207352.0, + "step": 21678 + }, + { + "epoch": 2.757791629563669, + "grad_norm": 51.25, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7915148735046387, + "num_tokens": 827242790.0, + "step": 21679 + }, + { + "epoch": 2.7579188398422594, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8016132116317749, + "num_tokens": 827281676.0, + "step": 21680 + }, + { + "epoch": 2.75804605012085, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8095341920852661, + "num_tokens": 827313676.0, + "step": 21681 + }, + { + "epoch": 2.7581732603994404, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7997123003005981, + "num_tokens": 827355017.0, + "step": 21682 + }, + { + "epoch": 2.758300470678031, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8178306221961975, + "num_tokens": 827391773.0, + "step": 21683 + }, + { + "epoch": 2.758427680956621, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8204052448272705, + "num_tokens": 827433034.0, + "step": 21684 + }, + { + "epoch": 2.758554891235212, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8140913248062134, + "num_tokens": 827475729.0, + "step": 21685 + }, + { + "epoch": 2.758682101513802, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8040481805801392, + "num_tokens": 827514127.0, + "step": 21686 + }, + { + "epoch": 2.758809311792393, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.8116635084152222, + "num_tokens": 827546709.0, + "step": 21687 + }, + { + "epoch": 2.758936522070983, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.5913, + "mean_token_accuracy": 0.8197789192199707, + "num_tokens": 827581450.0, + "step": 21688 + }, + { + "epoch": 2.7590637323495737, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7993345260620117, + "num_tokens": 827623161.0, + "step": 21689 + }, + { + "epoch": 2.7591909426281642, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7958531379699707, + "num_tokens": 827659460.0, + "step": 21690 + }, + { + "epoch": 2.7593181529067548, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7969871163368225, + "num_tokens": 827701110.0, + "step": 21691 + }, + { + "epoch": 2.7594453631853453, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8001206517219543, + "num_tokens": 827738992.0, + "step": 21692 + }, + { + "epoch": 2.759572573463936, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7877542972564697, + "num_tokens": 827775853.0, + "step": 21693 + }, + { + "epoch": 2.7596997837425263, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.806892991065979, + "num_tokens": 827819438.0, + "step": 21694 + }, + { + "epoch": 2.759826994021117, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7908765077590942, + "num_tokens": 827858833.0, + "step": 21695 + }, + { + "epoch": 2.7599542042997074, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7929970622062683, + "num_tokens": 827895069.0, + "step": 21696 + }, + { + "epoch": 2.760081414578298, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.8159545063972473, + "num_tokens": 827932970.0, + "step": 21697 + }, + { + "epoch": 2.7602086248568884, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.814788281917572, + "num_tokens": 827981859.0, + "step": 21698 + }, + { + "epoch": 2.760335835135479, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7953139543533325, + "num_tokens": 828022018.0, + "step": 21699 + }, + { + "epoch": 2.7604630454140695, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7950018644332886, + "num_tokens": 828061546.0, + "step": 21700 + }, + { + "epoch": 2.76059025569266, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7871493101119995, + "num_tokens": 828104464.0, + "step": 21701 + }, + { + "epoch": 2.7607174659712506, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7918664216995239, + "num_tokens": 828142527.0, + "step": 21702 + }, + { + "epoch": 2.760844676249841, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8093029260635376, + "num_tokens": 828189670.0, + "step": 21703 + }, + { + "epoch": 2.7609718865284316, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8089483976364136, + "num_tokens": 828231511.0, + "step": 21704 + }, + { + "epoch": 2.761099096807022, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8041547536849976, + "num_tokens": 828274387.0, + "step": 21705 + }, + { + "epoch": 2.7612263070856127, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7926971912384033, + "num_tokens": 828312263.0, + "step": 21706 + }, + { + "epoch": 2.7613535173642028, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7990379333496094, + "num_tokens": 828345550.0, + "step": 21707 + }, + { + "epoch": 2.7614807276427937, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7796998023986816, + "num_tokens": 828378633.0, + "step": 21708 + }, + { + "epoch": 2.761607937921384, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7889962196350098, + "num_tokens": 828414292.0, + "step": 21709 + }, + { + "epoch": 2.761735148199975, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.800166130065918, + "num_tokens": 828452682.0, + "step": 21710 + }, + { + "epoch": 2.761862358478565, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8190947771072388, + "num_tokens": 828494162.0, + "step": 21711 + }, + { + "epoch": 2.761989568757156, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8027040362358093, + "num_tokens": 828525997.0, + "step": 21712 + }, + { + "epoch": 2.762116779035746, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7949267625808716, + "num_tokens": 828562488.0, + "step": 21713 + }, + { + "epoch": 2.7622439893143365, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8105971813201904, + "num_tokens": 828591818.0, + "step": 21714 + }, + { + "epoch": 2.762371199592927, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7828997373580933, + "num_tokens": 828628872.0, + "step": 21715 + }, + { + "epoch": 2.7624984098715175, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8116475343704224, + "num_tokens": 828670939.0, + "step": 21716 + }, + { + "epoch": 2.762625620150108, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7820237278938293, + "num_tokens": 828708446.0, + "step": 21717 + }, + { + "epoch": 2.7627528304286986, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8094611167907715, + "num_tokens": 828749873.0, + "step": 21718 + }, + { + "epoch": 2.762880040707289, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7925859689712524, + "num_tokens": 828790066.0, + "step": 21719 + }, + { + "epoch": 2.7630072509858796, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7920072078704834, + "num_tokens": 828821911.0, + "step": 21720 + }, + { + "epoch": 2.76313446126447, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7911103367805481, + "num_tokens": 828856572.0, + "step": 21721 + }, + { + "epoch": 2.7632616715430607, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.8154661655426025, + "num_tokens": 828898865.0, + "step": 21722 + }, + { + "epoch": 2.763388881821651, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8071263432502747, + "num_tokens": 828933768.0, + "step": 21723 + }, + { + "epoch": 2.7635160921002417, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.8179869055747986, + "num_tokens": 828968869.0, + "step": 21724 + }, + { + "epoch": 2.7636433023788323, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7849201560020447, + "num_tokens": 829005621.0, + "step": 21725 + }, + { + "epoch": 2.763770512657423, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7860517501831055, + "num_tokens": 829045046.0, + "step": 21726 + }, + { + "epoch": 2.7638977229360133, + "grad_norm": 69.5, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.8180902004241943, + "num_tokens": 829084588.0, + "step": 21727 + }, + { + "epoch": 2.764024933214604, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.782770037651062, + "num_tokens": 829121010.0, + "step": 21728 + }, + { + "epoch": 2.7641521434931944, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.8135883808135986, + "num_tokens": 829159618.0, + "step": 21729 + }, + { + "epoch": 2.764279353771785, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7875461578369141, + "num_tokens": 829204826.0, + "step": 21730 + }, + { + "epoch": 2.7644065640503754, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8110044598579407, + "num_tokens": 829240741.0, + "step": 21731 + }, + { + "epoch": 2.7645337743289655, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.8216257095336914, + "num_tokens": 829278065.0, + "step": 21732 + }, + { + "epoch": 2.7646609846075565, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7978352308273315, + "num_tokens": 829319438.0, + "step": 21733 + }, + { + "epoch": 2.7647881948861466, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8109586238861084, + "num_tokens": 829357299.0, + "step": 21734 + }, + { + "epoch": 2.7649154051647375, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8022141456604004, + "num_tokens": 829398600.0, + "step": 21735 + }, + { + "epoch": 2.7650426154433276, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7892160415649414, + "num_tokens": 829439214.0, + "step": 21736 + }, + { + "epoch": 2.7651698257219186, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7873765230178833, + "num_tokens": 829477634.0, + "step": 21737 + }, + { + "epoch": 2.7652970360005087, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8087562322616577, + "num_tokens": 829514527.0, + "step": 21738 + }, + { + "epoch": 2.765424246279099, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8049684166908264, + "num_tokens": 829558531.0, + "step": 21739 + }, + { + "epoch": 2.7655514565576897, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8076087236404419, + "num_tokens": 829600769.0, + "step": 21740 + }, + { + "epoch": 2.7656786668362803, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8058179616928101, + "num_tokens": 829635938.0, + "step": 21741 + }, + { + "epoch": 2.765805877114871, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7992082238197327, + "num_tokens": 829672956.0, + "step": 21742 + }, + { + "epoch": 2.7659330873934613, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.8009704351425171, + "num_tokens": 829712658.0, + "step": 21743 + }, + { + "epoch": 2.766060297672052, + "grad_norm": 11.8125, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8066737651824951, + "num_tokens": 829755429.0, + "step": 21744 + }, + { + "epoch": 2.7661875079506424, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.8013512492179871, + "num_tokens": 829796373.0, + "step": 21745 + }, + { + "epoch": 2.766314718229233, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7957485914230347, + "num_tokens": 829834049.0, + "step": 21746 + }, + { + "epoch": 2.7664419285078234, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8030738830566406, + "num_tokens": 829870853.0, + "step": 21747 + }, + { + "epoch": 2.766569138786414, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7941076755523682, + "num_tokens": 829907519.0, + "step": 21748 + }, + { + "epoch": 2.7666963490650045, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7876128554344177, + "num_tokens": 829947162.0, + "step": 21749 + }, + { + "epoch": 2.766823559343595, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.8014810085296631, + "num_tokens": 829991981.0, + "step": 21750 + }, + { + "epoch": 2.7669507696221856, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8078852295875549, + "num_tokens": 830032409.0, + "step": 21751 + }, + { + "epoch": 2.767077979900776, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8085335493087769, + "num_tokens": 830072738.0, + "step": 21752 + }, + { + "epoch": 2.7672051901793666, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7974443435668945, + "num_tokens": 830110880.0, + "step": 21753 + }, + { + "epoch": 2.767332400457957, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7927765250205994, + "num_tokens": 830150148.0, + "step": 21754 + }, + { + "epoch": 2.7674596107365477, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7874703407287598, + "num_tokens": 830190717.0, + "step": 21755 + }, + { + "epoch": 2.767586821015138, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7994058132171631, + "num_tokens": 830226972.0, + "step": 21756 + }, + { + "epoch": 2.7677140312937283, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8111823797225952, + "num_tokens": 830262612.0, + "step": 21757 + }, + { + "epoch": 2.7678412415723193, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.8080222606658936, + "num_tokens": 830306856.0, + "step": 21758 + }, + { + "epoch": 2.7679684518509093, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7995470762252808, + "num_tokens": 830340726.0, + "step": 21759 + }, + { + "epoch": 2.7680956621295003, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.8018186688423157, + "num_tokens": 830380697.0, + "step": 21760 + }, + { + "epoch": 2.7682228724080904, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7832983136177063, + "num_tokens": 830414737.0, + "step": 21761 + }, + { + "epoch": 2.768350082686681, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.802475094795227, + "num_tokens": 830456518.0, + "step": 21762 + }, + { + "epoch": 2.7684772929652715, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7995742559432983, + "num_tokens": 830494192.0, + "step": 21763 + }, + { + "epoch": 2.768604503243862, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7959780693054199, + "num_tokens": 830533659.0, + "step": 21764 + }, + { + "epoch": 2.7687317135224525, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7958766222000122, + "num_tokens": 830569605.0, + "step": 21765 + }, + { + "epoch": 2.768858923801043, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7741738557815552, + "num_tokens": 830610814.0, + "step": 21766 + }, + { + "epoch": 2.7689861340796336, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8024740219116211, + "num_tokens": 830646811.0, + "step": 21767 + }, + { + "epoch": 2.769113344358224, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8039494156837463, + "num_tokens": 830680984.0, + "step": 21768 + }, + { + "epoch": 2.7692405546368146, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8018184900283813, + "num_tokens": 830713384.0, + "step": 21769 + }, + { + "epoch": 2.769367764915405, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7845523357391357, + "num_tokens": 830746922.0, + "step": 21770 + }, + { + "epoch": 2.7694949751939957, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.8154302835464478, + "num_tokens": 830789459.0, + "step": 21771 + }, + { + "epoch": 2.769622185472586, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8010554313659668, + "num_tokens": 830826736.0, + "step": 21772 + }, + { + "epoch": 2.7697493957511767, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8019266128540039, + "num_tokens": 830868118.0, + "step": 21773 + }, + { + "epoch": 2.7698766060297673, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7996264100074768, + "num_tokens": 830911658.0, + "step": 21774 + }, + { + "epoch": 2.770003816308358, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8004913330078125, + "num_tokens": 830948456.0, + "step": 21775 + }, + { + "epoch": 2.7701310265869483, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8096724152565002, + "num_tokens": 830988790.0, + "step": 21776 + }, + { + "epoch": 2.770258236865539, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7916543483734131, + "num_tokens": 831024551.0, + "step": 21777 + }, + { + "epoch": 2.7703854471441294, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7973414659500122, + "num_tokens": 831060082.0, + "step": 21778 + }, + { + "epoch": 2.77051265742272, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8039371967315674, + "num_tokens": 831100067.0, + "step": 21779 + }, + { + "epoch": 2.77063986770131, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8086445927619934, + "num_tokens": 831131923.0, + "step": 21780 + }, + { + "epoch": 2.770767077979901, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7807988524436951, + "num_tokens": 831167985.0, + "step": 21781 + }, + { + "epoch": 2.770894288258491, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7936781644821167, + "num_tokens": 831211737.0, + "step": 21782 + }, + { + "epoch": 2.771021498537082, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7926697731018066, + "num_tokens": 831248764.0, + "step": 21783 + }, + { + "epoch": 2.771148708815672, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8041834831237793, + "num_tokens": 831284140.0, + "step": 21784 + }, + { + "epoch": 2.771275919094263, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8064610958099365, + "num_tokens": 831318567.0, + "step": 21785 + }, + { + "epoch": 2.771403129372853, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7867413759231567, + "num_tokens": 831357128.0, + "step": 21786 + }, + { + "epoch": 2.7715303396514437, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7729030847549438, + "num_tokens": 831391551.0, + "step": 21787 + }, + { + "epoch": 2.771657549930034, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.8005365133285522, + "num_tokens": 831430332.0, + "step": 21788 + }, + { + "epoch": 2.7717847602086247, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.8076861500740051, + "num_tokens": 831471639.0, + "step": 21789 + }, + { + "epoch": 2.7719119704872153, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8054514527320862, + "num_tokens": 831506470.0, + "step": 21790 + }, + { + "epoch": 2.772039180765806, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.8009113073348999, + "num_tokens": 831547212.0, + "step": 21791 + }, + { + "epoch": 2.7721663910443963, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.785173773765564, + "num_tokens": 831593181.0, + "step": 21792 + }, + { + "epoch": 2.772293601322987, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.818094789981842, + "num_tokens": 831624889.0, + "step": 21793 + }, + { + "epoch": 2.7724208116015774, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8073159456253052, + "num_tokens": 831662389.0, + "step": 21794 + }, + { + "epoch": 2.772548021880168, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.8059813976287842, + "num_tokens": 831705282.0, + "step": 21795 + }, + { + "epoch": 2.7726752321587584, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8054330348968506, + "num_tokens": 831741604.0, + "step": 21796 + }, + { + "epoch": 2.772802442437349, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7892855405807495, + "num_tokens": 831784165.0, + "step": 21797 + }, + { + "epoch": 2.7729296527159395, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.8022794723510742, + "num_tokens": 831821101.0, + "step": 21798 + }, + { + "epoch": 2.77305686299453, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8058086037635803, + "num_tokens": 831855766.0, + "step": 21799 + }, + { + "epoch": 2.7731840732731206, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8102480173110962, + "num_tokens": 831895692.0, + "step": 21800 + }, + { + "epoch": 2.773311283551711, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.596, + "mean_token_accuracy": 0.8182141780853271, + "num_tokens": 831931079.0, + "step": 21801 + }, + { + "epoch": 2.7734384938303016, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7753363847732544, + "num_tokens": 831969742.0, + "step": 21802 + }, + { + "epoch": 2.773565704108892, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7931877374649048, + "num_tokens": 832011059.0, + "step": 21803 + }, + { + "epoch": 2.7736929143874827, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.8085365891456604, + "num_tokens": 832042768.0, + "step": 21804 + }, + { + "epoch": 2.7738201246660728, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8057599067687988, + "num_tokens": 832077699.0, + "step": 21805 + }, + { + "epoch": 2.7739473349446637, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8052944540977478, + "num_tokens": 832113528.0, + "step": 21806 + }, + { + "epoch": 2.774074545223254, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8023867607116699, + "num_tokens": 832149108.0, + "step": 21807 + }, + { + "epoch": 2.774201755501845, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8243899345397949, + "num_tokens": 832189279.0, + "step": 21808 + }, + { + "epoch": 2.774328965780435, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.8017478585243225, + "num_tokens": 832224423.0, + "step": 21809 + }, + { + "epoch": 2.774456176059026, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7909696102142334, + "num_tokens": 832260389.0, + "step": 21810 + }, + { + "epoch": 2.774583386337616, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8000637292861938, + "num_tokens": 832297591.0, + "step": 21811 + }, + { + "epoch": 2.7747105966162064, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.803766131401062, + "num_tokens": 832330195.0, + "step": 21812 + }, + { + "epoch": 2.774837806894797, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8006817102432251, + "num_tokens": 832361613.0, + "step": 21813 + }, + { + "epoch": 2.7749650171733875, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.807733416557312, + "num_tokens": 832402800.0, + "step": 21814 + }, + { + "epoch": 2.775092227451978, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8022664189338684, + "num_tokens": 832445161.0, + "step": 21815 + }, + { + "epoch": 2.7752194377305686, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7857517004013062, + "num_tokens": 832487905.0, + "step": 21816 + }, + { + "epoch": 2.775346648009159, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8021470308303833, + "num_tokens": 832521107.0, + "step": 21817 + }, + { + "epoch": 2.7754738582877496, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7926262617111206, + "num_tokens": 832557213.0, + "step": 21818 + }, + { + "epoch": 2.77560106856634, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7884371280670166, + "num_tokens": 832600445.0, + "step": 21819 + }, + { + "epoch": 2.7757282788449307, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8042681813240051, + "num_tokens": 832635994.0, + "step": 21820 + }, + { + "epoch": 2.775855489123521, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.8137551546096802, + "num_tokens": 832677151.0, + "step": 21821 + }, + { + "epoch": 2.7759826994021117, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8028991222381592, + "num_tokens": 832713995.0, + "step": 21822 + }, + { + "epoch": 2.7761099096807023, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7929878830909729, + "num_tokens": 832754201.0, + "step": 21823 + }, + { + "epoch": 2.776237119959293, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7957229018211365, + "num_tokens": 832796920.0, + "step": 21824 + }, + { + "epoch": 2.7763643302378833, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7983747124671936, + "num_tokens": 832835952.0, + "step": 21825 + }, + { + "epoch": 2.776491540516474, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.8031585812568665, + "num_tokens": 832878306.0, + "step": 21826 + }, + { + "epoch": 2.7766187507950644, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7899911403656006, + "num_tokens": 832914315.0, + "step": 21827 + }, + { + "epoch": 2.776745961073655, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7994303703308105, + "num_tokens": 832956661.0, + "step": 21828 + }, + { + "epoch": 2.7768731713522454, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8088309168815613, + "num_tokens": 832999446.0, + "step": 21829 + }, + { + "epoch": 2.7770003816308355, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7741457223892212, + "num_tokens": 833039001.0, + "step": 21830 + }, + { + "epoch": 2.7771275919094265, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7986331582069397, + "num_tokens": 833074334.0, + "step": 21831 + }, + { + "epoch": 2.7772548021880166, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7907039523124695, + "num_tokens": 833113089.0, + "step": 21832 + }, + { + "epoch": 2.7773820124666075, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.795769453048706, + "num_tokens": 833149755.0, + "step": 21833 + }, + { + "epoch": 2.7775092227451976, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7910000681877136, + "num_tokens": 833187718.0, + "step": 21834 + }, + { + "epoch": 2.7776364330237886, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.8087036609649658, + "num_tokens": 833221416.0, + "step": 21835 + }, + { + "epoch": 2.7777636433023787, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8190038800239563, + "num_tokens": 833261101.0, + "step": 21836 + }, + { + "epoch": 2.777890853580969, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.8073915243148804, + "num_tokens": 833296804.0, + "step": 21837 + }, + { + "epoch": 2.7780180638595597, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7963894605636597, + "num_tokens": 833332005.0, + "step": 21838 + }, + { + "epoch": 2.7781452741381503, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7957613468170166, + "num_tokens": 833371991.0, + "step": 21839 + }, + { + "epoch": 2.778272484416741, + "grad_norm": 62.25, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8170520067214966, + "num_tokens": 833407514.0, + "step": 21840 + }, + { + "epoch": 2.7783996946953313, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7963113784790039, + "num_tokens": 833444519.0, + "step": 21841 + }, + { + "epoch": 2.778526904973922, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7966440916061401, + "num_tokens": 833482252.0, + "step": 21842 + }, + { + "epoch": 2.7786541152525124, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7986365556716919, + "num_tokens": 833515513.0, + "step": 21843 + }, + { + "epoch": 2.778781325531103, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8064136505126953, + "num_tokens": 833551188.0, + "step": 21844 + }, + { + "epoch": 2.7789085358096934, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.812549889087677, + "num_tokens": 833589159.0, + "step": 21845 + }, + { + "epoch": 2.779035746088284, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7827146053314209, + "num_tokens": 833623581.0, + "step": 21846 + }, + { + "epoch": 2.7791629563668745, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.790998101234436, + "num_tokens": 833660466.0, + "step": 21847 + }, + { + "epoch": 2.779290166645465, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7849676609039307, + "num_tokens": 833697465.0, + "step": 21848 + }, + { + "epoch": 2.7794173769240555, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7954941391944885, + "num_tokens": 833735178.0, + "step": 21849 + }, + { + "epoch": 2.779544587202646, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.8134855031967163, + "num_tokens": 833770193.0, + "step": 21850 + }, + { + "epoch": 2.7796717974812366, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7979534864425659, + "num_tokens": 833807836.0, + "step": 21851 + }, + { + "epoch": 2.779799007759827, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7961752414703369, + "num_tokens": 833847605.0, + "step": 21852 + }, + { + "epoch": 2.7799262180384177, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8023561835289001, + "num_tokens": 833884239.0, + "step": 21853 + }, + { + "epoch": 2.780053428317008, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7987289428710938, + "num_tokens": 833921305.0, + "step": 21854 + }, + { + "epoch": 2.7801806385955983, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7905024290084839, + "num_tokens": 833959297.0, + "step": 21855 + }, + { + "epoch": 2.7803078488741892, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7751479744911194, + "num_tokens": 833992135.0, + "step": 21856 + }, + { + "epoch": 2.7804350591527793, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7919144630432129, + "num_tokens": 834035933.0, + "step": 21857 + }, + { + "epoch": 2.7805622694313703, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7911205291748047, + "num_tokens": 834070187.0, + "step": 21858 + }, + { + "epoch": 2.7806894797099604, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8067238330841064, + "num_tokens": 834107370.0, + "step": 21859 + }, + { + "epoch": 2.780816689988551, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7927939891815186, + "num_tokens": 834145357.0, + "step": 21860 + }, + { + "epoch": 2.7809439002671414, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8019746541976929, + "num_tokens": 834187250.0, + "step": 21861 + }, + { + "epoch": 2.781071110545732, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8158221244812012, + "num_tokens": 834227598.0, + "step": 21862 + }, + { + "epoch": 2.7811983208243225, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8199042081832886, + "num_tokens": 834267032.0, + "step": 21863 + }, + { + "epoch": 2.781325531102913, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7895687818527222, + "num_tokens": 834308479.0, + "step": 21864 + }, + { + "epoch": 2.7814527413815036, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8062649965286255, + "num_tokens": 834346185.0, + "step": 21865 + }, + { + "epoch": 2.781579951660094, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8088245391845703, + "num_tokens": 834384404.0, + "step": 21866 + }, + { + "epoch": 2.7817071619386846, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8061366081237793, + "num_tokens": 834424833.0, + "step": 21867 + }, + { + "epoch": 2.781834372217275, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8030446767807007, + "num_tokens": 834464215.0, + "step": 21868 + }, + { + "epoch": 2.7819615824958657, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7801627516746521, + "num_tokens": 834502452.0, + "step": 21869 + }, + { + "epoch": 2.782088792774456, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8023200631141663, + "num_tokens": 834539690.0, + "step": 21870 + }, + { + "epoch": 2.7822160030530467, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7875137329101562, + "num_tokens": 834583193.0, + "step": 21871 + }, + { + "epoch": 2.7823432133316373, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.8016259074211121, + "num_tokens": 834620028.0, + "step": 21872 + }, + { + "epoch": 2.782470423610228, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7870572805404663, + "num_tokens": 834657351.0, + "step": 21873 + }, + { + "epoch": 2.7825976338888183, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.5784, + "mean_token_accuracy": 0.8195650577545166, + "num_tokens": 834693687.0, + "step": 21874 + }, + { + "epoch": 2.782724844167409, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7991241216659546, + "num_tokens": 834736171.0, + "step": 21875 + }, + { + "epoch": 2.7828520544459994, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8169277906417847, + "num_tokens": 834775699.0, + "step": 21876 + }, + { + "epoch": 2.78297926472459, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7879867553710938, + "num_tokens": 834816053.0, + "step": 21877 + }, + { + "epoch": 2.78310647500318, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7945824861526489, + "num_tokens": 834857125.0, + "step": 21878 + }, + { + "epoch": 2.783233685281771, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8114138245582581, + "num_tokens": 834895676.0, + "step": 21879 + }, + { + "epoch": 2.783360895560361, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8106458187103271, + "num_tokens": 834927748.0, + "step": 21880 + }, + { + "epoch": 2.783488105838952, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7973591089248657, + "num_tokens": 834966722.0, + "step": 21881 + }, + { + "epoch": 2.783615316117542, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7982736825942993, + "num_tokens": 835003770.0, + "step": 21882 + }, + { + "epoch": 2.783742526396133, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8088052272796631, + "num_tokens": 835041504.0, + "step": 21883 + }, + { + "epoch": 2.783869736674723, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8016023635864258, + "num_tokens": 835075179.0, + "step": 21884 + }, + { + "epoch": 2.7839969469533137, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7887767553329468, + "num_tokens": 835113080.0, + "step": 21885 + }, + { + "epoch": 2.784124157231904, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7972632050514221, + "num_tokens": 835150586.0, + "step": 21886 + }, + { + "epoch": 2.7842513675104947, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7925835847854614, + "num_tokens": 835184049.0, + "step": 21887 + }, + { + "epoch": 2.7843785777890853, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7974585294723511, + "num_tokens": 835222745.0, + "step": 21888 + }, + { + "epoch": 2.784505788067676, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7896575331687927, + "num_tokens": 835263017.0, + "step": 21889 + }, + { + "epoch": 2.7846329983462663, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8155261278152466, + "num_tokens": 835303916.0, + "step": 21890 + }, + { + "epoch": 2.784760208624857, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8171477913856506, + "num_tokens": 835343148.0, + "step": 21891 + }, + { + "epoch": 2.7848874189034474, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8002340793609619, + "num_tokens": 835382337.0, + "step": 21892 + }, + { + "epoch": 2.785014629182038, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7991672158241272, + "num_tokens": 835423645.0, + "step": 21893 + }, + { + "epoch": 2.7851418394606284, + "grad_norm": 60.75, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.767181396484375, + "num_tokens": 835464590.0, + "step": 21894 + }, + { + "epoch": 2.785269049739219, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.5696, + "mean_token_accuracy": 0.830018162727356, + "num_tokens": 835496376.0, + "step": 21895 + }, + { + "epoch": 2.7853962600178095, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.784366250038147, + "num_tokens": 835541231.0, + "step": 21896 + }, + { + "epoch": 2.7855234702964, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.8053452968597412, + "num_tokens": 835584038.0, + "step": 21897 + }, + { + "epoch": 2.7856506805749905, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7933094501495361, + "num_tokens": 835627189.0, + "step": 21898 + }, + { + "epoch": 2.785777890853581, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8039492964744568, + "num_tokens": 835659701.0, + "step": 21899 + }, + { + "epoch": 2.7859051011321716, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8039119243621826, + "num_tokens": 835694045.0, + "step": 21900 + }, + { + "epoch": 2.786032311410762, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8016610741615295, + "num_tokens": 835732118.0, + "step": 21901 + }, + { + "epoch": 2.7861595216893527, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8111166954040527, + "num_tokens": 835772700.0, + "step": 21902 + }, + { + "epoch": 2.7862867319679427, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7826618552207947, + "num_tokens": 835811206.0, + "step": 21903 + }, + { + "epoch": 2.7864139422465337, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8023596405982971, + "num_tokens": 835844381.0, + "step": 21904 + }, + { + "epoch": 2.786541152525124, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7910621166229248, + "num_tokens": 835879926.0, + "step": 21905 + }, + { + "epoch": 2.7866683628037148, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7930197715759277, + "num_tokens": 835924014.0, + "step": 21906 + }, + { + "epoch": 2.786795573082305, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7929458618164062, + "num_tokens": 835956054.0, + "step": 21907 + }, + { + "epoch": 2.786922783360896, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.8014390468597412, + "num_tokens": 835988885.0, + "step": 21908 + }, + { + "epoch": 2.787049993639486, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8023526668548584, + "num_tokens": 836028885.0, + "step": 21909 + }, + { + "epoch": 2.7871772039180764, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8106643557548523, + "num_tokens": 836067340.0, + "step": 21910 + }, + { + "epoch": 2.787304414196667, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7960672974586487, + "num_tokens": 836104432.0, + "step": 21911 + }, + { + "epoch": 2.7874316244752575, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7956001162528992, + "num_tokens": 836141197.0, + "step": 21912 + }, + { + "epoch": 2.787558834753848, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.8174358606338501, + "num_tokens": 836176949.0, + "step": 21913 + }, + { + "epoch": 2.7876860450324386, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7932618856430054, + "num_tokens": 836206484.0, + "step": 21914 + }, + { + "epoch": 2.787813255311029, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8039728403091431, + "num_tokens": 836243056.0, + "step": 21915 + }, + { + "epoch": 2.7879404655896196, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7917321920394897, + "num_tokens": 836281154.0, + "step": 21916 + }, + { + "epoch": 2.78806767586821, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7978947162628174, + "num_tokens": 836319895.0, + "step": 21917 + }, + { + "epoch": 2.7881948861468007, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.805503249168396, + "num_tokens": 836357227.0, + "step": 21918 + }, + { + "epoch": 2.788322096425391, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7897698283195496, + "num_tokens": 836398442.0, + "step": 21919 + }, + { + "epoch": 2.7884493067039817, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.8104394674301147, + "num_tokens": 836436660.0, + "step": 21920 + }, + { + "epoch": 2.7885765169825723, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7957088947296143, + "num_tokens": 836479180.0, + "step": 21921 + }, + { + "epoch": 2.788703727261163, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8091299533843994, + "num_tokens": 836518874.0, + "step": 21922 + }, + { + "epoch": 2.7888309375397533, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.789555549621582, + "num_tokens": 836556910.0, + "step": 21923 + }, + { + "epoch": 2.788958147818344, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8127961158752441, + "num_tokens": 836598151.0, + "step": 21924 + }, + { + "epoch": 2.7890853580969344, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7929455041885376, + "num_tokens": 836637952.0, + "step": 21925 + }, + { + "epoch": 2.789212568375525, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8109403848648071, + "num_tokens": 836677138.0, + "step": 21926 + }, + { + "epoch": 2.7893397786541154, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.809960663318634, + "num_tokens": 836716880.0, + "step": 21927 + }, + { + "epoch": 2.7894669889327055, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.8139382004737854, + "num_tokens": 836752572.0, + "step": 21928 + }, + { + "epoch": 2.7895941992112965, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7941832542419434, + "num_tokens": 836794321.0, + "step": 21929 + }, + { + "epoch": 2.7897214094898866, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8105984926223755, + "num_tokens": 836829183.0, + "step": 21930 + }, + { + "epoch": 2.7898486197684775, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.789113461971283, + "num_tokens": 836864644.0, + "step": 21931 + }, + { + "epoch": 2.7899758300470676, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8122258186340332, + "num_tokens": 836895384.0, + "step": 21932 + }, + { + "epoch": 2.7901030403256586, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8070902824401855, + "num_tokens": 836943336.0, + "step": 21933 + }, + { + "epoch": 2.7902302506042487, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8188754320144653, + "num_tokens": 836980708.0, + "step": 21934 + }, + { + "epoch": 2.790357460882839, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.8015541434288025, + "num_tokens": 837015738.0, + "step": 21935 + }, + { + "epoch": 2.7904846711614297, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8129788637161255, + "num_tokens": 837053014.0, + "step": 21936 + }, + { + "epoch": 2.7906118814400203, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7944762706756592, + "num_tokens": 837083579.0, + "step": 21937 + }, + { + "epoch": 2.790739091718611, + "grad_norm": 10.5625, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8118162155151367, + "num_tokens": 837126201.0, + "step": 21938 + }, + { + "epoch": 2.7908663019972013, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.8151718378067017, + "num_tokens": 837162400.0, + "step": 21939 + }, + { + "epoch": 2.790993512275792, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8051695823669434, + "num_tokens": 837205733.0, + "step": 21940 + }, + { + "epoch": 2.7911207225543824, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7888285517692566, + "num_tokens": 837243838.0, + "step": 21941 + }, + { + "epoch": 2.791247932832973, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7972345948219299, + "num_tokens": 837280936.0, + "step": 21942 + }, + { + "epoch": 2.7913751431115634, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8013768196105957, + "num_tokens": 837321790.0, + "step": 21943 + }, + { + "epoch": 2.791502353390154, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8080229759216309, + "num_tokens": 837362023.0, + "step": 21944 + }, + { + "epoch": 2.7916295636687445, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8148743510246277, + "num_tokens": 837404876.0, + "step": 21945 + }, + { + "epoch": 2.791756773947335, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7952330112457275, + "num_tokens": 837442899.0, + "step": 21946 + }, + { + "epoch": 2.7918839842259255, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8099850416183472, + "num_tokens": 837477611.0, + "step": 21947 + }, + { + "epoch": 2.792011194504516, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.785696804523468, + "num_tokens": 837515928.0, + "step": 21948 + }, + { + "epoch": 2.7921384047831066, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7984737753868103, + "num_tokens": 837553042.0, + "step": 21949 + }, + { + "epoch": 2.792265615061697, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.813475489616394, + "num_tokens": 837589749.0, + "step": 21950 + }, + { + "epoch": 2.7923928253402877, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.8281453251838684, + "num_tokens": 837631104.0, + "step": 21951 + }, + { + "epoch": 2.792520035618878, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.810196042060852, + "num_tokens": 837668173.0, + "step": 21952 + }, + { + "epoch": 2.7926472458974683, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8081821799278259, + "num_tokens": 837704282.0, + "step": 21953 + }, + { + "epoch": 2.7927744561760592, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7950038313865662, + "num_tokens": 837750790.0, + "step": 21954 + }, + { + "epoch": 2.7929016664546493, + "grad_norm": 46.0, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8000471591949463, + "num_tokens": 837792041.0, + "step": 21955 + }, + { + "epoch": 2.7930288767332403, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.8021558523178101, + "num_tokens": 837818645.0, + "step": 21956 + }, + { + "epoch": 2.7931560870118304, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.8125450611114502, + "num_tokens": 837859816.0, + "step": 21957 + }, + { + "epoch": 2.793283297290421, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7960776090621948, + "num_tokens": 837903970.0, + "step": 21958 + }, + { + "epoch": 2.7934105075690114, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7986496090888977, + "num_tokens": 837941158.0, + "step": 21959 + }, + { + "epoch": 2.793537717847602, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8112403154373169, + "num_tokens": 837978157.0, + "step": 21960 + }, + { + "epoch": 2.7936649281261925, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.8207275867462158, + "num_tokens": 838017871.0, + "step": 21961 + }, + { + "epoch": 2.793792138404783, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.8108817934989929, + "num_tokens": 838050521.0, + "step": 21962 + }, + { + "epoch": 2.7939193486833735, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8094008564949036, + "num_tokens": 838084450.0, + "step": 21963 + }, + { + "epoch": 2.794046558961964, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7936260104179382, + "num_tokens": 838121505.0, + "step": 21964 + }, + { + "epoch": 2.7941737692405546, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8006278872489929, + "num_tokens": 838166148.0, + "step": 21965 + }, + { + "epoch": 2.794300979519145, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8021142482757568, + "num_tokens": 838203578.0, + "step": 21966 + }, + { + "epoch": 2.7944281897977357, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8034851551055908, + "num_tokens": 838240702.0, + "step": 21967 + }, + { + "epoch": 2.794555400076326, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8086261749267578, + "num_tokens": 838275639.0, + "step": 21968 + }, + { + "epoch": 2.7946826103549167, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7985594868659973, + "num_tokens": 838310045.0, + "step": 21969 + }, + { + "epoch": 2.7948098206335072, + "grad_norm": 9.4375, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.8016377091407776, + "num_tokens": 838351571.0, + "step": 21970 + }, + { + "epoch": 2.7949370309120978, + "grad_norm": 12.4375, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8168655633926392, + "num_tokens": 838393804.0, + "step": 21971 + }, + { + "epoch": 2.7950642411906883, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8083129525184631, + "num_tokens": 838435516.0, + "step": 21972 + }, + { + "epoch": 2.795191451469279, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8038299083709717, + "num_tokens": 838465425.0, + "step": 21973 + }, + { + "epoch": 2.7953186617478694, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.8031866550445557, + "num_tokens": 838512832.0, + "step": 21974 + }, + { + "epoch": 2.79544587202646, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8067989349365234, + "num_tokens": 838552806.0, + "step": 21975 + }, + { + "epoch": 2.79557308230505, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.8031032085418701, + "num_tokens": 838588592.0, + "step": 21976 + }, + { + "epoch": 2.795700292583641, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.8005654811859131, + "num_tokens": 838629402.0, + "step": 21977 + }, + { + "epoch": 2.795827502862231, + "grad_norm": 42.0, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7933725118637085, + "num_tokens": 838676580.0, + "step": 21978 + }, + { + "epoch": 2.795954713140822, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.796966016292572, + "num_tokens": 838715773.0, + "step": 21979 + }, + { + "epoch": 2.796081923419412, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7898960113525391, + "num_tokens": 838750270.0, + "step": 21980 + }, + { + "epoch": 2.796209133698003, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7807186841964722, + "num_tokens": 838794823.0, + "step": 21981 + }, + { + "epoch": 2.796336343976593, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7876962423324585, + "num_tokens": 838828901.0, + "step": 21982 + }, + { + "epoch": 2.7964635542551837, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7975746989250183, + "num_tokens": 838872648.0, + "step": 21983 + }, + { + "epoch": 2.796590764533774, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8124424815177917, + "num_tokens": 838905150.0, + "step": 21984 + }, + { + "epoch": 2.7967179748123647, + "grad_norm": 28.875, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8044599294662476, + "num_tokens": 838949027.0, + "step": 21985 + }, + { + "epoch": 2.7968451850909553, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7965904474258423, + "num_tokens": 838987320.0, + "step": 21986 + }, + { + "epoch": 2.796972395369546, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.8001174926757812, + "num_tokens": 839027396.0, + "step": 21987 + }, + { + "epoch": 2.7970996056481363, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8098905086517334, + "num_tokens": 839069779.0, + "step": 21988 + }, + { + "epoch": 2.797226815926727, + "grad_norm": 10.375, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8123328685760498, + "num_tokens": 839110919.0, + "step": 21989 + }, + { + "epoch": 2.7973540262053174, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.8158490657806396, + "num_tokens": 839143689.0, + "step": 21990 + }, + { + "epoch": 2.797481236483908, + "grad_norm": 61.5, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7801328301429749, + "num_tokens": 839181274.0, + "step": 21991 + }, + { + "epoch": 2.7976084467624984, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7858130931854248, + "num_tokens": 839217038.0, + "step": 21992 + }, + { + "epoch": 2.797735657041089, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8062640428543091, + "num_tokens": 839260104.0, + "step": 21993 + }, + { + "epoch": 2.7978628673196795, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.8022547960281372, + "num_tokens": 839294328.0, + "step": 21994 + }, + { + "epoch": 2.79799007759827, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7987425327301025, + "num_tokens": 839326662.0, + "step": 21995 + }, + { + "epoch": 2.7981172878768605, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7995041012763977, + "num_tokens": 839355997.0, + "step": 21996 + }, + { + "epoch": 2.798244498155451, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.8027894496917725, + "num_tokens": 839396596.0, + "step": 21997 + }, + { + "epoch": 2.7983717084340416, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.8056865334510803, + "num_tokens": 839432251.0, + "step": 21998 + }, + { + "epoch": 2.798498918712632, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8118326663970947, + "num_tokens": 839465987.0, + "step": 21999 + }, + { + "epoch": 2.7986261289912227, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8010073304176331, + "num_tokens": 839506147.0, + "step": 22000 + }, + { + "epoch": 2.7987533392698127, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7972947955131531, + "num_tokens": 839544620.0, + "step": 22001 + }, + { + "epoch": 2.7988805495484037, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7913967967033386, + "num_tokens": 839583217.0, + "step": 22002 + }, + { + "epoch": 2.799007759826994, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8239532709121704, + "num_tokens": 839621283.0, + "step": 22003 + }, + { + "epoch": 2.7991349701055848, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7907898426055908, + "num_tokens": 839658072.0, + "step": 22004 + }, + { + "epoch": 2.799262180384175, + "grad_norm": 55.75, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7849308252334595, + "num_tokens": 839700926.0, + "step": 22005 + }, + { + "epoch": 2.799389390662766, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8099243640899658, + "num_tokens": 839736160.0, + "step": 22006 + }, + { + "epoch": 2.799516600941356, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8191239833831787, + "num_tokens": 839776834.0, + "step": 22007 + }, + { + "epoch": 2.7996438112199464, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7950197458267212, + "num_tokens": 839818156.0, + "step": 22008 + }, + { + "epoch": 2.799771021498537, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.799254298210144, + "num_tokens": 839857133.0, + "step": 22009 + }, + { + "epoch": 2.7998982317771275, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7834373116493225, + "num_tokens": 839893129.0, + "step": 22010 + }, + { + "epoch": 2.800025442055718, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7971580624580383, + "num_tokens": 839924622.0, + "step": 22011 + }, + { + "epoch": 2.8001526523343085, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8186960220336914, + "num_tokens": 839961276.0, + "step": 22012 + }, + { + "epoch": 2.800279862612899, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7943921089172363, + "num_tokens": 840002554.0, + "step": 22013 + }, + { + "epoch": 2.8004070728914896, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8034912347793579, + "num_tokens": 840041642.0, + "step": 22014 + }, + { + "epoch": 2.80053428317008, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8110896348953247, + "num_tokens": 840083532.0, + "step": 22015 + }, + { + "epoch": 2.8006614934486707, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7975742220878601, + "num_tokens": 840124221.0, + "step": 22016 + }, + { + "epoch": 2.800788703727261, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8047914505004883, + "num_tokens": 840162309.0, + "step": 22017 + }, + { + "epoch": 2.8009159140058517, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7983830571174622, + "num_tokens": 840194778.0, + "step": 22018 + }, + { + "epoch": 2.8010431242844422, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8175855875015259, + "num_tokens": 840225081.0, + "step": 22019 + }, + { + "epoch": 2.8011703345630328, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8018368482589722, + "num_tokens": 840267698.0, + "step": 22020 + }, + { + "epoch": 2.8012975448416233, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7940375804901123, + "num_tokens": 840303255.0, + "step": 22021 + }, + { + "epoch": 2.801424755120214, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8046877384185791, + "num_tokens": 840344616.0, + "step": 22022 + }, + { + "epoch": 2.8015519653988044, + "grad_norm": 42.5, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7955962419509888, + "num_tokens": 840383293.0, + "step": 22023 + }, + { + "epoch": 2.801679175677395, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7950032353401184, + "num_tokens": 840424583.0, + "step": 22024 + }, + { + "epoch": 2.8018063859559854, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7914859056472778, + "num_tokens": 840458752.0, + "step": 22025 + }, + { + "epoch": 2.8019335962345755, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8073394894599915, + "num_tokens": 840498441.0, + "step": 22026 + }, + { + "epoch": 2.8020608065131665, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.8106591701507568, + "num_tokens": 840537838.0, + "step": 22027 + }, + { + "epoch": 2.8021880167917566, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7862065434455872, + "num_tokens": 840571770.0, + "step": 22028 + }, + { + "epoch": 2.8023152270703475, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7918409109115601, + "num_tokens": 840607559.0, + "step": 22029 + }, + { + "epoch": 2.8024424373489376, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7858421802520752, + "num_tokens": 840650156.0, + "step": 22030 + }, + { + "epoch": 2.8025696476275286, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7929655909538269, + "num_tokens": 840686590.0, + "step": 22031 + }, + { + "epoch": 2.8026968579061187, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7885820865631104, + "num_tokens": 840724445.0, + "step": 22032 + }, + { + "epoch": 2.802824068184709, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8147322535514832, + "num_tokens": 840765047.0, + "step": 22033 + }, + { + "epoch": 2.8029512784632997, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7947942018508911, + "num_tokens": 840802375.0, + "step": 22034 + }, + { + "epoch": 2.8030784887418903, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.800991952419281, + "num_tokens": 840839994.0, + "step": 22035 + }, + { + "epoch": 2.803205699020481, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7983880043029785, + "num_tokens": 840877144.0, + "step": 22036 + }, + { + "epoch": 2.8033329092990713, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7910298109054565, + "num_tokens": 840918478.0, + "step": 22037 + }, + { + "epoch": 2.803460119577662, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.806108295917511, + "num_tokens": 840958484.0, + "step": 22038 + }, + { + "epoch": 2.8035873298562524, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8047400116920471, + "num_tokens": 841000035.0, + "step": 22039 + }, + { + "epoch": 2.803714540134843, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7950517535209656, + "num_tokens": 841035836.0, + "step": 22040 + }, + { + "epoch": 2.8038417504134334, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7998592853546143, + "num_tokens": 841069295.0, + "step": 22041 + }, + { + "epoch": 2.803968960692024, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7937577962875366, + "num_tokens": 841108341.0, + "step": 22042 + }, + { + "epoch": 2.8040961709706145, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.8120823502540588, + "num_tokens": 841142945.0, + "step": 22043 + }, + { + "epoch": 2.804223381249205, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8106386661529541, + "num_tokens": 841174137.0, + "step": 22044 + }, + { + "epoch": 2.8043505915277955, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.8086225986480713, + "num_tokens": 841215914.0, + "step": 22045 + }, + { + "epoch": 2.804477801806386, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8006604909896851, + "num_tokens": 841256097.0, + "step": 22046 + }, + { + "epoch": 2.8046050120849766, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8043532371520996, + "num_tokens": 841288997.0, + "step": 22047 + }, + { + "epoch": 2.804732222363567, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.795203447341919, + "num_tokens": 841328053.0, + "step": 22048 + }, + { + "epoch": 2.8048594326421576, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8182487487792969, + "num_tokens": 841366848.0, + "step": 22049 + }, + { + "epoch": 2.804986642920748, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.807335615158081, + "num_tokens": 841398568.0, + "step": 22050 + }, + { + "epoch": 2.8051138531993383, + "grad_norm": 12.3125, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8035129904747009, + "num_tokens": 841435722.0, + "step": 22051 + }, + { + "epoch": 2.8052410634779292, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7957220077514648, + "num_tokens": 841473635.0, + "step": 22052 + }, + { + "epoch": 2.8053682737565193, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8096277713775635, + "num_tokens": 841512383.0, + "step": 22053 + }, + { + "epoch": 2.8054954840351103, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7883731126785278, + "num_tokens": 841543174.0, + "step": 22054 + }, + { + "epoch": 2.8056226943137004, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.8047176599502563, + "num_tokens": 841582185.0, + "step": 22055 + }, + { + "epoch": 2.805749904592291, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7961300611495972, + "num_tokens": 841623179.0, + "step": 22056 + }, + { + "epoch": 2.8058771148708814, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8102269172668457, + "num_tokens": 841661062.0, + "step": 22057 + }, + { + "epoch": 2.806004325149472, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7923028469085693, + "num_tokens": 841704259.0, + "step": 22058 + }, + { + "epoch": 2.8061315354280625, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.8046594262123108, + "num_tokens": 841748396.0, + "step": 22059 + }, + { + "epoch": 2.806258745706653, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.5785, + "mean_token_accuracy": 0.8225551247596741, + "num_tokens": 841787809.0, + "step": 22060 + }, + { + "epoch": 2.8063859559852435, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.8000925779342651, + "num_tokens": 841824046.0, + "step": 22061 + }, + { + "epoch": 2.806513166263834, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8112704753875732, + "num_tokens": 841856949.0, + "step": 22062 + }, + { + "epoch": 2.8066403765424246, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7843212485313416, + "num_tokens": 841898644.0, + "step": 22063 + }, + { + "epoch": 2.806767586821015, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.8079958558082581, + "num_tokens": 841934795.0, + "step": 22064 + }, + { + "epoch": 2.8068947970996057, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7884032726287842, + "num_tokens": 841976878.0, + "step": 22065 + }, + { + "epoch": 2.807022007378196, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.8005534410476685, + "num_tokens": 842010998.0, + "step": 22066 + }, + { + "epoch": 2.8071492176567867, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.5466, + "mean_token_accuracy": 0.8328314423561096, + "num_tokens": 842051410.0, + "step": 22067 + }, + { + "epoch": 2.8072764279353772, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8045328855514526, + "num_tokens": 842087954.0, + "step": 22068 + }, + { + "epoch": 2.8074036382139678, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7981045246124268, + "num_tokens": 842128326.0, + "step": 22069 + }, + { + "epoch": 2.8075308484925583, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7802273035049438, + "num_tokens": 842171143.0, + "step": 22070 + }, + { + "epoch": 2.807658058771149, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7861173748970032, + "num_tokens": 842212167.0, + "step": 22071 + }, + { + "epoch": 2.8077852690497394, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7941044569015503, + "num_tokens": 842252576.0, + "step": 22072 + }, + { + "epoch": 2.80791247932833, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7984655499458313, + "num_tokens": 842287812.0, + "step": 22073 + }, + { + "epoch": 2.80803968960692, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7931749820709229, + "num_tokens": 842328289.0, + "step": 22074 + }, + { + "epoch": 2.808166899885511, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7991260886192322, + "num_tokens": 842359182.0, + "step": 22075 + }, + { + "epoch": 2.808294110164101, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7955093383789062, + "num_tokens": 842399820.0, + "step": 22076 + }, + { + "epoch": 2.808421320442692, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.791481614112854, + "num_tokens": 842437852.0, + "step": 22077 + }, + { + "epoch": 2.808548530721282, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7860881090164185, + "num_tokens": 842476729.0, + "step": 22078 + }, + { + "epoch": 2.808675740999873, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7918051481246948, + "num_tokens": 842514590.0, + "step": 22079 + }, + { + "epoch": 2.808802951278463, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.8015788793563843, + "num_tokens": 842556502.0, + "step": 22080 + }, + { + "epoch": 2.8089301615570537, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7859969139099121, + "num_tokens": 842594299.0, + "step": 22081 + }, + { + "epoch": 2.809057371835644, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7988215088844299, + "num_tokens": 842633189.0, + "step": 22082 + }, + { + "epoch": 2.8091845821142347, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.8085789680480957, + "num_tokens": 842673626.0, + "step": 22083 + }, + { + "epoch": 2.8093117923928252, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7962708473205566, + "num_tokens": 842712574.0, + "step": 22084 + }, + { + "epoch": 2.8094390026714158, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7951545715332031, + "num_tokens": 842749534.0, + "step": 22085 + }, + { + "epoch": 2.8095662129500063, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7932689189910889, + "num_tokens": 842784989.0, + "step": 22086 + }, + { + "epoch": 2.809693423228597, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7950378656387329, + "num_tokens": 842823858.0, + "step": 22087 + }, + { + "epoch": 2.8098206335071874, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8006271719932556, + "num_tokens": 842858849.0, + "step": 22088 + }, + { + "epoch": 2.809947843785778, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.8225653171539307, + "num_tokens": 842895809.0, + "step": 22089 + }, + { + "epoch": 2.8100750540643684, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.5395, + "mean_token_accuracy": 0.8334919810295105, + "num_tokens": 842933650.0, + "step": 22090 + }, + { + "epoch": 2.810202264342959, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8077347278594971, + "num_tokens": 842967200.0, + "step": 22091 + }, + { + "epoch": 2.8103294746215495, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7925170660018921, + "num_tokens": 843002498.0, + "step": 22092 + }, + { + "epoch": 2.81045668490014, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8136559724807739, + "num_tokens": 843042627.0, + "step": 22093 + }, + { + "epoch": 2.8105838951787305, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7962530255317688, + "num_tokens": 843076862.0, + "step": 22094 + }, + { + "epoch": 2.810711105457321, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8006856441497803, + "num_tokens": 843115000.0, + "step": 22095 + }, + { + "epoch": 2.8108383157359116, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.8083693981170654, + "num_tokens": 843152408.0, + "step": 22096 + }, + { + "epoch": 2.810965526014502, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.804121732711792, + "num_tokens": 843189269.0, + "step": 22097 + }, + { + "epoch": 2.8110927362930926, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.800163984298706, + "num_tokens": 843227148.0, + "step": 22098 + }, + { + "epoch": 2.8112199465716827, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7872403860092163, + "num_tokens": 843259736.0, + "step": 22099 + }, + { + "epoch": 2.8113471568502737, + "grad_norm": 12.0625, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7981356382369995, + "num_tokens": 843302598.0, + "step": 22100 + }, + { + "epoch": 2.811474367128864, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.8216789364814758, + "num_tokens": 843337615.0, + "step": 22101 + }, + { + "epoch": 2.8116015774074548, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8217834234237671, + "num_tokens": 843378845.0, + "step": 22102 + }, + { + "epoch": 2.811728787686045, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7883502244949341, + "num_tokens": 843414128.0, + "step": 22103 + }, + { + "epoch": 2.811855997964636, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.817084550857544, + "num_tokens": 843454255.0, + "step": 22104 + }, + { + "epoch": 2.811983208243226, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8065007328987122, + "num_tokens": 843498332.0, + "step": 22105 + }, + { + "epoch": 2.8121104185218164, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7919988632202148, + "num_tokens": 843539196.0, + "step": 22106 + }, + { + "epoch": 2.812237628800407, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.80198073387146, + "num_tokens": 843573767.0, + "step": 22107 + }, + { + "epoch": 2.8123648390789975, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7849268317222595, + "num_tokens": 843611156.0, + "step": 22108 + }, + { + "epoch": 2.812492049357588, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.8049721121788025, + "num_tokens": 843644026.0, + "step": 22109 + }, + { + "epoch": 2.8126192596361785, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.801404595375061, + "num_tokens": 843683477.0, + "step": 22110 + }, + { + "epoch": 2.812746469914769, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7867231369018555, + "num_tokens": 843722815.0, + "step": 22111 + }, + { + "epoch": 2.8128736801933596, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8057252168655396, + "num_tokens": 843763843.0, + "step": 22112 + }, + { + "epoch": 2.81300089047195, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7939378619194031, + "num_tokens": 843805327.0, + "step": 22113 + }, + { + "epoch": 2.8131281007505406, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.8050802946090698, + "num_tokens": 843850132.0, + "step": 22114 + }, + { + "epoch": 2.813255311029131, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8032900094985962, + "num_tokens": 843887425.0, + "step": 22115 + }, + { + "epoch": 2.8133825213077217, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8086118102073669, + "num_tokens": 843920728.0, + "step": 22116 + }, + { + "epoch": 2.8135097315863122, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7983999848365784, + "num_tokens": 843959733.0, + "step": 22117 + }, + { + "epoch": 2.8136369418649028, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.8049581050872803, + "num_tokens": 844005616.0, + "step": 22118 + }, + { + "epoch": 2.8137641521434933, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.81273353099823, + "num_tokens": 844034216.0, + "step": 22119 + }, + { + "epoch": 2.813891362422084, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7979413270950317, + "num_tokens": 844071173.0, + "step": 22120 + }, + { + "epoch": 2.8140185727006743, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7967934608459473, + "num_tokens": 844113665.0, + "step": 22121 + }, + { + "epoch": 2.814145782979265, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.801557719707489, + "num_tokens": 844154063.0, + "step": 22122 + }, + { + "epoch": 2.8142729932578554, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.5791, + "mean_token_accuracy": 0.8218775391578674, + "num_tokens": 844185634.0, + "step": 22123 + }, + { + "epoch": 2.8144002035364455, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7879201173782349, + "num_tokens": 844221755.0, + "step": 22124 + }, + { + "epoch": 2.8145274138150365, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8088847398757935, + "num_tokens": 844256230.0, + "step": 22125 + }, + { + "epoch": 2.8146546240936265, + "grad_norm": 48.0, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7937617301940918, + "num_tokens": 844292823.0, + "step": 22126 + }, + { + "epoch": 2.8147818343722175, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.777803361415863, + "num_tokens": 844331843.0, + "step": 22127 + }, + { + "epoch": 2.8149090446508076, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8118656873703003, + "num_tokens": 844375823.0, + "step": 22128 + }, + { + "epoch": 2.815036254929398, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8009051084518433, + "num_tokens": 844415187.0, + "step": 22129 + }, + { + "epoch": 2.8151634652079887, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8082818388938904, + "num_tokens": 844454986.0, + "step": 22130 + }, + { + "epoch": 2.815290675486579, + "grad_norm": 34.75, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.812095046043396, + "num_tokens": 844495514.0, + "step": 22131 + }, + { + "epoch": 2.8154178857651697, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7993156909942627, + "num_tokens": 844532406.0, + "step": 22132 + }, + { + "epoch": 2.8155450960437602, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.8000826835632324, + "num_tokens": 844569561.0, + "step": 22133 + }, + { + "epoch": 2.8156723063223508, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8088019490242004, + "num_tokens": 844613504.0, + "step": 22134 + }, + { + "epoch": 2.8157995166009413, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7993341088294983, + "num_tokens": 844649255.0, + "step": 22135 + }, + { + "epoch": 2.815926726879532, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.798117458820343, + "num_tokens": 844687202.0, + "step": 22136 + }, + { + "epoch": 2.8160539371581224, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.8013765811920166, + "num_tokens": 844723096.0, + "step": 22137 + }, + { + "epoch": 2.816181147436713, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8020310401916504, + "num_tokens": 844759754.0, + "step": 22138 + }, + { + "epoch": 2.8163083577153034, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8027006387710571, + "num_tokens": 844798372.0, + "step": 22139 + }, + { + "epoch": 2.816435567993894, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7939266562461853, + "num_tokens": 844832738.0, + "step": 22140 + }, + { + "epoch": 2.8165627782724845, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8023011684417725, + "num_tokens": 844871684.0, + "step": 22141 + }, + { + "epoch": 2.816689988551075, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7904653549194336, + "num_tokens": 844912140.0, + "step": 22142 + }, + { + "epoch": 2.8168171988296655, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7938961386680603, + "num_tokens": 844948256.0, + "step": 22143 + }, + { + "epoch": 2.816944409108256, + "grad_norm": 11.25, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8184943795204163, + "num_tokens": 844983821.0, + "step": 22144 + }, + { + "epoch": 2.8170716193868466, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.798682451248169, + "num_tokens": 845020622.0, + "step": 22145 + }, + { + "epoch": 2.817198829665437, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7897167205810547, + "num_tokens": 845056748.0, + "step": 22146 + }, + { + "epoch": 2.8173260399440276, + "grad_norm": 57.0, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7812583446502686, + "num_tokens": 845096294.0, + "step": 22147 + }, + { + "epoch": 2.817453250222618, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8210750818252563, + "num_tokens": 845135132.0, + "step": 22148 + }, + { + "epoch": 2.8175804605012083, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.8110958337783813, + "num_tokens": 845172164.0, + "step": 22149 + }, + { + "epoch": 2.8177076707797992, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.8032102584838867, + "num_tokens": 845210270.0, + "step": 22150 + }, + { + "epoch": 2.8178348810583893, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8062782883644104, + "num_tokens": 845251863.0, + "step": 22151 + }, + { + "epoch": 2.8179620913369803, + "grad_norm": 11.75, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.812727689743042, + "num_tokens": 845290262.0, + "step": 22152 + }, + { + "epoch": 2.8180893016155704, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.809778094291687, + "num_tokens": 845326894.0, + "step": 22153 + }, + { + "epoch": 2.818216511894161, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8146156072616577, + "num_tokens": 845367545.0, + "step": 22154 + }, + { + "epoch": 2.8183437221727514, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7865245342254639, + "num_tokens": 845400920.0, + "step": 22155 + }, + { + "epoch": 2.818470932451342, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8023364543914795, + "num_tokens": 845436279.0, + "step": 22156 + }, + { + "epoch": 2.8185981427299325, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8105392456054688, + "num_tokens": 845476308.0, + "step": 22157 + }, + { + "epoch": 2.818725353008523, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.799292266368866, + "num_tokens": 845516624.0, + "step": 22158 + }, + { + "epoch": 2.8188525632871135, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8152039051055908, + "num_tokens": 845552385.0, + "step": 22159 + }, + { + "epoch": 2.818979773565704, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.8131564855575562, + "num_tokens": 845592600.0, + "step": 22160 + }, + { + "epoch": 2.8191069838442946, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8128210306167603, + "num_tokens": 845630028.0, + "step": 22161 + }, + { + "epoch": 2.819234194122885, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7982877492904663, + "num_tokens": 845665874.0, + "step": 22162 + }, + { + "epoch": 2.8193614044014756, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7877986431121826, + "num_tokens": 845702738.0, + "step": 22163 + }, + { + "epoch": 2.819488614680066, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.8089730739593506, + "num_tokens": 845744785.0, + "step": 22164 + }, + { + "epoch": 2.8196158249586567, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7987322807312012, + "num_tokens": 845779196.0, + "step": 22165 + }, + { + "epoch": 2.8197430352372472, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7881767153739929, + "num_tokens": 845819016.0, + "step": 22166 + }, + { + "epoch": 2.8198702455158378, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8035500049591064, + "num_tokens": 845855014.0, + "step": 22167 + }, + { + "epoch": 2.8199974557944283, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8107051849365234, + "num_tokens": 845895491.0, + "step": 22168 + }, + { + "epoch": 2.820124666073019, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7889071702957153, + "num_tokens": 845931560.0, + "step": 22169 + }, + { + "epoch": 2.8202518763516093, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8056525588035583, + "num_tokens": 845966701.0, + "step": 22170 + }, + { + "epoch": 2.8203790866302, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8197423219680786, + "num_tokens": 846002327.0, + "step": 22171 + }, + { + "epoch": 2.82050629690879, + "grad_norm": 52.75, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8014581203460693, + "num_tokens": 846042157.0, + "step": 22172 + }, + { + "epoch": 2.820633507187381, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7860122919082642, + "num_tokens": 846082851.0, + "step": 22173 + }, + { + "epoch": 2.820760717465971, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.79973304271698, + "num_tokens": 846116846.0, + "step": 22174 + }, + { + "epoch": 2.820887927744562, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7998289465904236, + "num_tokens": 846154373.0, + "step": 22175 + }, + { + "epoch": 2.821015138023152, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.8232112526893616, + "num_tokens": 846195329.0, + "step": 22176 + }, + { + "epoch": 2.821142348301743, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8087543845176697, + "num_tokens": 846233078.0, + "step": 22177 + }, + { + "epoch": 2.821269558580333, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7966755628585815, + "num_tokens": 846269030.0, + "step": 22178 + }, + { + "epoch": 2.8213967688589237, + "grad_norm": 53.5, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7985166907310486, + "num_tokens": 846307269.0, + "step": 22179 + }, + { + "epoch": 2.821523979137514, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.787182092666626, + "num_tokens": 846339116.0, + "step": 22180 + }, + { + "epoch": 2.8216511894161047, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7954786419868469, + "num_tokens": 846384968.0, + "step": 22181 + }, + { + "epoch": 2.8217783996946952, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7823076844215393, + "num_tokens": 846426235.0, + "step": 22182 + }, + { + "epoch": 2.8219056099732858, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7874261140823364, + "num_tokens": 846465715.0, + "step": 22183 + }, + { + "epoch": 2.8220328202518763, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7822158336639404, + "num_tokens": 846499972.0, + "step": 22184 + }, + { + "epoch": 2.822160030530467, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7989102602005005, + "num_tokens": 846535906.0, + "step": 22185 + }, + { + "epoch": 2.8222872408090574, + "grad_norm": 45.5, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.790357232093811, + "num_tokens": 846572723.0, + "step": 22186 + }, + { + "epoch": 2.822414451087648, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8023279905319214, + "num_tokens": 846613501.0, + "step": 22187 + }, + { + "epoch": 2.8225416613662384, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8302115201950073, + "num_tokens": 846650001.0, + "step": 22188 + }, + { + "epoch": 2.822668871644829, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7867765426635742, + "num_tokens": 846690636.0, + "step": 22189 + }, + { + "epoch": 2.8227960819234195, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7810735106468201, + "num_tokens": 846728513.0, + "step": 22190 + }, + { + "epoch": 2.82292329220201, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7851114869117737, + "num_tokens": 846771989.0, + "step": 22191 + }, + { + "epoch": 2.8230505024806005, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8180527687072754, + "num_tokens": 846808393.0, + "step": 22192 + }, + { + "epoch": 2.823177712759191, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7925094366073608, + "num_tokens": 846846928.0, + "step": 22193 + }, + { + "epoch": 2.8233049230377816, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7863322496414185, + "num_tokens": 846884637.0, + "step": 22194 + }, + { + "epoch": 2.823432133316372, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8071550130844116, + "num_tokens": 846925749.0, + "step": 22195 + }, + { + "epoch": 2.8235593435949626, + "grad_norm": 26.375, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7852743864059448, + "num_tokens": 846967191.0, + "step": 22196 + }, + { + "epoch": 2.8236865538735527, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8063066005706787, + "num_tokens": 847003280.0, + "step": 22197 + }, + { + "epoch": 2.8238137641521437, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7833311557769775, + "num_tokens": 847042179.0, + "step": 22198 + }, + { + "epoch": 2.8239409744307338, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7975201606750488, + "num_tokens": 847083008.0, + "step": 22199 + }, + { + "epoch": 2.8240681847093247, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7761543393135071, + "num_tokens": 847124009.0, + "step": 22200 + }, + { + "epoch": 2.824195394987915, + "grad_norm": 48.25, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7984753847122192, + "num_tokens": 847163526.0, + "step": 22201 + }, + { + "epoch": 2.824322605266506, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7999663352966309, + "num_tokens": 847201374.0, + "step": 22202 + }, + { + "epoch": 2.824449815545096, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7816282510757446, + "num_tokens": 847239469.0, + "step": 22203 + }, + { + "epoch": 2.8245770258236864, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.792375385761261, + "num_tokens": 847275964.0, + "step": 22204 + }, + { + "epoch": 2.824704236102277, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.8016036748886108, + "num_tokens": 847316815.0, + "step": 22205 + }, + { + "epoch": 2.8248314463808675, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8042101860046387, + "num_tokens": 847350949.0, + "step": 22206 + }, + { + "epoch": 2.824958656659458, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8085271716117859, + "num_tokens": 847388945.0, + "step": 22207 + }, + { + "epoch": 2.8250858669380485, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8139956593513489, + "num_tokens": 847435114.0, + "step": 22208 + }, + { + "epoch": 2.825213077216639, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7955139875411987, + "num_tokens": 847470815.0, + "step": 22209 + }, + { + "epoch": 2.8253402874952296, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.8034590482711792, + "num_tokens": 847509379.0, + "step": 22210 + }, + { + "epoch": 2.82546749777382, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.8002651929855347, + "num_tokens": 847547427.0, + "step": 22211 + }, + { + "epoch": 2.8255947080524106, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7814962863922119, + "num_tokens": 847584740.0, + "step": 22212 + }, + { + "epoch": 2.825721918331001, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7901842594146729, + "num_tokens": 847626201.0, + "step": 22213 + }, + { + "epoch": 2.8258491286095917, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8152511119842529, + "num_tokens": 847661311.0, + "step": 22214 + }, + { + "epoch": 2.8259763388881822, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7931351661682129, + "num_tokens": 847695234.0, + "step": 22215 + }, + { + "epoch": 2.8261035491667728, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8231891989707947, + "num_tokens": 847738848.0, + "step": 22216 + }, + { + "epoch": 2.8262307594453633, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8014719486236572, + "num_tokens": 847775952.0, + "step": 22217 + }, + { + "epoch": 2.826357969723954, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.8194178342819214, + "num_tokens": 847809381.0, + "step": 22218 + }, + { + "epoch": 2.8264851800025443, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7974358201026917, + "num_tokens": 847849835.0, + "step": 22219 + }, + { + "epoch": 2.826612390281135, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7931017875671387, + "num_tokens": 847882726.0, + "step": 22220 + }, + { + "epoch": 2.8267396005597254, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7964938879013062, + "num_tokens": 847920663.0, + "step": 22221 + }, + { + "epoch": 2.8268668108383155, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8086220026016235, + "num_tokens": 847962204.0, + "step": 22222 + }, + { + "epoch": 2.8269940211169065, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7898527383804321, + "num_tokens": 848003722.0, + "step": 22223 + }, + { + "epoch": 2.8271212313954965, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8129611015319824, + "num_tokens": 848045335.0, + "step": 22224 + }, + { + "epoch": 2.8272484416740875, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7941699028015137, + "num_tokens": 848088817.0, + "step": 22225 + }, + { + "epoch": 2.8273756519526776, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7871823906898499, + "num_tokens": 848120993.0, + "step": 22226 + }, + { + "epoch": 2.827502862231268, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7874624729156494, + "num_tokens": 848162542.0, + "step": 22227 + }, + { + "epoch": 2.8276300725098586, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.8119599223136902, + "num_tokens": 848198194.0, + "step": 22228 + }, + { + "epoch": 2.827757282788449, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7927181124687195, + "num_tokens": 848240575.0, + "step": 22229 + }, + { + "epoch": 2.8278844930670397, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.8049343228340149, + "num_tokens": 848289026.0, + "step": 22230 + }, + { + "epoch": 2.8280117033456302, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8010463714599609, + "num_tokens": 848326438.0, + "step": 22231 + }, + { + "epoch": 2.8281389136242208, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8099900484085083, + "num_tokens": 848368106.0, + "step": 22232 + }, + { + "epoch": 2.8282661239028113, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7968745231628418, + "num_tokens": 848409775.0, + "step": 22233 + }, + { + "epoch": 2.828393334181402, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8003416061401367, + "num_tokens": 848442149.0, + "step": 22234 + }, + { + "epoch": 2.8285205444599923, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7950917482376099, + "num_tokens": 848475468.0, + "step": 22235 + }, + { + "epoch": 2.828647754738583, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8034425377845764, + "num_tokens": 848509565.0, + "step": 22236 + }, + { + "epoch": 2.8287749650171734, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.8034533262252808, + "num_tokens": 848546357.0, + "step": 22237 + }, + { + "epoch": 2.828902175295764, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7908194661140442, + "num_tokens": 848587661.0, + "step": 22238 + }, + { + "epoch": 2.8290293855743545, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.8074995279312134, + "num_tokens": 848628250.0, + "step": 22239 + }, + { + "epoch": 2.829156595852945, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8042700290679932, + "num_tokens": 848670452.0, + "step": 22240 + }, + { + "epoch": 2.8292838061315355, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.8032768368721008, + "num_tokens": 848703592.0, + "step": 22241 + }, + { + "epoch": 2.829411016410126, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7791455984115601, + "num_tokens": 848739291.0, + "step": 22242 + }, + { + "epoch": 2.8295382266887166, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7989752292633057, + "num_tokens": 848776039.0, + "step": 22243 + }, + { + "epoch": 2.829665436967307, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8054255843162537, + "num_tokens": 848814799.0, + "step": 22244 + }, + { + "epoch": 2.8297926472458976, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8079122304916382, + "num_tokens": 848860123.0, + "step": 22245 + }, + { + "epoch": 2.829919857524488, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.820024311542511, + "num_tokens": 848900447.0, + "step": 22246 + }, + { + "epoch": 2.8300470678030782, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7990278005599976, + "num_tokens": 848945385.0, + "step": 22247 + }, + { + "epoch": 2.830174278081669, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7897962331771851, + "num_tokens": 848978783.0, + "step": 22248 + }, + { + "epoch": 2.8303014883602593, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7862347960472107, + "num_tokens": 849017324.0, + "step": 22249 + }, + { + "epoch": 2.8304286986388503, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8226634860038757, + "num_tokens": 849045928.0, + "step": 22250 + }, + { + "epoch": 2.8305559089174404, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7928847670555115, + "num_tokens": 849080855.0, + "step": 22251 + }, + { + "epoch": 2.830683119196031, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7986916899681091, + "num_tokens": 849114835.0, + "step": 22252 + }, + { + "epoch": 2.8308103294746214, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7882178425788879, + "num_tokens": 849152632.0, + "step": 22253 + }, + { + "epoch": 2.830937539753212, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8064315915107727, + "num_tokens": 849189911.0, + "step": 22254 + }, + { + "epoch": 2.8310647500318025, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7948494553565979, + "num_tokens": 849227847.0, + "step": 22255 + }, + { + "epoch": 2.831191960310393, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8078136444091797, + "num_tokens": 849265778.0, + "step": 22256 + }, + { + "epoch": 2.8313191705889835, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7992609739303589, + "num_tokens": 849302488.0, + "step": 22257 + }, + { + "epoch": 2.831446380867574, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.819658637046814, + "num_tokens": 849337483.0, + "step": 22258 + }, + { + "epoch": 2.8315735911461646, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.804502010345459, + "num_tokens": 849378491.0, + "step": 22259 + }, + { + "epoch": 2.831700801424755, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8071786761283875, + "num_tokens": 849418897.0, + "step": 22260 + }, + { + "epoch": 2.8318280117033456, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8087916970252991, + "num_tokens": 849457057.0, + "step": 22261 + }, + { + "epoch": 2.831955221981936, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8030269742012024, + "num_tokens": 849488825.0, + "step": 22262 + }, + { + "epoch": 2.8320824322605267, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8115513324737549, + "num_tokens": 849528336.0, + "step": 22263 + }, + { + "epoch": 2.8322096425391172, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8144122958183289, + "num_tokens": 849565373.0, + "step": 22264 + }, + { + "epoch": 2.8323368528177078, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.8179274201393127, + "num_tokens": 849600909.0, + "step": 22265 + }, + { + "epoch": 2.8324640630962983, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7909844517707825, + "num_tokens": 849639499.0, + "step": 22266 + }, + { + "epoch": 2.832591273374889, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8162579536437988, + "num_tokens": 849677232.0, + "step": 22267 + }, + { + "epoch": 2.8327184836534793, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8103637099266052, + "num_tokens": 849717187.0, + "step": 22268 + }, + { + "epoch": 2.83284569393207, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8133444786071777, + "num_tokens": 849756545.0, + "step": 22269 + }, + { + "epoch": 2.83297290421066, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.8025579452514648, + "num_tokens": 849788670.0, + "step": 22270 + }, + { + "epoch": 2.833100114489251, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.8005676865577698, + "num_tokens": 849825715.0, + "step": 22271 + }, + { + "epoch": 2.833227324767841, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.80484938621521, + "num_tokens": 849856435.0, + "step": 22272 + }, + { + "epoch": 2.833354535046432, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.8149821162223816, + "num_tokens": 849895634.0, + "step": 22273 + }, + { + "epoch": 2.833481745325022, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7785665392875671, + "num_tokens": 849938646.0, + "step": 22274 + }, + { + "epoch": 2.833608955603613, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8028183579444885, + "num_tokens": 849978642.0, + "step": 22275 + }, + { + "epoch": 2.833736165882203, + "grad_norm": 11.0, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8026126623153687, + "num_tokens": 850017820.0, + "step": 22276 + }, + { + "epoch": 2.8338633761607936, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.784040093421936, + "num_tokens": 850056619.0, + "step": 22277 + }, + { + "epoch": 2.833990586439384, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7943415641784668, + "num_tokens": 850092456.0, + "step": 22278 + }, + { + "epoch": 2.8341177967179747, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8041356205940247, + "num_tokens": 850131598.0, + "step": 22279 + }, + { + "epoch": 2.8342450069965652, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7932140827178955, + "num_tokens": 850170884.0, + "step": 22280 + }, + { + "epoch": 2.8343722172751558, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7979949712753296, + "num_tokens": 850206989.0, + "step": 22281 + }, + { + "epoch": 2.8344994275537463, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7956093549728394, + "num_tokens": 850248602.0, + "step": 22282 + }, + { + "epoch": 2.834626637832337, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7967047691345215, + "num_tokens": 850289790.0, + "step": 22283 + }, + { + "epoch": 2.8347538481109273, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8088110685348511, + "num_tokens": 850326688.0, + "step": 22284 + }, + { + "epoch": 2.834881058389518, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8129435777664185, + "num_tokens": 850366460.0, + "step": 22285 + }, + { + "epoch": 2.8350082686681084, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.8060028553009033, + "num_tokens": 850400046.0, + "step": 22286 + }, + { + "epoch": 2.835135478946699, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7823923826217651, + "num_tokens": 850434374.0, + "step": 22287 + }, + { + "epoch": 2.8352626892252895, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.799326479434967, + "num_tokens": 850469056.0, + "step": 22288 + }, + { + "epoch": 2.83538989950388, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.8030741214752197, + "num_tokens": 850505805.0, + "step": 22289 + }, + { + "epoch": 2.8355171097824705, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7861920595169067, + "num_tokens": 850550056.0, + "step": 22290 + }, + { + "epoch": 2.835644320061061, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8094194531440735, + "num_tokens": 850585075.0, + "step": 22291 + }, + { + "epoch": 2.8357715303396516, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8145318031311035, + "num_tokens": 850625758.0, + "step": 22292 + }, + { + "epoch": 2.835898740618242, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.8138440847396851, + "num_tokens": 850664608.0, + "step": 22293 + }, + { + "epoch": 2.8360259508968326, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8153290748596191, + "num_tokens": 850705897.0, + "step": 22294 + }, + { + "epoch": 2.8361531611754227, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7943611741065979, + "num_tokens": 850746195.0, + "step": 22295 + }, + { + "epoch": 2.8362803714540137, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.8104462027549744, + "num_tokens": 850785183.0, + "step": 22296 + }, + { + "epoch": 2.8364075817326038, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.806122362613678, + "num_tokens": 850819558.0, + "step": 22297 + }, + { + "epoch": 2.8365347920111947, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7918823957443237, + "num_tokens": 850857385.0, + "step": 22298 + }, + { + "epoch": 2.836662002289785, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7963926792144775, + "num_tokens": 850895297.0, + "step": 22299 + }, + { + "epoch": 2.836789212568376, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8035493493080139, + "num_tokens": 850930982.0, + "step": 22300 + }, + { + "epoch": 2.836916422846966, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7974203824996948, + "num_tokens": 850969032.0, + "step": 22301 + }, + { + "epoch": 2.8370436331255564, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8093668222427368, + "num_tokens": 851008325.0, + "step": 22302 + }, + { + "epoch": 2.837170843404147, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.803947925567627, + "num_tokens": 851044289.0, + "step": 22303 + }, + { + "epoch": 2.8372980536827375, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.8107097148895264, + "num_tokens": 851079887.0, + "step": 22304 + }, + { + "epoch": 2.837425263961328, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8056081533432007, + "num_tokens": 851124535.0, + "step": 22305 + }, + { + "epoch": 2.8375524742399185, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.5912, + "mean_token_accuracy": 0.8163958191871643, + "num_tokens": 851162982.0, + "step": 22306 + }, + { + "epoch": 2.837679684518509, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.8146530985832214, + "num_tokens": 851201864.0, + "step": 22307 + }, + { + "epoch": 2.8378068947970996, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8049126863479614, + "num_tokens": 851241659.0, + "step": 22308 + }, + { + "epoch": 2.83793410507569, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7943223714828491, + "num_tokens": 851278861.0, + "step": 22309 + }, + { + "epoch": 2.8380613153542806, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.8026413917541504, + "num_tokens": 851314442.0, + "step": 22310 + }, + { + "epoch": 2.838188525632871, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7956172823905945, + "num_tokens": 851356322.0, + "step": 22311 + }, + { + "epoch": 2.8383157359114617, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7921032309532166, + "num_tokens": 851395374.0, + "step": 22312 + }, + { + "epoch": 2.838442946190052, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7953591346740723, + "num_tokens": 851430957.0, + "step": 22313 + }, + { + "epoch": 2.8385701564686427, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8005627989768982, + "num_tokens": 851474357.0, + "step": 22314 + }, + { + "epoch": 2.8386973667472333, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.8131288290023804, + "num_tokens": 851510346.0, + "step": 22315 + }, + { + "epoch": 2.838824577025824, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.8021520376205444, + "num_tokens": 851546798.0, + "step": 22316 + }, + { + "epoch": 2.8389517873044143, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8070715665817261, + "num_tokens": 851587819.0, + "step": 22317 + }, + { + "epoch": 2.839078997583005, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8088613748550415, + "num_tokens": 851628764.0, + "step": 22318 + }, + { + "epoch": 2.8392062078615954, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.790820837020874, + "num_tokens": 851672206.0, + "step": 22319 + }, + { + "epoch": 2.8393334181401855, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7985925078392029, + "num_tokens": 851709969.0, + "step": 22320 + }, + { + "epoch": 2.8394606284187764, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7923531532287598, + "num_tokens": 851749015.0, + "step": 22321 + }, + { + "epoch": 2.8395878386973665, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.8089708089828491, + "num_tokens": 851787210.0, + "step": 22322 + }, + { + "epoch": 2.8397150489759575, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.795879602432251, + "num_tokens": 851824677.0, + "step": 22323 + }, + { + "epoch": 2.8398422592545476, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.8101305961608887, + "num_tokens": 851863614.0, + "step": 22324 + }, + { + "epoch": 2.839969469533138, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7994211912155151, + "num_tokens": 851908153.0, + "step": 22325 + }, + { + "epoch": 2.8400966798117286, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7955535650253296, + "num_tokens": 851946473.0, + "step": 22326 + }, + { + "epoch": 2.840223890090319, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7938418388366699, + "num_tokens": 851991600.0, + "step": 22327 + }, + { + "epoch": 2.8403511003689097, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8045946955680847, + "num_tokens": 852029468.0, + "step": 22328 + }, + { + "epoch": 2.8404783106475002, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8033425807952881, + "num_tokens": 852061400.0, + "step": 22329 + }, + { + "epoch": 2.8406055209260908, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8036291599273682, + "num_tokens": 852100596.0, + "step": 22330 + }, + { + "epoch": 2.8407327312046813, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.8033357858657837, + "num_tokens": 852140320.0, + "step": 22331 + }, + { + "epoch": 2.840859941483272, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7976692914962769, + "num_tokens": 852175969.0, + "step": 22332 + }, + { + "epoch": 2.8409871517618623, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7813050746917725, + "num_tokens": 852210816.0, + "step": 22333 + }, + { + "epoch": 2.841114362040453, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.8027795553207397, + "num_tokens": 852246684.0, + "step": 22334 + }, + { + "epoch": 2.8412415723190434, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.805840015411377, + "num_tokens": 852285283.0, + "step": 22335 + }, + { + "epoch": 2.841368782597634, + "grad_norm": 30.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8090978860855103, + "num_tokens": 852323310.0, + "step": 22336 + }, + { + "epoch": 2.8414959928762245, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.8066598176956177, + "num_tokens": 852360850.0, + "step": 22337 + }, + { + "epoch": 2.841623203154815, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7946825623512268, + "num_tokens": 852399320.0, + "step": 22338 + }, + { + "epoch": 2.8417504134334055, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7974388599395752, + "num_tokens": 852430980.0, + "step": 22339 + }, + { + "epoch": 2.841877623711996, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8141260147094727, + "num_tokens": 852465688.0, + "step": 22340 + }, + { + "epoch": 2.8420048339905866, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7970106601715088, + "num_tokens": 852504372.0, + "step": 22341 + }, + { + "epoch": 2.842132044269177, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8103703856468201, + "num_tokens": 852542232.0, + "step": 22342 + }, + { + "epoch": 2.8422592545477676, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7884833812713623, + "num_tokens": 852580005.0, + "step": 22343 + }, + { + "epoch": 2.842386464826358, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.8169898390769958, + "num_tokens": 852619912.0, + "step": 22344 + }, + { + "epoch": 2.8425136751049482, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7970069646835327, + "num_tokens": 852654349.0, + "step": 22345 + }, + { + "epoch": 2.842640885383539, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.791233241558075, + "num_tokens": 852691300.0, + "step": 22346 + }, + { + "epoch": 2.8427680956621293, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8027639389038086, + "num_tokens": 852731228.0, + "step": 22347 + }, + { + "epoch": 2.8428953059407203, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.8022341728210449, + "num_tokens": 852768333.0, + "step": 22348 + }, + { + "epoch": 2.8430225162193103, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7638277411460876, + "num_tokens": 852810962.0, + "step": 22349 + }, + { + "epoch": 2.843149726497901, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8145448565483093, + "num_tokens": 852844326.0, + "step": 22350 + }, + { + "epoch": 2.8432769367764914, + "grad_norm": 36.25, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7946885228157043, + "num_tokens": 852877737.0, + "step": 22351 + }, + { + "epoch": 2.843404147055082, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8109400272369385, + "num_tokens": 852922085.0, + "step": 22352 + }, + { + "epoch": 2.8435313573336725, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.8165948390960693, + "num_tokens": 852960742.0, + "step": 22353 + }, + { + "epoch": 2.843658567612263, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8064422607421875, + "num_tokens": 852993502.0, + "step": 22354 + }, + { + "epoch": 2.8437857778908535, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8109576106071472, + "num_tokens": 853036844.0, + "step": 22355 + }, + { + "epoch": 2.843912988169444, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7979205846786499, + "num_tokens": 853070851.0, + "step": 22356 + }, + { + "epoch": 2.8440401984480346, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.797266960144043, + "num_tokens": 853106116.0, + "step": 22357 + }, + { + "epoch": 2.844167408726625, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7985055446624756, + "num_tokens": 853152776.0, + "step": 22358 + }, + { + "epoch": 2.8442946190052156, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.8039101958274841, + "num_tokens": 853194176.0, + "step": 22359 + }, + { + "epoch": 2.844421829283806, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7945165038108826, + "num_tokens": 853236834.0, + "step": 22360 + }, + { + "epoch": 2.8445490395623967, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7927570343017578, + "num_tokens": 853274145.0, + "step": 22361 + }, + { + "epoch": 2.844676249840987, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.8039695024490356, + "num_tokens": 853312569.0, + "step": 22362 + }, + { + "epoch": 2.8448034601195777, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8093149065971375, + "num_tokens": 853351742.0, + "step": 22363 + }, + { + "epoch": 2.8449306703981683, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8026407957077026, + "num_tokens": 853389607.0, + "step": 22364 + }, + { + "epoch": 2.845057880676759, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8070345520973206, + "num_tokens": 853427218.0, + "step": 22365 + }, + { + "epoch": 2.8451850909553493, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.805820107460022, + "num_tokens": 853464773.0, + "step": 22366 + }, + { + "epoch": 2.84531230123394, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.8052864074707031, + "num_tokens": 853507250.0, + "step": 22367 + }, + { + "epoch": 2.84543951151253, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8042791485786438, + "num_tokens": 853551947.0, + "step": 22368 + }, + { + "epoch": 2.845566721791121, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7935847043991089, + "num_tokens": 853590557.0, + "step": 22369 + }, + { + "epoch": 2.845693932069711, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.8005112409591675, + "num_tokens": 853630870.0, + "step": 22370 + }, + { + "epoch": 2.845821142348302, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7963926792144775, + "num_tokens": 853662360.0, + "step": 22371 + }, + { + "epoch": 2.845948352626892, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.8000680804252625, + "num_tokens": 853703661.0, + "step": 22372 + }, + { + "epoch": 2.846075562905483, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.786798357963562, + "num_tokens": 853742398.0, + "step": 22373 + }, + { + "epoch": 2.846202773184073, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8031414151191711, + "num_tokens": 853781479.0, + "step": 22374 + }, + { + "epoch": 2.8463299834626636, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7770504951477051, + "num_tokens": 853822330.0, + "step": 22375 + }, + { + "epoch": 2.846457193741254, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7849113345146179, + "num_tokens": 853862747.0, + "step": 22376 + }, + { + "epoch": 2.8465844040198447, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8254369497299194, + "num_tokens": 853897124.0, + "step": 22377 + }, + { + "epoch": 2.8467116142984352, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7827052474021912, + "num_tokens": 853930516.0, + "step": 22378 + }, + { + "epoch": 2.8468388245770258, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8091531991958618, + "num_tokens": 853967016.0, + "step": 22379 + }, + { + "epoch": 2.8469660348556163, + "grad_norm": 58.5, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7845233082771301, + "num_tokens": 854008804.0, + "step": 22380 + }, + { + "epoch": 2.847093245134207, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.78911954164505, + "num_tokens": 854048393.0, + "step": 22381 + }, + { + "epoch": 2.8472204554127973, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8076171875, + "num_tokens": 854089480.0, + "step": 22382 + }, + { + "epoch": 2.847347665691388, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8147532343864441, + "num_tokens": 854125677.0, + "step": 22383 + }, + { + "epoch": 2.8474748759699784, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7905709147453308, + "num_tokens": 854160485.0, + "step": 22384 + }, + { + "epoch": 2.847602086248569, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7931833863258362, + "num_tokens": 854200635.0, + "step": 22385 + }, + { + "epoch": 2.8477292965271594, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8040221333503723, + "num_tokens": 854240046.0, + "step": 22386 + }, + { + "epoch": 2.84785650680575, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.8107665777206421, + "num_tokens": 854275880.0, + "step": 22387 + }, + { + "epoch": 2.8479837170843405, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.8030974864959717, + "num_tokens": 854311204.0, + "step": 22388 + }, + { + "epoch": 2.848110927362931, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7982090711593628, + "num_tokens": 854346076.0, + "step": 22389 + }, + { + "epoch": 2.8482381376415216, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8127068281173706, + "num_tokens": 854383075.0, + "step": 22390 + }, + { + "epoch": 2.848365347920112, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7804877161979675, + "num_tokens": 854416011.0, + "step": 22391 + }, + { + "epoch": 2.8484925581987026, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8075065016746521, + "num_tokens": 854449520.0, + "step": 22392 + }, + { + "epoch": 2.8486197684772927, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.79285728931427, + "num_tokens": 854484355.0, + "step": 22393 + }, + { + "epoch": 2.8487469787558837, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7863647937774658, + "num_tokens": 854526238.0, + "step": 22394 + }, + { + "epoch": 2.8488741890344738, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8035854697227478, + "num_tokens": 854562619.0, + "step": 22395 + }, + { + "epoch": 2.8490013993130647, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8058487772941589, + "num_tokens": 854604974.0, + "step": 22396 + }, + { + "epoch": 2.849128609591655, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8085028529167175, + "num_tokens": 854644066.0, + "step": 22397 + }, + { + "epoch": 2.849255819870246, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8074255585670471, + "num_tokens": 854682723.0, + "step": 22398 + }, + { + "epoch": 2.849383030148836, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8104650974273682, + "num_tokens": 854720292.0, + "step": 22399 + }, + { + "epoch": 2.8495102404274264, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7995046377182007, + "num_tokens": 854756917.0, + "step": 22400 + }, + { + "epoch": 2.849637450706017, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8084243535995483, + "num_tokens": 854798771.0, + "step": 22401 + }, + { + "epoch": 2.8497646609846075, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.80650395154953, + "num_tokens": 854838193.0, + "step": 22402 + }, + { + "epoch": 2.849891871263198, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7886667251586914, + "num_tokens": 854871650.0, + "step": 22403 + }, + { + "epoch": 2.8500190815417885, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.800247848033905, + "num_tokens": 854910757.0, + "step": 22404 + }, + { + "epoch": 2.850146291820379, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7841393947601318, + "num_tokens": 854945018.0, + "step": 22405 + }, + { + "epoch": 2.8502735020989696, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7922382354736328, + "num_tokens": 854988369.0, + "step": 22406 + }, + { + "epoch": 2.85040071237756, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8059650659561157, + "num_tokens": 855023935.0, + "step": 22407 + }, + { + "epoch": 2.8505279226561506, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.8006446361541748, + "num_tokens": 855062002.0, + "step": 22408 + }, + { + "epoch": 2.850655132934741, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8056239485740662, + "num_tokens": 855101257.0, + "step": 22409 + }, + { + "epoch": 2.8507823432133317, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7999728918075562, + "num_tokens": 855135834.0, + "step": 22410 + }, + { + "epoch": 2.850909553491922, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.8001600503921509, + "num_tokens": 855174156.0, + "step": 22411 + }, + { + "epoch": 2.8510367637705127, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.8050503730773926, + "num_tokens": 855207698.0, + "step": 22412 + }, + { + "epoch": 2.8511639740491033, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8025118112564087, + "num_tokens": 855240636.0, + "step": 22413 + }, + { + "epoch": 2.851291184327694, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8061841726303101, + "num_tokens": 855279923.0, + "step": 22414 + }, + { + "epoch": 2.8514183946062843, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7871419787406921, + "num_tokens": 855315460.0, + "step": 22415 + }, + { + "epoch": 2.851545604884875, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8070762753486633, + "num_tokens": 855351548.0, + "step": 22416 + }, + { + "epoch": 2.8516728151634654, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7852662801742554, + "num_tokens": 855390715.0, + "step": 22417 + }, + { + "epoch": 2.8518000254420555, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.792487382888794, + "num_tokens": 855431997.0, + "step": 22418 + }, + { + "epoch": 2.8519272357206464, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7971010208129883, + "num_tokens": 855470242.0, + "step": 22419 + }, + { + "epoch": 2.8520544459992365, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7894285917282104, + "num_tokens": 855508901.0, + "step": 22420 + }, + { + "epoch": 2.8521816562778275, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7961900234222412, + "num_tokens": 855545290.0, + "step": 22421 + }, + { + "epoch": 2.8523088665564176, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7909892797470093, + "num_tokens": 855579567.0, + "step": 22422 + }, + { + "epoch": 2.852436076835008, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7940222024917603, + "num_tokens": 855621117.0, + "step": 22423 + }, + { + "epoch": 2.8525632871135986, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.811972975730896, + "num_tokens": 855660494.0, + "step": 22424 + }, + { + "epoch": 2.852690497392189, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7995917797088623, + "num_tokens": 855694043.0, + "step": 22425 + }, + { + "epoch": 2.8528177076707797, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.787335991859436, + "num_tokens": 855734615.0, + "step": 22426 + }, + { + "epoch": 2.85294491794937, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8099848031997681, + "num_tokens": 855771141.0, + "step": 22427 + }, + { + "epoch": 2.8530721282279607, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7976117134094238, + "num_tokens": 855806877.0, + "step": 22428 + }, + { + "epoch": 2.8531993385065513, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8152202367782593, + "num_tokens": 855840076.0, + "step": 22429 + }, + { + "epoch": 2.853326548785142, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8152504563331604, + "num_tokens": 855877972.0, + "step": 22430 + }, + { + "epoch": 2.8534537590637323, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7965611219406128, + "num_tokens": 855917424.0, + "step": 22431 + }, + { + "epoch": 2.853580969342323, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7897061705589294, + "num_tokens": 855954853.0, + "step": 22432 + }, + { + "epoch": 2.8537081796209134, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8102058172225952, + "num_tokens": 855993300.0, + "step": 22433 + }, + { + "epoch": 2.853835389899504, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8090550899505615, + "num_tokens": 856035307.0, + "step": 22434 + }, + { + "epoch": 2.8539626001780944, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.802696704864502, + "num_tokens": 856071801.0, + "step": 22435 + }, + { + "epoch": 2.854089810456685, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7931715250015259, + "num_tokens": 856112035.0, + "step": 22436 + }, + { + "epoch": 2.8542170207352755, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8075839281082153, + "num_tokens": 856148275.0, + "step": 22437 + }, + { + "epoch": 2.854344231013866, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.8007161021232605, + "num_tokens": 856180330.0, + "step": 22438 + }, + { + "epoch": 2.8544714412924566, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7829734683036804, + "num_tokens": 856219354.0, + "step": 22439 + }, + { + "epoch": 2.854598651571047, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8120287656784058, + "num_tokens": 856257583.0, + "step": 22440 + }, + { + "epoch": 2.8547258618496376, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7811369895935059, + "num_tokens": 856298056.0, + "step": 22441 + }, + { + "epoch": 2.854853072128228, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7980396747589111, + "num_tokens": 856334109.0, + "step": 22442 + }, + { + "epoch": 2.8549802824068182, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.8077419996261597, + "num_tokens": 856376655.0, + "step": 22443 + }, + { + "epoch": 2.855107492685409, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.5965, + "mean_token_accuracy": 0.8155096173286438, + "num_tokens": 856409926.0, + "step": 22444 + }, + { + "epoch": 2.8552347029639993, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7899475693702698, + "num_tokens": 856451471.0, + "step": 22445 + }, + { + "epoch": 2.8553619132425903, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.8084973692893982, + "num_tokens": 856483237.0, + "step": 22446 + }, + { + "epoch": 2.8554891235211803, + "grad_norm": 26.0, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7925481796264648, + "num_tokens": 856521278.0, + "step": 22447 + }, + { + "epoch": 2.855616333799771, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8019627928733826, + "num_tokens": 856565048.0, + "step": 22448 + }, + { + "epoch": 2.8557435440783614, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8091861009597778, + "num_tokens": 856603756.0, + "step": 22449 + }, + { + "epoch": 2.855870754356952, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8068950772285461, + "num_tokens": 856640677.0, + "step": 22450 + }, + { + "epoch": 2.8559979646355425, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.8036838173866272, + "num_tokens": 856679982.0, + "step": 22451 + }, + { + "epoch": 2.856125174914133, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7896790504455566, + "num_tokens": 856720621.0, + "step": 22452 + }, + { + "epoch": 2.8562523851927235, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8139519095420837, + "num_tokens": 856757172.0, + "step": 22453 + }, + { + "epoch": 2.856379595471314, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7990421652793884, + "num_tokens": 856796876.0, + "step": 22454 + }, + { + "epoch": 2.8565068057499046, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.8016467094421387, + "num_tokens": 856831148.0, + "step": 22455 + }, + { + "epoch": 2.856634016028495, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7957940101623535, + "num_tokens": 856867091.0, + "step": 22456 + }, + { + "epoch": 2.8567612263070856, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7933961153030396, + "num_tokens": 856906118.0, + "step": 22457 + }, + { + "epoch": 2.856888436585676, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7687281966209412, + "num_tokens": 856942694.0, + "step": 22458 + }, + { + "epoch": 2.8570156468642667, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7935574054718018, + "num_tokens": 856977530.0, + "step": 22459 + }, + { + "epoch": 2.857142857142857, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.8042230010032654, + "num_tokens": 857012417.0, + "step": 22460 + }, + { + "epoch": 2.8572700674214477, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7955402135848999, + "num_tokens": 857053556.0, + "step": 22461 + }, + { + "epoch": 2.8573972777000383, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8082695007324219, + "num_tokens": 857094514.0, + "step": 22462 + }, + { + "epoch": 2.857524487978629, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8017720580101013, + "num_tokens": 857133670.0, + "step": 22463 + }, + { + "epoch": 2.8576516982572193, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7915160059928894, + "num_tokens": 857167934.0, + "step": 22464 + }, + { + "epoch": 2.85777890853581, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7795642614364624, + "num_tokens": 857208263.0, + "step": 22465 + }, + { + "epoch": 2.8579061188144, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7903005480766296, + "num_tokens": 857255107.0, + "step": 22466 + }, + { + "epoch": 2.858033329092991, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.8060086965560913, + "num_tokens": 857298368.0, + "step": 22467 + }, + { + "epoch": 2.858160539371581, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7991496324539185, + "num_tokens": 857336899.0, + "step": 22468 + }, + { + "epoch": 2.858287749650172, + "grad_norm": 10.3125, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8090907335281372, + "num_tokens": 857378482.0, + "step": 22469 + }, + { + "epoch": 2.858414959928762, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8127800226211548, + "num_tokens": 857413203.0, + "step": 22470 + }, + { + "epoch": 2.858542170207353, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7952988147735596, + "num_tokens": 857446706.0, + "step": 22471 + }, + { + "epoch": 2.858669380485943, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8075176477432251, + "num_tokens": 857483415.0, + "step": 22472 + }, + { + "epoch": 2.8587965907645336, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8081732988357544, + "num_tokens": 857525323.0, + "step": 22473 + }, + { + "epoch": 2.858923801043124, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8181353807449341, + "num_tokens": 857564400.0, + "step": 22474 + }, + { + "epoch": 2.8590510113217147, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8041740655899048, + "num_tokens": 857602756.0, + "step": 22475 + }, + { + "epoch": 2.859178221600305, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7998884916305542, + "num_tokens": 857636875.0, + "step": 22476 + }, + { + "epoch": 2.8593054318788957, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.8165525197982788, + "num_tokens": 857671950.0, + "step": 22477 + }, + { + "epoch": 2.8594326421574863, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.8126420974731445, + "num_tokens": 857705274.0, + "step": 22478 + }, + { + "epoch": 2.859559852436077, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7943196892738342, + "num_tokens": 857743730.0, + "step": 22479 + }, + { + "epoch": 2.8596870627146673, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7952612638473511, + "num_tokens": 857784055.0, + "step": 22480 + }, + { + "epoch": 2.859814272993258, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8131973743438721, + "num_tokens": 857822431.0, + "step": 22481 + }, + { + "epoch": 2.8599414832718484, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8100693225860596, + "num_tokens": 857859258.0, + "step": 22482 + }, + { + "epoch": 2.860068693550439, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.798492968082428, + "num_tokens": 857901209.0, + "step": 22483 + }, + { + "epoch": 2.8601959038290294, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.8189996480941772, + "num_tokens": 857935563.0, + "step": 22484 + }, + { + "epoch": 2.86032311410762, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7834950685501099, + "num_tokens": 857966777.0, + "step": 22485 + }, + { + "epoch": 2.8604503243862105, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.806406557559967, + "num_tokens": 858005078.0, + "step": 22486 + }, + { + "epoch": 2.860577534664801, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7885149121284485, + "num_tokens": 858049288.0, + "step": 22487 + }, + { + "epoch": 2.8607047449433916, + "grad_norm": 12.8125, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.8276557922363281, + "num_tokens": 858089834.0, + "step": 22488 + }, + { + "epoch": 2.860831955221982, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8087825179100037, + "num_tokens": 858124152.0, + "step": 22489 + }, + { + "epoch": 2.8609591655005726, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.8016401529312134, + "num_tokens": 858163202.0, + "step": 22490 + }, + { + "epoch": 2.8610863757791627, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8220240473747253, + "num_tokens": 858199173.0, + "step": 22491 + }, + { + "epoch": 2.8612135860577537, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.797465980052948, + "num_tokens": 858239693.0, + "step": 22492 + }, + { + "epoch": 2.8613407963363438, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.8138436675071716, + "num_tokens": 858280720.0, + "step": 22493 + }, + { + "epoch": 2.8614680066149347, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.813574492931366, + "num_tokens": 858314145.0, + "step": 22494 + }, + { + "epoch": 2.861595216893525, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8049740791320801, + "num_tokens": 858355177.0, + "step": 22495 + }, + { + "epoch": 2.861722427172116, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.8027044534683228, + "num_tokens": 858393070.0, + "step": 22496 + }, + { + "epoch": 2.861849637450706, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.799504280090332, + "num_tokens": 858431827.0, + "step": 22497 + }, + { + "epoch": 2.8619768477292964, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.8033720254898071, + "num_tokens": 858474990.0, + "step": 22498 + }, + { + "epoch": 2.862104058007887, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.8140560388565063, + "num_tokens": 858514698.0, + "step": 22499 + }, + { + "epoch": 2.8622312682864774, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.8001185059547424, + "num_tokens": 858557601.0, + "step": 22500 + }, + { + "epoch": 2.862358478565068, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8122028708457947, + "num_tokens": 858596830.0, + "step": 22501 + }, + { + "epoch": 2.8624856888436585, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7951793670654297, + "num_tokens": 858637555.0, + "step": 22502 + }, + { + "epoch": 2.862612899122249, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.5732, + "mean_token_accuracy": 0.8205053806304932, + "num_tokens": 858672813.0, + "step": 22503 + }, + { + "epoch": 2.8627401094008396, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8092870712280273, + "num_tokens": 858710743.0, + "step": 22504 + }, + { + "epoch": 2.86286731967943, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8124604821205139, + "num_tokens": 858739486.0, + "step": 22505 + }, + { + "epoch": 2.8629945299580206, + "grad_norm": 48.5, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7758761048316956, + "num_tokens": 858780678.0, + "step": 22506 + }, + { + "epoch": 2.863121740236611, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8025802969932556, + "num_tokens": 858826112.0, + "step": 22507 + }, + { + "epoch": 2.8632489505152017, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.8182897567749023, + "num_tokens": 858862158.0, + "step": 22508 + }, + { + "epoch": 2.863376160793792, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.807786226272583, + "num_tokens": 858905270.0, + "step": 22509 + }, + { + "epoch": 2.8635033710723827, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7951991558074951, + "num_tokens": 858941802.0, + "step": 22510 + }, + { + "epoch": 2.8636305813509733, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.8037724494934082, + "num_tokens": 858974672.0, + "step": 22511 + }, + { + "epoch": 2.863757791629564, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8014001250267029, + "num_tokens": 859015305.0, + "step": 22512 + }, + { + "epoch": 2.8638850019081543, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.79389888048172, + "num_tokens": 859050646.0, + "step": 22513 + }, + { + "epoch": 2.864012212186745, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.8139200806617737, + "num_tokens": 859090139.0, + "step": 22514 + }, + { + "epoch": 2.8641394224653354, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.8081923127174377, + "num_tokens": 859130661.0, + "step": 22515 + }, + { + "epoch": 2.8642666327439255, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8021731376647949, + "num_tokens": 859166399.0, + "step": 22516 + }, + { + "epoch": 2.8643938430225164, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.8094074726104736, + "num_tokens": 859208349.0, + "step": 22517 + }, + { + "epoch": 2.8645210533011065, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8078039884567261, + "num_tokens": 859243784.0, + "step": 22518 + }, + { + "epoch": 2.8646482635796975, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7817291021347046, + "num_tokens": 859278345.0, + "step": 22519 + }, + { + "epoch": 2.8647754738582876, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7914096117019653, + "num_tokens": 859317454.0, + "step": 22520 + }, + { + "epoch": 2.864902684136878, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7980892062187195, + "num_tokens": 859352076.0, + "step": 22521 + }, + { + "epoch": 2.8650298944154686, + "grad_norm": 10.75, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8124774694442749, + "num_tokens": 859389957.0, + "step": 22522 + }, + { + "epoch": 2.865157104694059, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8005698919296265, + "num_tokens": 859425167.0, + "step": 22523 + }, + { + "epoch": 2.8652843149726497, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.800404965877533, + "num_tokens": 859470999.0, + "step": 22524 + }, + { + "epoch": 2.86541152525124, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.8085123300552368, + "num_tokens": 859503763.0, + "step": 22525 + }, + { + "epoch": 2.8655387355298307, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.8119668364524841, + "num_tokens": 859537739.0, + "step": 22526 + }, + { + "epoch": 2.8656659458084213, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.8148621320724487, + "num_tokens": 859573497.0, + "step": 22527 + }, + { + "epoch": 2.865793156087012, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7972443103790283, + "num_tokens": 859606997.0, + "step": 22528 + }, + { + "epoch": 2.8659203663656023, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7895007133483887, + "num_tokens": 859642278.0, + "step": 22529 + }, + { + "epoch": 2.866047576644193, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8183491826057434, + "num_tokens": 859684728.0, + "step": 22530 + }, + { + "epoch": 2.8661747869227834, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7925451397895813, + "num_tokens": 859720946.0, + "step": 22531 + }, + { + "epoch": 2.866301997201374, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7935462594032288, + "num_tokens": 859752605.0, + "step": 22532 + }, + { + "epoch": 2.8664292074799644, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8149343729019165, + "num_tokens": 859784206.0, + "step": 22533 + }, + { + "epoch": 2.866556417758555, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7997756600379944, + "num_tokens": 859828142.0, + "step": 22534 + }, + { + "epoch": 2.8666836280371455, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7976683378219604, + "num_tokens": 859870032.0, + "step": 22535 + }, + { + "epoch": 2.866810838315736, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7925583124160767, + "num_tokens": 859911417.0, + "step": 22536 + }, + { + "epoch": 2.8669380485943265, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.8163167834281921, + "num_tokens": 859953929.0, + "step": 22537 + }, + { + "epoch": 2.867065258872917, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.8022935390472412, + "num_tokens": 859985402.0, + "step": 22538 + }, + { + "epoch": 2.8671924691515076, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7919065952301025, + "num_tokens": 860017374.0, + "step": 22539 + }, + { + "epoch": 2.867319679430098, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8106064796447754, + "num_tokens": 860058217.0, + "step": 22540 + }, + { + "epoch": 2.867446889708688, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7837061285972595, + "num_tokens": 860092317.0, + "step": 22541 + }, + { + "epoch": 2.867574099987279, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.797463595867157, + "num_tokens": 860125067.0, + "step": 22542 + }, + { + "epoch": 2.8677013102658693, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8176031708717346, + "num_tokens": 860163805.0, + "step": 22543 + }, + { + "epoch": 2.8678285205444602, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7887216210365295, + "num_tokens": 860200178.0, + "step": 22544 + }, + { + "epoch": 2.8679557308230503, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.8019700050354004, + "num_tokens": 860235283.0, + "step": 22545 + }, + { + "epoch": 2.868082941101641, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8253738880157471, + "num_tokens": 860276776.0, + "step": 22546 + }, + { + "epoch": 2.8682101513802314, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8078330755233765, + "num_tokens": 860318492.0, + "step": 22547 + }, + { + "epoch": 2.868337361658822, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7859813570976257, + "num_tokens": 860358556.0, + "step": 22548 + }, + { + "epoch": 2.8684645719374124, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7836323976516724, + "num_tokens": 860391114.0, + "step": 22549 + }, + { + "epoch": 2.868591782216003, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7968522310256958, + "num_tokens": 860430696.0, + "step": 22550 + }, + { + "epoch": 2.8687189924945935, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.809542179107666, + "num_tokens": 860471055.0, + "step": 22551 + }, + { + "epoch": 2.868846202773184, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7943276166915894, + "num_tokens": 860512472.0, + "step": 22552 + }, + { + "epoch": 2.8689734130517746, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7950356006622314, + "num_tokens": 860541014.0, + "step": 22553 + }, + { + "epoch": 2.869100623330365, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8156478404998779, + "num_tokens": 860585025.0, + "step": 22554 + }, + { + "epoch": 2.8692278336089556, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.8034591674804688, + "num_tokens": 860616361.0, + "step": 22555 + }, + { + "epoch": 2.869355043887546, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7971135377883911, + "num_tokens": 860655976.0, + "step": 22556 + }, + { + "epoch": 2.8694822541661367, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7843379378318787, + "num_tokens": 860703949.0, + "step": 22557 + }, + { + "epoch": 2.869609464444727, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.8055124282836914, + "num_tokens": 860739935.0, + "step": 22558 + }, + { + "epoch": 2.8697366747233177, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.808310866355896, + "num_tokens": 860782483.0, + "step": 22559 + }, + { + "epoch": 2.8698638850019083, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7888162136077881, + "num_tokens": 860816821.0, + "step": 22560 + }, + { + "epoch": 2.869991095280499, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7889118790626526, + "num_tokens": 860851313.0, + "step": 22561 + }, + { + "epoch": 2.8701183055590893, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8134487867355347, + "num_tokens": 860890540.0, + "step": 22562 + }, + { + "epoch": 2.87024551583768, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7934450507164001, + "num_tokens": 860929686.0, + "step": 22563 + }, + { + "epoch": 2.87037272611627, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8080859780311584, + "num_tokens": 860968186.0, + "step": 22564 + }, + { + "epoch": 2.870499936394861, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.792397677898407, + "num_tokens": 861007252.0, + "step": 22565 + }, + { + "epoch": 2.870627146673451, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.8054206967353821, + "num_tokens": 861049898.0, + "step": 22566 + }, + { + "epoch": 2.870754356952042, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.812656044960022, + "num_tokens": 861087711.0, + "step": 22567 + }, + { + "epoch": 2.870881567230632, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7803287506103516, + "num_tokens": 861128566.0, + "step": 22568 + }, + { + "epoch": 2.871008777509223, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.8058061599731445, + "num_tokens": 861156299.0, + "step": 22569 + }, + { + "epoch": 2.871135987787813, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8090836405754089, + "num_tokens": 861189376.0, + "step": 22570 + }, + { + "epoch": 2.8712631980664036, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7948817014694214, + "num_tokens": 861227156.0, + "step": 22571 + }, + { + "epoch": 2.871390408344994, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7985557317733765, + "num_tokens": 861268088.0, + "step": 22572 + }, + { + "epoch": 2.8715176186235847, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.8047183752059937, + "num_tokens": 861308892.0, + "step": 22573 + }, + { + "epoch": 2.871644828902175, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8050913214683533, + "num_tokens": 861350108.0, + "step": 22574 + }, + { + "epoch": 2.8717720391807657, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8157931566238403, + "num_tokens": 861390553.0, + "step": 22575 + }, + { + "epoch": 2.8718992494593563, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.8052419424057007, + "num_tokens": 861425878.0, + "step": 22576 + }, + { + "epoch": 2.872026459737947, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8178641200065613, + "num_tokens": 861460935.0, + "step": 22577 + }, + { + "epoch": 2.8721536700165373, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7886754274368286, + "num_tokens": 861500446.0, + "step": 22578 + }, + { + "epoch": 2.872280880295128, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7863348126411438, + "num_tokens": 861538000.0, + "step": 22579 + }, + { + "epoch": 2.8724080905737184, + "grad_norm": 39.5, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.8062992691993713, + "num_tokens": 861578052.0, + "step": 22580 + }, + { + "epoch": 2.872535300852309, + "grad_norm": 79.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7996682524681091, + "num_tokens": 861619065.0, + "step": 22581 + }, + { + "epoch": 2.8726625111308994, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.8040682673454285, + "num_tokens": 861660112.0, + "step": 22582 + }, + { + "epoch": 2.87278972140949, + "grad_norm": 31.625, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.8072012662887573, + "num_tokens": 861699312.0, + "step": 22583 + }, + { + "epoch": 2.8729169316880805, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8208723068237305, + "num_tokens": 861735167.0, + "step": 22584 + }, + { + "epoch": 2.873044141966671, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7967642545700073, + "num_tokens": 861778594.0, + "step": 22585 + }, + { + "epoch": 2.8731713522452615, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.8021432757377625, + "num_tokens": 861814754.0, + "step": 22586 + }, + { + "epoch": 2.873298562523852, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7880231142044067, + "num_tokens": 861849019.0, + "step": 22587 + }, + { + "epoch": 2.8734257728024426, + "grad_norm": 47.5, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7991107702255249, + "num_tokens": 861889471.0, + "step": 22588 + }, + { + "epoch": 2.8735529830810327, + "grad_norm": 55.5, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7938851118087769, + "num_tokens": 861926811.0, + "step": 22589 + }, + { + "epoch": 2.8736801933596237, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.797064483165741, + "num_tokens": 861963628.0, + "step": 22590 + }, + { + "epoch": 2.8738074036382137, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.786628246307373, + "num_tokens": 862002915.0, + "step": 22591 + }, + { + "epoch": 2.8739346139168047, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8069187998771667, + "num_tokens": 862047276.0, + "step": 22592 + }, + { + "epoch": 2.874061824195395, + "grad_norm": 38.75, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.8030576705932617, + "num_tokens": 862091241.0, + "step": 22593 + }, + { + "epoch": 2.8741890344739858, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.8038479089736938, + "num_tokens": 862126025.0, + "step": 22594 + }, + { + "epoch": 2.874316244752576, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7893264889717102, + "num_tokens": 862160822.0, + "step": 22595 + }, + { + "epoch": 2.8744434550311664, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8042888641357422, + "num_tokens": 862196808.0, + "step": 22596 + }, + { + "epoch": 2.874570665309757, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7830127477645874, + "num_tokens": 862233722.0, + "step": 22597 + }, + { + "epoch": 2.8746978755883474, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8132702112197876, + "num_tokens": 862271914.0, + "step": 22598 + }, + { + "epoch": 2.874825085866938, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7975746393203735, + "num_tokens": 862313696.0, + "step": 22599 + }, + { + "epoch": 2.8749522961455285, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7959628701210022, + "num_tokens": 862355366.0, + "step": 22600 + }, + { + "epoch": 2.875079506424119, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7935259342193604, + "num_tokens": 862395630.0, + "step": 22601 + }, + { + "epoch": 2.8752067167027096, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.790735125541687, + "num_tokens": 862429650.0, + "step": 22602 + }, + { + "epoch": 2.8753339269813, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7975594997406006, + "num_tokens": 862470921.0, + "step": 22603 + }, + { + "epoch": 2.8754611372598906, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8054971098899841, + "num_tokens": 862505204.0, + "step": 22604 + }, + { + "epoch": 2.875588347538481, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7952617406845093, + "num_tokens": 862548686.0, + "step": 22605 + }, + { + "epoch": 2.8757155578170717, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.8127517700195312, + "num_tokens": 862585920.0, + "step": 22606 + }, + { + "epoch": 2.875842768095662, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7909534573554993, + "num_tokens": 862619364.0, + "step": 22607 + }, + { + "epoch": 2.8759699783742527, + "grad_norm": 40.25, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7974592447280884, + "num_tokens": 862661317.0, + "step": 22608 + }, + { + "epoch": 2.8760971886528433, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.8069027662277222, + "num_tokens": 862698236.0, + "step": 22609 + }, + { + "epoch": 2.876224398931434, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8114207983016968, + "num_tokens": 862739224.0, + "step": 22610 + }, + { + "epoch": 2.8763516092100243, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8160648941993713, + "num_tokens": 862776534.0, + "step": 22611 + }, + { + "epoch": 2.876478819488615, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7899796366691589, + "num_tokens": 862815690.0, + "step": 22612 + }, + { + "epoch": 2.8766060297672054, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7979202270507812, + "num_tokens": 862851021.0, + "step": 22613 + }, + { + "epoch": 2.8767332400457954, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8016424775123596, + "num_tokens": 862885122.0, + "step": 22614 + }, + { + "epoch": 2.8768604503243864, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8097494840621948, + "num_tokens": 862922233.0, + "step": 22615 + }, + { + "epoch": 2.8769876606029765, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8138158917427063, + "num_tokens": 862961355.0, + "step": 22616 + }, + { + "epoch": 2.8771148708815675, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8133355975151062, + "num_tokens": 862998523.0, + "step": 22617 + }, + { + "epoch": 2.8772420811601576, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7990792989730835, + "num_tokens": 863039684.0, + "step": 22618 + }, + { + "epoch": 2.877369291438748, + "grad_norm": 61.5, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8017282485961914, + "num_tokens": 863077649.0, + "step": 22619 + }, + { + "epoch": 2.8774965017173386, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.8005868196487427, + "num_tokens": 863116816.0, + "step": 22620 + }, + { + "epoch": 2.877623711995929, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8119874000549316, + "num_tokens": 863155306.0, + "step": 22621 + }, + { + "epoch": 2.8777509222745197, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8107876181602478, + "num_tokens": 863194381.0, + "step": 22622 + }, + { + "epoch": 2.87787813255311, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7981276512145996, + "num_tokens": 863231470.0, + "step": 22623 + }, + { + "epoch": 2.8780053428317007, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7866899967193604, + "num_tokens": 863265617.0, + "step": 22624 + }, + { + "epoch": 2.8781325531102913, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7909923195838928, + "num_tokens": 863300111.0, + "step": 22625 + }, + { + "epoch": 2.878259763388882, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8096659183502197, + "num_tokens": 863338626.0, + "step": 22626 + }, + { + "epoch": 2.8783869736674723, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.795473575592041, + "num_tokens": 863374672.0, + "step": 22627 + }, + { + "epoch": 2.878514183946063, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7928056716918945, + "num_tokens": 863412717.0, + "step": 22628 + }, + { + "epoch": 2.8786413942246534, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8113532662391663, + "num_tokens": 863447776.0, + "step": 22629 + }, + { + "epoch": 2.878768604503244, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8165059089660645, + "num_tokens": 863485697.0, + "step": 22630 + }, + { + "epoch": 2.8788958147818344, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7954424619674683, + "num_tokens": 863528706.0, + "step": 22631 + }, + { + "epoch": 2.879023025060425, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7835277915000916, + "num_tokens": 863569181.0, + "step": 22632 + }, + { + "epoch": 2.8791502353390155, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7804502248764038, + "num_tokens": 863602985.0, + "step": 22633 + }, + { + "epoch": 2.879277445617606, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8042275905609131, + "num_tokens": 863634590.0, + "step": 22634 + }, + { + "epoch": 2.8794046558961965, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7990570068359375, + "num_tokens": 863677671.0, + "step": 22635 + }, + { + "epoch": 2.879531866174787, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.8112538456916809, + "num_tokens": 863716102.0, + "step": 22636 + }, + { + "epoch": 2.8796590764533776, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7833454608917236, + "num_tokens": 863751560.0, + "step": 22637 + }, + { + "epoch": 2.879786286731968, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.8223472237586975, + "num_tokens": 863792373.0, + "step": 22638 + }, + { + "epoch": 2.879913497010558, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.8004302382469177, + "num_tokens": 863829969.0, + "step": 22639 + }, + { + "epoch": 2.880040707289149, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8124610781669617, + "num_tokens": 863867080.0, + "step": 22640 + }, + { + "epoch": 2.8801679175677393, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.808963418006897, + "num_tokens": 863904230.0, + "step": 22641 + }, + { + "epoch": 2.8802951278463302, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8058633208274841, + "num_tokens": 863947309.0, + "step": 22642 + }, + { + "epoch": 2.8804223381249203, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8126182556152344, + "num_tokens": 863985557.0, + "step": 22643 + }, + { + "epoch": 2.880549548403511, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7900550961494446, + "num_tokens": 864021842.0, + "step": 22644 + }, + { + "epoch": 2.8806767586821014, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7942867279052734, + "num_tokens": 864060765.0, + "step": 22645 + }, + { + "epoch": 2.880803968960692, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.805453896522522, + "num_tokens": 864100624.0, + "step": 22646 + }, + { + "epoch": 2.8809311792392824, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8109360933303833, + "num_tokens": 864136599.0, + "step": 22647 + }, + { + "epoch": 2.881058389517873, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8001288175582886, + "num_tokens": 864169876.0, + "step": 22648 + }, + { + "epoch": 2.8811855997964635, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8028643131256104, + "num_tokens": 864213709.0, + "step": 22649 + }, + { + "epoch": 2.881312810075054, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.8026525974273682, + "num_tokens": 864245581.0, + "step": 22650 + }, + { + "epoch": 2.8814400203536445, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7860981822013855, + "num_tokens": 864287081.0, + "step": 22651 + }, + { + "epoch": 2.881567230632235, + "grad_norm": 68.5, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7830569744110107, + "num_tokens": 864321480.0, + "step": 22652 + }, + { + "epoch": 2.8816944409108256, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.8093311786651611, + "num_tokens": 864358580.0, + "step": 22653 + }, + { + "epoch": 2.881821651189416, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.785927414894104, + "num_tokens": 864396277.0, + "step": 22654 + }, + { + "epoch": 2.8819488614680067, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7970176339149475, + "num_tokens": 864433282.0, + "step": 22655 + }, + { + "epoch": 2.882076071746597, + "grad_norm": 27.375, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7961583733558655, + "num_tokens": 864469305.0, + "step": 22656 + }, + { + "epoch": 2.8822032820251877, + "grad_norm": 27.75, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.8022927641868591, + "num_tokens": 864507246.0, + "step": 22657 + }, + { + "epoch": 2.8823304923037782, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7966644167900085, + "num_tokens": 864547315.0, + "step": 22658 + }, + { + "epoch": 2.8824577025823688, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.794963002204895, + "num_tokens": 864590813.0, + "step": 22659 + }, + { + "epoch": 2.8825849128609593, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7871851325035095, + "num_tokens": 864625078.0, + "step": 22660 + }, + { + "epoch": 2.88271212313955, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.812960684299469, + "num_tokens": 864663714.0, + "step": 22661 + }, + { + "epoch": 2.88283933341814, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7972477674484253, + "num_tokens": 864697026.0, + "step": 22662 + }, + { + "epoch": 2.882966543696731, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8112658262252808, + "num_tokens": 864732337.0, + "step": 22663 + }, + { + "epoch": 2.883093753975321, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7973453998565674, + "num_tokens": 864776598.0, + "step": 22664 + }, + { + "epoch": 2.883220964253912, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7922108769416809, + "num_tokens": 864817279.0, + "step": 22665 + }, + { + "epoch": 2.883348174532502, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8122929334640503, + "num_tokens": 864854914.0, + "step": 22666 + }, + { + "epoch": 2.883475384811093, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.8088088631629944, + "num_tokens": 864892914.0, + "step": 22667 + }, + { + "epoch": 2.883602595089683, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7927630543708801, + "num_tokens": 864934189.0, + "step": 22668 + }, + { + "epoch": 2.8837298053682736, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8158597946166992, + "num_tokens": 864972085.0, + "step": 22669 + }, + { + "epoch": 2.883857015646864, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.779460608959198, + "num_tokens": 865008674.0, + "step": 22670 + }, + { + "epoch": 2.8839842259254547, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7893908023834229, + "num_tokens": 865046004.0, + "step": 22671 + }, + { + "epoch": 2.884111436204045, + "grad_norm": 12.0, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7979385852813721, + "num_tokens": 865082409.0, + "step": 22672 + }, + { + "epoch": 2.8842386464826357, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8047546744346619, + "num_tokens": 865124457.0, + "step": 22673 + }, + { + "epoch": 2.8843658567612263, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.8159166574478149, + "num_tokens": 865162290.0, + "step": 22674 + }, + { + "epoch": 2.884493067039817, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7924414873123169, + "num_tokens": 865201686.0, + "step": 22675 + }, + { + "epoch": 2.8846202773184073, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7828755378723145, + "num_tokens": 865236876.0, + "step": 22676 + }, + { + "epoch": 2.884747487596998, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8062845468521118, + "num_tokens": 865276994.0, + "step": 22677 + }, + { + "epoch": 2.8848746978755884, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7992160320281982, + "num_tokens": 865311990.0, + "step": 22678 + }, + { + "epoch": 2.885001908154179, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.8074939846992493, + "num_tokens": 865352324.0, + "step": 22679 + }, + { + "epoch": 2.8851291184327694, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.8029451370239258, + "num_tokens": 865391428.0, + "step": 22680 + }, + { + "epoch": 2.88525632871136, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7949348092079163, + "num_tokens": 865428478.0, + "step": 22681 + }, + { + "epoch": 2.8853835389899505, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.8003530502319336, + "num_tokens": 865466531.0, + "step": 22682 + }, + { + "epoch": 2.885510749268541, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8116765022277832, + "num_tokens": 865503177.0, + "step": 22683 + }, + { + "epoch": 2.8856379595471315, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7870991230010986, + "num_tokens": 865539454.0, + "step": 22684 + }, + { + "epoch": 2.885765169825722, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.8165729641914368, + "num_tokens": 865571937.0, + "step": 22685 + }, + { + "epoch": 2.8858923801043126, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.8007009625434875, + "num_tokens": 865609457.0, + "step": 22686 + }, + { + "epoch": 2.8860195903829027, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7853825688362122, + "num_tokens": 865647391.0, + "step": 22687 + }, + { + "epoch": 2.8861468006614936, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7972846031188965, + "num_tokens": 865680578.0, + "step": 22688 + }, + { + "epoch": 2.8862740109400837, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7982099056243896, + "num_tokens": 865714610.0, + "step": 22689 + }, + { + "epoch": 2.8864012212186747, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7975252270698547, + "num_tokens": 865755672.0, + "step": 22690 + }, + { + "epoch": 2.886528431497265, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8017335534095764, + "num_tokens": 865790111.0, + "step": 22691 + }, + { + "epoch": 2.8866556417758558, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7965925335884094, + "num_tokens": 865825580.0, + "step": 22692 + }, + { + "epoch": 2.886782852054446, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.8045694828033447, + "num_tokens": 865869099.0, + "step": 22693 + }, + { + "epoch": 2.8869100623330364, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7918915152549744, + "num_tokens": 865906957.0, + "step": 22694 + }, + { + "epoch": 2.887037272611627, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8096729516983032, + "num_tokens": 865945712.0, + "step": 22695 + }, + { + "epoch": 2.8871644828902174, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.784416913986206, + "num_tokens": 865984469.0, + "step": 22696 + }, + { + "epoch": 2.887291693168808, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8030574321746826, + "num_tokens": 866024624.0, + "step": 22697 + }, + { + "epoch": 2.8874189034473985, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7942178249359131, + "num_tokens": 866062446.0, + "step": 22698 + }, + { + "epoch": 2.887546113725989, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8158890604972839, + "num_tokens": 866100896.0, + "step": 22699 + }, + { + "epoch": 2.8876733240045795, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7835171222686768, + "num_tokens": 866134668.0, + "step": 22700 + }, + { + "epoch": 2.88780053428317, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.793268084526062, + "num_tokens": 866173834.0, + "step": 22701 + }, + { + "epoch": 2.8879277445617606, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.781733512878418, + "num_tokens": 866211691.0, + "step": 22702 + }, + { + "epoch": 2.888054954840351, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7944517135620117, + "num_tokens": 866240878.0, + "step": 22703 + }, + { + "epoch": 2.8881821651189417, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7870038151741028, + "num_tokens": 866289552.0, + "step": 22704 + }, + { + "epoch": 2.888309375397532, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.787379264831543, + "num_tokens": 866330361.0, + "step": 22705 + }, + { + "epoch": 2.8884365856761227, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8050758838653564, + "num_tokens": 866371557.0, + "step": 22706 + }, + { + "epoch": 2.8885637959547132, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.803107500076294, + "num_tokens": 866410627.0, + "step": 22707 + }, + { + "epoch": 2.8886910062333038, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7956917881965637, + "num_tokens": 866446036.0, + "step": 22708 + }, + { + "epoch": 2.8888182165118943, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7893885374069214, + "num_tokens": 866489655.0, + "step": 22709 + }, + { + "epoch": 2.888945426790485, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7977674603462219, + "num_tokens": 866532750.0, + "step": 22710 + }, + { + "epoch": 2.8890726370690754, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8071181774139404, + "num_tokens": 866570748.0, + "step": 22711 + }, + { + "epoch": 2.8891998473476654, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7938218116760254, + "num_tokens": 866609422.0, + "step": 22712 + }, + { + "epoch": 2.8893270576262564, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.8047155141830444, + "num_tokens": 866650865.0, + "step": 22713 + }, + { + "epoch": 2.8894542679048465, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7993572950363159, + "num_tokens": 866687605.0, + "step": 22714 + }, + { + "epoch": 2.8895814781834375, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7854741811752319, + "num_tokens": 866722974.0, + "step": 22715 + }, + { + "epoch": 2.8897086884620276, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7974281907081604, + "num_tokens": 866760694.0, + "step": 22716 + }, + { + "epoch": 2.889835898740618, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7774338722229004, + "num_tokens": 866795102.0, + "step": 22717 + }, + { + "epoch": 2.8899631090192086, + "grad_norm": 15.1875, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8189159631729126, + "num_tokens": 866831282.0, + "step": 22718 + }, + { + "epoch": 2.890090319297799, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7911533117294312, + "num_tokens": 866872661.0, + "step": 22719 + }, + { + "epoch": 2.8902175295763897, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7973437905311584, + "num_tokens": 866917576.0, + "step": 22720 + }, + { + "epoch": 2.89034473985498, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.8033955097198486, + "num_tokens": 866957245.0, + "step": 22721 + }, + { + "epoch": 2.8904719501335707, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7708298563957214, + "num_tokens": 866998695.0, + "step": 22722 + }, + { + "epoch": 2.8905991604121613, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7877058386802673, + "num_tokens": 867033502.0, + "step": 22723 + }, + { + "epoch": 2.890726370690752, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7860121130943298, + "num_tokens": 867076032.0, + "step": 22724 + }, + { + "epoch": 2.8908535809693423, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7983827590942383, + "num_tokens": 867107429.0, + "step": 22725 + }, + { + "epoch": 2.890980791247933, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.8030024170875549, + "num_tokens": 867139432.0, + "step": 22726 + }, + { + "epoch": 2.8911080015265234, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8025898933410645, + "num_tokens": 867180588.0, + "step": 22727 + }, + { + "epoch": 2.891235211805114, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.794646680355072, + "num_tokens": 867215170.0, + "step": 22728 + }, + { + "epoch": 2.8913624220837044, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8204315900802612, + "num_tokens": 867254958.0, + "step": 22729 + }, + { + "epoch": 2.891489632362295, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7988137006759644, + "num_tokens": 867291135.0, + "step": 22730 + }, + { + "epoch": 2.8916168426408855, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8061621189117432, + "num_tokens": 867326926.0, + "step": 22731 + }, + { + "epoch": 2.891744052919476, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7874211072921753, + "num_tokens": 867361142.0, + "step": 22732 + }, + { + "epoch": 2.8918712631980665, + "grad_norm": 31.75, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8205639719963074, + "num_tokens": 867398344.0, + "step": 22733 + }, + { + "epoch": 2.891998473476657, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.8128548264503479, + "num_tokens": 867437642.0, + "step": 22734 + }, + { + "epoch": 2.8921256837552476, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7887517809867859, + "num_tokens": 867471646.0, + "step": 22735 + }, + { + "epoch": 2.892252894033838, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7986239194869995, + "num_tokens": 867509682.0, + "step": 22736 + }, + { + "epoch": 2.892380104312428, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7910286784172058, + "num_tokens": 867544425.0, + "step": 22737 + }, + { + "epoch": 2.892507314591019, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7941219806671143, + "num_tokens": 867584978.0, + "step": 22738 + }, + { + "epoch": 2.8926345248696093, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.802478551864624, + "num_tokens": 867620360.0, + "step": 22739 + }, + { + "epoch": 2.8927617351482002, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8037824630737305, + "num_tokens": 867655246.0, + "step": 22740 + }, + { + "epoch": 2.8928889454267903, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7919443845748901, + "num_tokens": 867692792.0, + "step": 22741 + }, + { + "epoch": 2.893016155705381, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.8044546842575073, + "num_tokens": 867732365.0, + "step": 22742 + }, + { + "epoch": 2.8931433659839714, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7793294787406921, + "num_tokens": 867764800.0, + "step": 22743 + }, + { + "epoch": 2.893270576262562, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7984250783920288, + "num_tokens": 867807107.0, + "step": 22744 + }, + { + "epoch": 2.8933977865411524, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7909095287322998, + "num_tokens": 867850225.0, + "step": 22745 + }, + { + "epoch": 2.893524996819743, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7954719066619873, + "num_tokens": 867890457.0, + "step": 22746 + }, + { + "epoch": 2.8936522070983335, + "grad_norm": 10.625, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.8113241195678711, + "num_tokens": 867925880.0, + "step": 22747 + }, + { + "epoch": 2.893779417376924, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.8132727146148682, + "num_tokens": 867963923.0, + "step": 22748 + }, + { + "epoch": 2.8939066276555145, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8007190227508545, + "num_tokens": 868000489.0, + "step": 22749 + }, + { + "epoch": 2.894033837934105, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7977485656738281, + "num_tokens": 868034001.0, + "step": 22750 + }, + { + "epoch": 2.8941610482126956, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.8063369989395142, + "num_tokens": 868074956.0, + "step": 22751 + }, + { + "epoch": 2.894288258491286, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8103123903274536, + "num_tokens": 868110727.0, + "step": 22752 + }, + { + "epoch": 2.8944154687698767, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7765620946884155, + "num_tokens": 868151480.0, + "step": 22753 + }, + { + "epoch": 2.894542679048467, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7870792150497437, + "num_tokens": 868190620.0, + "step": 22754 + }, + { + "epoch": 2.8946698893270577, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7940205335617065, + "num_tokens": 868230598.0, + "step": 22755 + }, + { + "epoch": 2.8947970996056482, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7951418161392212, + "num_tokens": 868265933.0, + "step": 22756 + }, + { + "epoch": 2.8949243098842388, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8078162670135498, + "num_tokens": 868300282.0, + "step": 22757 + }, + { + "epoch": 2.8950515201628293, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.818852424621582, + "num_tokens": 868331031.0, + "step": 22758 + }, + { + "epoch": 2.89517873044142, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7968358397483826, + "num_tokens": 868368268.0, + "step": 22759 + }, + { + "epoch": 2.89530594072001, + "grad_norm": 14.75, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7825757265090942, + "num_tokens": 868407551.0, + "step": 22760 + }, + { + "epoch": 2.895433150998601, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8008927702903748, + "num_tokens": 868449301.0, + "step": 22761 + }, + { + "epoch": 2.895560361277191, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7969555258750916, + "num_tokens": 868482853.0, + "step": 22762 + }, + { + "epoch": 2.895687571555782, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7989397048950195, + "num_tokens": 868523960.0, + "step": 22763 + }, + { + "epoch": 2.895814781834372, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7934571504592896, + "num_tokens": 868566447.0, + "step": 22764 + }, + { + "epoch": 2.895941992112963, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.8168519735336304, + "num_tokens": 868608342.0, + "step": 22765 + }, + { + "epoch": 2.896069202391553, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8124308586120605, + "num_tokens": 868647354.0, + "step": 22766 + }, + { + "epoch": 2.8961964126701436, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.781280517578125, + "num_tokens": 868690716.0, + "step": 22767 + }, + { + "epoch": 2.896323622948734, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8081732988357544, + "num_tokens": 868728318.0, + "step": 22768 + }, + { + "epoch": 2.8964508332273247, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8148149251937866, + "num_tokens": 868764323.0, + "step": 22769 + }, + { + "epoch": 2.896578043505915, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.781218409538269, + "num_tokens": 868801371.0, + "step": 22770 + }, + { + "epoch": 2.8967052537845057, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8101764917373657, + "num_tokens": 868846679.0, + "step": 22771 + }, + { + "epoch": 2.8968324640630962, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7865851521492004, + "num_tokens": 868880076.0, + "step": 22772 + }, + { + "epoch": 2.8969596743416868, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.8012375235557556, + "num_tokens": 868917086.0, + "step": 22773 + }, + { + "epoch": 2.8970868846202773, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.8079953789710999, + "num_tokens": 868954952.0, + "step": 22774 + }, + { + "epoch": 2.897214094898868, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7904084920883179, + "num_tokens": 868990911.0, + "step": 22775 + }, + { + "epoch": 2.8973413051774584, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8129484057426453, + "num_tokens": 869033398.0, + "step": 22776 + }, + { + "epoch": 2.897468515456049, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.786596417427063, + "num_tokens": 869070513.0, + "step": 22777 + }, + { + "epoch": 2.8975957257346394, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.8091761469841003, + "num_tokens": 869113494.0, + "step": 22778 + }, + { + "epoch": 2.89772293601323, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8075945973396301, + "num_tokens": 869147433.0, + "step": 22779 + }, + { + "epoch": 2.8978501462918205, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7946509718894958, + "num_tokens": 869186763.0, + "step": 22780 + }, + { + "epoch": 2.897977356570411, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.8108600378036499, + "num_tokens": 869222624.0, + "step": 22781 + }, + { + "epoch": 2.8981045668490015, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8232739567756653, + "num_tokens": 869263027.0, + "step": 22782 + }, + { + "epoch": 2.898231777127592, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7950028777122498, + "num_tokens": 869304232.0, + "step": 22783 + }, + { + "epoch": 2.8983589874061826, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.8074493408203125, + "num_tokens": 869340141.0, + "step": 22784 + }, + { + "epoch": 2.8984861976847727, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7907081246376038, + "num_tokens": 869382428.0, + "step": 22785 + }, + { + "epoch": 2.8986134079633636, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8055387735366821, + "num_tokens": 869418890.0, + "step": 22786 + }, + { + "epoch": 2.8987406182419537, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7987620830535889, + "num_tokens": 869453704.0, + "step": 22787 + }, + { + "epoch": 2.8988678285205447, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7901564240455627, + "num_tokens": 869491845.0, + "step": 22788 + }, + { + "epoch": 2.898995038799135, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.813801646232605, + "num_tokens": 869523393.0, + "step": 22789 + }, + { + "epoch": 2.8991222490777258, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7959720492362976, + "num_tokens": 869561927.0, + "step": 22790 + }, + { + "epoch": 2.899249459356316, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7997105717658997, + "num_tokens": 869599568.0, + "step": 22791 + }, + { + "epoch": 2.8993766696349064, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8050285577774048, + "num_tokens": 869636329.0, + "step": 22792 + }, + { + "epoch": 2.899503879913497, + "grad_norm": 28.375, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.797694206237793, + "num_tokens": 869679720.0, + "step": 22793 + }, + { + "epoch": 2.8996310901920874, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.8028570413589478, + "num_tokens": 869719527.0, + "step": 22794 + }, + { + "epoch": 2.899758300470678, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7996997237205505, + "num_tokens": 869755139.0, + "step": 22795 + }, + { + "epoch": 2.8998855107492685, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7929291725158691, + "num_tokens": 869791860.0, + "step": 22796 + }, + { + "epoch": 2.900012721027859, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8219791054725647, + "num_tokens": 869827294.0, + "step": 22797 + }, + { + "epoch": 2.9001399313064495, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7938466668128967, + "num_tokens": 869859398.0, + "step": 22798 + }, + { + "epoch": 2.90026714158504, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8044193387031555, + "num_tokens": 869898215.0, + "step": 22799 + }, + { + "epoch": 2.9003943518636306, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7873550653457642, + "num_tokens": 869939787.0, + "step": 22800 + }, + { + "epoch": 2.900521562142221, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7900223135948181, + "num_tokens": 869976790.0, + "step": 22801 + }, + { + "epoch": 2.9006487724208116, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8026123642921448, + "num_tokens": 870013547.0, + "step": 22802 + }, + { + "epoch": 2.900775982699402, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8134154677391052, + "num_tokens": 870049540.0, + "step": 22803 + }, + { + "epoch": 2.9009031929779927, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.8084065914154053, + "num_tokens": 870088247.0, + "step": 22804 + }, + { + "epoch": 2.9010304032565832, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8070831894874573, + "num_tokens": 870129112.0, + "step": 22805 + }, + { + "epoch": 2.9011576135351738, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7986471056938171, + "num_tokens": 870162082.0, + "step": 22806 + }, + { + "epoch": 2.9012848238137643, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8081075549125671, + "num_tokens": 870199868.0, + "step": 22807 + }, + { + "epoch": 2.901412034092355, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8103801012039185, + "num_tokens": 870238121.0, + "step": 22808 + }, + { + "epoch": 2.9015392443709453, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.790246844291687, + "num_tokens": 870278008.0, + "step": 22809 + }, + { + "epoch": 2.9016664546495354, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7957369685173035, + "num_tokens": 870317671.0, + "step": 22810 + }, + { + "epoch": 2.9017936649281264, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.8004886507987976, + "num_tokens": 870352893.0, + "step": 22811 + }, + { + "epoch": 2.9019208752067165, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.810083270072937, + "num_tokens": 870387513.0, + "step": 22812 + }, + { + "epoch": 2.9020480854853075, + "grad_norm": 23.875, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8076720237731934, + "num_tokens": 870428649.0, + "step": 22813 + }, + { + "epoch": 2.9021752957638975, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.816021740436554, + "num_tokens": 870468329.0, + "step": 22814 + }, + { + "epoch": 2.902302506042488, + "grad_norm": 10.5625, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.8022361397743225, + "num_tokens": 870509273.0, + "step": 22815 + }, + { + "epoch": 2.9024297163210786, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7907192707061768, + "num_tokens": 870544206.0, + "step": 22816 + }, + { + "epoch": 2.902556926599669, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8203238248825073, + "num_tokens": 870578708.0, + "step": 22817 + }, + { + "epoch": 2.9026841368782597, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8127730488777161, + "num_tokens": 870620599.0, + "step": 22818 + }, + { + "epoch": 2.90281134715685, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8115285634994507, + "num_tokens": 870657828.0, + "step": 22819 + }, + { + "epoch": 2.9029385574354407, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7930892705917358, + "num_tokens": 870694786.0, + "step": 22820 + }, + { + "epoch": 2.9030657677140312, + "grad_norm": 10.25, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.8208448886871338, + "num_tokens": 870733850.0, + "step": 22821 + }, + { + "epoch": 2.9031929779926218, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8031538724899292, + "num_tokens": 870771556.0, + "step": 22822 + }, + { + "epoch": 2.9033201882712123, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8096847534179688, + "num_tokens": 870809335.0, + "step": 22823 + }, + { + "epoch": 2.903447398549803, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7981884479522705, + "num_tokens": 870844239.0, + "step": 22824 + }, + { + "epoch": 2.9035746088283934, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7995044589042664, + "num_tokens": 870879540.0, + "step": 22825 + }, + { + "epoch": 2.903701819106984, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7987379431724548, + "num_tokens": 870917219.0, + "step": 22826 + }, + { + "epoch": 2.9038290293855744, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8017656207084656, + "num_tokens": 870952332.0, + "step": 22827 + }, + { + "epoch": 2.903956239664165, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7910306453704834, + "num_tokens": 870991609.0, + "step": 22828 + }, + { + "epoch": 2.9040834499427555, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.8058188557624817, + "num_tokens": 871027456.0, + "step": 22829 + }, + { + "epoch": 2.904210660221346, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7898668050765991, + "num_tokens": 871071833.0, + "step": 22830 + }, + { + "epoch": 2.9043378704999365, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7974225878715515, + "num_tokens": 871105446.0, + "step": 22831 + }, + { + "epoch": 2.904465080778527, + "grad_norm": 33.0, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.80064457654953, + "num_tokens": 871141408.0, + "step": 22832 + }, + { + "epoch": 2.9045922910571176, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.805565595626831, + "num_tokens": 871173296.0, + "step": 22833 + }, + { + "epoch": 2.904719501335708, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.8016708493232727, + "num_tokens": 871210711.0, + "step": 22834 + }, + { + "epoch": 2.904846711614298, + "grad_norm": 11.875, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.8031071424484253, + "num_tokens": 871251179.0, + "step": 22835 + }, + { + "epoch": 2.904973921892889, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.5606, + "mean_token_accuracy": 0.8283716440200806, + "num_tokens": 871295758.0, + "step": 22836 + }, + { + "epoch": 2.9051011321714793, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7813268899917603, + "num_tokens": 871331145.0, + "step": 22837 + }, + { + "epoch": 2.9052283424500702, + "grad_norm": 22.75, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7723686099052429, + "num_tokens": 871372819.0, + "step": 22838 + }, + { + "epoch": 2.9053555527286603, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7983708381652832, + "num_tokens": 871414223.0, + "step": 22839 + }, + { + "epoch": 2.905482763007251, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8070422410964966, + "num_tokens": 871455013.0, + "step": 22840 + }, + { + "epoch": 2.9056099732858414, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7983036041259766, + "num_tokens": 871495462.0, + "step": 22841 + }, + { + "epoch": 2.905737183564432, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.8041313886642456, + "num_tokens": 871534999.0, + "step": 22842 + }, + { + "epoch": 2.9058643938430224, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.8000402450561523, + "num_tokens": 871569816.0, + "step": 22843 + }, + { + "epoch": 2.905991604121613, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.8075363039970398, + "num_tokens": 871610441.0, + "step": 22844 + }, + { + "epoch": 2.9061188144002035, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7867805361747742, + "num_tokens": 871649005.0, + "step": 22845 + }, + { + "epoch": 2.906246024678794, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7971714735031128, + "num_tokens": 871689263.0, + "step": 22846 + }, + { + "epoch": 2.9063732349573845, + "grad_norm": 14.0625, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8106420040130615, + "num_tokens": 871725440.0, + "step": 22847 + }, + { + "epoch": 2.906500445235975, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7974963188171387, + "num_tokens": 871767839.0, + "step": 22848 + }, + { + "epoch": 2.9066276555145656, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.805345892906189, + "num_tokens": 871801516.0, + "step": 22849 + }, + { + "epoch": 2.906754865793156, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.798926055431366, + "num_tokens": 871840359.0, + "step": 22850 + }, + { + "epoch": 2.9068820760717466, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8134509921073914, + "num_tokens": 871879396.0, + "step": 22851 + }, + { + "epoch": 2.907009286350337, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7926154136657715, + "num_tokens": 871915043.0, + "step": 22852 + }, + { + "epoch": 2.9071364966289277, + "grad_norm": 79.5, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7936290502548218, + "num_tokens": 871953391.0, + "step": 22853 + }, + { + "epoch": 2.9072637069075182, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.803148627281189, + "num_tokens": 871989879.0, + "step": 22854 + }, + { + "epoch": 2.9073909171861088, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7993193864822388, + "num_tokens": 872022428.0, + "step": 22855 + }, + { + "epoch": 2.9075181274646993, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8073773980140686, + "num_tokens": 872060644.0, + "step": 22856 + }, + { + "epoch": 2.90764533774329, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7940491437911987, + "num_tokens": 872098230.0, + "step": 22857 + }, + { + "epoch": 2.90777254802188, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7913388013839722, + "num_tokens": 872131217.0, + "step": 22858 + }, + { + "epoch": 2.907899758300471, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.8103224039077759, + "num_tokens": 872163912.0, + "step": 22859 + }, + { + "epoch": 2.908026968579061, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7748457193374634, + "num_tokens": 872200694.0, + "step": 22860 + }, + { + "epoch": 2.908154178857652, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8000580668449402, + "num_tokens": 872241156.0, + "step": 22861 + }, + { + "epoch": 2.908281389136242, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8019070625305176, + "num_tokens": 872273048.0, + "step": 22862 + }, + { + "epoch": 2.908408599414833, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.8095358610153198, + "num_tokens": 872312165.0, + "step": 22863 + }, + { + "epoch": 2.908535809693423, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8063325881958008, + "num_tokens": 872352740.0, + "step": 22864 + }, + { + "epoch": 2.9086630199720136, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7900140285491943, + "num_tokens": 872397800.0, + "step": 22865 + }, + { + "epoch": 2.908790230250604, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8140853047370911, + "num_tokens": 872429307.0, + "step": 22866 + }, + { + "epoch": 2.9089174405291947, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7929555177688599, + "num_tokens": 872465897.0, + "step": 22867 + }, + { + "epoch": 2.909044650807785, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.790793776512146, + "num_tokens": 872503252.0, + "step": 22868 + }, + { + "epoch": 2.9091718610863757, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.806662380695343, + "num_tokens": 872546276.0, + "step": 22869 + }, + { + "epoch": 2.9092990713649662, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.8046596050262451, + "num_tokens": 872585690.0, + "step": 22870 + }, + { + "epoch": 2.9094262816435568, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7909932136535645, + "num_tokens": 872621718.0, + "step": 22871 + }, + { + "epoch": 2.9095534919221473, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.8046460151672363, + "num_tokens": 872659349.0, + "step": 22872 + }, + { + "epoch": 2.909680702200738, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.8025859594345093, + "num_tokens": 872704468.0, + "step": 22873 + }, + { + "epoch": 2.9098079124793284, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7999171018600464, + "num_tokens": 872740003.0, + "step": 22874 + }, + { + "epoch": 2.909935122757919, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.78010094165802, + "num_tokens": 872780362.0, + "step": 22875 + }, + { + "epoch": 2.9100623330365094, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.809107780456543, + "num_tokens": 872823719.0, + "step": 22876 + }, + { + "epoch": 2.9101895433151, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.8060096502304077, + "num_tokens": 872856276.0, + "step": 22877 + }, + { + "epoch": 2.9103167535936905, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7912418842315674, + "num_tokens": 872893688.0, + "step": 22878 + }, + { + "epoch": 2.910443963872281, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.5823, + "mean_token_accuracy": 0.8235032558441162, + "num_tokens": 872931169.0, + "step": 22879 + }, + { + "epoch": 2.9105711741508715, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7936897277832031, + "num_tokens": 872970836.0, + "step": 22880 + }, + { + "epoch": 2.910698384429462, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.8201559782028198, + "num_tokens": 873010879.0, + "step": 22881 + }, + { + "epoch": 2.9108255947080526, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.8064689636230469, + "num_tokens": 873058018.0, + "step": 22882 + }, + { + "epoch": 2.9109528049866427, + "grad_norm": 40.75, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7921078205108643, + "num_tokens": 873096259.0, + "step": 22883 + }, + { + "epoch": 2.9110800152652336, + "grad_norm": 34.0, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7887054681777954, + "num_tokens": 873131955.0, + "step": 22884 + }, + { + "epoch": 2.9112072255438237, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7974526286125183, + "num_tokens": 873171024.0, + "step": 22885 + }, + { + "epoch": 2.9113344358224147, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8076034188270569, + "num_tokens": 873210568.0, + "step": 22886 + }, + { + "epoch": 2.9114616461010048, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8085662722587585, + "num_tokens": 873250025.0, + "step": 22887 + }, + { + "epoch": 2.9115888563795957, + "grad_norm": 13.125, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7940795421600342, + "num_tokens": 873285284.0, + "step": 22888 + }, + { + "epoch": 2.911716066658186, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.8040602207183838, + "num_tokens": 873320102.0, + "step": 22889 + }, + { + "epoch": 2.9118432769367764, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8043718338012695, + "num_tokens": 873354668.0, + "step": 22890 + }, + { + "epoch": 2.911970487215367, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7951376438140869, + "num_tokens": 873387415.0, + "step": 22891 + }, + { + "epoch": 2.9120976974939574, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.8235844969749451, + "num_tokens": 873428679.0, + "step": 22892 + }, + { + "epoch": 2.912224907772548, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8010117411613464, + "num_tokens": 873466973.0, + "step": 22893 + }, + { + "epoch": 2.9123521180511385, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.8188822269439697, + "num_tokens": 873510957.0, + "step": 22894 + }, + { + "epoch": 2.912479328329729, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7761207818984985, + "num_tokens": 873546456.0, + "step": 22895 + }, + { + "epoch": 2.9126065386083195, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.786811113357544, + "num_tokens": 873591972.0, + "step": 22896 + }, + { + "epoch": 2.91273374888691, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7972811460494995, + "num_tokens": 873629790.0, + "step": 22897 + }, + { + "epoch": 2.9128609591655006, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7977917194366455, + "num_tokens": 873667419.0, + "step": 22898 + }, + { + "epoch": 2.912988169444091, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.8085029125213623, + "num_tokens": 873705834.0, + "step": 22899 + }, + { + "epoch": 2.9131153797226816, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7970634698867798, + "num_tokens": 873748859.0, + "step": 22900 + }, + { + "epoch": 2.913242590001272, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7885943651199341, + "num_tokens": 873788068.0, + "step": 22901 + }, + { + "epoch": 2.9133698002798627, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.8082625865936279, + "num_tokens": 873827820.0, + "step": 22902 + }, + { + "epoch": 2.9134970105584532, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7979154586791992, + "num_tokens": 873866271.0, + "step": 22903 + }, + { + "epoch": 2.9136242208370438, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7937579154968262, + "num_tokens": 873907409.0, + "step": 22904 + }, + { + "epoch": 2.9137514311156343, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.808251142501831, + "num_tokens": 873947940.0, + "step": 22905 + }, + { + "epoch": 2.913878641394225, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7863423824310303, + "num_tokens": 873990062.0, + "step": 22906 + }, + { + "epoch": 2.9140058516728153, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8165181875228882, + "num_tokens": 874026660.0, + "step": 22907 + }, + { + "epoch": 2.9141330619514054, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.8083203434944153, + "num_tokens": 874064935.0, + "step": 22908 + }, + { + "epoch": 2.9142602722299964, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.8057413101196289, + "num_tokens": 874102807.0, + "step": 22909 + }, + { + "epoch": 2.9143874825085865, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8015129566192627, + "num_tokens": 874139688.0, + "step": 22910 + }, + { + "epoch": 2.9145146927871775, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8109142780303955, + "num_tokens": 874184505.0, + "step": 22911 + }, + { + "epoch": 2.9146419030657675, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.8022252321243286, + "num_tokens": 874223210.0, + "step": 22912 + }, + { + "epoch": 2.914769113344358, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7916772961616516, + "num_tokens": 874263939.0, + "step": 22913 + }, + { + "epoch": 2.9148963236229486, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8106772303581238, + "num_tokens": 874303946.0, + "step": 22914 + }, + { + "epoch": 2.915023533901539, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.8045297265052795, + "num_tokens": 874343701.0, + "step": 22915 + }, + { + "epoch": 2.9151507441801296, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7969499826431274, + "num_tokens": 874380425.0, + "step": 22916 + }, + { + "epoch": 2.91527795445872, + "grad_norm": 35.25, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7973884344100952, + "num_tokens": 874417643.0, + "step": 22917 + }, + { + "epoch": 2.9154051647373107, + "grad_norm": 40.5, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7926815748214722, + "num_tokens": 874463939.0, + "step": 22918 + }, + { + "epoch": 2.9155323750159012, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7959797978401184, + "num_tokens": 874503156.0, + "step": 22919 + }, + { + "epoch": 2.9156595852944918, + "grad_norm": 21.375, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7911880016326904, + "num_tokens": 874546754.0, + "step": 22920 + }, + { + "epoch": 2.9157867955730823, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.8079925775527954, + "num_tokens": 874590781.0, + "step": 22921 + }, + { + "epoch": 2.915914005851673, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8158432841300964, + "num_tokens": 874629592.0, + "step": 22922 + }, + { + "epoch": 2.9160412161302633, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.8049926161766052, + "num_tokens": 874667372.0, + "step": 22923 + }, + { + "epoch": 2.916168426408854, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7953615784645081, + "num_tokens": 874704687.0, + "step": 22924 + }, + { + "epoch": 2.9162956366874444, + "grad_norm": 29.875, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7989591360092163, + "num_tokens": 874740471.0, + "step": 22925 + }, + { + "epoch": 2.916422846966035, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.8044254779815674, + "num_tokens": 874780747.0, + "step": 22926 + }, + { + "epoch": 2.9165500572446255, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.8011373281478882, + "num_tokens": 874814204.0, + "step": 22927 + }, + { + "epoch": 2.916677267523216, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.8037574291229248, + "num_tokens": 874851872.0, + "step": 22928 + }, + { + "epoch": 2.9168044778018065, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7822574377059937, + "num_tokens": 874884215.0, + "step": 22929 + }, + { + "epoch": 2.916931688080397, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7967925071716309, + "num_tokens": 874919047.0, + "step": 22930 + }, + { + "epoch": 2.917058898358987, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7949856519699097, + "num_tokens": 874956469.0, + "step": 22931 + }, + { + "epoch": 2.917186108637578, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8117191791534424, + "num_tokens": 874994633.0, + "step": 22932 + }, + { + "epoch": 2.917313318916168, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.817032516002655, + "num_tokens": 875028530.0, + "step": 22933 + }, + { + "epoch": 2.917440529194759, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7860238552093506, + "num_tokens": 875059785.0, + "step": 22934 + }, + { + "epoch": 2.9175677394733492, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7968922853469849, + "num_tokens": 875096057.0, + "step": 22935 + }, + { + "epoch": 2.91769494975194, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.8089591860771179, + "num_tokens": 875141988.0, + "step": 22936 + }, + { + "epoch": 2.9178221600305303, + "grad_norm": 59.5, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8067980408668518, + "num_tokens": 875176973.0, + "step": 22937 + }, + { + "epoch": 2.917949370309121, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.8064611554145813, + "num_tokens": 875209271.0, + "step": 22938 + }, + { + "epoch": 2.9180765805877114, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.8074880838394165, + "num_tokens": 875247460.0, + "step": 22939 + }, + { + "epoch": 2.918203790866302, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.8009161949157715, + "num_tokens": 875286013.0, + "step": 22940 + }, + { + "epoch": 2.9183310011448924, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7971275448799133, + "num_tokens": 875324547.0, + "step": 22941 + }, + { + "epoch": 2.918458211423483, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7946732044219971, + "num_tokens": 875366655.0, + "step": 22942 + }, + { + "epoch": 2.9185854217020735, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7964775562286377, + "num_tokens": 875408571.0, + "step": 22943 + }, + { + "epoch": 2.918712631980664, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.794725775718689, + "num_tokens": 875450730.0, + "step": 22944 + }, + { + "epoch": 2.9188398422592545, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.8070048093795776, + "num_tokens": 875486850.0, + "step": 22945 + }, + { + "epoch": 2.918967052537845, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7976478338241577, + "num_tokens": 875528653.0, + "step": 22946 + }, + { + "epoch": 2.9190942628164356, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.813496470451355, + "num_tokens": 875564155.0, + "step": 22947 + }, + { + "epoch": 2.919221473095026, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.8012267351150513, + "num_tokens": 875598548.0, + "step": 22948 + }, + { + "epoch": 2.9193486833736166, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.8211381435394287, + "num_tokens": 875634587.0, + "step": 22949 + }, + { + "epoch": 2.919475893652207, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7942664623260498, + "num_tokens": 875671956.0, + "step": 22950 + }, + { + "epoch": 2.9196031039307977, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8040802478790283, + "num_tokens": 875706327.0, + "step": 22951 + }, + { + "epoch": 2.9197303142093882, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.8123469948768616, + "num_tokens": 875749320.0, + "step": 22952 + }, + { + "epoch": 2.9198575244879788, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.811622679233551, + "num_tokens": 875783871.0, + "step": 22953 + }, + { + "epoch": 2.9199847347665693, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8024663925170898, + "num_tokens": 875821716.0, + "step": 22954 + }, + { + "epoch": 2.92011194504516, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.8058385848999023, + "num_tokens": 875854339.0, + "step": 22955 + }, + { + "epoch": 2.92023915532375, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8028320074081421, + "num_tokens": 875889456.0, + "step": 22956 + }, + { + "epoch": 2.920366365602341, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7964168787002563, + "num_tokens": 875926659.0, + "step": 22957 + }, + { + "epoch": 2.920493575880931, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8126412630081177, + "num_tokens": 875972766.0, + "step": 22958 + }, + { + "epoch": 2.920620786159522, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7953155636787415, + "num_tokens": 876011824.0, + "step": 22959 + }, + { + "epoch": 2.920747996438112, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7805707454681396, + "num_tokens": 876050353.0, + "step": 22960 + }, + { + "epoch": 2.920875206716703, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7914752960205078, + "num_tokens": 876089358.0, + "step": 22961 + }, + { + "epoch": 2.921002416995293, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7837137579917908, + "num_tokens": 876126287.0, + "step": 22962 + }, + { + "epoch": 2.9211296272738836, + "grad_norm": 12.875, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.8156226873397827, + "num_tokens": 876165157.0, + "step": 22963 + }, + { + "epoch": 2.921256837552474, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.8012232780456543, + "num_tokens": 876201391.0, + "step": 22964 + }, + { + "epoch": 2.9213840478310646, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8076492547988892, + "num_tokens": 876243004.0, + "step": 22965 + }, + { + "epoch": 2.921511258109655, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.8029783964157104, + "num_tokens": 876279245.0, + "step": 22966 + }, + { + "epoch": 2.9216384683882457, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7967716455459595, + "num_tokens": 876312145.0, + "step": 22967 + }, + { + "epoch": 2.9217656786668362, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.8088638186454773, + "num_tokens": 876351354.0, + "step": 22968 + }, + { + "epoch": 2.9218928889454268, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7905448079109192, + "num_tokens": 876388621.0, + "step": 22969 + }, + { + "epoch": 2.9220200992240173, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7852078676223755, + "num_tokens": 876418377.0, + "step": 22970 + }, + { + "epoch": 2.922147309502608, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.8072478175163269, + "num_tokens": 876453351.0, + "step": 22971 + }, + { + "epoch": 2.9222745197811983, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.8068081140518188, + "num_tokens": 876484830.0, + "step": 22972 + }, + { + "epoch": 2.922401730059789, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8142064809799194, + "num_tokens": 876524318.0, + "step": 22973 + }, + { + "epoch": 2.9225289403383794, + "grad_norm": 24.0, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.8004598617553711, + "num_tokens": 876562421.0, + "step": 22974 + }, + { + "epoch": 2.92265615061697, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7984839677810669, + "num_tokens": 876605681.0, + "step": 22975 + }, + { + "epoch": 2.9227833608955605, + "grad_norm": 31.875, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7909601926803589, + "num_tokens": 876647371.0, + "step": 22976 + }, + { + "epoch": 2.922910571174151, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.813483476638794, + "num_tokens": 876685563.0, + "step": 22977 + }, + { + "epoch": 2.9230377814527415, + "grad_norm": 29.75, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7965283393859863, + "num_tokens": 876718404.0, + "step": 22978 + }, + { + "epoch": 2.923164991731332, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8017581105232239, + "num_tokens": 876751888.0, + "step": 22979 + }, + { + "epoch": 2.9232922020099226, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.8076381683349609, + "num_tokens": 876786813.0, + "step": 22980 + }, + { + "epoch": 2.9234194122885127, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.811734676361084, + "num_tokens": 876824511.0, + "step": 22981 + }, + { + "epoch": 2.9235466225671036, + "grad_norm": 27.125, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7975468039512634, + "num_tokens": 876856409.0, + "step": 22982 + }, + { + "epoch": 2.9236738328456937, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7832673192024231, + "num_tokens": 876889204.0, + "step": 22983 + }, + { + "epoch": 2.9238010431242847, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8033488988876343, + "num_tokens": 876927185.0, + "step": 22984 + }, + { + "epoch": 2.9239282534028748, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7982239723205566, + "num_tokens": 876965744.0, + "step": 22985 + }, + { + "epoch": 2.9240554636814657, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8065593242645264, + "num_tokens": 876997107.0, + "step": 22986 + }, + { + "epoch": 2.924182673960056, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7925596237182617, + "num_tokens": 877037894.0, + "step": 22987 + }, + { + "epoch": 2.9243098842386464, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7847902774810791, + "num_tokens": 877078724.0, + "step": 22988 + }, + { + "epoch": 2.924437094517237, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.800683856010437, + "num_tokens": 877119685.0, + "step": 22989 + }, + { + "epoch": 2.9245643047958274, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8075515031814575, + "num_tokens": 877157055.0, + "step": 22990 + }, + { + "epoch": 2.924691515074418, + "grad_norm": 14.0, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8061195611953735, + "num_tokens": 877188569.0, + "step": 22991 + }, + { + "epoch": 2.9248187253530085, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7953207492828369, + "num_tokens": 877222745.0, + "step": 22992 + }, + { + "epoch": 2.924945935631599, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7818593978881836, + "num_tokens": 877262510.0, + "step": 22993 + }, + { + "epoch": 2.9250731459101895, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7950320243835449, + "num_tokens": 877300164.0, + "step": 22994 + }, + { + "epoch": 2.92520035618878, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.8105043172836304, + "num_tokens": 877337187.0, + "step": 22995 + }, + { + "epoch": 2.9253275664673706, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7794530987739563, + "num_tokens": 877375902.0, + "step": 22996 + }, + { + "epoch": 2.925454776745961, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8016092777252197, + "num_tokens": 877414406.0, + "step": 22997 + }, + { + "epoch": 2.9255819870245516, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8028430938720703, + "num_tokens": 877448977.0, + "step": 22998 + }, + { + "epoch": 2.925709197303142, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.806354284286499, + "num_tokens": 877486766.0, + "step": 22999 + }, + { + "epoch": 2.9258364075817327, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8183545470237732, + "num_tokens": 877519320.0, + "step": 23000 + }, + { + "epoch": 2.925963617860323, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7767484188079834, + "num_tokens": 877557913.0, + "step": 23001 + }, + { + "epoch": 2.9260908281389137, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7964966297149658, + "num_tokens": 877591197.0, + "step": 23002 + }, + { + "epoch": 2.9262180384175043, + "grad_norm": 12.9375, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.8186500072479248, + "num_tokens": 877625018.0, + "step": 23003 + }, + { + "epoch": 2.926345248696095, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8065026998519897, + "num_tokens": 877662433.0, + "step": 23004 + }, + { + "epoch": 2.9264724589746853, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8145877718925476, + "num_tokens": 877703576.0, + "step": 23005 + }, + { + "epoch": 2.9265996692532754, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7969796657562256, + "num_tokens": 877734025.0, + "step": 23006 + }, + { + "epoch": 2.9267268795318664, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8079020380973816, + "num_tokens": 877771942.0, + "step": 23007 + }, + { + "epoch": 2.9268540898104565, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.8038276433944702, + "num_tokens": 877815662.0, + "step": 23008 + }, + { + "epoch": 2.9269813000890474, + "grad_norm": 40.0, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7852452397346497, + "num_tokens": 877855398.0, + "step": 23009 + }, + { + "epoch": 2.9271085103676375, + "grad_norm": 60.25, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7976092100143433, + "num_tokens": 877898544.0, + "step": 23010 + }, + { + "epoch": 2.927235720646228, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7997279167175293, + "num_tokens": 877937914.0, + "step": 23011 + }, + { + "epoch": 2.9273629309248186, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7876465320587158, + "num_tokens": 877979266.0, + "step": 23012 + }, + { + "epoch": 2.927490141203409, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.805915117263794, + "num_tokens": 878019762.0, + "step": 23013 + }, + { + "epoch": 2.9276173514819996, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7980467081069946, + "num_tokens": 878062174.0, + "step": 23014 + }, + { + "epoch": 2.92774456176059, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7966249585151672, + "num_tokens": 878098197.0, + "step": 23015 + }, + { + "epoch": 2.9278717720391807, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.8053370118141174, + "num_tokens": 878130478.0, + "step": 23016 + }, + { + "epoch": 2.9279989823177712, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7872265577316284, + "num_tokens": 878175670.0, + "step": 23017 + }, + { + "epoch": 2.9281261925963618, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7962031960487366, + "num_tokens": 878216415.0, + "step": 23018 + }, + { + "epoch": 2.9282534028749523, + "grad_norm": 13.3125, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.797897219657898, + "num_tokens": 878249582.0, + "step": 23019 + }, + { + "epoch": 2.928380613153543, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8141369223594666, + "num_tokens": 878290859.0, + "step": 23020 + }, + { + "epoch": 2.9285078234321333, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.815058171749115, + "num_tokens": 878322463.0, + "step": 23021 + }, + { + "epoch": 2.928635033710724, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8073961734771729, + "num_tokens": 878365754.0, + "step": 23022 + }, + { + "epoch": 2.9287622439893144, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.8191579580307007, + "num_tokens": 878410325.0, + "step": 23023 + }, + { + "epoch": 2.928889454267905, + "grad_norm": 14.8125, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8107579946517944, + "num_tokens": 878451604.0, + "step": 23024 + }, + { + "epoch": 2.9290166645464955, + "grad_norm": 15.3125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.803368330001831, + "num_tokens": 878485174.0, + "step": 23025 + }, + { + "epoch": 2.929143874825086, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7938079833984375, + "num_tokens": 878525867.0, + "step": 23026 + }, + { + "epoch": 2.9292710851036765, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7908734083175659, + "num_tokens": 878561327.0, + "step": 23027 + }, + { + "epoch": 2.929398295382267, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.8108709454536438, + "num_tokens": 878603045.0, + "step": 23028 + }, + { + "epoch": 2.929525505660857, + "grad_norm": 10.5625, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.806501567363739, + "num_tokens": 878645051.0, + "step": 23029 + }, + { + "epoch": 2.929652715939448, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7808626890182495, + "num_tokens": 878682003.0, + "step": 23030 + }, + { + "epoch": 2.929779926218038, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.8070822358131409, + "num_tokens": 878720509.0, + "step": 23031 + }, + { + "epoch": 2.929907136496629, + "grad_norm": 13.625, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.8116061091423035, + "num_tokens": 878757950.0, + "step": 23032 + }, + { + "epoch": 2.9300343467752192, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7987607717514038, + "num_tokens": 878801477.0, + "step": 23033 + }, + { + "epoch": 2.93016155705381, + "grad_norm": 29.125, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7939720749855042, + "num_tokens": 878838802.0, + "step": 23034 + }, + { + "epoch": 2.9302887673324003, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7901304960250854, + "num_tokens": 878875566.0, + "step": 23035 + }, + { + "epoch": 2.930415977610991, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7930940389633179, + "num_tokens": 878913463.0, + "step": 23036 + }, + { + "epoch": 2.9305431878895813, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7873064279556274, + "num_tokens": 878950769.0, + "step": 23037 + }, + { + "epoch": 2.930670398168172, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8117424249649048, + "num_tokens": 878989041.0, + "step": 23038 + }, + { + "epoch": 2.9307976084467624, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.8088390827178955, + "num_tokens": 879020195.0, + "step": 23039 + }, + { + "epoch": 2.930924818725353, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8036304712295532, + "num_tokens": 879057818.0, + "step": 23040 + }, + { + "epoch": 2.9310520290039435, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7961411476135254, + "num_tokens": 879094747.0, + "step": 23041 + }, + { + "epoch": 2.931179239282534, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7950044870376587, + "num_tokens": 879131127.0, + "step": 23042 + }, + { + "epoch": 2.9313064495611245, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.8006436228752136, + "num_tokens": 879168541.0, + "step": 23043 + }, + { + "epoch": 2.931433659839715, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7908191680908203, + "num_tokens": 879199409.0, + "step": 23044 + }, + { + "epoch": 2.9315608701183056, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8174896836280823, + "num_tokens": 879236447.0, + "step": 23045 + }, + { + "epoch": 2.931688080396896, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8019263744354248, + "num_tokens": 879275596.0, + "step": 23046 + }, + { + "epoch": 2.9318152906754866, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.8048590421676636, + "num_tokens": 879315620.0, + "step": 23047 + }, + { + "epoch": 2.931942500954077, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.8031987547874451, + "num_tokens": 879356220.0, + "step": 23048 + }, + { + "epoch": 2.9320697112326677, + "grad_norm": 48.75, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7781708240509033, + "num_tokens": 879397544.0, + "step": 23049 + }, + { + "epoch": 2.932196921511258, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.8139548897743225, + "num_tokens": 879438165.0, + "step": 23050 + }, + { + "epoch": 2.9323241317898487, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.8030869960784912, + "num_tokens": 879473601.0, + "step": 23051 + }, + { + "epoch": 2.9324513420684393, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7971503734588623, + "num_tokens": 879510998.0, + "step": 23052 + }, + { + "epoch": 2.93257855234703, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.8057799339294434, + "num_tokens": 879551469.0, + "step": 23053 + }, + { + "epoch": 2.93270576262562, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7963147163391113, + "num_tokens": 879585620.0, + "step": 23054 + }, + { + "epoch": 2.932832972904211, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8087881207466125, + "num_tokens": 879621749.0, + "step": 23055 + }, + { + "epoch": 2.932960183182801, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.8020262718200684, + "num_tokens": 879664529.0, + "step": 23056 + }, + { + "epoch": 2.933087393461392, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.8049665689468384, + "num_tokens": 879696293.0, + "step": 23057 + }, + { + "epoch": 2.933214603739982, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8047235012054443, + "num_tokens": 879733562.0, + "step": 23058 + }, + { + "epoch": 2.933341814018573, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8010914325714111, + "num_tokens": 879771740.0, + "step": 23059 + }, + { + "epoch": 2.933469024297163, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7931596636772156, + "num_tokens": 879805578.0, + "step": 23060 + }, + { + "epoch": 2.9335962345757536, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.8029889464378357, + "num_tokens": 879845175.0, + "step": 23061 + }, + { + "epoch": 2.933723444854344, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.811646580696106, + "num_tokens": 879878975.0, + "step": 23062 + }, + { + "epoch": 2.9338506551329346, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7914825081825256, + "num_tokens": 879919373.0, + "step": 23063 + }, + { + "epoch": 2.933977865411525, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7755391597747803, + "num_tokens": 879953821.0, + "step": 23064 + }, + { + "epoch": 2.9341050756901157, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7918210029602051, + "num_tokens": 879993505.0, + "step": 23065 + }, + { + "epoch": 2.934232285968706, + "grad_norm": 36.0, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8069143891334534, + "num_tokens": 880030124.0, + "step": 23066 + }, + { + "epoch": 2.9343594962472968, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7718808650970459, + "num_tokens": 880059480.0, + "step": 23067 + }, + { + "epoch": 2.9344867065258873, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8018158674240112, + "num_tokens": 880093402.0, + "step": 23068 + }, + { + "epoch": 2.934613916804478, + "grad_norm": 62.0, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7982674837112427, + "num_tokens": 880127874.0, + "step": 23069 + }, + { + "epoch": 2.9347411270830683, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8101376295089722, + "num_tokens": 880165755.0, + "step": 23070 + }, + { + "epoch": 2.934868337361659, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7925084829330444, + "num_tokens": 880203931.0, + "step": 23071 + }, + { + "epoch": 2.9349955476402494, + "grad_norm": 49.0, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8052234053611755, + "num_tokens": 880242844.0, + "step": 23072 + }, + { + "epoch": 2.93512275791884, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7873482704162598, + "num_tokens": 880280068.0, + "step": 23073 + }, + { + "epoch": 2.9352499681974304, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.8057990074157715, + "num_tokens": 880314044.0, + "step": 23074 + }, + { + "epoch": 2.935377178476021, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8012365102767944, + "num_tokens": 880355077.0, + "step": 23075 + }, + { + "epoch": 2.9355043887546115, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.81230628490448, + "num_tokens": 880388906.0, + "step": 23076 + }, + { + "epoch": 2.935631599033202, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8124055862426758, + "num_tokens": 880426834.0, + "step": 23077 + }, + { + "epoch": 2.9357588093117926, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.8006258010864258, + "num_tokens": 880463477.0, + "step": 23078 + }, + { + "epoch": 2.9358860195903826, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7898269891738892, + "num_tokens": 880500884.0, + "step": 23079 + }, + { + "epoch": 2.9360132298689736, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7970679998397827, + "num_tokens": 880539724.0, + "step": 23080 + }, + { + "epoch": 2.9361404401475637, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.799150824546814, + "num_tokens": 880577820.0, + "step": 23081 + }, + { + "epoch": 2.9362676504261547, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.8058651685714722, + "num_tokens": 880612055.0, + "step": 23082 + }, + { + "epoch": 2.9363948607047448, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8119459748268127, + "num_tokens": 880654907.0, + "step": 23083 + }, + { + "epoch": 2.9365220709833357, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8165231347084045, + "num_tokens": 880692418.0, + "step": 23084 + }, + { + "epoch": 2.936649281261926, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7840068340301514, + "num_tokens": 880727742.0, + "step": 23085 + }, + { + "epoch": 2.9367764915405163, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8142136335372925, + "num_tokens": 880766673.0, + "step": 23086 + }, + { + "epoch": 2.936903701819107, + "grad_norm": 27.875, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7965144515037537, + "num_tokens": 880810390.0, + "step": 23087 + }, + { + "epoch": 2.9370309120976974, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7978674173355103, + "num_tokens": 880846920.0, + "step": 23088 + }, + { + "epoch": 2.937158122376288, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.8104146718978882, + "num_tokens": 880885867.0, + "step": 23089 + }, + { + "epoch": 2.9372853326548785, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.8146629333496094, + "num_tokens": 880928408.0, + "step": 23090 + }, + { + "epoch": 2.937412542933469, + "grad_norm": 30.875, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7984791994094849, + "num_tokens": 880960681.0, + "step": 23091 + }, + { + "epoch": 2.9375397532120595, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7937601804733276, + "num_tokens": 881000358.0, + "step": 23092 + }, + { + "epoch": 2.93766696349065, + "grad_norm": 22.0, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.8022508025169373, + "num_tokens": 881037765.0, + "step": 23093 + }, + { + "epoch": 2.9377941737692406, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8129248023033142, + "num_tokens": 881076091.0, + "step": 23094 + }, + { + "epoch": 2.937921384047831, + "grad_norm": 15.6875, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7962692975997925, + "num_tokens": 881112784.0, + "step": 23095 + }, + { + "epoch": 2.9380485943264216, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.8086918592453003, + "num_tokens": 881149938.0, + "step": 23096 + }, + { + "epoch": 2.938175804605012, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7904316186904907, + "num_tokens": 881187541.0, + "step": 23097 + }, + { + "epoch": 2.9383030148836027, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.8009850978851318, + "num_tokens": 881221920.0, + "step": 23098 + }, + { + "epoch": 2.938430225162193, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7902532815933228, + "num_tokens": 881264804.0, + "step": 23099 + }, + { + "epoch": 2.9385574354407837, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7911988496780396, + "num_tokens": 881304482.0, + "step": 23100 + }, + { + "epoch": 2.9386846457193743, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7946125268936157, + "num_tokens": 881340822.0, + "step": 23101 + }, + { + "epoch": 2.938811855997965, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.797195315361023, + "num_tokens": 881374352.0, + "step": 23102 + }, + { + "epoch": 2.9389390662765553, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7981259226799011, + "num_tokens": 881413176.0, + "step": 23103 + }, + { + "epoch": 2.9390662765551454, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7956920862197876, + "num_tokens": 881454309.0, + "step": 23104 + }, + { + "epoch": 2.9391934868337364, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.8006730079650879, + "num_tokens": 881490702.0, + "step": 23105 + }, + { + "epoch": 2.9393206971123265, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.808749258518219, + "num_tokens": 881530394.0, + "step": 23106 + }, + { + "epoch": 2.9394479073909174, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8167356252670288, + "num_tokens": 881567286.0, + "step": 23107 + }, + { + "epoch": 2.9395751176695075, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7946794629096985, + "num_tokens": 881607017.0, + "step": 23108 + }, + { + "epoch": 2.939702327948098, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7995226383209229, + "num_tokens": 881645793.0, + "step": 23109 + }, + { + "epoch": 2.9398295382266886, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.8041667938232422, + "num_tokens": 881683596.0, + "step": 23110 + }, + { + "epoch": 2.939956748505279, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7850426435470581, + "num_tokens": 881717607.0, + "step": 23111 + }, + { + "epoch": 2.9400839587838696, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8172122240066528, + "num_tokens": 881750884.0, + "step": 23112 + }, + { + "epoch": 2.94021116906246, + "grad_norm": 44.0, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7962543368339539, + "num_tokens": 881789345.0, + "step": 23113 + }, + { + "epoch": 2.9403383793410507, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7962313294410706, + "num_tokens": 881828459.0, + "step": 23114 + }, + { + "epoch": 2.940465589619641, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7953915596008301, + "num_tokens": 881865294.0, + "step": 23115 + }, + { + "epoch": 2.9405927998982317, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.811049222946167, + "num_tokens": 881904936.0, + "step": 23116 + }, + { + "epoch": 2.9407200101768223, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.8062622547149658, + "num_tokens": 881936300.0, + "step": 23117 + }, + { + "epoch": 2.940847220455413, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.81191086769104, + "num_tokens": 881974092.0, + "step": 23118 + }, + { + "epoch": 2.9409744307340033, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7978255748748779, + "num_tokens": 882019339.0, + "step": 23119 + }, + { + "epoch": 2.941101641012594, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.8010616302490234, + "num_tokens": 882052837.0, + "step": 23120 + }, + { + "epoch": 2.9412288512911844, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8021558523178101, + "num_tokens": 882088449.0, + "step": 23121 + }, + { + "epoch": 2.941356061569775, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7849459052085876, + "num_tokens": 882124844.0, + "step": 23122 + }, + { + "epoch": 2.9414832718483654, + "grad_norm": 22.625, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7974640130996704, + "num_tokens": 882162246.0, + "step": 23123 + }, + { + "epoch": 2.941610482126956, + "grad_norm": 46.5, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7918821573257446, + "num_tokens": 882202825.0, + "step": 23124 + }, + { + "epoch": 2.9417376924055465, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7834243774414062, + "num_tokens": 882237951.0, + "step": 23125 + }, + { + "epoch": 2.941864902684137, + "grad_norm": 25.125, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7979822754859924, + "num_tokens": 882279147.0, + "step": 23126 + }, + { + "epoch": 2.941992112962727, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8279780745506287, + "num_tokens": 882322777.0, + "step": 23127 + }, + { + "epoch": 2.942119323241318, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.80633944272995, + "num_tokens": 882360681.0, + "step": 23128 + }, + { + "epoch": 2.942246533519908, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7814316749572754, + "num_tokens": 882403823.0, + "step": 23129 + }, + { + "epoch": 2.942373743798499, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.8079693913459778, + "num_tokens": 882443723.0, + "step": 23130 + }, + { + "epoch": 2.9425009540770892, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.785388708114624, + "num_tokens": 882480610.0, + "step": 23131 + }, + { + "epoch": 2.94262816435568, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.8008286952972412, + "num_tokens": 882518789.0, + "step": 23132 + }, + { + "epoch": 2.9427553746342703, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.790069043636322, + "num_tokens": 882556598.0, + "step": 23133 + }, + { + "epoch": 2.942882584912861, + "grad_norm": 14.5, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7940865755081177, + "num_tokens": 882593788.0, + "step": 23134 + }, + { + "epoch": 2.9430097951914513, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.8042817711830139, + "num_tokens": 882630861.0, + "step": 23135 + }, + { + "epoch": 2.943137005470042, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.8108604550361633, + "num_tokens": 882670950.0, + "step": 23136 + }, + { + "epoch": 2.9432642157486324, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7812391519546509, + "num_tokens": 882705680.0, + "step": 23137 + }, + { + "epoch": 2.943391426027223, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.8080174922943115, + "num_tokens": 882737589.0, + "step": 23138 + }, + { + "epoch": 2.9435186363058135, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7827950716018677, + "num_tokens": 882774502.0, + "step": 23139 + }, + { + "epoch": 2.943645846584404, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7892061471939087, + "num_tokens": 882815184.0, + "step": 23140 + }, + { + "epoch": 2.9437730568629945, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8147576451301575, + "num_tokens": 882856793.0, + "step": 23141 + }, + { + "epoch": 2.943900267141585, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8155529499053955, + "num_tokens": 882896128.0, + "step": 23142 + }, + { + "epoch": 2.9440274774201756, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8036885261535645, + "num_tokens": 882931253.0, + "step": 23143 + }, + { + "epoch": 2.944154687698766, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8242949843406677, + "num_tokens": 882975072.0, + "step": 23144 + }, + { + "epoch": 2.9442818979773566, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8145319223403931, + "num_tokens": 883013576.0, + "step": 23145 + }, + { + "epoch": 2.944409108255947, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.8003218173980713, + "num_tokens": 883046052.0, + "step": 23146 + }, + { + "epoch": 2.9445363185345377, + "grad_norm": 49.5, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.77772456407547, + "num_tokens": 883080398.0, + "step": 23147 + }, + { + "epoch": 2.944663528813128, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7899466753005981, + "num_tokens": 883120321.0, + "step": 23148 + }, + { + "epoch": 2.9447907390917187, + "grad_norm": 13.1875, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8065671324729919, + "num_tokens": 883161285.0, + "step": 23149 + }, + { + "epoch": 2.9449179493703093, + "grad_norm": 23.125, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.8047022819519043, + "num_tokens": 883199857.0, + "step": 23150 + }, + { + "epoch": 2.9450451596489, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.8070629835128784, + "num_tokens": 883238325.0, + "step": 23151 + }, + { + "epoch": 2.94517236992749, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7967127561569214, + "num_tokens": 883281814.0, + "step": 23152 + }, + { + "epoch": 2.945299580206081, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.804314136505127, + "num_tokens": 883317385.0, + "step": 23153 + }, + { + "epoch": 2.945426790484671, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.8103424906730652, + "num_tokens": 883353498.0, + "step": 23154 + }, + { + "epoch": 2.945554000763262, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.8094850778579712, + "num_tokens": 883391709.0, + "step": 23155 + }, + { + "epoch": 2.945681211041852, + "grad_norm": 43.0, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7839800119400024, + "num_tokens": 883426329.0, + "step": 23156 + }, + { + "epoch": 2.945808421320443, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7959325313568115, + "num_tokens": 883465797.0, + "step": 23157 + }, + { + "epoch": 2.945935631599033, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8047851324081421, + "num_tokens": 883501632.0, + "step": 23158 + }, + { + "epoch": 2.9460628418776236, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8137723207473755, + "num_tokens": 883537622.0, + "step": 23159 + }, + { + "epoch": 2.946190052156214, + "grad_norm": 13.25, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7987697124481201, + "num_tokens": 883576458.0, + "step": 23160 + }, + { + "epoch": 2.9463172624348046, + "grad_norm": 15.625, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7964350581169128, + "num_tokens": 883620929.0, + "step": 23161 + }, + { + "epoch": 2.946444472713395, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7712939381599426, + "num_tokens": 883652589.0, + "step": 23162 + }, + { + "epoch": 2.9465716829919857, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.8054327964782715, + "num_tokens": 883691844.0, + "step": 23163 + }, + { + "epoch": 2.946698893270576, + "grad_norm": 32.25, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7780166864395142, + "num_tokens": 883732838.0, + "step": 23164 + }, + { + "epoch": 2.9468261035491667, + "grad_norm": 38.25, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.8003373146057129, + "num_tokens": 883771415.0, + "step": 23165 + }, + { + "epoch": 2.9469533138277573, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.782812237739563, + "num_tokens": 883808807.0, + "step": 23166 + }, + { + "epoch": 2.947080524106348, + "grad_norm": 13.375, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8113735318183899, + "num_tokens": 883844300.0, + "step": 23167 + }, + { + "epoch": 2.9472077343849383, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.8171100616455078, + "num_tokens": 883879337.0, + "step": 23168 + }, + { + "epoch": 2.947334944663529, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7966181039810181, + "num_tokens": 883925810.0, + "step": 23169 + }, + { + "epoch": 2.9474621549421194, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8131150007247925, + "num_tokens": 883964753.0, + "step": 23170 + }, + { + "epoch": 2.94758936522071, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7916852235794067, + "num_tokens": 884005737.0, + "step": 23171 + }, + { + "epoch": 2.9477165754993004, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.8014581799507141, + "num_tokens": 884043414.0, + "step": 23172 + }, + { + "epoch": 2.947843785777891, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.8055391311645508, + "num_tokens": 884077399.0, + "step": 23173 + }, + { + "epoch": 2.9479709960564815, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.8013872504234314, + "num_tokens": 884116539.0, + "step": 23174 + }, + { + "epoch": 2.948098206335072, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.8050318956375122, + "num_tokens": 884149756.0, + "step": 23175 + }, + { + "epoch": 2.9482254166136626, + "grad_norm": 23.625, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.8182694911956787, + "num_tokens": 884181960.0, + "step": 23176 + }, + { + "epoch": 2.9483526268922526, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7950772047042847, + "num_tokens": 884223562.0, + "step": 23177 + }, + { + "epoch": 2.9484798371708436, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.8074225783348083, + "num_tokens": 884264541.0, + "step": 23178 + }, + { + "epoch": 2.9486070474494337, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7875145673751831, + "num_tokens": 884301171.0, + "step": 23179 + }, + { + "epoch": 2.9487342577280247, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7938299775123596, + "num_tokens": 884332481.0, + "step": 23180 + }, + { + "epoch": 2.9488614680066147, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8204469680786133, + "num_tokens": 884371616.0, + "step": 23181 + }, + { + "epoch": 2.9489886782852053, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.78850257396698, + "num_tokens": 884411245.0, + "step": 23182 + }, + { + "epoch": 2.949115888563796, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7957403659820557, + "num_tokens": 884445712.0, + "step": 23183 + }, + { + "epoch": 2.9492430988423863, + "grad_norm": 42.75, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.8110421895980835, + "num_tokens": 884484935.0, + "step": 23184 + }, + { + "epoch": 2.949370309120977, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.797105073928833, + "num_tokens": 884518714.0, + "step": 23185 + }, + { + "epoch": 2.9494975193995674, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7976318001747131, + "num_tokens": 884560430.0, + "step": 23186 + }, + { + "epoch": 2.949624729678158, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7909031510353088, + "num_tokens": 884595937.0, + "step": 23187 + }, + { + "epoch": 2.9497519399567484, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7926216125488281, + "num_tokens": 884633201.0, + "step": 23188 + }, + { + "epoch": 2.949879150235339, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8080933094024658, + "num_tokens": 884672602.0, + "step": 23189 + }, + { + "epoch": 2.9500063605139295, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8050609827041626, + "num_tokens": 884708378.0, + "step": 23190 + }, + { + "epoch": 2.95013357079252, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.8035205006599426, + "num_tokens": 884747527.0, + "step": 23191 + }, + { + "epoch": 2.9502607810711106, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.8215807676315308, + "num_tokens": 884781738.0, + "step": 23192 + }, + { + "epoch": 2.950387991349701, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7879195213317871, + "num_tokens": 884818884.0, + "step": 23193 + }, + { + "epoch": 2.9505152016282916, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8128794431686401, + "num_tokens": 884851751.0, + "step": 23194 + }, + { + "epoch": 2.950642411906882, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.8065844774246216, + "num_tokens": 884892935.0, + "step": 23195 + }, + { + "epoch": 2.9507696221854727, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.810710072517395, + "num_tokens": 884935820.0, + "step": 23196 + }, + { + "epoch": 2.950896832464063, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7985706925392151, + "num_tokens": 884976042.0, + "step": 23197 + }, + { + "epoch": 2.9510240427426537, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.5852, + "mean_token_accuracy": 0.82353675365448, + "num_tokens": 885013511.0, + "step": 23198 + }, + { + "epoch": 2.9511512530212443, + "grad_norm": 28.5, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.785915732383728, + "num_tokens": 885047291.0, + "step": 23199 + }, + { + "epoch": 2.951278463299835, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.792931079864502, + "num_tokens": 885083116.0, + "step": 23200 + }, + { + "epoch": 2.9514056735784253, + "grad_norm": 51.5, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7993252873420715, + "num_tokens": 885123282.0, + "step": 23201 + }, + { + "epoch": 2.9515328838570154, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8075209856033325, + "num_tokens": 885166277.0, + "step": 23202 + }, + { + "epoch": 2.9516600941356064, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.8216453790664673, + "num_tokens": 885206414.0, + "step": 23203 + }, + { + "epoch": 2.9517873044141965, + "grad_norm": 30.625, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.806848406791687, + "num_tokens": 885240977.0, + "step": 23204 + }, + { + "epoch": 2.9519145146927874, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8148137331008911, + "num_tokens": 885283190.0, + "step": 23205 + }, + { + "epoch": 2.9520417249713775, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8136885166168213, + "num_tokens": 885321340.0, + "step": 23206 + }, + { + "epoch": 2.952168935249968, + "grad_norm": 15.375, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8130643367767334, + "num_tokens": 885357524.0, + "step": 23207 + }, + { + "epoch": 2.9522961455285586, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7686538100242615, + "num_tokens": 885389696.0, + "step": 23208 + }, + { + "epoch": 2.952423355807149, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.8216140866279602, + "num_tokens": 885429574.0, + "step": 23209 + }, + { + "epoch": 2.9525505660857396, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8005738258361816, + "num_tokens": 885459961.0, + "step": 23210 + }, + { + "epoch": 2.95267777636433, + "grad_norm": 14.875, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7936301827430725, + "num_tokens": 885506725.0, + "step": 23211 + }, + { + "epoch": 2.9528049866429207, + "grad_norm": 44.25, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.8051149845123291, + "num_tokens": 885549409.0, + "step": 23212 + }, + { + "epoch": 2.952932196921511, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.809239387512207, + "num_tokens": 885589272.0, + "step": 23213 + }, + { + "epoch": 2.9530594072001017, + "grad_norm": 21.0, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7855781316757202, + "num_tokens": 885625951.0, + "step": 23214 + }, + { + "epoch": 2.9531866174786923, + "grad_norm": 15.25, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.804751992225647, + "num_tokens": 885665844.0, + "step": 23215 + }, + { + "epoch": 2.953313827757283, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.800888180732727, + "num_tokens": 885705616.0, + "step": 23216 + }, + { + "epoch": 2.9534410380358733, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.8173177242279053, + "num_tokens": 885752362.0, + "step": 23217 + }, + { + "epoch": 2.953568248314464, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7938112020492554, + "num_tokens": 885792485.0, + "step": 23218 + }, + { + "epoch": 2.9536954585930544, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.8046296834945679, + "num_tokens": 885834809.0, + "step": 23219 + }, + { + "epoch": 2.953822668871645, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.8049890995025635, + "num_tokens": 885870555.0, + "step": 23220 + }, + { + "epoch": 2.9539498791502354, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.8095118403434753, + "num_tokens": 885911911.0, + "step": 23221 + }, + { + "epoch": 2.954077089428826, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.8122580051422119, + "num_tokens": 885941094.0, + "step": 23222 + }, + { + "epoch": 2.9542042997074165, + "grad_norm": 29.0, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7995584607124329, + "num_tokens": 885978094.0, + "step": 23223 + }, + { + "epoch": 2.954331509986007, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8179289698600769, + "num_tokens": 886014859.0, + "step": 23224 + }, + { + "epoch": 2.954458720264597, + "grad_norm": 31.125, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.8103548288345337, + "num_tokens": 886048317.0, + "step": 23225 + }, + { + "epoch": 2.954585930543188, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.8002749681472778, + "num_tokens": 886081797.0, + "step": 23226 + }, + { + "epoch": 2.954713140821778, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7898568511009216, + "num_tokens": 886124111.0, + "step": 23227 + }, + { + "epoch": 2.954840351100369, + "grad_norm": 12.125, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8155621290206909, + "num_tokens": 886162823.0, + "step": 23228 + }, + { + "epoch": 2.954967561378959, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8012133836746216, + "num_tokens": 886204706.0, + "step": 23229 + }, + { + "epoch": 2.95509477165755, + "grad_norm": 12.375, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8136669993400574, + "num_tokens": 886241461.0, + "step": 23230 + }, + { + "epoch": 2.9552219819361403, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.794545590877533, + "num_tokens": 886279276.0, + "step": 23231 + }, + { + "epoch": 2.955349192214731, + "grad_norm": 19.625, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.8068299889564514, + "num_tokens": 886317703.0, + "step": 23232 + }, + { + "epoch": 2.9554764024933213, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.8074444532394409, + "num_tokens": 886355235.0, + "step": 23233 + }, + { + "epoch": 2.955603612771912, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.5561, + "mean_token_accuracy": 0.8288087248802185, + "num_tokens": 886387354.0, + "step": 23234 + }, + { + "epoch": 2.9557308230505024, + "grad_norm": 21.125, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.798427939414978, + "num_tokens": 886419742.0, + "step": 23235 + }, + { + "epoch": 2.955858033329093, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7993305325508118, + "num_tokens": 886455218.0, + "step": 23236 + }, + { + "epoch": 2.9559852436076834, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.8041418194770813, + "num_tokens": 886494454.0, + "step": 23237 + }, + { + "epoch": 2.956112453886274, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7749876976013184, + "num_tokens": 886532854.0, + "step": 23238 + }, + { + "epoch": 2.9562396641648645, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7985324859619141, + "num_tokens": 886564288.0, + "step": 23239 + }, + { + "epoch": 2.956366874443455, + "grad_norm": 14.9375, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.8067876100540161, + "num_tokens": 886597155.0, + "step": 23240 + }, + { + "epoch": 2.9564940847220456, + "grad_norm": 21.875, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.806674063205719, + "num_tokens": 886632015.0, + "step": 23241 + }, + { + "epoch": 2.956621295000636, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8151021599769592, + "num_tokens": 886665736.0, + "step": 23242 + }, + { + "epoch": 2.9567485052792266, + "grad_norm": 15.75, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.805665910243988, + "num_tokens": 886702652.0, + "step": 23243 + }, + { + "epoch": 2.956875715557817, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.798565149307251, + "num_tokens": 886743573.0, + "step": 23244 + }, + { + "epoch": 2.9570029258364077, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.8053181171417236, + "num_tokens": 886777932.0, + "step": 23245 + }, + { + "epoch": 2.957130136114998, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7887935638427734, + "num_tokens": 886813211.0, + "step": 23246 + }, + { + "epoch": 2.9572573463935887, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8058942556381226, + "num_tokens": 886849416.0, + "step": 23247 + }, + { + "epoch": 2.9573845566721793, + "grad_norm": 27.5, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7893334627151489, + "num_tokens": 886887983.0, + "step": 23248 + }, + { + "epoch": 2.95751176695077, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7909592986106873, + "num_tokens": 886930867.0, + "step": 23249 + }, + { + "epoch": 2.95763897722936, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.8041877746582031, + "num_tokens": 886971862.0, + "step": 23250 + }, + { + "epoch": 2.957766187507951, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.8082301616668701, + "num_tokens": 887005238.0, + "step": 23251 + }, + { + "epoch": 2.957893397786541, + "grad_norm": 89.5, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7852369546890259, + "num_tokens": 887035588.0, + "step": 23252 + }, + { + "epoch": 2.958020608065132, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.8034486770629883, + "num_tokens": 887079962.0, + "step": 23253 + }, + { + "epoch": 2.958147818343722, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.8237161636352539, + "num_tokens": 887114572.0, + "step": 23254 + }, + { + "epoch": 2.958275028622313, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.8065330386161804, + "num_tokens": 887149396.0, + "step": 23255 + }, + { + "epoch": 2.958402238900903, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.797109842300415, + "num_tokens": 887191916.0, + "step": 23256 + }, + { + "epoch": 2.9585294491794936, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.8127328157424927, + "num_tokens": 887228263.0, + "step": 23257 + }, + { + "epoch": 2.958656659458084, + "grad_norm": 24.875, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8163139820098877, + "num_tokens": 887258688.0, + "step": 23258 + }, + { + "epoch": 2.9587838697366746, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7989558577537537, + "num_tokens": 887301565.0, + "step": 23259 + }, + { + "epoch": 2.958911080015265, + "grad_norm": 46.75, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.797287106513977, + "num_tokens": 887337511.0, + "step": 23260 + }, + { + "epoch": 2.9590382902938557, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7998096942901611, + "num_tokens": 887370694.0, + "step": 23261 + }, + { + "epoch": 2.959165500572446, + "grad_norm": 52.25, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7942720651626587, + "num_tokens": 887412131.0, + "step": 23262 + }, + { + "epoch": 2.9592927108510367, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7991365790367126, + "num_tokens": 887447179.0, + "step": 23263 + }, + { + "epoch": 2.9594199211296273, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.795330286026001, + "num_tokens": 887486079.0, + "step": 23264 + }, + { + "epoch": 2.959547131408218, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8117307424545288, + "num_tokens": 887519092.0, + "step": 23265 + }, + { + "epoch": 2.9596743416868083, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7981191873550415, + "num_tokens": 887552008.0, + "step": 23266 + }, + { + "epoch": 2.959801551965399, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7967509031295776, + "num_tokens": 887585827.0, + "step": 23267 + }, + { + "epoch": 2.9599287622439894, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8034098744392395, + "num_tokens": 887628041.0, + "step": 23268 + }, + { + "epoch": 2.96005597252258, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8101270198822021, + "num_tokens": 887667210.0, + "step": 23269 + }, + { + "epoch": 2.9601831828011704, + "grad_norm": 41.75, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.8051685690879822, + "num_tokens": 887705429.0, + "step": 23270 + }, + { + "epoch": 2.960310393079761, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.786246120929718, + "num_tokens": 887753188.0, + "step": 23271 + }, + { + "epoch": 2.9604376033583515, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7846724987030029, + "num_tokens": 887795907.0, + "step": 23272 + }, + { + "epoch": 2.960564813636942, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.8025881052017212, + "num_tokens": 887834859.0, + "step": 23273 + }, + { + "epoch": 2.9606920239155325, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8008375763893127, + "num_tokens": 887879041.0, + "step": 23274 + }, + { + "epoch": 2.9608192341941226, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7931863069534302, + "num_tokens": 887913752.0, + "step": 23275 + }, + { + "epoch": 2.9609464444727136, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8117153644561768, + "num_tokens": 887950946.0, + "step": 23276 + }, + { + "epoch": 2.9610736547513037, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7940669059753418, + "num_tokens": 887993155.0, + "step": 23277 + }, + { + "epoch": 2.9612008650298947, + "grad_norm": 17.75, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.8010083436965942, + "num_tokens": 888027831.0, + "step": 23278 + }, + { + "epoch": 2.9613280753084847, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.8114922642707825, + "num_tokens": 888066941.0, + "step": 23279 + }, + { + "epoch": 2.9614552855870753, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.8111984729766846, + "num_tokens": 888099739.0, + "step": 23280 + }, + { + "epoch": 2.961582495865666, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.8048003911972046, + "num_tokens": 888139675.0, + "step": 23281 + }, + { + "epoch": 2.9617097061442563, + "grad_norm": 43.25, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7862731218338013, + "num_tokens": 888179374.0, + "step": 23282 + }, + { + "epoch": 2.961836916422847, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7850866913795471, + "num_tokens": 888214139.0, + "step": 23283 + }, + { + "epoch": 2.9619641267014374, + "grad_norm": 13.9375, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7984467148780823, + "num_tokens": 888246365.0, + "step": 23284 + }, + { + "epoch": 2.962091336980028, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7977772951126099, + "num_tokens": 888290903.0, + "step": 23285 + }, + { + "epoch": 2.9622185472586184, + "grad_norm": 13.0, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.8002758026123047, + "num_tokens": 888331329.0, + "step": 23286 + }, + { + "epoch": 2.962345757537209, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.79041987657547, + "num_tokens": 888372411.0, + "step": 23287 + }, + { + "epoch": 2.9624729678157995, + "grad_norm": 24.5, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7990301847457886, + "num_tokens": 888409778.0, + "step": 23288 + }, + { + "epoch": 2.96260017809439, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.8108495473861694, + "num_tokens": 888453593.0, + "step": 23289 + }, + { + "epoch": 2.9627273883729806, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.814435601234436, + "num_tokens": 888488245.0, + "step": 23290 + }, + { + "epoch": 2.962854598651571, + "grad_norm": 27.0, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.8031224608421326, + "num_tokens": 888524306.0, + "step": 23291 + }, + { + "epoch": 2.9629818089301616, + "grad_norm": 25.625, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8148723840713501, + "num_tokens": 888563791.0, + "step": 23292 + }, + { + "epoch": 2.963109019208752, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.8120951056480408, + "num_tokens": 888605021.0, + "step": 23293 + }, + { + "epoch": 2.9632362294873427, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.8054260015487671, + "num_tokens": 888647382.0, + "step": 23294 + }, + { + "epoch": 2.963363439765933, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.8003154993057251, + "num_tokens": 888682174.0, + "step": 23295 + }, + { + "epoch": 2.9634906500445237, + "grad_norm": 14.4375, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.8055906891822815, + "num_tokens": 888724401.0, + "step": 23296 + }, + { + "epoch": 2.9636178603231143, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8120361566543579, + "num_tokens": 888766230.0, + "step": 23297 + }, + { + "epoch": 2.963745070601705, + "grad_norm": 17.125, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7779028415679932, + "num_tokens": 888811865.0, + "step": 23298 + }, + { + "epoch": 2.9638722808802953, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.8022596836090088, + "num_tokens": 888849549.0, + "step": 23299 + }, + { + "epoch": 2.9639994911588854, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7990211248397827, + "num_tokens": 888881305.0, + "step": 23300 + }, + { + "epoch": 2.9641267014374764, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8077784776687622, + "num_tokens": 888917862.0, + "step": 23301 + }, + { + "epoch": 2.9642539117160664, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.8056327700614929, + "num_tokens": 888953883.0, + "step": 23302 + }, + { + "epoch": 2.9643811219946574, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8251365423202515, + "num_tokens": 888991052.0, + "step": 23303 + }, + { + "epoch": 2.9645083322732475, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.8048872947692871, + "num_tokens": 889022523.0, + "step": 23304 + }, + { + "epoch": 2.964635542551838, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7877078652381897, + "num_tokens": 889063470.0, + "step": 23305 + }, + { + "epoch": 2.9647627528304286, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.8107919096946716, + "num_tokens": 889101160.0, + "step": 23306 + }, + { + "epoch": 2.964889963109019, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7993056178092957, + "num_tokens": 889139128.0, + "step": 23307 + }, + { + "epoch": 2.9650171733876096, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7804019451141357, + "num_tokens": 889175574.0, + "step": 23308 + }, + { + "epoch": 2.9651443836662, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7971204519271851, + "num_tokens": 889216520.0, + "step": 23309 + }, + { + "epoch": 2.9652715939447907, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.8172353506088257, + "num_tokens": 889256746.0, + "step": 23310 + }, + { + "epoch": 2.965398804223381, + "grad_norm": 29.25, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7977578639984131, + "num_tokens": 889298868.0, + "step": 23311 + }, + { + "epoch": 2.9655260145019717, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7954028844833374, + "num_tokens": 889337758.0, + "step": 23312 + }, + { + "epoch": 2.9656532247805623, + "grad_norm": 19.375, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.8124455213546753, + "num_tokens": 889375308.0, + "step": 23313 + }, + { + "epoch": 2.965780435059153, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8099406361579895, + "num_tokens": 889413735.0, + "step": 23314 + }, + { + "epoch": 2.9659076453377433, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.8013027906417847, + "num_tokens": 889451585.0, + "step": 23315 + }, + { + "epoch": 2.966034855616334, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.785040020942688, + "num_tokens": 889487405.0, + "step": 23316 + }, + { + "epoch": 2.9661620658949244, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.781504213809967, + "num_tokens": 889523051.0, + "step": 23317 + }, + { + "epoch": 2.966289276173515, + "grad_norm": 20.0, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7809849381446838, + "num_tokens": 889559822.0, + "step": 23318 + }, + { + "epoch": 2.9664164864521054, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7970130443572998, + "num_tokens": 889591598.0, + "step": 23319 + }, + { + "epoch": 2.966543696730696, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.8109786510467529, + "num_tokens": 889629787.0, + "step": 23320 + }, + { + "epoch": 2.9666709070092865, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.8044965267181396, + "num_tokens": 889668730.0, + "step": 23321 + }, + { + "epoch": 2.966798117287877, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7971718311309814, + "num_tokens": 889705215.0, + "step": 23322 + }, + { + "epoch": 2.966925327566467, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7977465391159058, + "num_tokens": 889738528.0, + "step": 23323 + }, + { + "epoch": 2.967052537845058, + "grad_norm": 21.625, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7989888191223145, + "num_tokens": 889771694.0, + "step": 23324 + }, + { + "epoch": 2.967179748123648, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8184996843338013, + "num_tokens": 889806096.0, + "step": 23325 + }, + { + "epoch": 2.967306958402239, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.8057511448860168, + "num_tokens": 889848514.0, + "step": 23326 + }, + { + "epoch": 2.967434168680829, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.5832, + "mean_token_accuracy": 0.8255620002746582, + "num_tokens": 889889689.0, + "step": 23327 + }, + { + "epoch": 2.96756137895942, + "grad_norm": 14.6875, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8077700138092041, + "num_tokens": 889925550.0, + "step": 23328 + }, + { + "epoch": 2.9676885892380103, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7864765524864197, + "num_tokens": 889961797.0, + "step": 23329 + }, + { + "epoch": 2.967815799516601, + "grad_norm": 31.0, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.8044390082359314, + "num_tokens": 889997935.0, + "step": 23330 + }, + { + "epoch": 2.9679430097951913, + "grad_norm": 17.5, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8155182600021362, + "num_tokens": 890037900.0, + "step": 23331 + }, + { + "epoch": 2.968070220073782, + "grad_norm": 42.25, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8019815683364868, + "num_tokens": 890074814.0, + "step": 23332 + }, + { + "epoch": 2.9681974303523724, + "grad_norm": 76.0, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8127084970474243, + "num_tokens": 890107091.0, + "step": 23333 + }, + { + "epoch": 2.968324640630963, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.8125849962234497, + "num_tokens": 890140558.0, + "step": 23334 + }, + { + "epoch": 2.9684518509095534, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.8016624450683594, + "num_tokens": 890174563.0, + "step": 23335 + }, + { + "epoch": 2.968579061188144, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.8010764718055725, + "num_tokens": 890210584.0, + "step": 23336 + }, + { + "epoch": 2.9687062714667345, + "grad_norm": 13.5, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.8223997950553894, + "num_tokens": 890251955.0, + "step": 23337 + }, + { + "epoch": 2.968833481745325, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.8008257746696472, + "num_tokens": 890289095.0, + "step": 23338 + }, + { + "epoch": 2.9689606920239155, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.8052573204040527, + "num_tokens": 890327864.0, + "step": 23339 + }, + { + "epoch": 2.969087902302506, + "grad_norm": 33.75, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.803034782409668, + "num_tokens": 890364244.0, + "step": 23340 + }, + { + "epoch": 2.9692151125810966, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7891076803207397, + "num_tokens": 890397652.0, + "step": 23341 + }, + { + "epoch": 2.969342322859687, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.8139433860778809, + "num_tokens": 890435598.0, + "step": 23342 + }, + { + "epoch": 2.9694695331382777, + "grad_norm": 22.5, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.8093950748443604, + "num_tokens": 890473445.0, + "step": 23343 + }, + { + "epoch": 2.969596743416868, + "grad_norm": 20.25, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8099809885025024, + "num_tokens": 890514263.0, + "step": 23344 + }, + { + "epoch": 2.9697239536954587, + "grad_norm": 12.5625, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.807350754737854, + "num_tokens": 890554123.0, + "step": 23345 + }, + { + "epoch": 2.9698511639740492, + "grad_norm": 27.25, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.8023113012313843, + "num_tokens": 890593938.0, + "step": 23346 + }, + { + "epoch": 2.9699783742526398, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7907543182373047, + "num_tokens": 890632260.0, + "step": 23347 + }, + { + "epoch": 2.97010558453123, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7960336804389954, + "num_tokens": 890666852.0, + "step": 23348 + }, + { + "epoch": 2.970232794809821, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7994173765182495, + "num_tokens": 890707959.0, + "step": 23349 + }, + { + "epoch": 2.970360005088411, + "grad_norm": 25.375, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7829053401947021, + "num_tokens": 890748729.0, + "step": 23350 + }, + { + "epoch": 2.970487215367002, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.813066303730011, + "num_tokens": 890782116.0, + "step": 23351 + }, + { + "epoch": 2.970614425645592, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.791388213634491, + "num_tokens": 890827309.0, + "step": 23352 + }, + { + "epoch": 2.970741635924183, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7914660573005676, + "num_tokens": 890863879.0, + "step": 23353 + }, + { + "epoch": 2.970868846202773, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8026164770126343, + "num_tokens": 890901184.0, + "step": 23354 + }, + { + "epoch": 2.9709960564813636, + "grad_norm": 33.25, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.8140484094619751, + "num_tokens": 890937779.0, + "step": 23355 + }, + { + "epoch": 2.971123266759954, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.8221741914749146, + "num_tokens": 890978055.0, + "step": 23356 + }, + { + "epoch": 2.9712504770385446, + "grad_norm": 35.0, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7613600492477417, + "num_tokens": 891010985.0, + "step": 23357 + }, + { + "epoch": 2.971377687317135, + "grad_norm": 18.5, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.809839129447937, + "num_tokens": 891045904.0, + "step": 23358 + }, + { + "epoch": 2.9715048975957257, + "grad_norm": 49.75, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8071945905685425, + "num_tokens": 891086240.0, + "step": 23359 + }, + { + "epoch": 2.971632107874316, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7970043420791626, + "num_tokens": 891126482.0, + "step": 23360 + }, + { + "epoch": 2.9717593181529067, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7943377494812012, + "num_tokens": 891165982.0, + "step": 23361 + }, + { + "epoch": 2.9718865284314973, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7808071970939636, + "num_tokens": 891199101.0, + "step": 23362 + }, + { + "epoch": 2.972013738710088, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.8054423928260803, + "num_tokens": 891231511.0, + "step": 23363 + }, + { + "epoch": 2.9721409489886783, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7914212942123413, + "num_tokens": 891262369.0, + "step": 23364 + }, + { + "epoch": 2.972268159267269, + "grad_norm": 16.875, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.794136643409729, + "num_tokens": 891304206.0, + "step": 23365 + }, + { + "epoch": 2.9723953695458594, + "grad_norm": 21.5, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.790397047996521, + "num_tokens": 891348006.0, + "step": 23366 + }, + { + "epoch": 2.97252257982445, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8086450099945068, + "num_tokens": 891386851.0, + "step": 23367 + }, + { + "epoch": 2.9726497901030404, + "grad_norm": 44.75, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.81245356798172, + "num_tokens": 891423582.0, + "step": 23368 + }, + { + "epoch": 2.972777000381631, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.8200651407241821, + "num_tokens": 891463773.0, + "step": 23369 + }, + { + "epoch": 2.9729042106602215, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.8138382434844971, + "num_tokens": 891498117.0, + "step": 23370 + }, + { + "epoch": 2.973031420938812, + "grad_norm": 16.0, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.804908275604248, + "num_tokens": 891529055.0, + "step": 23371 + }, + { + "epoch": 2.9731586312174025, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8148775696754456, + "num_tokens": 891567370.0, + "step": 23372 + }, + { + "epoch": 2.9732858414959926, + "grad_norm": 31.25, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8117457032203674, + "num_tokens": 891602737.0, + "step": 23373 + }, + { + "epoch": 2.9734130517745836, + "grad_norm": 23.25, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.8003126382827759, + "num_tokens": 891639713.0, + "step": 23374 + }, + { + "epoch": 2.9735402620531737, + "grad_norm": 13.4375, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8153716325759888, + "num_tokens": 891673931.0, + "step": 23375 + }, + { + "epoch": 2.9736674723317646, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.8036453723907471, + "num_tokens": 891708015.0, + "step": 23376 + }, + { + "epoch": 2.9737946826103547, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7981388568878174, + "num_tokens": 891746023.0, + "step": 23377 + }, + { + "epoch": 2.9739218928889453, + "grad_norm": 23.0, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7970972657203674, + "num_tokens": 891777767.0, + "step": 23378 + }, + { + "epoch": 2.974049103167536, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.8052855730056763, + "num_tokens": 891810816.0, + "step": 23379 + }, + { + "epoch": 2.9741763134461263, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7958920001983643, + "num_tokens": 891848810.0, + "step": 23380 + }, + { + "epoch": 2.974303523724717, + "grad_norm": 14.625, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.8085423111915588, + "num_tokens": 891892553.0, + "step": 23381 + }, + { + "epoch": 2.9744307340033074, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8152496218681335, + "num_tokens": 891927539.0, + "step": 23382 + }, + { + "epoch": 2.974557944281898, + "grad_norm": 17.25, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8037636876106262, + "num_tokens": 891959577.0, + "step": 23383 + }, + { + "epoch": 2.9746851545604884, + "grad_norm": 15.8125, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7767611145973206, + "num_tokens": 892000799.0, + "step": 23384 + }, + { + "epoch": 2.974812364839079, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7998823523521423, + "num_tokens": 892042200.0, + "step": 23385 + }, + { + "epoch": 2.9749395751176695, + "grad_norm": 15.9375, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.793508768081665, + "num_tokens": 892084838.0, + "step": 23386 + }, + { + "epoch": 2.97506678539626, + "grad_norm": 13.875, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7889848351478577, + "num_tokens": 892126442.0, + "step": 23387 + }, + { + "epoch": 2.9751939956748505, + "grad_norm": 25.75, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7898294925689697, + "num_tokens": 892165019.0, + "step": 23388 + }, + { + "epoch": 2.975321205953441, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.806128978729248, + "num_tokens": 892202291.0, + "step": 23389 + }, + { + "epoch": 2.9754484162320316, + "grad_norm": 37.25, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7873282432556152, + "num_tokens": 892235159.0, + "step": 23390 + }, + { + "epoch": 2.975575626510622, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7910960912704468, + "num_tokens": 892271967.0, + "step": 23391 + }, + { + "epoch": 2.9757028367892127, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7848135232925415, + "num_tokens": 892306230.0, + "step": 23392 + }, + { + "epoch": 2.975830047067803, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7991645336151123, + "num_tokens": 892342118.0, + "step": 23393 + }, + { + "epoch": 2.9759572573463937, + "grad_norm": 12.25, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.8087738156318665, + "num_tokens": 892380764.0, + "step": 23394 + }, + { + "epoch": 2.9760844676249842, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7991184592247009, + "num_tokens": 892421113.0, + "step": 23395 + }, + { + "epoch": 2.9762116779035748, + "grad_norm": 33.5, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8098431825637817, + "num_tokens": 892457512.0, + "step": 23396 + }, + { + "epoch": 2.9763388881821653, + "grad_norm": 43.5, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.8000978827476501, + "num_tokens": 892493599.0, + "step": 23397 + }, + { + "epoch": 2.9764660984607554, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7886752486228943, + "num_tokens": 892528474.0, + "step": 23398 + }, + { + "epoch": 2.9765933087393464, + "grad_norm": 24.375, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.8097766041755676, + "num_tokens": 892569591.0, + "step": 23399 + }, + { + "epoch": 2.9767205190179364, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.8056765794754028, + "num_tokens": 892607185.0, + "step": 23400 + }, + { + "epoch": 2.9768477292965274, + "grad_norm": 24.125, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8052884340286255, + "num_tokens": 892641453.0, + "step": 23401 + }, + { + "epoch": 2.9769749395751175, + "grad_norm": 39.25, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7938292026519775, + "num_tokens": 892676802.0, + "step": 23402 + }, + { + "epoch": 2.977102149853708, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8099054098129272, + "num_tokens": 892711694.0, + "step": 23403 + }, + { + "epoch": 2.9772293601322986, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7953394651412964, + "num_tokens": 892752662.0, + "step": 23404 + }, + { + "epoch": 2.977356570410889, + "grad_norm": 30.5, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7928504943847656, + "num_tokens": 892790170.0, + "step": 23405 + }, + { + "epoch": 2.9774837806894796, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7862101197242737, + "num_tokens": 892831437.0, + "step": 23406 + }, + { + "epoch": 2.97761099096807, + "grad_norm": 16.75, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8228759169578552, + "num_tokens": 892869022.0, + "step": 23407 + }, + { + "epoch": 2.9777382012466607, + "grad_norm": 41.25, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.8006856441497803, + "num_tokens": 892910738.0, + "step": 23408 + }, + { + "epoch": 2.977865411525251, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.8126055598258972, + "num_tokens": 892952055.0, + "step": 23409 + }, + { + "epoch": 2.9779926218038417, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7905644178390503, + "num_tokens": 892985843.0, + "step": 23410 + }, + { + "epoch": 2.9781198320824323, + "grad_norm": 14.1875, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7868772745132446, + "num_tokens": 893031197.0, + "step": 23411 + }, + { + "epoch": 2.978247042361023, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.8180812001228333, + "num_tokens": 893068113.0, + "step": 23412 + }, + { + "epoch": 2.9783742526396133, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.8191283941268921, + "num_tokens": 893109148.0, + "step": 23413 + }, + { + "epoch": 2.978501462918204, + "grad_norm": 24.625, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.8069060444831848, + "num_tokens": 893151406.0, + "step": 23414 + }, + { + "epoch": 2.9786286731967944, + "grad_norm": 14.25, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.806753396987915, + "num_tokens": 893190209.0, + "step": 23415 + }, + { + "epoch": 2.978755883475385, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.8006463646888733, + "num_tokens": 893232949.0, + "step": 23416 + }, + { + "epoch": 2.9788830937539754, + "grad_norm": 28.625, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.8024361729621887, + "num_tokens": 893273931.0, + "step": 23417 + }, + { + "epoch": 2.979010304032566, + "grad_norm": 35.75, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7836529612541199, + "num_tokens": 893317465.0, + "step": 23418 + }, + { + "epoch": 2.9791375143111565, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7854719161987305, + "num_tokens": 893354089.0, + "step": 23419 + }, + { + "epoch": 2.979264724589747, + "grad_norm": 26.5, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7964848279953003, + "num_tokens": 893388051.0, + "step": 23420 + }, + { + "epoch": 2.979391934868337, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7939242124557495, + "num_tokens": 893426191.0, + "step": 23421 + }, + { + "epoch": 2.979519145146928, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7928735017776489, + "num_tokens": 893462134.0, + "step": 23422 + }, + { + "epoch": 2.979646355425518, + "grad_norm": 15.0625, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7981149554252625, + "num_tokens": 893503159.0, + "step": 23423 + }, + { + "epoch": 2.979773565704109, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7791959047317505, + "num_tokens": 893538379.0, + "step": 23424 + }, + { + "epoch": 2.979900775982699, + "grad_norm": 26.625, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7872415781021118, + "num_tokens": 893580431.0, + "step": 23425 + }, + { + "epoch": 2.98002798626129, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7939370274543762, + "num_tokens": 893622412.0, + "step": 23426 + }, + { + "epoch": 2.9801551965398803, + "grad_norm": 20.625, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.8019781708717346, + "num_tokens": 893658575.0, + "step": 23427 + }, + { + "epoch": 2.980282406818471, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.813071608543396, + "num_tokens": 893702812.0, + "step": 23428 + }, + { + "epoch": 2.9804096170970613, + "grad_norm": 31.375, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7890788912773132, + "num_tokens": 893742785.0, + "step": 23429 + }, + { + "epoch": 2.980536827375652, + "grad_norm": 17.625, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.8086478114128113, + "num_tokens": 893782310.0, + "step": 23430 + }, + { + "epoch": 2.9806640376542424, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8150994181632996, + "num_tokens": 893822794.0, + "step": 23431 + }, + { + "epoch": 2.980791247932833, + "grad_norm": 12.5, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.8139882683753967, + "num_tokens": 893870462.0, + "step": 23432 + }, + { + "epoch": 2.9809184582114234, + "grad_norm": 35.5, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.803036630153656, + "num_tokens": 893910654.0, + "step": 23433 + }, + { + "epoch": 2.981045668490014, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7984337210655212, + "num_tokens": 893948819.0, + "step": 23434 + }, + { + "epoch": 2.9811728787686045, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8119094371795654, + "num_tokens": 893983325.0, + "step": 23435 + }, + { + "epoch": 2.981300089047195, + "grad_norm": 13.0625, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7987570762634277, + "num_tokens": 894023712.0, + "step": 23436 + }, + { + "epoch": 2.9814272993257855, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.8109191656112671, + "num_tokens": 894059719.0, + "step": 23437 + }, + { + "epoch": 2.981554509604376, + "grad_norm": 38.5, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.8021754622459412, + "num_tokens": 894100150.0, + "step": 23438 + }, + { + "epoch": 2.9816817198829666, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7757043838500977, + "num_tokens": 894134336.0, + "step": 23439 + }, + { + "epoch": 2.981808930161557, + "grad_norm": 19.0, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7863202095031738, + "num_tokens": 894168972.0, + "step": 23440 + }, + { + "epoch": 2.9819361404401477, + "grad_norm": 22.25, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7995175719261169, + "num_tokens": 894204641.0, + "step": 23441 + }, + { + "epoch": 2.982063350718738, + "grad_norm": 39.0, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.8037983775138855, + "num_tokens": 894248403.0, + "step": 23442 + }, + { + "epoch": 2.9821905609973287, + "grad_norm": 11.9375, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8219550848007202, + "num_tokens": 894284035.0, + "step": 23443 + }, + { + "epoch": 2.9823177712759192, + "grad_norm": 12.625, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.8043978214263916, + "num_tokens": 894321804.0, + "step": 23444 + }, + { + "epoch": 2.9824449815545098, + "grad_norm": 23.75, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.8081908226013184, + "num_tokens": 894355648.0, + "step": 23445 + }, + { + "epoch": 2.9825721918331, + "grad_norm": 26.875, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7998255491256714, + "num_tokens": 894392947.0, + "step": 23446 + }, + { + "epoch": 2.982699402111691, + "grad_norm": 54.25, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7939543724060059, + "num_tokens": 894433437.0, + "step": 23447 + }, + { + "epoch": 2.982826612390281, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.8056287169456482, + "num_tokens": 894481428.0, + "step": 23448 + }, + { + "epoch": 2.982953822668872, + "grad_norm": 31.5, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.8041871786117554, + "num_tokens": 894518055.0, + "step": 23449 + }, + { + "epoch": 2.983081032947462, + "grad_norm": 22.875, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7951593399047852, + "num_tokens": 894562693.0, + "step": 23450 + }, + { + "epoch": 2.983208243226053, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.801223874092102, + "num_tokens": 894604973.0, + "step": 23451 + }, + { + "epoch": 2.983335453504643, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.8065989017486572, + "num_tokens": 894645672.0, + "step": 23452 + }, + { + "epoch": 2.9834626637832335, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.8002519011497498, + "num_tokens": 894686702.0, + "step": 23453 + }, + { + "epoch": 2.983589874061824, + "grad_norm": 34.25, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8101280927658081, + "num_tokens": 894722768.0, + "step": 23454 + }, + { + "epoch": 2.9837170843404146, + "grad_norm": 23.5, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.8163784742355347, + "num_tokens": 894759216.0, + "step": 23455 + }, + { + "epoch": 2.983844294619005, + "grad_norm": 37.75, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.80609130859375, + "num_tokens": 894799691.0, + "step": 23456 + }, + { + "epoch": 2.9839715048975957, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7959619760513306, + "num_tokens": 894835869.0, + "step": 23457 + }, + { + "epoch": 2.984098715176186, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7782901525497437, + "num_tokens": 894875706.0, + "step": 23458 + }, + { + "epoch": 2.9842259254547767, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.8109861612319946, + "num_tokens": 894920135.0, + "step": 23459 + }, + { + "epoch": 2.9843531357333672, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8029007911682129, + "num_tokens": 894960825.0, + "step": 23460 + }, + { + "epoch": 2.9844803460119578, + "grad_norm": 38.0, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.8048272132873535, + "num_tokens": 895001935.0, + "step": 23461 + }, + { + "epoch": 2.9846075562905483, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.8108749389648438, + "num_tokens": 895033559.0, + "step": 23462 + }, + { + "epoch": 2.984734766569139, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.8044626116752625, + "num_tokens": 895070958.0, + "step": 23463 + }, + { + "epoch": 2.9848619768477294, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7990855574607849, + "num_tokens": 895111533.0, + "step": 23464 + }, + { + "epoch": 2.98498918712632, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7989513874053955, + "num_tokens": 895149861.0, + "step": 23465 + }, + { + "epoch": 2.9851163974049104, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.8096050024032593, + "num_tokens": 895192958.0, + "step": 23466 + }, + { + "epoch": 2.985243607683501, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.8124852180480957, + "num_tokens": 895226710.0, + "step": 23467 + }, + { + "epoch": 2.9853708179620915, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.8037947416305542, + "num_tokens": 895267695.0, + "step": 23468 + }, + { + "epoch": 2.985498028240682, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.8161410689353943, + "num_tokens": 895303101.0, + "step": 23469 + }, + { + "epoch": 2.9856252385192725, + "grad_norm": 30.125, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8070226907730103, + "num_tokens": 895340437.0, + "step": 23470 + }, + { + "epoch": 2.9857524487978626, + "grad_norm": 10.875, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8187023401260376, + "num_tokens": 895377695.0, + "step": 23471 + }, + { + "epoch": 2.9858796590764536, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7954025268554688, + "num_tokens": 895418599.0, + "step": 23472 + }, + { + "epoch": 2.9860068693550437, + "grad_norm": 37.5, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.8033403754234314, + "num_tokens": 895454832.0, + "step": 23473 + }, + { + "epoch": 2.9861340796336346, + "grad_norm": 24.25, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7844590544700623, + "num_tokens": 895490607.0, + "step": 23474 + }, + { + "epoch": 2.9862612899122247, + "grad_norm": 26.25, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7999138236045837, + "num_tokens": 895531340.0, + "step": 23475 + }, + { + "epoch": 2.9863885001908153, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.8002408742904663, + "num_tokens": 895568688.0, + "step": 23476 + }, + { + "epoch": 2.986515710469406, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7958660125732422, + "num_tokens": 895607316.0, + "step": 23477 + }, + { + "epoch": 2.9866429207479963, + "grad_norm": 14.3125, + "learning_rate": 1e-06, + "loss": 0.5813, + "mean_token_accuracy": 0.8252134323120117, + "num_tokens": 895648694.0, + "step": 23478 + }, + { + "epoch": 2.986770131026587, + "grad_norm": 17.0, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7992435693740845, + "num_tokens": 895687996.0, + "step": 23479 + }, + { + "epoch": 2.9868973413051774, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.8089264035224915, + "num_tokens": 895726727.0, + "step": 23480 + }, + { + "epoch": 2.987024551583768, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.8078712224960327, + "num_tokens": 895760688.0, + "step": 23481 + }, + { + "epoch": 2.9871517618623584, + "grad_norm": 24.75, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7816870808601379, + "num_tokens": 895797303.0, + "step": 23482 + }, + { + "epoch": 2.987278972140949, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.8052340745925903, + "num_tokens": 895834078.0, + "step": 23483 + }, + { + "epoch": 2.9874061824195395, + "grad_norm": 18.0, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7895950078964233, + "num_tokens": 895872640.0, + "step": 23484 + }, + { + "epoch": 2.98753339269813, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.8166066408157349, + "num_tokens": 895907646.0, + "step": 23485 + }, + { + "epoch": 2.9876606029767205, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.8170262575149536, + "num_tokens": 895944183.0, + "step": 23486 + }, + { + "epoch": 2.987787813255311, + "grad_norm": 11.375, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.8000386953353882, + "num_tokens": 895983987.0, + "step": 23487 + }, + { + "epoch": 2.9879150235339016, + "grad_norm": 15.5625, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7909870147705078, + "num_tokens": 896020928.0, + "step": 23488 + }, + { + "epoch": 2.988042233812492, + "grad_norm": 26.125, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.8102755546569824, + "num_tokens": 896057459.0, + "step": 23489 + }, + { + "epoch": 2.9881694440910826, + "grad_norm": 18.25, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8016800880432129, + "num_tokens": 896095049.0, + "step": 23490 + }, + { + "epoch": 2.988296654369673, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.8157375454902649, + "num_tokens": 896127433.0, + "step": 23491 + }, + { + "epoch": 2.9884238646482637, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8096410036087036, + "num_tokens": 896165662.0, + "step": 23492 + }, + { + "epoch": 2.9885510749268542, + "grad_norm": 36.75, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7972355484962463, + "num_tokens": 896207753.0, + "step": 23493 + }, + { + "epoch": 2.9886782852054448, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.8028606176376343, + "num_tokens": 896249105.0, + "step": 23494 + }, + { + "epoch": 2.9888054954840353, + "grad_norm": 32.0, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.8022890090942383, + "num_tokens": 896291728.0, + "step": 23495 + }, + { + "epoch": 2.9889327057626254, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7872810363769531, + "num_tokens": 896331729.0, + "step": 23496 + }, + { + "epoch": 2.9890599160412163, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.790804922580719, + "num_tokens": 896371804.0, + "step": 23497 + }, + { + "epoch": 2.9891871263198064, + "grad_norm": 25.5, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7996092438697815, + "num_tokens": 896408398.0, + "step": 23498 + }, + { + "epoch": 2.9893143365983974, + "grad_norm": 41.0, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.8185471892356873, + "num_tokens": 896440496.0, + "step": 23499 + }, + { + "epoch": 2.9894415468769875, + "grad_norm": 13.5625, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.8038768768310547, + "num_tokens": 896475212.0, + "step": 23500 + }, + { + "epoch": 2.989568757155578, + "grad_norm": 16.625, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7991974353790283, + "num_tokens": 896517122.0, + "step": 23501 + }, + { + "epoch": 2.9896959674341685, + "grad_norm": 13.6875, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8098662495613098, + "num_tokens": 896557259.0, + "step": 23502 + }, + { + "epoch": 2.989823177712759, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7908530831336975, + "num_tokens": 896600237.0, + "step": 23503 + }, + { + "epoch": 2.9899503879913496, + "grad_norm": 11.5, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8233200311660767, + "num_tokens": 896640564.0, + "step": 23504 + }, + { + "epoch": 2.99007759826994, + "grad_norm": 27.625, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8120038509368896, + "num_tokens": 896674164.0, + "step": 23505 + }, + { + "epoch": 2.9902048085485307, + "grad_norm": 20.875, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7965220212936401, + "num_tokens": 896716613.0, + "step": 23506 + }, + { + "epoch": 2.990332018827121, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7956426739692688, + "num_tokens": 896755586.0, + "step": 23507 + }, + { + "epoch": 2.9904592291057117, + "grad_norm": 43.75, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7994473576545715, + "num_tokens": 896797641.0, + "step": 23508 + }, + { + "epoch": 2.9905864393843022, + "grad_norm": 13.75, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7952657341957092, + "num_tokens": 896835113.0, + "step": 23509 + }, + { + "epoch": 2.9907136496628928, + "grad_norm": 25.25, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.8026061058044434, + "num_tokens": 896866496.0, + "step": 23510 + }, + { + "epoch": 2.9908408599414833, + "grad_norm": 15.0, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.8024252653121948, + "num_tokens": 896907535.0, + "step": 23511 + }, + { + "epoch": 2.990968070220074, + "grad_norm": 18.125, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.8216376304626465, + "num_tokens": 896948574.0, + "step": 23512 + }, + { + "epoch": 2.9910952804986644, + "grad_norm": 14.5625, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.797001838684082, + "num_tokens": 896992377.0, + "step": 23513 + }, + { + "epoch": 2.991222490777255, + "grad_norm": 30.375, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.8022940158843994, + "num_tokens": 897035411.0, + "step": 23514 + }, + { + "epoch": 2.9913497010558454, + "grad_norm": 20.5, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.8045307397842407, + "num_tokens": 897074661.0, + "step": 23515 + }, + { + "epoch": 2.991476911334436, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.8060163259506226, + "num_tokens": 897115550.0, + "step": 23516 + }, + { + "epoch": 2.9916041216130265, + "grad_norm": 18.375, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7920219898223877, + "num_tokens": 897155091.0, + "step": 23517 + }, + { + "epoch": 2.991731331891617, + "grad_norm": 11.125, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7992228269577026, + "num_tokens": 897196551.0, + "step": 23518 + }, + { + "epoch": 2.991858542170207, + "grad_norm": 12.75, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8101092576980591, + "num_tokens": 897231885.0, + "step": 23519 + }, + { + "epoch": 2.991985752448798, + "grad_norm": 63.0, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.8017897605895996, + "num_tokens": 897267944.0, + "step": 23520 + }, + { + "epoch": 2.992112962727388, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.8052047491073608, + "num_tokens": 897307330.0, + "step": 23521 + }, + { + "epoch": 2.992240173005979, + "grad_norm": 29.625, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.8049498796463013, + "num_tokens": 897344028.0, + "step": 23522 + }, + { + "epoch": 2.992367383284569, + "grad_norm": 23.375, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7960233688354492, + "num_tokens": 897391213.0, + "step": 23523 + }, + { + "epoch": 2.99249459356316, + "grad_norm": 44.5, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7972558736801147, + "num_tokens": 897428222.0, + "step": 23524 + }, + { + "epoch": 2.9926218038417502, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7828271985054016, + "num_tokens": 897457737.0, + "step": 23525 + }, + { + "epoch": 2.992749014120341, + "grad_norm": 19.25, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8104256391525269, + "num_tokens": 897498925.0, + "step": 23526 + }, + { + "epoch": 2.9928762243989313, + "grad_norm": 32.5, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7940637469291687, + "num_tokens": 897533157.0, + "step": 23527 + }, + { + "epoch": 2.993003434677522, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.8019813895225525, + "num_tokens": 897574159.0, + "step": 23528 + }, + { + "epoch": 2.9931306449561124, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7917149662971497, + "num_tokens": 897606970.0, + "step": 23529 + }, + { + "epoch": 2.993257855234703, + "grad_norm": 15.5, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7767945528030396, + "num_tokens": 897647725.0, + "step": 23530 + }, + { + "epoch": 2.9933850655132934, + "grad_norm": 28.25, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8047756552696228, + "num_tokens": 897687109.0, + "step": 23531 + }, + { + "epoch": 2.993512275791884, + "grad_norm": 20.75, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.8109108209609985, + "num_tokens": 897720994.0, + "step": 23532 + }, + { + "epoch": 2.9936394860704745, + "grad_norm": 12.6875, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8166255950927734, + "num_tokens": 897749855.0, + "step": 23533 + }, + { + "epoch": 2.993766696349065, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7976138591766357, + "num_tokens": 897792846.0, + "step": 23534 + }, + { + "epoch": 2.9938939066276555, + "grad_norm": 20.125, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.8018443584442139, + "num_tokens": 897836765.0, + "step": 23535 + }, + { + "epoch": 2.994021116906246, + "grad_norm": 52.0, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.8059840798377991, + "num_tokens": 897880918.0, + "step": 23536 + }, + { + "epoch": 2.9941483271848366, + "grad_norm": 54.0, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7978913187980652, + "num_tokens": 897916676.0, + "step": 23537 + }, + { + "epoch": 2.994275537463427, + "grad_norm": 28.125, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8114397525787354, + "num_tokens": 897953968.0, + "step": 23538 + }, + { + "epoch": 2.9944027477420176, + "grad_norm": 18.625, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.8049216866493225, + "num_tokens": 897993055.0, + "step": 23539 + }, + { + "epoch": 2.994529958020608, + "grad_norm": 21.25, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8161548376083374, + "num_tokens": 898030202.0, + "step": 23540 + }, + { + "epoch": 2.9946571682991987, + "grad_norm": 17.875, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.8004288673400879, + "num_tokens": 898067234.0, + "step": 23541 + }, + { + "epoch": 2.9947843785777892, + "grad_norm": 14.375, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8097091913223267, + "num_tokens": 898105423.0, + "step": 23542 + }, + { + "epoch": 2.9949115888563798, + "grad_norm": 16.125, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.8249247670173645, + "num_tokens": 898144550.0, + "step": 23543 + }, + { + "epoch": 2.99503879913497, + "grad_norm": 19.5, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.8018869757652283, + "num_tokens": 898179277.0, + "step": 23544 + }, + { + "epoch": 2.995166009413561, + "grad_norm": 11.625, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.8036192655563354, + "num_tokens": 898218518.0, + "step": 23545 + }, + { + "epoch": 2.995293219692151, + "grad_norm": 17.375, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7892003655433655, + "num_tokens": 898259266.0, + "step": 23546 + }, + { + "epoch": 2.995420429970742, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.8013321757316589, + "num_tokens": 898292648.0, + "step": 23547 + }, + { + "epoch": 2.995547640249332, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.8058183789253235, + "num_tokens": 898328413.0, + "step": 23548 + }, + { + "epoch": 2.995674850527923, + "grad_norm": 29.5, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.8062362670898438, + "num_tokens": 898376963.0, + "step": 23549 + }, + { + "epoch": 2.995802060806513, + "grad_norm": 26.75, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8054556846618652, + "num_tokens": 898407256.0, + "step": 23550 + }, + { + "epoch": 2.9959292710851035, + "grad_norm": 29.375, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.8054927587509155, + "num_tokens": 898448800.0, + "step": 23551 + }, + { + "epoch": 2.996056481363694, + "grad_norm": 68.0, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7996683120727539, + "num_tokens": 898488912.0, + "step": 23552 + }, + { + "epoch": 2.9961836916422846, + "grad_norm": 21.75, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7985756397247314, + "num_tokens": 898530434.0, + "step": 23553 + }, + { + "epoch": 2.996310901920875, + "grad_norm": 16.5, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.8032300472259521, + "num_tokens": 898565044.0, + "step": 23554 + }, + { + "epoch": 2.9964381121994657, + "grad_norm": 25.875, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8140817880630493, + "num_tokens": 898601798.0, + "step": 23555 + }, + { + "epoch": 2.996565322478056, + "grad_norm": 15.125, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7952897548675537, + "num_tokens": 898644789.0, + "step": 23556 + }, + { + "epoch": 2.9966925327566467, + "grad_norm": 53.25, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7981269359588623, + "num_tokens": 898683928.0, + "step": 23557 + }, + { + "epoch": 2.9968197430352372, + "grad_norm": 15.4375, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.8031554222106934, + "num_tokens": 898726217.0, + "step": 23558 + }, + { + "epoch": 2.9969469533138278, + "grad_norm": 30.25, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.8141093850135803, + "num_tokens": 898760575.0, + "step": 23559 + }, + { + "epoch": 2.9970741635924183, + "grad_norm": 18.875, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.8019090890884399, + "num_tokens": 898801232.0, + "step": 23560 + }, + { + "epoch": 2.997201373871009, + "grad_norm": 32.75, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.8057658672332764, + "num_tokens": 898837174.0, + "step": 23561 + }, + { + "epoch": 2.9973285841495994, + "grad_norm": 30.0, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7890905141830444, + "num_tokens": 898875849.0, + "step": 23562 + }, + { + "epoch": 2.99745579442819, + "grad_norm": 22.125, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.8082905411720276, + "num_tokens": 898909975.0, + "step": 23563 + }, + { + "epoch": 2.9975830047067804, + "grad_norm": 16.25, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7957477569580078, + "num_tokens": 898948250.0, + "step": 23564 + }, + { + "epoch": 2.997710214985371, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8186217546463013, + "num_tokens": 898989122.0, + "step": 23565 + }, + { + "epoch": 2.9978374252639615, + "grad_norm": 34.5, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7989555597305298, + "num_tokens": 899028715.0, + "step": 23566 + }, + { + "epoch": 2.997964635542552, + "grad_norm": 13.8125, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.8056244254112244, + "num_tokens": 899064101.0, + "step": 23567 + }, + { + "epoch": 2.9980918458211425, + "grad_norm": 18.75, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8053359985351562, + "num_tokens": 899101759.0, + "step": 23568 + }, + { + "epoch": 2.9982190560997326, + "grad_norm": 15.875, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7983601689338684, + "num_tokens": 899142273.0, + "step": 23569 + }, + { + "epoch": 2.9983462663783236, + "grad_norm": 20.375, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.8005895614624023, + "num_tokens": 899175878.0, + "step": 23570 + }, + { + "epoch": 2.9984734766569137, + "grad_norm": 28.0, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8187930583953857, + "num_tokens": 899208937.0, + "step": 23571 + }, + { + "epoch": 2.9986006869355046, + "grad_norm": 36.5, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7989977598190308, + "num_tokens": 899244088.0, + "step": 23572 + }, + { + "epoch": 2.9987278972140947, + "grad_norm": 25.0, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7978545427322388, + "num_tokens": 899280111.0, + "step": 23573 + }, + { + "epoch": 2.9988551074926852, + "grad_norm": 19.875, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.795586109161377, + "num_tokens": 899310882.0, + "step": 23574 + }, + { + "epoch": 2.9989823177712758, + "grad_norm": 28.75, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7890257835388184, + "num_tokens": 899346820.0, + "step": 23575 + }, + { + "epoch": 2.9991095280498663, + "grad_norm": 50.75, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8202102184295654, + "num_tokens": 899382044.0, + "step": 23576 + }, + { + "epoch": 2.999236738328457, + "grad_norm": 19.75, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.8072090148925781, + "num_tokens": 899418389.0, + "step": 23577 + }, + { + "epoch": 2.9993639486070474, + "grad_norm": 16.375, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.8042621612548828, + "num_tokens": 899458486.0, + "step": 23578 + }, + { + "epoch": 2.999491158885638, + "grad_norm": 39.75, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7999926805496216, + "num_tokens": 899498168.0, + "step": 23579 + }, + { + "epoch": 2.9996183691642284, + "grad_norm": 22.375, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.813302218914032, + "num_tokens": 899538655.0, + "step": 23580 + }, + { + "epoch": 2.999745579442819, + "grad_norm": 14.125, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7914713621139526, + "num_tokens": 899581693.0, + "step": 23581 + }, + { + "epoch": 2.9998727897214095, + "grad_norm": 37.0, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.8001245260238647, + "num_tokens": 899623817.0, + "step": 23582 + }, + { + "epoch": 3.0, + "grad_norm": 19.125, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.8061271905899048, + "num_tokens": 899664226.0, + "step": 23583 + }, + { + "epoch": 3.0, + "step": 23583, + "total_flos": 5.62815163329864e+19, + "train_loss": 0.6996442060487161, + "train_runtime": 42634.2383, + "train_samples_per_second": 8.85, + "train_steps_per_second": 0.553 + } + ], + "logging_steps": 1, + "max_steps": 23583, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 11792, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 5.62815163329864e+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..d913875 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5faca76471275d4c0ece2841f33bf46a1c8e9716388f09bc37a54d572dde953 +size 11921